Continuous Glucose Monitoring (CGM Real-Time)

EUnetHTA Joint Action 3 WP4

® Rapid assessment of other technologies using the HTA Core Model for Rapid Relative Effectiveness Assessment

Continuous glucose monitoring (CGM real -time) and flash glucose monitoring (FGM) as personal, standalone systems in patients with diabetes mellitus treated with insulin

Project ID: OTJA08

Version 1.4, 27 July 2018

This report is part of the project / joint action ‘724130 / EUnetHTA JA3’ which has received funding from the European Union’s Health Programme (2014-2020)

Dec2015 ©EUnetHTA, 2015. Reproduction is authorised provided EUnetHTA is explicitly acknowledged 1 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

DOCUMENT HISTORY AND CONTRIBUTORS

Version Date Description V1.0 22/04/2018 First draft. V1.1 28/05/2018 Input from dedicated reviewers has been processed. V1.2 21/06/18 Input from external experts and manufacturer(s) has been processed. V1.3 07/07/18 Input from medical editor has been processed. V1.4 27/07/18 Final version.

Disclaimer

The assessment represents a consolidated view of the EUnetHTA assessment team members and is in no case the official opinion of the participating institutions or individuals.

EUnetHTA Joint Action 3 is supported by a grant from the European Commission. The sole responsibility for the content of this document lies with the authors and neither the European Com- mission nor EUnetHTA are responsible for any use that may be made of the information con- tained therein.

Assessment team

Author(s) Agency for Quality and Accreditation in Health Care and Social Welfare (AAZ), Croatia Co-Author(s) Main Association of Austrian Social Security Institutions (HVB), Austria The Norwegian Institute of Public Health (NIPHNO), Norway Dedicated Agency for Health Quality and Assessment of Catalonia (AQUAS), Spain Reviewer(s) Healthcare Improvement Scotland (HIS), Scotland Regione Emilia-Romagna (RER), Italy Observer(s) Administracao Central do Sistema de Saude (ACSS), Portugal Agency for Health Technology Assessment and Tariff System (AOTMiT), Poland Project Ludwig Boltzmann Institute for Manager Health Technology Assessment (LBI-HTA), Austria

Further contributors

International Diabetes Federation European Region, Brussels; Diabetes Scotland, Scotland and Zagreb Diabetes Association, Croatia were involved to present the range of experiences and views of patient groups and patients with the disease/condition for which the health intervention is being assessed to help HTA reviewers to better understand the impact of the condition, unmet needs and the impact of interventions that are currently available. Details of their involvement are described through this assessment.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 2 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Consultation of the draft Rapid Assessment

External experts [v1.1] Institute of Cardiovascular and Medical Sciences BHF Glasgow, Cardiovascular Research Centre University of Glasgow, Scotland Department of Clinical and Molecular Medicine, Norwegian University of Science and Technology, Trondheim, Norway Manufacturers [v1.1] Abbott Diabetes Care (factual accuracy check) Dexcom, Inc. Medtronic Contributor [v1.1] Health Technology Wales, United Kingdom Medical editor [v1.2] Rogor Editing, Croatia

Conflict of interest

All authors, dedicated reviewers and external experts involved in the production of this assessment have declared they have no conflicts of interest in relation to the technology and comparators assessed according to the EUnetHTA Declaration of interest and confidentiality undertaking of interest (DOICU) statement form.

International Diabetes Federation European Region, Brussels:

International Diabetes Federation Europe received 76.45% funding from industry in 2016. The percentage of the highest contribution from a single organisation is 23.43% from Eli Lilly (see https://www.idf.org/images/2016FundingSourcesSummary.pdf).

Diabetes Scotland, Scotland:

Diabetes Scotland fund raise and operate as part of Diabetes UK. The main sources are supporter, legacy, grants, trust and corporates.

Participants involved in Focus groups have no conflicts of interest in relation to the technology assessed according to the EUnetHTA Declaration of interest and confidentiality undertaking of interest (DOICU) statement form.

How to cite this assessment

Please, cite this assessment as follows:

Agency for Quality and Accreditation in Health Care and Social Welfare (AAZ), Main Association of Austrian Social Security Institutions (HVB), The Norwegian Institute of Public Health (NIPHNO). Continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) as personal, standalone systems in patients with diabetes mellitus treated with insulin. Joint Assessment. Zagreb: EUnetHTA; 2018. Report No.: OTJA08.

The document is available on the website of EUnetHTA.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 3 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

TABLE OF CONTENTS

LIST OF ABBREVIATIONS ...... 7 GLOSSARY ...... 10 SUMMARY ...... 12 SCOPE ...... 12 INTRODUCTION ...... 12 METHODS ...... 15 RESULTS ...... 16 SUMMARY TABLES AND FIGURES ...... 19 DISCUSSION ...... 25 CONCLUSION ...... 26 1 SCOPE ...... 27 2 METHODS AND EVIDENCE INCLUDED ...... 29 2.1 ASSESSMENT TEAM ...... 29 2.2 SOURCE OF ASSESSMENT ELEMENTS ...... 29 2.3 SEARCH ...... 30 2.4 STUDY SELECTION ...... 30 2.5 DATA EXTRACTION AND ANALYSES ...... 31 2.6 QUALITY RATING...... 32 2.7 PATIENT INVOLVEMENT ...... 33 2.8 DESCRIPTION OF THE EVIDENCE USED ...... 33 2.9 DEVIATIONS FROM PROJECT PLAN ...... 39 3 DESCRIPTION AND TECHNICAL CHARACTERISTICS OF TECHNOLOGY (TEC) ...... 40 3.1 RESEARCH QUESTIONS ...... 40 3.2 RESULTS ...... 40 4 HEALTH PROBLEM AND CURRENT USE OF THE TECHNOLOGY (CUR)...... 76 4.1 RESEARCH QUESTIONS ...... 76 4.2 RESULTS ...... 76 5 CLINICAL EFFECTIVENESS (EFF) ...... 93 5.1 RESEARCH QUESTIONS ...... 93 5.2 RESULTS ...... 93 6 SAFETY (SAF)...... 118 6.1 RESEARCH QUESTIONS ...... 118 6.2 RESULTS ...... 118 7 POTENTIAL ETHICAL, ORGANISATIONAL, PATIENT AND SOCIAL, AND LEGAL ASPECTS (ETH, ORG, SOC, LEG) ...... 125 8 PATIENT INVOLVEMENT ...... 126 9 DISCUSSION ...... 135 10 CONCLUSION ...... 141 11 REFERENCES ...... 142

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 4 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

APPENDIX 1: METHODS AND DESCRIPTION OF THE EVIDENCE USED ...... 152 DOCUMENTATION OF THE SEARCH STRATEGIES ...... 152 DESCRIPTION OF THE EVIDENCE USED ...... 162 Guidelines for diagnosis and management ...... 162 Evidence tables of individual studies included for clinical effectiveness and safety ...... 174 rtCGM MDII patients EVIDENCE TABLES ...... 174 rtCGM MDII or CSII mixed patients EVIDENCE TABLES ...... 278 rtCGM CSII patients EVIDENCE TABLES ...... 323 rtCGM vs FGM EVIDENCE TABLES ...... 340 FGM vs SMBG EVIDENCE TABLES ...... 349 nRCT for SAF EVIDENCE TABLES ...... 394 List of ongoing and planned studies ...... 422 Risk of bias and GRADE tables ...... 424 Applicability tables ...... 461 Mean and standard deviation estimation for Forest plots visualization ...... 461 Patient Involvement ...... 467 APPENDIX 2: REGULATORY AND REIMBURSEMENT STATUS ...... 484 APPENDIX 3: CHECKLIST FOR POTENTIAL ETHICAL, ORGANISATIONAL, PATIENT AND SOCIAL AND LEGAL ASPECTS ...... 500

LIST OF TABLES AND FIGURES

Tables

Table 1: Summary of findings on HbA1c changes ...... 19 Table 2: Summary of findings on time spent in normoglyceamia ...... 20 Table 3: Summary of findings on time spent in hypoglycaemia ...... 21 Table 4: Summary of findings on hypoglycaemia and severe hypoglycaemia events...... 23 Table 5: Frequency and severity of local adverse events in RCTs and nRCTs related to device (rtCGM and FGM) or procedures ...... 23 Table 6: Main characteristics of studies included ...... 34 Table 7: Features of the technologies ...... 60 Table 8: Commercially available glucose meters ...... 61 Table 9: Features of the comparators medical devices ...... 70 Table 10: Overview by country ...... 75 Table 11: Classification of DM on the basis of the pathogenic process ...... 77 Table 12: Staging of type 1 diabetes ...... 78 Table 13: Main characteristics of included RCTs ...... 96 Table 14: Changes in HbA1c percentage and number of patients with HbA1c <7% from baseline to the end of the study/follow-up ...... 99 Table 15: Changes in HbA1c percentage from baseline to study end ...... 100 Table 16: Changes in HbA1c from baseline to the end of the study/follow-up ...... 100 Table 17: Time spent in the target glycaemic range 70-180 mg/dL (3.9-10.0 mmol/l) ...... 101 Table 18: Time spent in the target glycaemic range 70-180 mg/dL (3.9-10.0 mmol/l) ...... 101 Table 19: Time (percentage) spent in the target glycaemic ranges ...... 102 Table 20: Time spent in hypoglycaemic ranges ...... 102 Table 21: Time spent in hypoglycaemia ...... 104

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 5 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 22: Percentage time within defined glucose range (hypoglycaemia) ...... 106 Table 23: Time spent in hyperglycaemic range ...... 106 Table 24: Time spent in hyperglycaemia, FGM patients vs SMBG ...... 107 Table 25: Time (percentage) spent in hyperglycaemia ...... 107 Table 26: Results for hypoglycaemia and severe hypoglycaemia (adults unless specified otherwise) ...... 108 Table 27: QoL and user satisfaction data ...... 112 Table 28: QoL and User Satisfaction data FGM vs SMBG ...... 114 Table 29: Patient-reported outcome and quality of life (QoL) measures ...... 114 Table 30: QoL and patient satisfaction for Oskarsson et al. 2018, MDII subgroup of Impact trial* (n=78 vs n=70) ...... 115 Table 31: Gold Score, Hypoglycaemia fear (HFS-II) and Diabetes-related emotional distress (PAID questionnaire) ...... 116 Table 32: Frequency and severity of local adverse events in RCTs and nRCTs related to devices (rtCGM and FGM) or procedures ...... 119 Table A1: Overview of guidelines ...... 162 Table A2: List of ongoing studies with rtCGM or FGM medical devices ...... 422 Table A3: Risk of bias – study level (RCTs) ...... 425 Table A4: Risk of bias – outcome level (randomised controlled trials) rtCGM vs SMBG ...... 427 Table A5: Risk of bias – outcome level (randomised controlled trials) FGM vs rtCGM and FGM vs SMBG ...... 430 Table A6: rtCGM compared with SMBG for different outcomes ...... 432 Table A7: rtCGM compared with SMBG for different outcomes (continued) ...... 449 Table A8: rtCGM compared with FGM ...... 453 Table A9: FGM compared with SMBG ...... 455 Table A10: Summary table characterising the applicability of a body of studies ...... 461 Table A11: Regulatory status ...... 484 Table A12: Summary of (reimbursement) recommendations in European countries for the technology ...... 496

Figures

Figure 1: Forest plot of meta-analysis comparing the changes in HbA1c from baseline to the end of the study: rtCGM vs SMBG, patients with T1DM on MDII ...... 16 Figure 2: Summary of findings on Time spent in hyperglycaemia ...... 21 Figure 3: Flow chart ...... 31 Figure 4: G4® PLATINUM Sensor, Transmitter and receiver ...... 47 Figure 5: Dexcom G5® Mobile CGM System Sensor, Transmitter and receiver ...... 49 Figure 6: Dexcom G6® CGM System Sensor, Transmitter and display device (receiver or smart device) ...... 52 Figure 7: Freestyle Navigator II® Transmitter and Receiver ...... 53 Figure 8: EnliteTM Sensor, GuardianTM Connect Transmitter and GuardianTM Connect App ...... 54 Figure 9: Eversense® Sensor, Transmitter and Smartphone App ...... 55 Figure 10: Eversense® XL Sensor, Transmitter and a smartphone App ...... 56 Figure 11: Freestyle Libre® ...... 58

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 6 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Figure 12: World Health Organization – Diabetes country profiles, 2016. Graph developed by authors of this assessment...... 90 Figure 13: Risk of bias of included studies at study level ...... 94 Figure 14: Forest plot of meta-analysis comparing the changes in HbA1c from baseline to the end of the study ...... 99 Figure A1: Risk of Bias at study level, with summary, for all RCTs included in this assessment ...... 424

LIST OF ABBREVIATIONS

2-h PG Two hour plasma glucose ADA American Diabetes Association AE Adverse event ARD Absolute relative difference AUC Area under the curve BG Blood glucose BGRI Blood glucose risk index BMI Body mass index CAD Coronary artery disease CE Conformite European mark CEG Consensus Error Grid CG-EGA Continuous Glucose-Error Grid Analysis CGM Continuous glucose monitoring CGM-SAT CGM Satisfaction Scale CI Confidence intervals CIDS Confidence in Diabetes Self-Care CONGA Continuous overall net glycaemic action CSII Continuous subcutaneous insulin infusion CT Computed Tomography CUR Current use of technology domain CV Coefficient of variation CVD Cardiovascular disease DCCT Diabetes Control and Complications Trial DDS Diabetes Distress Scale DKA Diabetic ketoacidosis DM Diabetes mellitus DM1 Diabetes mellitus Type 1 DM2 Diabetes mellitus Type 2 DOICU Declaration of conflict of interest and confidentiality undertaking

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 7 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

DQoL Diabetes Quality of Life [questionnaire] DR Diabetic retinopathy DTSQ Diabetes Treatment Satisfaction Questionnaire DVLA Driver and Vehicle Licensing Agency EFF Clinical effectiveness domain EGA Error Grid Analysis EQ-5D European Quality of Life 5 Dimensions questionnaire ESRD End-stage renal disease EU European Union EUR Europe Region FGM Flash glucose monitoring FPG Fasting plasma glucose GDM Gestational diabetes mellitus GDP Gross domestic product GFR Glomerular filtration rate GLP-1 Glucagon-like peptide-1 GMDN Global medical device nomenclature GRADE Grading of Recommendations, Assessment, Development and Evaluation HbA1c Hemoglobin A1c HBGI High blood glucose index HCP Healthcare professional HCS Hypoglycemic confidence scale HFS Hypoglycaemia fear survey HHKN Hyperosmotic hyperglycemic nonketotic state HHS Hyperosmolar hyperglycemic state HLA Human Leukocyte Antigen HTA Health technology assessment HTAi Health technology assessment international HUS Hypoglycaemia Unawareness Score ICD International Classification of Diseases iCGM Intermittently viewed CGM IDF International Diabetes Federation IFG Impaired fasting glucose IGT Impaired glucose tolerance IHA Impaired hypoglycaemia awareness IQR Interquartile range ISIG Interstitial signal ITT Intention to treat IWRS Interactive web response system

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 8 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

LBGI Low blood glucose index LOCF Last observation carried forward LSM Least square mean MA Meta-analysis MAD Mean absolute difference (deviation) MAGE Mean amplitude of glycemic excursions MARD Mean Absolute Relative Difference MDI Multiple daily injections MDII Multiple daily insulin injections MeSH Medical Subject Headings MI Myocardial infarction MODD Mean of daily differences MODY Maturity-onset diabetes of the young MRD Mean relative difference MRI Magnetic Resonance Imaging NFC Near-field communication NGSP National Glycohemoglobin Standardization Program nRCT Non-randomized controlled trial OGTT Oral glucose tolerance test PAID Problem Areas in Diabetes Questionnaire PARD Paired or precision absolute relative difference PDM Personal Diabetes Manager PG Plasma glucose PHI Private Health Information PICO Population-Intervention-Control-Outcome (scheme) PII Personally Identifiable Information PLWD People living with diabetes PP Per protocol QoL Quality of life RCT Randomised Controlled Trial REA Relative Effectiveness Assessment RF Radio Frequency RFID Radio frequency identification RoB Risk of Bias RPG Random plasma glucose RR Relative risk rt-CGM Real-time continuous glucose monitoring SAE Serious adverse event

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 9 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

SAF Safety domain SAP Sensor-augmented (or enabled) insulin pump SAP Session Announcement Protocol SD Standard deviation SE Standard error SG Sensor glucose SGLT-2 Sodium-glucose co-transporter-2 SMBG Self-monitoring of blood glucose SR Systematic review SWE-HFS Swedish version of the Hypoglycaemia Fear Survey SWE-PAID Swedish version of the Problem Areas in Diabetes Questionnaire TDD Total daily dose TEC Technical characteristics of technology domain TIR Time in range US $ American dollar WHO World Health Organization YSI Yellow Springs Instruments

GLOSSARY [1-9]

CGM system: continuous glucose monitoring system Error Grid (Clarke, Parkes) plots: related to device accuracy CSII: continuous subcutaneous insulin infusion CSII + CGM: non-enabled continuous subcutaneous insulin infusion and stand-alone continuous glucose monitoring CSII + SMBG: non-enabled continuous subcutaneous insulin infusion with self-monitoring of blood glucose by capillary blood testing FGM system: flash glucose monitoring system (called also iCGM: intermittently viewed continuous glucose monitoring) – provides the current glucose value plus retrospective glucose data for a specified time period upon “scanning” HbA1c: glycated haemoglobin IAH: Impaired awareness of hypoglycaemia: when people with diabetes, usually type 1 diabetes mellitus, are frequently unable to notice when they have low blood sugar MARD scores: Mean Absolute Relative Difference related to device accuracy MDII: Multiple daily insulin injections MDII + CGM: multiple daily insulin injections with continuous monitoring of blood glucose MDII + SMBG: multiple daily insulin injections with self-monitoring of blood glucose by capillary blood testing Reference standard: the best currently available diagnostic test, against which the index test is compared rtCGM: real-time CGM: provides real-time numerical and graphical information about the current glucose level, glucose trends, the direction/rate of change of glucose and alarms and alerts at present treshold

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 10 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

SMBG: Self-monitoring blood glucose SAP: sensor-augmented (or enabled) insulin pump or CGM-enabled insulin pumps: Sensor enabled (or augmented) insulin pump systems compatible (connected) with specific CGM system – sensor data display on pump; Sensor-integrated pump: sensor data display on pump and pump acts on sensor data (suspends insulin at pre-set sensor threshold or suspend insulin in advance of pre-set sensor threshold) T1DM: type 1 diabetes mellitus T2DM: type 2 diabetes mellitus Time spent in the target glucose range: Time in range

Classifications of hypoglycaemia, based on clinical evaluation [1], [2]: Level 1: a hypoglycaemia alert glucose value of 70–54 mg/dL (3.9–3.0 mmol/L) with or without symptoms. This should be considered an alert that the individual may be at risk for developing hypoglycaemia and should work to minimize the time spent in this range to reduce the risk of developing more clinically significant hypoglycaemia. Level 2: a glucose level of 54 mg/dL (3.0 mmol/L) with our without symptoms. This should be considered clinically significant hypoglycaemia requiring immediate attention. Level 3: severe hypoglycaemia: This denotes cognitive impairment requiring external assistance for recovery but is not defined by a specific glucose value. Hypoglycaemia should be quantified in the following ways [1], [2]: As the percentage of CGM values that are below a given threshold (70 mg/dL [3.9 mmol/L] or 54 mg/dL [3.0 mmol/L]) or the number of minutes or hours below these thresholds and as the number of hypoglycaemic events that occur over the given CGM reporting period. A hypoglycaemic event [1], [2]: Beginning of a CGM event readings below the threshold for at least 15 min is considered an event. For example at least 15min, 54 mg/dL (3.0 mmol/L) to define a clinically significant (level 2) hypoglycaemic event. End of a CGM event: readings for 15 min at 70 mg/dL (3.9 mmol/L). A second hypoglycaemic event outcome of prolonged hypoglycaemia is considered when CGM levels are 54 mg/dL (3.0 mmol/L) for consecutive 120 min or more. Time in range (TIR) [1], [2] generally refers to the time spent in an individual’s target glucose range (usually 70–180 mg/dL [3.9–10 mmol/L] but occasionally 70–140mg/dL [3.9–7.8 mmol/L]). TIR measurements add valuable information to assess the level of current glycaemic control in addition to what is known from the HbA1c. It is also necessary to quantitate the times below and above target range, using a few severity thresholds for each level: Percentage of time in level 2 hypoglycaemic range (54 mg/dL [3.0 mmol/L]). Percentage of time in level 1 hypoglycaemic range (70–54 mg/dL [3.9–3.0 mmol/L]). Percentage of time in target range: 70–180 mg/dL (3.9–10.0 mmol/L) (default); 70–140 mg/dL (3.9–7.8 mmol/L) (secondary); Percentage of time in level 1 hyperglycaemic range (180 mg/dL [10.0 mmol/L]). Percentage of time in level 2 hyperglycaemic range (250 mg/dL [13.9 mmol/L]). Episodes of hypoglycaemia and hyperglycaemia

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 11 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

SUMMARY OF THE RELATIVE EFFECTIVENESS OF CONTINUOUS GLUCOSE MONITORING (CGM REAL-TIME) AND FLASH GLUCOSE MONITORING (FGM) AS PERSONAL, STANDALONE SYSTEMS IN PATIENTS WITH DIABETES MELLITUS TREATED WITH INSULIN

Scope

The objective of this rapid assessment was to evaluate relative effectiveness and safety of real- time continuous glucose monitoring (rtCGM) and flash glucose monitoring (FGM) as personal, standalone systems in patients with diabetes mellitus treated with insulin. Specifically, we addressed the research question whether the use of rtCGM and FGM medical devices as personal, standalone systems, being adjunctive (cannot be used to make treatment decision without confirmatory finger-stick testing) or non-adjunctive (can be used to make treatment decision without confirmatory finger-stick testing) in patients with diabetes mellitus type 1 and type 2 (T1DM and T2DM) in adults and children, gestational DM patients treated with insulin, either through insulin pump therapy (CSII) or multiple daily insulin injections (MDII), is more effective and/or safer than using self-monitoring blood glucose (SMBG) medical devices. The relative effectiveness and safety between rtCGM and FGM was to be assessed as well (head-to-head). Potential ethical, organisational, patient and social and legal aspects were addressed if relevant.

The scope can be found here: Scope.

Introduction

The purpose of this project was first to collectively produce a rapid core HTA on real-time continuous glucose monitoring (rtCGM) and flash glucose monitoring (FGM) systems using the EUnetHTA HTA Core Model® for Relative Effectiveness Assessment (REA) [10, 11] and next to initiate local HTA productions based on this assessment.

Different definitions and/or category names are currently being used for rtCGM and FGM systems in the available published literature. FGM, for instance, may be described as a separate entity from CGM between a traditional blood glucose meter and a continuous glucose monitoring (CGM) system, or as special case of CGM or subset of CGM or intermittently viewed CGM (iCGM) or “flash” CGM/[1, 2, 4, 6, 9]. Authors of this assessment have therefore decided to use definitions according their specific “Instruction for Use” documents – “Indication for use” sections, and thus are referring to them as flash glucose monitoring (FGM) system and real-time continuous glucose monitoring (rtCGM) systems.

The topic of this assessment was chosen based on the initial request from the Croatian national payer organisation who commissioned AAZ to carry out an HTA on rtCGM and FGM devices in patients with diabetes. The topic is highly relevant as many of these glucose monitoring devices have become available on the market. In light of potential clinical and quality of life benefits, the number of brands on the market, the high costs of purchase and further use, and differences among countries in implementation status and use, many EUnetHTA partners have communicated their interest in this assessment. In addition, stakeholders obviously may benefit from this report.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 12 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Description of technology and comparators

The health technologies assessed include rtCGM systems and FGM systems as personal, standalone medical devices for patients with diabetes mellitus type 1 and type 2 (T1DM and T2DM), including adults, children, and patients with gestational DM treated with insulin, either through insulin pump therapy (CSII) or multiple daily insulin injections (MDII). Professional (retrospective) devices are not included in this assessment (B0001).

Standalone medical devices for glucose monitoring can be adjunctive, what means that they cannot be used to make treatment decision without confirmatory finger-stick testing, or non-adjunctive, which means they can be used to make treatment decision without confirmatory finger-stick testing and may or may not demand calibration by finger sticks measurement [12] (B0001).

The following rtCGM systems are within the scope of this assessment: G4® PLATINUM Continuous Glucose Monitoring System, Dexcom; G5® Mobile Continuous Glucose Monitoring System, Dex- com; G6® Continuous Glucose Monitoring System, Dexcom GuardianTM Connect Continuous Glu- cose Monitoring system, Medtronic; Eversense® Continuous Glucose Monitoring system, Senseon- ics Incorporated; Eversense® XL Continuous Glucose Monitoring (CGM) System, Senseonics In- corporated, and FreeStyle Navigator II® Continuous Glucose Monitoring System, Abbott (B0001).

The only FGM system within the scope of this assessment is the FreeStyle Libre® Flash Glucose Monitoring System, Abbott B0001.

RtCGM and FGM devices allow continuous evaluation of glycaemic control, providing trends and fluctuations of interstitial glucose levels over time, so they may affect adherence to glucose self- monitoring and patient quality of life and enable proactive therapeutic interventions to maintain glycaemic control. Their potential benefit would seem particularly relevant in children, patients with poorly controlled diabetes, pregnant women, and patients with hypoglycaemia unawareness. RtCGM systems can be combined with an insulin pump, integrated in insulin pump, or used as standalone glucose monitoring device. RtCGM data can be viewed from an external receiver or smartphone device (Apple or Android) (B0002).

A number of different systems for rtCGM have been available on the market for approximately 10 years. The number of medical devices for glucose monitoring available is growing rapidly, as the technologies are rapidly developing producing new improved generations of existing devices, which provide more options for the user [4, 140, 141].

Comparators were self-monitoring blood glucose (SMBG) medical devices as the reference standard, but head-to-head comparisons are also presented. SMBG refers to home blood glucose testing for people with diabetes. SMBG devices are blood glucose monitors, measuring blood glucose concentration using a drop of venous, arterial or, mainly, capillary blood from a finger puncture. SMBG is recognized as one approach to improve glycaemic control and reduce hypoglycaemic events, allowing patients with diabetes to modify their hypoglycaemic drug dose, diet, and physical activities. Various blood glucose monitors are available on the market [14] (B0001).

Health problem

Diabetes mellitus (DM) refers to a group of common metabolic disorders that share the phenotype of hyperglycaemia. Diabetes mellitus can be classified into the following general categories: 1. Type 1 diabetes (due to autoimmune b-cell destruction, usually leading to absolute insulin deficiency)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 13 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

2. Type 2 diabetes (due to a progressive loss of b-cell insulin secretion frequently on the background of insulin resistance) 3. Gestational diabetes mellitus (GDM) (diabetes diagnosed in the second or third trimester of pregnancy that was not clearly overt diabetes prior to gestation) 4. Specific types of diabetes due to other causes, e.g., monogenic diabetes syndromes (such as neonatal diabetes and maturity-onset diabetes of the young [MODY]), diseases of the exocrine pancreas (such as cystic fibrosis and pancreatitis), and drug- or chemical-induced diabetes (such as with glucocorticoid use, in the treatment of HIV/AIDS, or after organ transplantation).

Type 1 diabetes and type 2 diabetes are heterogeneous diseases in which clinical presentation and disease progression may vary considerably [15] (A0002).

Globally, registries of the WHO DIAMOND Project recorded large differences in the incidence and prevalence of type 1 diabetes, with incidence from over 60 to under 0.5 cases annually per 100 000 children aged under 15 years; differences in case ascertainment may have contributed to the variability. In high-income countries, the prevalence of type 2 diabetes is frequently highest among poor people. There are few data on the income gradient of diabetes in low- and middle-income countries, but data that do exist suggest that, although the prevalence of diabetes is often highest among wealthy people, this trend is reversing in some middle-income countries. The proportion of undiagnosed type 2 diabetes varies widely – a recent review of data from seven countries found that between 24% and 62% of people with diabetes were undiagnosed and untreated. Prevalence of gestational diabetes: The frequency of previously undiagnosed diabetes in pregnancy and gestational diabetes varies among populations but probably affects 10–25% of pregnancies. It has been estimated that most (75–90%) of cases of high blood glucose during pregnancy are gestational diabetes [16] (A0023).

People diagnosed with T1DM, T2DM, or gestational diabetes, who are willing and able to monitor and manage their DM themselves are target population of this assessment (A0007).

Hypoglycaemia is the major limiting factor in the glycaemic management of type 1 and type 2 diabetes [89]. Symptoms of hypoglycaemia include, but are not limited to, shakiness, irritability, confusion, tachycardia, and hunger. Hypoglycaemia may be inconvenient or frightening to patients with diabetes. Severe hypoglycaemia may be recognized or unrecognized and can progress to loss of consciousness, seizure, coma, or death. Clinically significant hypoglycaemia can cause acute harm to the person with diabetes or others, especially if it causes falls, motor vehicle accidents, or other injury. Young children with type1 diabetes and the elderly, including those with type 1 and type 2 diabetes, are noted as particularly vulnerable to clinically significant hypoglycaemia because of their reduced ability to recognize hypoglycaemic symptoms and effectively communicate their needs [89]. Hypoglycaemia mortality estimates range from 4 to 10 percent of deaths of patients with type 1 diabetes. Hypoglycaemia mortality rates in type 2 diabetes are currently unknown, but fatal hypoglycaemia has been documented in type 2 diabetes. Severe hypoglycaemia may also be associated with an increased risk of cardiovascular disease in patients with type 2 diabetes. Recurrent severe hypoglycaemia has been associated with cognitive impairment in young children or older persons with diabetes. Nocturnal hypoglycaemia is a particular problem, which can lead to disruption of sleep and delays in correction of the hypoglycaemia [17]. Hypoglycaemia unawareness could severely influence diabetes control and quality of life [89] (A0005).

In 2018 American Diabetes Association provided recommendations related to assessment of glycaemic control [89] (A0025).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 14 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Methods

For Description and technical characteristics of technology (TEC) and Health problem and current use of technology (CUR) domains no quality assessment tool was used, but multiple sources were used in order to validate individual, possibly biased, sources. We performed descriptive analyses of information from the various sources explored. The completed part of EUnetHTA submission file from the manufacturer was used as starting point. The Medical Devices Evidence Submission template was sent to all relevant manufacturers of the technology that agreed to participate. Man- ufacturers were asked to submit non-confidential evidence, focusing on the technical characteristics and current use of the technology. The documentation provided was used in addition to the literature identified by the literature search. We performed descriptive analyses of information from the various sources explored.

An update of existing systematic reviews (SRs) was not possible, and a new SR of RCTs with a meta-analysis on one clinical outcome – HbA1C change – was performed. For the Summary of findings on relative effectiveness related to outcomes where MA was not possible, we presented the Forest plots for visualisation, i.e. without pooling estimates. To achieve this, we converted median values into mean values. The method used to estimate means and standard deviations for outcomes that were originally reported using nonparametric summary statistics such as medians (IQR) is described by Wan et al 2014 [18].

Risk of bias of included RCTs was evaluated independently by two researchers. Study and outcomes validity and level of evidence were assessed according to the EUnetHTA guidelines [19]. The Cochrane Risk of Bias Tool both on study and outcome levels, and GRADE (Grading of Rec- ommendations, Assessment, Development, and Evaluation) to assess the certainty of the body of evidence were used. Outcomes were changed in HbA1c from baseline to the end of the study, time spent in normoglycaemia; time spent in hypoglycaemia; hypoglycaemia events, serious hypoglycaemia, and QoL and patient satisfaction) [20-22]. Relevant subgroup analyses including meta-analyses were planned where possible, as well as indirect comparisons (through network meta-analysis), but not performed.

Two different methods were used for involving patients, one involving individual patients and the second involving patient organisations. A focus group (with individual patients) was held in Croa- tia, one for adults and one for children with informal caregivers (i.e. two separate focus groups were held). Four to five hours’ meeting was planned and four flipovers were used with predefined questions related to impact of condition, experience with currently available medical devices, experiences with and expectations of the medical devices being assessed, and additional information which patients believed would be helpful to the HTA researchers. The entire focus group discussion was recorded, and transcriptions were made in Croatian language. No ethical approval was necessary for the focus group in Croatia but patients and informal caregivers (parents) needed to sign an informed consent form. Two patient groups were contacted, one at the European level and one at the national level. The Patient Group Submission template was sent to the Inter- national Diabetes Federation European Region, Brussels and Diabetes Scotland, Scotland. The Patient Group Submission template was prepared for this assessment by modifying the HTAi Pa- tient Group Submission template for HTA of health interventions (not medicines) [23] and with

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 15 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin inclusion of topics and issues from EUnetHTA Core Model® 3.0 related to Patients and Social Domain aspect [10]. Payer representatives at EU level (The International Association of Mutual Benefit Societies (AIM), Brussels) were planned to be contacted related to reimbursement status of technologies under assessment.

Results

Available evidence

Twelve RCTs were included which reported on the use of rtCGM (as standalone devices or with insulin pumps) and FGM devices compared with SMBG [13, 24-38]. Only one head-to-head trial was identified comparing rtCGM and FGM [34]. Three nRCTs [39-41] were included in the safety domain in addition to those RCTs included in the clinical effectiveness domain. Due to heterogeneity between populations, interventions and outcomes measures, MA was done only for one outcome, i.e. HbA1c change from baseline to the end of the study, pooling the data from two RCTs (DIAMOND and GOLD trials) [25, 29] Figure 1. Our systematic review provides a narrative summary. All RCTs had RoBs, the most frequent due to lack of blinding (Table A3). The certainty of evidence was low to very low for the outcomes time spent in normo- and hypoglycaemia, hypoglycaemia, and severe hypoglycaemia events, QoL and user satisfaction, and moderate for the outcome HbA1c change from baseline to the end of the study [13, 24-38].

Clinical effectiveness

HbA1c changes from baseline to the end of the study

Meta-analysis of the data pooled from two RCTs (with T1DM patients on MDII treatment) comparing rtCGM vs SMBG (Beck 2017 T1DM and Lind 2017 T1DM) [25, 29] demonstrated a statistically significant benefit of rtCGM in reducing HbA1c levels (heterogeneity among studies was low).

Figure 1: Forest plot of meta-analysis comparing the changes in HbA1c from baseline to the end of the study: rtCGM vs SMBG, patients with T1DM on MDII (Beck 2017 T1DM and Lind 2017 T1DM)

RtCGM led to a statistically significant reduction in HbA1c levels compared with SMBG in the majority of studies on MDII patients (Beck 2017; Ruedy 2017; Lind 2017; Beck 2017 T2DM) [25], [37], [29], [26] and the two studies on MDII and CSII patients gathered (Battelino 2011, Riveline 2012) [24, 36]. In one study on MDII patients (Heinemann 2018) [13], two studies with MDII and CSII patients (Mauras 2012; van Beers 2016) [31], [38] and in one study on CSII patients (Ly 2013) [30] no statistically significant difference was found. In addition, no statistically significant difference was identified in the only and small head-to-head study comparing rtCGM vs FGM (Reddy 2018) [34] and two RCTs comparing FGM vs SMBG in T1DM patients (Bolinder 2016) [27] and in T2DM patients (Haak 2017) [28] including patients on MDII and CSII treatments. Four studies with impaired hypoglycaemia awareness patients (Heinemann 2018; Van Beers 2016; Ly 2013) [13], [38],

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 16 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[30] comparing rtCGM vs SMBG and comparing rtCGM vs FGM (Reddy 2018) [34] showed no statistically significant difference between intervention and control groups. Notably, in the latter studies, HbA1c was not the primary outcome and participants were selected on the basis of IHA or previous severe hypoglycaemia events. The certainty of the evidence varied from moderate to very low.

Time spent in the target glycaemic range

Results from three RCTs comparing rtCGM vs SMBG in T1DM patients (Beck 2017 T1 patients; van Beers 2016 and Battelino 2011) [25], [38], [24] favoured rtCGM (p<0.05)). In one small head-to- head trial comparing rtCGM vs FGM (Reddy et al, 2017) [34] no statistically significant difference was observed. No difference was identified when comparing FGM vs SMBG in T2DM patients (Haak 2017) [41], while in T1DM patients (Bolinder et al 2016) there was a statistically significant increase in the intervention group compared with the control group at 6 months [27]. The certainty of the evidence varied from low to very low.

Time spent in hyperglycaemia

When comparing rtCGM and SMBG in terms of time spent in the hyperglycaemic range, the time was statistically significant reduced with rtCGM in patients with T1DM in one study (Beck 2017) [25], but was not in another study (Heinemann 2018) [13], while two studies did not report on statistical significance (Lind 2017 and Beck 2017) [29, 26]. No statistically significant differences were found in the small rtCGM vs FGM trial (Reddy et al, 2018) [34]. When comparing FGM vs SMBG in T2DM patients (Haak et al 2017) there was no difference; a statistically significant difference was observed in patients with T1DM (Bolinder et al 2016) [28, 27]. The certainty of the evidence varied from low to very low.

Time spent in hypoglycaemia

When comparing rtCGM vs SMBG, statistically significant results which favoured continuous glucose monitoring over control were identified in studies with MDII patients (Beck 2017 T1 patients; Heinemann 2018) [25, 13], MDII and CSII patients combined (Battelino 2011, van Beers 2016) [24, 38], and CSII patients (Ly 2013) [30]. The difference was not statistically significant in one study (Mauras 2012 [31], and not another (Lind 2017) [29]. One head-to-head trial comparing rtCGM vs FGM (Reddy et al, 2018) [34], along with two RCTs comparing rtCGM and SMBG in T1DM (Bolinder et al 2016) and in T2DM patients (Haak 2017) reported statistically significant decrease in time in hypoglycaemia in the intervention group [27, 28]. The certainty of the evidence varied from low to very low.

Hypoglycaemia and severe hypoglycaemia events

There was a statistically significant difference in hypoglycaemic events when rtCGM and SMBG were compared in four studies including all insulin regimes (Riddlesworth 2017; Heinemann 2018; Ly 2013; van Beers 2016) [35, 13, 30, 38], while three other studies did not show any differences (Battelino 2011, Mauras 2012; Riveline 2012) [24, 31, 36]. The certainty of the evidence for varied from low to very low.

The single and small head-to-head trial comparing rtCGM vs FGM (Reddy et al, 2018) [34], as well as two RCTs with respectively T1DM and T2DM patients comparing FGM vs SMBG (Bolinder 2016; Haak 2017) [27, 28], reported statistically significant decrease of hypoglycaemia events in the intervention group. The certainty of the evidence varied from low to very low.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 17 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

All four studies involving T1DM patients with IHA or previous severe hypoglycaemia events, i.e. three comparing rtCGM vs SMBG (Heinemann 2018; Ly 2013; van Beers 2016 [13, 30, 38]) and the head-to-head study comparing rtCGM vs FGM (Reddy et al, 2018) [34] showed statistically significant results in favour of rtCGM in terms of time spent in hypoglycaemia, as well as for clinical outcomes – hypoglycaemia and severe hypoglycaemia events. The certainty of the evidence varied from low to very low.

Patient-reported outcome, quality of life (QoL) measures and user satisfaction

Results were inconsistent across studies, probably due to differences in types of outcomes or survey tools. The certainty of the evidence for these outcomes varied from low to very low.

Safety

Systemic AEs were reported in all but one study. Systemic SAEs were reported in the majority of studies, and those related to severe hypoglycaemia and ketoacidosis as well, but investigators found a majority of them were not related to the intervention (C0008).

Devices (or procedures) related local AEs were reported in 5 RCTs (3 related to FGM device) [27- 30, 32] and 3 nRCTs (all related to the use of FGM) [39-41]. According to Haak et al 2017 [41], anticipated symptoms referred to those typically expected using a sensor device and were equal to symptoms normally experienced with blood glucose finger-stick testing, e.g., pain, bleeding, bruising, and were resolved without medical intervention. Allergic reactions were also reported. The small head-to-head study comparing rtCGM vs FGM did not consider AEs at all [34] (C0008).

Ethical, organisational, patient and social and legal aspects

Some specific aspects concerning ethical, organisational, patient and social, and legal aspects were identified from the Rapid REA Checklist [11]. Where deemed relevant, these issues should be taken into consideration at national, regional, and/or local levels.

Patient involvement

Adults, children and parents of children with T1DM reported very positive experiences with rtCGM and FGM devices as they could see the glucose trends and react properly. Benefits were major in terms of emotional and social impact, changes to sleeping patterns, better quality of life, independence, and better control and normal life. Importance of education and motivation were stressed and the most important barriers, but concerns were also related to high cost and unavailability of these medical devices in some countries.

Upcoming evidence

Several ongoing RTCs and nRCTs include the use of FGM and rtCGM (Dexcom G4® and G5®) devices in adults and the paediatric population.

Reimbursement

Reimbursement status of rtCGM and FGM devices for adults or children only with T1DM and T2DM varied in EU countries at national or regional levels.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 18 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Summary tables and figures: HbA1c changes, time spent in normoglyceamia, hypoglycaemia and hyperglycaemia, events of hypoglycaemia and severe hypoglycaemia

Table 1: Summary of findings on HbA1c changes

HbA1c changes Studies Certainty of evidence according to GRADE rtCGM vs SMBG Statistically significant Studies on MDII patients (Beck 2017; Ruedy 2017; From moderate to reduction in HbA1c Lind 2017; Beck 2017 T2DM) and the two studies on very low levels than usual care MDII and CSII patients (Battelino 2011, Riveline 2012) No statistically MDII patients (Heinemann 2018), two studies with MDII significant difference and CSII patients (Mauras 2012; van Beers 2016) and in one study on CSII patients (Ly 2013) FGM vs SMBG No statistically IMPACT (on T1 DM) (Bolinder 2016) and REPLACE From low to very low significant difference (on T2 DM patients) (Haak 2017), patients on MDII and CSII treatment rtCGM vs FGM No statistically Reddy 2018 Very low significant difference

Change in HbA1c % rtCGM vs SMBG

Change in HbA1c % FGM vs SMBG

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 19 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 2: Summary of findings on time spent in normoglyceamia

Time spent in the target glycemic Studies Certainty of evidence (normoglycaemic) range according to GRADE rtCGM vs SMBG Statistically significant in favour Beck 2017 T1 patients; van Beers 2016 From low to very low of rtCGM over usual care and Battelino 2011 No statistically significant MDII patients (Heinemann 2018), study difference with MDII and CSII patients (Mauras 2012) FGM vs SMBG Statistically significant in favour IMPACT (on T1 DM) (Bolinder 2016), From low to very low of FGM over usual care patients on MDII and CSII treatment No statistically significant REPLACE (on T2 DM patients) (Haak difference 2017), patients on MDII and CSII treatment rtCGM vs FGM No statistically significant difference Reddy 2018 Very low

Time in normoglycaemia 70-180 mg/dl (or 3.9-10.0 mmol/l) rtCGM vs SMBG

*For Mauras 2012 and Ly 2013 mean difference: not estimable Time in normoglycaemia FGM vs SMBG

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 20 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time in hyperglycaemia >180 mg/dl rtCGM vs SMBG

Time in hyperglycaemia >250 mg/dl (or 13.9 mmo/l) rtCGM vs SMBG

Time in hyperglycaemia >240 mg/dl (or 13.3 mmol/l) rtFGM vs SMBG

Figure 2: Summary of findings on Time spent in hyperglycaemia

Table 3: Summary of findings on time spent in hypoglycaemia (part 1)

Results for for time spent Studies Certainty of evidence in hypoglycemic range according to GRADE rtCGM vs SMBG Statistically significant MDII patients (Beck 2017 T1 patients; From low to very low in favour of rtCGM over Heinemann 2018) [25, 13], both MDII and CSII usual care patients (Battelino 2011, van Beers 2016) [24, 38] and CSII patients (Ly 2013) [30] No statistically significant Study with MDII and CSII patients difference (Mauras 2012) [31] FGM vs SMBG Statistically significant in IMPACT (on T1 DM) (Bolinder 2016) [27], From low to very low favour of FGM over usual care patients on MDII and CSII treatment rtCGM vs FGM Statistically significant in Reddy 2018 [34] Very low favour of rtCGM over FGM

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 21 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 3: Summary of findings on time spent in hypoglycaemia (part 2)

Time in hypoglycaemia <70 mg/dl rtCGM vs SMBG

*For Mauras 2012 mean difference: not estimable Time in hypoglycaemia <50-60 mg/dl (or 2.8-3.3 mmol/l) rtCGM vs SMBG

Time in hypo <70 mg/dl FGM vs SMBG

Time in hypo <55 mg/dl FGM vs SMBG

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 22 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 4: Summary of findings on hypoglycaemia and severe hypoglycaemia events

Results for Number of Studies Certainty of evidence hypoglycaemia events and according to GRADE severe hypoglycaemia events rtCGM vs SMBG Statistically significant in favour Riddlesworth 2017 and Heinemann 2018 From low to very low of rtCGM over usual care studies in MDII patients, Ly 2013 on CSII patients and van Beers 2016 study related to MDII and CSII patients No statistically significant Battelino 2011, Mauras 2012 and difference Riveline 2012 FGM vs SMBG Statistically significant in favour IMPACT (on T1 DM) (Bolinder 2016), From low to very low of FGM over usual care patients on MDII and CSII treatment rtCGM vs FGM Statistically significant in favour Reddy 2018 Very low of rtCGM over FGM IHA or previous severe hypoglycaemia patients: outcome hypoglycaemia events or severe hypoglycaemia

Table 5: Frequency and severity of local adverse events in RCTs and nRCTs related to device (rtCGM and FGM) or procedures

Studies Device (or procedure) related local AEs RCTs rtCGM MDII patients None reported Beck 2017 DIAMOND [25]; Ruedy 2017 [37]; Beck 2017 T2DM [26]; Heinemann 2018 [13] Lind 2017 GOLD [29] One patient in the CGM group discontinued use because of an allergic reaction to the sensor rtCGM CSII patients Ly 2013 [30] Device failure occurred on 2 occasions and was corrected with replacement of the sensor transmitter (Minilink™) rtCGM MDII and CSII patients None reported Battelino 2011[24]; Mauras 2012 [31]; Riveline 2012 [36]; Van Beers 2016 [38] FGM vs CGM None reported Reddy 2018 [34]

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 23 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Studies Device (or procedure) related local AEs FGM vs SMBG Bolinder 2016 [27] 13 adverse events were reported by ten participants related to the sensor – four allergy events (one severe, three moderate); one itching (mild); one rash (mild); four insertion-site symptom (severe); two erythema (one severe, one mild); and one oedema (moderate); all were resolved. There were 248 sensor insertion-site signs and symptoms experienced by 65 participants across both groups. Signs are subdivided into those expected due to sensor insertion: pain (38), bleeding (25), oedema (eight), induration (five), and bruising (five), and those associated with sensor wear: erythema (85), itching (51), and rash (31). Seven participants withdrew from the study due to device-related adverse events or repetitive occurrences of sensor insertion-related symptoms. Haak 2017 [28] Six intervention participants (4%) reported nine adverse events for sensor-wear reactions (two severe, six moderate, one mild). These were sensor-adhesive reactions, primarily treated with topical preparations. All were resolved at study exit. Anticipated symptoms refer to those typically expected using a sensor device and equate to symptoms normally experienced with blood glucose finger-stick testing, e.g., pain, bleeding, bruising. There were 158 anticipated sensor insertion site symptoms observed for 41 (27.5%) intervention and 9 (12.0%) control participants. These symptoms were primarily (63%) due to the sensor adhesive (erythema, itching, and rash) and resolved without medical intervention. Oskarsson 2018 [32] Eight adverse events for six (7%) intervention participants were related to wearing the study device. Four participants withdrew because of these adverse events. There were 144 sensor insertion-site symptoms experienced by 34 participants. The numbers of participants affected by expected signs or symptoms due to sensor insertion were: pain, n = 14; bleeding, n = 9; oedema, n = 3; and induration, n = 3. The symptoms associated with sensor wear were erythema, n = 23; itching, n = 14; and rash n = 8. nRCT or single arm extension results of RCT FGM vs SMBG Bailey 2015 [39] Skin issues observed in 202 site exams of 72 study participants: moderate to severe itching 0.5% of the time, moderate erythema 4.0% of the time, and 98.6% of the insertions had a pain rating of ≤ 2. Rate of mild incidences was < 9% for any individual category of skin issues mentioned above, including oedema, rash, induration, bruising, bleeding, and others. Edge 2017 [40] Five device related AEs were reported in total from five (6%) participants, aged 6, 9, 10, 12 and 15 years: allergic reaction, blister, pink mark/scabbing and abrasion on sensor removal (n=2) – four were mild, one was moderate, all were resolved at study completion. Site exams performed for all sensor insertions checked for anticipated AEs associated with sensor application or insertion sites – moderate erythema was observed on 11.6% of occasions, mild erythema and pain 13.6% and 4.1%, respectively, and mild instances of bleeding, bruising, itching and oedema were each on <3% of occasions. There were no trends in rate of anticipated AEs, including erythema, with age. Haak 2017 Nine participants reported 16 instances of device-related adverse events single arm results (e.g. infection, allergy) and 28 participants (20.1%) experienced 134 occurrences at 12 months [41] of anticipated skin symptoms/sensor-insertion events expected with device use (e.g. erythema, itching and rash).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 24 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Discussion

The population (mainly adults) included in RCTs is representative of patients usually included in such clinical trials (T1 and T2 DM patients). The majority of trials included T1 DM patients [13, 24, 25, 27, 29-40], hence results from these trials not be applied to T2 DM patients. Baseline characteristics showed that the studies included similar groups of patients. There were a small number of paediatric patients in only 1 RCT [31] and no RCTs on patients in pregnancy. The studies varied in terms of patients’ age and DM type, and inclusion criteria. The studies assessed various rtCGM and FGM devices. Different rtCGM devices were compared with SMBG in the different studies, with only one exception. Only one small head-to-head study compared rtCGM and FGM [34]. Two RCTs compared FGM with SMBG, one in T1 and one in T2 DM patients [27, 28].

The most commonly used CE marked-approved SMBG devices used in the majority of RCTs were included as comparators in this assessment. The choice of outcomes were in line with clinical guidelines (i.e. HbA1c changes from baseline to the end of the study; time spent in the target glycaemic range; time spent in hypoglycaemia; time spent in hyperglycaemia; hypoglycaemia and severe hypoglycaemia events; patient-reported outcome, quality of life (QoL) measures and user satisfaction) [1-9], and majority of included RCTs reported these outcomes [13, 24-41]. But outcomes measures were highly varied among studies, which prevented data pooling in MA (MA was done only for one outcome, HbA1c change from baseline to the end of the study) [25, 29]. Mortality was not specified as an outcome or reported in any of the RCTs included in this assessment. Hypogly- caemia was reported differently between studies as well as measures for user satisfaction and QoL. Patients were followed up from 8 weeks to 12 months. Different time periods of follow-up also hin- dered pooling of results.

RCTs included patients worldwide in an outpatient setting which is representative for the expected use.

Our SR provides a narrative summary of the best scientific evidence available, without providing any recommendations. That task should be left to the national/regional/local HTA and appraisal processes and decision making.

Assessing nonpharmacological treatments, including medical devices raises specific methodological issues, like difficulties of blinding [42]. If blinding is impossible, at least blinded endpoint evaluation is recommended. In 2017, CONSORT Statement extension for RCTs of nonpharmacological treatment was published with the aim of helping authors increase transparency of their reports and facilitate an accurate interpretation of study results [43]. Two of three new items added address description of attempts to limit bias if blinding is not possible and whether and how adherence of participants to intervention is assessed or enhanced. Blinding is one of important item evaluated in Risk of Bias (RoB) tool [43]; if there are no written attempts to limit bias if blinding is not possible, RoB will be deemed to be a major issue. According the GRADE Handbook, systematic reviewers should provide a comprehensive summary of the evidence but should not typically include healthcare recommendations. Separating judgements about strength of evidence from judgements about the strength of recommendations is important because both the direction and the strength of a recommendations may be modified after taking into account other factors such as implications for resource utilization, equity, acceptability, and feasibility of alternative management strategies [44]. An important limitation in this assessment was that MA was not possible for the majority of outcomes assessed due to methodological and clinical heterogeneity.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 25 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Conclusion

Based on a narrative summary of studies with important risk of biases and moderate to very low certainty of evidence results, use of CGM devices assessed was associated with a reduction in HbA1c in the majority of the studies that included MDII treated patients and in the two studies with MDII and CSII treated patients; both CGM and FGM devices assessed were associated with reduction in hypo- and hyperglycaemia outcomes and improved treatment satisfaction in patients with T1 or T2DM, compared with SMBG. No RCTs were found investigating devices assessed with CE mark authorisation in pregnancy. Only one RCT included solely a child population [31], only one study was head-to-head [34], and only four studies were performed on patients with impaired hypoglycaemia awareness [13, 30, 34, 38]. In all RCTs, patients were followed up from 8 weeks to 12 months. Due to the heterogeneity of populations, interventions, and outcomes measures, MA was done only for one outcome, i.e. HbA1c change from baseline to the end of the study (the DIAMOND and GOLD studies) [25, 29], showing statistically significant results in favour of CGM.

All four studies involving T1DM patients with IHA or previous severe hypoglycaemia events, including only one head-to-head study which compared CGM vs FGM with a follow-up of 8 weeks [34], showed statistically significant results which favoured CGM in terms of time spent in hypoglycaemia, as well as hypoglycaemia and severe hypoglycaemia events. These results (of low to very low evidential certainty) are of high clinical importance for both patients on MDII therapy and CSII patients, because both conditions predispose such patients to future hypoglycaemia episodes.

Further high quality head-to-head studies are needed on long-term relative effectiveness and safety comparing CGM and FGM devices, especially in children and pregnancy.

Adults, children, and parents with T1 DM reported very positive experiences with CGM and FGM devices; different benefits were shown as well as the most important barriers – high cost and unavailability of these medical devices in some countries.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 26 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

1 SCOPE

Description Project scope Population Patients with diabetes mellitus (Type 1 or Type 2, including adults and children, gestational DM) treated with insulin (insulin pump therapy or multiple daily injections – MDII) Intended use of the technology: ICD 10 codes: E10, E11, O24.4 Mesh-terms and non-Mesh terms: Diabetes Mellitus, Type 1; Diabetes Mellitus, Type 2; Diabetes, Gestational; Insulin; Injections, Subcutaneous; Insulin Infusion Systems Intervention Continuous glucose monitoring (CGM) systems (real-time) and flash glucose monitoring (FGM) systems as personal, standalone medical devices Mesh-terms and non-Mesh terms: Continuous glucose monitoring; Flash glucose monitoring Comparison Comparisons to the reference standard (SMBG) and head-to-head comparisons Patients on multiple daily insulin injection (MDII) MDII + Stand-alone CGM vs MDII +SMBG MDII + Stand-alone FGM vs MDII + SMBG MDII + Stand-alone CGM vs MDII + Stand-alone FGM MDII + Stand-alone CGM vs MDII + Stand-alone CGM Patients on insulin pump therapy (CSII) CSII + Stand-alone CGM vs CSII + SMBG CSII + Stand-alone FGM vs CSII + SMBG CSII + Stand-alone CGM vs CSII + Stand-alone FGM CSII + Stand-alone CGM vs CSII + Stand-alone CGM CSII + Stand-alone CGM vs sensor-integrated/augmented (enabled) CSII CSII + Stand-alone FGM vs sensor-integrated/augmented (enabled) CSII CSII + SMBG vs sensor-integrated/augmented (enabled) CSII Rationale: Comparators will be selected based on the recommendations from the relevant HTAs, clinical guidelines [1-9] and the EUnetHTA Guidelines [19]. Mesh-terms and non-Mesh terms: Blood Glucose Self-Monitoring; Multiple daily insulin injections; Insulin pump therapy; Continuous glucose monitoring; Flash glucose monitoring; Sensor-integrated/augmented (enabled) pump Outcomes COMET database was consulted to identify standardised core outcome sets, without success. TEC Domain Clinical validity: Device Accuracy A combination of the MARD score and Error Grid plots (Clarke Error Grid; Parkes Error Grid) define the accuracy of rtCGM and FGM systems. EFF Domain Clinical utility: Mortality Glycaemic control: change in HbA1c (glycosylated haemoglobin) Incidence of hypoglycaemia (i.e., level 1, 2 and 3 hypoglycaemia) Incidence of hyperglycaemia Time spent in range Time spent in hypoglycaemia Time spent in hyperglycaemia Quality of life Patient satisfaction Hypoglycaemia fear Incidence of diabetic ketoacidosis (in type 1 diabetes) Incidence of hyperosmolar, hyperglycaemic coma (in type 2 diabetes) Resource utilization related to diabetes mellitus (i.e., number of visits to emergency room, primary care, specialists; hospitalizations)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 27 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Description Project scope Number of daily finger-sticks tests Number of calibration Need of re-calibration Compliance/adherence: Percentage of time using rtCGM Number of sensor scans per day (in FGM system) Outcomes for EFF domain possibly used in meta-analysis will be: Change in HbA1c; Incidence of hypoglycaemia (i.e., level 1, 2 and 3 hypoglycaemia); Incidence of hyperglycaemia; Time spent in range; Time spent in hypoglycaemia; Quality of life SAF Domain Adverse events (AEs) (device related, i.e. Pain or discomfort related to glucose monitoring and not related to device)/Any AEs, Serious AE (SAE), most frequent AEs and SEAs, Death as SAE, withdrawals due AEs/Outcomes for SAF domain possibly used in meta-analysis will be: Frequency of any AEs and SAE. From the Checklist for potential ethical, organisational, patient and social, and legal aspects, if needed. Rationale: Outcomes will be selected based on the recommendations from the relevant HTAs, clinical guidelines [1-9] and the EUnetHTA Guidelines on Clinical and Surrogate Endpoints and Safety [19]. Study Clinical validity: design SR of accuracy studies or primary accuracy studies Effectiveness: If suitable evidence syntheses (SRs/HTA reports) are available: evidence syntheses (SRs/HTA reports) and primary studies (as described in next bullet) published after the last search date of the latest SR/HTA document If suitable evidence syntheses (SRs/HTA reports) are NOT available: Randomised controlled trials Prospective controlled studies, if no RCTs available Safety: If suitable evidence syntheses (SRs/HTA reports) are available: evidence syntheses (SRs/HTA reports) and primary studies (as described in next bullets) published after the last search date of the latest SR/HTA document If suitable evidence syntheses (SRs/HTA reports) are NOT available: Randomised controlled trials Non-randomised controlled trials Prospective studies with or without a control group Medical device adverse event registers and Post marketing surveillance data on device-related adverse events Organisational, ethical, patient and social, legal aspects: Qualitative and quantitative studies, reports or opinions (according to the ® EUnetHTA Core HTA Model 3.0), if needed [10]. Only English language studies will be included. Subgroup Type 1 DM-Type 2 DM-gestational DM; adults-children; Analysis adjunctive*-nonadjunctive**; (if possible) insulin pump therapy-multiple daily injections (MDII); patients with impaired awareness of hypoglycaemia; patients with awareness of hypoglycaemia

* Adjunctive: cannot be used to make treatment decision without confirmatory finger-stick testing ** Non-adjunctive: can be used to make treatment decision without confirmatory finger-stick testing CGM real-time: continuous glucose monitoring; FGM: flash glucose monitoring; CSII: continuous subcutaneous insulin infusion; MDII: multiple daily insulin injections; SMBG: self-monitoring blood glucose; Note: Professional (retrospective) devices will not be included in this assessment.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 28 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

2 METHODS AND EVIDENCE INCLUDED

2.1 Assessment Team

Agency for Quality and Accreditation in Health Care and Social Welfare (AAZ) Develop first draft of the project plan; Perform the literature search; Carry out the assessment: select and answer assessment elements (of all four domains TEC, CUR, EFF and SAF), fill in the checklist on potential “ethical, organisational, patient and social and legal aspects” of the HTA Core Model® for rapid REA; Send “draft versions” to reviewers for comments, compile feedback from reviewers and incorporate relevant changes to the draft; Prepare all draft versions and the final assessment including an executive summary.

Main Association of Austrian Social Security Institutions (HVB) Review the project plan draft; Support the production of TEC and CUR domains and quality check all steps of their production (data, information, sources); Contribute in answering questions related to potential ethical, organisational, patient and social, and legal aspects if needed. Approve/endorse conclusions drawn as well as all draft versions and the final assessment including the executive summary.

The Norwegian Institute of Public Health (NIPHNO) Review the project plan draft; Support the production of EFF and SAF domains and quality check all steps of their production; Contribute to data extraction, quality check and approve all assessment steps (data extraction, assessment of risk of bias, data analyses, and evidence syntheses); Review drafts of EFF and SAF assessments, propose amendments where necessary, and provide written feedback. Contribute in answering questions related to potential ethical, organisational, patient and social, and legal aspects if needed; Approve/endorse conclusions drawn as well as all draft versions and the final assessment including the executive summary.

Agency for Health Quality and Assessment of Catalonia (AQUAS), Spain; Healthcare Improvement Scotland (HIS), Scotland; Regione Emilia-Romagna (RER), Italy Thorough review of 1st draft project plan and 1st draft assessment report incl. studies and results.

2.2 Source of assessment elements

The selection of assessment elements was based on the EUnetHTA Core Model® Application for Rapid Relative Effectiveness (REA) Assessments (4.2) [11]. The Checklist for potential ethical, organisational, patient and social, and legal aspects of the HTA Core Model R for rapid REA was filled in as well. Additionally, further assessment elements from the EUnetHTA Core Model® domains: ethical analysis, organisational aspects, patients and social aspects, and legal aspects were included if deemed relevant (3.0) [10]. The selected issues (generic questions) were translated into actual research questions (answerable questions).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 29 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

2.3 Search

For Effectiveness (EFF) and Safety (SAF) domains, a systematic literature search according to the predefined search strategy (without limitations) was performed according to the Cochrane methodology in March 2018 [20], in standard medical and HTA databases (The Cochrane Central Reg- ister of Controlled Trials, The Database of Abstracts of Reviews of Effects, The Health Technolo- gy Assessment Database, NHS Economic Evaluation Database, MEDLINE, EMBASE, EBSCO CINAHL). Manual searches (from reference lists of relevant studies) were also carried out. The following clinical trials registries: ClinicalTrials.gov (http://www.clinicaltrials.gov/), WHO Internation- al Clinical Trials Registry Platform (http://apps.who.int/trialsearch/Default.aspx) and EU Clinical Trials Register (https://www.clinicaltrialsregister.eu/) were also searched for registered ongoing clinical trials and observational studies.

Relevant references (after duplicates removed) were screened and assessed for eligibility independently by two researchers. References were included or excluded according to the overall research question, Population-Intervention-Control-Outcome (PICO)-scheme (as described in Project Scope), and the predefined inclusion/exclusion criteria and presented according to the PRISMA Statement [45]. Detailed tables on search strategy can be found in Appendix 1.

All manufacturers whose products were included in the current assessment (as intervention) were contacted and invited to participate. Abbott Diabetes Care, Medtronic and Dexcom, Inc. agreed to participate and scoping face to face meetings were held on February 1st (Medtronic, Dexcom) and 2nd (Abbott Diabetes Care), 2018 in Vienna together with authors and co-authors. Furthermore, EUnetHTA evidence submission files were submitted by these three companies.

2.4 Study selection

1536 records were identified through database searching and 8 additional records were identified through other sources; (1120 results left after automatic deduplication; after manual deduplication 988 results left – a total of 416+132 duplicates = 548) 996 remained after duplicates were removed. 131 full-text articles were assessed for eligibility and after the exclusion of 112 full-text articles, 12 RCTs were included in our SR for EFF domain (published in 16 scientific articles). For the SAF domain, 3 prospective non-randomised studies were also included (two interventional non- randomised clinical accuracy and safety studies and one interventional single arm study), adding up to a total of 15 studies. Two studies are included in meta-analysis for EFF Domain.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 30 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Records identified through Additional records identified database searching through other sources (n = 1536) (n = 8) Identification Records after duplicates removed (n = 996)

Records screened

Screening (n = 996) Records excluded (n = 865)

Full-text articles assessed for eligibility (n = 131)

Eligibility Full-text articles excluded, with reasons (n = 112) Studies included in qualitative synthesis Exclusion criteria were: EFF • background literature; RCTs n = 12 (16 articles) • wrong population;

SAF n = 15 (n = 12 RCTs + n = 3 nRCTs) • not English; • preliminary study results; • duplicate of original Included publication; Studies included in quantitative synthesis • abstracts (meta-analysis) EFF (n = 2)

Figure 3: Flow chart

2.5 Data extraction and analyses

Data extraction related to Study characteristics (authors, year of publication, setting, study design, clinical trial identification number/registry identifier and funding source, objectives, study duration, statistical analysis), Patient characteristics (number of participants in the trial, age, sex, diagnosis, time since diagnosis with DM, comorbidities, insulin treatment), Intervention (type of medical devices), Comparators (type of medical devices), and Outcomes related to effectiveness (clinical utility): Mortality; Glycaemic control: Change in HbA1c (glycosylated haemoglobin); Incidence of hypoglycaemia (i.e., level 1, 2 and 3 hypoglycaemia); Incidence of hyperglycaemia; Time spent in range; Time spent in hypoglycaemia; Time spent in hyperglycaemia; Quality of life; Patient satisfaction; Hypoglycaemia fear; Incidence of diabetic ketoacidosis (in type 1 diabetes); Incidence of hyperosmolar, hyperglycaemic coma (in type 2 diabetes); Resource utilization related to diabetes mellitus (i.e., number of visits to emergency room, primary care, specialists, hospitalizations); Number of daily finger-sticks tests; Number of calibrations; Need for re-calibration; Compliance/ adherence: Percentage of time using CGM and Number of sensor scans per day (in FGM system); Safety /Adverse events (AEs) device related, i.e. Pain or discomfort related to glucose moni-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 31 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin toring and not related to device, Any AEs, Serious AE (SAE), most frequent AEs and SEAs, Death as SAE, withdrawals due AEs/ was performed by one researcher on pre-defined extraction tables and double-checked regarding completeness and accuracy by a second researcher. Any differences in extraction results were discussed to achieve consensus; any disagreements were resolved by consulting a third researcher. Organisational, ethical, patient and social, and legal aspects were considered if deemed relevant; please see the Checklist for potential ethical, organisational, patient and social, and legal aspects in Appendix 3.

An update of existing systematic reviews (SRs) was not possible, so a new SR of RCTs with a meta-analysis on one clinical outcome in EFF Domain was performed.

The meta-analysis combined results of studies (or included only the studies) that were considered clinically homogenous in terms of participants, interventions, and outcomes using the RevMan3 software. We used the mean difference along with the appropriate 95% confidence intervals (CI) and the appropriate Mantel-Haenszel method of meta-analysis. We performed a random-effects model of meta-analysis. Statistical significance was considered to be p<0.05. Forest plots were used for graphical display of the results. The level of heterogeneity was classified as either low (I2 less than 25%), moderate (I2 between 25% to 75%) or high (I2 over 75%) [46].

To summarize findings on relative effectiveness for outcomes on which MA was not possible, we used Forest plots for visualisation of effects, without pooling estimates. For this purpose, in some cases we had to convert medians into means. The method used to estimate means and standard deviations for outcomes that were originally reported using nonparametric summary statistics is described by Wan et al 2014 [18]. Details are provided in Appendix 1.

2.6 Quality rating

No quality assessment tool was used for the Description and technical characteristics of technology (TEC) and Health problem and current use of technology (CUR) domains, but multiple sources were used in order to validate individual, possibly biased, sources. We performed descriptive analyses of information from the various sources explored. The completed part of the EUnetHTA submission file from the manufacturer was used as a starting point. The Medical Devices Evidence Submission template was sent to all relevant manufacturers of the technology. Manufacturers were asked to submit non-confidential evidence, focusing on the technical characteristics and current use of the technology. The documentation provided was used in addition to the literature identified by the literature search. We performed descriptive analyses of information from the various sources explored.

The results from the included SRs were planned to be included according to the methodology suggested by Whitlock 2008 [47] and Robinson 2014 [48] on how to integrate existing SRs into new SRs. To answer our research questions, only one out of four possible approaches in using existing SRs, described in Robinson et al. 2014 [48], was used: (1) using the existing SR(s)’ list- ing of included studies as a quality check for the literature search and screening strategy conducted for the new review (Scan References). It was not possible to use the other three approaches: (2) using the existing SR(s) to completely or partially provide the body of included studies for one or more Key Questions in the new review (Use Existing Search); (3) using the data abstraction, risk of bias assessments and/or analyses from existing SRs for one or more Key Questions in the new review (Use Data Abstraction/Syntheses), and (4) using the existing SR(s), including conclusions, to fully or partially answer one or more Key Questions in this SR (Use Complete Review).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 32 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The risk of bias of included RCTs was evaluated independently by two researchers. The Cochrane risk of bias assessment approach was used for RCTs [20], on study level, on few critical outcomes (change in HbA1c from baseline to the end of the study; time spent in normoglycaemia; time spent in hyperglycaemia; time spent in hypoglycaemia; hypoglycaemia events and serious hypoglycaemia and QoL and patient satisfaction), and on safety outcome – incidence of adverse events (AEs) related to medical devices (local AEs). Quality of data in RCTs, related to a few critical outcomes according the clinical guidelines (change in HbA1c from baseline to the end of the study; time spent in normoglycaemia; time spent in hypoglycaemia; hypoglycaemia events and serious hypoglycaemia; QoL and patient satisfaction) was assessed using the GRADE methodology [20- 22]. This approach specifies four levels of quality: High: further research is very unlikely to change our confidence in the estimate of effect; Moderate: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimates; Low: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; Very low: we are very uncertain about the estimate.

2.7 Patient involvement

Two different methods for patient involvement were used in this assessment, one involving individual patients and second involving patient organisations. A Focus Group (with individual patients) was held in Croatia – one for adults and one for children with informal caregivers (i.e. two separate focus groups were held). Patients were recruited through the Zagreb Diabetes Associa- tion, Croatia. To minimise the risk of last minute cancelation, all participants were received a written confirmation of the invitation from the Focus Group (by email), including an information and consent form. All participants received a reminder (by phone or by email) one or two days before the Focus Group. Four to five-hours meeting was planned and four flipovers were used with predefined questions related to impact of condition; experience with currently available medical devices; experiences with, and expectations of, the medical devices being assessed; and additional information which patients believe would be helpful to the HTA researchers. The entire Focus Group discussion was recorded and transcriptions were made in the Croatian language. Written notes were taken as well. No ethical approval was necessary for the focus group in Croatia, but patients and informal caregivers (parents) were asked to sign an Informed consent form in the national language as well as EUnetHTA DOICU form. Two patient groups were contacted, one at the European level and one at the national level. A Patient Group Submission template was sent to the International Diabetes Federation European Region, Brussels and Diabetes Scotland, Scotland. The Patient Group Sub- mission template was prepared for this assessment by modifying the HTAi Patient Group Submis- sion template for HTA of health interventions (not medicines) [23] and with inclusion of topics and issues from EUnetHTA Core Model® 3.0 related to Patients and Social Domain aspect [10]. This was helpful in the assessment of the value of health technologies, specifically CGM and FGM medical devices. The form was intended to help patient groups present the range of experiences and views of patients with the disease/condition for which the health intervention is being assessed.

2.8 Description of the evidence used

For the Effectiveness domain, data from 12 RCTs were analysed (published in 16 scientific articles), and for the Safety domain, in addition to the already mentioned RCTs, data from 3 prospective non-randomised studies were also included (two interventional non-randomised clinical accuracy and safety studies and one interventional single arm study), adding up to a total of 15 studies (see table below).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 33 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 6: Main characteristics of studies included

Author and year or Study type/ Population (T1/T2)/ Number Intervention(s)/ Main endpoints Domain study name/Country or Study period of patients Control region/Registry number or follow-up randomised/analysed/Age RCTs rtCGM MDII patients Beck 2017 [25] Diamond trial RCT T1 MDII Dexcom G4® Platinum with Change in HbA1c levels; EFF, USA 24 weeks 158/158 an enhanced algorithm, Percentage of participants with HbA1c level less SAF 505 software/SMBG NCT02282397 ≥25 y (26-73 y) than 7.0%;Time in various glycaemic ranges; Patient satisfaction; Quality-of-life; AEs Ruedy 2017 [37] RCT T1 and T2 MDII Dexcom G4® Platinum with Change in HbA1c levels: EFF, A subset analysis of the 24 weeks 116/114 an enhanced algorithm, Time in various glycaemic ranges; Glucose SAF 505 software /SMBG DIAMOND trial ≥60 y variability; Frequency of blood glucose self- NCT02282397 monitoring; AEs Riddlesworth 2017 [35] RCT T1 MDII Dexcom G4® Platinum with Hypoglycaemic event frequency EFF A subset analysis of 24 weeks 158/156 an enhanced algorithm, 505 software /SMBG DIAMOND trial ≥25 y (26-73 y) NCT02282397 Polonsky 2017 [33] RCT T1 MDII Dexcom G4® Platinum with QoL, Satisfaction, Hypoglycaemia fear EFF A subset analysis of 24 weeks 158/155 an enhanced algorithm, 505 software /SMBG DIAMOND trial ≥25 y (26-73 y) NCT02282397 Lind 2017 [29] RCT two T1 MDII Dexcom G4® Change in HbA1c levels; Time in various glycaemic EFF, Sweden cross-over 161/142 (142 received Platinum/SMBG ranges; Well-being; satisfaction, diabetes distress, SAF arms design Hypoglycaemic fear and confidence; treatment Gold trial rtCGM, 142 received 26 weeks conventional therapy) adherence; self-measurement of glucose; NCT0209205 events of hypoglycaemia; AEs ≥18 y

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 34 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author and year or Study type/ Population (T1/T2)/ Number Intervention(s)/ Main endpoints Domain study name/Country or Study period of patients Control region/Registry number or follow-up randomised/analysed/Age Beck 2017[26] RCT T2 MDII Dexcom G4® Platinum with Change in HbA1c levels; Percentage of EFF, USA 24 weeks 158/158 an enhanced algorithm, participants with HbA1c level less than 7.0%; SAF 505 software /SMBG Time in various glycaemic ranges Part of DIAMOND trial ≥25 y Protocol Patient satisfaction; Quality-of-life; AEs NCT02282397 Heinemann 2018 [13] RCT T1 MDII Dexcom G5®/SMBG Number of hypoglycaemic and hyperglycaemic EFF, Germany 26 weeks 149/149 events Impaired hypoglycaemia awareness SAF assessed with the hypoglycaemia unawareness; HypoDE study ≥18 y QoL NCT02671968 Impaired hypoglycaemia Fear; Self-reported health status Satisfaction with awareness glucose measurement assessed with the Glucose Monitoring Satisfaction Survey MDII and CSII patients Battelino 2011 [24] RCT T1 MDII or CSII CGM (FreeStyle Navigator, Time spent in hypoglycaemia (<63 mg/dL) EFF, EU and Israel 24 weeks 120/116 Abbott Diabetes Care) vs during the 26-week study period; Time spent in SAF SMBG hyperglycaemia (>180 mg/dL or >250 mg/dL); The NCT00843609 10-65 y (stratified 10-17 y; number of hypoglycaemic excursions 18-65 y) (< 55 and < 63 mg/dL) per day; The number of hypoglycaemic excursions (< 55 and < 63 mg/dL) during the night period of 0000–0600 h; The risk associated with glucose concentration outside the recommended range; AEs Mauras 2012 [31] RCT T1 MDII or CSII CGM (FreeStyle Navigator, Decrease in HbA1c≥0.5% with no severe EFF, USA 26 weeks 146/146 Abbott Diabetes Care or for hypoglycaemic events; severe hypoglycaemia; SAF patients on Medtronic sensor use; biochemical hypoglycaemia; NCT00760526 Children 4-9.9 y Paradigm™ system – measures of variability; parental quality of life; Medtronic MiniMed™ patient satisfaction, hypoglycaemia fear; AEs MiniLink™ REAL-time transmitter) vs SMBG Riveline 2012 [36] RCT T1 MDII or CSII CGM (FreeStyle Navigator, HbA1C, glucose stability, hypoglycaemia, EFF, France 12 months Group 1 n=62/Control n=61 Abbott Diabetes Care) vs ketoacidosis, sensor use, QoL, patient satisfaction SAF SMBG NCT00726440 8-60 y

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 35 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author and year or Study type/ Population (T1/T2)/ Number Intervention(s)/ Main endpoints Domain study name/Country or Study period of patients Control region/Registry number or follow-up randomised/analysed/Age Van Beers 2016 [38] RCT T1 CSII and MDII Paradigm™ Veo system Change in HbA1c EFF, Netherlands cross-over 52/52 used solely as a monitor Time in glycaemic ranges SAF 16 weeks with a MiniLink™ transmitter IN CONTROL trial 18-75 y (Medtronic), and the Enlite™ Severe hypoglycaemia events NCT01787903 Impaired hypoglycaemia glucose sensor/CSII-treated Frequency and duration of CGM-derived hypo awareness patients continued using glycaemic episodes CSII=23 their own pump for insulin Self-reported hypoglycaemia awareness QoL; treatment with SMBG MDII=29 Satisfaction with use of CGM; AEs CSII patients Ly 2013 [30] RCT T1 CSII Medtronic Paradigm™ Change in HbA1c levels, % (95% CI); EFF, Australia 24 weeks 95/95 Veo/CSII with SMBG Number of people with hypoglycaemic events; SAF Hypoglycaemic incidence rate; HUS ACTRN12610000024044 Mixed population 4-50 y CSII + CGM + Suspend: (70% children <18 y) sensor-integrated pump (Medtronic Paradigm ™ Impaired hypoglycaemia Veo System, Medtronic awareness MiniMed™) with automated insulin suspension n=46; CSII + SMBG: continue using their insulin pump n=49 CGM vs FGM Reddy 2018 [34] RCT T1 40/40 CGM vs FGM Change in time spent in glycaemic ranges; EFF, UK 8 weeks ≥ 18 years Dexcom G5® vs Abbott Low blood glucose index; Severe hypoglycaemia; SAF Freestyle Libre® Hypoglycaemia risk; HbA1c levels; Gold Score; NCT03028220 Impaired hypoglycaemia Hypoglycaemia fear; Diabetes-related emotional awareness distress

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 36 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author and year or Study type/ Population (T1/T2)/ Number Intervention(s)/ Main endpoints Domain study name/Country or Study period of patients Control region/Registry number or follow-up randomised/analysed/Age FGM MDII and CSII patients Bolinder 2016 [27] Impact trial RCT T1 MDII or CSII FGM vs SMBG Time spent in glycaemic ranges EFF, EU 24 weeks 241/239 Abbott Freestyle Libre® HbA1c SAF NCT02232698 ≥ 18 years System utilisation Number of sensor scans per day Frequency of glucose finger-sticks tests; Emergency room visits or admissions and non- protocol related additional clinic time; Patient- recorded outcome measures; Adverse events and sensor insertion-site symptoms Haak 2017 [28]Replace trial RCT T2 MDII or CSII FGM vs SMBG Difference in HbA1c; Time in glycaemic ranges; EFF, EU 24 weeks 224/201 Abbott Freestyle Libre® number and duration of hypoglycaemic and hyper- SAF glycaemic events; Mean glucose, and glucose NCT02082184 ≥ 18 years variability measures; frequency of glucose finger- sticks and sensor scans per day during the study period, system utilization; change in total daily dose of insulin, body mass index (BMI), weight, participant questionnaire, blood pressure, lipid levels, HCP questionnaire responses, emergency room visits, hospital admissions, additional clinic time, lancet use and non-insulin medication use. AEs MDII patients Oskarsson 2018 [32] RCT T1 MDII FGM vs SMBG Mean time in hypoglycaemia; sensor scanning EFF, MDII subgroup of Impact trial 24 weeks 167/161 Abbott Freestyle Libre® frequency; number of self-monitored blood glucose SAF tests; treatment satisfaction; perception of hypo/ EU ≥ 18 years hyperglycaemia; AEs NCT02232698

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 37 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author and year or Study type/ Population (T1/T2)/ Number Intervention(s)/ Main endpoints Domain study name/Country or Study period of patients Control region/Registry number or follow-up randomised/analysed/Age nRCTs Bailey 2015 [39] nRCT T1 MDII or CSII FGM Accuracy, safety, usability SAF USA 14 days 75/72 Abbott Freestyle Libre® NCT02073058 Prospective, 18-71 y single-arm, clinical study Edge 2017 [40] nRCT T1 MDII or CSII FGM Accuracy, safety, user acceptability SAF UK 14 days 89/89 Abbott Freestyle Libre® NCT02388815 Prospective, 4-17 y single-arm, clinical study Haak 2017 [41] RCT T2 MDII or CSII 139/125 FGM Safety endpoints: incorporated all adverse events, SAF EU 12 months ≥ 18 years Abbott Freestyle Libre® including severe hypoglycaemia (requiring third- party assistance), hypoglycaemic events and NCT02082184 sensor insertion or sensor wear-related symptoms, Single arm results at diabetic ketoacidosis or hyperosmolar hyper- 12 month of Replace trial glycaemic state episodes and cardiac events. (participants in the interventional group continued into the further 6 months open-access phase)

Abbreviations: AE - adverse event; AUC, area under the curve; DKA - diabetic ketoacidosis; EQ-5D - European Quality of Life-5 Dimensions scale; NR - not reported; NS - not significant; U - units. The AUC is the product of the magnitude and duration of the sensor measured glucose level above or below a specified cut-off level. Higher values for this calculation indicate more numerous, severe or protracted glycaemic events. HUS, Hypoglycaemia Unawareness Score (Clarke questionnaire), higher is worse Sources: [13, 24-41]

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 38 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

2.9 Deviations from project plan

An update of existing systematic reviews (SRs) was not possible due to different scope and thus a new SR of RCTs. Meta-analysis was possible for only one clinical outcome in the EFF Domain (HbA1c changes) due to the high heterogeneity between trials. Neither NMA nor subgroup analyses were performed.

Payer representatives at the EU level (The International Association of Mutual Benefit Societies (AIM), Brussels) were not contacted related to reimbursement status of technologies under assessment since relevant data were received in the Manufacturer Submission file.

SR including accuracy studies was not performed, hence this assessment does not provide accuracy data related to medical devices under assessment.

In the later phase of this assessment a new medical device, Dexcom G6® Continuous Glucose Monitoring System, Dexcom, was CE marked and was therefore included in the 3rd version of this assessment in the TEC Domain (Manufacturer updated the Submission file and sent it together with IFU on 22/06/2018).

The manufacturer of the medical device SugarBeat®, Nemaura Medical Inc. informed us during the assessment that the 1st generation SugarBeat® (which consists of a disposable adhesive skin-patch connected to a rechargeable transmitter, with an app displaying glucose readings) is not on the market, hence information on this device was deleted from TEC Domain text. The 2nd generation SugarBeat® device does not yet have a CE-mark (expected to receive it later this year).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 39 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

3 DESCRIPTION AND TECHNICAL CHARACTERISTICS OF TECHNOLOGY (TEC)

3.1 Research questions

Element ID Research question B0001 What are the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) and the comparators medical devices? A0020 For which indication has the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – received CE marking? B0002 What is the claimed benefit of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – in relation to the comparator(s) medical devices? B0003 What is the phase of development and implementation of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices? B0004 Who administers the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices and in what context and level of care are they provided? B0008 What kind of special premises are needed to use the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices? B0009 What equipment and supplies are needed to use the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices? A0021 What is the reimbursement status of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems)?

3.2 Results

Features of the technology and comparators

[B0001] – What are the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) and the comparators medical devices? [A0020] – For which indication has the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – received CE marking?

Technology in this assessment includes real-time continuous glucose monitoring (rtCGM) systems and flash glucose monitoring (FGM) systems as personal, standalone medical devices for patients with diabetes mellitus type 1 and type 2 (DM1 and DM2), including adults and children, and gestational DM treated with insulin, either through insulin pump therapy or multiple daily insulin injections (MDII).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 40 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Standalone medical devices for glucose monitoring can be adjunctive, what means that cannot be used to make treatment decision without confirmatory finger-stick testing, or non-adjunctive, which can be used to make treatment decision without confirmatory finger-stick testing and may or may not demand calibration by finger sticks measurement [12].

Different definitions and/or category names are currently being used for continuous glucose monitoring (CGM) systems and flash glucose monitoring (FGM) systems in the published literature. FGM, for instance, may be described as a separate entity from CGM between a traditional blood glucose meter and a continuous glucose monitoring (CGM) system, or as special case of CGM or subset of CGM or intermittently viewed CGM (iCGM) or “flash” CGM [1, 2, 4, 6, 9].

Therefore, the authors of this assessment decided to use definitions according to the specific “Instruction for Use” documents – “Indication for use” sections – and are thus referring to them as flash glucose monitoring (FGM) system and continuous glucose monitoring (CGM) systems.

Professional (retrospective) devices are not included in this assessment.

STAND-ALONE REAL-TIME CONTINUOUS OR FLASH GLUCOSE MONITORING SYSTEMS

Following stand-alone systems (with approved indications) for measuring interstitial fluid glucose levels – real-time continuous or flash glucose monitoring systems – are assessed in this Rapid REA:

Real time continuous or flash glucose monitoring system with CE Mark Indications for use

Real time continuous or flash CE Mark Indications for use glucose monitoring system G4® PLATINUM Continuous For detecting trends and tracking patterns in persons (age 2 and older) Glucose Monitoring with diabetes. The system is intended for use by patients at home and System, Dexcom in healthcare facilities. The Dexcom G4® PLATINUM System is indicated for use as an adjunctive device to complement, not replace, information obtained from standard home glucose monitoring devices. The Dexcom G4® PLATINUM System aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments, which may minimize these excursions. Interpretation of the Dexcom G4 PLATINUM® System results should be based on the trends and patterns seen with several sequential readings over time. G5® Mobile Continuous For the management of diabetes in persons age 2 years and older. Glucose Monitoring The Dexcom G5® Mobile CGM System is designed to replace finger- System, Dexcom stick blood glucose testing for diabetes treatment decisions. Interpretation of the Dexcom G5® Mobile CGM System results should be based on the glucose trends and several sequential readings over time. The Dexcom G5® Mobile CGM System also aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments. The Dexcom G5® Mobile CGM System is intended for use by patients at home and in healthcare facilities. G6® Continuous Glucose A glucose monitoring system indicated for persons age 2 years and older, Monitoring System, designed to replace finger-stick blood glucose (BG) testing for treatment Dexcom decisions. Interpretation of the Dexcom G6® System results should be based on the glucose trends and several sequential readings over time. The Dexcom G6® System also aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments. The Dexcom G6® System is intended for use by patients at home and in healthcare facilities.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 41 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Real time continuous or flash CE Mark Indications for use glucose monitoring system FreeStyle Navigator II® For continually measuring interstitial fluid glucose levels in people (age 6 Continuous Glucose and older) with diabetes mellitus. The indication for children (age 6 - 17) Monitoring System, is limited to those who are supervised by a caregiver who is at least 18 Abbott years of age. The caregiver is responsible for managing or assisting the child to manage the FreeStyle Navigator II® System and also for interpreting or assisting the child to interpret FreeStyle Navigator II® readings. The FreeStyle Navigator II® Continuous Glucose Monitoring System is designed to replace blood glucose testing in the self-management of diabetes with the exceptions listed below. Under the following circumstances, use a blood glucose meter to check the current glucose readings from the FreeStyle Navigator II® Continuous Glucose Monitoring System Sensor: During times of rapidly changing glucose levels, interstitial glucose levels as measured by the Sensor and reported as current may not accurately reflect blood glucose levels. When glucose levels are falling rapidly, glucose readings from the Sensor may be higher than blood glucose levels. Conversely, when glucose levels are rising rapidly, glucose readings from the Sensor may be lower than blood glucose levels; In order to confirm hypoglycaemia or impending hypoglycaemia as reported by the Sensor; If symptoms do not match the FreeStyle Navigator II® Continuous Glucose Monitoring System reading. Do not ignore symptoms that may be due to low blood glucose or high blood glucose. Guardian Connect® For continuous monitoring of glucose levels in the interstitial fluid under Continuous Glucose the skin, in persons with diabetes mellitus. Monitoring system, ® Medtronic The Guardian Connect app (CSS7200): intended for continuous or periodic monitoring of glucose levels in the interstitial fluid under the skin, in persons with diabetes mellitus. The Guardian Connect® app is intended for use with a compatible consumer mobile electronic device. It allows users to track patterns in glucose concentrations and to possibly identify episodes of low and high glucose. The Guardian Connect® app displays alerts if a glucose level reaches, falls below, or rises above set values. Sensor glucose values displayed on the screen are not intended to be used directly for making therapy adjustments, but rather to provide an indication of when a meter blood glucose measurement may be required. Guardian Connect® transmitter (MMT-7821*): for single-patient or multiple-patient use as a component of select Medtronic CGM systems Enlite™ sensor (MMT-7008*): intended for use with Medtronic Diabetes (Medtronic) glucose sensing systems to continuously monitor glucose levels in persons with diabetes. CareLink™ Connect feature: intended to work with the Guardian Connect® CGM system. The CareLink™ Connect feature is intended to provide a secondary display of continuous glucose monitoring on a supported consumer electronic device for users of a Guardian Connect® CGM system and their designated care partners. The CareLink™ Connect feature is not intended to replace the real-time display of continuous glucose monitoring. All therapy decisions should be based on blood glucose measurements obtained from a blood glucose meter. The CareLink™ Connect feature is not intended to analyse or modify the continuous glucose monitoring data that it receives. Nor is it intended to control any function of the continuous glucose monitoring system to which it is connected.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 42 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Real time continuous or flash CE Mark Indications for use glucose monitoring system Eversense® Continuous The Eversense® CGM System is indicated for continually measuring Glucose Monitoring system, interstitial fluid glucose levels in adults (18 years and older) with diabetes Senseonics, Incorporated for the operating life of the sensor. The system is intended to: Aid in the management of diabetes; Provide real-time glucose readings; Provide glucose trend information; Provide alerts for the detection and prediction of episodes of low blood glucose (hypoglycaemia) and high blood glucose (hyperglycaemia). Historical data from the system can be interpreted to aid in providing therapy adjustments; These adjustments should be based on patterns and trends seen over time. The system is indicated for use as an adjunctive device to complement, not replace, information obtained from standard home blood glucose monitoring devices. Eversense® XL Continuous The Eversense® XL CGM System is indicated for continually measuring Glucose Monitoring system, interstitial fluid glucose levels in adults (18 years and older) with diabetes Senseonics, Incorporated for the operating life of the sensor. The system is intended to: Aid in the management of diabetes; Provide real-time glucose readings; Provide glucose trend information; Provide alerts for the detection and prediction of episodes of low blood glucose (hypoglycaemia) and high blood glucose (hyperglycaemia). Historical data from the system can be interpreted to aid in providing therapy adjustments. These adjustments should be based on patterns and trends seen over time. The system is indicated for use as an adjunctive device to complement, not replace, information obtained from standard home blood glucose monitoring devices. FreeStyle Libre® Flash Measuring interstitial fluid glucose levels in people (age 4 and older) Glucose Monitoring with diabetes mellitus, including pregnant women. System, Abbott The indication for children (age 4 - 12) is limited to those who are supervised by a caregiver who is at least 18 years of age. The caregiver is responsible for managing or assisting the child to manage the FreeStyle Libre® Flash Glucose Monitoring System and also for interpreting or assisting the child to interpret FreeStyle Libre® readings. It is designed to replace blood glucose testing in the self-management of diabetes with the exceptions listed below. Under the following circumstances, blood glucose meter should be used to check the current glucose readings from the FreeStyle Libre® Flash Glucose Monitoring System Sensor: During times of rapidly changing glucose levels, interstitial glucose levels as measured by the Sensor and reported as current may not accurately reflect blood glucose levels. When glucose levels are falling rapidly, glucose readings from the Sensor may be higher than blood glucose levels. Conversely, when glucose levels are rising rapidly, glucose readings from the Sensor may be lower than blood glucose levels. In order to confirm hypoglycaemia or impending hypoglycaemia as reported by the Sensor. If symptoms do not match the FreeStyle Libre® Flash Glucose Monitoring System reading.

* These numbers are subject to change and may therefore no longer be valid for the mentioned device or system Source: [49-57]

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 43 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

COMPARATORS MEDICAL DEVICES

Comparators include self-monitoring blood glucose (SMBG) medical devices as the reference standard, but head-to-head comparisons are also planned [58].

Comparisons to the reference standard (SMBG) and head-to-head comparisons

Patients on multiple daily insulin injection (MDII) MDII + Stand-alone CGM vs MDII +SMBG MDII + Stand-alone FGM vs MDII + SMBG MDII + Stand-alone CGM vs MDII + Stand-alone FGM MDII + Stand-alone CGM vs MDII + Stand-alone CGM

Patients on insulin pump therapy (CSII) CSII + Stand-alone CGM vs CSII + SMBG CSII + Stand-alone FGM vs CSII + SMBG CSII + Stand-alone CGM vs CSII + Stand-alone FGM CSII + Stand-alone CGM vs CSII + Stand-alone CGM CSII + Stand-alone CGM vs sensor-integrated/augmented (enabled) CSII CSII + Stand-alone FGM vs sensor-integrated/augmented (enabled) CSII CSII + SMBG vs sensor-integrated/augmented (enabled) CSII

CGM real-time: continuous glucose monitoring; FGM: flash glucose monitoring; CSII: continuous subcutaneous insulin infusion; MDII: multiple daily insulin injections; SMBG: self-monitoring blood glucose; Note: Professional (retrospective) devices will not be included in this assessment.

Self-monitoring blood glucose (SMBG) medical devices

Self-monitoring of blood glucose or SMBG refers to home blood glucose testing for people with diabetes. SMBG medical devices are blood glucose monitors, measuring blood glucose concentration using a drop of venous, arterial or, mainly, capillary blood from a finger puncture. Self-monitoring of blood glucose is recognized as one approach to improve glycaemic control and reduce hypoglycaemic events by alerting patients with diabetes to modify their hypoglycaemic drug dose, diet, and physical activities; several different blood glucose monitors are available on the market.

Sensor integrated and sensor augmented (or enabled) insulin pump systems

The following sensor integrated and sensor augmented (or enabled) insulin pump systems compatible (connected) with specific CGM systems (with approved indications), were used as comparators in head-to-head analysis (if clinical studies are available) and are listed below:

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 44 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Sensor integrated and sensor augmented (or enabled) insulin pump systems

Sensor integrated and sensor enabled CE Mark Indication for use (or augmented) insulin pump systems compatible (connected) with specific CGM system Paradigm™ Veo Pump system, For the continuous delivery of insulin, at set and variable rates, integrated with MiniLink™ transmitter for the management of diabetes mellitus in persons requiring and Enlite™ Glucose Sensor, insulin. In addition, the pump system is indicated for continuous Medtronic or periodic monitoring of glucose levels in the fluid under the skin, and possible low and high blood glucose episodes. MiniMed™ 640G system, integrated For the continuous delivery of insulin, at set and variable rates, with Guardian 2 Link transmitter and for the management of diabetes mellitus in persons requiring Enlite™ Glucose Sensor, Medtronic insulin. In addition, the system is indicated for continuous or periodic monitoring of glucose levels in the fluid under the skin, and detecting possible low and high glucose episodes. t:slim X2 Insulin Pump Compatible For the subcutaneous delivery of insulin, at set and variable with Dexcom G5® CGM, Tandem rates, for the management of diabetes mellitus in persons Diabetes Care, Inc. requiring insulin. The t:slim X2 Insulin Pump can be used solely for continuous insulin delivery and as part of the t:slim X2 System to receive and display continuous glucose measurements from the Dexcom G5 Mobile® Sensor and Transmitter. The t:slim X2 System also includes continuous glucose monitoring (CGM) indicated for the management of diabetes. The Dexcom G5 Mobile® CGM is designed to replace finger-stick blood glucose testing for diabetes treatment decisions. Omnipod® System, Compatible with For subcutaneous delivery of insulin at set and variable rates Dexcom G5® CGM, Insulet Corporation for the management of diabetes mellitus in persons requiring insulin and for the quantitative measurement of glucose in fresh capillary whole blood (in vitro) from the finger; It consists of a Personal Diabetes Manager (PDM) with integrated blood glucose meter and the “pod,” which delivers insulin; CGM’s provide real-time glucose readings every five minutes for people with type 1 or type 2 diabetes, and many Podders™ count on the Dexcom CGM to provide glucose readings throughout the day and night, including the speed and direction of glucose trends.

Source: Instructions for use documents [59-62]

Continuous glucose monitoring systems

Continuous glucose monitoring systems measure glucose levels in the interstitial fluid surrounding skin cells. These measurements supplement conventional SMBG by monitoring the glucose fluctuations continuously over a stipulated period of time, thereby identifying fluctuations that would not be identified with SMBG alone.

Two types of CGM are available: • Retrospective systems that measure glucose concentrations during a certain time span; the information is stored in a monitor and can be downloaded later. • Real time systems (rtCGM) that continuously provide the actual glucose concentration on a display [63].

CGM systems employ minimally-invasive approaches to determine the glucose concentration in interstitial fluid. All currently available systems consist of a small needle which is inserted in the abdominal subcutaneous fat. The tip of the needle houses a small glucose sensor which can measure glucose levels in the fluid which surrounds the fatty tissue [64].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 45 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The sensor is wired to a transmitter. The device (except Dexcom G6®, the first factory calibrated rtCGM) requires calibration using a capillary blood glucose measurement. Each sensor continuously measures glucose every 5-10 seconds averaging these values every 5 minutes and storing this data in the monitor’s memory. Depending on the device used, the algorithm in the device can measure glucose over a 3, 6, 7, or 10 day period using one sensor. After the 3, 6, 7, or 10 day period, a new sensor is required. The device is equipped with alarms which warn the patient of impending hypo- or hyperglycaemia [63].

The claimed benefit of CGM is the continuous provision of information regarding the blood-glucose concentration, to facilitate the adjustment of the insulin dosage. Disadvantages of CGM are the several minute delay of the measurements which may impede optimal monitoring and some patients may not like the continuous provision of information that confronts them with their illness at all times.

CGM may be useful for children (to reduce the often high number of finger punctures in this group), for patients with poorly controlled diabetes, for pregnant women in whom tight glucose control is essential with respect to the outcome of pregnancy, and for patients with hypoglycaemia unawareness (to prevent episodes of hypoglycaemia).

CGM system can be combined with an insulin pump, integrated in insulin pump or used as standalone glucose monitoring device [64].

External receivers are available for the Dexcom G4®, G5®, and G6® devices; the G5® and G6® devices can also be connected to a smartphone device that can be used as a receiver.

The following real-time continuous glucose monitoring systems are within the scope of this assessment: The G4® PLATINUM Continuous Glucose Monitoring System, Dexcom; G5® Mobile Contin- uous Glucose Monitoring System, Dexcom; G6® Continuous Glucose Monitoring System, Dexcom; GuardianTM Connect Continuous Glucose Monitoring system, Medtronic; Eversense® Continuous Glucose Monitoring system, Senseonics Incorporated; Eversense® XL Continuous Glucose Moni- toring (CGM) System, Senseonics Incorporated; and FreeStyle Navigator II® Continuous Glucose Monitoring System, Abbott. The only Flash glucose monitoring system within the scope of this assessment is the FreeStyle Libre® Flash Glucose Monitoring System, Abbott.

G4® PLATINUM Continuous Glucose Monitoring System, Dexcom

Dexcom’s G4® PLATINUM CGM System includes the sensor, the transmitter, and the receiver. The sensor is a disposable unit that is inserted under the skin to continuously monitor glucose levels for up to 7 days. In adults aged 18 or older, the sensor should be inserted in the belly, while for children and adolescents between 2 and 17 years old the sensor can be inserted in the belly (front of the body) or the upper buttocks (back of the body). The sensor can be placed above or below the belt line. The best areas to insert the sensor are usually flat and “pinchable“. Sensor insertion should be avoided in places where something may rub or press against the sensor, and the area of insertion should be at least 7.62 cm from where patient plans to inject insulin. The sensor is water resistant when showering, bathing, or swimming if the transmitter is fully snapped in place. Continuous sensor glucose readings are updated every 5 minutes for up to 7 days. The transmitter is a reusable device that wirelessly sends sensor-measured glucose information to the receiver. Its battery lasts at least 6 months. The receiver is a small hand-held device (10.2 cm × 4.6 cm × 1.3 cm) that shows sensor glucose readings, a direction and rate of change arrow, and trend graph.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 46 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Dexcom G4® PLATINUM Continuous Glucose Monitoring System requires calibration 2 hours after insertion of the sensor and every 12 hours after the initial startup calibration. Calibrating less often than every 12 hours might cause sensor glucose readings to be inaccurate. Dexcom G4® PLATINUM does not replace a blood glucose meter and for treatment decisions, such as how much insulin patient should take, the blood glucose value from blood glucose meter should be used. The direction, rate of glucose change, and trend graph on Dexcom G4 PLATINUM® System provide additional information to help with diabetes management decisions.

® Figure 4: G4 PLATINUM Sensor, Transmitter and receiver (Source: [49])

Dexcom G4® PLATINUM CGM System allows patients to see glucose information through the 1, 3, 6, 12, and 24 hour trend graphs. The trend graph provides additional information that blood glucose meter does not. It shows the current glucose value, the direction it is changing and how fast it is changing. The system only reports glucose information in a range between 2.2-22.2 mmol/L.

The Dexcom G4® PLATINUM System lets the patient to create his/her own personal settings for how he/she wants the receiver to tell them what is going on. The low and high glucose alerts will tell the patient when his/her sensor glucose readings are outside his/her target glucose range. Rise and fall (rate of change) alerts let the user know when their glucose levels are changing rapidly. The Dexcom G4® PLATINUM System also features a low glucose alarm that cannot be adjusted or turned off and is set at 3.1 mmol/L. This is a safety feature which tells the user that his/her glucose level may be dangerously low. In addition to the alert screens that appear on receiver display, patients can also set high and low glucose alerts to notify them with vibrations and beeps. This feature can be helpful during times such as sleeping, driving, exercising, or during meetings.

The so called “Events feature” allows the patient to record information about diabetes management that may help him/her and his/her healthcare professionals better understand glucose patterns and trends. Details about carbohydrate intake, insulin intake, exercise, and health issues can be entered. Trends and track the patterns can be viewed using the Dexcom Studio software.

Contraindications The Dexcom G4® PLATINUM Sensor, Transmitter, and Receiver must be removed prior to Magnetic Resonance Imaging (MRI), CT scan, or diathermy treatment. The Dexcom G4 PLATINUM System has not been tested during MRI or CT scans or with diathermy treatment, and it is unknown if there are safety or performance issues. Taking acetaminophen (paracetamol) containing products (such as Tylenol) while wearing the sensor may falsely raise your sensor glucose readings. The level of inaccuracy depends on the amount of acetaminophen (paracetamol) active in a patient´s body. The Dexcom G4® PLATINUM System is not approved for use in pregnant women or persons on dialysis.

Source: [49]

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 47 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Dexcom G5® mobile Continuous Glucose Monitoring System, Dexcom

The Dexcom G5® Mobile Continuous Glucose Monitoring (CGM) System is a medical device which allows user to track glucose trends and monitor the speed and direction of glucose changes. Sen- sor glucose readings can be seen continually, updated every five minutes for up to 7 days, without taking constant finger-stick measurements. The system is comprised of three key parts: single use sensor, reusable transmitter, and display devices: rechargeable receiver (optional in some countries); Dexcom G5® Mobile App (downloaded to the patient’s smart device, the app is not available for all smart devices and countries); Dexcom Share/Follow.

Sensor glucose readings are measured by a single use sensor inserted under the belly’s (if between the ages of 2 and 17, belly or upper buttocks) skin. The sensor wire is made of silver and platinum with polymer membranes. Once inserted, the thin and flexible wire measures glucose levels in the interstitial fluid for up to seven days. The sensor comprises an applicator, a sensor probe, and a sensor pod. The disposable applicator is used to insert the sensor probe. The patient inserts the sensor probe just beneath the skin (subcutaneous tissue) of the abdominal wall at an angle using a small needle inside the applicator, which is then withdrawn. At no time is the needle visible. The sensor probe measures glucose levels in surrounding interstitial tissue fluid. The sensor pod adheres to the patient’s abdomen to hold the transmitter in place. The sensor is left in place for up to seven days and is then replaced with a new one. The sensor can be used with both Dexcom G5® and G4® systems [65].

A reusable transmitter sends the data to the display device once every five minutes. It is water resistant so that user can shower, bathe, or swim and can transmit data to the display devices for up to six meters (range is lower if in or under water). The battery within the transmitter lasts approximately three months. The Dexcom G5® Mobile CGM System transmitter works with a number of display devices giving flexibility to the patient to use what is best for his/her situation, or lifestyle.

The Dexcom G5® Mobile CGM System is the first CGM system where a smart device (smart phone or similar device) can act as a receiver. Multiple smart devices cannot be used at the same time, but the receiver can be combined with a smart device during a sensor session. The app is not available for all smart devices and countries. A list of compatible smart devices is available on manufacturer’s internet page [48]. Just like the dedicated G5® receiver, the mobile app receives data from the sensor and displays glucose sensor readings, trend graphs, trend arrows, and alerts. Dexcom Share in the Dexcom G5® app allows sensor glucose readings, trends, and other glucose data to be viewed remotely on a smart device. The data can be shared with up to five people (“followers”), as determined by the user. After being invited by the “Sharer” and by downloading the Dexcom Follow™ App, an individual becomes a “Follower”. The user determines what a Follower can see, including the user’s sensor glucose readings, trends, alarm/alerts when the user’s glucose is low or high, and messages. The Dexcom Share/G5® System has been developed with technology that provides a high level of security for Personally Identifiable Information (PII) and Private Health Information (PHI) of registered users (G4® PII, G5® PII) [65].The primary difference between the Dexcom G5® Receiver and Dexcom G5® App is not the information they give, but how that information is presented [65].

The rechargeable receiver measures 10.2 cm × 4.6 cm × 1.3 cm. It is programmed to collect and process data from the sensor and to display the results as the current glucose value, glucose trend and direction, and speed of glucose level change. The optional receiver must be kept within 20 feet (~600 cm) of the sensor and transmitter at all times for optimal performance. The rechargeable battery lasts for about three days before requiring a further charge. The receiver is neither water resistant nor waterproof and can get damaged if moisture gets inside [65].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 48 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

There are important times when the system must be calibrated: Initial or Startup Calibration: two hours after insertion of the sensor; 12 Hour Update: every 12 hours after two hour startup calibration; when system prompts the patient. Without calibrations the sensor may be inaccurate, and as a result the display device’s sensor glucose readings, alerts, and prompts as well.

® Figure 5: Dexcom G5 Mobile CGM System Sensor, Transmitter and receiver (Source: [65])

Sensor glucose readings from Dexcom G5® Mobile CGM System can be used to make treatment decisions, including insulin doses. Other features that were not available in previous generations include: Dexcom G5® Mobile App for smart device; Updates to the Dexcom G5® Mobile Receiver screens and Dexcom Share™ in the Dexcom G5® Mobile App. BG meter values should be used as a backup when CGM data does not reflect how the patient feels or if patient has sensor reading gaps.

Contraindications Remove the Dexcom G5® Mobile CGM System sensor, transmitter, and receiver before Magnetic Resonance Imaging (MRI), Computed Tomography (CT) scan, or high-frequency electrical heat (diathermy) treatment. The Dexcom G5® Mobile CGM System has not been tested during MRI or CT scans or with diathermy treatment. The magnetic fields and heat could damage the components of the Dexcom G5® Mobile CGM System, which may cause it to display inaccurate blood glucose readings or may prevent alerts. Taking medications with paracetamol/acetaminophen while wearing the Dexcom G5® Mobile CGM System may inaccurately raise the glucose readings generated by the Dexcom G5® Mobile CGM System. The level of inaccuracy depends on the amount of paracetamol/acetaminophen active in your body and is different for each person. Do not rely on continuous glucose monitoring (CGM) data produced by the Dexcom G5® Mobile CGM System if you have recently taken paracetamol/acetaminophen. The Dexcom G5® Mobile CGM System was not evaluated or approved for the pregnant women and persons on dialysis. The Dexcom G5® Mobile CGM System’s accuracy has not been tested in people within these groups and the system’s glucose readings may be inaccurate.

Source: [50]

The Dexcom G5® Mobile CGM System provides personalized trend alerts, prompting patients to proactively react when their glucose levels are getting too low or too high. Alerts and alarm values can be customised by the user in respect of blood glucose (BG) levels and rates of glucose rise and fall. They can be changed to alert the user using any or all of visual, auditory, or tactile signals. The Dexcom G5® pre-set low glucose alarm at 3.1 mmol/L cannot be adjusted or turned off.

Dexcom provides web-based reports reflecting glucose trends and patterns and the reports can be shared with healthcare professional when developing patient’s diabetes management treatment plans. Dexcom Clarity is a data management software program that allows the transfer of glucose

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 49 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin data to remote servers for data management. Dexcom Clarity can transfer data from the Dexcom G4® PLATINUM (via PC upload) and the Dexcom G5® Mobile CGM System (WiFi, Bluetooth or PC upload). The web-based Dexcom Clarity software is intended for use by both home users and healthcare professionals to assist people with diabetes in the review, analysis, and evaluation of historical CGM data to support effective diabetes management. It is intended for use as an accessory to CGM devices with data interface capabilities. The Dexcom data overlay graph provides a synopsis of daily glycaemic control [65].

Dexcom G6® Continuous Glucose Monitoring System, Dexcom

The Dexcom G6® rtC-GM System is a glucose-monitoring device indicated for detecting trends and tracking patterns in persons aged ≥ 2 years with diabetes. It is intended for use at home and but can also be used in healthcare facilities. The Dexcom rtCGM system aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments, which may minimize excursions in glucose levels and increase time in the glucose target range. Interpretation of data is based on the trends and patterns seen with several sequential readings over time. The Dexcom G6® System is designed to replace SMBG testing for making treatment decisions. G6® is the first and only rt-CGM factory calibrated rt-CGM that requires no routine finger-sticks tests, and can be used in conjunction with paracetamol/acetaminophen. Dex- com G6® is also intended to autonomously communicate with digitally connected devices, including automated insulin dosing systems. Dexcom G6® can be used alone or in conjunction with these digitally connected medical devices for the purpose of managing diabetes [66].

Dexcom G6® is a glucose monitoring device consisting of three major components: sensor, transmitter, and display device (receiver or smart device) (Figure 6).

The sensor is a flexible, round, miniature wire that is placed just under the skin to read glucose levels. The Sensor is inside the applicator and can be inserted with the push of a button. The sensor attaches to the skin with its adhesive patch. The sensor comprises an applicator, a sensor probe, and a sensor pod. The disposable applicator is used to insert the sensor probe. The patient inserts the sensor probe just beneath the skin (subcutaneous tissue) of the abdominal wall at an angle using a small needle inside the applicator, which is then withdrawn. The needle includes a permselective membrane to limit the interference of paracetamol. At no time is the needle visible. The sensor probe measures glucose levels in surrounding interstitial tissue fluid. The sensor pod adheres to the patient’s abdomen to hold the transmitter in place. The sensor is left in place for up to ten days and is then replaced with a new one [66].

The Dexcom G6® automated sensor applicator is intended to simplify the insertion and result in more consistent sensor wire deployment, enhance ease of use, reduce discomfort, and to minimise the likelihood of misuse and safety related errors. The sensor wire is inside the applicator and can be inserted with the push of a button. The sensor attaches to the skin with its adhesive patch. Each applicator has a unique sensor code that should be entered at the start a new sensor session. This code is entered to the display device to use the Dexcom G6® rt-CGM without finger-stick calibrations. The patient’s sensor automatically shuts off after 10 days. The patients display device will alert them at 6 hours, 2 hours, and 30 minutes before their sensor session ends [66].

The Dexcom G6® rt-CGM transmitter and holder has a 28% lower profile than the Dexcom G5® Mobile rt-CGM. Like the Dexcom G5®, the sensor snaps into the sensor pod and sends glucose information wirelessly to the receiver or smart device. When they are properly connected, the transmitter and sensor are water resistant, so that user can shower, bathe, or swim. The same transmitter is used when the sensor is replaced. The battery life of the Dexcom G6® transmitter

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 50 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin depends on how the patient uses the system. More power is required when the user is viewing glucose information on the phone than on the receiver. The transmitter is designed to meet the specification of 3 months of battery life even when the use of power is consistently high over the life of the transmitter. From three weeks before the transmitter battery life comes to an end at a minimum of 90 days, the user is alerted to change the transmitter to avoid interruption in receiving glucose data without notification to the user [66].

Display Device

Data collected by the sensor can be processed and displayed using a receiver or a smart device running the Dexcom G6® Mobile Application. The displays on the Dexcom G6® receiver and those on the app are different in appearance but relate the same information. This information, together with trend arrows to display the direction and velocity of changes in glucose level and alarms and alerts give the patient the information to optimise the management of their diabetes.

Receiver notes: Review readings; Set and receive Alarm/Alerts; Keep within 20 feet of transmitter; Keep away from water. The receiver is not water resistant or waterproof and can get damaged if moisture gets inside.

The Dexcom G6® app is available for both Apple and Android devices. Not every Android phone is supported. There are over 30,000 Android devices and Dexcom supports the most popular hand- sets.

The alerts provided by Dexcom rtCGM are a critical and potentially lifesaving feature, allowing patients or their caregivers to take appropriate actions to correct hyper- and hypoglycaemic events. Alerts and alarm values can be customised by the user in respect of blood glucose (BG) levels and rates of glucose rise and fall. They can be changed to alert the user using any or all of visual, auditory, or tactile signals. The Dexcom G6® pre-set urgent low glucose alarm at 3.1 mmol/L cannot be adjusted or turned off. The Dexcom G6® will alert the user when their glucose levels are falling at a potentially dangerous rage. The urgent low soon alert sounds when the patient will be at 3.1 mmol/L within 20 minutes. This function provides the patient with time to take appropriate action to correct their glucose levels before a dangerous hypoglycaemic event occurs.

Dexcom Share in the Dexcom G6® app allows sensor glucose readings, trends and other glucose data to be viewed remotely on a smart device. The data can be shared with up to five people (“Followers”), as determined by the user. After being invited by the “Sharer” and by downloading the Dexcom Follow™ App, an individual becomes a “Follower”. The user determines what a Fol- lower can see, including the user’s sensor glucose readings, trends, alarm/alerts when the user’s glucose is low or high, and messages [66].

The Dexcom Share/G6® System has been developed with technology that provides a high level of security for Personally Identifiable Information (PII) and Private Health Information (PHI) of registered users [66].

Dexcom Clarity® is a data management software program that allows the transfer of glucose data from the Dexcom G6® to remote servers for data management. The cloud-based Dexcom Clarity® software is intended for use by both home users and healthcare professionals to assist people with diabetes in the review, analysis, and evaluation of historical CGM data to support effective diabetes management. The software provides summary reports, which include average glucose, frequency of calibrations, and patterns of low and high glucose. Healthcare professionals can use the retrospective information presented in Dexcom Clarity® to modify their recommendations for a patient’s diabetes management plan [66].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 51 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Display Device • See real-time glucose data and trends • Compatible on Apple and Android smart devices (For a list of compatible devices, please visit https://www.dexcom.com/ ous-compatibility-page) • Receiver with touch screen display is an optional component

Sensor & Transmitter Sensor Measures interstitial glucose levels just underneath the skin Indicated for up to 10 days of use Transmitter The transmitter sends glucose information to display devices Auto-Applicator (Dexcom G6 app and/or • Simple, auto-applicator insertion system receiver) via Bluetooth

Figure 6: Dexcom G6® CGM System Sensor, Transmitter and display device (receiver or smart device) Source: [66]

Contraindications Do not wear your CGM (sensor, transmitter, receiver, or smart device) for magnetic resonance imaging (MRI), computed tomography (CT) scan, or high-frequency electrical heat (diathermy) treatment. The Dexcom G6® has not been tested in those situations. The magnetic fields and heat could damage the components of the G6®, which may cause it to display inaccurate G6® sensor glucose readings (G6® readings) or may prevent alerts. Without G6® readings or alarm/alert notifications, you might miss a severe low or high glucose event.

The FreeStyle Navigator II® Continuous Glucose Monitoring System, Abbott

The FreeStyle Navigator II® System consists of two kits; a System Kit and a Sensor Kit. The Sys- tem Kit includes: Receiver, Transmitter, Charging Cable, A/C Wall Charger, Adapters, and Receiv- er Skin. The Receiver’s silicone skin is an optional accessory, not required for use. The Receiver is a handheld controller that wirelessly communicates with the Transmitter and displays glucose measurements (at distances up to 30 metres). It uses a rechargeable battery and should never be submerged in liquids; however, Sensor and Transmitter are water resistant. The Receiver also has a built-in FreeStyle Lite Blood Glucose Meter to confirm the continuous glucose result. When cor- rectly worn on the body along with the Sensor and Sensor Support Mount the Transmitter measures continuous glucose and communicates data to the Receiver. The Sensor Kit includes: Sensor De-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 52 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin livery Unit, Sensor Inserter, and Sensor Support mount. The Sensor Delivery Unit is the combination of 2 parts: the Sensor Support Mount and the Sensor Inserter (with pre-installed Sensor). The Sensor Delivery Unit inserts the FreeStyle Navigator II® Sensor about 5 mm under user’s skin. The Sensor Inserter is a single-use device that guides the Sensor into the skin. The Sensor Sup- port Mount is a single-use component that attaches to the skin with an adhesive pad. It is designed to hold the Transmitter and Sensor on the body for up to 5 days. The combination of the Sensor Support Mount (with Sensor) and the Transmitter that is worn on patient’s body is known as the Transmitter/Sensor Unit. The Sensor must be changed at least every 5 days. The system automatically ends the Sensor session after 5 days. Calibrating is the process the FreeStyle Nav- igator II® System uses to match interstitial fluid glucose readings with blood glucose readings. The system needs to be calibrated by checking blood glucose at approximately 1, 2, 10, 24, and 72 hours after Sensor insertion [52].

The electrochemical sensor inserted into the tissue is 5.5 mm long, 600 µm wide, and 250 µm thick [67]. It should be inserted only in the abdomen or back of the upper arm. An area of skin that stays flat during normal daily activities (no bending or creasing) should be selected for sensor placement. Site for sensor placement should be at least 2.5 cm (1 inch) away from an insulin infusion site and/or previous insertion site.

® Figure 7: Freestyle Navigator II Transmitter and Receiver (Source: [52])

The Home Screen of the device shows current continuous glucose reading and a trend arrow that indicates how fast glucose is changing and in what direction (increasing or decreasing). The device has possibility of alarm for low glucose, high glucose, and projected low and projected high glucose (early warning of an event that is likely to occur if the current trend continues). The Low Glu- cose Alarm cannot be set below 60 mg/dL (3.3 mmol/L) while the High Glucose Alarm cannot be set above 300 mg/dL (16.7 mmol/L).Therefore, the device is not intended to notify of severe hypoglycaemia or hyperglycaemia. The user has the possibility to respond to the alarm by temporarily muting or clearing the alarm. The alarms should always be used with other indications of glycaemic state such as your glucose level, trend, Timeline Graph, etc. [52].

GuardianTM Connect Continuous Glucose Monitoring system, Medtronic

The GuardianTM Connect is a, real time CGM system made up of various components that perform different functions. The EnliteTM Sensor, which is inserted under the skin using a small insertion device (serter) and taped in place for a single-use, six-days period, measures glucose levels in the fluid surrounding the cells below the skin (interstitial fluid). The sensor sends glucose readings approximately every 5 minutes to the phone app via Bluetooth. It does not replace finger-stick tests for determining insulin requirements for meals and activities. It also requires a minimum of two finger-stick calibrations against the system’s glucose meter every day. The GuardianTM Con- nect Transmitter, in conjunction with the glucose sensor, collects and wirelessly transmits interstitial glucose values to the App. The GuardianTM Connect system can store infinite days of glucose

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 53 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin sensor data. The GuardianTM Connect App displays sensor glucose data, trends, and alerts directly to the SmartPhone: 24/7 glucose monitoring and trends, as well as predictive alerts for high and low and real-time remote monitoring and SMS alerts for care partners.

CareLinkTM personal is a diabetes therapy management platform with a set of reports enabling patients to understand patterns and discuss therapy adjustment with the physician.

Figure 8: EnliteTM Sensor, GuardianTM Connect Transmitter and GuardianTM Connect App (Source: [68])

Medtronic Guardian™ Connect personal CGM standalone system features include: Alerts on situations that may require immediate attention, e.g. predictive alerts anywhere from 10 to 60 min in advance before high or low; 24/7 monitoring of glucose levels & trends, displayed directly on their mobile device; An integrated logbook enabling tracking of daily events, e.g. insulin records, carbohydrates, exercise, medication, etc.; Automated data upload to CareLink Personal; For healthcare providers’ data access: Access to CGM data enabling further insights and a more informed therapy decision; Automatic data upload to CareLink Professional (if accounts are linked); Personalized reports on patients’ glucose levels patterns; Enables care partners to remotely track patient’s glucose real time information on their own device and Optional SMS alerts for high and low glucose levels.

Contraindications EnliteTM Sensor: None known GuardianTM Connect Transmitter: Do not expose your transmitter to MRI equipment, diathermy devices, or other devices that generate strong magnetic fields. If your transmitter is inadvertently exposed to a strong magnetic field, discontinue use and contact the 24 Hour HelpLine or your local representative for further assistance. GuardianTM Connect App: None known

Eversense® Continuous Glucose Monitoring system, Senseonics, Incorporated

Eversense® CGM System includes: a small sensor inserted subcutaneously by a physician, a removable smart transmitter worn over the sensor, and a mobile app to display the glucose readings.

The Eversense® Sensor is a miniaturised fluorometer that uses fluorescent intensity to measure glucose in interstitial fluid. The sensor is implanted subcutaneously (under the skin) on the upper arm, leaving no part of the sensor protruding from the skin. The sensor remains in place and provides CGM measurements for up to 90 days. The sensor is encased in a biocompatible material and utilises a unique fluorescent, glucose indicating polymer. A light emitting diode embedded in the sensor excites the polymer, and the polymer then rapidly signals changes in glucose concen-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 54 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin tration via a change in light output. The measurement is then relayed to the smart transmitter. Measurements are completed automatically and require no action by the user. The sensor is approximately 3.5 mm x 18.3 mm and has a silicone ring that contains a small amount of dexamethasone acetate, an anti-inflammatory steroid drug. The dexamethasone acetate minimises inflammatory responses, very similarly to common medical devices such as pacemakers. The sensor requires 24 hours to stabilise within the insertion site; this period is known as the Warm-up Phase.

The removable smart transmitter is worn externally over the sensor and powers the sensor. It sends wirelessly glucose data (via Bluetooth) to the mobile device app. The smart transmitter also provides on-body vibe alerts based on the glucose settings user choose. It has a rechargeable battery and is reusable for up to one year. The smart transmitter is water-resistant to a depth of 1 metre (3.2 feet) for 30 minutes. The adhesive patch on smart transmitter should be replaced daily.

® Figure 9: Eversense Sensor, Transmitter and Smartphone App (Source: [69])

The Eversense® App is a software application that runs on a mobile device (e.g., smartphone or tablet) and displays glucose data in a variety of ways. It also provides alerts based on the glucose settings user choose. A compatible smartphone for Android (version 4.4 or higher) or Apple iPh- one® or iPod® or iPad® (iOS version 8.0 or higher) that has Bluetooth Smart (or Bluetooth Low Energy). The Eversense® App also works with the Apple Watch®. The Eversense® DMS Program is a web-based application that enables patients, caregivers, and healthcare professionals to view and analyse glucose data that has been transmitted from the Eversense® Smart Transmitter or the Eversense® CGM System mobile app.

A separate blood glucose monitoring system is required for calibrating the CGM System and to make treatment decisions. When used properly, these components work together to help ensure the user gets continuous glucose monitoring for up to 90 days. There are two calibration phases: the Initialisation Phase – after the 24-hour Warm-Up Phase, the user must complete 4 finger-stick calibration tests, spaced 2 to 12 hours apart – and daily Calibration Phase – after the Initialisation Phase, user must complete 2 finger-stick calibration tests per day, spaced 10 to 14 hours apart.

Some of the features of the Eversense® CGM System include: Wireless communication with the sensor, smart transmitter, and app; Long-term sensor wear in the upper arm for up to 90 days, Alerts when pre-set Low or High Glucose Alert levels (hypoglycaemia or hyperglycaemia) are reached; Predictive Alerts to let user know before reaching pre-set Low or High Glucose Alert levels; Use of mobile device (e.g., smartphone) to display glucose readings; On-body vibe alerts with the smart transmitter even when mobile device is not nearby; Readings provided within 2.2 - 22.2 mmol/ L range every 5 minutes; Trend arrows that show whether glucose values are rising or falling and how fast; Graphs and statistics that show glucose results in easy-to-understand formats; Remov- able and rechargeable smart transmitter; Event entry capabilities (like meals, exercise, and insulin); Storing glucose data in the app and on the smart transmitter.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 55 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Contraindications The sensor and smart transmitter are incompatible with magnetic resonance imaging (MRI) procedures. Patients should not undergo an MRI procedure while the sensor is inserted or when wearing the smart transmitter. Should an MRI be required, the sensor must be removed before the procedure. The system is contraindicated in people for whom dexamethasone or dexamethasone acetate may be contraindicated. Therapeutics products such as mannitol intravenous and irrigation solutions may increase blood mannitol concentrations and cause falsely elevated readings of sensor glucose results.

Source: [55]

Eversense® XL Continuous Glucose Monitoring (CGM) System, Senseonics, Incorporated

The system includes a small sensor inserted subcutaneously by a physician, a removable smart transmitter worn over the sensor, and a mobile app to display the glucose readings. The Ever- sense® Sensor is a miniaturised fluorometer that uses fluorescent intensity to measure glucose in interstitial fluid. The sensor is implanted subcutaneously (under the skin) on the upper arm, leaving no part of the sensor protruding from the skin. The sensor remains in place and provides CGM measurements for the operating life of the sensor. The sensor is encased in a biocompatible material and utilises a unique fluorescent, glucose indicating polymer. A light emitting diode embedded in the sensor excites the polymer, and the polymer then rapidly signals changes in glucose concentration via a change in light output. The sensor is approximately 3.5 mm × 18.3 mm and has a silicone ring that contains a small amount of dexamethasone acetate.

Measured glucose levels are calculated by the smart transmitter (37.6 mm × 48.0 mm × 8.8 mm) and sent to the app. The Eversense® XL Sensor lasts up to 180 days. The system provides with notifications through the mobile app so the user can schedule a replacement. The removable smart transmitter is worn externally over the sensor and powers the sensor. It wirelessly sends glucose data (via Bluetooth) to the mobile device app. The smart transmitter also provides on-body vibe alerts based on the glucose settings user choose. It has a rechargeable battery and is reusable for up to one year. The Eversense® XL App is a software application that runs on a mobile device (e.g. smartphone or tablet) and displays glucose data in a variety of ways. It also provides alerts based on the glucose settings user choose. Disposable adhesive patches for daily use are also included as part of the system. The patch has an acrylic adhesive side that attaches to the back of the smart transmitter and a silicone adhesive side that attaches to the skin.

® Figure 10: Eversense XL Sensor, Transmitter and a smartphone App (Source: [56])

A separate blood glucose monitoring system (not provided by Senseonics) is required for calibrating the CGM System, and to make treatment decisions.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 56 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

During the Initialisation phase, 4 finger-stick blood glucose meter tests are required. The 4 calibration tests must be spaced 2 to 12 hours apart, and all 4 tests must be completed within a 36 hour period. The Daily Calibration Phase requires 2 blood glucose meter tests at the scheduled morning and evening calibration times. Daily Calibration times must be spaced 10 to 14 hours apart.

Some of the features of the Eversense® XL CGM System: Wireless communication with the sensor, smart transmitter and app; Long-term sensor wear in the upper arm for the operating life of the sensor; Alerts when pre-set Low or High Glucose Alert levels (hypoglycaemia or hyperglycaemia) are reached; Predictive Alerts to let you know before reaching pre-set Low or High Glu- cose Alert levels; Use of mobile device (e.g., smartphone) to display glucose readings; On-body vibe alerts with the smart transmitter even when mobile device is not nearby; Provides readings within the 2.2 - 22.2 mmol/L range every 5 minutes; Trend arrows that show whether your glucose values are rising or falling and how fast; Graphs and statistics that show your glucose results in easy-to-understand formats; Removable and rechargeable smart transmitter; Event entry capabilities (like meals, exercise and insulin); Stores glucose data in the app and on the smart transmitter; Provides remote monitoring capability to others using the Eversense® NOW mobile app.

Contraindications The sensor and smart transmitter are incompatible with magnetic resonance imaging (MRI) procedures. Patients should not undergo an MRI procedure while the sensor is inserted or when wearing the smart transmitter. Should an MRI be required the sensor must be removed before the procedure. The system is contraindicated in people for whom dexamethasone or dexamethasone acetate may be contraindicated. Therapeutics products such as mannitol intravenous and irrigation solutions may increase blood mannitol concentrations and cause falsely elevated readings of sensor glucose results.

Source: [56]

Flash glucose monitoring (FGM) systems

Glucose values are reported only when the user scans the sensor by passing a reader or a cell phone close to the sensor. Flash glucose monitoring forfeits the abilities to display a continuous real-time graph of glucose versus time, rate of change of glucose, alarms, remote monitoring, and suitability for use in closed-loop systems. Flash glucose monitoring can still generate retrospective graphical displays of glucose versus date and time of day. Current flash glucose monitoring systems have good accuracy, factory calibration, a 2-week sensor lifetime, small size, light weight, ex- cellent usability, good user acceptance, low cost, and an improved method for sensor insertion [4].

Frestyle Libre®

FreeStyle Libre® system has 2 main parts: a handheld Reader and a disposable Sensor, which patients wear on their body. Patients use the Reader to wirelessly scan the Sensor and get their glucose readings.

The FreeStyle Libre® is an enzymatic amperometric 3-electrode sensor system which utilizes wired enzyme technology. This technique uses mediator molecules which are crosslinked together with the enzyme into a polymer matrix. Glucose molecules diffusing from the interstitial tissue through the outer membrane into the enzyme matrix are oxidized by the enzyme glucose oxidase. The resulting electrons are transferred from the enzyme to mediator molecules (an osmium complex) and then shuttled to the working electrode using neighbouring mediator molecules.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 57 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The disposable sensor has a thin, sterile filament (0.4 mm wide, inserted approximately 5 mm under skin) attached to a small disc (30 mm × 5 mm; approximately the size of a € 2 coin), and medical grade adhesive is used to keep the sensor in place on top of the skin once applied. The sensor is water-resistant up to 1 m depth of water for up to 30 minutes, and can therefore be worn while bathing, swimming, and exercising. The FreeStyle Libre® sensor is applied by the patient to the back of the upper arm and continuously measures glucose levels in interstitial fluid for up to 14 days. The filament draws interstitial fluid from the muscle into the sensor, where glucose levels are automatically measured every minute and stored at 15-minute intervals for 8 hours. Glucose levels can be seen at any time by scanning the reader over the sensor. Values are updated every minute. It is factory calibrated sensor, which means it does not require any additional finger-stick calibration during the 14 days wearing time.

The Reader is handheld and light weight, with a backlit colour touchscreen, it is reusable and has a rechargeable battery that must be charged every 7 days. It also has built-in blood glucose and blood ketone meters, which can be used with FreeStyle Optium blood glucose strips or Optium Beta ketone test strips to test finger-prick blood samples.

To scan the sensor, the reader is held 1cm to 4cm above the sensor for 1 second. Readings can be taken through the wearer's clothes. At each scan, the reader displays current glucose levels, levels over the previous 8 hours, and whether glucose levels are trending upwards or downwards (and how fast). For a full 24 hours of data, users must scan the sensor at least once every 8 hours.

® Figure 11: Freestyle Libre (Source: [70])

Three digital products have also been created to enhance the value of the FreeStyle Libre® platform and include two mobile medical apps (FreeStyle LibreLink and LibreLinkUp), and a cloud- based diabetes management system (LibreView). As an alternative to using the reader, the sensor can be scanned with a mobile device capable of near-field communication (NFC) on which the Libre Link companion app has been installed. It is an optional alternative to the FreeStyle Libre® reader, but the two can be used interchangeably. The Libre Link app can be used on smartphone (Android mobile devices and iPhone) and has similar features to the reader. LibreView is a free and secure cloud-based diabetes management system which enables patients and Health Care Professionals (HCPs) to generate a series of reports which reveal trends and patterns that can be used to make faster, more informed treatment decisions. LibreView can also share data in near real-time with the LibreLinkUp app, which enables authorized caregivers to remotely monitor the scan activity of their loved ones.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 58 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The FreeStyle Libre® system is designed to replace blood glucose testing in the self-management of diabetes with a few exceptions: During times of rapidly changing glucose levels, interstitial glucose levels as measured by the sensor and reported as current may not accurately reflect blood glucose levels. When glucose levels are falling rapidly, glucose readings from the sensor may be higher than blood glucose levels. Conversely, when glucose levels are rising rapidly, glucose readings from the sensor may be lower than blood glucose levels; In order to confirm hypoglycaemia or impending hypoglycaemia as reported by the sensor; If symptoms do not match the FreeStyle Libre® Flash Glucose Monitoring System reading. Users should not ignore symptoms that may be caused by low blood glucose or high blood glucose [70].

Various reports are available from the reader: individual glucose readings and user‐entered notes; daily overview of glucose readings, including how they fall within the target glucose range; average glucose readings along with four 6‐hours periods during the day; Daily patterns: indicates when glucose levels are in the target range and the variability of glucose levels; Time in target: indicates the percentage of time glucose readings are in the target range and above or below the target range; Low glucose events: indicates the number of low glucose events at four different times of the day; and Sensor usage: indicates average number of scans per day and what percentage of glucose data has been captured by these scans [71].

Contraindications The FreeStyle Libre® Flash Glucose Monitoring System must be removed prior to Magnetic Resonance Imaging (MRI).

Main features of the technologies which are within the scope of this assessment are listed in the table below.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 59 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 7: Features of the technologies

Device type CGM CGM CGM CGM CGM CGM CGM FGM Name Dexcom G4® Dexcom G5® Dexcom G6® FreeStyle Navigator II® GuardianTM Eversense® Eversense® XL FreeStyle Libre® Platinum Mobile Continues Connect Glucose Monitor Manufacturer Dexcom, Inc Dexcom, Inc. Dexcom, Inc. Abott Diabetes Care Medtronic Senseonic Senseonic Abott Diabetes Care Reference codes mmol/L mmol/L No data available MMT-7820WE No data No data MT22012 MT23284 available available Class/GMDN code System: Ilb/ Ilb/ 44611 Class Ilb 44611 GC EnliteTM No data No data 44611 44611 FreeStyle Navigator II® sensor available available FreeStyle Libre® Sensor: CGM Sensor Kit: IIb Class III/59016 FGM System Class IIb FreeStyle Navigator II® GC Transmitter (Reader kit): IIb Transmitter CGM System Kit: IIb Class IIa/44611 FreeStyle Libre® and reciever: GC App FGM System Class IIa Class IIa/44106 (Sensor kit): IIb Adjunctive Yes No No No Yes Yes Yes No Non-Adjunctive No Yes Yes Yes No No No Yes Calibration needed Yes Yes No Yes Yes Yes Yes No Sensor Duration (days) 7 7 10 5 6 90 180 14

Abbreviations: CGM - Continuous glucose monitoring system, FGM - Flash glucose monitoring system, GC - GuardianTM Connect, GMDN - Global medical device nomenclature Sources: Dexcom G4® Platinum EC Declaration of confirmity; User’s guide, G5® Mobile, Dexcom Submission file; User’s guide, GuardianTM Connect, Medtronic Submission file, Eversense® User guide, Eversense® XL User guide, , FreeStyle Libre® FGM System Declaration of confirmity; Frestyle Libre® Abott Submission file [65, 51, 72, 73, 55, 56, 70].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 60 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

COMPARATORS MEDICAL DEVICES

In this assessment, comparisons to the reference standard/self-monitoring of blood glucose (SMBG) and head-to-head comparisons were made.

Self-monitoring blood glucose

Conventional self-monitoring of blood glucose (SMBG) is performed by finger capillary blood sample, where the blood glucose is usually measured using a small handheld device. This provides a value of the blood glucose at the moment when the blood was sampled. Blood glucose meter systems consist of a meter and test strips [63, 74]. Some commercially available glucose meters are listed below.

Table 8: Commercially available glucose meters

Device Manufacturer FreeStyle Freedom Lite Abbott FreeStyle Lite Abbott AgaMatrix JAZZ AgaMatrix AgaMatrix PRESTO AgaMatrix BREEZE®2 Bayer Contour Next Bayer HemoCue Glucose 201 HemoCue SideKick Nipro TRUEresult Nipro Nova Max Nova StatStrip Xpress Nova OneTouch Ultra2 LifeScan OneTouch VerioIQ LifeScan ReliOn Micro ReliOn ReliOn Prime ReliOn Accu-Chek Aviva Plus Roche Accu-Chek Nano Roche

Source: [74]

Sensor integrated and sensor augmented (or enabled) insulin pump systems

Sensor integrated and sensor augmented (or enabled) insulin pump systems compatible (connected) with specific CGM systems (with approved indications), which are used as comparators in head-to-head analysis (if clinical studies are available) are described below.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 61 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Paradigm™Veo Pump system, Medtronic

The Paradigm™ 754 pump system is indicated for the continuous delivery of insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin. In addition, the pump system is indicated for continuous or periodic monitoring of glucose levels in the fluid under the skin, and possible low and high blood glucose episodes. The pump displays continuous glucose values and stores this data so that it can be analysed to track patterns and improve diabetes management. Pump history can be downloaded to a computer for analysis of historical glucose values. Pump therapy is not recommended for people who are unwilling or unable to perform a minimum of four blood glucose tests per day and maintain contact with their healthcare professional. Successful insulin pump therapy requires sufficient vision or hearing to allow recognition of the pump signals and alarms.

The pumps use disposable reservoirs and infusion sets for insulin delivery. The 754 pump can be used with either the 300-unit Paradigm™ reservoir (MMT-332A*) or the 176-unit reservoir, depending on insulin needs. Medtronic Diabetes provides a variety of Paradigm-compatible infusion sets to fit the user´s needs. The infusion set needs to be changed every two to three days. The pump uses only fast-acting insulin.

The pump can be used with an optional blood glucose meter powered by MWT1 technology (where or if available). MWT1 is the wireless Radio Frequency (RF) technology that is used to transmit information from the meter to the pump. The pump can be programed to automatically receive blood glucose reading from this meter. All referenced meters are blood glucose meters supported by MWT1 technology. The optional Paradigm™ remote control can be used with the pump to deliver normal boluses and suspend/resume the pump from a distant location.

The optional sensor and transmitter can provide continuous glucose measurements to help the patient to control glucose levels better. The sensor measures the glucose levels in the fluid under the skin. The Medtronic MiniLink® transmitter (MMT-7703*) is a small device that connects to the sensor. It comes with a tester (MMT-7706*) and a charger (MMT-7705* or MMT-7715*). When connected to a sensor that is inserted in the body, the transmitter automatically initializes the sensor and begins to periodically send glucose data to the pump using a radio signal. The Glucose alerts must be turned on in order for the system to send an alert when the sensor glucose measurements reach or exceed Glucose Limits.

The Enlite™ Sensor (MMT-7008*) can be inserted using the Enlite™ Serter (MMT-7510*). Two hours after the pump is used to start the sensor, the pump will alert the user to enter a meter blood glucose value. This meter blood glucose entry will represent the first calibration for the sensor. 10-15 minutes after calibration, the first sensor glucose reading can be seen on the pump screen. Sensor calibration is necessary for optimal glucose sensor performance and the sensor must be calibrated every 12 hours. If the entry is missed the pump displays an alert. Only blood glucose entries in the range of 2.2-22.2 mmol/L (40-400 mg/dL) are accepted for sensor calibration. Blood glucose values for calibration can be entered manually or through a linked meter. The pump shows an updated, continuous glucose measurement. This measurement is generated by data sent from the sensor to the transmitter and then to the pump every five minutes. The pump converts these measurements to glucose graphs.

The continuous glucose values provided by the Paradigm™ 754 pump system are not intended to be used directly for making therapy adjustments. Rather, they provide an indication that a confirmation finger-stick measurement may be required. All therapy adjustments should be based on measurements obtained using a home glucose monitor and not based on the value displayed by the pump. Blood glucose values should be tested at least four to six times per day. The recom-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 62 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin mended times to test include: overnight (occasionally, at approximately 2:00-3:00), pre-breakfast (fasting), post-breakfast (approximately two hours after eating), pre-lunch, post-lunch (approximately two hours after eating), pre-dinner, post-dinner (approximately two hours after eating), bedtime, before driving, and at any other time that patient feels his/her blood glucose is high or low. The pump can be set up to automatically receive blood glucose reading from the linked meter. This meter may not be available in all countries.

Insulin pumps deliver insulin closer to the way the human pancreas delivers insulin than any other method of treating diabetes. The user controls when and at what rate insulin is delivered. Insulin pump therapy allows setting a basal rate, or background insulin. This is delivered throughout the day and night for normal body function without food. The basal rate can be reduced or increased according to need. Insulin pump therapy allows giving a bolus, or dose of insulin on demand – when the patient eats. Meal bolus can also be increased or decreased based on the food which is to be eaten. A bolus may also be used to lower elevated blood glucose, which is called a correction bolus. Square Wave bolus delivers a bolus evenly over a period of time (30 minutes to 8 hours). This bolus can be used for insulin delivery when users have eaten a long meal or extended snack- ing. Dual Wave bolus delivers a combination of an immediate Normal bolus followed by a Square Wave bolus. The Square Wave portion is delivered evenly over a period of time. A Dual Wave bolus is useful for meals with both rapidly and slowly absorbed carbohydrates.

The Paradigm™ pump also contains an optional feature called the Bolus Wizard™ which does the math for the required bolus amount based on patient personal settings. The Bolus Wizard™ will use blood glucose reading, carbohydrate intake, and active insulin when coming up with bolus amount. The system has the option “Suspend” which stops all insulin delivery, including the current basal and any bolus or Fill Cannula deliveries that are in progress. While suspended, the pump will not deliver insulin until it is resumed. When the pump is resumed the basal delivery will continue. The pump will beep or vibrate about every 15 minutes on the hour to remind that it is not delivering insulin.

After patient receives and clears a High Glucose, Rise Rate of Change, High Predictive alert, Low Glucose, Fall Rate of Change, or Low Predictive alert, the alert will repeat until the condition that has caused the alert is resolved. The High Repeat feature and the Low Repeat feature allows the patient to set how frequently he/she wants the alert to repeat after he/she clears it the first time. The Predictive alerts calculate when Low or High Glucose Limits may be reached, and sends an alert before those limits are reached. A Predictive alert tells the user that if sensor glucose measurements keep falling or rising at the current rate, the Glucose Limit in may be reached the displayed number of minutes.

The Rate of Change alerts tell the patient when sensor glucose (SG) changes at or faster than the per-minute rate pre-selected by user. There are two alerts: fall rate for sensor glucose decrease at or faster than the pre-selected rate. The pump plays two consecutive tones, in falling pitch, if an audible beep has been selected as the alert type and rise rate for SG increases at or faster than the pre-selected rate.

The daily total screen provides a day-by-day history of the total amount of insulin for the last 32 days. The pump data management feature allows the patient and his/her healthcare professional to view and manage basal and bolus insulin delivery, food intake, blood glucose data, sensor glucose data, and Area Under the Curve data with averages. The details of each day individually can be seen or average data over a specified number of days (up to 32 days) can be shown.

The system automatically records different types of: insulin delivery and food intake information, sensor glucose information and meter blood glucose information.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 63 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The Medtronic Diabetes CareLink® USB (MMT-7305*) is used to download the Paradigm™ 754 pump data to the diabetes management software using a USB port on your computer. CareLink® Personal Software is a Web-based system designed to help the patient to manage diabetes. It copies (uploads) data from the devices: insulin pump and supported blood glucose meters; device data is stored on a centralized database; there is an online logbook where patient can record self- reported information, such as how many carbohydrates he/she consumed. Uploaded data and other information stored on the system can be viewed through several different types of treatment reports.

The pump must not be used in water and needs to be removed if planning water activities [59].

Note: * These numbers are subject to change and may therefore no longer be valid for the mentioned device or system

MiniMed™ 640G system, Medtronic

The MiniMed™ 640G system is indicated for the continuous delivery of insulin at set and variable rates for the management of diabetes mellitus in persons requiring insulin. In addition, the system is indicated for continuous or periodic monitoring of glucose levels in the fluid under the skin and detecting possible low and high glucose episodes. When using a sensor and transmitter, the pump displays continuous sensor glucose values and stores this data so that it can be analysed to track patterns and improve diabetes management. This data can be uploaded to a computer for analysis of historical glucose values.

The continuous sensor glucose values provided by the MiniMed™ 640G system are not intended to be used directly for making therapy adjustments. Rather, they provide an indication that a confirmation finger-stick measurement may be required. All therapy adjustments should be based on measurements obtained using a home blood glucose monitor and not based on the value displayed by the pump.

Pump therapy is not recommended for people who are unwilling or unable to perform a minimum of four blood glucose tests per day, people who are unwilling or unable to maintain contact with their healthcare professional, and people whose vision or hearing does not allow recognition of pump signals and alarms.

The pump is intended for use with U100 insulin. The following insulins have been tested by Med- tronic Diabetes and found to be safe for use with the MiniMed™ 640G insulin pump: Humalog™, NovoLog™, NovoRapid™. Before using different insulin with the pump, the insulin label needs to be checked to make sure the insulin can be used with the pump.

Optional devices include: compatible Bayer Glucose meter, Guardian™ 2 Link transmitter, Enlite™ glucose sensor, and CareLink™ USB. The 640G system was compatible with Bayer blood glucose meter, and now the system is also compatible with the Ascensia blood glucose meter. It wirelessly connects to the pump, allowing sending blood glucose meter readings to the pump and can be used to upload system data to diabetes management software using the USB port on the computer. The Guardian™ 2 Link transmitter and Enlite ™ glucose sensor are used with the pump for Continuous Glucose Monitoring (CGM). The Guardian™ 2 Link transmitter (MMT-7731*) is a device that connects to a glucose sensor and collects data measured by the sensor and wirelessly sends this data to monitoring devices. The Enlite ™ glucose sensor is inserted just below skin to measure glucose levels in the interstitial fluid and is a single-use device. It must be calibrated at a minimum of every 12 hours to ensure continuous receiving of sensor glucose data. The CareLink™ USB (MMT-7306*) can be used to upload system data to the diabetes management software using a USB port on the computer. The MiniLink™ transmitter (MMT-7703*) should not be used with the MiniMed™ 640G insulin pump as this device does not communicate with this insulin pump.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 64 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

At the time of manufacture and when the reservoir and tubing are properly inserted, the pump is waterproof. It is protected against the effects of being underwater to a depth of up to 3.6 meters (12 feet) for up to 24 hours. Exposure to extreme temperatures can damage the device.

Basal insulin is the “background” insulin that patient needs throughout the day and night to maintain target blood glucose values when he/she is not eating. It is the specific amount of insulin that the pump continuously delivers each hour. Basal insulin delivery settings include: Basal Pattern: a set of one or more basal rates that cover a 24-hour period; Temp Basal: a basal rate that is used in place of the patient´s scheduled basal rate for short-term situations; Preset Temp: a temporary basal rate that can be define ahead of time, and Max Basal Rate: maximum amount of basal insulin that pump can deliver per hour. The Temp Basal feature and Preset Temp feature allows to set temporary basal rates to manage blood glucose levels during short-term activities or conditions that require a basal rate different than the user’s current one, such as an illness or a change in physical activity.

A bolus is the amount of insulin taken to cover an expected rise in blood glucose, typically when person eats a meal or snack. Bolus can also be used to correct a high blood glucose reading. There are different types of bolus deliveries patients can use, depending on insulin needs at the time. There are also different ways bolus can be delivered. Available bolus types with MiniMed™ 640G include: Normal – provides a single immediate dose of insulin; Square Wave™ – delivers a single bolus evenly over an extended period of time (30 minutes to 8 hours); and Dual Wave™ – delivers a combination of an immediate Normal bolus™ followed by a Square Wave™ bolus.

There are different ways for delivering bolus: Bolus Wizard™, Manual, Preset Bolus, Easy Bolus™. Bolus Wizard enables entering the BG meter reading and carbohydrates users plan to eat, with which it then calculates an estimated bolus amount based on individual settings. The manual way means that the patient does his/her own calculation and manually enters the bolus amount. Bolus settings include: maximum amount of bolus insulin (in units) the pump can deliver in a single bolus; the amount of insulin (in units) that is increased or decreased with each button press when adjusting bolus amount (bolus increment); and the speed at which the pump delivers bolus insulin.

If user does not connect a meter to the pump, blood glucose readings must be entered manually. The History feature includes the Summary, Daily History, and Alarm History screens. The sensor glucose review and ISIG History screens are available if the sensor feature is used. The Summary screen shows details about past insulin deliveries and meter readings. Historical details for a single day or for multiple days (average of all the results for the number of days that the user selected) can be viewed. The Daily History screen displays a list of actions that users performed on the pump or event entries that were made for the selected day, such as BG meter readings, bolus deliveries, and any temp basal rates the patient has used, while the Alarm History screen displays a list of alarms and alerts that occurred on the selected day. ISIG represents a signal measured by the sensor that is used to calculate sensor glucose value. The ISIG History feature shows the history of your ISIG values over the previous 24-hour period.

MiniMed™ 640G enables specific reminders that prompt patient to check BG after a bolus, give a food bolus, check reservoir level, and change infusion set. There are also personal reminders that can be used for any purpose. If the sensor feature is turned on, the calibration reminder prompts patient to calibrate sensor.

The Sensor feature on the pump lets patients integrate and use continuous glucose monitoring (CGM). SmartGuard is a feature that can automatically stop and resume insulin delivery based on sensor glucose values and low limit. The SmartGuard setting “suspend before low” automatically suspends insulin delivery when sensor glucose is predicted to fall below a predefined low limit.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 65 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

There are several types of glucose alerts that can be set to notify the patient if glucose values are changing at a particular rate or if they are approaching or have reached a specified low or high limit. The sensor graph displays the current sensor glucose (SG) reading that is wirelessly sent to the pump by the transmitter. The sensor graph includes the following information: the most recent sensor glucose reading, historical sensor glucose readings for the last 3-hour, 6-hour, 12-hour, or 24-hour periods, high and low glucose alert limit, the bolus deliveries the patient has received during the time period shown on the graph and any suspend events that have occurred. When using a sensor, trend arrows appear on the Home screen if SG has been rising or falling faster than a certain per-minute rate. The number of arrows that appear tell the patient how quickly SG has been changing. The Alert Silence feature allows the patient to make sensor glucose alerts silent for a set period of time, for example in situations where he/she does not want to disturb others such as in a business meeting or in a movie theatre. If a glucose alert occurs when the Alert Silence feature is used, the notification light begins to flash and the Sensor alert occurred message appears.

The pump has a sophisticated safety network. If this safety network detects anything unusual, it conveys this information in the form of notifications. Notifications include alarms, alerts, and messages. An alarm warns the patient that the pump detected something that prevents insulin from being delivered. It is important that users respond to an alarm. It also covers most common or serious alarms, alerts, and messages related to the sensor glucose readings, as well as the status of the transmitter and sensor [60].

Note: * These numbers are subject to change and may therefore no longer be valid for the mentioned device or system t:slim X2™ Insulin Pump, Tandem Diabetes Care Inc.

The t:slim X2TM Insulin Pump is the smallest pump available and the only pump capable of remote feature updates. It is also the only available pump that is compatible with Dexcom G5® Mobile continuous glucose monitoring and approved to allow making treatment decisions without pricking the finger. It is up to 38% smaller than other pumps, yet it can hold a surprisingly large amount of insulin – up to 300 units. The t:slim X2 TM Insulin Pump is made up of the t:slim® Insulin Pump and the t:slim 3mL (300 units) cartridge. The t:slim X2 Insulin Pump delivers insulin in two ways: continuous, or basal insulin delivery, and bolus insulin delivery to cover carbohydrates eaten (food bolus) and to lower high blood glucose (correction bolus). The disposable cartridge is filled with up to 300 units of U-100 insulin and attached to the pump. The cartridge is replaced every few days.

The t:slim X2 TM Insulin Pump with Dexcom G5® Mobile CGM (“t:slim X2 System”) consists of the t:slim X2 TM Insulin Pump paired with the Dexcom G5® Mobile Sensor and Transmitter. The t:slim X2 TM Insulin Pump is intended for the subcutaneous delivery of insulin, at set and variable rates, for the management of diabetes mellitus in persons requiring insulin. The t:slim X2 TM Insulin Pump can be used solely for continuous insulin delivery and as part of the t:slim X2 System to receive and display continuous glucose measurements from the Dexcom G5® Mobile Sensor and Transmitter.

The t:slim X2 TM System also includes continuous glucose monitoring indicated for the management of diabetes. The Dexcom G5® Mobile CGM is designed to replace finger-stick blood glucose testing for diabetes treatment decisions. The t:slim X2 TM System aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments, which may minimize these excursions. Interpretation of the t:slim X2 TM System results should be based on the trends and patterns seen with several sequential readings over time.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 66 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The t:slim X2 TM System is indicated for use in individuals 6 years of age and greater. The t:slim X2 TM System is intended for single patient use and requires a prescription. The device is indicated for use with NovoLog or Humalog U-100 insulin.

The t:slim X2 TM System is not intended for anyone unable or unwilling to: test blood glucose (BG) levels as recommended by your healthcare provider, demonstrate adequate carbohydrate-counting skills (preferred, not required), maintain sufficient diabetes selfcare skills, see his/her healthcare provider(s) regularly. The User must also have adequate vision and/or hearing in order to recognize System alerts. The t:slim X2 TM Pump, Dexcom G5® Mobile Transmitter, and Dexcom G5® Mobile Sensor must be removed before Magnetic Resonance Imaging (MRI), Computed Tomog- raphy (CT) scan, or diathermy treatment. Exposure to MRI, CT, or diathermy treatment can damage the System. Taking medications with acetaminophen (such as Tylenol) while wearing the sensor may falsely raise your sensor glucose readings. The level of inaccuracy depends on the amount of acetaminophen active in the body and may be different for each person.

The Dexcom G5® Mobile Sensor is a disposable device that is inserted under the skin to continuously monitor glucose levels for up to 7 days. The Dexcom G5® Mobile Transmitter connects to the sensor pod and wirelessly sends readings to the pump display every 5 minutes. The display shows sensor glucose readings, trend graph, and direction and rate of change arrows. The sensor is discarded after a session of up to 7 days. The transmitter is reusable and is replaced about every 3 months. The sensor measures glucose in the fluid under the skin – not in blood – and sensor readings are not identical to readings from a blood glucose meter. The blood glucose meter is still needed to calibrate CGM on a regular basis to help ensure the accuracy of sensor glucose readings. The t:slim X2 TM Pump and Dexcom G5® Mobile Transmitter wirelessly pair together using BluetoothTM Low Energy communication. This allows the pump and transmitter to communicate securely and only with each other. During an active sensor session, CGM readings are sent to the t:slim X2 TM Pump every 5 minutes. The trend graph provides additional information that a blood glucose meter does not. It shows current glucose value, the direction it is changing and how fast it is changing. The trend graph can also show the patient what the glucose value has been over time. CGM History displays the historical log of CGM events. At least 90 days of data can be viewed in the History.

The User can create personal settings for how and when he/she wants the System to tell him/her what is happening. The High and Low Alerts tell the patient when sensor glucose readings are outside target glucose range. Rise and Fall (rate of change) Alerts let the patient know when his/her glucose levels are changing fast. The System also has a 55 mg/dL Fixed Low Alert that cannot be changed or turned off. This safety feature tells the patient that the glucose level may be dangerously low. The Out of Range Alert notifies the patient when the transmitter and pump are not communicating. The transmitter and the pump should be kept within 20 feet of each other without ob- structions. When the transmitter and the pump are too far apart, sensor glucose readings or alerts will not be displayed.

The t:slim X2 TM Insulin Pump can be used for basal and bolus insulin delivery with or without Dex- com G5® Mobile CGM. If the Dexcom G5® Mobile Sensor and Transmitter are not used, sensor glucose readings will not be sent to the pump display and you will not receive any sensor glucose alerts.

A Personal Profile is a group of settings that define basal and bolus delivery within specific time segments throughout a 24-hour period. Each profile can be personalized with a name. The following can be set within the Personal Profile: Timed Settings (Basal Rate, Correction Factor, Carb Ratio, and Target blood glucose) and Bolus Settings (Insulin Duration, Max Bolus, and Carbohy-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 67 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin drates setting (on/off)). Up to 6 different Personal Profiles can be created and up to 16 different time segments can be set in each Personal Profile, which provides more flexibility for body and lifestyle of the patient.

The t:slim X2 TM Pump offers the ability to deliver different boluses to cover carbohydrate intake (food bolus) and bring blood glucose back to target values (correction bolus). Food and correction boluses can also be programmed together. The Extended Bolus feature allows delivering part of the bolus immediately and part of the bolus slowly over a period of up to 8 hours. This can be helpful for high fat meals such as pizza or if patient has gastroparesis. A Temp Rate is used to increase or decrease (by percentage) the current basal rate for a period of time. This feature can be helpful for situations such as exercise or illness. Setting up the Quick Bolus function enables patients to deliver a bolus by simply pressing a button. It is a way to deliver a bolus by following beep/vibration commands without navigating through or viewing the pump screen. The patient can stop all insulin delivery at any time. When the user stops all insulin delivery, any active bolus and any active temp rate are immediately stopped.

The t:slim X2 TM Pump lets patients know important information about the System with Reminders, Alerts, and Alarms. Reminders are displayed to notify patient of an option that he/she has set (for example, a reminder to check blood glucose after a bolus). Alerts display automatically to notify on safety conditions that the patient needs to know (for example, an alert that insulin level is low). Alarms display automatically to notify of an actual or potential stopping of insulin delivery (for example, an alarm that the insulin cartridge is empty). User should pay special attention to Alarms [61].

Omnipod®, Insulet Corporation

The OmniPod® Insulin Management System is an innovative continuous insulin delivery system that provides all the proven benefits of continuous subcutaneous insulin infusion (CSII) therapy in a way no conventional insulin pump can. It is a simple system consisting of just 2 parts: the tubeless Pod and the handheld Personal Diabetes Manager (PDM) that the patient keeps nearby to wirelessly program insulin delivery. The Pod is a small, lightweight, self-adhesive device that is filled with insulin and worn directly on the body. The Pod delivers precise, personalized doses of insulin into the body through a small flexible tube (called a cannula) based on instructions that are programed into its wireless companion, the Personal Diabetes Manager. The cannula is inserted only once with each Pod. The Pod needs to be replaced at least once every 48-72 hours or up to 200 units of insulin (2-3 days). It is waterproof to a depth of 25 feet for up to 60 minutes (IPX8). The Personal Diabetes Manager is a wireless, handheld device that: programs the Pod with personalized insulin-delivery instructions, wirelessly monitors the Pod’s operation, and includes a Free- Style® blood glucose meter. There is no tubing connecting the Pod to the PDM. The patient wears the Pod comfortably and discreetly under the clothes. PDM can be carried separately in a back- pack, briefcase, or purse.

The user does not have to insert the infusion set manually or carry around a separate inserter. He/she simply presses a button on the PDM and the Pod’s automated insertion system safely and consistently inserts the cannula beneath the skin, virtually pain free. It then begins delivering insulin according to the programmed basal rate.

The patient needs to determine the amount of insulin that needs to be inserted into the Pod. For example, if the patient will be using the Pod for 48 hours, he/she needs enough insulin to last 48 hours. The Pod requires a minimum of 85 units of insulin to begin operation and can deliver up to 200 units of insulin. After filling the Pod with insulin, the Pod change process should be finished within 60 minutes. As a reminder that the Pod has to be filled, it will beep every 5 minutes to indicate that time is passing. If patient does not set up the Pod within 60 minutes, it must be deac-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 68 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin tivated and discarded. Before applying a new Pod, an appropriate infusion site must be selected. Due to ease of access and viewing, the abdomen is often used. The healthcare provider may suggest other potential sites that, like the abdomen, typically have a layer of fatty tissue, such as the hip, back of upper arm, upper thigh, or lower back.

With a fully integrated two-part design, there is no need to carry separate infusion sets, reservoirs, or inserters. It is all integrated into the Pod. With the PDM, patient can check blood glucose level using FreeStyle® blood glucose test strips, but without the hassle of carrying a separate blood glucose meter. If the user prefers to use another blood glucose meter, he/she can enter the readings manually into the PDM.

Another convenient part of the OmniPod® System is record keeping. The data storage system in the Personal Diabetes Manager displays up to 90 days’ worth of information. This includes blood glucose readings, basal rates and bolus doses, carbohydrates, and alarms.

The Personal Diabetes Manager can store up to 7 different basal programs. Each program can contain 24 rates, programmed in half-hour increments. A temporary basal rate lets the patient to adjust basal rate for a predetermined period of time, for example during and after physical activities. This is called a “one-time temporary basal rate”. Some temporary changes are easy to predict and respond to because they happen routinely and the patient may know from experience how they affect insulin needs (e.g. the same exercise class twice a week for a few weeks or months, a monthly hormonal change for woman). To easily handle those predictable, short-term changes, the patient can “preset” a temporary basal rate. The PDM can remember up to 7 temporary basal presets. Temporary basal rate can be set for a duration of 30 minutes to 12 hours. Once the time limit is reached, the Pod automatically returns to the active basal program.

After the patient checks blood glucose, he/she enters the carbohydrates for a snack or meal. Based on individual settings, the System displays a suggested bolus dose. The suggestion can be accepted, changed, or cancelled. To make carbohydrate counting easier, the OmniPod® System includes a reference food library which gives information about carbohydrates and other values for many standard food items. The user can also enter their own favourite foods, snacks, or entire meals as “carb presets.” Presets contain the grams of carbohydrate in the food item or meal. Next time the same food is eaten, the carbs do not have to be counted; the user can select their carb preset and the System does the calculating. The System stores up to 36 carb presets.

The OmniPod® System offers the following bolus dose options: Suggested bolus calculator – when the patient wants the System to calculate suggested bolus, based on personal settings, current blood glucose, the amount of insulin still active in the body from previous boluses, and the grams of carbohydrate he/she is about to eat; Normal bolus: when the patient needs a dose of insulin right away to cover a meal or snack or to reduce a high blood glucose level; extended bolus: when the patient is eating high-fat or high-protein foods (which take longer to digest and are slower to affect blood glucose).

In addition to automatic safety alarms, the OmniPod® System offers a number of personal settings. These features are optional and can be turned on or off at any time, except for the alerts. Those the patient can set at levels that he/she finds convenient to remind them to change the Pod. (62)

The OmniPod® Insulin Management System can be used with any available CGM for tubeless insulin delivery with unparalleled peace of mind. CGM’s provide real-time glucose readings every five minutes for people with type 1 or type 2 diabetes, and many Podders™ count on the Dexcom CGM to provide glucose readings throughout the day and night, including the speed and direction of glucose trends. With the Dexcom G5® Mobile CGM System, dynamic glucose data can be ac-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 69 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin cessed and shared safely and conveniently anywhere, anytime to your compatible smart device. If CGM shows the patient is trending high or low, blood sugar can be tested using the PDM’s built-in FreeStyle® blood glucose meter and immediately adjust insulin. (62)

Table 9: Features of the comparators medical devices

Comparators Name Self-monitoring Insulin Pump Insulin Pump; CGM Enabled Insulin Blood glucose and continuous sensor Insulin Pump Management meters interstitial glucose augmented System monitoring device pump (SAP) Proprietary Various Paradigm™ Veo MiniMed™ t:slim X2™ Omnipod® name 640G Insulin Pump Manufacturer Various Medtronic Medtronic Tandem Insulet Diabetes Corporation Care Inc. Names in other NA NA NA NA NA countries Reference NA Data not Data not Data not Data not codes available available available available Class/GMDN IIa IIb IIb II Data not code available

Abbreviations: CGM - Continuous glucose monitoring system; SAP - Sensor augmented pump; GMDN - Global medical device nomenclature Sources: [59-62]

[B0002] – What is the claimed benefit of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – in relation to the comparator(s) medical devices?

CGM and FGM medical devices allow continuous evaluation of glycaemic control, providing trends and fluctuations of interstitial glucose levels over time, so they may affect adherence to glucose self- monitoring, patient quality of life and enable proactive therapeutic interventions to maintain glycaemic control. Their potential benefit would seem particularly relevant in children, patients with poorly controlled diabetes, pregnant women, and patients with hypoglycaemia unawareness. (58)

The success of continuous glucose sensing technology is dependent on the Mean Absolute Rela- tive Difference (MARD) of the glucose measurements versus YSI (Yellow Springs Instruments) which is a global standard for accuracy in liquid measurements. This is in contrast to accuracy standards that are made versus manufacturers’ proprietary SMBG values.

MARD represents the mean of all percentage errors between CGM measurements and the reference values (SMBG or YSI), where the lower the percentage, the more accurate the sensor per- forms in the various glucose ranges. The 9% MARD of the Dexcom G5® in adults and 10% MARD in children (if the sensor is placed on the upper buttocks) is comparable to the MARD of FDA- approved and CE-marked SMBG meters. The Dexcom G5® is the only stand-alone real time CGM system with the FDA annotation of “therapeutic CGM”.

Dexcom G5® Mobile CGM and G6® are the Real Time CGM (rtCGM) system available that does not need confirmation with finger-stick readings to support treatment decisions in the full range of Blood Glucose (2.2 to 22 mmol/L). The G5® transmitter allows a patient’s glucose data to be stored in the Cloud via a compatible smart phone or similar device and offers the option of sharing this data with up to five people.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 70 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

GuardianTM Connect (real-time CGM with predictive alarms) is the only device on the market with predictive alerts for hypoglycaemia and hyperglycaemia from 10 minutes to 1 hour ahead [68].

FreeStyle Libre® has a replacement claim for SMBG. It offers a complete and continuous glycaemic picture, whereby patients can see a current glucose reading, 8-hour glucose history and a trend arrow showing the direction and rate of change of glucose levels.

FreeStyle Libre® system is clinically proven to be accurate and consistent over 14 days. Unlike other CGMs (except Dexcom G6®), FreeStyle Libre® does not require SMBG calibration, which may improve adherence to device and eliminate patient-calibration errors [75].

[B0003] – What is the phase of development and implementation of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices?

A number of different systems for real-time continuous glucose monitoring (CGM) have been available for approximately 10 years. The number of medical devices for glucose monitoring available on the market is rapidly growing in terms of the development of the new medical devices and as well as development of new generations of existing devices, which are being improved and provide more options for the user.

• The Dexcom G4® PLATINUM Continuous Glucose Monitoring System received the Conformite European (CE) mark on October 25, 2012 and the FDA approval on October 5, 2012 [65].

• The Dexcom G5® Mobile Continuous Glucose Monitoring System received the CE mark on May 15, 2015 and received approval by the FDA on August 19, 2015 [65].

• The Dexcom G6® Continuous Glucose Monitoring System received the CE mark on June 12, 2018 [66].

• GuardianTM Connect Continuous Glucose Monitoring System received the CE mark on July 16, 2016 [68].

• FreeStyle Navigator II® Continiuous Glucose Monitoring System received the CE mark on June 5, 2007 [73].

• FreeStyle Libre® received CE mark in August 2014 [70].

[B0004] – Who administers the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices and in what context and level of care are they provided?

Medical devices for CGM and FGM are usually prescribed by secondary care healthcare professionals (Diabetologist/Endocrinologist) and are used in an outpatient setting by the patients themselves or their caregivers.

For most CGM and FGM devices, a sensor can be applied with an applicator/inserter to the subcutaneous tissue on back of the upper arm, abdominal wall, or upper buttocks by the patient themselves or the caregiver. In case of the Eversense® CGM and Eversense® XL System, the sensor is inserted subcutaneously by a physician.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 71 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

To use CGM and FGM medical devices for successful diabetic management, the user and/or caregiver should maintain sufficient diabetes selfcare skills, see his/her healthcare provider regularly, and become familiar with the features and possibilities which technology allows. The user should also have adequate vision and/or hearing in order to recognize System alerts [49-57].

[B0008] – What kind of special premises are needed to use the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices?

There are no special facilities required in order to use any of the continuous glucose monitoring [65] and flash glucose monitoring medical devices [49-57].

[B0009] – What equipment and supplies are needed to use the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s) medical devices?

Equipment and supplies needed for use continuous glucose monitoring and flash glucose monitoring devices are listed below.

Equipment and supplies needed for use of the technologies

Technology Equipment/Supplies ® Dexcom G4® G4 PLATINUM Sensor (with applicator) PLATINUM G4® PLATINUM Transmitter G4® PLATINUM Receiver Receiver’s USB charging and download cable AC power wall charger Adapters for wall outlets Blood glucose meter ® Dexcom G5® G5 Mobile Sensor (with applicator) Mobile G5® Mobile Transmitter G5® Mobile Receiver (Receiver is optional as the system can run on smart device only) Receiver’s USB charging and download cable AC power wall charger Adapters for wall outlets Blood glucose meter ® Dexcom G6® G6 Sensor (with applicator) G6® reusable Transmitter G6® chargeable Receiver (optional) Receiver’s USB charging and download cable, AC power wall charger Adapters for wall outlets FreeStyle System kit ® Navigator II Freestyle Navigator Receiver Transmitter Charging cable Wall charger Adapters Receiver skin

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 72 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Technology Equipment/Supplies FreeStyle Sensor kit ® Navigator II Sensor delivery unit (continuation) Sensor insertor Sensor support mount GuardianTM Enlite Serter Connect EnliteTM Sensor Guardian Connect transmitter Two testers Charger Compatibile Smartphone device CareLink™ Connect Any internet-enabled device Blood glucose meter ® Eversense® Eversense Sensor Eversense® Transmitter A compatible smartphone for Android (version 4.4 or higher) or Apple iPhone® Stand-alone reader (optional) Charging Cradle Power Supply (USB cable and AC power adapter) Blood glucose meter Eversense® XL Sensor Transmitter A compatible smartphone for Android (version 4.4 or higher) or Apple iPhone® Mobile app Stand-alone reader (optional) Charging Cradle Power Supply (USB cable and AC power adapter) Blood glucose meter FreeStyle Reader Kit: ® Libre FreeStyle Libre Reader USB Cable Power Adapter Sensor Kit: FreeStyle Libre® Sensor Sensor Applicator Alcohol wipe Product insert FreeStyle Libre software (optional)

Source: [49-57]

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 73 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[A0021] – What is the reimbursement status of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems)?

Dexcom submission file [65]: Positive reimbursement for Real Time CGM (stand-alone or in combination with CSII): USA employers and Medicare; UK for T 1 DM adults and for clinical commissioning groups on individual funding request; Sweden in regional tenders; Norway in regional tenders; France nationally by HAS from June 2018; Germany full reimbursement since 2016-09- 06; Switzerland by national private insurances; Italy full reimbursement in Piemonte, Friuli, Blozano, Toscana, EmilioRomagna, Marche and Basalicata; limited reimbursement in Sardegna and Lazio; Positive reimbursement for (stand-alone) rt-CGM: Spain full reimbursement in 3 provinces; Belgium full reimbursement is granted; Slovenia – reimbursed for paediatrics; NL paid ibn hospital DRG only; Czech Republic – partial reimbursement includes receivers and transmitters from Oct 2016, patient co-pay between 20-40% of treatment costs; Australia – for paediatrics; Slovakia for paediatrics, with patient co-pay; South Africa – Reimbursed up to Savings Limit/patient co-pay; Israel – paediatrics.

Medtronic submission file [68]: Austria for type 1 DM; Belgium full for type 1, co-payment for type 2 DM; Czech republic for type 1 kids fully reimbursed, for adults 25% co-payment; Denmark regional funding/tender market; Finland funding/tender market; Germany national coverage for type 1 and type 2 on insulin treatment; Ireland national coverage; Italy – funding at the regional level for DM patients; NL – national coverage type 1 DM; Norway funding/tender market; Slovenia – national coverage; Spain – regional coverage; Sweden – funding/tender market; Switzerland – national coverage in the basic insurance for type 1 DM; UK – funding based on guideline recommendation.

Abbott submission file [70]: FreeStyle Libre®: France, Austria, Belgium, Luxembourg, Italia (depending on regions), UAE Qatar – Saudi Arabia, UK, Sweden, Switzerland, Denmark, Lebanon – full reimbursement for T1 and T2 DM patients on insulin; Portugal, Israel – full reimbursement for T1; Croatia reimbursement for T1 DM patients with hypoglycaemia; Finland, Germany in individual payer contract, not nationwide; Spain in some regions, especially for paediatrics; Norway for paediatrics.

Based on the submission files from three companies (Dexcom, Abbott, and Medtronic) [65, 68, 70], an overview of the reimbursement status is shown by country below. Detailed tables can be found in Appendix 2.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 74 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 10: Overview by country

National coverage National Regional Funding Funding for type 1 and coverage for funding on DRGs on Guideline type 2 diabetes type 1 diabetes recommendations patients on patients insulin treatment Germany, Belgium Switzerland, Austria, Austria, UK (with co-payment), Netherlands, Denmark, Denmark, France (strong Czech republic Italy, Spain Finland indication) (partly with co-payment)

National coverage for type 1 and type 2 diabetes patients on insulin treatment

National coverage for type 1 diabetes patients

Regional fundings

Funding on DRGs

Funding on Guideline recommendations

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 75 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

4 HEALTH PROBLEM AND CURRENT USE OF THE TECHNOLOGY (CUR)

4.1 Research questions

Element ID Research question A0002 What is diabetes mellitus (Type 1 and Type 2, gestational DM)? A0004 What is the natural course of the diabetes mellitus? A0005 What are the symptoms and the burden of diabetes mellitus for the patient? A0006 What are the consequences of the diabetes mellitus for the society? A0024 How is diabetes mellitus currently diagnosed according to published guidelines and in practice? A0025 How is diabetes mellitus currently managed according to published guidelines and in practice? A0007 What is the target population in this assessment? A0023 How many people belong to the target population? A0011 How much are the technologies – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – utilized?

4.2 Results

Overview of the disease or health condition

[A0002] – What is diabetes mellitus (Type 1 and Type 2, gestational DM)?

Diabetes mellitus (DM) refers to a group of common metabolic disorders that share the phenotype of hyperglycaemia due to relative insulin deficiency, resistance or both. Several distinct types of DM are caused by a complex interaction of genetics and environmental factors. The metabolic dysregulation associated with DM causes secondary pathophysiologic changes in multiple organ systems that impose a tremendous burden on the individual with diabetes and on the healthcare system [76, 77].

Diabetes mellitus can be classified into the following general categories [15]: 1. Type 1 diabetes (due to autoimmune b-cell destruction, usually leading to absolute insulin deficiency) 2. Type 2 diabetes (due to a progressive loss of b-cell insulin secretion frequently on the background of insulin resistance) 3. Gestational diabetes mellitus (GDM) (diabetes diagnosed in the second or third trimester of pregnancy that was not clearly overt diabetes prior to gestation) 4. Specific types of diabetes due to other causes, e.g., monogenic diabetes syndromes (such as neonatal diabetes and maturity-onset diabetes of the young [MODY]), diseases of the exocrine pancreas (such as cystic fibrosis and pancreatitis), and drug- or chemical- induced diabetes (such as with glucocorticoid use, in the treatment of HIV/AIDS, or after organ transplantation)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 76 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Type 1 diabetes and type 2 diabetes are heterogeneous diseases in which clinical presentation and disease progression may vary considerably. Classification is important for determining therapy, but some individuals cannot be clearly classified as having type 1 or type 2 diabetes at the time of diagnosis. The traditional paradigms of type 2 diabetes occurring only in adults and type 1 diabetes only in children are no longer accurate, as both diseases occur in both age-groups. Chil- dren with type 1 diabetes typically present with the hallmark symptoms of polyuria/polydipsia, and approximately one-third present with diabetic ketoacidosis (DKA). The onset of type 1 diabetes may be more variable in adults, and they may not present with the classic symptoms seen in children. Occasionally, patients with type 2 diabetes may present with DKA, particularly ethnic minorities. Although difficulties in distinguishing diabetes type may occur in all age-groups at onset, the true diagnosis becomes more obvious over time [15].

DM is classified on the basis of the pathogenic process that leads to hyperglycaemia, as opposed to earlier criteria such as age of onset or type of therapy.

Table 11: Classification of DM on the basis of the pathogenic process

Hyperglycaemia Pre-diabetes Diabetes mellitus Type of Diabetes Normal Impaired fasting Not Insulin Insulin glucose glucose or impaired insulin required required tolerance glucose tolerance requiring for control for survival Type 1 * Type 2 * Other specific types * Gestational Diabetes * Time (years) * FPG <5.6 mmol/L 5.6 - 6.9 mmol/L ≥7.0 mmol/L (100 mg/dL) (100 -125 mg/dL) (126 mg/dL) 2-h PG <7.8 mmol/L 7.8 - 11 mmol/L ≥11.1 mmol/L (140 mg/dL) (140 -199 mg/dL) (200 mg/dL) A1C <5.6% 5.7-6.4% ≥6.5%

Source [77] * Arrows indicate that changes in glucose tolerance may be bidirectional in some types of diabetes, i.e., individuals with type 2 DM may return to the impaired glucose tolerance category with weight loss; in gestational DM, diabetes may revert to impaired glucose tolerance or even normal glucose tolerance after delivery.

Type 1 Diabetes

Two subtypes of type 1 diabetes mellitus, depending on their aetiology, can be distinguished [15]: • Immune mediated diabetes • Idiopathic diabetes

Immune-Mediated Diabetes [1] was previously called “insulin dependent diabetes” or “juvenile-onset diabetes”, accounts for 5-10% of diabetes, and is due to cellular-mediated autoimmune destruction of the pancreatic beta-cells. The disease has strong HLA associations. The rate of b-cell destruction is quite variable. At the latter stage of the disease there is little or no insulin secretion, as manifested by low or undetectable levels of plasma C-peptide. Immune-mediated diabetes commonly occurs in childhood and adolescence, but it can occur at any age, even in the 8th and 9th decades of life.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 77 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Idiopathic Type 1 Diabetes [1] has no known aetiologies. These patients have permanent insulin- openia and are prone to DKA, but have no evidence of b-cell autoimmunity. Although only a minority of patients with type 1 diabetes fall into this category, of those who do, most are of African or Asian ancestry. This form of diabetes is strongly inherited and is not HLA associated. An absolute requirement for insulin replacement therapy in affected patients may be intermittent.

The American Diabetes Association defines three distinct stages of type 1 diabetes [1]: stage 1 with autoimmunity, normoglycaemia, and being presymptomatic; stage 2 with autoimmunity, dysglycaemia, and being presymptomatic, and stage 3 with new-onset hyperglycaemia and being symptomatic [1].

Table 12: Staging of type 1 diabetes

Stage 1 Stage 2 Stage 3 Characteristics Autoimmunity Autoimmunity New-onset Normoglycaemia Dysglycaemia hyperglycaemia Presymptomatic Presymptomatic Symptomatic Diagnostic criteria Multiple Multiple autoantibodies Clinical symptoms autoantibodies Dysglycaemia: IFG and/or Diabetes by No IGT or IFG IGT FPG 100-125 mg/dL standard criteria (5.6-6.9 mmol/L) 2-h PG 140-199 mg/dL (7.8-11.0 mmol/L) A1C 5.7-6.4% (39-47 mmol/mol) or ≥10% increase in A1C

Source: [78]

Type 2 Diabetes

Type 2 diabetes was previously referred to as “noninsulin-dependent diabetes” or “adult-onset diabetes”. It accounts for 90-95% of all diabetes. This form encompasses individuals who have relative insulin deficiency and have peripheral insulin resistance. There are various causes of type 2 diabetes. Although the specific aetiologies are not known, autoimmune destruction of beta-cells does not occur and patients do not have any of the other known causes of diabetes. Type 2 diabetes frequently goes undiagnosed for many years because hyperglycaemia develops gradually. Insulin secretion is defective in these patients and insufficient to compensate for insulin resistance. The risk of developing type 2 diabetes increases with age, obesity, and lack of physical activity. It occurs more frequently in women with prior GDM, in those with hypertension or dyslipidemia. The genetics of type 2 diabetes is poorly understood [1].

Gestational diabetes mellitus

Diabetes mellitus (GDM) is diabetes diagnosed in the second or third trimester of pregnancy that is not clearly either preexisting type 1 or type 2 diabetes. Women diagnosed with diabetes by standard diagnostic criteria in the first trimester should be classified as having preexisting pregesta- tional diabetes (type 2 diabetes or, very rarely, type 1 diabetes or monogenic diabetes) [79].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 78 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[A0004] – What is the natural course of diabetes mellitus?

Type 1 diabetes

In immune-mediated diabetes, the rate of b-cell destruction is quite variable, being rapid mainly in infants and children and slow mainly in adult patients. Some patients, particularly children and adolescents, may present with ketoacidosis as the first manifestation of the disease. Others have modest fasting hyperglycaemia that can rapidly change to severe hyperglycaemia and/or ketoacidosis in the presence of infection or other stress. At this latter stage of the disease, there is little or no insulin secretion, as manifested by low or undetectable levels of plasma C-peptide. The degree of hyperglycaemia (if any) may change over time, depending on the extent of the underlying disease process [80].

There are three main consequences which are result of untreated type I diabetes [81]: • increased blood glucose • increased utilization of fats for energy and for formation of cholesterol by the liver, and • depletion of the body’s proteins

The high blood glucose causes loss of glucose through urine. The level of the blood “threshold” for the appearance of glucose in the urine is considered to be blood glucose concentration above 180 mg/dL (10 mmol/L). Thus, polyuria, intracellular and extracellular dehydration, and increased thirst are classic symptoms of diabetes. Chronic high glucose concentration over long periods causes the function abnormality and structural changes of the blood vessels in multiple tissues throughout the body, resulting with inadequate blood supply to the tissues. This in turn leads to increased risk for heart attack, stroke, end-stage kidney disease, retinopathy and blindness, and ischemia and gangrene of the limbs. Many other tissues are also affected by the chronic hyperglycaemia. For example, peripheral neuropathy and autonomic nervous system dysfunction are frequent complications which can result in impaired cardiovascular reflexes, impaired bladder control, decreased sensation in the extremities, and other symptoms of peripheral nerve damage. Hypertension, secondary to renal injury, and atherosclerosis, secondary to abnormal lipid metabolism, often develop in patients with diabetes and amplify the tissue damage caused by the elevated glucose.

Type 1 diabetes causes depletion of the body’s proteins as a result of increased utilization and decreased storage of proteins and fat in attempt to compensate for inability to use glucose as a source of energy. Therefore, a person with severe untreated diabetes mellitus type 1 suffers rapid weight loss and asthenia despite polyphagia. Untreated, it can cause death within a few weeks [81].

Type 2 diabetes

Type 2 diabetes (T2D) is now understood to be a complex disorder that involves multiple organ systems. Several systems have been recognized for their role in the pathophysiology of T2D. Con- tributing pathologies include reduced glucose reuptake in skeletal muscle, increased hepatic glucose production in the liver, a reduced incretin effect in the gut, increased glucagon secretion from pancreatic α cells, increased lipolysis in fat cells, increased glucose retention by the kidneys, and even hypothalamic insulin resistance within the brain. More recently, additional work has suggested that catecholamines, vitamin D, the renin-angiotensin system, and testosterone may also impact T2D [82].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 79 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

In the long term, Type 2 diabetes mellitus increases the risk of microvascular damage (retinopathy, nephropathy, and neuropathy). It is associated with reduced life expectancy, significant morbidity due to specific diabetes related microvascular complications, increased risk of macrovascular complications (ischaemic heart disease, stroke, and peripheral vascular disease), and diminished quality of life. Additionally, Type 2 diabetes mellitus is associated with increased risk of further diseases such as cancer, psychiatric diseases, cognitive decline or chronic liver disease.

Many people with Type 2 DM have the same risk of a cardiovascular event as someone without diabetes who has already had their first heart attack; people with diabetes and a previous cardiovascular event are at very high risk – around 10 times the background population [83].

Gestational diabetes

Women with a history of GDM have a greatly increased risk of conversion to type 2 diabetes over time and not solely within the 4- to 12-week postpartum time frame [84].

In general, specific risks of uncontrolled diabetes in pregnancy include spontaneous abortion, foetal anomalies, preeclampsia, foetal demise, macrosomia, neonatal hypoglycaemia, and neonatal hyperbilirubinemia, among others. Diabetes in pregnancy may increase the risk of obesity and type 2 diabetes in offspring later in life [79, 84].

Long-term metabolic and cardiovascular implications of hyperglycaemia during pregnancy for both the mother and the child are now recognized as having major implications for public health. It has also been demonstrated that in utero exposure to either nutrient excess or deprivation affects foetal metabolic programming, resulting in an increased long-term risk of obesity, DM, and cardiovascular diseases during adult life [85].

Effects of diabetes mellitus

[A0005] – What are the symptoms and the burden of diabetes mellitus for the patient?

Children with type 1 diabetes typically present with the hallmark symptoms of polyuria/polydipsia, and approximately one-third present with diabetic ketoacidosis (DKA). The onset of type 1 diabetes may be more variable in adults, and they may not present with the classic symptoms seen in children. Occasionally, patients with type 2 diabetes may present with DKA, particularly ethnic minorities [79]. The severity of symptoms depends on the blood sugar level. Individuals with type 2 DM usually have a much more gradual onset of symptoms, and many people with type 2 DM are asymptomatic for years. [77].

Acute complications are a significant contributor to mortality, costs and poor quality of life.

Diabetic ketoacidosis (DKA) and hyperosmolar hyperglycaemic state (HHS, also known as hyperosmotic hyperglycaemic nonketotic state [HHNK]) are two of the most serious acute complications of diabetes. They are part of the spectrum of hyperglycaemia, and each represents an extreme in the spectrum [86]. DKA is more common in young (<65 years) patients, whereas hyperosmolar hyperglycaemic state (HHS) most commonly develops in individuals older than 65 years.

Diabetic ketoacidosis (DKA) is an acute, major, life-threatening complication of diabetes. This condition is a complex disordered metabolic state characterized by hyperglycaemia, ketoacidosis, and ketonuria. The most common early symptoms of DKA are the insidious increase in polydipsia and polyuria. Other signs and symptoms of DKA are malaise, generalized weakness, fatigability, nausea and vomiting; may be associated with diffuse abdominal pain, decreased appetite, and anorexia, and altered consciousness [87].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 80 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Hyperosmolar Hyperglycaemic Nonketotic Syndrome (HHNS), is a dangerous condition resulting from very high blood glucose levels. HHNS can affect both types of diabetics, yet it usually occurs amongst people with type 2 diabetes. The mortality rate in HHS can be as high as 20% which is about 10 times higher than the mortality seen in diabetic ketoacidosis [88].

Hypoglycaemia is the major limiting factor in the glycaemic management of type 1 and type 2 diabetes [89]. Symptoms of hypoglycaemia include, but are not limited to, shakiness, irritability, confusion, tachycardia, and hunger. Hypoglycaemia may be inconvenient or frightening to patients with diabetes. Severe hypoglycaemia may be recognized or unrecognized and can progress to loss of consciousness, seizure, coma, or death. Clinically significant hypoglycaemia can cause acute harm to the person with diabetes or others, especially if it causes falls, motor vehicle accidents, or other injury. Young children with type 1 diabetes and the elderly, including those with type 1 and type 2 diabetes, are noted as particularly vulnerable to clinically significant hypoglycaemia because of their reduced ability to recognize hypoglycaemic symptoms and effectively communicate their needs [89]. Hypoglycaemia mortality estimates ranging from 4 to 10 percent of deaths of patients with type 1 diabetes. Hypoglycaemia mortality rates in type 2 diabetes are currently unknown, but fatal hypoglycaemia has been documented in type 2 diabetes. Severe hypoglycaemia may also be associated with an increased risk of cardiovascular disease in patients with type 2 diabetes. Recurrent severe hypoglycaemia has been associated with cognitive impairment in young children or older persons with diabetes. Nocturnal hypoglycaemia is a particular problem, which can lead to disruption of sleep and delays in correction of the hypoglycaemia [17].

Hypoglycaemia unawareness (hypoglycaemia-associated autonomic failure) is syndrome characterized by deficient counterregulatory hormone release, especially in older adults, and a diminished autonomic response, which both are risk factors for, and caused by, hypoglycaemia. This syndrome could severely influence diabetes control and quality of life [89].

Microvascular complications of diabetes are those long-term complications that affect small blood vessels. These typically include retinopathy, nephropathy, and neuropathy. The prevalence of retinopathy increased progressively in patients with both type 1 and type 2 diabetes with increasing duration of disease [90]. The Wisconsin Epidemiologic Study of Diabetic Retinopathy is one of the most comprehensive studies documenting the natural history of retinal disease in diabetic patients. In patients with type 1 DM, 13% have retinopathy at 5 years and 90% have retinopathy after 10 to 15 years; approximately 25% will develop proliferative retinopathy after 15 years [91]. In patients with type 2 DM, 40% of patients taking insulin and 24% of patients taking oral hypoglycaemic agents will develop retinopathy at 5 years. After 15 to 19 years, the percentages increase to 84% and 53%, respectively. Proliferative retinopathy develops in 2% of patients with type 2 DM for longer than 5 years and in 25% of patients with diabetes for 25 years or longer [91].

The prevalence of DR has significantly decreased as intensive insulin therapy for the management of type 1 diabetes has become more widespread [92]. Prevalence rates for retinopathy at 8 to 10 years duration of type 1 diabetes vary between 32 and 59 percent in reports from Finland, Sweden, and a follow-up cohort from Wisconsin [93]. The severity of DR has decreased as well, with only 18 percent of retinopathy patients found to have vision-threatening levels of retinopathy at 20 years follow-up, compared with 43 percent at 20 years in the earlier Wisconsin study [94]. Similarly, in study from the United Kingdom, rates of DR and the proportion of patients with type 2 diabetes requiring laser therapy have decreased over a six-year interval [95].

The prevalence of nephropathy in diabetes has not been determined. Approximately 30% of patients with type 1 DM – and 5% to 10% of those with type 2 DM – become uremic. Diabetic nephropathy is a leading cause of end-stage renal disease [96]. With intensive glycaemic control and the use of ACE inhibitors and ARBs to control blood pressure, patients diagnosed with type 1

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 81 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin diabetes may now have lower rates of overt nephropathy and ESRD [97]. In the DCCT/EDIC type 1 diabetes cohort, less than 2 percent (10 of 711) of the intensively treated subjects developed renal insufficiency (serum creatinine >2.0 mg/dL or renal replacement therapy) over an average of 30 years of diabetes duration [92].

Prevalence of neuropathy is a function of disease duration and about 50 percent of patients with diabetes will eventually develop neuropathy [98]. The prevalence of neuropathy in patients with diabetes is 7% at 1 year, increasing to 50% at 25 years for both type 1 and type 2 DM [99, 100].

Macrovascular complications of diabetes are primarily diseases of the coronary arteries, peripheral arteries, and cerebrovasculature. Early macrovascular disease is associated with atherosclerotic plaque in the vasculature supplying blood to the heart, brain, limbs, and other organs. Late stages of macrovascular disease involve complete obstruction of these vessels, which can increase the risks of myocardial infarction (MI), stroke, claudication, and gangrene. Cardiovascular disease (CVD) is the major cause of morbidity and mortality in patients with diabetes [101]. Macrovascular complications in patients with diabetes cause an estimated two- to four-fold increased risk of coronary artery disease (CAD), peripheral arterial disease, and cerebrovascular disease. An estimated 37% to 42% of all ischemic strokes in Americans are attributable to the effects of diabetes, alone or in combination with hypertension. The prevalence of CAD or stroke in patients with diabetes is approximately 34% in both men and women. The prevalence of peripheral vascular disease in patients with diabetes aged 30 years or older is 26% [102-104].

In addition to the traditional complications described above, diabetes has been associated with increased rates of specific cancers, and increased rates of physical and cognitive disability [16].

[A0006] – What are the consequences of diabetes mellitus for the society?

Diabetes mellitus and related complications have significant impact on the population level, as a serious threat to population health, stability of healthcare systems, and with regard to significant economic burden [16].

According to WHO estimates, in 2014 there were 422 million adults aged over 18 years living with diabetes, resulting with a prevalence of 8.5% among the adult population. The number of people with diabetes has steadily risen over the past few decades, due to population growth, the increase in the average age of the population, and the rise in prevalence of diabetes at each age.

Forty percent of the worldwide increase between 1980 and 2014 (from 108 million in 1980 to 422 million adults in 2014) is estimated to result from population growth and ageing, 28% from a rise in age-specific prevalence, and 32% from the interaction of the two.

Diabetes is among the leading causes of death in The International Diabetes Federation (IDF) and Europe Region (EUR) and continues to increase in prevalence with diabetic macro- and microvascular complications resulting in increased disability and enormous healthcare costs. In 2013, the number of people with diabetes was estimated to be 56 million in EUR with an overall estimated prevalence of 8.5%. However, estimates of diabetes prevalence in 2013 vary widely in the 56 diverse countries in EUR from 2.4% in Moldova to 14.9% in Turkey. Trends in diabetes prevalence also vary between countries with stable prevalence since 2002 for many countries, but there was a doubling of diabetes prevalence in Turkey. For 2035, a further increase of nearly 10 million people with diabetes is projected for the EUR. Prevalence of type 1 has also increased over the past 20 years in EUR and there was estimated to be 129,350 cases in children aged 0-14 years in 2013. Registries provide valid information on incidence of type 1 diabetes with more complete data available for children than for adults [105].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 82 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

A total of 281 thousand men and 317 thousand women worldwide died with DM in 2011, most from CVD. The healthcare expenditure for DM in Europe was about 75 billion Euros in 2011 and is projected to increase to 90 billion by 2030 [7].

In 2012, the total burden of deaths from high blood glucose has been estimated to be up to 3.7 million, with 1.5 million deaths directly related to diabetes, and an additional 2.2 million deaths from cardiovascular diseases, chronic kidney disease, and tuberculosis related to higher-than-optimal blood glucose. Many of these deaths (43%) occurred under the age of 70.

Distinct global estimates of diabetes prevalence for type 1 and type 2 do not exist since it is often difficult to distinguish between type 1 and type 2 diabetes [16].

Prevalence of type 1 diabetes

Globally, registries of the WHO DIAMOND Project recorded large differences in the incidence and prevalence of type 1 diabetes, ranging from over 60 to under 0.5 cases annually per 100 000 children aged under 15 years; differences in case ascertainment may have contributed to the variability [16].

Prevalence of type 2 diabetes

In high-income countries, the prevalence of type 2 diabetes is frequently highest among poor people. There are few data on the income gradient of diabetes in low- and middle-income countries, but data that do exist suggest that, although the prevalence of diabetes is often highest among wealthy people, this trend is reversing in some middle-income countries. The proportion of undiagnosed type 2 diabetes varies widely – a recent review of data from seven countries found that between 24% and 62% of people with diabetes were undiagnosed and untreated [16].

Prevalence of gestational diabetes

The frequency of previously undiagnosed diabetes in pregnancy and gestational diabetes varies among populations but probably affects 10-25% of pregnancies. It has been estimated that most (75-90%) of cases of high blood glucose during pregnancy are gestational diabetes.

Economic burden

Diabetes imposes a large economic burden on the global healthcare system and the wider global economy, measured through direct medical costs, indirect costs associated with productivity loss, premature mortality and the negative impact of diabetes on nations’ gross domestic product (GDP).

Direct medical costs associated with diabetes include expenditures for preventing and treating diabetes and its complications: costs of outpatient and emergency care, inpatient hospital care, medications and medical supplies such as injection devices and self-monitoring consumables, and long-term care.

The costs of diabetes internationally range from 5% to 10% of the total healthcare spending [106]. A cost-of-illness study that covered 8 European countries estimated annual direct medical costs per patient of € 2834 and total costs of € 29 billion [83].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 83 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Current clinical management of diabetes mellitus

[A0024] – How is diabetes mellitus currently diagnosed according to published guidelines and in practice?

In individual European countries, responsible bodies and expert societies are responsible for publishing national guidelines on diabetes diagnosis.

General Diagnostic Tests for Diabetes [79]

Diabetes can be diagnosed based on plasma glucose criteria, either based on the fasting plasma glucose, 2-h PG after 75-g OGTT or haemoglobin A1C criteria.

There is incomplete concordance between A1C, FPG, and 2-h PG: 2-h PG diagnoses more people with diabetes than FPG or A1C. Marked discrepancies between measured A1C and plasma glucose levels should prompt consideration that the A1C assay may not be reliable for that individual, since a relatively small percentage of patients have conditions such as sickle cell trait or haemoglobinopathies that skew A1C results.

If patients have test results near the margins of the diagnostic threshold, the healthcare professional should follow the patient closely and repeat the test within 3 to 6 months.

General criteria for diagnosis of diabetes (Type 1 and Type 2) are listed below.

Criteria for the diagnosis of diabetes:

Criteria for the diagnosis of diabetes FPG ≥ 126 mg/dL (7.0 mmol/L). Fasting is defined as no caloric intake for at least 8 h.*

OR 2-h PG ≥ 200 mg/dL (11.1 mmol/L) during an OGTT. The test should be performed as described by the WHO, using a glucose load containing the equivalent of 75 g anhydrous glucose dissolved in water.*

OR Hemoglobin A1C ≥ 6.5% (48 mmol/mol). The test should be performed in a laboratory using a method that is NGSP certified and standardized to the DCCT assay.*

OR In a patient with classic symptoms of hyperglycaemia or hyperglycemic crisis, a random plasma glucose ≥ 200 mg/dL (11.1 mmol/L).

* In the absence of unequivocal hyperglycaemia, results should be confirmed by repeat testing.

Confirmation of diagnosis via second test is necessary unless there is a clear clinical diagnosis (e.g., patient in a hyperglycemic crisis or with classic symptoms of hyperglycaemia and a random plasma glucose ≥ 200 mg/dL [11.1 mmol/L]. It is recommended that the same test be repeated without delay using a new blood sample for confirmation. If two different tests (such as HbA1C and FPG) are both above the diagnostic threshold, this also confirms the diagnosis.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 84 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Type 1 Diabetes

In a patient with classic symptoms, measurement of plasma glucose is sufficient to diagnose diabetes (symptoms of hyperglycaemia or hyperglycemic crisis plus a random plasma glucose ≥ 200 mg/dL [11.1 mmol/L]). Also, the A1C levels can be used to determine how long a patient has had hyperglycaemia.

Summary of recommendations on type 1 diabetes diagnosis, according to the ADA, includes as follows: • Plasma blood glucose rather than haemoglobin A1C should be used to diagnose the acute onset of type 1 diabetes in individuals with symptoms of hyperglycaemia. • Screening for type 1 diabetes with a panel of autoantibodies is currently recommended only in the setting of a research trial or in first-degree family members of a proband with type 1 diabetes. • Persistence of two or more autoantibodies predicts clinical diabetes and may serve as an indication for intervention in the setting of a clinical trial.

Detailed diagnostic criteria for type 1 diabetes diagnosis are listed below.

Type 2 Diabetes

Screening for prediabetes and type 2 diabetes through an informal assessment of risk factors or with an assessment tool, such as the ADA risk test (diabetes.org/socrisktest), is recommended to guide providers on whether performing a diagnostic test is appropriate.

Summary of recommendations on type 2 diabetes diagnosis, according to the ADA, includes as follows: • Screening for type 2 diabetes with an informal assessment of risk factors or validated tools should be considered in asymptomatic adults. • Testing for type 2 diabetes in asymptomatic people should be considered in adults of any age who are overweight or obese (BMI ≥ 25 kg/m2 or ≥ 23 kg/m2 in Asian Americans) and who have one or more additional risk factors for diabetes. • For all people, testing should begin at age 45 years. • If tests are normal, repeat testing carried out at a minimum of 3-year intervals is reasonable. • To test for type 2 diabetes, fasting plasma glucose, 2-h PG after 75-g oral glucose tolerance test, and haemoglobin A1C are equally appropriate. • In patients with diabetes, other cardiovascular disease risk factors should be identified and treated. • Testing for type 2 diabetes should be considered in children and adolescents who are overweight or obese (BMI > 85th percentile for age and sex, weight for height > 85th percentile, or weight > 120% of ideal for height) and who have additional risk factors for diabetes.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 85 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Gestational Diabetes Mellitus

GDM diagnosis can be accomplished with either of two strategies: 1. “One-step” 75-g OGTT or 2. “Two-step” approach with a 50-g (non-fasting) screen followed by a 100-g OGTT for those who screen positive.

Summary of recommendations on GDM, according to the ADA, includes as follows: • Test for undiagnosed diabetes should be carried out at the first prenatal visit in those with risk factors, using standard diagnostic criteria. • Test for gestational diabetes mellitus should be carried out at 24–28 weeks of gestation in pregnant women not previously known to have diabetes. • Testing for persistent diabetes should be performed in women with gestational diabetes mellitus at 4-12 weeks postpartum, using the oral glucose tolerance test and clinically appropriate nonpregnancy diagnostic criteria. • Women with a history of gestational diabetes mellitus should have lifelong screening for the development of diabetes or prediabetes at least every 3 years. • Women with a history of gestational diabetes mellitus found to have prediabetes should receive intensive lifestyle interventions or metformin to prevent diabetes.

[A0025] – How is diabetes mellitus currently managed according to published guidelines and in practice?

People with diabetes require access to systematic, ongoing, and organized care delivered by a team of skilled healthcare providers.

The majority of care can be provided at the primary care level with basic interventions involving medication, health education and counselling, and consistent follow-up. A periodic referral for specialist care is required in order to perform more complicated interventions (for example, comprehensive eye examinations, laser and surgical treatment of eye complications, complex kidney function tests, and tests of the heart and arteries in the limbs). All cases of acute cardiovascular disease, diabetic coma, kidney failure, and infected foot ulcers should be managed in a hospital.

Patient education is an important component of diabetes management since the effectiveness of diabetes management ultimately depends on patient compliance with recommendations and treatment [16].

National guidelines and management protocols developed for (or adapted to) individual settings are useful tools in achieving a standardized and consistent management approach. They should cover these basic principles of diabetes management [16]: • Interventions to promote and support healthy lifestyles, including healthy diet, physical activity, avoidance of tobacco use, and harmful use of alcohol. • Medication for blood glucose control – insulin or oral hypo-glycaemic agents as required. • Medication to control cardiovascular disease risk. • Regular exams for early detection of complications: comprehensive eye examination, measurement of urine protein, and assessment of feet for signs of neuropathy. • Standard criteria for referral of patients from primary care to secondary or tertiary care.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 86 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

For the proper management of diabetes, achieving optimal glycaemic control while avoiding hypoglycaemia is essential. In accordance with the national standards for diabetes self-management education and support, all people with diabetes should participate in diabetes self-management education to facilitate the knowledge, skills, and ability necessary for diabetes self-care and in diabetes self-management support to assist with implementing and sustaining skills and behaviours needed for ongoing self-management, both at diagnosis and as needed thereafter [15]. In some diabetic patients, adequate glycaemic control can be achieved with weight reduction, exercise, and/ or oral glucose-lowering agents and they therefore do not require insulin. Individuals with extensive b-cell destruction and with no residual insulin secretion require insulin for survival [80].

A list of relevant guidelines from several European countries, with some key points on aspects of management of diabetes highlighted, provided by the manufacturers is provided in Appendix 1.

In 2018 American Diabetes Association provided further recommendations related to assessment of glycaemic control [89, 107]:

Recommendations Most patients using intensive insulin regimens (multiple-dose insulin or insulin pump therapy) should perform self-monitoring of blood glucose (SMBG) prior to meals and snacks, at bedtime, occasionally postprandially, prior to exercise, when they suspect low blood glucose, after treating low blood glucose until they are normoglycemic, and prior to critical tasks such as driving. B* When prescribed as part of a broad educational program, SMBG may help to guide treatment decisions and/or self-management for patients taking less frequent insulin injections B or noninsulin therapies. E* When prescribing SMBG, ensure that patients receive ongoing instruction and regular evaluation of SMBG technique, SMBG results, and their ability to use SMBG data to adjust therapy. E* When used properly, continuous glucose monitoring (CGM) in conjunction with intensive insulin regimens is a useful tool to lower A1C in adults with type 1 diabetes who are not meeting glycaemic targets. A* CGM may be a useful tool in those with hypoglycaemia unawareness and/or frequent hypoglycaemic episodes. C* Given the variable adherence to CGM, assess individual readiness for continuing CGM use prior to prescribing. E* When prescribing CGM, robust diabetes education, training, and support are required for optimal CGM implementation and ongoing use. E* People who have been successfully using CGM should have continued access after they turn 65 years of age. E*

Recommendations All children and adolescents with type 1 diabetes should self-monitor blood glucose levels multiple times daily, including premeal, prebedtime, and as needed for safety in specific clinical situations such as exercise, driving, or for symptoms of hypoglycaemia. B* Continuous glucose monitoring should be considered in children and adolescents with type 1 diabetes, whether using injections or continuous subcutaneous insulin infusion, as an additional tool to help improve glycaemic control. Benefits of continuous glucose monitoring correlate with adherence to ongoing use of the device. B* Automated insulin delivery systems improve glycaemic control and reduce hypoglycaemia in adolescents and should be considered in adolescents with type 1 diabetes. B* An A1C goal of 7.5% (58 mmol/dL) is recommended across all paediatric age-groups. E*

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 87 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Recommendations Hypoglycaemia unawareness or one or more episodes of severe hypoglycaemia should trigger reevaluation of the treatment regimen. E* Insulin-treated patients with hypoglycaemia unawareness or an episode of clinically significant hypoglycaemia should be advised to raise their glycaemic targets to strictly avoid hypoglycaemia for at least several weeks in order to partially reverse hypoglycaemia unawareness and reduce risk of future episodes. A*

* ADA evidence-grading system for “Standards of Medical Care in Diabetes” [1]

Target population

[A0007] – What is the target population of this assessment?

People diagnosed with DM type1, type 2, or gestational diabetes, who are willing and able to monitor and manage their DM themselves (see Scope).

Dexcom submission file [65]

The target population for the Dexcom rtCGM systems are people with Type 1 or Type 2 diabetes on intensive insulin regimens.

Real time CGM (using Dexcom G4® or Dexcom G5®) is particularly suitable for people with diabetes on intensive insulin regimens: • with impaired or absent awareness of hypoglycaemia; • with symptomatic hypoglycaemic events despite optimal management using SMBG or alternatives • who have a significant fear of hypoglycaemia which prevents them from managing their diabetes optimally • whose HbA1c remains high despite optimal management using SMBG or alternative glucose monitoring • who have a need for particularly tight control of BG in order to continue in their occupation (such as driving cars and operating heavy machinery) • with disabilities (such as poor vision) which make self-management of diabetes difficult • who are cared for by others (children, especially young children; some people with mental illness or special needs; elderly persons who are frail or have dementia)

Abbott submission file [70]

While FreeStyle Libre® is indicated for all diabetes population, reimbursement is typically limited to patients on intensive insulin therapy (also known as Multiple Daily Injection of insulin – MDI) and gestational diabetes. FreeStyle Libre® is intended to be used as an alternative to routine blood glucose monitoring for: • Adults and children (aged 4 or older) with type 1 or type 2 diabetes, who have multiple daily injections of insulin or who use insulin pumps and are self-managing their diabetes. • Gestational diabetes

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 88 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

FreeStyle Libre® should be available for patients either through specialist/secondary care or primary care as per local pathway approvals.

Also, real-time transmission of data should be considered in children who need alarms and in non-autonomous or isolated patients, which raises the issue of care organization geared towards telemedicine.

Medtronic submission file [68]

The target population for Guardian™ Connect personal standalone CGM is patients with type 1 diabetes of all ages in particular patients: • not reaching their glycaemic targets • with impaired hypoawareness or recurrent severe hypoglycaemia • patients with high glycaemic variability suffering from recurrent hypoglycaemia • during (pre-)pregnancy

These indications are aligned with clinical evidence, products characteristics (e.g. alarms and predictive alerts) and clinical guidelines by endocrinologists on the position of real time CGM with alarms.

[A0023] – How many people belong to the target population?

The increasing prevalence of DM worldwide has led to a situation where approximately 360 million people had DM in 2011, of whom more than 95% would have had type 2 DM (T2DM). This number is estimated to increase to 552 million by 2030 and it is thought that about half of those will be unaware of their diagnosis [7].

The WHO Global report on diabetes demonstrates that the number of adults living with diabetes has almost quadrupled since 1980 to 422 million adults [16].

In the UK in 2016, DM type 1 affected more than 370.000 adults [5] and in 2013 more than 3.2 million adults were diagnosed with DM type 2 with prevalence rates of 6.0% and 6.7% in England and Wales, respectively. About 90% of adults currently have type 2 DM [108].

According to the 2014 Norwegian Public Health report, diabetes affected an estimated 4.3% of the Norwegian population. Type 1 and 2 are the two main types, with the prevalence of type 2 account- ing for the majority (>85%) of diabetes [109].

Below the total diabetes prevalence and the proportional mortality of diabetes per country is presented. These numbers were extracted from the WHO diabetes country profiles for the EU countries and put in an excel graph [110].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 89 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

WHO Diabetes Country Profiles 2016

WHO Diabetes Country profiles 2016 [6] Sweden Slovenia Slovakia Romania Portugal Poland Norway Netherlands Malta Latvia Luxembourg Italy Hungary Croatia Greece UK France Finland Estonia Espania Denmark Germany Czech republic Cyprus Switzerland Bulgaria Belgiu Austria 0% 2% 4% 6% 8% 10% 12% Prevalence of diabetes total proportional mortality of diabetes (% of total deaths, all ages)

Figure 12: World Health Organization – Diabetes country profiles, 2016. Graph developed by authors of this assessment.

Dexcom: Estimated size of the target population (numbers stated in the submission file) [65]

The population of the European Union has been estimated to be 508 million. The prevalence of type 1 diabetes is estimated at 10-15% worldwide; because of these estimates it is hypothesized that the total population could be as high as 76.2 million. Due to the nature of type 1 diabetes, the totality of this population will be intensive insulin-using patients, and as such would be appropriate for rtCGM.

In the type 1 population, the prevalence of impaired awareness of hypoglycaemia is estimated to be 25%. If the calculations are based on this population, the total population could be as low as 19 million. It must be emphasized that in this population Flash glucose monitoring may not be appropriate due to the absence of alarms and alerts.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 90 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Abbott: Estimate the size of the target population (numbers stated in the submission file) [70] • For Sweden 88 000, 48 000 Type 1 (includes 7000 children) and 40 000 type 2 MDI. • For Germany 787 650, Type 1 and Type 2 intensive insulin users. • For Spain 116 000. • For France 300 000. • For UK 330 000 Type 1 and Type 2 intensive insulin users. • For Norway 28 000 Type 1, 8220 Type 2 MDI users, GSD 2280-4800. • For Denmark Type 1 32 000, 13 000 Type 2 MDI users, GSD 1800. • For Italy 400 000.

Medtronic: Estimate the size of the target population (numbers stated in the submission file) [68] • 1 in 3 adults cannot achieve optimal A1c levels • Impaired awareness of hypoglycaemia is observed in around a quarter of people with type 1 diabetes • About 4% of T1DM patients using basal-bolus insulin regimens are hospitalized at least once due to severe hypoglycaemia • At least 1 severe hypo is recorded in 7.1% of patients with type 1 diabetes • The percentage of T1 kids with at least 1 severe hypo event ranges from 3.0 to 5.8%

Up to 5% of women who give birth each year have either pre-existing diabetes or gestational diabetes. Of these, it is estimated that approximately 7.5% have type 1 diabetes.

[A0011] – How much are the technologies – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – utilised?

There were no results provided for utilization in the submission files of the companies. To get an overview of the utilization aspects we searched unsystematically in PubMed using the search- strategy “((((((diabetes) AND self-management) OR self-management)) AND continuous glucose monitoring)) AND ((utilization) OR utilization) (18 items)“. An unsystematic update was performed in June 2018, also including the search on flash glucose monitoring medical device.

Previous data from the T1D Exchange (T1DX), related to period 2013-2014, showed that 9% of participants were using the CGMs; new data as of early 2017 show that around 24% of patients in the T1D Exchange Clinic Registry were using the CGM [111].

Patients have strong intentions to use the technology. Using continuous subcutaneous insulin in- fusions, continuous glucose monitoring systems, and applications with patients was reported to be occasional [112] More innovations in human factors, data delivery, reporting, and interpretation are needed to foster expanded use [113] The utilization of technology is still suboptimal among patients of transition-age (pages 13-25) [114]. Adherence to diabetes regimens remains complex and often difficult to predict. Human factors, such as patient perceptions and behavioural self-regulation, are central to adherence to prescribed regimens, as well as to adoption and utilization of diabetes technology, and they will continue to be crucial as diabetes management evolves [115]. Pa-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 91 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin tients who do not understand or practice the basics of intensive insulin therapy face the greatest challenges. Those who do best watch the receiver frequently, continue with frequent home blood glucose monitoring, use the trending information to make insulin adjustments, and understand the limitations of the technology [116]. Cost-effectiveness and cost-utility analysis are needed to prior- itize and allow health management services to make the correct choices [117].

Reasons for discontinuing rt-CGM were: • problematic equipment and inaccuracy [4]; insurance issues, reimbursement [4]; lack of approval [4]; need for recalibrations [4]; time consuming uploads for data analysis [4]; inexpe- rience of physicians or other healthcare professionals[4]; lack of standardization software methods for data analysis [4]; cost; discomfort with wearing the devices, including sensors falling off; and finding the alarms disruptive [63]

Related to flash glucose monitoring, Dunn et al. in 2018 published the results of glucose testing patterns in users worldwide under real life settings with aim to establish testing frequency and association with glycaemic parameters: data between September 2014 and May 2016, comprising 50,831 readers and 279,446 sensors worldwide were analysed showing that users performed a mean of 16.3 scans/day [median (IQR): 14 (10–20)] with 86.4 million hours of readings and 63.8 million scans. Germany was the most frequent (46%), followed by Spain (11%), France (10%), United Kingdom (8.8%), Italy (8.2%), Sweden (5.2%), and the Netherlands (3.8%). Austria (2.5%) and Belgium (0.9%) were the only other countries with more than 0.5% of the readers, with the remaining 4% collected from 37 different countries around the world [118].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 92 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

5 CLINICAL EFFECTIVENESS (EFF)

5.1 Research questions

Element ID Research question D0001 What is the expected beneficial effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on mortality? D0005 How do the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) affect symptoms and findings (severity, frequency) of the diabetes mellitus? D0006 How do the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) affect progression of the diabetes mellitus? D0011 What is the effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on patients’ body functions? D0016 How does the use of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) affect activities of daily living? D0012 What is the effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on generic health-related quality of life? D0013 What is the effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on disease-specific quality of life? D0017 Were patients satisfied with the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems)?

5.2 Results

Included studies

Twelve RCTs reported clinical effectiveness outcomes in a total of 16 published scientific articles (Table 13, Appendix 1) [13, 24-38]. Only one small RCT was a head-to-head study comparing a CGM device with a FGM device [34]. Two RCTs had cross-over designs (Lind 2017, van Beers 2016) [29, 38]. Duration of monitoring, i.e. from start to follow-up examination, varied from 8 weeks to 12 months.

These RCTs were published between 2011 and 2018, with a total of 1667 randomised patients. No RCTs were identified investigating effect of use of health technologies under assessment with CE mark authorisation in pregnancy. Only one RCT exclusively reported on children (Mauras 2012) [31]. Four RCTs included patients with impaired hypoglycaemia awareness (Heinemann 2018, Ly 2013, Reddy 2018, van Beers 2016) [13, 30, 34, 38]. Sample size was calculated in all trials. Du- ration of treatment period varied from 8 weeks to 12 months.

Results are presented in categories and sequentially according to comparisons (CGM vs SMBG, FGM vs SMBG, and CGM vs FGM), clinical outcomes, type of diabetes (T1DM and/or T2DM), age

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 93 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(adults or children or mixed populations), insulin regiment (MDII, CSII, or both). Notably, trials where the insulin treatments were different between the two arms (i.e. in one arm MDII and in the other CSII) were excluded, as well as studies with monitoring durations less than 8 weeks.

All included studies had risks of bias mainly due to lack of blinding. The comparative study by Reddy et al. [34] had a particularly high risk of bias (Figure 13 and Table A3 in the Appendix 1).

Figure 13: Risk of bias of included studies at study level

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 94 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Studies comparing CGM with SMBG

Four RCTs included adults on MDII therapy, where three had patients with T1 DM (Beck 2017 T1DM; Heinemann 2018; Lind 2017) [13, 25, 29] and one had patients with T2 DM (Beck 2017 T2DM) [26]. The study by Heinemann and colleagues specifically examined patients with impaired hypoglycaemia awareness [13]. Notably, results from one RCT (DIAMOND trial) were presented in four scientific articles (Beck 2017 T1DM, Polonsky 2017, Riddlesworth 2017, Ruedy 2017) [25, 33, 35, 37]. The duration (time to follow-up examination) of the glucose monitoring in these studies varied from 24 to 26 weeks. Details can be found in Appendix 1.

Three RCTs including a mix of MDII and CSII patients with respectively 6 months (Battellino 2011) [24], 26 weeks (Mauras 2012) [31], and 12 months (Riveline 2012) [36] of follow-up used Free- Style Navigator, while one RCT, also a mixed MDII and CSII population, used the Paradigm™ Veo system solely as a monitor with a MiniLink™ transmitter and the Enlite™ glucose sensor (Van Beers 2016) [38]. The study by Van Beers et al. included patients with impaired hypoglycaemia awareness, and were followed-up for 16 weeks.

For patients exclusively on CSII, one RCT was identified, which included a mix of children <18 years (70%) and adults (30%) with impaired hypoglycaemia followed up for 24 weeks (Ly 2013) [30].

Studies comparing FGM with SMBG

Two RCTs with a 24 week monitoring period comparing Freestyle Libre® with SMBG included patients respectively with T1DM (IMPACT) [27] and T2DM (REPLACE) [28] on both MDII and CSII therapies. Subsequently, another study examined the MDII sub-population of the IMPACT trial [32].

Studies comparing CGM with FGM

Only one small head-to-head study (40 participants) was identified, which compared CGM (using Dexcom G5®) vs FGM (using Freestyle Libre®) [34]. Details can be found in Appendix 1.

Due to large heterogeneities between populations, interventions and outcomes measures, a meta-analysis was only possible for changes in HbA1c levels, which included two studies (Beck 2017 T1DM, Lind 2017) [25, 29].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 95 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 13: Main characteristics of included RCTs

Author, year, study name Study type/period Population (T1/T2)/Number of patients Intervention(s)/Control (Country or region) Registry number or follow-up randomised/analysed/ Age CGM vs SMBG Patients on MDII Beck 2017 [25] RCT T1 MDII Dexcom G4® Platinum/SMBG Diamond trial (USA) 24 weeks 158/158 NCT02282397 ≥25 y (26-73 y) Ruedy 2017 [37] RCT T1 and T2 MDII Dexcom G4® Platinum/SMBG A subset analysis of DIAMOND trial (USA) 24 weeks 116/114 NCT02282397 ≥60 y Riddlesworth 2017 [35] RCT T1 MDII Dexcom G4® Platinum/SMBG A subset analysis of DIAMOND trial (USA) 24 weeks 158/156 NCT02282397 ≥25 y (26-73 y) Polonsky 2017 [33] RCT T1 MDII Dexcom G4® Platinum/SMBG A subset analysis of DIAMOND trial (USA) 24 weeks 158/155 NCT02282397 ≥25 y (26-73 y) Lind 2017 [29] RCT T1 MDII Dexcom G4® Platinum/SMBG Gold trial (Sweden) (cross-over) 142/142 NCT0209205 26 weeks ≥18 y Beck 2017 [26] RCT T2 MDII Dexcom G4® Platinum/SMBG Part of DIAMOND trial Protocol (USA) 24 weeks 158/158 NCT02282397 ≥25 y Heinemann 2018 [13] RCT T1 MDII Dexcom G5®/SMBG HypoDE study (Germany) 26 weeks 149/149 NCT02671968 ≥18 y Impaired hypoglycaemia awareness Patients on MDII or CSII (mixed populations) Battelino 2011 [24] RCT T1 MDII or CSII CGM (FreeStyle Navigator, Abbott Diabetes Care)/SMBG (EU and Israel) 24 weeks 120/116 NCT00843609 10-65 y (stratified 10-17 y; 18-65 y)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 96 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, year, study name Study type/period Population (T1/T2)/Number of patients Intervention(s)/Control (Country or region) Registry number or follow-up randomised/analysed/ Age Mauras 2012 [31] RCT T1 MDII or CSII CGM (FreeStyle Navigator, Abbott Diabetes Care or (USA) 26 weeks 146/146 for patients on Medtronic Paradigm system – Medtronic NCT00760526 Children 4-9.9 y MiniMed™ MiniLink™ REAL-time transmiter)/SMBG Riveline 2012 [36] RCT T1 MDII or CSII CGM (FreeStyle Navigator, Abbott Diabetes Care)/SMBG (France) 12 months 62/61 NCT00726440 8-60 y Van Beers 2016 [38] RCT cross-over T1 CSII (23) and MDII (29) Paradigm™ Veo system used solely as a monitor with IN CONTROL trial (Netherlands) 16 weeks 52/52 a MiniLink™ transmitter (Medtronic), and the Enlite™ NCT01787903 18-75 y glucose sensor/CSII-treated patients continued using Impaired hypoglycaemia awareness their own pump for insulin treatment with SMBG Patients on CSII Ly 2013 [30] RCT T1 CSII Medtronic Paradigm™ Veo/CSII with SMBG (Australia) 24 weeks 95/95 (CSII + CGM + Suspend: sensor-integrated pump ACTRN12610000024044 Mixed population 4-50 y (70% children <18 y) (Medtronic Paradigm™ Veo System, Medtronic Impaired hypoglycaemia awareness MiniMed™) with automated insulin suspension n=46; CSII + SMBG: continue using their insulin pump n=49) FGM vs SMBG Patients on MDII or CSII (mixed populations) Bolinder 2016 [27] RCT T1 MDII or CSII FGM/SMBG Impact trial (EU) 24 weeks 241/239 (Abbott Freestyle Libre®) NCT02232698 ≥ 18 years Haak 2017 [28] RCT T2 MDII or CSII FGM/SMBG Replace trial (EU) 24 weeks 224/201 (Abbott Freestyle Libre®) NCT02082184 ≥ 18 years Patients on MDII Oskarsson 2018 [32] RCT T1 MDII FGM/SMBG MDII subgroup of Impact trial (EU) 24 weeks 167/161 (Abbott Freestyle Libre®) NCT02232698 ≥ 18 years CGM vs FGM Reddy 2018 [34] RCT T1 20/20 CGM/FGM (UK) 8 weeks ≥ 18 years (Dexcom G5®/Abbott Freestyle Libre®) NCT03028220 Impaired hypoglycaemia awareness

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 97 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Mortality

[D0001] – What is the expected beneficial effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on mortality?

Mortality was not specified as an outcome nor reported in any of the included RCTs.

Morbidity

[D0005] – How does the technology affect symptoms and findings (severity, frequency) of the disease or health condition? [D0006] – How do the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) affect progression of the diabetes mellitus? [D0011] – What is the effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on patients’ body functions? [D0016] – How does the use of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) affect activities of daily living?

Changes in HbA1c levels rtCGM vs SMBG

Differences in changes in HbA1c levels (mg/dL or mmol/L and %) from baseline to the end of the study are summarized below in Table 14.

Continuous glucose monitoring led to a statistically significant reduction in HbA1c levels compared with usual care (SMBG) in the majority of studies with MDII patients (Beck 2017; Beck 2017 T2DM; Lind 2017; Ruedy 2017) [25, 26, 29, 37] and the two studies on MDII and CSII patients (Battelino 2011, Riveline 2012) [24, 36]. In one study on MDII patients (Heinemann 2018) [13], two studies with MDII and CSII patients (Mauras 2012; van Beers 2016) [31, 38], and in one study on CSII patients (Ly 2013) [30] no statistically significant difference was found. The average HbA1c values at the end of follow-up were higher than 7% for all but one study. Beck et al 2017 [26] reported that 18% of people who used continuous glucose monitoring achieved an HbA1c ≤7.0%; only 4% of the usual care group reached this threshold. The certainty of the evidence for changes in HbA1c levels varied from moderate to very low, according to GRADE [44]. Details can be found in Ap- pendix 1.

Notably, four of the studies indicating no statistically significant in HbA1c levels included patients with impaired hypoglycaemia awareness (Heinemann 2018; Van Beers 2016; Ly 2013; Reddy 2018) (Table 14 and Table 16). In these studies, HbA1c change was not among the primary outcomes.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 98 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 14: Changes in HbA1c percentage and number of patients with HbA1c <7% from baseline to the end of the study/follow-up

Author, Outcome Intervention Mean Control Mean Mean (Adjusted) P-value Year (SD or 95% CI) (unless (SD or 95% CI) (unless Difference (SD or specified otherwise) specified otherwise) 95% CI) (unless n=analysed n=analysed specified otherwise) Patients on MDII Beck 2017 Changes in -1 (0.8) -0.4 (0.7) -0.60 (-0.84, -0.36) <0.001 (T1DM) HbA1c (%) n=105 n=53 Number (N) of N=18 (18% of the group) N=2 (4% of the group) 15 (0, 30) 0.01 patients with n=103 n=53 HbA1c <7% Ruedy Changes in −0.9 (0.7) −0.5 (0.7) −0.4 (0.1) <0.001 2017 HbA1c (%) n=61 n=53 Lind 2017 Changes in -0.57 (0.2) -0.1 (0.3) -0.47 (-0.53, -0.41) <0.001 HbA1c (%) n=142 n=142 Beck 2017 Changes in -0.8 (-1.0, -0.7) -0.5 (-0.7, -0.3) -0.3 (-0.5, 0.0) 0.022 (T2DM) HbA1c (%) n=79 n=79 Number (N) of 11 (14% of the group) 9 (12% of the group) 3 (-9, 14) % 0.88 patients HbA1c n=79 n=79 <7% Heinemann Changes in -0.2* -0.1* 0.03 (-0.12, 0-19) 0.66 2018 HbA1c (%) n=75 n=66 Patients on MDII or CSII (mixed population) Battelino Changes in -0.23* 0.04* -0.27 (-0.47, -0.07) 0.008 2011 HbA1c (%) n=62 n=54 Mauras Changes in -0.1 (0.6) -0.1 (0.6) 0* 0.79 2012 HbA1c (%) n=69 n=68 Riveline Changes in -0.50 (-0.70, -0.29) 0.02 (-0.18, 0.23) -0.52* 0.0006 2012 HbA1c (%) n=62 n=61 van Beers Changes in -0.1 (-0.2, 0.1) -0.1 (-0.2, 0.0) -0.1 (-0.2, 0.1) 0.449 2016 HbA1c (%) n=26 n=26 Patients on CSII Ly 2013 Changes in −0.1 (−0.3, 0.03) −0.06 (−0.2, 0.09) 0.07 (−0.2, 0.3) 0.55 HbA1c (%) n=46 n=49

* Estimation (as no mean difference of changes was reported). Abbreviations: HbA1c, glycated haemoglobin; CGM, continuous glucose monitoring; CI, confidence interval; CSII, continuous subcutaneous insulin infusion (insulin pump); MDI, multiple daily injections; n.s., non-significant; RCT, randomized controlled trial; SD, standard deviation.

Due to heterogeneities, only two studies with T1DM patients on MDII treatment could be pooled in a meta-analysis of changes in HbA1C levels (Beck 2017 T1DM and Lind 2017) [25, 29]; the pooled estimate indicated a statistically significant benefit of CGM in reducing the HbA1c level. Figure 14 The certainty of evidence was moderate following GRADE assessment [44] (Table A6). Details are described in Appendix 1.

Figure 14: Forest plot of meta-analysis comparing the changes in HbA1c from baseline to the end of the study (Beck 2017 T1DM and Lind 2017 T1DM) (Source: [21, 26])

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 99 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

FGM vs SMBG

Two RCTs, i.e. the IMPACT (on T1 DM) (Bolinder 2016) [27] and REPLACE (on T2 DM patients) (Haak 2017) [28] trials, included adult patients on either MDII or CSII treatment (mixed population). Results were published in four scientific articles (Bolinder 2016; Haak 2017; one of them was used only in safety domain, Haak 2017 at 12 months; Oskarsson 2018 as MDII subgroup of Impact trial) [27, 28, 32, 41]. No statistically significant difference in HbA1c changes between baseline and the end of the study was recorded (Table 15). Certainty of evidence was very low for this outcome (Table A9, Appendix 1).

Table 15: Changes in HbA1c percentage from baseline to study end (Bolinder 2016; Haak 2017)

Author, Year Outcome FGM SMBG Mean (Adjusted) P-value Mean (SD) Mean (SD) Difference (SE) n=analysed n=analysed Patients on MDII or CSII (mixed population) Bolinder 2016 Change in 0.15 0.17 0.00 (0.059) 0.9556 (T1 DM) (27) HbA1c (%) Baseline: 6.79 (0.52) Baseline: 6.78 (0.64) Follow-up: 6.94 (0.65) Follow-up: 6.95 (0.66) n=119 n=119 Haak 2017 Change in 0.28 0.41 0.03 (0.114) 0.8222 (T2 DM) (28) HbA1c (%) Baseline: 8.65 (1.01) Baseline: 8.75 (0.98) Follow-up: 8.37 (0.83) Follow-up: 8.34 (1.14) n=149 n=75 rtCGM vs FGM

Only one head-to-head RCT compared CGM and FGM for 8 weeks in a total of 40 adult T1DM patients on either MDII or CSII regimes (Reddy et al, 2018). There was no statistically significant difference in HbA1c change between start and the end of the study (Table 16) [34]. Certainty of evidence was very low according to GRADE assessment (Table A8, Appendix 1).

Table 16: Changes in HbA1c from baseline to the end of the study/follow-up (Reddy et al, 2018)

Author, Year Outcome rtCGM FGM Mean (Adjusted) P-value Median (95% CI) Median (95% CI) Difference (SD or n=analysed n=analysed 95% CI) Patients on MDII or CSII (mixed population) Reddy 2018 Change in −1.5 (−8.6 to −1.0) −4.5 (−5.8 to 0.0) NR 0.91 [34] HbA1c (%) n=19 n=20 Change in −0.15 (−0.8 to −0.05) −0.35 (−0.6 to 0.0) NR NR HbA1c n=19 n=20 (mmol/L)

Abbreviations: NR: not reported

Time spent in the target (normoglycemic) range rtCGM vs SMBG

Results from eight RCTs which measured the time spent in the target glycaemic range, i.e. 70- 180 mg/dL (3.9-10.0 mmol/l) are presented in Table 17. Three of these favoured CGM over SMBG with a p<0.05 (Battelino 2011;Beck 2017 T1 patients; van Beers 2016) [24, 25, 38].The certainty of the evidence for this outcome varied from low to very low. Details can be found in Table A6, Appendix 1.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 100 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 17: Time spent in the target glycaemic range 70-180 mg/dL (3.9-10.0 mmol/l)

Author, Outcome Intervention Control Difference p-value Year n=analysed n=analysed Patients on MDII Beck 2017 Minutes per day 736 (206) 650 (194) Mean adj. difference: 77 0.005 (T1DM) [25] Mean (SD) Change: +76 Change: 0 (99% CI 6 to 147) Change from baseline n=105 n=53 Ruedy [37] Minutes per day 889 (251) 732 (252) 114* <0.001 2017 Mean (SD) Change: +93 Change: -21 Change from baseline n=58 n=50 Lind 2017 % of time spent 42.28 (14.95) 39.81 (12.56) 2.47* NR [29] Mean (SD) n=123 n=125 Beck 2017 Minutes per day 882 (647-1077) 836 (551-965) 38* NR (T2DM) [26] Median (IQR) Change=+80 Change=+42 n=74 n=72 Heinemann Minutes per day Mean: NR Mean: NR Mean adj. difference: 0.0513 2018 [13] Mean (SD) Change: 11 Change: -36.8 44.9 (95% CI -0.30 to 90-0) Change from baseline n=75 n=66 Patients on MDII and CSII van Beers % of time spent 65.0 (62.8–67.3) 55.4 (53.1–57.7) Mean adj. difference: <0.0001 2016 [38] Mean (95% CI) n=26 n=26 9.6 (95% CI 8.0 to 11.2) Minutes per day 936 (906-972) 798 (762-828) Mean adj. difference: <0.0001 Mean (95% CI) n=26 n=26 138 (95% CI 114 to 162) Battelino Hours per day 17.6 (3.2) 16.0 (3.4) MD: NR 0.009 2011 [24] Mean (SD) n=62 n=54 Ratio of means: 1.10 (95% CI 1.02 to 1.18) Mauras CGM glucose values in 48 49 MD: NR 0.60 2012 [31] normal range Median n=62 n=67

* Estimation (as no mean difference of changes was reported). Abbreviations: CI, confidence interval; RCT, randomized controlled trial; SE, standard error; MD: mean difference; NR: not reported. CIs, p-values, and SD were calculated by the authors of Ontario health technology assessment, 2018 or authors of this assessment.

FGM vs SMBG

Results from two RCTs including patients with T1DM (Bolinder 2016) [27] and patients with T2 DM (Haak 2017) [28] are presented below (Table 18). While there was a statistically significant difference in patients with T1DM [27], no difference was recorded in patients with T2DM. Certainty of evidence was very low. Details can be found in Table A9, Appendix 1.

Table 18: Time spent in the target glycaemic range 70-180 mg/dL (3.9-10.0 mmol/l) [Bolinder 2016; Haak 2017]

Author, Year Outcome Intervention Control Mean (Adjusted) P-value Mean (SD) Mean (SD) Difference (SE) n=analysed n=analysed Patients on MDII or CSII (mixed population) Bolinder 2016 Hours per day Baseline: 15.0 (2.5) Baseline: 14.8 (2.8) 1.0 (0.30) 0.0006 (T1 DM) (27] Mean (SD) Follow-up: 15.8 (2.9) Follow-up: 14.6 (2.9) n=119 n=119 Haak 2017 Hours per day Baseline: 13.9 (4.5) Baseline: 13.5 (5.2) 0.2 (0.58) 0.7925 (T2 DM) (28] Mean (SD) Follow-up: 13.6 (4.6) Follow-up: 13.2 (4.9) n=149 n=75

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 101 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin rtCGM vs FGM

One small study compared CGM and FGM with a follow-up of eight weeks (Reddy et al, 2018) [34]. No statistically significant differences between the two groups in terms of change in time spent in target were observed (Table 19). The certainty of evidence was very low. Details can be found in Table A8, Appendix 1.

Table 19: Time (percentage) spent in the target glycaemic ranges [34]

Author, Year Outcome Intervention Control Mean (Adjusted) P-value Median (95% CI) Median (95% CI) Difference n=analysed n=analysed (SD or 95% CI) Patients on MDII or CSII (mixed population) Reddy 2018 % of time spent in 10.6 (3.3 to 14.4) 5.9 (−2.4 to 9.0) NR 0.15 ]34] 3.9-7.8 mmol/l per day n=19 n=20 % of time spent in 12.7 (7.2 to 15.8) 5.3 (1.1 to 11.7) NR 0.05 3.9-10 mmol/l per day n=19 n=20 % of time spent in 13.0 (−4.1 to 19.6) 4.1 (−1.0 to 11.0) NR 0.16 3.9-10 mmol/l overnight (22.00–07.00) % of time spent in 14.1 (−1.5 to 23.7) 5.2 (0.7 to 11.6) NR 0.20 3.9-10 mmol/l overnight (22.00–07.00)

Time spent in hypoglycaemic range rtCGM vs SMBG

Table 20 summarizes the findings on time spent in hypoglycaemic range.

Statistically significant results which favoured CGM compared with SMBG were identified in studies with MDII patients (Beck 2017 T1 patients; Heinemann 2018) [13, 25], a mixture of both MDII and CSII patients (Battelino 2011,van Beers 2016) [24, 38] and in CSII patients (Ly 2013) [30]. In the study by Beck 2017 [26] in T2 patients, the CGM-measured hypoglycaemic values were very low at baseline. Median time in glycaemic range below 3.89 mmol/L (70 mg/dL) was 11 minutes per day in the CGM group versus 12 minutes per day in the control group. This prevented detecting any possible effect of CGM in reducing hypoglycaemia in this RCT (Beck 2017 T2 patients) [26]. Differences were not statistically significant in the study by Mauras 2012 [31], and not reported in the study by Lind 2017 [29].The certainty of the evidence varied from low to very low. Details can be found in Table A6, Appendix 1.

Table 20: Time spent in hypoglycaemic ranges

Author, Outcome Intervention Control Difference p-value Year n=analysed n=analysed Patients on MDII Beck 2017 Minutes spent in <70 mg/dL 43 (27-69) 80 (36–111) –37* 0.002 (T1DM) [25] or 3.9 mmo/L per day n=105 n=53 Median (IQR) Minutes spent in <60 mg/dL 20 (9–30) 40 (16–68) –20* 0.002 or 3.3 mmo/L per day n=105 n=53 Median (IQR) Minutes spent in <50 mg/dL 6 (2–12) 20 (4–42) –14* 0.001 or 2.8 mmo/L per day n=105 n=53 Median (IQR) Ruedy 2017 Minutes per day (<60 mg/dL) 3 (0–15) 4 (0–24) -1* 0.11 [37] Median (IQR) n=58 n=50

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 102 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, Outcome Intervention Control Difference p-value Year n=analysed n=analysed Patients on MDII Lind 2017 % time spent in 2.79 (2.97) 4.79 (4.03) -2* NR [29] hypoglycaemia <70 mg/dL 2.2 (0.0-20.4) 3.5 (0.0-17.3) or 3.9 mmol/l n=123 n=125 Mean (SD) Median (Range) % time spent in 0.79 (1.23) 1.89 (2.12) -1.1* NR hypoglycaemia <54 mg/dL 0.4 (0.0-10.5) 1.1 (0.0-9.5) or 3.0 mmol/l n=123 n=125 Mean (SD) Median (Range) Beck 2017 Minutes spent in 4 12 -8* NR (T2DM) [26] hypoglycaemia <70 mg/dL (0-17) (0-34) or 3.9 mmol/l n=74 n=72 Median (IQR) Minutes spent in 0 0 0* NR hypoglycaemia < 50 mg/dL (0-1) (0-5) Median (IQR) Heinemann Minutes duration ≤3.9 23.9 92.2 -68.3* <0.0001 2018 [13] mmol/l per day (12.9 to 54.5) (51.8 to 172.6) Median (IQR) n=75 n=66 Minutes duration ≤3.0 3.8 (1.1 to 11.9) 32.9 -29.1* <0.0001 mmol/l per day (13.1 to 83.9) Median (IQR) % of time in ≤3.9 mmol/l 1.6% 6.4% -4.8* <0.0001 Median (IQR) % of time in ≤3.0 mmol/l 0.3% 2.5% -2.2* <0.0001 Median (IQR) Patients on MDII and CSII van Beers Hours per day in 1.6 2.7 MD: <0.0001 2016 [38] hypoglycaemia ≤ 3.9 mmol/L (1.3 to 2.0) (2.4 to 3.1) −1.1 (−1.4 to −0.8) Mean (95% CI) n=26 n=26 Battelino Hours per day in 0.48±0.57 0.97±1.55 Difference NR 0.03 2011 [24] hypoglycaemia <63 mg/dL 0.26 (0.14-0.54) 0.54 (0.23-1.31) Ratio of means 0.49 Mean (SD) n=62 n=54 (95% CI 0.26-0.76) Median (IQR) Mauras CGM glucose values (mg/dl) 1.5 2.1 NR 0.78 2012 [31] % ≤70 mg/dL n=62 n=67 Median CGM glucose values (mg/dl) 0.4 0.6 NR 0.31 % ≤60 mg/dL Median Patients on CSII Ly 2013 [30] % of hours spent in Daya: 1.5 (0.9-3.7) Daya:3.3 (1.6-5.9) Daya: −1.8 Daya: 0.01 hypoglycaemia <60 mg/dL Nighta:2.4(0.4-5.3) Nighta:6.2 (4.2-9.9) Nighta: −3.8 Nighta: Median (IQR) n=46 n=49 <0.001 a Day, 6 a.m. to 10 p.m.; night, 10 p.m. to 6 a.m. Abbreviations: CI, confidence interval; IQR, interquartile range.

FGM vs SMBG [27, 28]

The study by Bolinder et al. included T1DM patients, and reported statistically significant reduction in mean time in hypoglycaemia between by a mean adjusted difference of −1.24 (SE 0.239; p<0.0001) between intervention and control, corresponding to a 38% reduction in time in hypoglycaemia in the FGM group compared with SMBG [27].

The between-group differences for time in hypoglycaemia defined as sensor glucose lower than 3.1 mmol/L, 2.5 mmol/L, and 2.2 mmol/L were statistically significant in favour of FGM. The number of hypoglycaemic events registered at each hypoglycaemic threshold was statistically signifi-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 103 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin cant reduced. Analysis by day and night showed that time below all hypoglycaemic thresholds and number of episodes were statistically significantly improved with FGM. The between-group differences for area under the curve (h × mg/dL) (AUC) were also statistically significant in favour of FGM. At 6 months, 77 (65%) of the intervention group compared with 39 (33%) of the control group reduced their time in hypoglycaemia (<3.9 mmol/L) by at least 30% (p<0.0001). Time spent in hypoglycaemia was reduced almost immediately as sensor-based results became visible to participants (Table 21). Certainty of evidence was nevertheless very low using the GRADE tool. De- tails can be found in Table A9, Appendix 1.

In patients with T2DM, similar results were provided by Haak et al. 2017 [28]. Statistically significant reductions in all sensor measures of time spent in hypoglycaemia, number of events, and AUC were observed for intervention participants as compared with control. Time in hypoglycaemia <3.9 mmol/L (70 mg/dL) was reduced by a mean 0.47 ± SE 0.13 h/day (p=0.0006), while time in <3.1 mmol/L (55 mg/dL) was reduced by a mean 0.22 ± SE 0.07 h/day (p=0.0014), which corresponded to reductions of 43% and 53%, respectively. Nocturnal (23.00–06.00 h) hypoglycaemia <3.9 mmol/L (70mg/dL) was reduced by 54% (-0.29 ± 0.08 h per 7 h), while daytime (06.00– 23.00 h) hypoglycaemia <3.9 mmol/L (70 mg/dL) was reduced by 31% (-0.16 ± 0.08 h per 17 h) with p=0.0374.

The frequency of events with glucose <3.9 mmol/L (70 mg/dL) was reduced by 28% (-0.16 ± 0.065 per day mean ± SE) with FGM (p=0.0164). Events <3.1 mmol/L (55 mg/dL) were reduced by 44% (-0.12 ± 0.037) with FGM (p=0.0017). Frequency of events <2.5 mmol/L (45 mg/dL) was reduced by 49% (-0.06 ± 0.02) with p=0.0098.

For participants below 65 years of age, time in hypoglycaemia <3.9 mmol/L (70 mg/dL) was reduced by 35% with FGM (-0.37 ± 0.168 h/day, p=0.0279) with a 40% reduction in the area under the curve (p=0.0305), but there was no difference in the number of hypoglycaemic events. For participants of 65 years or more, time in hypoglycaemia <3.9 mmol/L (70 mg/dL) was reduced by 56% using FGM (-0.60 ± 0.220, p = 0.0083) with a 71% reduction in the area under the curve (p=0.0061). No difference was detected in number of events (p=0.0513). Certainty of evidence was nevertheless very low. Details can be found in Appendix 1.

Table 21: Time spent in hypoglycaemia

Author, Outcome Difference in adjusted Percentage P-value Year means (SE) difference (%) Patients on MDII or CSII (mixed population) Glucose <3.9 mmol/L (70 mg/dL) within 24 h Bolinder 2016 (T1 DM) [27] Number of events −0.45 (0.089) −25.8 <0.0001 n=119/119 Time (hours) −1.24 (0.239) −38.0 <0.0001 AUC (h × mg/dL) –25.14 (5.32) –46.7 <0.0001 Haak 2017 (T2 DM) [28] Number of events −0.16 (0.065) −27.7 0.0164 n=149/75 Time (hours) −0.47 (0.134) −43.1 0.0006 AUC (h × mg/dL) –7.80 (2.20) –51.1 0.0005 Glucose <3.9 mmol/L (70 mg/dL) at night (23.00–06.00 h) within 7 h Bolinder 2016 (T1 DM) [27] Number of events −0.14 (0.029) −33.2 <0.0001 Time (hours) −0.47 (0.118) −39.8 <0.0001 Haak 2017 (T2 DM) [28] Number of events −0.12 (0.03) −44.9 0.0003 Time (hours) −0.29 (0.08) −54.3 0.0001

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 104 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, Outcome Difference in adjusted Percentage P-value Year means (SE) difference (%) Patients on MDII or CSII (mixed population) Glucose <3.1 mmol/L (55 mg/dL) within 24 h Bolinder 2016 (T1 DM) [27] Number of events −0.38 (0.074) −41.3 <0.0001 Time (hours) −0.82 (0.175) −50.3 <0.0001 AUC (h × mg/dL) −9.67 (2.29) −56.1 <0.0001 Haak 2017 Number of events −0.12 (0.037) −44.3 0.0017 (T2 DM) [28] Time (hours) −0.22 (0.068) −53.1 0.0014 AUC (h × mg/dL) −2.51 (0.76) −60.3 0.0012 Glucose <3.1 mmol/L (55 mg/dL) at night (23.00–06.00 h) within 7 h Bolinder 2016 (T1 DM) [27] Number of events −0.11 (0.03) −34.9 0.0005 Time (hours) −0.32 (0.07) −48.9 <0.0001 Haak 2017 (T2 DM) [28] Number of events −0.07 (0.02) −53.0 0.0012 Time (hours) −0.12 (0.04) −58.1 0.0032 Glucose <2.5 mmol/L (45 mg/dL) within 24 h* Bolinder 2016 (T1 DM) [27] Number of events −0.26 (0.06) −48.5 <0.0001 Time (hours) −0.55 (0.14) −59.5 <0.0001 AUC (h × mg/dL) –2.88 (0.75) –63.1 0.0002 Haak 2017 (T2 DM) [28] Number of events −0.06 (0.02) −48.8 0.0098 Time (hours) −0.14 (0.04) −64.1 0.0013 AUC (h × mg/dL) –0.70 (0.22) –66.7 0.0015 Glucose <2.5 mmol/L (45 mg/dL) at night (23.00–06.00 h) within 7 h* Bolinder 2016 (T1 DM) [27] Number of events −0.09 (0.02) −44.9 <0.0001 Time (hours) −0.25 (0.06) −60.4 <0.0001 Haak 2017 (T2 DM) [28] Number of events −0.04 (0.02) −57.8 0.0086 Time (hours) −0.08 (0.03) −68.3 0.0041 Glucose <2.2 mmol/L (40 mg/dL) within 24 h Bolinder 2016 (T1 DM) [27] Number of events −0.22 (0.050) −55.0 <0.0001 Time (hours) −0.46 (0.122) −65.3 0.0003 Haak 2017 (T2 DM) [28] Number of events −0.05 (0.02) −52.6 0.0199 Time (hours) −0.10 (0.03) −66.7 0.0020

* Post-hoc endpoint

In the study by Oskarsson et al. 2018 [32], time in hypoglycaemia was statistically significant reduced in the MDII subgroup of patients with T1DM from the IMPACT trial. At 6 months, mean time in hypoglycaemia was reduced by 46.0%, i.e. from 3.44 h/day to 1.86 h/day in the intervention group (baseline adjusted mean change was −1.65 h/day), and from 3.73 h/day to 3.66 h/day in the control group (baseline adjusted mean change, 0.00 h/day), with a between-group difference of −1.65 (95% CI −2.21 to −1.09 and p<0.0001).

rtCGM vs FGM

Reddy et al, 2018 [34] showed that CGM more effectively reduces time spent in hypoglycaemia in people with T1DM and impaired awareness of hypoglycaemia, compared with FGM. Time spent in hypoglycaemia at all thresholds was statistically significantly different between groups, favouring CGM (Table 22). However, this study only included 40 patients and the certainty of evidence was very low. Details can be found in Table A8, Appendix 1.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 105 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 22: Percentage time within defined glucose range (hypoglycaemia)

Author, Outcome rtCGM FGM P-value Year Median (95% CI) Median (95% CI) Change from baseline Change from baseline Patients on MDII or CSII (mixed population) Reddy 2018 [34] % of time spent in <2.8 mmol/l 10.6 (3.3 to 14.4) 5.9 (−2.4 to 9.0) 0.003 n=19/20 per day % of time spent in <3.3 mmol/l 12.7 (7.2 to 15.8) 5.3 (1.1 to 11.7) 0.006 per day % of time spent in <3.5 mmol/l 13.0 (−4.1 to 19.6) 4.1 (−1.0 to 11.0) 0.004 per day % of time spent in <3.9 mmol/l 14.1 (−1.5 to 23.7) 5.2 (0.7 to 11.6) 0.01 per day % of time spent in <2.8 mmol/l −1.2 (−4.3 to −0.5) 1.3 (−1.0 to 2.4) 0.003 overnight* % of time spent in <3.3 mmol/l −3.0 (−5.0 to −0.3) 1.3 (−1.4 to 3.6) 0.006 overnight* % of time spent in <3.5 mmol/l −2.8 (−4.7 to −0.3) 2.0 (−1.0 to 4.7) 0.004 overnight* % of time spent in <3.9 mmol/l −2.7 (−6.1 to −0.1) 0.6 (−2.1 to 5.4) 0.01 overnight*

* Overnight (22.00–07.00)

Time spent in hyperglycaemic range

RCTs which measured time spent in hyperglycaemic range reported a statistically significant reduction using CGM in patients with T1DM in the study by Beck et al. 2017 [25] but not in the study by Heinemman et al. [13]. Lind et al. 2017 and Beck et al. 2017 T2DM indicated reduced time as well, but did not report any p-value [26, 29] (Table 23).

Table 23: Time spent in hyperglycaemic range

Author, Year Outcome Intervention Control Difference p-value Patients on MDII Beck 2017 [25] Minutes per day in >180 mg/dL 638 (503-807) 740 (625-854) −102 0.03 n=105/53 Median (IQR) Minutes per day in >250 mg/dL 223 (128-351) 347 (241-429) −124 <0.001 Median (IQR) Ruedy 2017 [37] Minutes per day in >250 mg/dL 89 (37-208) 179 (83-316) -90 0.006 n=61/53 Median (IQR) Lind 2017 [29] % of time spent in hyperglycaemia 18.48 (12.28) 21.92 (12.39) -3.44 NR n=123/125 > 180 mg/dl or 10.0 mmol/l 16.8 (1.3-55.3) 18.7 (2.1-59.6) Mean (SD) Median (range) % of time spent in hyperglycaemia 44.90 (15.68) 46.98 (14.01) -2.08 NR > 250 mg/dl or 13.9 mmol/l 46.1 (9.0-81.5) 46.0 (19.4-81.0) Mean (SD) Median (range) Beck 2017 (T2 Minutes per day in hyperglycaemia 549 571 -22 NR DM) [26] > 180 mg/dl or 10.0 mmol/l (353-789) (422-883) n=79/79 Median (range) Minutes per day in hyperglycaemia 105 118 -13 NR > 250 mg/dl or 13.9 mmol/l (37-246) (48-288) Median (IQR) Heinemann Mean minutes per day in hyper- 558.6 (268.4) 509.1 (219.1) -18.7 (95% CI 0.47 2018 [13] glycaemia > 180 mg/dl 10.0 mmol/l -70.3 to 32.9) n=75/66 % of values > 180 mg/dl or 10.0 mmol/l 38.8 (18.7) 35.3 (15.2) 1.3 (95% CI 0.47 Mean (SD) -2.3 to 4.9)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 106 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

FGM vs SMBG [27, 28]

In patients with T1DM, Time spent in hyperglycaemia >13.3 mmol/L (240 mg/dL) was statistically significantly reduced more in the intervention group than in the control group (Table 24) [27]. How- ever, there was no effect on time spent with sensor glucose concentrations higher than 10.0 mmol/L.

Similar results were provided by Haak et al 2017 for T2DM patients [28]. There was no difference in time in hyperglycaemia > 10.0 mmol/L (180 mg/dL) or > 13.3 mmol/L (240 mg/dL) between the two groups (Table 24).

Table 24: Time spent in hyperglycaemia, FGM patients vs SMBG Bolinder et al. 2016, Haak et al. 2017

Author, Outcome Difference in adjusted Percentage P-value Year means (SE) difference (%) Patients on MDII or CSII (mixed population) Glucose >13.3 mmol/L (240 mg/dL) within 24 h Bolinder 2016 T1 DM [27] Time (hours) −0.37 (0.163) −19.1 0.0247 n=119/119 Haak 2017 T2 DM [28] Time (hours) 0.1 (0.46) 2.1 0.8729 n=149/75

rtCGM vs FGM [34]

Reddy et al. 2018 reported no statistically significant differences in change in time spent above hyperglycaemic thresholds between the two arms (Table 25 [34]

Table 25: Time (percentage) spent in hyperglycaemia Reddy et al.2018

Author, Outcome rtCGM FGM P-value Year Median (95% CI) Median (95% CI) Change from baseline Change from baseline Patients on MDII or CSII (mixed population) Reddy et % of time spent in −8.6 (−13.0 to−1.1) −7.0 (−16.9 to 1.7) 0.71 al.2018 [34] >10.0 mmol/L mmol/l per day n=19/20 % of time spent in −4.9 (−8.6 to −0.7) −3.1 (−5.3 to −0.4) 0.48 >15 mmol/l per day % of time spent in 10.6 (3.3 to 14.4) 5.9 (−2.4 to 9.0) 0.15 3.9-7.8 mmol/l per day % of time spent in −4.4 (−15.4 to 9.5) −9.9 (−15.7 to −4.3) 0.36 >10.0 mmol/l overnight* % of time spent in −4.1 (−6.1 to 0.0) −2.9 (−6.1 to −1.4) 0.70 >15 mmol/l overnight*

* Overnight (22.00–07.00)

Number of hypoglycaemia events and severe hypoglycaemia events rtCGM vs SMBG

Table 26 summarizes the results on hypoglycaemia and severe hypoglycaemia events for rtCGM vs SMBG patients. There were huge variations in how hypoglycaemia was reported between studies. Different measures of hypoglycaemia and severe hypoglycaemia were used, including AUC of rate of patients with blood glucose <50 mg/dL per day, mean number of hypoglycaemic episodes <4.0 mmol/L per day, mean number of hypoglycaemic episodes ≤ 3.1 mmol/L per 4

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 107 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin weeks, AUC of rate of patients having blood glucose <70 mg/dL per day and severe hypoglycaemia requiring the assistance of another person, annualized rate, severe hypoglycaemia, including seizure or coma, 6-month rate per 100 patient-months (change from baseline), and number of severe hypoglycaemic events (Appendix 1).There was a statistically significant difference in hypoglycaemic events between the continuous glucose monitoring groups and the control groups presented in the Riddlesworth 2017 [35] and Heinemann 2018 studies in MDII patients [13], Ly 2013 on CSII patients [30] and the van Beers 2016 study [38] on MDII and CSII patients, but not in Battelino 2011, Mauras 2012, and Riveline 2012 [24, 31, 36]. The certainty of the evidence for this outcome varied from low to very low. The certainty of the evidence for severe hypoglycaemic events also varied from low to very low. Details can be found in Table A6, Appendix 1.

Table 26: Results for hypoglycaemia and severe hypoglycaemia (adults unless specified otherwise)

Author, Year Outcome Intervention Control Difference p-value Patients on MDII Riddlesworth Number of hypoglycaemic BL: 3 (2,7) BL: 4 (1,7) NR NR 2017 [35] events per 2 weeks* FU: 2 (1,4) FU: 4 (1 to 6) n=103/53 Median (IQR) Number of hypoglycaemic BL: 10 (10%) BL: 6 (11%) NR NR events per 2 weeks* FU: 25 (24%) FU: 9 (17%) Hypoglycaemic event rate BL: 0.23 (0.15, 0.46) BL: 0.31 (0.08, NR NR per 24 h* vs 0.16 (0.07, 0.30) 0.54) FU: 0.30 Median (IQR) (0.09, 0.46) Change in hypoglycaemic -0.08 (-0.23, 0.07) -0.00 (-0.19, 0.10) -0.08 0.03 event rate per 24 h* Median (IQR) Heinemann Number of hypoglycaemic BL: 10·8 (10·0) BL: 14·4 (12·4) Mean adj. <0.0001 2018 [13] events per 28 days FU: to 3·5 (4·7) FU: 13·7 (11·6) difference 0.28 n=75/66 Mean (SD) (95% CI 0.20-0.39) Incidence of decreased by 72% NR IRR 0·28, (95% <0.0001 hypoglycaemic events CI 0·20-0·39) Mean number of from 10.4 (SD 9.6) (13.2 [11.4] to IRR of 0.27 <0.0001 hypoglycaemic events to 3.4 (4.5) 13.2 [10.9]) (95% CI 0.20-0.38) Mean (SD) Number of nocturnal reduced from increased from Mean adj. <0.0001 hypoglycaemic events 2.3 (2.4) to 1.0 (1.0) 2.4 (2.6) to 2.7 (2.8) difference 0.35 per 28 days Change = -1.3 Change = 0.3 (95% CI 0.22-0.56) Mean (SD) Patients on MDII and CSII (mixed) van Beers Number of severe 14 34 -20 0.033 2016 [38] hypoglycaemic events n=26/26 Patients with more 10 (19%) 18 (35%) OR=0.48 0.062 than one severe (95% CI 0.2-1.0) hypoglycaemic event Battelino The number of 0.28 ± 0.54 0.37 ± 0.40 Ratio of means 0.07 2011 [24] hypoglycaemic excursions 0.76 n=62/54 <55 mg/dL per day** (95% CI 0.47-1.43) Mean (SD) The number of 0.53 ± 0.60 0.76 ± 0.94 Ratio of means 0.08 hypoglycaemic excursions 0.70 <63 mg/dL per day** (95% CI 0.43-1.03) Mean (SD) Mauras Subjects with at least 1 3 5 NR 0.49 2012 [31] event severe n=69/68 hypoglycaemic events Incidence rate (per 100 8.6 17.6 NR 0.80 person-years) Riveline Subjects with at least 15 (24.2%) 6 (9.8%) NR 0.9962 2012 [33] 1 event severe n=62/61 hypoglycaemic events

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 108 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, Year Outcome Intervention Control Difference p-value Patients on CSII Ly 2013 [30] Severe hypoglycaemia, BL: 1.8 (0.6 to 4.3) BL: 2.1 (0.8 to 4.6) NR NR n=46/49 including seizure or coma, FU: 0 (0 to 2.4) to 2.2 (0.5 to 6.5) 6-month rate per 100 patient-months (95% CI) Number of severe BL: 5 (46) BL: 6 (49) NR NR hypoglycaemic events (total FU: 0 (41) FU: 6 (45) number of patients) Incidence rate difference NR NR 1.5 (0.3 to 2.7) 0.02 from baseline to end point (95% CI) Moderate hypoglycaemia BL: 128.1 BL: 20 (13.2-29.1) NR NR 6-month rate per 100 (109.8-148.7) FU: 9.6 (5.1-16.5) patient-months (95% CI) FU: 28.5 (19.8-39.6) Number of moderate 173 (45) to 35 (41) 27 (45) to 13 (45) NR NR hypoglycaemic events (total no. of patients) Incidence rate ratio per 100 9.6 (5.1-18.1) 26.3 (15.4-45.0) 2.7 (1.2-6.1) 0.01 patient-months (95% CI) Sum of severe and moderate BL: 129.6 BL: 20.7 (13.8-30) NR NR hypoglycaemia 6-month rate (111.1-150.3) FU: 1.9 (6.8-19.3) per 100 patient-months FU: 28.4 (19.8-39.6) (95% CI) Sum of severe and moderate 175 (45) vs 35 (41) 28 (45) vs 13 (45) NR NR hypoglycaemia: number of events (no. of patients) Incidence rate per 100 9.5 (5.2-17.4) 34.2 (22.0-53.3) 3.6 <0.001 patient-months (95% CI) (95% CI, 1.7-7.5) Ly 2013 [30] Sum of severe and moderate BL: 302.2 BL: 42.2 NR NR n=46/49 hypoglycaemia 6-month rate (253.6-357.5) (25.4-65.9) per 100 patient-months Sensitivity FU: 64.4 (43.2-92.6) FU: 17.8 (7.7-35.0) (95% CI) analysis for patients Sum of severe and moderate BL: 136 BL: 19 younger than hypoglycaemia: Number of FU: 29 FU: 8 12 years events (total No. of patients) Sum of severe and moderate Adj. incidence ratio: <0.001 hypoglycaemia rate 5.5 (2.0-15.7) favouring the low-glucose suspension group

Abbreviations: CI, confidence interval; BL: baseline; FU: follow-up; NR: not reported * Hypoglycaemia events lasting at least 20 min at less than 3.0 mmol/L (54 mg/dL) separated by at least 15 min. ** An excursion was defined as all consecutive recordings outside the boundary covering at least 10 min. The duration of an excursion was defined as the elapsed time from first excursion to the first reading indicating return inside the excursion boundary./

FGM vs SMBG

Bolinder et al 2016, in T1 DM patients [27], reported that the number of hypoglycaemic events registered at each hypoglycaemic threshold was significantly reduced. Similar results were provided by Haak et al 2017, for T2 DM patients [28]: significant reduction in number of events was observed for intervention participants compared with controls. The frequency of events with glucose <3.9 mmol/L (70 mg/dL) was reduced by 28% (-0.16 ± 0.065 per day mean ± SE) for intervention participants compared with controls (p=0.0164). Events <3.1 mmol/L (55 mg/dL) were reduced by 44% (-0.12 ± 0.037) for intervention participants compared with controls (p= 0.0017). Frequency of events <2.5 mmol/L (45 mg/dL) was reduced by 49% (-0.06 ± 0.02) for intervention participants compared with controls (p=0.0098). For the prespecified subgroup aged less than 65 years, no difference in the number of events was found. For participants aged 65 years or more,

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 109 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin no difference was observed in the number of events (p=0.0513) either. These results were already presented in Tables above. Certainty of evidence for both studies was very low. Details can be found in Table A9, Appendix 1. rtCGM vs FGM

Results from one head-to-head trial comparing rtCGM vs FGM after for 8 weeks (Reddy et al, 2018) were already presented above, showing that CGM more effectively reduced time spent in hypoglycaemia in people with T1 DM and impaired awareness of hypoglycaemia, compared with FGM [34]. Certainty of evidence was very low. Details can be found in Table A8, Appendix 1.

Health-related quality of life

[D0012] – What is the effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on generic health-related quality of life? [D0013] – What is the effect of the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) on disease-specific quality of life?

Satisfaction

[D0017] – Were patients satisfied with the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems)?

QoL and User Satisfaction rtCGM vs SMBG

Of the studies that reported patient-reported outcome, quality of life (QoL) measures and user satisfaction, most used well-known validated questionnaires: Diabetes Distress Scale (DDS), Diabetes Quality of Life (DQoL), and Diabetes Treatment Satisfaction (DTSQs and DTSQc). Results are summarized below in Tables. Results were still inconsistent across studies, probably due to differences in types of outcomes or survey tools. The certainty of the evidence for these outcomes varied from low to very low. Details can be found in Table A6, Appendix 1.

Beck et al 2017 T1 patients [25]: In the CGM group, satisfaction with use of CGM was high, as indicated by the mean (SD) score of 4.2 (0.4) on the CGM Satisfaction Survey, with mean (SD) scores of 4.2 (0.5) on the benefits subscale and 4.3 (0.5) on the subscale for lack of hassles.

Ruedy et al 2017 [37]: In the CGM group, satisfaction with use of CGM was high as indicated by the mean score of 4.2 ± 0.4 on the CGM Satisfaction Survey (possible score range 1 to 5), with mean scores of 4.3 ± 0.5 on the Benefits subscale and 1.8 ± 0.5 on the Hassles subscale, indicating that perceived benefits were high while perceived hassles were few.

Polonsky et al 2017 [33]: presented data from the DIAMOND randomized controlled trial, related to QoL. CGM contributed to significant improvement in diabetes-specific QOL (i.e. diabetes distress, hypoglycaemic confidence) in adults with T1D, but not with QOL measures not specific to diabetes (i.e. wellbeing, health status). CGM satisfaction was associated with most of the QOL outcomes but not with glycaemic outcomes. The CGM group had statistically significant increased hypoglycaemic confidence (p=0.01) and a greater decrease in diabetes distress (p=0.01) than the

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 110 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

SMBG group. No statistically significant differences in wellbeing, health status, or hypoglycaemic fear were observed. CGM satisfaction was not significantly associated with glycaemic changes but was associated with reductions in diabetes distress (p<0.001), hypoglycaemic fear (p=0.02), and increases in hypoglycaemic confidence (p<0.001) and well-being (p=0.01).

Lind et al 2017 [29]: Results of the pre-specified analyses of patient-reported outcomes of wellbeing and diabetes treatment satisfaction shown that overall well-being, estimated with the WHO-5 questionnaire, statistically significantly improved during CGM use (66.1 vs 62.7; p=0.02). Treatment satisfaction was statistically significantly higher during CGM use as measured by the Diabetes Treatment Satisfaction Questionnaire status version (30.21 vs 26.62; p<0.001) and also for the change version (13.20 vs 5.97; p<0.001). The Hypoglycaemia Confidence Questionnaire scale showed statistically significantly less hypoglycaemia fear in favour of CGM (3.40 vs 3.27; p<0.001).

Heinemann 2018 [13]: The diabetes distress total score was reduced in both groups. A statistically significant between-group effect was observed only for the hypoglycaemia distress subscale score of the T1-DDS, favouring rtCGM. Self-reported health status, measured by the EQ-5D questionnaire, showed no statistically significant difference between both groups. The glucose monitoring satisfaction score showed that participants in the rtCGM group were more satisfied with their method of glucose monitoring than those in the control group. At the study end, fear of hypoglycaemia was lowered in both groups (between-group difference p=0.067).

Beck et al 2017 [26]: T2DM treatment groups did not differ statistically significant in any of the 5 quality-of-life measures. The CGM group had high satisfaction with use of CGM, as indicated by the mean score of 4.3 (SD, 0.4) on the CGM Satisfaction Scale (score range, 1 to 5). Mean scores were 4.4 (SD, 0.5) on the benefits subscale and 1.8 (SD, 0.5) on the hassles subscale, indicating that perceived benefits were high and perceived hassles low. On almost all items, most participants responded with scores indicating high satisfaction.

Van Beers et al 2016 [38]: No statistically significant differences were found in self-reported hypoglycaemia awareness scores, with no relevant between-group differences in 16-week hypoglycaemia awareness scores or change in hypoglycaemia awareness scores from baseline to endpoint. No between-group statistically significant differences were noted in quality of life from scores on the HFS Behaviour subscale, PAID-5, CIDS, EQ5D, or WHO-5 between the CGM and SMBG phases (data not shown). Scores on the HFS Worry subscale, transformed to a 0-100 scale, were statistically significantly lower after the CGM phase compared with the SMBG phase (32.5 vs 38.9; mean difference 6.4, 95% CI 1.4-11.4; p=0.014). CGM-SAT scores after the CGM phase were higher than neutral (3.0 on a 5.0 scale), with a mean score of 3.8 (SD 0.6).

Ly et al, 2013 [30]: The modified Clarke questionnaire for hypoglycaemia unawareness was completed by patients who were 12 years old or older. Parents of participants aged 4 to 18 years completed the parent version of HUS. There was a statistically significant improvement in the HUS score in both groups, from 6.4 (95% CI, 5.9-6.8) to 5.1 (95% CI, 4.5-5.6) in the pump-only group (p<0.001) and from 5.9 (95% CI, 5.5-6.4) to 4.7 (95% CI, 4.0-5.1) in the low-glucose suspension group (p<0.001). There were no statistically significant between-group differences (least square mean difference low-glucose suspension pump only (95% CI, −0.2; −0.9 to 0.5; p=0.58)).

Mauras 2012 [31]: Quality of life assessments was performed at 26 weeks and there were no statistically significant differences between treatment groups on the Hypoglycaemia Fear or the PAID questionnaire survey. Scores on the Blood Glucose Monitoring System Rating Scale were indicative of fewer problems/concerns perceived by the CGM group compared with the control group. On the CGM Satisfaction Scale at 26 weeks, parents generally reported a high degree of satisfaction with CGM, with an average item score of 3.9 and 86% of scores ≥3.5 (on a 5-point

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 111 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Likert-type scale, with 3 being neutral). Mean item scores were more favourable than neutral (>3.0) on all 43 items. Scores on the Benefits of CGM subscale tended to be slightly higher than scores on the Lack of Hassles of CGM subscale (mean 4.1 ± 0.4 vs 3.9 ± 0.6, respectively). 90% of parents responded that use of CGM makes adjusting insulin easier, shows patterns in blood glucose not seen before, and makes them feel safer knowing that they will be warned about low blood glucose before it happens. No one responded that he or she would not recommend CGM for other children with type 1 diabetes.

Riveline 2012 [33]: QoL was investigated at 1 year, and a statistically significant improvement in the physical component summary of the SF-36 questionnaire was reported in groups 1 and 2 in comparison with group 3 (groups 1 + 2: 1.47 ± 6.52; group 3: 22.48 ± 6.52; p=0.0042). The mental component summary did not differ statistically significantly between the groups (groups 1 + 2: 0.65 ± 10.55; group 3: 21.03 ± 10.62; NS) and neither did the global DQoL score. Patient satisfaction, assessed by one scale of the DQoL, statistically significantly improved in groups 1 and 2 (groups 1 + 2:2.83 ± 12.61 vs group 3: 22.12 ± 12.61; p=0.0447).

Table 27: QoL and user satisfaction data

Author, Year Outcome Intervention Control Difference p-value Patients on MDII Beck 2017 CGM satisfaction survey, 4.2 (0.4) NR NR NR [25] mean score (SD) n=101 Satisfaction Change in -0.2 ± 1.3 -0.3 ± 1.6 -0.1 (-0.7 to 0.64 Clarke Hypoglycemia n=102 n=53 +0.5) Unawareness Questionnaire Ruedy 2017 CGM satisfaction survey, 4.2 ± 0.4 NR NR NR [37] mean score n=60 Polonsky QoL WHO-5 (mean±SD) 71.28 ± 14.71 vs 69.06 ± 14.89 vs -1.26 (-5.42 to 0.62 2017 [33] 70.47 ± 16.68 67.32 ± 16.86 2.91) n=102/53 QoL EQ-5D-5L (mean±SD) 0.90 ± 0.11 vs 0.89 ± 0.11 vs 0.00 (-0.03 to 0.86 0.89 ±0.10 0.88 ±0.10 0.03) QoL Diabetes distress (DDS) 1.78 ± 0.65 vs 1.69 ± 0.62 vs 0.22 (0.08 to 0.009 (mean±SD) 1.61 ±0.48 1.78 ± 0.65 0.36) QoL Hypoglycaemia fear 15.75 ± 12.30 vs 17.30 ± 13.22 vs 3.17 (0.19 to 0.07 (worry subscale of HFS-II) 13.48 ± 10.63 17.73 ± 14.92 6.14) (mean±SD) Lind 2017 [29] DTSQ status version, 30.21 (95% CI 26.62 (95% CI 3.43 (95% CI <0.001 scale total 29.47–30.96) 25.61–27.64) 2.31–4.54) n=141 n=141 Beck 2017 WHO-5 (mean±SD) 16 (5) 17 (4) NR NR (T2DM) [26] n=77 n=73 EQ-5D-5L (mean±SD) 0.82 (0.14) 0.82 (0.16) NR NR Diabetes distress (DDS) 1.8 (0.9) 1.8 (0.6) NR NR (mean±SD), total Hypoglycaemia fear 0.8 (0.6) 0.7 (0.5) NR NR (worry subscale of HFS-II) (mean±SD) Patient satisfaction 4.3 (0.4) NR NR NR CGM Satisfaction Scale (score range 1 to 5) Mean Change in Clarke 0.1 (1.5) 0.2 (1.7) NR Hypoglycemia Unawareness Total Score from baseline SD

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 112 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, Year Outcome Intervention Control Difference p-value Heinemann QoL The diabetes distress total score was reduced in both groups. 2018 [13] A significant between-group effect was observed only for the hypo- n=75/66 glycaemia distress subscale score of the T1-DDS. Self-reported health status, measured by the EQ-5D questionnaire, showed no significant difference between both groups. Patient satisfaction The glucose monitoring satisfaction score showed that participants in the rtCGM group were more satisfied with their method of glucose monitoring than were those in the control group. Hypoglycaemia fear Lowered in both p=0.067 groups Patients on MDII and CSII (mixed) van Beers Change in Clarke score –0.1 –0.4 –0.3 0.216 2016 [38] from baseline (–0.5 to 0.3) (–0.8 to 0.0) (–0.9 to 0.2) n=26 n=26 QoL from scores on the HFS NR NR No difference NR Behaviour subscale, PAID-5, CIDS, EQ5D, or WHO-5 between the CGM and SMBG phases Scores on the HFS Worry 32.5 38.9 6.4, 95% CI p=0.014 subscale 1.4-11.4 CGM satisfaction survey, 3.8 (SD 0.6) NR NR NR mean score Mauras 2012 QoL: Hypoglycaemia fear 38±17 42±19 NR 0.38 [31] n=69 n=68 QoL: PAID questionnaire 44±17 49±16 NR 0.42 CGM Satisfaction Scale NR NR NR NR Overall 3.9±0.5 NR NR NR Benefits of CGM subscale 4.1±0.4 NR NR NR Lack of Hassles of CGM 3.9±0.6 NR NR NR subscale Riveline 2012 Reported for Group 1+2, not NA NA NA NA [36] separately Patients on CSII Ly 2013 [30] Modified Clarke questionnaire 5.9 (95% CI, 6.4 (95% CI, Square mean 0.58 for hypoglycemia unawareness 5.5-6.4) to 4.7 5.9-6.8) to 5.1 difference (95% CI, 4.0-5.1) low-glucose (95% CI, 4.5-5.6) (p< 0.001) suspension (p< 0.001) pump only n=46 n=49 −0.2; −0.9 to 0.5

Abbreviations: CGM, continuous glucose monitoring; CI, confidence interval; DQOL, Diabetes Quality of Life [questionnaire]; DTSQ, Diabetes Treatment Satisfaction Questionnaire; NR, not reported; RCT, randomized controlled trials; SD, standard deviation; Hypoglycemia unawareness score (HUS)

FGM vs SMBG [27, 28, 32]

Bolinder 2016 [27]: Patient satisfaction with treatment was statistically significantly improved for the intervention group compared with controls as well as the total treatment satisfaction and perceived frequency of hyperglycaemia (Table 28). Diabetes quality of life score did not statistically significantly favour either group in the full analysis set but was statistically significantly improved in the per-protocol set. There was no statistically significant difference in diabetes distress or hypoglycaemia fear behaviour or worry scores. Certainty of evidence for these outcomes was however very low. Details can be found in Table A9, Appendix 1.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 113 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 28: QoL and User Satisfaction data FGM vs SMBG (Bolinder 2016 Impact trial) (n=119 vs n=119)

Quality of life Neither group in the full analysis set statistically significant favoured (−0.08 [0.039]; p=0.0524) Patient satisfaction Patient satisfaction with treatment was statistically significant improved for intervention compared with control (adjusted between- group difference −0.24 [SE 0.049]; p<0.0001) Total treatment satisfaction Statistically significantly improved in the intervention group compared with the control group (6.1 [0.84]; p<0.0001) Perceived frequency of Statistically significantly improved in the intervention group hyperglycaemia compared with the control group(−1.0 [0.22]; p<0.0001) Diabetes distress No statistically significant differences between the study groups (−0.03 [SE 0.089]; p=0.7634) Hypoglycaemia fear behaviour No statistically significant differences between the study groups (0.0 [0.72]; p=0.9834) Worry scores No statistically significant differences between the study groups (−1.2 [1.48]; p=0.4154)

In the Haak 2017 REPLACE study [28]: patient-reported outcome and quality of life (QoL) measures were assessed using validated questionnaires: Diabetes Distress Scale (DDS), Diabetes Quality of Life (DQoL), and Diabetes Treatment Satisfaction (DTSQs and DTSQc) (Table 29). However, certainty of evidence for these outcomes was very low. Details can be found in Table A9, Appen- dix 1.

Table 29: Patient-reported outcome and quality of life (QoL) measures Haak 2017 Replace trial (n=119 vs n=75)

Patient Treatment satisfaction was statistically significant higher in intervention compared with satisfaction controls (DTSQ 13.1 ± 0.50 (mean ± SE) and 9.0 ± 0.72, respectively; p<0.0001). Total treatment satisfaction score for DTSQ (status versus change) was statistically significantly improved for intervention group participants (13.1 ± 0.50, mean ± SE) compared with controls (9.0 ± 0.72), p<0.0001. Satisfaction with treatment results using DQoL demonstrated statistically significant improvement for the intervention group (-0.2 ± 0.04, mean ± SE) versus the control group (0.0 ± 0.06), p=0.0259, for this element of the questionnaire. There were no other statistically significant differences observed in other aspects of DTSQ and DQoL or for the DDS scales. User questionnaire results showed intervention participants agreed with positive aspects of the device including use, comfort, and utilization of sensor glucose information.

Oskarsson 2018 [32]: DQoL satisfaction with treatment score was statistically significantly improved for the intervention group compared with control participants (p<0.0001), as was the DTSQ overall treatment satisfaction score (p<0.0001), perception of hypoglycaemia (p=0.010), and perception of hyperglycaemia scores (p<0.0001). Hypoglycaemia fear behaviour (p=0.76) or worry (p=0.59) scores and diabetes distress score (p=0.98) were similar for both groups (Table 30). Certainty of evidence for these outcomes varied from low to very low. Details can be found in Table A9, Ap- pendix 1.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 114 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 30: QoL and patient satisfaction for Oskarsson et al. 2018, MDII subgroup of Impact trial* (n=78 vs n=70)

Quality of life Questionnaire item Difference in adjusted p value Patient (mean (SD) unless otherwise stated) means in intervention and satisfaction control (95% CI) DTSQ Total treatment satisfaction score 6.4 (4.4, 8.4) <0.0001 /28.3 (4.7) to 13.3 (5.4) vs 27.7 (5.3) to 6.8 (6.2)/ Perceived frequency of hypoglycaemia −0.6 (−1.1, −0.2) 0.010 Perceived frequency of hyperglycaemia −1.2 (−1.7, −0.7) <0.0001 DQoL Total core scale score −0.1 (−0.2, 0.0) 0.15 /1.9 (0.3) to 1.9 (0.4) vs 21. (0.5) to 2.1 (0.5)/ Satisfaction with treatment −0.3 (−0.4, −0.1) <0.0001 Social worry 0.0 (−0.1, 0.2) 0.74 Diabetes worry −0.1 (−0.3, 0.1) 0.23 Impact of treatment −0.0 (−0.1, 0.1) 0.82 DDS Total DDS score 0.0 (−0.2, 0.2) 0.98 Emotional burden subscore −0.0 (−0.3, 0.2) 0.77 Physician distress subscore 0.1 (−0.2, 0.4) 0.45 Regimen distress subscore −0.0 (−0.3, 0.2) 0.71 Interpersonal distress subscore 0.0 (−0.2, 0.3) 0.74 HFS Behavioural subscale −0.3 (−2.0, 1.4) 0.76 Worry subscale −1.0 (−4.6, 2.6) 0.59

* Participants were requested to complete questionnaires at 6 months. Data from n = 5 participants who did not complete the study (n = 4 from the intervention group and n = 1 from the control group) are included in the analysis. Questionnaire data was not available for n = 1 individual in the intervention group who did complete the study.

FGM vs rtCGM [34]

Reddy 2018 [34]: No statistically significant difference was observed in overall Gold score from baseline to end-point between the two groups. A statistically significant difference was observed in HFS total score and changes in the worry sub-score of the HFSII (p=0.02). No within- or between group differences were noted in HFS-II behaviour sub-score, and PAID scores (Table 31). Quality of this study for these outcomes was very low. Details can be found in Table A8, Appendix 1.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 115 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 31: Gold Score, Hypoglycaemia fear (HFS-II) and Diabetes-related emotional distress (PAID questionnaire)

Author, Outcome rtCGM FGM P-value Year Median (95% CI) Median (95% CI) Change from baseline Change from baseline Patients on MDII or CSII (mixed population) Reddy Gold score 0.0 (−1.0 to 0.0) 0.0 (−0.8 to 0.0) 0.23 2018 [34] HFS total score −6.5 (−10.8 to −2.2) −2.0 (−3.8 to 2.8) 0.02 n=19/20 HFS-Behaviour subscore −2.0 (−3.8 to −0.1) −0.5 (−3.0 to 1.8) 0.36 HFS- Worry subscore −4.5 (−7.8 to −0.1) 0.5 (−3.0 to 2.8) 0.02 PAID score −1.0 (−5.7 to 4.8) −1 (−5.0 to 2.0) 0.82

Other outcomes

Sensor use (Compliance/Adherence) and number of daily finger-sticks tests

Beck et al 2017 [25]: Among the 102 participants in the CGM group who completed the trial, median CGM use was 7.0 d/wk (IQR, 7.0-7.0) at 4, 12, and 24 weeks; only 2 (2%) discontinued CGM before the 24-week visit. During month 6 (weeks 21-24), CGM use was 6 or more d/wk for 93% of the 102 participants.

Ruedy et al 2017 [37]: Among the 61 in the CGM group completing the trial, 97% used CGM ≥6 days/week in month 6. The mean reduction in the number of daily blood glucose tests from baseline to week 24 was statistically significantly greater for the CGM group compared with the control group (−1.2 ± 1.6 vs −0.2 ± 1.4, p=0.001).

Lind et al 2017 [29]: Overall mean time of CGM use (estimated by the proportion of CGM data downloaded in relation to follow-up time) was 87.8% during CGM treatment periods. CGM use ranged between 86.5% and 91.9% during various study visits. HbA1c was reduced by 0.46% (0.31%-0.61%) in patients using the CGM sensor more than 70% of the time, and there was no statistically significant difference in HbA1c for those using the CGM sensor for less than 70% of the time. Patients performed a mean (SD) of 2.75 (1.39) self-measurements of blood glucose during CGM therapy and 3.66 (2.30) during conventional therapy.

Beck et al 2017 [25]: The CGM group averaged 6.7 days (SD, 0.9) of CGM use per week.

Battelino 2011[24]: Median sensor wear in the CGM group in month 6 was 5.9 days per week. Median sensor wear was 5.6 (paediatric 5.6; adults 4.9) and 6.1 (paediatric 6.1; adults 6.1) day per week.

Mauras 2012 [31]: Frequency of sensor use in the CGM group A total of 63 (91%) of 69 participants who completed the 26-week visit were wearing a sensor on at least 1 day a week at the end of 26 weeks. The amount of CGM sensor wear decreased during the 26 weeks of the study (p<0.001), with only 41% averaging at least 6 days/week of wear in month 6. The amount of sensor wear in month 6 did not vary with age overall (r=20.07) and was not associated with baseline HbA1c (r=20.02). There was no association between change in HbA1c from baseline to 26 weeks and the overall amount of CGM sensor wear during the entire 26 weeks (Spearman rs=20.09, p=0.44) or during month 6 (Spearman rs=20.11, p=0.37). 28 participants who wore a sensor ≥6 days/week during month 6 tended to have a slightly greater reduction in HbA1c

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 116 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin compared with the 41 participants who wore a sensor less frequently (mean change from baseline to 26 weeks 20.3 ±0.7% vs. 0.0 ± 0.5%, p=0.01; 25 vs. 15% with a reduction in HbA1c ≥0.5% without a severe hypoglycaemia event, p=0.33). Among those wearing a sensor ≥6 days/week in month 6, the median percentage of glucose values in the target range of 71-180 mg/dL was 51%, with 38% >200 mg/dL, 16% >250 mg/dL, and 0.3% ≤60 mg/dL compared with 43, 44, 23, and 0.4%, respectively, in those wearing a sensor less frequently.

Riveline 2012 [36]: In group 1, actual sensor use was 3.42 per month (median [Q1–Q3] [2.20-3.91]), that is, 57 ± 20% of the prescribed time. No correlation between sensor use and HbA1c level was noted in this group (p=0.117, R2=0.0449). Compliance in terms of sensor use was negatively correlated with HbA1c level (p=0.026, R2=0.1050). Total sensor consumption during 1 year was statistically significant lower (34%) in group 2 than in group 1 (p=0.001). From baseline to the end of the study, SMBG per week statistically significantly decreased in both CGM groups versus the control group (29 ± 12 vs. 1 ± 12, p<0.0001).

Ly et al, 2013 [30]: presented results on sensor use over 6 months (median) for Sensor-Augmented Pump With Low-Glucose Suspension (n=46) as follows: 68% in all age groups, 71% in participants younger than 12; 54% in the 12-18 group, and 81% in the >18 group.

In van Beers et al, 2016 [38]: the percentage of time using a sensor was 89.4% (IQR 80.8-95.5).

Bolinder et al 2016 in T1 DM patients [27]: The authors reported the mean number of self-monitored blood glucose tests performed per day by the intervention group immediately reduced from 5.5 (SD 2.0) tests per day in the 14 day baseline phase to 0.5 (0.7) tests per day during the treatment phase of the trial. This was an unprompted response by intervention participants that clinically equates to one self-monitoring of blood glucose test every 2-5 days. The mean number of sensor scans per day for the intervention group was 15.1 (SD 6.9) during the treatment phase. System utilisation, defined as the percentage of data collected assuming continuous device wear for 6 months by the intervention group (n=112) was 92.8% (SD 7.3). The number of self-monitoring blood glucose tests performed by participants in the control group was consistent throughout the study, from 5.8 tests (SD 1.7) per day at baseline to 5.6 (2.2) per day at 6 months.

Haak 2017 in T2 DM patients [28]: Self-monitoring blood glucose frequency for intervention participants fell from 3.8 ± 1.4 tests/day mean ± SD (3.8 tests/day median) at baseline to 0.5 ± 1.1 (0.1 median) from the first unblinded sensor wear with full access to sensor glucose data (day 15– 31), reducing further to 0.4 ± 1.0 tests/day (0.0 median) by the study end (day 208). The overall blood glucose monitoring rate over 6 months was 0.3 ± 0.7, median 0.1. Self-monitoring of blood glucose frequency for control participants was 3.9 ± 1.5 test/day (median 3.9) at baseline and this rate was maintained until the study end [3.8 ± 1.9 (median 3.9)]. Control group participants <65 years performed less blood glucose monitoring tests (2.78 ± 1.08 test/day) than those ≥65 years (3.46 ± 0.94), p=0.0247. During the treatment phase (day 15 onwards), the average sensor-scanning frequency was 8.3 ± 4.4 (mean ± SD) times/day (median 6.8), i.e. double the frequency of blood glucose testing. There was no significant difference in the number of scans performed by those <65 years and >65 years of age [8.1 ± 4.6 (median 6.8) and 8.5 ± 4.1 (median 6.9), respectively, p= 0.6627]. Device use for the intervention group (n = 138) was 88.7 ± 9.2% (defined as the percentage of data collected, assuming continuous device wear for 6 months).

Oskarsson et al 2018 [32]: For participants in the intervention group, the mean ± SD daily sensor scanning frequency was 14.7 ± 10.7 (median 12.3) and the mean number of self-monitored blood glucose tests performed per day reduced from 5.5 ± 2.0 (median 5.4) at baseline to 0.5 ± 1.0 (median 0.1). The baseline frequency of self-monitored blood glucose tests by control participants was maintained [from 5.6 ± 1.9 (median 5.2) to 5.5 ± 2.6 (median 5.1) per day].

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 117 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

6 SAFETY (SAF)

6.1 Research questions

Element ID Research question C0008 How safe is the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – in relation to the comparator(s)? C0004 How does the frequency or severity of harms change over time or in different settings? C0005 What are the susceptible patient groups that are more likely to be harmed through the use of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems)? B0010 What kind of data/records and/or registry is needed to monitor the use of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s)?

6.2 Results

Included studies

According to the protocol, in addition to the 12 RCTs included in the Effectiveness domain [13, 24, 25, 27-31, 34, 36-38], 2 prospective non-randomised studies and 1 single-arm extension study of a RCT were included for the assessment of safety [39-41] (Table 6, Appendix 1). One out of 12 RCTs (with high risk of bias at the study level) did not specify AEs as an outcome or report them [34]. Out of 3 prospective non-randomised studies included for the assessment of safety, 2 were an interventional non-randomised accuracy and safety studies and one was an interventional single- arm study as extension of RCT. Neither the risk of bias nor the quality of evidence according to GRADE [44] (in which observational studies are primarily graded as low quality unless upgraded by review authors to moderate or high quality, if the effect is large enough) were assessed for these studies.

Frequency and severity of local adverse events in RCTs [13, 24, 25, 27-31, 34, 36-38] and nRCTs [39-41] related to device (rtCGM and FGM) or procedures are described in Table 32. In brief, local AEs were mentioned in 12 RCTs (14 articles) and 3 nRCTs. Risk of Bias on outcome level for this outcome varied from unclear to high; details can be found in Table A4, Appendix 1.

Systemic AEs were reported in all but one study [34] and SAEs in the majority of studies, but investigators found the majority of them were unrelated to the intervention. Statistical analysis was not performed for the majority of studies.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 118 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Patient safety

[C0008] – How safe is the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – in relation to the comparator(s)?

Frequency and severity of local adverse events in 12 RCTs [13, 24, 25, 27-31, 34, 36-38] published in 14 articles and 3 nRCTs [39-41] related to devices (rtCGM and FGM) or procedures are described in Table 32. In brief, local AEs were mentioned in 5 RCTs (3 related to FGM device) [27-30, 32] and 3 nRCTs (all related to FGM device) [39-41].

Table 32: Frequency and severity of local adverse events in RCTs and nRCTs related to devices (rtCGM and FGM) or procedures

Studies Device (or procedure) related local AEs RCTs rtCGM MDII patients Beck 2017 DIAMOND None reported [25] Ruedy 2017 [37] None reported Lind 2017 GOLD [29] One patient in the CGM group discontinued use because of an allergic reaction to the sensor. Beck 2017 T2DM [26] None reported Heinemann 2018 [13] None reported rtCGM CSII patients Ly 2013 [30] Device failure occurred on 2 occasions and was corrected with replacement of the sensor transmitter (Minilink™) rtCGM MDII and CSII patients Battelino 2011 [24] None reported Mauras 2012 [31] None reported Riveline 2012 [36] None reported Van Beers 2016 [38] None reported FGM vs CGM Reddy 2018 [34] None reported FGM vs SMBG Bolinder 2016 [27] 13 adverse events were reported by ten participants related to the sensor – four allergy events (one severe, three moderate); one itching (mild); one rash (mild); four insertion-site symptom (severe); two erythema (one severe, one mild); and one oedema (moderate); all were resolved. There were 248 sensor insertion-site signs and symptoms experienced by 65 participants across both groups. Signs are subdivided into those expected due to sensor insertion: pain (38), bleeding (25), oedema (eight), induration (five), and bruising (five), and those associated with sensor wear: erythema (85), itching (51), and rash (31). Seven participants withdrew from the study due to device-related adverse events or repetitive occurrences of sensor insertion-related symptoms.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 119 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Studies Device (or procedure) related local AEs RCTs Haak 2017 [28] Six intervention participants (4%) reported nine adverse events for sensor-wear reactions (two severe, six moderate, one mild). These were sensor-adhesive reactions, primarily treated with topical preparations. All were resolved at study exit. Anticipated symptoms refer to those typically expected using a sensor device and equate to symptoms normally experienced with blood glucose finger-stick testing, e.g., pain, bleeding, bruising. There were 158 anticipated sensor insertion site symptoms observed for 41 (27.5%) intervention and 9 (12.0%) control participants. These symptoms were primarily (63%) due to the sensor adhesive (erythema, itching, and rash) and resolved without medical intervention. Oskarsson 2018 [32] Eight adverse events for six (7%) intervention participants were related to wearing the study device. Four participants withdrew because of these adverse events. There were 144 sensor insertion-site symptoms experienced by 34 participants. The numbers of participants affected by expected signs or symptoms due to sensor insertion were: pain, n=14; bleeding, n=9; oedema, n=3; and induration, n=3. The symptoms associated with sensor wear were erythema, n=23; itching, n=14; and rash n=8. nRCT or single arm extension results of RCT FGM vs SMBG Bailey 2015 [39] Skin issues observed in 202 site exams of 72 study participants: moderate to severe itching 0.5% of the time, moderate erythema 4.0% of the time, and 98.6% of the insertions had a pain rating of ≤2. Rate of mild incidences was <9% for any individual category of skin issues mentioned above, including oedema, rash, induration, bruising, bleeding, and others. Edge 2016 [40] Five device-related AEs were reported in total from five (6%) participants, aged 6, 9, 10, 12, and 15 years: allergic reaction, blister, pink mark/scabbing, and abrasion on sensor removal (n=2) – four were mild, one was moderate; all were resolved at study completion. Site exams performed for all sensor insertions checked for anticipated AEs associated with sensor application or insertion sites: moderate erythema was observed on 11.6% of occasions, mild erythema and pain 13.6% and 4.1%, respectively, and mild instances of bleeding, bruising, itching and oedema were each found in <3% of cases. There were no trends in the rate of anticipated AEs, including erythema, with age. Haak 2017 Nine participants reported 16 instances of device-related adverse events single-arm results (e.g. infection, allergy) and 28 participants (20.1%) experienced 134 occurrences at 12 months [41] of anticipated skin symptoms/sensor-insertion events expected with device use (e.g. erythema, itching, and rash).

Systemic SAEs were reported in the majority of studies [13, 24-27, 29-32, 36-41] and are briefly described below, but investigators found the majority of them were unrelated to the intervention.

CGM MDII patients

In Beck 2017 (DIAMOND Trial) [25], severe hypoglycaemic events occurred in 2 participants in each group (p=0.67). There were no occurrences of diabetic ketoacidosis. Other serious adverse events, unrelated to the study intervention, occurred in 2 participants in the CGM group (inner ear disorder, pulmonary mass, and trigeminal neuralgia) and none in the control group.

In Ruedy 2017 [37] there were no severe hypoglycaemia or diabetic ketoacidosis events in either group.

In the Lind 2017 GOLD Trial [29], there were in total 77 patients with 137 adverse events during CGM and 67 patients with 122 adverse events during conventional therapy. There were no major differences for any adverse event between the treatments. The most adverse event in both groups

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 120 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin was nasopharyngitis. There were 7 patients with a total of 9 serious adverse events during CGM treatment and 3 patients with total of 9 serious adverse events during conventional treatment. Ketoacidosis was not reported during the study. Five patients in the conventional treatment group and 1 patient in the continuous glucose monitoring group had severe hypoglycaemia/24-week Incidence Rate (per 100 person-years): 4.2 vs 12.2, p=0.27). During washout when patients used conventional therapy, 7 patients had severe hypoglycaemia. One patient in the CGM group discontinued use because of an allergic reaction to the sensor.

Beck 2017 Type 2 DM [26] reported that severe hypoglycaemia or diabetic ketoacidosis did not occur in either group. In the CGM group, 1 participant died of a myocardial infarction and 2 others were hospitalized for chest pain and fully recovered, all considered to be unrelated to CGM use. No serious adverse events occurred in the control group.

Heinemann 2018 [13] reported 18 serious adverse events: seven in the control group (two severe episodes of hypoglycaemia, one kidney transplantation, one myocardial infarction, two colon polyps, and one seizure), ten in the rtCGM group (four episodes of severe hypoglycaemia, two diabetic foot ulcers, one allergic reaction following a wasp sting, two fractures, and one kidney tumour removal), and one before randomisation (whiplash after a car accident). No event was considered to be related to the investigational device.

CGM CSII patients

In Ly 2013 [30] there were no episodes of diabetic ketoacidosis or hyperglycaemia with ketosis. Device failure occurred on 2 occasions and was corrected with replacement of the sensor transmitter (Minilink™).

CGM MDII and CSII patients

In Battelino 2011 [24], four serious adverse events were reported although none were related to the study or device (prolonged cephalea in the control group and mild diabetic ketoacidosis, myasthenia gravis, and laparoscopic repair of the spermatic vein in the CGM device group). This incident of mild diabetic ketoacidosis in a patient in the continuous monitoring group was due the patient disconnecting his or her insulin pump. However, this patient had stopped wearing the continuous glucose monitoring system 2 weeks before the incident. No incidents of severe hypoglycaemia were reported.

In Mauras 2012 [31] there were no cases of diabetic ketoacidosis and no serious adverse events attributable to the study interventions, including no serious skin reactions.

In the Riveline 2012 study [36], the frequency of severe hypoglycaemia episodes did not differ among groups after adjustment for both age and the presence of at least one severe hypoglycaemia episode during the year before inclusion. A nonsignificant increase was observed in group 1, due to one patient who experienced seven episodes of severe hypoglycaemia. Ketoacidosis events were infrequent in all groups. In patients on CSII, the frequency of severe and mild hypoglycaemia was similar in the CGM groups compared with the control group.

Van Beers 2016 [38] reported that during CGM, the number of severe hypoglycaemic events was lower (14 events vs 34 events, p=0.033). Five serious adverse events other than severe hypoglycaemia occurred during the trial, but all were deemed unrelated to the trial intervention. Addition- ally, no mild to moderate adverse events were related to the trial intervention. In the washout phase, one event each of anaphylactic reaction to eye drops, cerebral contusion, rupture of the

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 121 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Achilles tendon, and rupture of the biceps tendon occurred; one hospital admission for erysipelas (not at the CGM insertion site) occurred during the CGM phase. No ketoacidosis occurred during the trial. 11 mild to moderate adverse events occurred during the CGM phase, 16 mild to moderate adverse events occurred during the SMBG phase, and two mild to moderate adverse events occurred during the wash-out phase. The mild to moderate adverse events were deemed unrelated to the study intervention and consisted of adverse events related to the musculoskeletal system (CGM phase, n=2; SMBG phase, n=6), urinary tract infection (CGM phase, n=2), dermal infection (CGM phase, n=2; SMBG phase, n=1; wash-out, n=1), gastrointestinal infection (CGM phase, n=1; SMBG phase, n=3), dermal burn (CGM phase, n=1), fever for less than 1 week (CGM phase, n=2; SMBG phase, n=3), excision of lipoma (CGM phase, n=1), dyspnoea (washout, n=1), perio- dontitis (SMBG phase, n=1), infection of the upper respiratory tract (SMBG phase, n=1), and glaucoma (SMBG phase, n=1). No adverse events resulted in discontinuation of the study.

FGM vs SMBG RCTs

In the Bolinder 2016 RCT [27], 13 adverse events related to the sensor were reported by ten participants – four of allergy events (one severe, three moderate); one itching (mild); one rash (mild); four insertion-site symptom (severe); two erythema (one severe, one mild); and one oedema (moderate); all were resolved. There were 276 AEs reported (138 in each group); there were ten serious adverse events (five in each group) reported by nine participants; none were related to the device. There were six hypoglycaemia-related serious adverse events (requiring hospitalisation or third party intervention) in five participants: two in the intervention group (n=2) and three in the control group (n=3). Additionally, there were three hypoglycaemia-related adverse events reported in the control group (n=2). None of the hypoglycaemic events were considered device-related. There were no reported events of diabetic ketoacidosis during the study. There were 248 sensor insertion-site signs and symptoms experienced by 65 participants across both groups. Signs can be subdivided into those expected due to sensor insertion: pain (38), bleeding (25), oedema (eight), induration (five), and bruising (five), and those associated with sensor wear: erythema (85), itching (51), and rash (31). Seven participants withdrew from the study due to device-related adverse events or repetitive occurrences of sensor insertion-related symptoms.

As Haak 2017 reported in this 6-months RCT [41], serious adverse or adverse events (n=515) were experienced by 114 (76.5%) intervention and 47 (62.7%) control participants. No serious adverse events or severe hypoglycaemic events were reported related to the device or study procedure. Forty-two serious events [16 (10.7%) intervention participants, 12 (16.0%) controls] were not device-related. Four hypoglycaemia serious adverse events were experienced by four participants (three interventions and one control) and 57 hypoglycaemia adverse events by 10 (7%) intervention and seven (9%) control participants. None of the severe hypoglycaemic episodes or hypoglycaemic adverse events were associated with the device. Three participants (one intervention, two controls) experienced an adverse event leading to withdrawal from the study; none were associated with the device. There were no reported events of diabetic ketoacidosis or hyperosmolar hyperglycaemic state. Seven cardiac events were reported for four (2.7%) intervention and three (4.0%) control participants (none were considered to be related to study procedures or the device). Six intervention participants (4%) reported nine adverse events for sensor-wear reactions (two severe, six moderate, one mild). These were sensor-adhesive reactions, primarily treated with topical preparations. All were resolved at study exit. Anticipated symptoms refer to those typically expected using a sensor device and equate to symptoms normally experienced with blood glucose finger- stick testing, e.g., pain, bleeding, bruising. There were 158 anticipated sensor insertion site symp-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 122 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin toms observed for 41 (27.5%) intervention and 9 (12.0%) control participants. These symptoms were primarily (63%) due to the sensor adhesive (erythema, itching, and rash) and resolved without medical intervention.

In the Oskarsson 2018 study [32], no device-related hypoglycaemia or safety-related issues were reported. Nine serious adverse events were reported for eight participants (four in each group), none related to the device. Eight adverse events for six of the participants in the intervention group were also reported, which were related to sensor insertion/wear; four of these participants withdrew because of the adverse event. There were 178 adverse events, including serious adverse events, experienced by 85 participants (52% for each group). Nine serious adverse events were reported for eight participants (four in each group), none related to the study device or procedure. Five hypoglycaemia-related serious adverse events were reported for four participants; one in the intervention group (one participant) and four in the control group (three participants). One control participant discontinued the study because of severe hypoglycaemia. In addition, one participant in the control group experienced two hypoglycaemia-related adverse events. There were no diabetic ketoacidosis events reported. Eight adverse events for six (7%) intervention participants were related to wearing the study device. Four participants withdrew because of these adverse events. There were 144 sensor insertion-site symptoms experienced by 34 participants. The numbers of participants affected by expected signs or symptoms due to sensor insertion were: pain, n=14; bleeding, n=9; oedema, n=3; and induration, n=3. The symptoms associated with sensor wear were erythema, n=23; itching, n=14; and rash n=8.

FGM vs SMBG nRCT or single arm results of RCT

In Bailey 2015 nRCT [39] there were no unexpected adverse device effects reported during the clinical study. One participant had a serious adverse event (severe hypoglycaemia prior to sensor insertion) not related to the study or device. Skin issues observed in 202 site exams of 72 study participants were as follows: moderate to severe itching 0.5% of the time, moderate erythema 4.0% of the time, and 98.6% of the insertions had a pain rating of ≤2. Rate of mild incidences was <9% for any individual category of skin issues mentioned above, including oedema, rash, induration, bruising, bleeding, and others.

In the Edge 2017 study [40], one participant had a serious AE that was not related to the study or device (pain and lack of feeling in leg). Five device-related AEs were reported in total from five (6%) participants, aged 6, 9, 10, 12, and 15: allergic reaction, blister, pink mark/scabbing, and abrasion on sensor removal (n=2) – four were mild, one was moderate; all were resolved at study completion. Site exams performed for all sensor insertions checked for anticipated AEs associated with sensor application or insertion sites: moderate erythema was observed on 11.6% of occasions, mild erythema and pain on 13.6% and 4.1%, respectively, and mild instances of bleeding, bruising, itching, and oedema were each observed on <3% of occasions. There were no trends in rate of anticipated AEs, including erythema, with age.

Haak 2017 [41] presented single-arm results at 12 months. During the 6-month extension period, no device-related serious adverse events were reported. Nine participants reported 16 instances of device-related adverse events (e.g. infection, allergy) and 28 participants (20.1%) experienced 134 occurrences of anticipated skin symptoms/sensor-insertion events expected with device use (e.g. erythema, itching, and rash).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 123 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

FGM vs CGM

In the Reddy 2018 trial [34], no episodes of severe hypoglycaemia were reported during the 8- week intervention phase in either group. AEs were not mentioned as aim or outcome or reported; there were no pre-specified primary or secondary endpoints; thus the study does not report the results for a key outcome that could reasonably be expected for a study of its nature.

[C0004] – How does the frequency or severity of harms change over time or in different settings?

No published data were found to answer this question.

[C0005] – What are the susceptible patient groups that are more likely to be harmed through the use of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems)?

No published data were found to answer this question.

[B0010] – What kind of data/records and/or registry is needed to monitor the use of the technology – the continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) medical devices (as personal, standalone systems) – and the comparator(s)?

Observational registries studies could be important for safety monitoring of these health technologies already registered on the market; establishment of the new registries could be important for the new health technologies for the prospective follow-up and real-word data collection and analysis.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 124 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

7 POTENTIAL ETHICAL, ORGANISATIONAL, PATIENT AND SOCIAL, AND LEGAL ASPECTS (ETH, ORG, SOC, LEG)

Some specific questions concerning ethical, organisational, patient and social aspects were identified from the Rapid REA Checklist (Appendix 3). If deemed relevant, these should be answered at national/regional level.

Questions related to equity could be important to take into consideration. Could potential inequalities prevent access to the technology? And more specifically, are there factors that could prevent a group or person from gaining access to the technology? If so, is it possible to influence these factors or manage the utilisation of the technology in a way that gives equal access to those in equal need?

Questions related to supportive actions and information could be important to focus on. Is there any specific information or any support patients (or decision-makers?) should seek to decide upon adopting the technology? Are there any particular challenges related to the use of the technology that the patient and/or care-givers need to be aware of? Is there clear and sufficient information available to understand the technology and possible risk related to unappropriated use?

Questions related to the involvement of patients and caregivers as well as proper education and training could be important to raise. What kind of involvement of patients/participants and/or caregivers is the most suitable, and what kinds of co-operation and communication of activities are needed?

Questions related to patients' perspectives and perceptions as well as their expectations from using the technology could be important to discuss. This includes any positive or negative experiences that may arise as a consequence of using the technology (i.e., worries, satisfaction, stigmatisation, social status, etc.). A new technology may allow patients to return to work, however since the technology can be seen or the alarm sound heard by co-workers, it may lead to undesired attention from the surroundings.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 125 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

8 PATIENT INVOLVEMENT

Predefined questions related to impact of condition; experience with currently available medical devices; experiences with, and expectations of, the medical devices being assessed; and additional information which patients believe would be helpful to the HTA researchers were used with aim of helping in the assessment of the value of health technologies, specifically CGM and FGM medical devices.

A focus group (with individual patients) was held in Croatia – one for adults and one for children with informal caregivers. All participants had T1 DM; 12 patients in the adult group (9 men, 3 women aged 25-74; time period of DM range 4-61 years) and 7 children (2 girls, 5 boys aged 12-18; DM range 2 years to 16 years).

Two patient groups were contacted, one at the European level and one at the national level. The Patient Group Submission template was sent to The International Diabetes Federation European Region, Brussels and Diabetes Scotland, Scotland. The Patient Group Submission template was prepared for this assessment by modifying the HTAi Patient Group Submission template for HTA of health interventions (not medicines) [23] and with inclusion of topics and issues from the EU- netHTA Core Model® 3.0 related to the Patients and Social Domain aspect [10].

The compleded Patient Group Submission template received from the International Diabetes Fed- eration European Region, Brussels and Diabetes Scotland, Scotland as well as a summary from the two Focus Groups conducted in Croatia can be found in Appendix 1.

Summary of patient involvement can be found below.

Impact of condition – Diabetes mellitus treated with insulin

Diabetes Serious, life-long health condition; two main types of diabetes: Type 1 and mellitus Type 2, as different conditions, with differing modes of onset; if left untreated, high blood glucose levels can cause serious health complications, even death; Living with diabetes is difficult and exhausting; variability of glucose values in different life cycles; People face managing both the short and long term complications on a daily basis, which is frequently described as ‘relentless’, ‘life limiting’, ‘stressful’, ‘unpredictable’, and ‘exhausting’; Healthy eating and exercise are the foundation of good diabetes management; Understanding the nutritional profile of foods and drinks, including the calorie, fat, sugar and salt content levels, is vital for day to day and long term management of diabetes. This information is often limited or missing from menus in restaurants and other out of home food outlets. This lack of information often prohibits or stops people from going out for meals with family and friends and can lead to social isolation; Growing up with type 1 diabetes, there was not much awareness about the condition, and everyone kept it a secret. Some patients are ashamed of the diagnosis and always hide their treatment from their loved ones; more difficult to accept the condition and have control; diet and exercise are important factors, however, type 1 diabetes comes with a lot of anger, frustration, anxiety, stress and much more; weight gain is a common side effect for people who take insulin. This can cause a greater stress on the person with diabetes, especially young girls. Gaining weight may be very stressful on the person, and may want to give up entirely causing more damage to the person’s health; women with Type 1 have close to two and a half times the chance of developing an eating disorder. Diabetes treated with insulin affects many areas in

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 126 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

a person’s life; social life, ability to engage in a specific activity or task at a certain moment due to high/low blood sugar; Stigma definitely exists, but it’s impact is different by countries; People living with diabetes do not get a day off from the condition. Those dependent on insulin are required to undertake self-monitoring of blood glucose (SMBG) and self-manage 24 hours a day, 365 days a year. Not everyone with diabetes comes to terms with the fact that they are living with a long term condition, or are able to sustain the intense daily vigilance required to keep healthy. This need for constant vigilance can lead to ‘diabetes burnout’, anxiety, obsession and eating disorders such as anorexia and diabulimia. Not taking the correct amount of insulin may occur for many reasons such as fear of hypos or under-estimating carbohydrate intake; when it is associated with weight control and happens over a prolonged period this is called diabulimia. Eating disorders in T1 diabetes contribute to poor diabetes control, rapid development of secondary complications such as retinopathy and neuropathy, and increased rates of severe hypo, hyperglycaemia leading to Diabetes Ketone Acidosis, and mortality compared with a person with T1 without eating disorders.

Insulin therapy Required to inject multiple times a day and recommended to test blood glu- and glucose cose levels between 4-10 times a day: upon waking, before and after meals, measuring before and after physical activity, before driving, before going to bed; If the person develops a mild illness such as cold or gastric upset, this not only impacts on blood glucose control quite markedly, they are also required to increase the frequencies of blood glucose testing; For those already struggling with needle phobia this can be an enormous struggle and cause further emotional and mental distress; the need to constantly finger prick to test blood glucose is painful and inconvenient; discrimination in the work place for doing blood tests in public which results in them limiting the number of times they tests their levels

Regular review Whilst diabetes is recognised under the Disability Discrimination Act, em- appointments on ployers and education establishments frequently do not make reasonable a monthly 3, 6, 9 adjustments. Individuals have reported being subjected to competency or or 12 monthly disciplinary proceedings as result of periods of absence due to attending basis, depending frequent diabetes appointments or seeking treatment/support for diabetes on the stability related complications. A survey in 2017 by Diabetes Scotland revealed that and complexity 65% of people with diabetes found it difficult to manage their condition at of their diabetes work. The survey revealed that 32 per cent of people felt it was ‘not very easy’ or ‘not at all easy’ to take time away during work to self-manage their condition with blood testing, taking medication etc. Furthermore, 12 per cent of respondents said they had been refused time off to attend a diabetes healthcare appointment.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 127 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The most challenging aspects of living with diabetes

Prevention and If left untreated severe hypo can lead to seizures, coma, lasting neuro def- management of icits, and even death. The fear of, and anxiety associated with, hypos is one hypoglycaemia of the most talked about topics covered in calls to Diabetes UK Helpline, in (hypos), focus and support groups, and online forums. Hypos are distressing not especially only for the person living with the condition but also for parents, spouses nocturnal (night and family members. They can be difficult and distressing to manage, the time) hypos person may become aggressive, irritable, uncooperative, unsteady, confused etc. The impact and time required to recover from a severe hypo varies from individual to individual. On average it can take several hours before the person is able to resume normal daily tasks. Recovery from a nocturnal hypo can be longer due to disturbed sleep and length of time to reach safe target levels. In some instances the person may not be able to attend school, college, or university, thus missing vital education; or, if employed, attend work. All of which can impact on attainment, future prospects, security of tenure of employment and close relationships due to financial insecurity.

Potential loss of This frequently results in individuals testing excessively, running their blood hypo awareness glucose level higher than advised targets – thus increasing the risk of micro and macro vascular damage and serious lasting complications such as blindness, stroke, kidney disease, neuropathy and amputation; and/or withdrawing from social events/ interaction because of the fear of hypoglycaemic episodes and the possible consequences (loss of control, hospital admission, injury through falling, discrimination, and stigma).

Burden of disease in different patient groups

Young, Optimum management, administering insulin and testing blood glucose, and pre-school and balancing insulin and food intake, requires skills in numeracy and dexterity; early years for young, pre-school and early years children the day to day management children and of the condition is largely reliant on the intervention of the main care giver/ families parents; Parents frequently report feelings of guilt, denial, anger and anxiety at diagnosis – why did this happen? Is it my fault? How will I/we cope?; Families find themselves faced with unexpected hospital stays, constant medication adjustments and lifestyle changes in order to cope with health complications from diabetes. Simple things such as going out, family events, and holidays require more planning which can lead to frustration, anger, resentment and more stress. Family dynamics can change due to one family member dominating the attention of others, causing feelings of jealously, abandonment etc. The strain of managing a child with diabetes, especially in the early months and years after diagnosis can cause breakdown of parental relationships. Similarly, a parent diagnosed with diabetes may require help from his or her children thus altering the established family roles. Parents of young children frequently report interrupted sleeps for protracted periods (years) because they have to check their child’s blood glucose levels during the night to avoid life threatening hypos and, where necessary, take remedial actions such as waking the child for blood testing or treatment. For single parents this strain can be enormous, it can affect the parent’s ability to attend work, sustain employment etc. Children with diabetes can experience isolation, depression, unwillingness to interact with peers because of the fear that they might not be understood

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 128 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

and their situation taken seriously. Diabetes requires a different, new lifestyle and family dynamic, where everyone needs to play an active role. A number of families who have a relative with type 1 diabetes share the same emotionally charged reaction to this chronic illness when they were informed about the diagnosis. It adversely affected them during the first few weeks/months after the diagnosis. They were traumatised by getting to know the diagnosis of diabetes, expressing strong disbelief and heartbreak. Some of the fears that they face are related to the stigma diabetes carries within society that sometimes drive the person to avoid informing others such as peers and school staff about it. There is also the constant fear experienced by family members of whether the person was managing his or her condition properly and the complications such as physical injury that could arise if it is not controlled. Managing their child’s diabetes is a great responsibility and constant care needs to be given. As parents, they want their children with type 1 diabetes to be independent and learn to self- manage however, their fear and anxiety on the condition never fades away. Most families want to give the best to their child, whether it is insulin, new methods of glucose reading or others, This may sometimes put the family in a financial burden since all diabetes medication are costly. This can result in other stresses which some families cannot handle. Great burden for care-givers; they need to be available most of the time (if the patient is a child), they have to implement more daily tasks into their routine and that can be very draining, both physically and mentally; have to be aware that at all times acute diabetes complications e.g. hypoglycaemia can be deadly at any moment. Negative experience related to disease: without support of institutions, school, discriminations, could not go alone on school trips; parent must go with them; fears of hypoglycaemia; huge oscillations during the childhood; could not eat everything; education in school staffs, children, could not eat all kind of food; anxiety, stress; huge burden for all family.

Diabetes in Diabetes (T1 and T2) can have an impact on the wellbeing of mothers dur- pregnancy ing pregnancy and early motherhood. During pregnancy, blood glucose control can be very labile. Pregnant women can experience unpredictable swings in blood glucose levels. To ensure the health, and reduce the risk of diabetes related complications for both mother and baby, it is important that the mother maintains blood glucose levels within a safe target range.

Special groups Individuals with limited dexterity can encounter numerous challenges from of patients inability to manipulate insulin vials, syringes or cartridges, to handling blood glucose strips, using glucose meters, and self-injections. Those with limited visual acuity can face similar challenges requiring meters with larger text display windows and/or audible voice facilities. Poor visual acuity can make it difficult for the person to administer anti-diabetes treatments, making them reliant on third party assistance. For individuals with language, literacy and numeracy disabilities, the need to understand carbohydrate counting and ratios to match intake of carbohydrates to insulin dosage often proves problematic and difficult to master. Women have other factors such as pregnancy complications, menstruation (which causes a disruption of the glucose levels), social factors (such as not feeling comfortable in their own skin) and more. The lack of awareness in our societies may also cause issues in managing diabetes mellitus. Nu- merous time people pass comments such as, ‘you get diabetes from eating too much sugar’, ‘it is not a serious illness, not like cancer’, ‘but you do not

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 129 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

look like you have diabetes’, ‘have a good diet and exercise, then you can stop taking insulin’, ‘are you sure you want to do your shot here?’, ‘should you really be eating that?’, ‘I would die if I had to give myself shots’ and much more. Such comments make the person with diabetes either feel angry or sad. On the rare occasions, they feel empowered to educate others. No insulin-treated PWD has it easy. But some groups – such as mothers with diabetes that are active in the work force face issues of not taking enough care of themselves since they are taking care of their whole nuclear and often even, extended families. Children and adolescents have a tendency to hide from peers while injecting insulin or checking blood sugar, since they don’t want to be labelled abnormal by others.

Experiences with currently available health interventions: Self-monitoring blood glucose (SMBG) medical devices

Advantages Integral component in the management of insulin dependent diabetes; As part of the day-to-day routine it can help with necessary lifestyle and treatment choices; it helps to monitor for signs of hypo or hyperglycaemia and prevent any long-term complications from developing. It is therefore essential that the individual is taught how to carry out a test properly from onset as poor technique may lead to incorrect results which could lead to inaccurate medication. Disadvantages Frequent testing can be painful, inconvenient and difficult to achieve due to the person’s daily work routine. For example it is not always practical/ easy for a teacher to SMBG in front of pupils, leave in the middle of class to SMBG, or take appropriate action if their blood glucose levels are rising/ falling. It is not always easy for people to wash their hands in order to test when they are on the move; problems with finger sticks, pain, sensation lost; SMBG with tests – not enough (only 4 per day), some problems with device, visible, inconvenient in some situations, visible blood, not know the trend of glucose; problems on very low temperature outside, on the swimming pool (wrong value of glucose), indiscrete; not visible trends, stress from finger sticks; finger stics as short picture; problems with SMBG, pain, disinfections of hands; problems with small amount on blood, several finger sticks needed, problems with waste, financial, organisational issues Barriers for Access to essential meters and strips have been subjected to restrictions access across some parts of UK; Limiting access to essential test strips on the grounds of fiscal policy is both dangerous to patient safety and a false economy; Achieving even a small percentage reduction in blood glucose levels can have a marked reduction in the risk of developing secondary complications and improve the quality of life for patients; Access to appropriate SMBG devices and sufficient test strips is essential to achieving this.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 130 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Experiences with, and expectations of, the medical devices being assessed: continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) as personal, standalone systems in patients with diabetes mellitus treated with insulin

Experience with More control; Improved HBA1C; More readings; Less stress; More financial CGM and FGM burden since it is an expensive device for families who do not afford; Not all devices have an alarm that advises you when blood sugar is going dangerously low/high which most parents seek for their children with diabetes; Some individuals are allergic to the components used forcing them to stop using the device; Fingers are no longer sore/numb from all the pricks done; Doctors/healthcare professionals have a clear picture of what insulin needs to be adjusted if needed; Carers are at ease since when in doubt (for example during the night) one can easily check without waking up the child; For example, if surgery is needed with general anaesthesia, the professionals can continuously check the blood glucose and have a clear picture whether the blood will be going low/high; better quality of life with devices; much better control with CGM or FGM when big glucose oscillations are present; better regulation of glucose, better knowledge on food impact on glucose is provided; better sleep, better regulation of insulin dose and precision; CGM or FGM as film; normal life and quality, can be used more discreetly; must be affordable, life is much better controlled; important for children and young people, with positive impact on caregivers; positive impact and normal life, will prevent chronic complications; cost-benefit will be much better; experience with currently available medical devices; less stress; no need to stick on nigh, just scan, better quality of life; life becomes easier; trend of glucose is of utmost importance; hypo-hyper oscillations, no need for high number of finger-sticks, arrows and trend; education, communication, importance of insulin pumps; knowing when high or low is coming, proper reactions; not visible, easy to use; positive impact and normal life; more freedom to children and parents, more peaceful, more autonomy, importance of education and motivation; life becomes easier and more quality, could see the trends and could react properly; major emotional and social impact, changes to sleeping patterns; better quality of life; independence, more control and normal life.

Importance of education, importance of alarms, problems if sensors are lost, CGM or FGM for different periods of life, price should be lower, devices are visible, people must be educated and motivated; malfunctioning could occur; medical devices could not be available, problems related to unavailability, non-accessibility, finance, could be lost in water, need to be proper protection, not proper accessibility; should be more accessible; the price, many families could not paid for such devices; high cost and unavailability of rtCGM and FGM medical devices.

CGM The device provides data on how blood glucose may vary in response to advantages everyday tasks such as sleeping, after meals, physical activity or if unwell. CGM is potentially useful to anyone with diabetes especially those on multiple daily injections and insulin pump users. Some devices are integrated with the insulin pump with the capacity to suspend insulin delivery if the person’s glucose levels are falling. Results with CGM are not 100% accurate and regular blood glucose testing is still required. A key aspect of CGM is the alarm that warns of hypo or hyperglycaemia for people who have lost hypo awareness; improved many lives of people with type 1 diabetes. This is because it gives you a clear picture of how your bloods are 24/7 without having to prick your finger multiple times a day. It does not take as much

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 131 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

time as the normal meters therefore, people who are busy at school, work or in their daily lives, can check their blood whilst walking to their next class or meeting. It is discreet and painless. One can immediately know when a low or high is coming and can treat accordingly. This device also improves the HBA1C since a number of people start checking more often and do their best to keep their bloods in the correct range. Self-monitoring devices present a huge advantage in diabetes self-management. They provide you with more specific information at any given moment. That takes off a bit of the burden and fear of sudden changes in BG levels and the need to constantly prick your fingers. It can also be very useful for the care-giver who can be informed about the person’s status at all times even if they are not close by (most devices have this feature). In the end, they help you improve your self-management and directly prevent complications. The device is very easy to use and other persons with diabetes who use the device are all willing to help another set up the device and show them how it is used. The device is also easily covered by clothing, and if not it is a small sensor. Other people who do not know what it is usually think it is a device that helps you stop smoking. There where cases where others wanted to see what it is and took it of the person with diabetes. This caused the sensor to stop working, therefore the money spent to buy it was lost. Since there is no actual awareness about it, not a lot of people know what it is and usually do not ask

“I’m woken by an alarm if my levels are low ... My life is totally different, I can make plans ... I never want to go back to worrying that I might go to bed and not wake up” (Female, age 24, living with diabetes) “Technology has as made my life with T1 diabetes better. It gives the freedom and confidence to manage my diabetes and enjoy life”

Disadvantages Not always accurate; sometimes still have to check with a normal blood glucose reader; but most of the time, it is accurate. Self-monitoring devices present a huge advantage in diabetes self-management; some teenagers do not wear it during the summer so that it is not seen by others; others may see it as an eye sore, especially when you are part of the wedding party. Most self-monitoring devices are quite easy to use, though the process of insertion for some can be somewhat complicated. The biggest barrier though are the costs of using such a system since in many countries they are not reimbursed by health insurance; need to calibrate the device every 12 hours

FGM advantages User access results by scanning the sensor. Flash GM shows if blood glucose level are rising or falling, allowing the user to take remedial action sooner to avoid hypos or hyperglycaemia. Flash GM can be used for short- term investigations into an individual’s glucose levels where an individual is having difficulty in managing their condition or for longer diabetes management. Those utilising the device have said it is “life changing”; a new technology that is potentially life-changing for many people living with diabetes. Those utilising the device have indicated that they can see a clear ‘direction of travel’ in terms of understanding how their diabetes affects their blood glucose levels. They have described the device as ‘discreet’ allowing them to scan blood glucose levels easily in public without drawing too much attention and/or negative responses. People indicate they feel more confident in managing their diabetes. People are relieved not to have to do as many painful finger prick tests. Carers/parents can monitor blood glucose levels during activities/sleep without disturbing/waking the child/PLWD un-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 132 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

necessarily; the most useful aspect of the Flash GM device (Freestyle Li- bre®) is the directional arrow that tells them whether their blood glucose is rising or falling

Disadvantages The Flash GM does not have a built alarm facility to alert the users to highs and lows or suspend insulin delivery. Flash GM would not be appropriate in those who have irreversibly lost their hypoglycaemia awareness, which means they cannot recognise when their glucose levels are low. CGM devices which provide alarms are more appropriate in this cohort of PLWD. Users are still required to do ‘regular’ SGBM, especially if they are a driver (Flash GM is not recognised by Driving Vehicle Licensing Authority (DVLA) in the UK.; not currently available in all areas of the UK; Many people have chosen to self-fund the technology but this is not an option for others, particularly those who live in areas of multiple deprivation or are on low incomes.; the monthly cost since they are not covered by health insurance and that is the most important area that needs to be tackled in the future – making them more accessible; FGM could be lost during the training or pool use. “We are lucky to be able to afford the Libre, it has made a huge difference to my daughter’s lifestyle and her diabetes care. It’s a travesty that those who can’t afford it don’t have access to such a life changing technology.”

Additional information which patients believe would be helpful to the HTA researchers were used with aim help in the assessment of the value of health technologies, specifically CGM and FGM medical devices

Those who use SMBG devices should make sure they have access to and follow the guidelines for usage, checking equipment for any signs of damage or breakage regularly. Choice of SMBG devices and monitoring equipment should be based on individual need, and ease of use. This involves discussing options and making joint decisions based on individual need. Consultation should discuss use and frequency of testing and agree on targets for blood glucose levels.

Those commencing SMBG, CGM, or Flash GM should receive structured education to ensure they can maximise their use and benefit from such technology. Healthcare professionals working with people living with diabetes require training on use of the technology and will need to understand how to use and interpret glucose data in order to support patients.

People who use a Flash GM monitor still require regular testing and will need adequate supplies of test strips in order to perform the finger-prick test: If glucose levels are changing rapidly; If scanned levels do not correspond with physical symptoms; If the reader indicates low glucose; To meet current essential Driving and Vehicle Licensing Authority requirements.

Flash GM should not be considered an alternative to Continuous Glucose Monitoring (CGM). Peo- ple who meet the guidance and criteria for the use of CGM in people with Type 1 diabetes, as set out clearly in clinical guidance, should still be provided with CGM.

The patient should be able to make the treatment choice in collaboration with the endocrinologist. This is especially true for these systems e.g. does the patient want a pre-calibrated system or a one where you have to calibrate it yourself, is the ability to have alarms crucial or not, etc. Patients should also go through an educational course at the start of using these devices. They must be aware that they make mistakes and be taught what to do when malfunction occurs.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 133 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Key messages from patients

Hypoglycaemia (severe/moderate) is one of most significant and challenging aspects associated with diabetes (T1 and T2). Hypoglycaemia has an enormous impact on the person’s quality of life. There is related cost not only to NHS in terms of resources (ambulance call-outs, hospital admission, increased length of stay associated with poor management when admitted for non-diabetes- related reasons) but also to the individual in terms of stress and anxiety caused, working days lost, missed education and effect on family relationships.

Current medical devices for self-monitoring blood glucose (SMBG) provide just a snap shot of blood glucose at a single point time, and the results obtained are dependent on the skills/technique and circumstances of the users. Devices can produce aberrant readings, caused by defects in calibration or test strip accuracy (finite shelf life), and lack of optimum conditions for testing (e.g. handwashing facilities). There is also the potential for the misinterpretation of readings due to in- adequacies in size and configuration of display windows.

The medical devices such as Continuous Glucose Monitoring (CGM) and Flash Glucose Monitor- ing (Flash GM) are beneficial because they can provide a more comprehensive picture of glycaemic control and trends. CGM devices can alert the users of oncoming episodes of potentially de- bilitating hypo and hyperglycaemia, especially at night or when sleeping. They are of particular benefit in those with loss of hypo awareness. The data recorded on such devices allows PLWD and their healthcare professionals to view events, activities, examine their impact on blood glucose level, and make informed joint decisions to help improve self-management and quality of life. Flash technology is new, but the experience of users indicates that when integrated in normal/usual care the clinical and quality of life benefits are significant. These benefits include: Reducing the need for painful and inconvenient finger prick glucose monitoring; Helping people to better manage their diabetes and engage people with self-management of their condition; Reducing stress and anxiety for people with diabetes and their families; Reducing hypoglycaemia and increasing time in optimum range; Improving HbA1c to reduce the risk of developing costly complications.

Further comparative study comparing Flash GM and traditional CGM is required to help us understand the advantages of Flash GM and its role in diabetes management, both T1 and T2.

Technology associated with the management of diabetes is developing rapidly and constantly changing. It is essential to be vigilant and ensure that diabetes services are tested in the future, fit for purpose, and enable those living with diabetes to effectively self-manage, avoid unnecessary complications, and live well with the condition.

Current medical devices have been effective over the past years, but with the increasingly fast- paced lifestyle of today, they are outdated. People treated with insulin need these new medical devices in order to have a better, more controlled life. There are thousands of people using these devices and they are able to lead a “normal” life, without the fear of how their blood is, because they can check discreetly in a single second. These devices are very useful and should be accessible by all by making them affordable and available in different countries.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 134 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

9 DISCUSSION

The purpose of this assessment was to jointly produce a rapid core HTA on real-time continuous glucose monitoring (rtCGM) and flash glucose monitoring (FGM) systems (with CE marks) using the EUnetHTA HTA Core Model® for Relative Effectiveness Assessment (REA) [10, 11] and to initiate local production in different member states based on this assessment. Our SR provides a narrative summary of the best available scientific evidence. We do not provide any recommendations; this is left to national/regional and local HTA processes, appraisal processes, and decision-making.

Assessment of medical devices is in general more challenging than pharmaceuticals for a number of reasons. Drummond et al. 2009 reported six specific reasons for that [42]. Assessing nonpharmacological treatment, including medical devices, raises specific methodological issues, such as difficulties of blinding [42].

In the EUnetHTA Guideline “Therapeutic medical devices”, three major issues relevant for REA in which MDs differ from drugs were noted: the short life-cycle and rapid and predominantly incremental development; stronger user dependency of the treatment effect and learning curves; evaluation of long-term effects for high risk devices. The 1st recommendation is related to specifics of HTA on MD and stated that HTA on medical device interventions should generally be done with currently established methods for finding, selecting, analysing, synthesizing, and interpreting evidence on clinical effectiveness; the need for specific methods mainly derives from the incremental development of MDs and their user and context dependency, and some implications of the physical mode of action [119]. If blinding is impossible, at least blinded endpoint evaluation is recommended.

Recently, in 2017, the CONSORT Statement extension for RCTs of nonpharmacological treatment was published with the aim of helping authors to increase transparency of their reports and facilitate an accurate interpretation of study results [43]. Two of the three new items added address description of attempts to limit bias if blinding is not possible and whether and how adherence of participants to intervention is assessed or enhanced. Blinding is one of important item evaluated in the Risk of Bias (RoB) tool [43]; if there no written attempts to limit bias if blinding is not possible, RoB will be described as high.

Schnell-Inderst et al. 2018 reviewed existing guidance on the methods for evaluation of comparative effectiveness of therapeutic medical devices and developed ten recommendations for assessment [120], taking into account the more complex composition of the intervention, its rapid incremental development, dependency of treatment effects on contextual factors, and user proficiency. The same authors published recommendations for primary studies evaluating therapeutic medical devices to support improvement of the evidence base for health technology assessments [121]. Motte et al. 2017 concluded that no existing reporting guideline is fully suitable for implantable medical devices [122].

Another important challenge is that trials assessing devices are often not standardised in terms of methodology and what clinical outcome units is used, although this may vary between therapeutic areas. In our case, this was a major issue and limited pooling data into MA to a large extent.

The GRADE approach is a system for rating the quality of a body of evidence in SRs and other evidence synthesis (such as HTA and guidelines) and grading recommendations in healthcare [44]. According the GRADE Handbook, systematic reviewers should provide a comprehensive summary of the evidence but they should not typically include healthcare recommendations. Separating judgements about quality or certainty of evidence from judgements about the strength of recom-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 135 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin mendations is important because both the direction and the strength of recommendations may be modified after taking into account factors other than the strength of evidence, such as implications for resource utilization, equity, acceptability, and feasibility of alternative management strategies [44].

TEC and CUR Domains

A number of different systems for real-time continuous glucose monitoring (CGM) have been available for approximately 10 years. The number of medical devices for glucose monitoring available on the market is rapidly growing, in terms of development of the new medical devices and as well for development of new generations of existing devices, which are being improved and provide more options for the user [4, 140, 141].

People diagnosed with DM type 1, type 2 DM, or gestational diabetes and who are willing and able to monitor and manage their DM themselves are the target population of this assessment. Hypo- glycaemia and hypoglycaemia unawareness are recognized as important limiting factors in the glycaemic management of type 1 and type 2 diabetes [89]. Hypoglycaemia may be inconvenient or frightening to patients with diabetes. Severe hypoglycaemia may be recognized or unrecognized and can progress to loss of consciousness, seizure, coma, or death. Clinically significant hypoglycaemia can cause acute harm to the person with diabetes or others, especially if it causes falls, motor vehicle accidents, or other injury. Young children with type 1 diabetes and the elderly, including those with type 1 and type 2 diabetes, are noted as particularly vulnerable to clinically significant hypoglycaemia because of their reduced ability to recognize hypoglycaemic symptoms and effectively communicate their needs [89]. Hypoglycaemia mortality estimates ranging from 4 to 10 percent of deaths of patients with type 1 diabetes. Hypoglycaemia mortality rates in type 2 diabetes are currently unknown, but fatal hypoglycaemia has been documented in type 2 diabetes. Se- vere hypoglycaemia may also be associated with an increased risk of cardiovascular disease in patients with type 2 diabetes. Recurrent severe hypoglycaemia has been associated with cognitive impairment in young children or older persons with diabetes. Nocturnal hypoglycaemia is a particular problem, which can lead to disruption of sleep and delays in correction of the hypoglycaemia [17]. Hypoglycaemia unawareness could severely influence diabetes control and quality of life [89].

EFF Domain outcomes

We included twelve RCTs that reported on the use of CGM devices (as standalone devices or with insulin pumps) and FGM devices compared with SMBG. Only one small head-to-head study was identified, comparing CGM and FGM for an 8-week follow-up [34]. Due to heterogeneity of populations, interventions, and outcome measures MA was done only for one outcome, i.e. HbA1c change from baseline to the end of the study, pooling the data from 2 RCTs (DIAMOND and GOLD trials) [25, 29]. Our systematic review provides a narrative summary. Notably, endpoints in our assessment were not indicated as primary or secondary endpoints as opposed to the included primary studies. The format of the EUnetHTA Core Model was followed [10], and the prioritisation of endpoints was done by the EUnetHTA assessment team for the present evaluation.

RoB was important for all RCTs both at the study level and outcome level. Certainty of evidence varied from moderate in the case of HbA1c changes from baseline to the end of the study (two RCTs DIAMOND and GOLD) [25, 29], to low or very low for other outcomes such as time spent in normo- and hypoglycaemia, hypoglycaemia and severe hypoglycaemia events, QoL and user satisfaction, and local AEs.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 136 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

HbA1c changes from baseline to the end of the study

Meta-analysis of the data pooled from 2 RCTs (with T1DM patients on MDII treatment) comparing CGM vs SMBG (Beck 2017 T1DM and Lind 2017) [25, 29] demonstrated a statistically significant benefit of CGM in reducing HbA1c levels (heterogeneity among included RCTs was low). CGM led to a statistically significantly larger reduction in HbA1c levels than SMBG in the majority of studies including MDII patients (Beck 2017; Beck 2017 T2DM; Lind 2017; Ruedy 2017) [25, 26, 29, 37] and in the two studies including MDII and CSII patients (Battelino 2011, Riveline 2012) [24, 36]. In one study with MDII patients (Heinemann 2018) [13], two studies with MDII and CSII patients (Mauras 2012; van Beers 2016) [31, 38], and in one study including CSII patients (Ly 2013) [30] no statistically significant differences was identified. In addition, no statistically significant difference was found in one head-to-head study comparing CGM vs FGM (Reddy 2018) [34] and two RCTs comparing FGM vs SMBG in T1DM (Bolinder 2016) [27] and T2DM patients (Haak 2017) [28] on MDII and CSII treatments.

Four studies with impaired hypoglycaemia awareness patients (Heinemann 2018; Ly 2013; Van Beers 2016;, related to rtCGM vs SMBG and Reddy 2018, comparing CGM vs FGM) [13, 30, 34, 38] also provided no statistically significant difference between intervention and control groups, but in these studies the change in HbA1c levels was not a primary outcome and participants were selected on the basis of IHA or previous severe hypoglycaemia events.

Time spent in the target glycaemic range

Results from three randomized controlled trials comparing CGM with SMBG (Battelino 2011; Beck 2017 T1 patients; van Beers 2016) [24, 25, 38] favoured continuous glucose monitoring over control (p<0.05).The certainty of the evidence varied from low to very low. In one head-to-head trial comparing CGM vs FGM for after an 8-week period (Reddy et al, 2017) [34], no statistically significant difference was observed. The same applied for FGM vs SMBG in T2DM patients (Haak et al 2017 REPLACE) [28]. In T1DM patients, Bolinder et al (Bolinder 2016 IMPACT) [27], reported a statistically significant increase in the intervention group compared with the control group at 6 months.

Time spent in hyperglycaemia

In studies comparing CGM vs SMBG, time spent in the hyperglycaemic range was statistically significantly reduced using CGM in T1DM patients in one study (Beck 2017 T1DM) [25] but not in another study (Heinemann 2018) [13].Two studies remained inconclusive: one with T1DM patients (Lind 2017) [29] and one with T2DM patients (Beck 2017) [26], as no p-value was reported. No statistically significant differences were found for one small study comparing CGM vs FGM (Reddy et al, 2018) [34], as well as in one RCT comparing FGM and SMBG in T2DM patients (Haak et al 2017 REPLACE) [28, 41]. Statistically significant differences were found for the RCT that included T1DM patients for this comparison (Bolinder et al 2016) [27].

Time spent in hypoglycaemia, and hypoglycaemia and severe hypoglycaemia events

For rtCGM vs SMBG patients, statistically significant results favoured rtCGM in studies with MDII patients (Beck 2017 T1 patients; Heinemann 2018) [13, 25], MDII and CSII patients combined (van Beers 2016, Battelino 2011) [38, 24], and CSII patients (Ly 2013) [30]. The difference was not statistically significant in one study (Mauras 2012) [31] and not reported in another study (Lind 2017) [29].The certainty of the evidence varied from low to very low. One small head-to-head trial comparing rtCGM vs FGM (Reddy et al, 2017) [34], as well as two RCTs comparing FGM vs SMBG in T1DM and T2DM patients (Bolinder et al 2016; Haak et al 2017) [27, 28] reported statistically significant decreased time in hypoglycaemia in the intervention groups.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 137 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

There was a statistically significant difference in hypoglycaemic events between the rtCGM and control groups in MDII patients (Heinemann et al. 2018 and Riddlesworth et al. 2017) [13, 35], in MDII and CSII treatments combined with CSII patients (van Beers 2016) [38], and CSII patients assesses separately (Ly 2013) [30], but not in the studies by Battelino et al 2011 [24], Mauras et al 2012 [31] and Riveline et al 2012 [36].

One small head-to-head trial comparing rtCGM vs FGM (Reddy et al, 2017) [34], as well as two RCTs (Bolinder 2016; Haak 2017) [27, 28] with T1 DM and T2DM patients respectively, and comparing FGM vs SMBG patients, reported statistically significant decreased hypoglycaemia events in the intervention group.

Four studies involved T1DM patients with IHA or previous severe hypoglycaemia events: three comparing rtCGM with SMBG (Heinemann 2018 in MDII patients; Ly 2013 on CSII patients and van Beers 2016 in patients on MDII and CSII together) [13, 30, 38] and one head-to-head study comparing rtCGM vs FGM (Reddy et al, 2017) [34] showed statistically significant results with rtCGM in terms of time spent in hypoglycaemia as well as hypoglycaemia and severe hypoglycaemia events. These results (with low or very low certainty of evidence) are of high clinical importance for both patients on MDII therapy and CSII patients, because both conditions predispose such patients to future hypoglycaemia episodes [123, 124]. Karges et al 2017 reported results on the association of insulin pump therapy vs insulin injection therapy with severe hypoglycaemia, ketoacidosis and glycaemic control among children, adolescents, and young adults with T1DM, but the use of rtCGM was not analysed in this study [125].

In April 2018, results from the GOLD-3 study (Olafsdottir et al) were published, showing that rtCGM reduced time in both nocturnal and daytime hypoglycaemia in persons with T1DM treated with MDI and improved hypoglycaemia-related confidence, especially in social situations, thus contributing to greater well-being and quality of life [126].

Patient-reported outcome, quality of life (QoL) measures and user satisfaction

Results were inconsistent across studies, probably due to differences in types of outcomes and/or survey tools. The certainty of the evidence for these outcomes varied from low to very low.

None of the SRs identified in our literature search could be used and updated due to different scope of assessment and different health technologies available at the market at the time of assessment [127, 64, 128, 129, 63]. The most recent SR [63] showed that rtCGM (using Medtronic or Dexcom devices that has been included in peer-reviewed publications since 2010) was more effective than SMBG in managing T1DM for some outcomes, such as time spent in the target glucose range and time spent outside the target glucose range (moderate certainty in this evidence). Authors were less certain that rtCGM would reduce the number of severe hypoglycaemic events. Because the scope of this assessment was limited to rtCGM devices by manufacturers with Health Canada li- cences at the time of writing, devices such as the Dexcom SEVEN and the Abbott FreeStyle Nav- igator were not included in this HTA.

Riemsma 2017 [129] concluded that the Veo system was better than the other systems considered in reducing hypoglycaemic events. Future trials in adolescent and adult populations, but specifically in children, should include longer terms follow-up and ratings on the European Quality of Life-5 Dimensions scale at various time points.

Pickup 2011 [127] SR and MA of individual patient data (with the primary outcome of final HbA1c percentage and area under the curve of hypoglycaemia <3.9 mmol/L) during either treatment, followed by one-step meta-regression exploring patient-level determinants of HbA1c and hypogly-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 138 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin caemia), concluded that rtCGM was associated with a significant reduction in HbA1c percentage, which was greatest in those with the highest HbA1c at baseline and those who most frequently used the sensors. Exposure to hypoglycaemia was also reduced during continuous glucose monitoring.

Langendam 2012 [64] concluded that there is limited evidence for the effectiveness of rtCGM use in children, adults, and patients with poorly controlled diabetes. The largest improvements in glycaemic control were seen for sensor-augmented insulin pump therapy in patients with poorly controlled diabetes who had not used an insulin pump before. There are indications that higher compliance of wearing the CGM device improves HbA1c to a larger extent. Yeh et al 2012) concluded that for glycaemic control, rt-CGM was superior to SMBG in adults with T1 DM [128].

Parker et al, 2018 [142] and Dunn et al, 2018 [118] found in real-time CGM users that time in range increased as screen view frequency increased from the lowest to highest frequency cohorts. RtCGM should not be introduced in patients who are unwilling to use it consistently or are incapa- ble of using it beneficially [140].

No RCTs were found investigating devices assessed with CE mark authorisation in pregnancy. Only one RCT included only children [31].

One RCT, CONCEPTT, published by Feig et al 2017 [130], presented for the first time that the use of CGM (Guardian REAL-Time or MiniMed™ Minilinksystem, both Medtronic, Northridge, CA) during pregnancy, in women with type 1 diabetes, which was associated with improved neonatal health outcomes attributed to reduced exposure to maternal hyperglycaemia. The numbers of pregnant women needed to treat with CGM to prevent one new-born complication were six for both neonatal intensive care admission and large for gestational age, and eight for neonatal hypoglycaemia. In this multicentre international trial, authors showed that women had a small but significantly greater reduction in HbA1c levels than the control participants, accompanied by increased time in target, reduced hyperglycaemia, and less glycaemic variability at 34 weeks’ gestation.

SAF Domain outcomes

All but one of the included studies reported AEs [13, 24-27, 29-32, 36-41]. Systemic SAEs were reported in the majority of the studies, as well as those related to severe hypoglycaemia and ketoacidosis, but investigators found the majority of them were not related to the intervention. Devices (or procedures) related local AEs were reported in 5 RCTs (3 related to FGM device) [27-30, 32] and 3 nRCTs (all related to FGM device) [39-41]. According to Haak et al 2017 [41], anticipated symptoms referred to those typically expected using a sensor device and were equal to symptoms normally experienced with blood glucose finger-stick testing, e.g., pain, bleeding, bruising, and resolved without medical intervention.

Allergic reactions were also reported. In the literature data, Herman et al 2017 [131] reported a case series of allergic contact dermatitis cause by FreeStyle Libre®; all 15 patients (children and adults) were tested and reacted to the adhesive part of the sensor. Twelve patients were shown to be sensitized to isobornyl acrylate (presented in the sensor) as the relevant culprit allergen. Some patients needed to discontinue use of the device because complete avoidance of or a substantial decrease in exposure to the allergen responsible in contact allergy was the only effective solution [132]. Allergic contact dermatitis should not be confused with contact irritation at application site associated with a burning or stinging sensation instead of a profound itch [132]. Jadviscokova et al 2017 [133] reported that around 18% of patients using CGM devices may suffer from hypersensi-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 139 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin tivity reactions and Schwensen et al 2016 [134] reported sensitization to ethyl cyanoacrylate from Dexcom G4® Platinum glucose sensor.

All but one of the included studies reported AEs. In the study by Reddy 2018 [34], AEs were neither mentioned as an aim or outcome, nor reported and were not pre-specified as primary or secondary endpoints.

In the sample of the RCTs analysed in a study published by Huic et al, 2011 [135], in which technologies other than pharmaceuticals were presented in 30% of the total sample, serious and non- serious AEs were mentioned in more than 80% of the published articles.

AEs data reporting should be according to evidence-based reporting guidelines, specifically the CONSORT Statement extension on better reporting of harms in RCTs and trials assessing nonpharmacological treatments [136], as well as the PRISMA harm checklist [137]. New recommendations to improve AE reporting on medical devices in clinical trial publications, like those recently published on pharmaceuticals, are clearly needed [138].

Patient involvement

In our assessment, adults, children and parents with T1DM reported very positive experiences with rtCGM and FGM devices. Life became easier with better life quality. Patients could see the glucose trends and could react properly. The use of these devices had a major emotional and social impact, changed sleeping patterns, provided independence, and gave more control and contributed to a normal life. The importance of education and motivation should be emphasized, as well as the most important barriers and the high cost and unavailability of these medical devices in some countries.

Lawton et al 2018 [139] performed an in-depth review with a total of 24 adults, adolescents, and parents (using insulin pumps) who had used CGM ≥4 weeks. They found CGM was an empower- ing tool due to effortlessly access to blood glucose data. Trend arrows enabled them to see blood glucose rising or dropping and at what speed, allowing them to take preventive action related to hypo- or hyperglycaemia as well aid short-term lifestyle planning. Having glucose data on a continuous basis allowed them to develop better understanding on how food, activities, and insulin im- pacted blood glucose. Some difficulties were recognised however, related to calibration, problems inserting and/or removing the device, and device failure. Regarding the alarms, authors found that these could cause frustration and disrupted sleep, but on the other hand they provided comfort and reassurance by alerting in timely manner, reinforcing a sense of hypoglycaemic safety.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 140 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

10 CONCLUSION

Based on a narrative summary of studies with important risk of biases and moderate to very low certainty of evidence results, use of CGM devices assessed was associated with a reduction in HbA1c in majority of the studies that included MDII treated patients and in the two studies with MDII and CSII treated patients; both CGM and FGM devices assessed were associated with reduction in hypo- and hyperglycaemia outcomes and improved treatment satisfaction in patients with T1 or T2DM, compared with SMBG. No RCTs were found investigating devices assessed with CE mark authorisation in pregnancy. Only one RCT included solely a child population [31], only one study was head-to-head [34], and only four studies were performed on patients with impaired hypoglycaemia awareness [13, 30, 34, 38]. In all RCTs, patients were followed up from 8 weeks to 12 months. Due to the heterogeneity of populations, interventions, and outcomes measures, MA was done only for one outcome, i.e. HbA1c change from baseline to the end of the study (the DIA- MOND and GOLD studies) [25, 29], showing statistically significant results in favour of CGM.

All four studies involving T1DM patients with IHA or previous severe hypoglycaemia events, including the single head-to-head study which compared CGM vs FGM with a follow-up of 8 weeks [34], showed statistically significant results which favoured CGM in terms of time spent in hypoglycaemia as well as hypoglycaemia and severe hypoglycaemia events. These results (of low to very low evidential certainty) are of high clinical importance for both patients on MDII therapy and CSII patients, because both conditions predispose such patients to future hypoglycaemia episodes.

Further high quality head-to-head studies on long-term relative effectiveness and safety comparing CGM and FGM devices are needed, especially in children and pregnancy.

Adults, children and parents with T1 DM reported very positive experiences with CGM and FGM devices; different benefits were shown as well as the most important barriers – high cost and unavailability of these medical devices in some countries.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 141 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

11 REFERENCES

[1] American Diabetes Association. Introduction:Standards of Medical Care in Diabetes – 2018.2018. S1-S2 p.

[2] Danne T, Nimri R, Battelino T, Bergenstal RM, Close KL, DeVries JH, et al. International Consensus on Use of Continuous Glucose Monitoring. Diabetes Care. 2017;40(12):1631-40.

[3] Pfutzner A, Klonoff DC, Pardo S, Parkes JL. Technical aspects of the Parkes error grid. J Diabetes Sci Technol. 2013;7(5):1275-81.

[4] Rodbard D. Continuous Glucose Monitoring: A Review of Recent Studies Demonstrating Improved Glycemic Outcomes. Diabetes technology & therapeutics. 2017;19(S3):S25-s37.

[5] National Institute for Health and Care Excellence. Type 1 diabetes in adults: diagnosis and management. NICE guideline. 2015. Last updated: July 2016.

[6] Mankin G. Freestyle Libre Flash Glucose Monitoring System. Regional Drug & Therapeutic Centre (Newcastle), Northern Treatment Advisory Group (NTAG). 2016.

[7] Ryden L, Grant PJ, Anker SD, Berne C, Cosentino F, Danchin N, et al. ESC Guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the EASD: the Task Force on diabetes, pre-diabetes, and cardiovascular diseases of the European Society of Cardiology (ESC) and developed in collaboration with the European Association for the Study of Diabetes (EASD). European heart journal. 2013;34(39):3035-87.

[8] Obley A, Hackett R, Mosbaek C, King V, Shaffer W. Coverage guidance: Continuous glucose monitoring in diabetes mellitus. Portland, OR: Center for Evidence-based Policy, Oregon Health & Science University. 2017.

[9] Wright N, Ng S, Agwu S, Adolfsson P, Drew J, Pemberton J, et al. ACDC Clinical Guideline. A Practical Approach to the Management of Continuous Glucose Monitoring (CGM)/Real-Time Flash Glucose Scanning (FGS) in Type 1 Diabetes Mellitus in Children and Young People Under 18 years. 2017.

[10] European network for Health Technology Assessment (EUnetHTA). HTA Core Model. 2016.

[11] European network for Health Technology Assessment (EUnetHTA). HTA Core Model for Rapid Relative Effectiveness (REA) Assessments (4.2).2015. [cited 2017 Dec 10]. Available from: http://meka.thl.fi/htacore/model/HTACoreModel_ForRapidREAs4.2.pdf .

[12] Forlenza GP, Argento NB, Laffel LM. Practical Considerations on the Use of Continuous Glucose Monitoring in Pediatrics and Older Adults and Nonadjunctive Use. Diabetes technology & therapeutics. 2017;19(S3):S13-s20.

[13] Heinemann L, Freckmann G, Ehrmann D, Faber-Heinemann G, Guerra S, Waldenmaier D, et al. Real-time continuous glucose monitoring in adults with type 1 diabetes and impaired hypoglycaemia awareness or severe hypoglycaemia treated with multiple daily insulin injections (HypoDE): a multicentre, randomised controlled trial. Lancet (London, England). 2018;391(10128):1367-77.

[14] Canadian Agency for Drugs and Technologies in Health. Blood Glucose Monitors and Test Strips: A Review of the Comparative Clinical Evidence and Cost-Effectiveness. 2011.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 142 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[15] American Diabetes Association. Classification and Diagnosis of Diabetes. Diabetes Care. 2017;40(Supplement 1):S11-S24.

[16] World Health Organization. Global report on diabetes. 2016.

[17] Cryer PE. Management of hypoglycemia during treatment of diabetes mellitus. UpToDateMay 2018.

[18] Wan X, Wang W, Liu J, Tong T. Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC medical research methodology. 2014;14:135.

[19] European network for Health Technology Assessment (EUnetHTA). Updated EUnetHTA methodological Guidelines. 2015. [cited 2017 Dec 10]. Available from: http://wwweunethtaeu/eunethta-guidelines.

[20] Higgins J, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. The Cochrane Collaboration. 2011. [updated 2011 Mar; cited 2017 Dec 10]. Available from: http://handbook-5-1.cochrane.org/.

[21] Guyatt GH, Oxman AD, Vist GE, Kunz R, Falck-Ytter Y, Alonso-Coello P, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ (Clinical research ed). 2008;336(7650):924-6.

[22] Balshem H, Helfand M, Schunemann HJ, Oxman AD, Kunz R, Brozek J, et al. GRADE guidelines: 3. Rating the quality of evidence. Journal of clinical epidemiology. 2011;64(4):401-6.

[23] HTAi Patient Group Submission template – Non Medicines HTA. [cited 2017 Dec 10]. Available from: https://www.htai.org/interestgroups/patient-and-citizen- involvement/resources/for-patients-and-patient-groups/.

[24] Battelino T, Phillip M, Bratina N, Nimri R, Oskarsson P, Bolinder J. Effect of continuous glucose monitoring on hypoglycemia in type 1 diabetes. Diabetes Care. 2011;34(4):795-800.

[25] Beck RW, Riddlesworth T, Ruedy K, Ahmann A, Bergenstal R, Haller S, et al. Effect of Continuous Glucose Monitoring on Glycemic Control in Adults With Type 1 Diabetes Using Insulin Injections: The DIAMOND Randomized Clinical Trial. Jama. 2017;317(4):371-8.

[26] Beck RW, Riddlesworth TD, Ruedy K, Ahmann A, Haller S, Kruger D, et al. Continuous Glucose Monitoring Versus Usual Care in Patients With Type 2 Diabetes Receiving Multiple Daily Insulin Injections: A Randomized Trial. Annals of internal medicine. 2017;167(6):365-74.

[27] Bolinder J, Antuna R, Geelhoed-Duijvestijn P, Kroger J, Weitgasser R. Novel glucose-sensing technology and hypoglycaemia in type 1 diabetes: a multicentre, non-masked, randomised controlled trial. Lancet (London, England). 2016;388(10057):2254-63.

[28] Haak T, Hanaire H, Ajjan R, Hermanns N, Riveline J-P, Rayman G. Flash Glucose-Sensing Technology as a Replacement for Blood Glucose Monitoring for the Management of Insulin- Treated Type 2 Diabetes: a Multicenter, Open-Label Randomized Controlled Trial. Diabetes Therapy. 2017;8(1):55-73.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 143 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[29] Lind M, Polonsky W, Hirsch IB, Heise T, Bolinder J, Dahlqvist S, et al. Continuous Glucose Monitoring vs Conventional Therapy for Glycemic Control in Adults With Type 1 Diabetes Treated With Multiple Daily Insulin Injections: The GOLD Randomized Clinical Trial. Jama. 2017;317(4):379-87.

[30] Ly TT, Nicholas JA, Retterath A, Lim EM, Davis EA, Jones TW. Effect of sensor-augmented insulin pump therapy and automated insulin suspension vs standard insulin pump therapy on hypoglycemia in patients with type 1 diabetes: a randomized clinical trial. Jama. 2013;310(12):1240-7.

[31] Mauras N, Beck R, Xing D, Ruedy K, Buckingham B, Tansey M, et al. A randomized clinical trial to assess the efficacy and safety of real-time continuous glucose monitoring in the management of type 1 diabetes in young children aged 4 to <10 years. Diabetes Care. 2012;35(2):204-10.

[32] Oskarsson P, Antuna R, Geelhoed-Duijvestijn P, Krger J, Weitgasser R, Bolinder J. Impact of flash glucose monitoring on hypoglycaemia in adults with type 1 diabetes managed with multiple daily injection therapy: a pre-specified subgroup analysis of the IMPACT randomised controlled trial. Diabetologia. 2018;61(3):539-50.

[33] Polonsky WH, Hessler D, Ruedy KJ, Beck RW. The Impact of Continuous Glucose Monitoring on Markers of Quality of Life in Adults With Type 1 Diabetes: Further Findings From the DIAMOND Randomized Clinical Trial. Diabetes Care. 2017;40(6):736-41.

[34] Reddy M, Jugnee N, El Laboudi A, Spanudakis E, Anantharaja S, Oliver N. A randomized controlled pilot study of continuous glucose monitoring and flash glucose monitoring in people with Type 1 diabetes and impaired awareness of hypoglycaemia. Diabetic medicine: a journal of the British Diabetic Association. 2018;35(4):483-90.

[35] Riddlesworth T, Price D, Cohen N, Beck RW. Hypoglycemic Event Frequency and the Effect of Continuous Glucose Monitoring in Adults with Type 1 Diabetes Using Multiple Daily Insulin Injections. Diabetes Ther. 2017;8(4):947-51.

[36] Riveline JP, Schaepelynck P, Chaillous L, Renard E, Sola-Gazagnes A, Penfornis A, et al. Assessment of patient-led or physician-driven continuous glucose monitoring in patients with poorly controlled type 1 diabetes using basal-bolus insulin regimens: a 1-year multicenter study. Diabetes Care. 2012;35(5):965-71.

[37] Ruedy KJ, Parkin CG, Riddlesworth TD, Graham C. Continuous Glucose Monitoring in Older Adults With Type 1 and Type 2 Diabetes Using Multiple Daily Injections of Insulin: Results From the DIAMOND Trial. J Diabetes Sci Technol. 2017;11(6):1138-46.

[38] van Beers CA, DeVries JH, Kleijer SJ, Smits MM, Geelhoed-Duijvestijn PH, Kramer MH, et al. Continuous glucose monitoring for patients with type 1 diabetes and impaired awareness of hypoglycaemia (IN CONTROL): a randomised, open-label, crossover trial. The lancet Diabetes & endocrinology. 2016;4(11):893-902.

[39] Bailey T BB, Christiansen MP, Klaff LJ, Alva S. The Performance and Usability of a Factory- Calibrated Flash Glucose Monitoring System. Diabetes Technol Ther 2015;;17(11):787-94.

[40] Edge J, Acerini C, Campbell F, Hamilton-Shield J, Moudiotis C, Rahman S, et al. An alternative sensor-based method for glucose monitoring in children and young people with diabetes. Archives of disease in childhood. 2017;102(6):543-9.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 144 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[41] Haak T, Hanaire H, Ajjan R, Hermanns N, Riveline JP, Rayman G. Use of Flash Glucose- Sensing Technology for 12 months as a Replacement for Blood Glucose Monitoring in Insulin-treated Type 2 Diabetes. Diabetes Ther. 2017;8(3):573-86.

[42] Drummond M, Griffin A, Tarricone R. Economic evaluation for devices and drugs – same or different? Value in health: the journal of the International Society for Pharmacoeconomics and Outcomes Research. 2009;12(4):402-4.

[43] Boutron I, Altman DG, Moher D, Schulz KF, Ravaud P. CONSORT Statement for Randomized Trials of Nonpharmacologic Treatments: A 2017 Update and a CONSORT Extension for Nonpharmacologic Trial Abstracts. Annals of internal medicine. 2017;167(1):40-7.

[44] Schünemann H, Brożek J, Guyatt G, Oxman A, The GRADE Working Group. GRADE Handbook for Grading Quality of Evidence and Strength of Recommendations. Available from: gdtguidelinedevelopmentorg/app/handbook/handbookhtml. Updated October 2013.

[45] Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS medicine. 2009;6(7):e1000100.

[46] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ (Clinical research ed). 2003;327(7414):557-60.

[47] Whitlock EP, Lin JS, Chou R, Shekelle P, Robinson KA. Using existing systematic reviews in complex systematic reviews. Annals of internal medicine. 2008;148(10):776-82.

[48] Robinson KA, Whitlock EP, O'Neil ME, Anderson JK, Hartling L, Dryden DM, et al. AHRQ Methods for Effective Health Care. Integration of Existing Systematic Reviews. Rockville (MD): Agency for Healthcare Research and Quality (US); 2014.

[49] Dexcom. G4® PLATINUM CGM System, User’s guide.

[50] Dexcom. Dexcom G5® Mobile CGM System. User’s guide.

[51] Dexcom. Dexcom G6® CGM System. User’s guide.

[52] Abbott. Freestyle Navigator II® Continious Glucose Monitoring System User’s Manual.

[53] Medtronic. GuardianTM Connect Transmitter Instructions for use.

[54] Medtronic. EnliteTM Sensor Instructions for use.

[55] Senseonics. Eversense User Guide – A guide for using the Eversense Continuous Glucose Monitoring System. 2018. [cited 2018 Feb 27]. Available from: https://ouseversensediabetescom/wp-content/uploads/ 2017/08/LBL-0902-12- 101_Rev_A_Eversense_User_Guide_mmolL_ZAFpdf;2722018.

[56] Senseonics. Eversense XL User Guide. 2017. [cited 2018 Feb 28]. Available from: https://ous.eversensediabetes.com/wp-content/uploads/2018/03/LBL-1202-12- 101_Rev_B_RM_RM_Eversense_User_Guide_mmol_ZAF.pdf.

[57] Abbott. FreeStyle Libre® User’s Manual.

[58] EUnetHTA JA3 WP4. Project plan. Continuous glucose monitoring (CGM real-time) and flash glucose monitoring (FGM) as personal, standalone systems in patients with diabetes mellitus treated with insulin. Project ID: OTJA08. 2017.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 145 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[59] Medtronic. Paradigm™ Veo User Guide Instructions for use Paradigm™ Veo Pump system, integrated with MiniLink® transmitter and Enlite Glucose Sensor.

[60] Medtronic. MiniMed™ 640G System User Guide Instructions for use.

[61] Tandem Diabetes Care. t:slim X2 TM Insulin Pump Dexcom G5® Mobile CGM Enabled User Guide Instructions for use t:slim X2 Insulin Pump Compatible with Dexcom G5® CGM.

[62] OmniPod®, Podder’s handbook, UST400 USER GUIDE. [cited 2018 Mar 16] Available from: https://www.myomnipod.com/sites/default/files/inline-files/17845- 5A%20Guide%2C%20Eros%20US%20User%20Guide%20Rev%20B.pdf.

[63] Health Quality Ontario. Continuous monitoring of glucose for type 1 diabetes: a health technology assessment. Ont Health Technol Assess Ser [Internet] 2018;18((2)):1-160. [cited 2018 Feb 28 ]. Available from: http://www.hqontario.ca/Evidence-to-Improve-Care/Health-Technology-Assessment/Journal- Ontario-Health-Technology-Assessment-Series.

[64] Langendam M, Luijf YM, Hooft L, Devries JH, Mudde AH, Scholten RJ. Continuous glucose monitoring systems for type 1 diabetes mellitus. The Cochrane database of systematic reviews. 2012;1:Cd008101.

[65] Dexcom. EUnetHTA Submission file. 2018.

[66] Dexcom. EUnetHTA Submission file. 2018.Update June 2018.

[67] Weinstein RL, Schwartz SL, Brazg RL, Bugler JR, Peyser TA, McGarraugh GV. Accuracy of the 5-day FreeStyle Navigator Continuous Glucose Monitoring System: comparison with frequent laboratory reference measurements. Diabetes Care. 2007;30(5):1125-30.

[68] Medtronic. EUnetHTA Submission file 2018.

[69] Senseonics. Eversense® Sensor, Transmitter and Smartphone App. [cited 2018 Feb 27]. Available from: https://ous.eversensediabetes.com/eversense-cgm-system/.

[70] Abbott. EUnetHTA Submission file. 2018.

[71] Bidonde J, Fagerlund B, Frønsdal K, Lund U, Robberstad B. FreeStyle Libre flash glucose self‐monitoring system: a single‐technology assessment. Norwegian Institute of Public Health (Folkehelseinstitutttet) Oslo. 2017.

[72] Medtronic. User’s guide, GuardianTM Connect, Medtronic Submission file.

[73] Abbott. FreeStyle Navigator II® Declaration of confirmity.

[74] Ekhlaspour L, Mondesir D, Lautsch N, Balliro C, Hillard M, Magyar K, et al. Comparative Accuracy of 17 Point-of-Care Glucose Meters. J Diabetes Sci Technol. 2017;11(3):558-66.

[75] Abbott. FreeStyle Libre for glucose monitoring NICE Medtech innovation briefing. 2017.

[76] Gale E, Anderson J. Diabetes mellitus and other disorders of metabolism. In Clinical Medicine. Ed. Kumar P and Clark M. 8th Edition Elsevier Ltd. ; 2012.

[77] Longo DL FA, Kasper DL, Hauser SL, Jameson JL, Loscalzo J. Harrison's Principles of Internal Medicine. 344. Diabetes Mellitus. In: McGraw-Hill, editor. 18 ed. New York2012.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 146 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[78] Skyler JS, Bakris GL, Bonifacio E, Darsow T, Eckel RH, Groop L, et al. Differentiation of Diabetes by Pathophysiology, Natural History, and Prognosis. Diabetes. 2017;66(2):241-55.

[79] American Diabetes Association. 2. Classification and Diagnosis of Diabetes:Standards of Medical Care in Diabetes—2018. Diabetes Care. 2018;41(Supplement 1):S13-S27.

[80] American Diabetes Association. Diagnosis and Classification of Diabetes Mellitus. Diabetes Care. 2013;37(Supplement_1):S81-S90.

[81] Hall J, Guyton A. Guyton and Hall textbook of medical physiology. 13 ed. Philadelphia, Pennsylvania: Elsevier; 2016.

[82] Brunton S. Pathophysiology of Type 2 Diabetes: The Evolution of Our Understanding. The Journal of family practice. 2016;65(4 Suppl).

[83] Agency for Quality and Accreditation in Health Care and Social Welfare; Finnish Medicines Agency; Assessment of Pharmacotherapies Regione Veneto EnfHTAE. Canagliflozin for the treatment of diabetes mellitus. 2014.

[84] American Diabetes Association. 13. Management of Diabetes in Pregnancy: Standards of Medical Care in Diabetes—2018<. Diabetes Care. 2018;41(Supplement 1):S137-S43.

[85] Carreiro MP, Nogueira AI, Ribeiro-Oliveira A. Controversies and Advances in Gestational Diabetes-An Update in the Era of Continuous Glucose Monitoring. Journal of clinical medicine. 2018;7(2).

[86] Hirsh IB EM. Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Epidemiology and pathogenesis. UpToDate. May 2018.

[87] Hamdy O, Khardori R. Diabetic Ketoacidosis. [cited 2018 Feb 09]. Available from: https://emedicine.medscape.com/article/118361-overviewUpdated.

[88] Adeyinka A, Kondamudi NP. Hyperosmolar Hyperglycemic Nonketotic Coma (HHNC, Hyperosmolar Hyperglycemic Nonketotic Syndrome). StatPearls. Treasure Island (FL): StatPearls Publishing StatPearls Publishing LLC.; 2018.

[89] American Diabetes Association. 6. Glycemic Targets: Standards of Medical Care in Diabetes – 2018. Diabetes Care. 2018;41(Supplement 1):S55-S64.

[90] Fraser CE DAD. Diabetic retinopathy: Classification and clinical features. UpToDate. May 2018.

[91] Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The Wisconsin epidemiologic study of diabetic retinopathy. II. Prevalence and risk of diabetic retinopathy when age at diagnosis is less than 30 years. Archives of ophthalmology (Chicago, Ill : 1960). 1984;102(4):520-6.

[92] Nathan DM, Zinman B, Cleary PA, Backlund JY, Genuth S, Miller R, et al. Modern-day clinical course of type 1 diabetes mellitus after 30 years' duration: the diabetes control and complications trial/epidemiology of diabetes interventions and complications and Pittsburgh epidemiology of diabetes complications experience (1983-2005). Archives of internal medicine. 2009;169(14):1307-16.

[93] Lecaire T, Palta M, Zhang H, Allen C, Klein R, D'Alessio D. Lower-than-expected prevalence and severity of retinopathy in an incident cohort followed during the first 4-14 years of type 1 diabetes: the Wisconsin Diabetes Registry Study. American journal of epidemiology. 2006;164(2):143-50.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 147 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[94] LeCaire TJ, Palta M, Klein R, Klein BE, Cruickshanks KJ. Assessing progress in retinopathy outcomes in type 1 diabetes: comparing findings from the Wisconsin Diabetes Registry Study and the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Diabetes Care. 2013;36(3):631-7.

[95] Vallance JH, Wilson PJ, Leese GP, McAlpine R, MacEwen CJ, Ellis JD. Diabetic retinopathy: more patients, less laser: a longitudinal population-based study in Tayside, Scotland. Diabetes Care. 2008;31(6):1126-31.

[96] Friedman E. Diabetic Renal Disease. In Rifkin H, Porte D, Eds. Diabetes Mellitus/Theory and Practice 1990.

[97] Bakris GL. Overview of diabetic nephropathy. UpToDate. 2018.

[98] Feldman E. Epidemiology and classification of diabetic neuropathy. UpToDate. May 2018.

[99] Vinik AI MB, Leichter SB, Wagner AL, O'Brian IT, Georges LP. . Epidemiology of the complications of diabetes. In: Press CU, editor. Diabetes: Clinical Science in Practice Cambridge, MA1995.

[100] Edwards JL, Vincent AM, Cheng HT, Feldman EL. Diabetic neuropathy: mechanisms to management. Pharmacology & therapeutics. 2008;120(1):1-34.

[101] Zimmerman R. Diabetes Mellitus: Management of Microvascular and Macrovascular Complications.[cited 2018 Feb 28]. Available from: http://www.clevelandclinicmeded.com/medicalpubs/diseasemanagement/ endocrinology/diabetes-mellitus/2016.

[102] Vinik AI PG, McNitt P, Stansberry KB. Diabetic neuropathies: an overview of clinical aspects, pathogenesis, and treatment. . Diabetes Mellitus: A Fundamental and Clinical Text 3ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2003. p. 911–34.

[103] Kissela BM, Khoury J, Kleindorfer D, Woo D, Schneider A, Alwell K, et al. Epidemiology of ischemic stroke in patients with diabetes: the greater Cincinnati/Northern Kentucky Stroke Study. Diabetes Care. 2005;28(2):355-9.

[104] Melton LJ, 3rd, Macken KM, Palumbo PJ, Elveback LR. Incidence and prevalence of clinical peripheral vascular disease in a population-based cohort of diabetic patients. Diabetes Care. 1980;3(6):650-4.

[105] Tamayo T, Rosenbauer J, Wild SH, Spijkerman AM, Baan C, Forouhi NG, et al. Diabetes in Europe: an update. Diabetes research and clinical practice. 2014;103(2):206-17.

[106] Rieder A, Rathmanner T, Kiefer I, Dorner T, Kunze M. Österreichischer Diabetesbericht. 2004.

[107] American Diabetes Association. 12. Children and Adolescents: Standards of Medical Care in Diabetes – 2018. Diabetes Care. 2018;41(Supplement 1):S126-S36.

[108] National Institute for Health and Care Excellence. Type 2 diabetes in adults: diagnosis and management.NICE guideline. 2015.

[109] Norwegian Institute of Public Health. Freestyle Libre Flash Glucose Self-Monitoring System: A single technology assessment. 2017.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 148 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[110] World Health Organization. WHO diabetes country profiles. [cited 2018 Jan 19]. Available from: http://www.who.int/diabetes/country-profiles/en/.

[111] Wong JC, Foster NC, Maahs DM, Raghinaru D, Bergenstal RM, Ahmann AJ, et al. Real-time continuous glucose monitoring among participants in the T1D Exchange clinic registry. Diabetes Care. 2014;37(10):2702-9.

[112] James S, Perry L, Gallagher R, Lowe J. Diabetes Educators’ Intended and Reported Use of Common Diabetes-Related Technologies:Discrepancies and Dissonance. Journal of Diabetes Science and Technology. 2016;10(6):1277-86.

[113] Fonseca VA, Grunberger G, Anhalt H, Bailey TS, Blevins T, Garg SK, et al. CONTINUOUS GLUCOSE MONITORING: A CONSENSUS CONFERENCE OF THE AMERICAN ASSOCIATION OF CLINICAL ENDOCRINOLOGISTS AND AMERICAN COLLEGE OF ENDOCRINOLOGY. Endocrine practice : official journal of the American College of Endo- crinology and the American Association of Clinical Endocrinologists. 2016;22(8):1008-21.

[114] Los E, Ulrich J, Guttmann-Bauman I. Technology Use in Transition-Age Patients With Type 1 Diabetes: Reality and Promises. J Diabetes Sci Technol. 2016;10(3):662-8.

[115] Gonder-Frederick L, Shepard J, Peterson N. Closed-loop glucose control: psychological and behavioral considerations. J Diabetes Sci Technol. 2011;5(6):1387-95.

[116] Hirsch IB. Clinical review: Realistic expectations and practical use of continuous glucose monitoring for the endocrinologist. The Journal of clinical endocrinology and metabolism. 2009;94(7):2232-8.

[117] Martin-Vaquero P, Martinez-Brocca MA, Garcia-Lopez JM. Position statement on efficiency of technologies for diabetes management. Endocrinologia y nutricion : organo de la Sociedad Espanola de Endocrinologia y Nutricion. 2014;61(10):e45-63.

[118] Dunn TC, Xu Y, Hayter G, Ajjan RA. Real-world flash glucose monitoring patterns and associations between self-monitoring frequency and glycaemic measures: A European analysis of over 60 million glucose tests. Diabetes research and clinical practice. 2018;137:37-46.

[119] European network for Health Technology Assessment (EUnetHTA). Guideline “Therapeutic medical devices”. 2015.

[120] Schnell-Inderst P, Hunger T, Conrads-Frank A, Arvandi M, Siebert U. Ten recommendations for assessing the comparative effectiveness of therapeutic medical devices: a targeted review and adaptation. Journal of clinical epidemiology. 2018;94:97-113.

[121] Schnell-Inderst P, Hunger T, Conrads-Frank A, Arvandi M, Siebert U. Recommendations for primary studies evaluating therapeutic medical devices were identified and systematically reported through reviewing existing guidance. Journal of clinical epidemiology. 2018;94:46-58.

[122] Motte AF, Diallo S, van den Brink H, Chateauvieux C, Serrano C, Naud C, et al. Existing reporting guidelines for clinical trials are not completely relevant for implantable medical devices: a systematic review. Journal of clinical epidemiology. 2017;91:111-20.

[123] Geddes J, Schopman JE, Zammitt NN, Frier BM. Prevalence of impaired awareness of hypoglycaemia in adults with Type 1 diabetes. Diabetic medicine : a journal of the British Diabetic Association. 2008;25(4):501-4.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 149 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[124] Gubitosi-Klug RA, Braffett BH, White NH, Sherwin RS, Service FJ, Lachin JM, et al. Risk of Severe Hypoglycemia in Type 1 Diabetes Over 30 Years of Follow-up in the DCCT/EDIC Study. Diabetes Care. 2017;40(8):1010-6.

[125] Karges B, Schwandt A, Heidtmann B, et al. Association of insulin pump therapy vs insulin injection therapy with severe hypoglycemia, ketoacidosis, and glycemic control among children, adolescents, and young adults with type 1 diabetes. JAMA. 2017;318(14):1358-66.

[126] Olafsdottir AF, Polonsky W, Bolinder J, Hirsch IB, Dahlqvist S, Wedel H, et al. A Randomized Clinical Trial of the Effect of Continuous Glucose Monitoring on Nocturnal Hypoglycemia, Daytime Hypoglycemia, Glycemic Variability, and Hypoglycemia Confidence in Persons with Type 1 Diabetes Treated with Multiple Daily Insulin Injections (GOLD-3). Diabetes technology & therapeutics. 2018;20(4):274-84.

[127] Pickup JC, Freeman SC, Sutton AJ. Glycaemic control in type 1 diabetes during real time continuous glucose monitoring compared with self monitoring of blood glucose: meta-analysis of randomised controlled trials using individual patient data. BMJ (Clinical research ed). 2011;343:d3805.

[128] Yeh HC, Brown TT, Maruthur N, Ranasinghe P, Berger Z, Suh YD, et al. Comparative effectiveness and safety of methods of insulin delivery and glucose monitoring for diabetes mellitus: a systematic review and meta-analysis. Annals of internal medicine. 2012;157(5):336-47.

[129] Riemsma R, Corro Ramos I, Birnie R, Buyukkaramikli N, Armstrong N, Ryder S, et al. Integrated sensor-augmented pump therapy systems [the MiniMed(R) Paradigm Veo system and the Vibe and G4(R) PLATINUM CGM (continuous glucose monitoring) system] for managing blood glucose levels in type 1 diabetes: a systematic review and economic evaluation. Health technology assessment (Winchester, England). 2016;20(17):v-xxxi, 1-251.

[130] Feig DS, Donovan LE, Corcoy R, Murphy KE, Amiel SA, Hunt KF, et al. Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): a multicentre international randomised controlled trial. Lancet (London, England). 2017;390(10110):2347-59.

[131] Herman A, Aerts O, Baeck M, Bruze M, De Block C, Goossens A, et al. Allergic contact dermatitis caused by isobornyl acrylate in Freestyle(R) Libre, a newly introduced glucose sensor. Contact dermatitis. 2017;77(6):367-73.

[132] Aerts O, Herman A, Bruze M, Goossens A, Mowitz M. FreeStyle Libre: contact irritation versus contact allergy. Lancet (London, England). 2017;390(10103):1644.

[133] Jadviscokova T, Fajkusova Z, Pallayova M, Luza J, Kuzmina G. Occurence of adverse events due to continuous glucose monitoring. Biomedical papers of the Medical Faculty of the University Palacky, Olomouc, Czechoslovakia. 2007;151(2):263-6.

[134] Schwensen JF, Friis UF, Zachariae C, Johansen JD. Sensitization to cyanoacrylates caused by prolonged exposure to a glucose sensor set in a diabetic child. Contact dermatitis. 2016;74(2):124-5.

[135] Huic M, Marusic M, Marusic A. Completeness and changes in registered data and reporting bias of randomized controlled trials in ICMJE journals after trial registration policy. PloS one. 2011;6(9):e25258.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 150 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

[136] Ioannidis JP, Evans SJ, Gotzsche PC, O'Neill RT, Altman DG, Schulz K, et al. Better reporting of harms in randomized trials: an extension of the CONSORT statement. Annals of internal medicine. 2004;141(10):781-8.

[137] Zorzela L, Loke YK, Ioannidis JP, Golder S, Santaguida P, Altman DG, et al. PRISMA harms checklist: improving harms reporting in systematic reviews. BMJ (Clinical research ed). 2016;352:i157.

[138] Lineberry N, Berlin JA, Mansi B, Glasser S, Berkwits M, Klem C, et al. Recommendations to improve adverse event reporting in clinical trial publications: a joint pharmaceutical industry/journal editor perspective. BMJ (Clinical research ed). 2016;355:i5078.

[139] Lawton J, Blackburn M, Allen J, Campbell F, Elleri D, Leelarathna L, et al. Patients' and caregivers' experiences of using continuous glucose monitoring to support diabetes self- management: qualitative study. BMC endocrine disorders. 2018;18(1):12.

[140] Welsh JB. Role of Continuous Glucose Monitoring in Insulin-Requiring Patients with Diabetes. Diabetes Technol Ther. 2018;20; S2-42-49.

[141] Garg SK, Akturk HK. A New Era in Continuous Glucose Monitoring: Food and Drug Administration Creates a New Category of Factory-Calibrated Nonadjunctive, Interoperable Class II Medical Devices. Diabetes Technol Ther. 2018; 20;391-4.

[142] Parker AS, Welsh JB, Dunn LJ, et al.: Insights from big data. (1): viewing of real-time continuous glucose monitoring data and its impact on time in range. Diabetes Technol Ther 2018; 20:A-121.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 151 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

APPENDIX 1: METHODS AND DESCRIPTION OF THE EVIDENCE USED

DOCUMENTATION OF THE SEARCH STRATEGIES

The systematic literature search was performed between 9 -15 March 2018.

MEDLINE

Database: Ovid MEDLINE(R) Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE(R) Daily and Ovid MEDLINE(R) <1946 to Present>

Search Strategy:

------

1 diabetes mellitus/ or exp diabetes mellitus, type 1/ or exp diabetes mellitus, type 2/ or diabetes, gestational/ (276478)

2 diabetes mellitus type 1.tw. (915)

3 diabetes mellitus type 2.tw. (1926)

4 type 1 diabetes.tw. (32930)

5 type 2 diabetes.tw. (100017)

6 gestational diabetes.tw. (11377)

7 or/1-6 (321866)

8 exp Insulin/ (174470)

9 Insulin Infusion Systems/ (4613)

10 exp Injections, Subcutaneous/ (38189)

11 8 and 10 (2408)

12 insulin.tw. (321899)

13 8 or 9 or 11 or 12 (357485)

14 continuous glucose monitoring.tw. (2702)

15 flash glucose monitoring.tw. (55)

16 FreeStyle Libre.tw. (32)

17 FreeStyle Navigator II.tw. (1)

18 G4 PLATINUM.tw. (42)

19 G5 Mobile.tw. (7)

20 Guardian Connect.tw. (1)

21 Eversense.tw. (2)

22 Monitoring system.tw. (8943)

23 SugarBEAT.tw. (0)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 152 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

24 or/14-23 (11029)

25 Blood Glucose Self-Monitoring/ (5418)

26 blood glucose self-monitoring.tw. (248)

27 blood sugar self-monitoring.tw. (7)

28 blood glucose monitoring.tw. (1637)

29 blood sugar monitoring.tw. (50)

30 MiniMed Paradigm Veo.tw. (1)

31 MiniLink transmitter.tw. (1)

32 Enlite Glucose Sensor.tw. (4)

33 MiniMed 640G.tw. (11)

34 Guardian 2 Link transmitter.tw. (0)

35 t:slim X2 Insulin Pump.tw. (0)

36 Dexcom G5 CGM.tw. (3)

37 Omnipod.tw. (26)

38 or/25-37 (6679)

39 7 and 13 and 24 and 38 (526)

***************************

CENTRAL

Database: EBM Reviews - Cochrane Central Register of Controlled Trials

Search Strategy:

------

1 diabetes mellitus/ or exp diabetes mellitus, type 1/ or exp diabetes mellitus, type 2/ or diabetes, gestational/ (16404)

2 insulin dependent diabetes.kw. (7704)

3 diabetes mellitus type 1.tw. (57)

4 diabetes mellitus type 2.tw. (223)

5 type 1 diabetes.tw. (3263)

6 type 2 diabetes.tw. (16222)

7 gestational diabetes.tw. (1059)

8 or/1-7 (27764)

9 exp Insulin/ (9941)

10 Insulin Infusion Systems/ (527)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 153 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

11 exp Injections, Subcutaneous/ (4011)

12 9 and 11 (434)

13 insulin.tw. (27389)

14 9 or 10 or 12 or 13 (28591)

15 continuous glucose monitoring.tw. (779)

16 flash glucose monitoring.tw. (11)

17 FreeStyle Libre.tw. (6)

18 FreeStyle Navigator II.tw. (1)

19 G4 PLATINUM.tw. (15)

20 G5 Mobile.tw. (3)

21 Guardian Connect.tw. (0)

22 Eversense.tw. (3)

23 Monitoring system.tw. (749)

24 SugarBEAT.tw. (0)

25 or/15-24 (1367)

26 Blood Glucose Self-Monitoring/ (573)

27 blood glucose monitoring.kw. (1146)

28 blood glucose self-monitoring.tw. (36)

29 blood sugar self-monitoring.tw. (2)

30 blood glucose monitoring.tw. (217)

31 blood sugar monitoring.tw. (3)

32 MiniMed Paradigm Veo.tw. (0)

33 MiniLink transmitter.tw. (4)

34 Enlite Glucose Sensor.tw. (8)

35 MiniMed 640G.tw. (11)

36 Guardian 2 Link transmitter.tw. (0)

37 t:slim X2 Insulin Pump.tw. (0)

38 Dexcom G5 CGM.tw. (0)

39 Omnipod.tw. (6)

40 or/26-39 (1742)

41 8 and 14 and 25 and 40 (261)

***************************

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 154 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

EMBASE

Database: Embase Classic+Embase <1947 to 2018 Week 11>

Search Strategy:

------

1 *insulin dependent diabetes mellitus/ (63676)

2 *pregnancy diabetes mellitus/ (14658)

3 diabetes mellitus type 1.tw. (1555)

4 diabetes mellitus type 2.tw. (3762)

5 type 1 diabetes.tw. (50601)

6 type 2 diabetes.tw. (151304)

7 gestational diabetes.tw. (18075)

8 or/1-7 (248002)

9 *insulin/ (120088)

10 *insulin infusion/ (2310)

11 *subcutaneous drug administration/ (1375)

12 9 and 11 (149)

13 insulin.tw. (442863)

14 9 or 10 or 12 or 13 (460725)

15 continuous glucose monitoring.tw. (4883)

16 flash glucose monitoring.tw. (104)

17 FreeStyle Libre.tw. (79)

18 FreeStyle Navigator II.tw. (2)

19 G4 PLATINUM.tw. (90)

20 G5 Mobile.tw. (15)

21 Guardian Connect.tw. (1)

22 Eversense.tw. (9)

23 Monitoring system.tw. (12828)

24 SugarBEAT.tw. (1)

25 or/15-24 (16679)

26 *blood glucose monitoring/ (5809)

27 blood glucose self-monitoring.tw. (328)

28 blood sugar self-monitoring.tw. (11)

29 blood glucose monitoring.tw. (2577)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 155 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

30 blood sugar monitoring.tw. (114)

31 MiniMed Paradigm Veo.tw. (5)

32 MiniLink transmitter.tw. (7)

33 Enlite Glucose Sensor.tw. (15)

34 MiniMed 640G.tw. (41)

35 Guardian 2 Link transmitter.tw. (0)

36 t:slim X2 Insulin Pump.tw. (0)

37 Dexcom G5 CGM.tw. (2)

38 Omnipod.tw. (89)

39 or/26-38 (8027)

40 8 and 14 and 25 and 39 (525)

***************************

CINAHL with Full Text (EBSCOhost)

S41 (S9 AND S15 AND S26 AND S40) (207)

S40 (S27 OR S28 OR S29 OR S30 OR S31 OR S32 OR S33 OR S34 OR S35 OR S36 OR S37 OR S38 OR S39) (3,170)

S39 TI (Omnipod) OR AB (Omnipod) (12)

S38 TI (Dexcom G5 CGM) OR AB (Dexcom G5 CGM) (3)

S37 TI (t:slim X2 Insulin Pump) OR AB (t:slim X2 Insulin Pump) (1)

S36 TI (Guardian 2 Link transmitter) OR AB (Guardian 2 Link transmitter) (0)

S35 TI (MiniMed 640G) OR AB (MiniMed 640G) (9)

S34 TI (Enlite Glucose Sensor) OR AB (Enlite Glucose Sensor) (5)

S33 TI (MiniLink transmitter) OR AB (MiniLink transmitter) (0)

S32 TI (MiniMed Paradigm Veo) OR AB (MiniMed Paradigm Veo) (3)

S31 TI (blood glucose monitoring) OR AB (blood glucose monitoring) (1,415)

S30 TI (blood sugar monitoring) OR AB (blood sugar monitoring) (47)

S29 TI (blood sugar self-monitoring) OR AB (blood sugar self-monitoring) (2)

S28 TI (blood glucose self-monitoring) OR AB (blood glucose self-monitoring) (117)

S27 MH "Blood Glucose Self-Monitoring" (2,390)

S26 S16 OR S17 OR S18 OR S19 OR S20 OR S21 OR S22 OR S23 OR S24 OR S25 (3,324)

S25 TI (SugarBEAT) OR AB (SugarBEAT) (0)

S24 TI (Monitoring system) OR AB (Monitoring system) (2,633)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 156 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

S23 TI (Eversense) OR AB (Eversense) (1)

S22 TI (Guardian Connect) OR AB (Guardian Connect) (1)

S21 TI (G5 Mobile) OR AB (G5 Mobile) (4)

S20 TI (G4 PLATINUM) OR AB (G4 PLATINUM) (18)

S19 TI (FreeStyle Navigator II) OR AB (FreeStyle Navigator II) (1)

S18 TI (FreeStyle Libre) OR AB (FreeStyle Libre) (11)

S17 TI (flash glucose monitoring) OR AB (flash glucose monitoring) (23)

S16 TI (continuous glucose monitoring) OR AB (continuous glucose monitoring) (907)

S15 S10 OR S11 OR S13 OR S14 (32,079)

S14 TI (insulin) OR AB (insulin) (26,734)

S13 (S10 AND S12) (571)

S12 MH "Injections, Subcutaneous+" (2,379)

S11 MH "Insulin Infusion Systems" (1,448)

S10 MH "insulin+" (16,456)

S9 (S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 OR S8) (82,192)

S8 TI (gestational diabetes) OR AB (gestational diabetes) (3,380)

S7 TI (type 2 diabetes) OR AB (type 2 diabetes) (23,681)

S6 TI (type 1 diabetes) OR AB (type 1 diabetes) (9,324)

S5 TI (diabetes mellitus type 2) OR AB (diabetes mellitus type 2) (6,068)

S4 TI (diabetes mellitus type 1) OR AB (diabetes mellitus type 1) (1,717)

S3 MH "diabetes mellitus, type 2+" (31,182)

S2 MH "diabetes mellitus, type 1+" (12,649)

S1 MH "diabetes mellitus" (35,389)

HTA

Database: EBM Reviews - Health Technology Assessment <4th Quarter 2016>

Search Strategy:

------

1 diabetes mellitus/ or exp diabetes mellitus, type 1/ or exp diabetes mellitus, type 2/ or diabetes, gestational/ (314)

2 diabetes mellitus type 1.tw. (70)

3 diabetes mellitus type 2.tw. (121)

4 type 1 diabetes.tw. (53)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 157 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

5 type 2 diabetes.tw. (135)

6 gestational diabetes.tw. (16)

7 or/1-6 (378)

8 exp Insulin/ (66)

9 Insulin Infusion Systems/ (33)

10 exp Injections, Subcutaneous/ (18)

11 8 and 10 (2)

12 insulin.tw. (167)

13 8 or 9 or 11 or 12 (167)

14 continuous glucose monitoring.tw. (18)

15 flash glucose monitoring.tw. (0)

16 FreeStyle Libre.tw. (0)

17 FreeStyle Navigator II.tw. (0)

18 G4 PLATINUM.tw. (0)

19 G5 Mobile.tw. (0)

20 Guardian Connect.tw. (0)

21 Eversense.tw. (0)

22 Monitoring system.tw. (24)

23 SugarBEAT.tw. (0)

24 or/14-23 (37)

25 Blood Glucose Self-Monitoring/ (42)

26 blood glucose self-monitoring.tw. (42)

27 blood sugar self-monitoring.tw. (0)

28 blood glucose monitoring.tw. (10)

29 blood sugar monitoring.tw. (0)

30 MiniMed Paradigm Veo.tw. (0)

31 MiniLink transmitter.tw. (0)

32 Enlite Glucose Sensor.tw. (0)

33 MiniMed 640G.tw. (0)

34 Guardian 2 Link transmitter.tw. (0)

35 t:slim X2 Insulin Pump.tw. (0)

36 Dexcom G5 CGM.tw. (0)

37 Omnipod.tw. (0)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 158 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

38 or/25-37 (47)

39 7 and 13 and 24 and 38 (8)

***************************

DARE

Database: EBM Reviews - Database of Abstracts of Reviews of Effects <1st Quarter 2016>

Search Strategy:

------

1 diabetes mellitus.kw. (1004)

2 diabetes mellitus, type 1.kw. (132)

3 diabetes mellitus, type 2.kw. (651)

4 diabetes, gestational.kw. (49)

5 diabetes mellitus type 1.tw. (133)

6 diabetes mellitus type 2.tw. (652)

7 type 1 diabetes.tw. (121)

8 type 2 diabetes.tw. (602)

9 gestational diabetes.tw. (81)

10 or/1-9 (1155)

11 Insulin.kw. (266)

12 Insulin Infusion Systems.kw. (23)

13 Injections, Subcutaneous.kw. (47)

14 11 and 13 (10)

15 insulin.tw. (502)

16 11 or 12 or 14 or 15 (502)

17 continuous glucose monitoring.tw. (15)

18 flash glucose monitoring.tw. (0)

19 FreeStyle Libre.tw. (0)

20 FreeStyle Navigator II.tw. (0)

21 G4 PLATINUM.tw. (0)

22 G5 Mobile.tw. (0)

23 Guardian Connect.tw. (0)

24 Eversense.tw. (0)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 159 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

25 Monitoring system.tw. (13)

26 SugarBEAT.tw. (0)

27 or/17-26 (23)

28 Blood Glucose Self-Monitoring.kw. (45)

29 blood glucose self-monitoring.tw. (49)

30 blood sugar self-monitoring.tw. (0)

31 blood glucose monitoring.tw. (14)

32 blood sugar monitoring.tw. (0)

33 MiniMed Paradigm Veo.tw. (0)

34 MiniLink transmitter.tw. (0)

35 Enlite Glucose Sensor.tw. (0)

36 MiniMed 640G.tw. (0)

37 Guardian 2 Link transmitter.tw. (0)

38 t:slim X2 Insulin Pump.tw. (0)

39 Dexcom G5 CGM.tw. (0)

40 Omnipod.tw. (1)

41 or/28-40 (60)

42 10 and 16 and 27 and 41 (8)

***************************

NHS EED

Database: EBM Reviews - NHS Economic Evaluation Database <1st Quarter 2016>

Search Strategy:

------

1 diabetes mellitus/ or exp diabetes mellitus, type 1/ or exp diabetes mellitus, type 2/ or diabetes, gestational/ (583)

2 diabetes mellitus type 1.tw. (93)

3 diabetes mellitus type 2.tw. (392)

4 type 1 diabetes.tw. (57)

5 type 2 diabetes.tw. (316)

6 gestational diabetes.tw. (26)

7 or/1-6 (615)

8 exp Insulin/ (139)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 160 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

9 Insulin Infusion Systems/ (13)

10 exp Injections, Subcutaneous/ (52)

11 8 and 10 (9)

12 insulin.tw. (257)

13 8 or 9 or 11 or 12 (257)

14 continuous glucose monitoring.tw. (2)

15 flash glucose monitoring.tw. (0)

16 FreeStyle Libre.tw. (0)

17 FreeStyle Navigator II.tw. (0)

18 G4 PLATINUM.tw. (0)

19 G5 Mobile.tw. (0)

20 Guardian Connect.tw. (0)

21 Eversense.tw. (0)

22 Monitoring system.tw. (14)

23 SugarBEAT.tw. (0)

24 or/14-23 (16)

25 Blood Glucose Self-Monitoring/ (23)

26 blood glucose self-monitoring.tw. (25)

27 blood sugar self-monitoring.tw. (0)

28 blood glucose monitoring.tw. (7)

29 blood sugar monitoring.tw. (0)

30 MiniMed Paradigm Veo.tw. (0)

31 MiniLink transmitter.tw. (0)

32 Enlite Glucose Sensor.tw. (0)

33 MiniMed 640G.tw. (0)

34 Guardian 2 Link transmitter.tw. (0)

35 t:slim X2 Insulin Pump.tw. (0)

36 Dexcom G5 CGM.tw. (0)

37 Omnipod.tw. (0)

38 or/25-37 (30)

39 7 and 13 and 24 and 38 (1)

***************************

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 161 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

DESCRIPTION OF THE EVIDENCE USED

Guidelines for diagnosis and management

Table A1: Overview of guidelines

Name of society/ Date of Country/ies to Summary of recommendations Organisation issu- issue or last which guide- ing guidelines update line applies (Level of evidence/grade of recommendation for the indication under assessment)

Endocrine Society 2016 Global 6. Real-Time Continuous Glucose Monitors in Adult Outpatients doi: 10.1210/jc.2016- 6.1 We recommend real-time continuous glucose 2534 monitoring (RT-CGM) devices for adult patients

with T1DM who have A1c levels above target and who are willing and able to use these devices on Dexcom Submis- sion file 2018 a nearly daily basis. (1 QQQQ)*

6.2 We recommend RT-CGM devices for adult patients with well-controlled T1DM who are willing

and able to use these devices on a nearly daily basis. (1 QQQQ)*

Use of Continuous Glucose Monitoring in Adults with Type 2 Diabetes

6.3 We suggest short-term, intermittent RT-CGM use in adult patients with T2DM (not on prandial

insulin) who have A1c levels _≥7% and are willing and able to use the device. (2 QQOO)*

NICE July 2016 UK https://www.nice.org.uk/guidance/ng17

Abbott T1 diabetes in adults:diagnosis and management Submission Nov 16 Test at least 4 times a day, including before each file 2018 meal and before bed.

Test up to 10 times a day in certain circumstances,such as in illness, pregnancy or during Dexcom Aug 15 sport. Submission file

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 162 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

2018

https://www.nice.org.uk/guidance/ng18

May 17 Diabetes (T1 and T2) in children and young people:diagnosis and management

Perform at least 5 capillary blood glucose tests per day. More frequent testing is often needed (for example with physical activity and during intercur-

rent illness). 2015

https://www.nice.org.uk/guidance/ng3

Diabetes in pregnancy:management from pre- conception to the post natal period

Pregnant women with type 1 diabetes should test blood glucose levels daily.

Pregnant women with type 2 diabetes or gestational diabetes who are on a multiple daily

insulin injection regimen should test blood glucose levels daily. Pregnant women with type

2 diabetes or gestational diabetes should test blood glucose levels daily if they are on diet

and exercise therapy, taking oral therapy (with or without diet and exercise therapy) or

single dose intermediate acting or long acting insulin.

https://www.nice.org.uk/guidance/ng28

Type 2 diabetes in adults:management

Do not routinely offer self-monitoring of blood glucose levels for adults with type 2

diabetes unless the person is on insulin, there is evidence of hypoglycaemic episodes, the

person is on medication that may increase their

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 163 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

risk of hypoglycaemia while driving or

operating machinery or the person is pregnant, or is planning to become pregnant.

Consider short-term self-monitoring of blood glucose levels in adults with type 2 diabetes

when starting treatment with oral or intravenous corticosteroids or to confirm suspected hypoglycaemia.

Continuous glucose monitoring

Do not offer real-time continuous glucose monitoring routinely to adults with type 1 diabetes.

Consider real-time continuous glucose monitoring for adults with type 1 diabetes who are

willing to commit to using it at least 70% of the time and to calibrate it as needed, and who

have any of the following despite optimised use of insulin therapy and conventional blood

glucose monitoring:

• More than 1 episode a year of severe hypoglycaemia with no obviously preventable precipitating cause.

• Complete loss of awareness of hypoglycaemia.

• Frequent (more than 2 episodes a week) asymptomatic hypoglycaemia that is causing problems with daily activities.

• Extreme fear of hypoglycaemia.

• Hyperglycaemia (HbA1c level of 75 mmol/mol [9%] or higher) that per- sists despite testing at least 10 times a day (see recommendations 1.6.11 and 1.6.12).

• Continue real-time continuous glucose monitoring only if HbA1c can be sustained at or below 53

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 164 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

mmol/mol (7%) and/or there has been a fall in HbA1c of 27 mmol/mol (2.5%) or more.

• Patients using rtCGM should be treated using the principles of flexible insulin therapy with either MDI or CSII.

• NICE Clinical Guideline #NG18 recom- mends that rtCGM with alarms should be offered to children and young people with T1D who have any of:

frequent SH;

IAH associated with adverse consequences (for example, seizures or anxiety);

inability to recognise, or communicate about, symptoms of hypoglycaemia (for example, because of age, cognitive or neurological disabilities).

The Norwegian June 2017 Norway Fehler! Hyperlink-Referenz ungültig. directorate of https://helsedirektoratet.no/Retningslinjer/Diabetes Health .pdf

Abbott Submission

file 2018 National Guidelines’ (Last update: June 2017) key highlights:

o There are no pure FGM guidelines published by the authorizes

o SMBG MDI: before and after meals, driving, training, illness, travelling, pregnancy

o T1: 4-8 measurements a day is the referral made in the national guidelines

o CGM: Unexplainable Hypos/Hyper, Hypo unawareness (especially for patients in jobs

with potential critical consequences of such), patients living alone, pregnancy, patients who experience sever and rapid changes in glucose levels during training and patients who measures a lot or “to much” (12-14 strips/ day). The guidelines also discuss the CGM effect on HbA1c and Hypo reduction.

The Danish society Oct. 2016 Denmark Type 1:

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 165 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

for endocrinology http://www.endocrinology.dk/index.php/nbvhoved menu/1-diabetes-mellitus/3-type-1-diabetes- 2014 mellitus

Abbott Submission Type 2 Insulin users: file 2018 http://www.endocrinology.dk/PDF/Insulinbehandlin Dec. 2017 gDM2rev2014.pdf

FGM and CGM: http://endocrinology.dk/index.php/nbvhovedmenu/ 1-diabetes-mellitus/nbv-cgm-og-fgm-til-born-unge- og-voksne

Associazione Med- 2016 Italy SMBG ici Diabetologi Class 1. Patient in intensive insulin treatment (AMD) - Societa (basal-bolus or pump) Italiana di Diabeto- Patients in basal-bolus insulin therapy: recom- logia (SID) – mended 150 tests/month Patients with pump, pre-gestational diabetes in Standard italiani pregnancy and patients < 18: recommended 250 per la cura del tests/month diabete mellito Children < 6 years: recommended 300 tests/month

Patients that start insulin treatment: recommended Abbott Submission 200 tests/month for the first quarter Patients with T2DM in basal-bolus insulin treat- file 2018 ment with stable glycemic control: recommended 125 tests/month Recommended an unlimited number of tests in conditions of glycemic unbalance or in presence of comorbidities, for a period limited to the duration of the event

CGM Utility of CGM has been shown in real time mode and in particular in association with SAP protocols, in selected patients and adequately educated, whilst the evidence for patients with DMT1 in MDI is weaker. No specific recommendation is made on CGM.

FGM In the clinical practice, feedback on FGM experience are positive Scientific evidence is still limited.

Between 2016 and 2018, 16 Italian Regions is-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 166 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

sued regional guidelines on the usage of FGM and, only part of them, regulated the access to CGM. FGM is mainly addressed to patients in MDI therapy, whilst CGM is generally recommended in association to SAP.

DDG= German 2011 Germany 1) People with type 1 diabetes should perform at Diabetes Associa- least 4 times a day (before eating and tion Novel guideline before going to bed) a blood glucose self- expected measurement. Recommendation grade a = strong 2018 recommendation

2) More frequent daily blood glucose self- measurements are among others in

Abbott Submission following situations: file 2018 • before, possibly during and after intensive physical activity / sports

Avoidance of hypoglycaemia,

2011 Novel • after hypoglycaemia, guideline expected • in case of illness (including the need for gluco- 2018 corticoid administration)

• in case of planned pregnancy and during preg-

nancy,

• active participation in road traffic and prolonged

participation

also in between,

2017 • to travel

DDG Guideline for Gestational Diabe- https://www.deutsche-diabetes- tes gesell-

schaft.de/fileadmin/Redakteur/Leitlinien/Evidenzba 2017 sierte_Leitlinien/AktualisierungTherapieTyp1Diabe tes_1_20120319_TL.pdf

Evidence-based guideline on diagnostics, therapy a. aftercare of the German Diabetes DDG Guideline: for Society (DDG) and the German Society for

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 167 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

children and ado- Gynecology and Obstetrics (DGGG) lescents 1) At the first presentation, the pregnant woman DDG Practical should start with 4 self-measured values (4-point Guideline profile) for the first 1-2 weeks, in the morning and 1 or 2 hours after the start of the main meals.

(Grade A).

2) If at least 2 measurements have been taken during at least 2 days within the first two weeks, a

6-point profile can be taken on the following day after the second conspicuous day (3 pre-prandial measurements and 3 measurements 1 or 2 hours 2016 after the main meals (grade C).

3) In case of insulin therapy, daily measurements with a 4-point profile or a 6-point profile should be

carried out as instructed by the caregiver. Addi-

tional measurements should be determined individually (grade B).

http://www.deutsche-diabetes- gesell- schaft.de/fileadmin/Redakteur/Leitlinien/Evidenzba G-BA guideline sierte_Leitlinien/Gestationsdiabetes_EbLL_Endfas sung_2011_08_11_.pdf

Diagnosis, Therapy and Follow-Up of Diabetes

Mellitus in Children and Adolescents

The average frequency of glucose control should be between 5 and 6 times a day, but may be significantly higher in individual cases. https://www.deutsche-diabetes- gesell- schaft.de/fileadmin/Redakteur/Leitlinien/Praxisleitli nien/2017/dus_2017_S2_Praxisempfehlungen_39 73540_Neu_DM_im_Kindes- _und_Jugendalter__11__Online-PDF.PDF

Practical recommendation of the DDG: Glu- cose control in patients with type 1 or type 2 diabetes

1) A clear indication for SMBG exists in patients

with type 1 diabetes or insulin mandatory type 2

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 168 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

diabetes. Patients with type 1 diabetes and multiple insulin injections daily or an insulin pump should perform at least

4x daily SMBG (pre-prandial and before bedtime) and all 2-3 weeks also a blood control test during the night.

2) Additional measurements are recommended in special situations, e.g. to check meal effects, suspected hypoglycaemia, sports, illness, vaca- tion, driving a car, etc. This results in an average need for glucose test strips of at least 5 per day.

3) A special group of patients are patients with type 1 diabetes and hypoglycaemia unawareness. Additional controls are recommended before each car ride, during exercise, sports and everyday work. This creates a quarterly requirement of at least 800 test strips.

4) Patients with insulin-dependent type 2 diabetes should perform 4-5 blood glucose self- measurements (also pre-prandial and occasionally postprandial as well before bedtime).

5) Pregnant women with a pre-existing type 1 or type 2 diabetes should perform also postprandial glucose measurements so that there is a need for at least 7 test strips per day.

https://www.deutsche-diabetes- gesell- schaft.de/fileadmin/Redakteur/Leitlinien/Praxi sleitlinien/2017/Praxisempfehlung_Glukosem onitoring_finale.pdf

Continuous interstitial glucose measurement with real-time measuring devices (rtCGM) may be provided at the expense of the statutory health insurance

• in patients with insulin- dependent diabetes mellitus,

• who need intensified insulin treatment, are trained in

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 169 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

this and already apply it,

• especially when the individual therapy goals for metabolic adjustment established between the doc- tor and the patient cannot be achieved even if the respective life situation of the patient is taken into account

• and if the requirements for quality assurance are met

HAS 2007 France Indications and prescription of a glycemic self- monitoring in a diabetic patient

https://www.has- sante.fr/portail/jcms/r_1437977/fr/indications-et- prescription-d-une-autosurveillance-glycemique-

chez-un-patient-diabetique-fiche-buts

Abbott Submission Self-monitoring must be: file 2018 • Systematic and multi-day in type 1 diabetes.

• Limited to certain patients, depending on clinical situations, in type 2 diabetes.

2018 • Inscribed in a process of patient education.

Glycemic self-monitoring should NOT be:

• A measure automatically generalized to all diabetics.

• A passive measure, with no immediate therapeutic consequences.

People with type 1 diabetes, T2 MDI diabetes and gestational diabetes, should perform at least 4 times a day SMBG.

CGM- Flash (publication S Borot et al. 2018)

Given the data in the literature, the working group considers FreeStyle Libre® as an alternative to

SMBG through the use of faster, easier, more French Diabetes frequent and more informative self-monitoring.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 170 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Society Its use is recommended as a replacement for SMBG in patients with T1D or T2D (adults and children) on intensified insulin therapy (pump or multiple injections; Fig. below). Abbott Submission file 2018 CGM systems are recommended in patients with T1D presenting with major hypoglycaemic problems (severe hypoglycaemia, hypoglycaemia unawareness, hypoglycaemia phobia). The working group suggests that institutional recommendations pay particular attention to populations at risk of hypoglycaemia in terms of indications for and reimbursement of CGM devices (Fig. below). Also, real-time transmission of data should be considered in children and in non-autonomous or isolated patients and raise the issue of care organization geared towards telemedicine.

SKL 28 sept 2017 Sweden http://dagensdiabetes.info/index.php/alla-senaste- nyheter/2628-nu-fardigt-vardprogram-nationellt- vardprogram-for-beh-med-insulinpump-cgm-fgm- vuxna-med-typ-1-diabetes-skl

Abbott Submission Use of CGM should be considered in adults file 2018 with type 1 diabetes, with or without insulin pump, if at least one of the following criteria is met.

1. Recurring problems with hyper and hypoglycaemia.

2. The person has had severe hypoglycaemia last year that required the help of another person.

3. The person has left HbA1c ≥70 mmol / mole or has not achieved individual HbA1c targets.

4. The person tests blood glucose frequently and

is in need of at least 10 blood glucose measurements per day to avoid hypoglycaemia and / or to

achieve individual HbA1c targets.

5. Hypoglycemic unawareness.

6. Before and during pregnancy of fluctuating blood glucose, hypoglycaemia or unsatisfactory HbA1c.

7. Expressed anxiety and fear of hypoglycaemia.

8. If the risk of hypoglycaemia is an obstacle to the work.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 171 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

9. Increased opportunities for physical activity without adverse effects like episodes of hypoglycaemia.

Indications for the use of FGM in adults with type 1 diabetes

Use of FGM should be considered in adults with Type 1 diabetes if at least one of the following criteria is met.

1. Recurring problems with hyper and hypoglycaemia where CGM is not working or not deemed necessary.

2. The person has had severe hypoglycaemia or hypoglycemic unawareness where CGM has not worked or is not deemed necessary.

3. The person has a residual HbA1c ≥70 mmol / mole or has not achieved individual HbA1c targets, and where CGM is not deemed necessary.

5. The person frequently tests blood glucose to avoid hypoglycaemia and / or to achieve individual HbA1c targets.

6. Expressed anxiety and fear of hypoglycaemia. If the risk of hypoglycaemia is an obstacle to the work.

7. Increased opportunities for physical activity without adverse effects such as episodes of hypoglycaemia.

8. Conditions that prevent adequate blood glucose measurement eg sexual dysfunction, disability or work in dirty environments or with food.

http://www.janusinfo.se/Documents/Nationellt_info rande_av_nya_lakemedel/FreeStyle-Libre- 180205.pdf

"Nursing program. National care program for insulin pump treatment, CGM, FGM, Type 1 diabetes adult". SKL describing that a total of about 70-75% of all type 1 diabetes patients meet the continuous glucose measurement (CGM or FGM) indications.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 172 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

NDF 2017 Netherlands http://www.zorgstandaarddiabetes.nl/

CGM is indicated for patients (with no psychosocial contra-indications and who adhere to thera- Dexcom Submis- py) who: sion file 2018 and update in June are on intensive insulin treated (with or without 2018 CSII) and who have

o Problematic hypoglycaemia

o Hypoglycaemia Unawareness, based on a self scored questionnaire, or

o uncontrolled DM for last 3 months, with a minimum of 4 SMBG per day, or

o T1D who have had > 2 DKA episodes for which they were admitted to hospital in the last year

o have unexplained glucose variability which cannot be solved with current treatment

o People with diabetes who have a wish to become pregnant

People with diabetes who have impaired hypo awareness are not eligible for reimbursement for Freestyle Libre® and should be on rtCGM devices only

Sources: (65, 70) *In terms of the strength of the recommendation, strong recommendations use the phrase “we recommend” and the number 1, and weak recommendations use the phrase “we suggest” and the number 2. Cross-filled circles indicate the quality of the evidence, such that ⊕○○○ denotes very low quality evidence; ⊕⊕○○, low quality; ⊕⊕⊕○, moderate quality; and ⊕⊕⊕⊕, high quality.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 173 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Evidence tables of individual studies included for clinical effectiveness and safety

In this chapter you can find evidence tables of individual studies included for clinical effectiveness and safety as well as corresponding RoB assessment tables of these individual studies. Summarised RoB assessment tables as well as GRADE assessment tables can be found in section “Risk of Bias and GRADE tables.

Characteristics of randomised controlled studies and other studies included (Tables) rtCGM MDII patients EVIDENCE TABLES

RCTs

rtCGM

MDII patients

® Beck 2017 RCT T1 MDII Dexcom G4 Plati- Change in HbA1c level; Percentage of participants with HbA1c Clinical effectiveness and safety domains Diamond level less than 7.0%; CGM-measured time in range (70-180 158/158 num/SMBG trial mg/dL);Duration of hypoglycaemia (<70 mg/dL, <60 mg/dL, and 24 weeks <50 mg/dL);Duration of hyperglycaemia (>180 mg/dL, >250 ≥25 y (26-73 USA mg/dL, and >300 mg/dL);Glucose variability (coefficient of y) variation);Change in hypoglycaemia unawareness; Change in NCT022823 frequency of blood glucose meter testing; CGM-measured mean 97 glucose concentration;HbA1c level less than 7.5%; Relative HbA1c reduction greater than or equal to 10%; HbA1c reduction of 1% or more;HbA1c level less than 7.0% or reduction of 1% or more; CGM-measured area above the curve 70 mg/dL and area under the curve 180 mg/dL; Change in insulin dose; Change in body weight; Satisfaction with CGM; Quality-of-life and health economic outcomes- separate article AEs

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 174 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 175 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author and Study Number of Intervention (s)/Control Main Included in clinical effectiveness and/ or year or study type/T1 or patients endpoints safety domain name/MDII or T2 DM or randomised/a CSII or both both/Study nalysed period ® Lind 2017 RCT T1 MDII Dexcom G4 Diference in HbA1c; other glycemic outcomes; well-being, Clinical effectiveness and safety domains Platinum/SMBG satisfaction, diabetes distress, Hypoglycaemic fear and Sweden 161/142 confidence; treatment adhetence; self-measurement of glucose; 26 weeks Hypoglycaemia; AEs Gold trial ≥18 y NCT020920 5 ® Beck 2017 RCT T2 MDII Dexcom G4 Change in HbA1c; Proportions of participants with HbA1c levels Clinical effectiveness and safety domains Platinum/SMBG below 7.0%, HbA1c levels below 7.5; Relative reduction of at least USA 158/158 10%, Reduction of at least 1%, Reduction of at least 1% or HbA1c 24 weeks level below 7.0%; Length of time per day the glucose Part of ≥25 y concentration was hypoglycemic (<3.89, <3.33, and <2.78 mmol/L DIAMOND [<70, <60, and <50 mg/dL]), hyperglycemic (>9.99, >13.88, and trial >16.65 mmol/L [>180, >250, and >300 mg/dL]), and in the target Protocol range (3.89 to 9.99 mmol/L [70 to 180 mg/dL]). Glucose variability; Scores on the Clarke Hypoglycaemia Unawareness Survey; 2 NCT022823 general quality-of-life measures (5-level EuroQol-5D and 5-item 97 World Health Organization Well-Being Index); 3 diabetes-specific quality-of-life measures (Hypoglycaemia Fear Survey, Diabetes Distress Scale, and Hypoglycemic Confidence Scale); The CGM group's satisfaction; Insulin use,Body weight;Frequency of blood glucose meter testing;All device or study-related adverse events;Severe hypoglycaemia (defined as an event that required assistance from another person to administer carbohydrates or other resuscitative action), Diabetic ketoacidosis or severe hyperglycaemia if treatment was received at a health care Facility; Serious adverse events regardless of causality

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 176 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author and Study Number of Intervention (s)/Control Main Included in clinical effectiveness and/ or year or study type/T1 or patients endpoints safety domain name/MDII or T2 DM or randomised/a CSII or both both/Study nalysed period ® Heinemann RCT T1 MDII Dexcom G5 /SMBG Number of hypoglycaemic events measured by rtCGM during the Clinical effectiveness and safety domains 2018 follow-up phase compared with baseline; changes in nocturnal 149/149 hypoglycaemic events (0000 h to 0600 h); percentage and Germany 26 weeks duration of glucose readings derived from continuous glucose ≥18 y monitoring per day in different glucose ranges (≤3·0 mmol/L [≤54 HypoDE Impaired mg/dL], ≤3·9 mmol/L [≤70 mg/dL], >3·9 mmol/L to ≤10·0 mmol/L study [>70 mg/dL to 180 mg/dL], and >10·0 mmol/L [>180 mg/dL]), and hypoglycae percentage of blood glucose readings based on SMBG NCT026719 mia measurements in these different glucose ranges; Impaired 68 awareness hypoglycaemia awareness assessed with the hypoglycaemia unawareness questionnaire; diabetes distress assessed with the Diabetes Distress Scale for type 1diabetes (T1-DDS); fear of hypoglycaemia assessed with the Hypoglycaemia Fear Survey; self-reported health status assessed with the European Quality of Life 5 Dimensions questionnaire (EQ-5D); and satisfaction with glucose measurement assessed with the Glucose Monitoring Satisfaction Survey

Author, year, reference Beck et al. 2017

Beck R, Riddlesworth T, Ruedy K et al. JAMA. 2017;317(4):371-378. doi:10.1001/jama.2016.19975

Study title/objectives Effect of Continuous Glucose Monitoring on Glycemic Control in Adults With Type 1 Diabetes Using Insulin Injec- tions: The DIAMOND Randomized Clinical Trial

Objective: To determine the effectiveness of CGM in adults with type 1 diabetes treated with insulin injections

Study characteristics

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 177 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Study design Randomized, open label, parallel group clinical trial

Study Registration number NCT02282397

Country of recruitment United Stetes

Centre (single or multicentre) Multicentre (24 endocrinology practices)

Ethics Committee Approval The protocol and Health Insurance Portability and Accountability Act–compliant informed consent forms were approved by institutional review boards (central commercial board for 17 sites and local boards for the other 7 sites)

Sponsor

Study period (study start, study end) October 2014 – May 2016

Duration of follow-up 24 weeks

Inclusion criteria 1. Age 25 years of age and older

2. Diagnosis of type 1 diabetes

3. Followed regularly by a physician or diabetes educator for their diabetes management – with at least 2 office

visits in last year as documented by clinical history

4. Using multiple daily injections of insulin for at least 12 months prior to study entry

5. Sub-optimal glycemic control, defined as persistent hyperglycaemia, confirmed initially by historical or local

lab (POC or site’s lab) A1C of ≥7.7% to ≤10%, then followed with a confirmatory result by central lab of ≥7.5% to ≤10%

6. Desire to lower A1C such as a goal of 7%

7. Stable control of diabetes, as determined per investigator assessment

8. Stable diabetes medication regimen for 3 months prior to study entry

9. Stable weight maintained 3 months prior to study entry, per investigator’s assessment, and not planning any

structured weight reduction interventions such as prescription weight loss medications, bariatric surgery, or

protein sparing modified fast during the course of the study.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 178 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

10. Willing to wear a device (CGM)

11. Willing to avoid use of acetaminophen medications throughout the study

12. Currently performing self-monitoring blood glucose testing (by history) an average of >3 times per day

13. Able to speak, read, and write English

Exclusion criteria 1. Use of personal RT-CGM 3 months prior to study entry (professional CGM use, blinded or un-blinded, is acceptable)

2. Use of CSII 3 months prior to study entry (including patch pumps)

3. Plan to use personal CGM and/or pump during the course of the study

4. Addition of any new oral or injectable hypoglycemic agents (including GLP-1 analogues, Pramlintide, and SGLT-2 inhibitors – these agents are only for T2DM participants) within 3 months prior to study entry. (Use of these agents does not affect eligibility if used 3 or more months prior to study entry.) For GLP-1 medications, must be on stable dose and the GLP-1 medication will be maintained throughout the study.

Note: These agents should not be added or modified during course of the study. If use of this class medication is planned, the patient is not eligible.

5. Use of pre-mixed insulin (e.g. 70/30 or 50/50) 6 months prior to study entry

6. Current or anticipated acute uses of glucocorticoids (oral, injectable, or IV), that will affect glycemic control and impact A1C – such as frequent steroid bursts required for inflammatory arthritis or inflammatory bowel disease, recurrent lumbar epidural steroid injections, etc. (Long-term stable glucocorticoid doses are allowed, such as when used for the treatment of rheumatoid arthritis or Addison’s disease).

7. Pregnancy (as demonstrated by a positive test at study entry) at time of screening or are planning to become pregnant during the study

8. Medical conditions that, per investigator determination, make it inappropriate or unsafe to target an A1C of <7%. Condi- tions may include but are not limited to:

• Unstable. recent cardiovascular disease,

• Recent myocardial infarction

• Significant heart failure

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 179 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• Ventricular rhythm disturbances

• Recent transient ischemic attack, or cerebrovascular accident

• Significant malignancy

• Other conditions resulting in physical or cognitive decline

• Recurrent severe hypoglycaemia

9. History of visual impairment which would hinder participant’s participation in the study and perform all study procedures safely, as determined by investigator

10. History of psychiatric, psychological disorder, or psycho-social issues that could limit adherence to the required study tasks

11. Renal disease defined as estimated Glomerular Filtration Rate eGFR <45

12. Extensive skin changes/disease that preclude wearing the sensor on normal skin (e.g. extensive psoriasis, recent burns or severe sunburn, extensive eczema, extensive scarring, extensive tattoos, dermatitis herpetiformis)

13. Known allergy to medical-grade adhesives

14. Current participation in another investigational study (must have completed any previous studies at least 30 days prior to being enrolled in this study)

15. Recent hospitalization or emergency room visit in the 6 months prior to screening resulting in a primary diagnosis of uncontrolled diabetes

16. Currently abusing illicit drugs, alcohol, or prescription drugs

17. Any condition, per investigator assessment, that could impact reliability of the A1C measurement, such as (but not limited to) hemoglobinopathy, hemolytic anemia, chronic liver disease; chronic GI blood loss, red blood cell transfusion or erythropoietin administration within 3 months prior to screening

Patient characteristics

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 180 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Age of patients Mean age - 48 years (SD,13) [range, 26-73 y]; 34(22%) ≥60 y

Baseline CGM Control

Age, Mean, (SD)[ 46 (14) [26-72] 51 (11)[26-73]

range], y

Sex 44% woman

Baseline CGM Control

group

No, (%) Woman 47 (45) 23 (43)

BMI CGM group Control

BMI mean, (SD) 28 (6) 27 (5)

Diagnosis (Type I, II or gestational) Type 1

Comorbidities (i.e. obesity…)

Time since diagnosis with DM, median (IQR), y 19 y (IQR, 10-31 years)

Baseline CGM group Control

Time since diagnosis 19(9-29) 19 (11-35)

with DM, median (IQR), y

Insulin treatment (CSII or MDII) MDII

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 181 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Pregnancy (yes, no) No

Exclusion criteria - Pregnancy (as demonstrated by a positive test at study entry) at time of screening or are planning to become pregnant during the study

Special subgroup pf patient (i.e., hypoglycaemia fear…) NR

Intervention

Type of medical device (CGM or FGM) CGM

® Name (Description) of medical device Dexcom G4 Platinum

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical SMBG (Bayer Contour Next) devices)

Name (Description) of medical device Bayer Contour Next meter with test stripes

Sensor integrated (Yes, No) NR

Sensor augmented (or enabled) insulin pump systems compatible (con- NR nected) with specific CGM systems (Yes, No)

Outcomes

Primary Change in the central laboratory measured HbA1c level

Secondary Percentage of participants with HbA1c level less than 7.0%

CGM-measured time in range (70-180 mg/dL),

Duration of hypoglycaemia (<70 mg/dL, <60 mg/dL, and <50 mg/dL),

Duration of hyperglycaemia (>180 mg/dL, >250 mg/dL, and >300 mg/dL),

Glucose variability (coefficient of variation)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 182 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Change in hypoglycaemia unawareness

Change in frequency of blood glucose meter testing

CGM-measured mean glucose concentration

HbA1c level less than 7.5%

Relative HbA1c reduction greater than or equal to 10%

HbA1c reduction of 1% or more,

HbA1c level less than 7.0% or reduction of 1% or more

CGM-measured area above the curve 70 mg/dL and area under the curve 180 mg/dL

Change in insulin dose

Change in body weight

Satisfaction with CGM

Quality-of-life and health economic outcomes- separate article

Safety outcomes

Severe hypoglycaemia (defined as an event that required assistance from another person to administer carbohydrate, glucagon, or other resuscitative actions)

Diabetic ketoacidosis

Serious adverse events regardless of causality

Flow of patients

No of patients enrolled 186

No of randomized 158

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 183 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Allocated per arms

CGM group Control

No of patients 105 53

Received int. per arms

CGM group Control

No of patients 105 53

Lost to follow-up per arms

CGM group Control

No of patients 1 lost of follow up 0

1 Site withdrew

1 requested to withdraw

Completed study

CGM group Control

No of patients 102 53

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 184 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

No of analysed per arm

CGM group Control

No of patients 105 53

4 Imputation used for HbA1c

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) ITT

Impute for missing data – Rubin method

Results

Effectiveness results

n (%) 95% CI

Mortality NR

Change in HbA1c Primary outcome, 12 weeks 24 weeks Mean adjusted P value mean (SD) difference, %(95% CI)

CGM Control CGM Control

N=103 N=52 N=105 N=53

Change in HbA1c -1.1 -0.5 -1.0 -0.4 (0.7) -0.6 <0.001 from baseline (0.8) (0.7) (0.7) (-0.8 to -0.3)

Other HbA1c outcomes

Primary outcome, 12 weeks 24 weeks Mean adjusted P value mean (SD) difference,

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 185 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

%(95% CI)

CGM Control CGM Control

N=103 N=52 N=105 N=53

HbA1c 7.6 (0.7) 8.1 (0.7) 7.7 (0.8) 8.2 (0.8) -0.6 <0.001

(-0.8 to -0.3)

HbA1c <7% 14 (14) 2 (4) 18 (18) 2 (4) 15 (0 to 30) 0.01

HbA1c <7.5% 49 (48) 6 (12) 39 (38) 6 (11) 31 (12 to 51) <0.001

Relative reduction in 62 (60) 12 (23) 58 (57) 10 (19) 37 (16 to 58) <0.001 HbA1c ≥10%

HbA1c reduction 55 (53) 12 (23) 53 (52) 10 (19) 33 (11 to 54) <0.001 ≥1%

HbA1c level <7.0% 57 (55) 12 (23) 53(52) 11 (21) 31 (9 to 52) <0.001 or reduction ≥1%

Hypoglycaemic Event Frequency

Hypoglycaemia events lasting at least 20 min at less than 3.0 mmol/L (54 mg/dL) separated by at least 15 min

Separate article: Median (IQR) CGM Control group P VALUE Riddlesworth T, Price D , Cohen N, Beck RW Hypoglycemic Event

Frequency and the Effect of Continuous Glucose Monitoring in Adults with Type 1 Diabetes Using Multiple Daily Insulin Injections Baseline Follow up Baseline Follow up Diabetes Ther (2017) 8:947–951 N=105 N=103 N=105 N=103 DOI 10.1007/s13300-017-0281-4

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 186 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of 3 (2,7) 2 (1,4) 4 4 events per 2 (1,7) (1 to 6) weeks, Median (IQR)

Number of 10 (10%) 25 (24%) 6 (11%) 9 (17%) events per 2 weeks, N(%)

Event rate per 0.23 0.16 0.31 0.30 24 h, Median (0.15, 0.46) (0.07, 0.30) (0.08, 0.54) (0.09, 0.46) (IQR)

Change in event - 0.08 - 0.00 0.03 rate per 24 h, (-0.23, 0.07) (-0.19, 0.01) Median (IQR)

Higher precentage of participants in the control group had reported severe hypoglycaemia (seizure or loss of consciousness)

level 1 hypoglycaemia NR

level 2 hypoglycaemia NR

level 3 hypoglycaemia NR

Incidence of hyperglycaemia NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 187 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time spent in range, mean (SD)

mean (SD) Baseline 12 and 24 weeks Pooled Mean ad- P value justed difference, %(95% CI)

CGM Control CGM Control

N=105 N=53 N=103 N=53

Minutes per day 660 (179) 650 (170) 736 650 77 (6 to 147) 0.005 in range 70 -180 (206) (194) mg/dL

Time spent in hypoglycaemia, median (IQR)

median (IQR) Baseline 12 and 24 weeks Pooled P value

CGM Control CGM Control

N=105 N=53 N=103 N=53

Minutes per day < 65 (33 to 103) 72 (35 to 136) 43 80 0.002 70mg/dL (27 to 69) (36 to 111)

Minutes per day < 32 (15 to 61) 39 (15 to 78) 20 40 0.002 60mg/dL (9 to 30) (16 to 68)

Minutes per day < 13 (5 to 29) 18 (4 to 39) 6 (2 to 12) 20 (4 to 42) 0.001 50mg/dL

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 188 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Area above curve 70 mg/dL

median (IQR) Baseline 12 and 24 weeks Pooled P value

CGM Control CGM Control

N=105 N=53 N=103 N=53

Area above curve 0.5 (0.3 to 1.1) 0.7 (0.2 to 1.4) 0.3 (0.2 to 0.5) 0.7 (0.2 to 1.3) <0.001 70 mg/dL

Time spent in hyperglycaemia

median (IQR) Baseline 12 and 24 weeks Pooled P value

CGM Control CGM Control

N=105 N=53 N=103 N=53

Minutes per day > 687 (554 to 810) 725 (537 to 798) 638 740 0.003 180mg/dL (503 to 807) (625 to 854)

Minutes per day > 301 (190 to 401) 269 (184 to 383) 223 347 <0.001 250mg/dL (241 to 429) (241 to 429)

Minutes per day 129 (66 to 201) 109 (71 to 204) 78 (36 to 142) 167 (89 to <0.001 >300mg/dL 226)

Area under curve 180 mg/dL

median (IQR) Baseline 12 and 24 weeks Pooled P value

CGM Control CGM Control

N=105 N=53 N=103 N=53

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 189 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Area above curve 34 (25 to 46) 33 (26 to 45) 27 (17 to 40) 40 (31 to 51) <0.001 180 mg/dL

Glucose variability:coefficient of variation, mean (SD)

mean (SD) Baseline 12 and 24 weeks Pooled Mean adjusted P value difference, %(95% CI)

CGM Control CGM Control

N=105 N=53 N=103 N=53

Glucose variabil- 42 (7) 42 (7) 38 42 -4 (-6 to -2) <0.001 ity: coefficient of (6) (7) variation

Mean Glucose, mean (SD), mg/dL

mean (SD) Baseline 12 and 24 weeks Pooled Mean adjusted P value difference, %(95% CI)

Mean Glucose, CGM Control CGM Control mean (SD), N=105 N=53 N=103 N=53 mg/dL

187 (7) 186 (30) 180 189 -9 (-19 to 0) 0.01

(27) (25)

Quality of life NR; reported in Polonsky 2017

Patient satisfaction with use of CGM, mean (SD)

CGM group

Patient satisfaction with use of CGM mean (SD)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 190 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

CGM Satisfaction survey 4.2 (0.4)

Benefit subscale 4.2 (0.5)

Subscale for lack of hassles 4.3 (0.5)

Results for all 44 questions –Table 19 Supplement

Hypoglycaemia fear NR

Hypoglycaemia unawareness

mean ± SD Baseline 24 weeks Mean adjusted P difference, value %(95% CI)

CGM Control CGM Control

N=105 N=53 N=102 N=53

Clarke Hypogly- 2.1 ± 1.8 2.7 ± 2.1 2.0 ± 1.8 2.5 ± 2.1 - - caemia

Unawareness Questionnaire

Total Score ,

mean ± SD

Change in Clarke - - -0.2 ± 1.3 -0.3 ± 1.6 -0.1 (-0.7 to 0.64 Hypoglycaemia +0.5)

Unawareness Questionnaire,

Total Score

mean ± SD

Incidence of diabetic ketoacidosis

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 191 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

CGM (N=105) Control (N=53)

Diabetic ketoacidosis 0 0

Incidence of hyperosmolar, hyperglycaemic coma NR

Resource utilization related to DM NR

Number of visits to emergency room NR

Number of visits to primary care NR

Number of visits to specialists NR

Number of hospitalizations NR

Number of daily finger-sticks tests, mean (SD)

Number of daily finger-sticks tests, CGM Control Adjusted mean difference mean (SD) for the change

Baseline period of blinded CGM (2 5.1 (1.8) 5.1 (1.4) weeks)

At 24 weeks 3.6 (1.6) 4.6 (1.8) -1.0; 99%CI, -1.7 to -0.4; P<0.001

Number of calibration NR

Need (Yes, with number or No) of re-calibration NR

Compliance/adherence NR

Percentage of time using CGM

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 192 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Continuous Glucose Week 4 Visit Week 12 Visit Week 24 Visit Overall Monitoring Use in (N=105 ) (N=103) (N=102 ) (N=102 ) CGM Group

N (%) unless otherwise stated

median (interquartile 7.0 (7.0-7.0) 7.0 (7.0-7.0) 7.0 (7.0-7.0) 7.0 (6.9-7.0) range)

Range 3.3-7.0 0.0-7.0 0.0-7.0 2.3-7.0

Zero use 0 1(<1%) 2 (2%) 0

<1 day/week 0 0 0 0

1-<2 days/week 0 0 0 0

2-<3 days/week 0 0 0 1 (<1%)

3-<4 days/week 1 (1%) 0 0 0

4-<5 days/week 0 0 1 (1%) 3 (3%)

5-<6 days/week 0 3 (3%) 4 (4%) 2 (2%)

6-<7 days/week 11 (11%) 7 (7%) 12 (12%) 22 (22%)

7 days/week 88 (88%) 92 (89%) 79 (81%) 74 (73%)

<6 days/week 1 (1%) 4 (4%) 7 (7%) 6 (6%)

≥6 days/week 99 (99%) 99 (96%) 91 (93%) 96 (94%)

% of Possible Read- 98% 96% (92%-98%) 96% 96% (92%-98%) ings b median (inter- (95%-99%) (90%-98%) quartile range)

Number of sensor scans per day (in FGM system) NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 193 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Safety results

n (%) 95% CI

Any AEs NR

Serious AE (SAE)

CGM (N=105) Control (N=53) P value

Diabetic ketoacidosis 0 0

Severe Hypoglycaemia

# Events Per Participant 0.67

0 103 51

1 2 1

2 0 1

24-week Kaplan-Meier Incidence +2% (-4% to +8%) +4% (-6% to +14%) 0.49

(95% Confidence Interval)

24-week Incidence Rate (per 100 person-years) 4.2 12.2 0.27

Other Serious Adverse Events(No 3/2 0/0 events/No Serious adverse events participants) were inner ear disorder, pulmonary mass, and trigeminal neuralgia

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 194 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Most frequent AEs (by arms) NR

Most frequent SEAs (by arms) NR

Death as SAE NR

Withdrawals due AEs NR

Costs (only for national assessment) Reported in separate articles

Author Disclosure (Conflict of interest) All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dexcom Inc provided funding for the trial to each investigator’s institution. Dr Beck reports receiving a study grant from Dexcom and that his institution received supplies for research from Dexcom and Abbott Diabetes Care for other studies. Dr Ahmann reports receiving grants for the study and consulting for Dexcom Inc; receiving grants for research support from Medtronic, Novo Nordisk, Lexicon, and Sanofi; consulting for Novo Nordisk, Sanofi, and AstraZeneca; and serving on advisory boards for Lilly, Janssen, and AstraZeneca. Dr Bergenstal reports receiving a study grant from Dexcom and NIH; reports serving on the advisory boards for and/or receiving study funding from Abbott Diabetes Care, AstraZeneca, Becton Dickinson, Boehringer Ingelheim, Calibra, Eli Lilly, Halozyme, Hygieia, Johnson & Johnson, Medtronic, Novo Nordisk, Roche, Sanofi, and Takeda; and reports holding stock in Merck. Ms Kruger reports holding stock in Dexcom. Dr McGill reports receiving grant funding from Novartis, Novo Nordisk, Lexicon, Bristol-Myers Squibb, and Dexcom; and consulting fees from Boehringer Ingelheim, Dexcom, Lilly, Merck, Novo Nordisk, Intarcia, Dynavax, Valeritas, Janssen, and Calibra. Dr Polonsky reports consulting for Dexcom. Dr Wolpert reports receiving grant funding from Abbott Diabetes Care. Dr Price is an employee of Dexcom, Inc and reports holding stock in the company. No other disclosures were reported.

RR: relative risk ITT: intention to treat PP: per protocol

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 195 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Risk of Bias

Study (Author, year): Beck et al. 2017 DIAMOND TRIAL NCT02282397

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Each participant was assigned randomly from a computer-generated sequence to either the CGM or control group in a 2:1 ratio, with a permuted block design (block sizes of 3 and 6) stratified by HbA1c level (<8.5% and ≥8.5%). A 2:1 randomization was used rather than 1:1 to provide a larger sample size for a separate follow-on randomized trial assessing glycemic benefits of initiating pump therapy in CGM users using insulin injections.

Allocation concealment (Selection bias) Unclear Not described

Blinding of participants (Performance bias) High Open label

Blinding of personnel (Performance bias) High Open label

Blinding of outcome assessment (Detection bias) Unclear Not described

Incomplete outcome data (Attrition bias) Low Intent-to-treat principle; The Rubin method was used to impute for missing data; The 24-week primary study outcome visit was completed by 102 participants (97%) in the CGM group and all 53 (100%) in the control group. Overall visit completion was 99% and 98%, respectively.

Selective reporting (Reporting bias) Low No

Other source of bias (Other bias) Unclear In light of the eligibility criteria, the results may not apply to individuals with type 1 diabetes who are younger than 26 years or have HbA1c levels outside the range of 7.5% to 9.9% and should not be applied to individuals with type 2 diabetes who receive multiple daily injections of insulin. The informed consent process and the run-in phase had the potential to exclude individuals who might be less adherent with CGM than the cohort that was studied; Funding source

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 196 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Risk of Bias

Study (Author, year): Riddlesworth et al. 2017 From DIAMOND TRIAL NCT02282397; hypoglycaemic event frequency

Judgement (Low, Unclear, Support for judgement High)

Allocation concealment (Selection bias) Unclear Not described

Blinding of participants (Performance bias) High Open label

Blinding of personnel (Performance bias) High Open label

Blinding of outcome assessment (Detection bias) Unclear Not described

Selective reporting (Reporting bias) Low No

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 197 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

define a hypoglycaemic event. A hypoglycemic event was defined as a series of at least two sensor glucose values less than 3.0 mmol/L (54 mg/dL), lasting at least 20 min, with no intervening values of 3.0 mmol/L or more. The end of a hypoglycemic event was defined as a minimum of 15 consecutive minutes with at least two sensor glucose values of at least 3.0 mmol/L and at least 0.6 mmol/L (10 mg/dL) above the nadir of the event. A new event was temporally separated from any

previous event by 15 min or more, with no intervening values less than 3.0 mmol/L.

Author, year, reference Ruedy et al. 2017

Ruedy KJ, Parkin CG, Riddlesworth TD, Graham C; DIAMOND Study Group. Continuous Glucose Monitoring in Older Adults With Type 1 and Type 2 Diabetes Using Multiple Daily Injections of Insulin: Results From the DIAMOND Trial. J Diabetes Sci Technol. 2017 Nov;11(6):1138-1146. doi:10.1177/1932296817704445.

Study title/objectives Continuous Glucose Monitoring in Older Adults With Type 1 and Type 2 Diabetes Using Multiple Daily Injections of Insulin: Results From the DIAMOND Trial.

Objective: The objective was to determine the effectiveness of real-time continuous glucose monitoring (CGM) in adults ≥ 60 years of age with type 1 (T1D) or type 2 (T2D) diabetes using multiple daily insulin injections (MDI).

This paper present findings from a subset analysis of the full DIAMOND study population that assessed the effectiveness of CGM use in MDI-treated T1D and T2D individuals ≥60 years of age.

Study characteristics

Study design Multicenter, randomized trial

Study Registration number NCT02282397 www.clinicaltrials.gov

Country of recruitment United States and Canada

Centre (single or multicentre) Multicenter (29 endocrinology practices)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 198 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Ethics Committee Approval The protocol and Health Insurance Portability and Accountability Act (HIPAA)-compliant informed consent forms were approved by a central and multiple local institutional review

Boards.

Sponsor Dexcom, Inc provided funding for the trial to each of the investigator’s institutions.

Study period (study start, study end) 24 weeks

Duration of follow-up (days) Follow-up visits (both treatment groups): after 4, 12 and 24 weeks

CGM group : additional visit 1 week after randomization.

Control group: two additional visits 1 week before the 12- and 24-week visits (to initiate blinded CGM use for 1week).

Phone contacts for both groups occurred 2 and 3 weeks after randomization.

Inclusion criteria • Age ≥60 years • Diagnosis of T1D or T2D • Multiple daily injections of insulin treatment for at least one year • Central laboratory measured HbA1c 7.5% to 10.0% • Stable diabetes medication regimen and weight over the prior 3 months • Selfreported blood glucose meter testing averaging 2 or more times per day for T2D and 3 or more for T1D • Estimated glomerular filtration rate ≥45

Exclusion criteria • Use of real-time CGM within 3 months of screening • Any medical condition(s) that would make it inappropriate or unsafe to target an HbA1c of <7.0%.

Patient characteristics CGM group (n=63) Control group (n=53)

Age of patients, years (mean ± SD) 67 ± 5 67 ± 5

Sex (n, %) Female: 34 (54) Female: 26 (49)

BMI, kg/m2, mean (SD) 33 ± 7 34 ± 8

Diagnosis (Type I, II or gestational) Type I: 20 (32) Type I: 14 (26)

Type II: 43 (68) Type II: 39 (74)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 199 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Comorbidities (i.e. obesity…) - -

Time since diagnosis with DM, duration, years median 20 (14, 32) 21 (16, 30)

(quartiles)

Insulin treatment (CSII or MDI) MDI MDI

Pregnancy (yes, no) - -

Special subgroup of patient : Hypoglycaemia awareness

• Reduced awareness 11 (17) 7 (13) • Uncertain 10 (16) 5 (9) • Aware 42 (67) 41 (77)

Intervention

Type of medical device (CGM or FGM) CGM (adjunctive to SMBG)

Name (Description) of medical device DexcomG4® Platinum CGM System with an enhanced algorithm (software 505) (Dexcom, Inc, San Diego, CA)

Contour Next USB meter and test strips for SMBG (Ascensia Diabetes Care, Parsippany, NJ, USA)

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device Contour Next USB meter and test strips (Ascensia Diabetes Care, Parsippany, NJ, USA).

The Control group was asked to perform home blood glucose monitoring at least 4 times daily.

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible (con- - nected) with specific CGM systems (Yes, No)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 200 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Outcomes

Primary Change in the central-laboratory measured HbA1c from baseline to 24 weeks.

Secondary Amount of time per day the glucose concentration was hypoglycemic (<60 mg/dL)

Amount of time per day the glucose concentration was hyperglycemic (>250 mg/dL)

Amount of time per day the glucose concentration was in the target range of 70 to 180 mg/dL.

Glucose variability

Frequency of blood glucose self-monitoring

Flow of patients

No of patients enrolled 143

No of randomized 116

Allocated per arms CGM group: 63 Control group:53

Received int. per arms CGM group: 63 Control group:53

Lost to follow-up per arms CGM group: 2 (1 lost to follow up; 1 site withdraws partic- Control group:0 ipant)

No of analysed per arm CGM group: 61 Control group:53

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) -

Results

Effectiveness results n (%) 95% CI

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 201 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Mortality -

Change in HbA1c CGM Control

Baseline 12 Reduction from Baseline 12 weeks Reduction from Adjusted difference in mean change Adjusted difference in mean change weeks baseline at 24 baseline at 24 after 12 weeks (CGM vs control) and after 24 weeks (CGM vs control) and weeks weeks P value P value

Mean HbA1c 8.4 ± 0.6% 7.5 ± −0.9 ± 0.7% 8.6 ± 0.7% 8.0 ± 0.7% −0.5 ± 0.7% −0.4 ±0.1% (P < .001) −0.4 ± 0.1% (P < .001) 0.7%

Incidence of hypoglycaemia - level 1 hypoglycaemia - level 2 hypoglycaemia - level 3 hypoglycaemia -

Incidence of hyperglycaemia -

CGM Control

Baseline (n=63) 12 weeks (n=61) 24 weeks (n=58) Baseline (n=53) 12 weeks (n=52) 24 weeks (n=50) P value

Mean glucose, mg/dL 175 ± 25 167 ± 27 168 ± 29 179 ± 30 178 ± 28 180 ± 28 0.01 (mean ± SD)

Glycemic variability, 34 (28, 42) 33 (28, 37) 31 (28, 36) 34 (29, 38) 33 (28, 38) 33 (27, 39) 0.02 coefficient of variation % (median (IQR))

Time spent in range: 796 ± 236 892 ± 256 889 ± 251 753 ± 253 767 ± 265 732 ± 252 <0.001 70-180 mg/dL, min/day (mean ± SD)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 202 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time spent inhyper- 172 (83, 281) 93 (30, 180) 89 (37, 208) 208 (112, 294) 180 (81, 251) 179 (83, 316) 0.006 glycaemia : >250 mg/dL, min/day (median (IQR))

Time spent in hypo- 10 (1, 38) 4 (0, 15) 3 (0, 15) 8 (1, 23) 4 (0, 27) 4 (0, 24) 0.11 glycaemia : <60 mg/dL, min/day (median (IQR))

Quality of life -

Patient satisfaction

CGM satisfaction CGM Satisfaction Survey (possible Benefits subscale Hassles subscale score range 1 to 5)

• Mean value (overall) 4.2 ± 0.4 4.3 ± 0.5 1.8 ± 0.5 • Detailed table (per questions) → Listed at the end of the document

Hypoglycaemia fear -

Incidence of diabetic ketoacidosis -

Incidence of hyperosmolar, hyperglycaemic coma -

Resource utilization related to DM -

Number of visits to emergency room -

Number of visits to primary care -

Number of visits to specialists -

Number of hospitalizations -

Mean reduction in the number of daily blood glucose tests from baseline to CGM group Control group P value

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 203 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

week 24 −1.2 ± 1.6 −0.2 ± 1.4 0 .001

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Compliance/adherence -

Percentage of time using CGM -

Mean CGM use (days/week) (CGM group only, n=61) Month 1 (weeks 1-4) Month 6 (weeks 21-24)

6.9 ± 0.2 days/week 6.8 ± 1.1 days/week

Percentage of patients used CGM≥6 days/week in month 6 97%

Number of sensor scans per day (in FGM system) -

Safety results There were no severe hypoglycaemia or diabetic ketoacidosis events in either group. n (%) 95% CI

Any AEs -

Serious AE (SAE) -

Most frequent AEs (by arms) -

Most frequent SEAs (by arms) -

Death as SAE -

Withdrawals due AEs -

Costs (only for national assessment) -

Author Disclosure (Conflict of interest) The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dexcom, Inc, San Diego, CA: David Price; Eileen Casal; Claudia Graham.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 204 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Andrew Ahmann Andrew Ahmann reports consulting from Dexcom, Inc as well as grants from Medtronic and T1D Exchange, grants and consulting from Novo Nordisk and Sanofi, and consulting from Lilly and AstraZeneca.

Roy Beck Roy Beck reports that his institution has received supplies for research studies from Dexcom, Inc and Abbott Diabetes Care.

Ronnie Aronson Ronnie Aronson reports grants from Sanofi, Takeda, BD, GlaxoSmithKline, Sanofi, Merck, Janssen, Medtronic, Boehringer Ingelheim, Regeneron, Bristol-Myers Squibb, AstraZeneca, Novo Nordisk, Novartis, Eli Lilly, Abbott, Quin- tiles, ICON, and Medpace, and consulting from Novo Nordisk, Janssen, Sanofi, Medtronic, Bristol-Myers Squibb, AstraZeneca, Takeda, and Amgen.

Richard Bergenstal Richard Bergenstal sits on the advisory board for Abbott Diabetes Care, AstraZeneca, Becton Dickinson, Boehringer Ingelheim, Calibra, Eli Lilly, Hygieia, Johnson & Johnson, Medtronic, Novo Nordisk, Roche, Sanofi, Takeda, and ResMed; in addition, Dr. Bergenstal holds stock in Merck.

Davida Kruger Davida Kruger is an advisor, CME speaker, and stockholder for Dexcom, Inc; she also advises and speaks for Abbott Diabetes Care and Animas.

Janet McGill Janet McGill reports grant funding from Novartis, Lexicon, and Dexcom as well as consulting fees from Boehringer Ingelheim, Dexcom, Lilly, Merck, Novo Nordisk, Janssen, 8 Journal of Diabetes Science and Technology 00(0) and Calibra.

William Polonsky William Polonsky reports personal fees from Dexcom, Inc.

David Price David Price is an employee of Dexcom, Inc and is a stock holder.

Stephen Aronoff Nothing to report

Stacie Haller Nothing to report

Craig Kollman Nothing to report

Tonya Riddlesworth, Nothing to report

Katrina Ruedy Nothing to report

Elena Toschi Nothing to report

RR: relative risk ITT: intention to treat PP: per protocol

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 205 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

“CGM satisfaction”

CGM satisfaction Mean score Agree strongly Agree (%) Neutral (%) Disagree (%) Disagree strongly (%) (%)

1. Causes me to be more worried about controlling blood sugars. 3.3 18 15 15 18 33

2. ►Makes adjusting insulin easier. 4.4 52 40 5 3 0

3. ►Helps me to be sure about making diabetes decisions. 4.4 52 38 7 2 0

4. Causes others to ask too many questions about diabetes 3.7 5 12 20 35 28

5. Makes me think about diabetes too much. 3.7 2 13 22 37 25

6. ►Helps to keep low blood sugars from happening. 4.4 53 32 12 2 0

7. ►Has taught me new things about diabetes that I didn’t know before. 4.4 52 38 10 0 0

8. Causes too many hassles in daily life. 4.2 3 2 15 35 45

9. ►Teaches me how eating affects blood sugar. 4.5 58 37 5 0 0

10. ►Helps me to relax, knowing that unwanted changes in blood sugar will be detected 4.4 50 40 10 0 0 quickly.

11. ►Has helped me to learn about how exercise affects blood sugar. 4.2 38 43 18 0 0

12. ►Helps with keeping diabetes under control on sick days. 4.1 32 45 22 0 0 4.1 32 45 22 0 0

13. ►Has shown me that blood sugar is predictable and orderly. 3.5 22 42 12 17 8

14. Sometimes gives too much information to work with. 3.9 2 8 13 52 25

15. ►Has made it easier to accept doing blood sugar tests. 4.1 37 40 15 3 2

16. Is uncomfortable or painful. 4.2 2 3 12 40 43

17. ►Has helped me to learn how to treat low sugars better. 4.2 47 32 17 5 0

18. Is more trouble than it is worth. 4.5 3 0 5 27 65

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 206 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

19. ►Has helped my family to get along better about diabetes. 3.9 28 37 30 3 2

20. ►Shows patterns in blood sugars that we didn’t see before. 4.4 47 50 3 0 0

21. ►Helps prevent problems rather than fixing them after they’ve happened. 4.4 48 45 7 0 0

22. ►Allows more freedom in daily life. 4.1 37 43 15 5 0

23. ►Makes it clearer how some everyday habits affect blood sugar levels. 4.4 45 50 5 0 0

24. ►Makes it easier to complete other diabetes self-care duties. 4.2 33 50 15 2 0

25. Has caused more family arguments. 4.4 0 7 10 23 60

26. Is too hard to get it to work right. 4.4 0 0 8 45 45

27. Has been harder or more complicated than expected. 4.3 0 5 8 38 48

28. ►Has helped to control diabetes better even when not wearing it. 3.5 13 43 28 10 5

29. Causes our family to talk about blood sugars too much. 4.0 0 3 20 47 28

30. Makes it harder for me to sleep. 4.2 2 5 13 33 47

31. Causes more embarrassment about feeling different from others. 4.5 2 0 3 38 57

32. Shows more “glitches” and “bugs” than it should. 4.1 3 7 8 45 37

33. Interferes a lot with sports, outdoor activities, etc. 4.3 0 2 10 50 38

34. Skips too many readings to be useful. 4.4 0 2 5 45 48

35. Gives a lot of results that don’t make sense. 4.2 2 3 7 52 37

36. Causes too many interruptions during the day. 4.4 0 2 3 45 50

37. Alarms too often for no good reason. 4.3 0 3 5 48 42

38. ►Has helped to adjust pre-meal insulin doses. 4.2 38 45 13 3 0

39. The feedback from the device is not easy to understand or useful. 4.2 0 3 8 48 38

40. I don’t recommend this for others with diabetes. 4.7 0 2 2 20 77

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 207 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

41. ►Has made me worry less about having low blood sugars. 4.2 43 37 15 3 2

42. ►If possible, I want to use this device when the research study is over. 4.5 65 27 5 2 2

43. ►Helps in adjusting doses of insulin needed through the night. 4.2 38 47 13 2 0

44. ►Makes me feel safer knowing that I will be warned about low blood sugar before it 4.6 68 27 3 2 0 happens.

Overall mean ± SD score 4.2 ± 0.4. Benefits subscale mean ± SD score 4.3 ± 0.5 (items 2, 3, 6, 7, 9, 10, 11, 12, 17, 20, 21, 22, 23, 24, 38, 41, 42, 43,

44). Hassles subscale mean ± SD score 1.8 ± 0.5 (items 4, 5, 8, 14, 16, 18, 25, 26, 27, 29, 30, 31, 32, 33, 34, 35, 36, 37, 39, 40). One participant did not complete the questionnaire at the conclusion of the study. Responses were missing from 1 participant for items 3, 5, 6, 12, 26, 29, 37, and 39 and missing from 2 participants for item 15. Items with a ► symbol are positively worded (agreeing corresponds to more satisfaction) and those without the symbol are negatively worded (agreeing corresponds to less satisfaction). To calculate the mean value for each item and the overall mean value, the scores for the positively worded items were reversed so that a higher score always corresponds to greater satisfaction. For example, a value of 5 corresponds to

“agree strongly” with a positively worded item, or “disagree strongly” with a negatively worded item. To calculate the subscale mean values, scores for all questions were reversed so that a higher score on the Benefits subscale denotes greater satisfaction and a higher score on the Hassles subscale denotes less satisfaction [37].

Risk of Bias

Study (Author, year): Ruedy 2017. Subgroup analysis from DIAMOND Trial NCT02282397; individuals ≥60 years of age

Judgement (Low, Support for judgement Unclear, High)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 208 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Allocation concealment (Selection bias) Unclear Not described

Blinding of participants and personnel (Performance bias) High Open label

Blinding of outcome assessment (Detection bias) High Open label

Blinding of outcome assessment (Detection bias) Unclear Not described

Incomplete outcome data (Attrition bias) Low Intent-to-treat principle; The 24-week primary study outcome visit was completed by 67 participants (97%) in the CGM group and all 53 (100%) in the control group.

Selective reporting (Reporting bias) Low No

Other source of bias (Other bias) Unclear In light of the eligibility criteria, the results may not apply to individuals with type 1 diabetes who are younger than 26 years or have HbA1c levels outside the range of 7.5% to 9.9% and should not be applied to individuals with type 2 diabetes who receive multiple daily injections of insulin. The informed consent process and the run-in phase had the potential to exclude individuals who might be less adherent with CGM than the cohort that was studied; Funding source

Author, year, reference Polonsky et al. 2017.

Polonsky WH, Hessler D, Ruedy KJ, Beck RW, for the DIAMOND Study Group. The Impact of Continuous Glucose Monitoring on Markers of Quality of Life in Adults With Type 1 Diabetes: Further Findings From the DIAMOND Randomized Clinical Trial. Diabetes Care 2017 Jun; 40(6): 736-741. https://doi.org/10.2337/dc17-0133

Study title/objectives The Impact of Continuous Glucose Monitoring on Markers of Quality of Life in Adults With Type 1 Diabetes: Further Findings From the DIAMOND Randomized Clinical Trial.

Objective: To investigate the impact of 24 weeks of CGM use on QOL in adults with type 1 diabetes (T1D) who use multiple daily insulin injections.

In this study, QOLmeasures from the DIAMOND study, both diabetes specific and non–diabetes specific,

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 209 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

administered to participants at baseline and at 24 weeks were examined.

Study characteristics

Study design DIAMOND trial: prospective, randomized, parallel, two group, controlled study

Study Registration number Clinical trial reg. no. NCT2282397, clinicaltrials.gov.

Country of recruitment U.S.

Centre (single or multicentre) Multicentre (24 endocrinology practices in the U.S.)

Ethics Committee Approval -

Sponsor Dexcom, Inc., provided funding for the study to the Behavioral Diabetes Institute and to the Jaeb Center for Health Research.

Study period (study start, study end) 24 weeks

Duration of follow-up (days) Follow-up visits (both treatment groups): after 4, 12, and 24 weeks.

CGM group : additional visit 1 week after randomization.

Control group: two additional visits 1 week before the 12- and 24-week visits (to

initiate blinded CGM use for 1week).

Phone contacts for both groups occurred 2 and 3 weeks after randomization.

Inclusion criteria • Age ≥25 years • Diagnosis of T1D • MDI treatment for at least 1 year • Central laboratory– measured HbA1c 7.5–10.0% • No CGM use in the 3months pretrial

Exclusion criteria - (See DIAMOND Study)

Patient characteristics CGM group (n=102) Control group (n=53)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 210 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Age of patients (years) (mean ± SD) 46 ± 14 51 ± 11

Sex (n (%)) Female: 46 (45%) Female: 23 (43%)

BMI - -

Diagnosis (Type I, II or gestational) Type I diabetes Type I diabetes

Comorbidities (i.e. obesity…) - -

Time since diagnosis with DM (years) (mean ± SD) 20± 13 24 ± 14

Insulin treatment (CSII or MDII) MDI MDI

Pregnancy (yes, no) - -

Special subgroup pf patient (i.e., hypoglycaemia fear…):

• Hypoglycaemia fear (worry subscale of HFS-II) (mean ± SD) 17.3 ± 13.2 15.8 ± 12.3

• Hypoglycemic confidence (HCS) (mean ± SD) 3.2 ± 0.6 3.3 ± 0.6

Intervention

Type of medical device (CGM or FGM) CGM + SMBG

® Name (Description) of medical device Dexcom G4 Platinum CGM System with an enhanced algorithm (Dexcom, Inc., San Diego, CA).

Bayer Contour Next USB meter and test strips.

Instructed to use the CGM daily, calibrate the CGM twice daily, and verify the CGM glucose concentration with their meter before making diabetes management decisions.

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devices) SMBG

Name (Description) of medical device Bayer Contour Next USB meter and test strips.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 211 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Participants in the control group were asked to perform home blood glucose monitoring with the study meter at least four times daily.

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible (connected) with - specific CGM systems (Yes, No)

Outcomes

Primary

• Non–diabetes-specific QOL WHO-5

EQ-5D-5L

• Diabetes-specific QOL self-report measures Diabetes distress Scale (DDS)

• Total • Regimen • Emotional burden • Interpersonal • Physician

Hypoglycemic confidence scale (HCS)

Hypoglycaemia fear survey (worry subscale of HFS-II)

• Satisfaction with CGM * Assessed at 24 weeks with the CGM Satisfaction Survey; only for CGM group

Secondary

Flow of patients

No of patients enrolled 186

No of randomized 158

Allocated per arms CGM group: 105 Control group: 53

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 212 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Received int. per arms CGM group:102 Control group:53

Lost to follow-up per arms CGM group:3 (1 lost to follow up, 1 site withdrew Control group:0 participant, 1 participant requested to withdraw)

No of analysed per arm CGM group:105 Control group: 53

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) Intention-to-treat

Results

Effectiveness results n (%) 95% CI

Quality of life (table below)

Quality of life CGM group Control group Model 1 (resulted from mixed linear Model 2 (values are further adjusted for the participant regression models adjusted for baseline demographic factors of age, sex, and number of years levels of the outcome and clinical site as since diagnosis) a random effect)

Baseline 24 weeks Baseline 24 weeks Mean differ- 95% CI P value Mean differ- 95% CI P value d ence in ence in

change change between between arms arms • WHO-5 (mean±SD) 71.28 ± 70.47 ± 69.06 ± 67.32 ± -1.26 -5.42 to 0.62 -1.63 -5.88 to 0.50 0.12 14.71 16.68 14.89 16.86 2.91 2.61 • EQ-5D-5L (mean±SD) 0.90 ± 0.11 0.89 ±0.10 0.89 ± 0.11 0.88 ±0.10 0.00 -0.03 to 0.86 0.00 -0.03 to 0.92 0.00

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 213 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

0.03 0.03 • Diabetes distress

(DDS) (mean±SD)

• Total 1.78 ± 0.65 1.61 ±0.48 1.69 ± 0.62 1.78 ± 0.65 0.22 0.08 to 0.009 0.23 0.09 to 0.03 0.44 0.36 0.36

• Regimen 2.09 ± 0.87 1.81 ± 0.68 2.08 ±0.99 2.05 ±0.87 0.25 0.05 to 0.04 0.26 0.05 to 0.04 0.31 0.46 0.47

• Emotional burden 2.06 ± 0.90 1.93 ± 0.80 1.91 ± 0.83 2.03 ±0.95 0.21 0.01 to 0.08 0.21 0.00 to 0.09 0.33 0.41 0.41

• Interpersonal 1.54 ± 0.81 1.43 ± 0.61 1.45 ± 0.70 1.73 ± 1.04 0.37 0.16 to 0.009 0.37 0.16 to 0.01 0.51 0.56 0.58

• Physician 1.19 ± 0.63 1.09 ± 0.25 1.12 ± 0.25 1.18 ± 0.69 0.10 -0.04 to 0.21 0.12 -0.03 to 0.15 0.18 0.25 0.27 • Hypoglycemic confi- 3.27 ± 0.57 3.47 ± 0.55 3.15 ± 0.57 3.18 ± 0.63 0.23 0.06 to 0.03 0.23 0.05 to 0.03 0.40 dence (HCS) (mean±SD) 0.40 0.41 • Hypoglycaemia fear 15.75 ± 13.48 ± 17.30 ± 17.73 ± 3.17 0.19 to 0.07 2.46 -0.58 to 0.15 0.25 (worry subscale of HFS-II) (mean±SD) 12.30 10.63 13.22 14.92 6.14 5.51

CGM benefits (CGM group only) See table below

CGM hasseles(CGM group only) See table below

CGM total satisfaction (CGM group only) See table below

CGM benefits CGM hasseles CGM total satisfaction

B (SE) 95% CI P value B (SE) 95% CI P value B (SE) 95% CI P value

ΔWHO-5 4.55 (2.57) -0.56 to 9.66 0.09 -7.00 (2.39) -11.75 to -2.25 0.01 7.61 (2.80) 2.05 to 13.17 0.02

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 214 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ΔEQ-5D-5L 0.02 (0.02) -0.02 to 0.05 0.38 -0.04 (0.02) -0.08 to -0.01 0.03 0.04 (0.02) -0.01 to 0.08 0.08

ΔDiabetes distress (DDS)

• Total -0.18 (0.07) -0.32 to -0.04 0.02 0.31 (0.07) 0.17 to 0.44 < 0.001 -0.31 (0.08) -0.47 to -0.16 <0.001 • Regimen -0.26 (0.11) -0.48 to -0.03 0.04 0.46 (0.10) 0.25 to 0..66 <0.001 -0.46 (0.12) -0.70 to -0.22 <0.001 • Emotional bur- -0.23 (0.12) -0.48 to 0.02 0.08 0.47 (0.11) 0.23 to 0.70 < 0.001 -0.46 (0.13) -0.73 to -0.18 0.003 den • Interpersonal 0.03 (0.08) -0.14 to 0.19 0.74 0.07 (0.08) -0.10 to 0.22 0.43 -0.03 (0.09) -0.21 to 0.16 0.79 • Physician -0.11 (0.05) -0.21 to -0.01 0.04 0.11 (0.05) 0.02 to 0.21 0.03 -0.14 (0.05) -0.24 to -0.02 0.02

Δ Hypoglycemic confi- 0.30 (0.09) 0.11 to 0.48 0.003 -0.44 (0.09) -0.62 to -0.26 < 0.001 0.49 (0.10) 0.29 to 0.70 < 0.001 dence

(HCS)

Δ Hypoglycaemia fear -2.29 (1.56) -5.38 to 0.81 0.16 4.13 (1.50) 1.15 to 7.11 0.02 -4.22 (1.73) -7.66 to -0.78 0.03

(worry subscale of HFS- II)

Δ HbA1c -0.15 (0.15) -0.45 to 0.15 0.33 0.21 (0.15) -0.09 to 0.49 0.17 -0.23 (0.17) -0.57 to 0.10 0.18

Δ % Time within range 0.04 (0.03) -0.02 to 0.09 0.21 0.01 (0.03) -0.05 to 0.06 0.98 0.02 (0.03) -0.04 to 0.09 0.50

Δ % Time < 70 mg/dL 0.01 (0.01) -0.01 to 0.02 0.48 0.00 (0.01) - 0.01 to 0.02 0.92 0.00 (0.01) -0.01 to 0.02 0.73

Δ % Time .180 mg/dL -0.04 (0.03) -0.10 to 0.02 0.21 0.01 (0.03) -0.06 to 0.06 0.88 -0.03 (0.04) -0.11 to 0.05 0.43

Mortality -

Change in HbA1c -

Incidence of hypoglycaemia - level 1 hypoglycaemia -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 215 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

level 2 hypoglycaemia - level 3 hypoglycaemia -

Incidence of hyperglycaemia -

Time spent in range -

Time spent in hypoglycaemia -

Time spent in hyperglycaemia -

Quality of life -

Patient satisfaction -

Hypoglycaemia fear -

Incidence of diabetic ketoacidosis -

Incidence of hyperosmolar, hyperglycaemic coma -

Resource utilization related to DM -

Number of visits to emergency room -

Number of visits to primary care -

Number of visits to specialists -

Number of hospitalizations -

Number of daily finger-sticks tests -

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Compliance/adherence -

Percentage of time using CGM -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 216 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of sensor scans per day (in FGM system) -

Safety results - n (%) 95% CI

Any AEs -

Serious AE (SAE) -

Most frequent AEs (by arms) -

Most frequent SEAs (by arms) -

Death as SAE -

Withdrawals due AEs -

Costs (only for national assessment) -

Author Disclosure (Conflict of interest) -

Polonsky WH W.H.P. reports consulting fees from Dexcom and Abbott Diabetes Care.

Beck RW R.W.B. reports that his institution has received supplies for research studies from Dexcomand Abbott Diabe- tes Care.

RR: relative risk ITT: intention to treat PP: per protocol

Risk of Bias

Study (Author, year): Polonsky et al. From DIAMOND Trial, QoL

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Each participant was assigned randomly from a computer-generated sequence to either the CGM or con-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 217 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

trol group in a 2:1 ratio, with a permuted block design (block sizes of 3 and 6) stratified by HbA1c level (<8.5% and ≥8.5%). A 2:1 randomization was used rather than 1:1 to provide a larger sample size for a separate follow-on randomized trial assessing glycemic benefits of initiating pump therapy in CGM users using insulin injections.

Allocation concealment (Selection bias) Unclear Not described

Blinding of participants and personnel (Performance bias) High Open label

Blinding of outcome assessment (Detection bias) High Open label

Blinding of outcome assessment (Detection bias) Unclear Not described

Incomplete outcome data (Attrition bias) Low Intent-to-treat principle; The 24-week primary study outcome visit was completed by 102 participants (97%) in the CGM group and all 53 (100%) in the control group. Overall visit completion was 99% and 98%, respectively. Initially, analyses were conducted without imputation for missing values (i.e., used the missing-at-random approach) and then repeated by using multiple imputation to supply values for missing data. Because missing data were minimal and no meaningful differences existed between the results of the analyses with or without data imputation, only the results of the initial analysis that used the missing-at- random (nonimputed) data are presented.

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear In light of the eligibility criteria, the results may not apply to individuals with type 1 diabetes who are younger than 26 years or have HbA1c levels outside the range of 7.5% to 9.9% and should not be applied to individuals with type 2 diabetes who receive multiple daily injections of insulin. The informed consent process and the run-in phase had the potential to exclude individuals who might be less adherent with CGM than the cohort that was studied; Funding source; study participants were racially homogeneous, with the majority of participants being non-Hispanic white with a high education level (with more than one-half of the sample reporting college degrees; potential clinical significance is unknown

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 218 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, year, reference Lind et al. 2017

Lind M, Polonsky W, Hirsch I B, Heise T, Bolinder J, Dahlqvist S et al. Continuous Glucose Monitoring vs Conven- tional Therapy for Glycemic Control in Adults With Type 1 Diabetes Treated With Multiple Daily Insulin Injections : The GOLD Randomized Clinical Trial. JAMA. 2017;317(4):379-387

Study title/objectives Continuous Glucose Monitoring vs Conventional Therapy for Glycemic Control in Adults With Type 1 Diabetes Treated With Multiple Daily Insulin Injections : The GOLD Randomized Clinical Trial

Objective: To evaluate the effects of continuous glucose monitoring in adults with type 1 diabetes treated with multiple daily insulin injections.

Separate article: Design and methods Lind M, Polonsky W, Hirsch IB, Heise T, Bolinder J, Dahlqvist S, Pehrsson NG, Moström P. Design and Methods of a Randomized Trial of Continuous Glucose Monitoring in Persons With Type 1 Diabetes With Impaired Glycemic Control Treated With Multiple Daily Insulin Injections (GOLD Study). J Diabetes Sci Technol. 2016 May 3;10(3):754-61. doi: 10.1177/1932296816642578.

Study characteristics

Study design Randomized, open-label, cross-over clinical trial

Study Registration number NCT02092051

Country of recruitment Sweden

Centre (single or multicentre) Multicentre

Ethics Committee Approval Approved by Ethics committee at the University of Gothenburg, Gothenburg, Sweden

Sponsor The trial was investigator initiated and sponsored by the NU Hospital Group, Trollhättan and Uddevalla, Sweden.

Study period (study start, study end) February 24, 2014 - June, 1, 2016

Duration of follow-up (days) 69 weeks

Treatment phase 1 : 26 weeks (Visits weeks 0,2 ,4, 13, 26) - Wash out period 17 weeks (Visit week 40-43 for masked CGM) Treatment phase 2: 26 weeks (Visit weeks 43, 45, 47, 56, 69)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 219 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Follow up time: 182 days

Inclusion criteria Type 1 diabetes

Adults 18 years or older

Written Informed Consent

HbA1c greater than or equal to 58 mmol/mol (7.5% DCCT standard)

Fasting C-peptide level of less than 0.91 ng/mL (to convert to nmol/L, multiply by 0.331)

Diabetes duration of greater than 1 year

Exclusion criteria 1. Pregnancy, planned pregnancy for the study duration or pregnancy during the last six months

2. Severe cognitive dysfunction or other disease, which is judged by the physician to be not suitable for inclusion

3. Requred continuous use of paracetamol. Paracetamol must not have been used the week before the study and shall not be used during CGM-use because it disturbs the interpretation of blood glucose levels estimated by the Dexcom G4®. How- ever, other pain killers can be used throughout the study duration

4. Current CGM use (within the past 4 months)

5. History of allergic reaction to any of the CGMS materials or adhesives in contact with the skin.

6. History of allergic reaction to chlorhexidine or alcoholanti-septic solution.

7. Abnormal skin at the anticipated glucose sensor attachment sites (excessive hair, burn, inflammation, infection, rash, and/or tattoo)

8. Patient is uncomfortable by using the sensor during the blinded run-in period and believes it is unlikely that he/she will use the sensor more than 80% of the time during the trial

9. The patient has on average performed 12 or less calibrations per week during the run-in period.

10. Insulin pump therapy=Continuous subcutaneous insulin infusion (CSII)

11. Diabetes duration < 1 year

12. Participation in another study

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 220 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

13. Own insulin production (If this is not clear, C-peptide should be checked by a local blood sample)

14. Other investigator-determined criteria making patients unsuitable for participation

Patient characteristics

Age of patients, mean (SD), y Baseline In full analysis set population

43.7 years 44.6 (12.7)

CGM group Conventional therapy

Age, mean (SD), y 46.7 (13.0) 42.6 (12.2)

Sex - No. (%) Baseline In full analysis set population

45.3% woman 56.3 % men

CGM group Conventional therapy

No Men 37 (53.6) 43 (58.9) (%) Woman 32 (46.4) 30 (41.1)

BMI at randomization visit - mean, (SD)

CGM group Conventional therapy

BMI mean, (SD) 27 (4.1) 27.2 (4.8)

Diagnosis (Type I, II or gestational) Type I

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 221 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Comorbidities (i.e. obesity…), No, (%)

Medical history at inclusion CGM group Conventional therapy visit, No, (%)

Previous laser 14 (20.3) 14 (19.2)

photocoagulation

of the retina

Previous myocardial 3 (4.3) 0

infarction

Previous stroke 1 (1.4) 1 (1.4)

Previous bypass graft 1 (1.4) 0

Previous PCI 2 (2.9) 0

Previous amputation 0 1 (1.4)

Previous diabetic foot 1 (1.4) 5 (6.8)

(or leg) ulcer

Current diabetic foot 0 3 (4.1)

(or leg) ulcer

Time since diagnosis with DM, mean (SD), y In full analysis set population

Baseline CGM group Conventional 22.2 (11.8) therapy

Time since diagnosis 23.4 (11.9) 21 (11.7) with DM, mean (SD), y

Insulin treatment (CSII or MDII) MDII

Pregnancy (yes, no) No

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 222 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Exclusion criteria: Pregnancy, planned pregnancy for the study duration or pregnancy during the last six months

Special subgroup pf patient (i.e., hypoglycaemia fear…) NR

Intervention

Type of medical device (CGM or FGM) CGM

Name (Description) of medical device Dexcom G4® Platinum, Dexcom Inc, San Diego

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medi- Only SMBG cal devices)

Name (Description) of medical device NR

Sensor integrated (Yes, No) No

Sensor augmented (or enabled) insulin pump systems compatible No (connected) with specific CGM systems (Yes, No)

Outcomes

Primary Difference in HbA1c between CGM and conventional therapy at weeks 26 and 69

Secondary • The difference in mean glucose level (measured by CGM during 2 weeks) between weeks 23-26 and 66-69.

• The difference in MAGE (measured by CGM during 2 weeks) between weeks 23-26 and 66-69.24

• The difference in standard deviation of glucose levels measured by CGM during 2 weeks between weeks 23-26 and weeks 66-69, measured by CGM.

• The difference in DTSQ scores between weeks 26 and 69.

• DTSQc score at week 69

• The difference in WHO-5 scores between weeks 26 and 69.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 223 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• The difference in SWE-HFS scores between weeks 26 and 69.

• The difference in SWE-PAID-20 scores between weeks 26 and 69.

• The difference in the proportion of time with low glucose levels measured by CGM during 2 weeks between week 23- 26 and week 66-69 measured by CGM (below 54 mg/dl [3.0 mmol/l] and below 72 mg/dl [4.0 mmol/l] respectively).

• The difference in the proportion of time with high glucose levels measured by CGM during 2 weeks between week 23-26 and week 66-69 measured by CGM (above 180 mg/dl [10.0 mmol/l] and above 250 mg/dl [13.9 mmol/l] respectively).

• The difference in the proportion of time with euglycemic levels measured by CGM during 2 weeks between weeks 23-26 and weeks 66-69 (99-180 mg/dl [5.5-10.0 mmol/l] and 70-180 mg/dl [3.9-10.0 mmol/l] respectively).

• The difference in the proportion of patients reducing their HbA1c by 0.5% (5 mmol/mol) or more.

• The difference in the proportion of patients lowering their HbA1c 1% (10 mmol/mol) or more.

• The difference in the mean number of severe hypoglycemic events between weeks 1-26 and weeks 44-69 defined as unconsciousness due to hypoglycaemia or need of assistance from another person to resolve the hypoglycaemia.

• The difference in mean number of capillary glucose measurements per day between weeks 1-26 and weeks 43-69, from time periods when values are available in glucometers.

Flow of patients

No of patients enrolled 205

No of randomized 161

Allocated per arms

CGM group Conventional therapy

No of patients 82 79

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 224 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Received int. per arms,No

CGM group SMBG

Period 1 , No 82 79

Period 2, No 70 Crossed over and received con- 73 Crossed over and received CGM ventional therapy after 17-week after 17-week wash out wash out

During the 17-week washout period, patients used conventional therapy and masked CGM was performed for 2 weeks (to evaluate time in hypoglycaemia, time in euglycaemia, time in hyperglycaemia, and glycemic variability)

Lost to follow-up per arms, No

CGM group Conventional therapy

Excluded (no follow up data in 13 6 both periods), No

No of analysed per arm

CGM group Conventional therapy

Included in primary 69 73

analysis, No

142 Toatal included in Primary 142 Recieved CGM 142 Recieved Conventional therapy Analysis

Statistical analysis

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 225 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ITT, modified ITT, Per protocol; other (specify) Per protocol

The full analysis set consists of all randomized patients who received at least 1 follow-up measurement in each treatment phase.

The per-protocol population (PP-population) consists of all patients in the full analysis set without any significant protocol deviations. The PP-population is defined at the clean-file meeting before the database is locked.

The safety population consists of all randomized patients who received treatment (conventional or CGM) at any time. In the safety analysis patients will be assigned to treatment given, not the randomized treatment.

The primary efficacy analysis was the difference in HbA1c at weeks 26 and 69 between CGM and conventional therapy for the full analysis set using a general linear model adjusted for both period’s effect and subject effect.

If efficacy measurements from 26 and 69 weeks follow-up, respectively, are missing, the last observation carried forward (LOCF) principle will be applied.

Secondary efficacy analyses, the differences in the secondary endpoints between CGM and conventional therapy, will also be analyzed adjusted for period’s effect and subject effect on the full analysis set. The theory of sequential multiple test procedures will be applied for the primary analysis and for secondary analyses. If a test gives a significant result at the 5% significance level, the total test mass will be transferred to the following number in the test sequence until a nonsignificant result is achieved. The above efficacy analyses will also be performed on the PP-population. All significance tests will be performed 2-sided at significance level of 0.05.

The study was powered to detect a difference of 0.3% (3 mmol/mol) in HbA1c between weeks 26 and 69, at 90% power, assuming a standard deviation of 1.1%, which requires 144 participants. Assuming a drop-out rate of 10% 160 individuals were required to be enrolled. All analyses will be performed by using SAS® v9.4 (Cary, NC).

Results

Effectiveness results n (%) 95% CI

Mortality NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 226 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

HbA1c

CGM, Mean Conventional Therapy, Least Square Means or Mean P value (95% CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

HbA1c, % 7.92 (7.79 to 8.35 (8.19 to 8.51) −0.43 (−0.57 to −0.29) <0.001 8.05)

HbA1c, mmol/mol 63 (61.6 to 64.5) 68 (66.0 to 69.4) −4.7 (−6.27 to −3.13)

Number of patients 142 142

HbA1c Change

CGM (Dexcom G4 ®) Conventional

therapy

Percentage of Patients Reducing their HbA1c by 0.5% (5 105 (73.9%) 78 (56.5%) mmol/mol) (LOCF)

Percentage of Patients Reducing their HbA1c by 1% (10 46 (32.4%) 14 (10.1%) mmol/mol) (LOCF)

Patient s who used sensor Patient s who used sensor more than 70% time less than 70% time

HbA1c reduce 0.46 % No significant difference

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 227 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(0.31%-0.61%)

SD of glucose levels, mg/dL

CGM, Mean Conventional Therapy, Least Square Means or Mean P value (95% CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

SD of glucose levels, 68.49 (66.36 to 77.23 (74.96 to 79.5) −8.69(−10.76 to −6.61) <0.001 mg/dl 70.63)

No of patients 133 133

Mean glucose level, mg/dL

CGM, Mean (95% Conventional Therapy, Least Square Means or Mean P value CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

Mean glucose level, 186.93 (181.66 to 193.68 (188.31 to −6.61(−12.01 to −1.2) 0.02 mg/dL 192.2) 199.04)

No of patients 133 133

Data were measured by CGM during 2 weeks.

Mean amplitude glycemic exursions, mg/dL

CGM, Mean Conventional Therapy, Least Square Means or Mean P value (95% CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

Mean amplitude glycemic 161.93 (156.94 180.96 (175.72 to −19.36(−24.26 to −14.46) <0.001

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 228 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

exursions ,mg/dl to 166.91) 186.20)

No of patients 123 127

Data were measured by CGM during 2 weeks.

Incidence of hypoglycaemia

CGM Conventional Therapy During Washout period

(Convetional thera-

py)

Events of severe hypo- 1 (event rate, 5 (event rate, 0.19 per 7 (event rate, 0.41 glycaemia, No 0.04 per 1000 1000 patient days) per 1000 patient patient days) days)

level 1 hypoglycaemia NR level 2 hypoglycaemia NR level 3 hypoglycaemia NR

Time spent in range Mean (SD), Median (Range) and n CGM Conventional Therapy

Percentage of Time with Euglycaemic Levels 99-180 34.69 (11.76) 31.85 (9.85) mg/dl [5.5-10.0 mmol/l] 34.5 (7.9-60.0) 33.1 (8.2-53.0)

n=123 n=125

Percentage of Time with Euglycaemic Levels (Measured 42.28 (14.95) 39.81 (12.56) by CGM During 2 Weeks) 70-180 mg/dl [3.9-10.0 41.7 (8.0-73.8) 41.0 (8.4-65.2) mmol/l] n=123 n=125

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 229 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time spent in hypoglycaemia, Mean (SD) percentage Mean (SD), Median (Range) and n CGM, Mean Conventional Therapy, Mean (SD) (SD) percent- percentage age

Time spent in hypoglycaemia (<70 mg/dL) 2.79% (2.97%) 4.79% (4.03%)

Time spent in hypoglycaemia (<54 mg/dL) 0.79% (1.23%) 1.89% (2.12%)

Time spent in hyperglycaemia

Mean (SD), Median CGM Conventional Therapy (Range) and n

Time spent in hypergly- 18.48 (12.28) 21.92 (12.39) caemia above 250 mg/dl 16.8 (1.3-55.3) 18.7 (2.1-59.6) [13.9 mmol/l] n=123 n=125

Time spent in hyperglace- 44.90 (15.68) 46.98 (14.01) mia above 180 mg/dl [10.0 46.1 (9.0-81.5) 46.0 (19.4-81.0) mmol/l] n=123 n=125

Quality of life NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 230 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Patient satisfaction (Diabetes Treatment Satisfaction Question- naire (DTSQ) status version) CGM, Mean (95% Conventional Therapy, Least Square Means or Mean P value CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

DTSQ status ver- 30.21 (29.47 to 26.62 (25.61 to 27.64) 3.43(2.31 to 4.54) <0.001 sion, scale total 30.96)

No of patients 136 137 131

Patient satisfaction (Diabetes Treatment Satisfaction Question- naire (DTSQ) change version) CGM, Mean (95% Conventional Therapy, Least Square Means or Mean P value CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

DTSQ change 13.2 (12.13 to 5.97 (3.64 to 8.30) 3.76(1.70 to 5.82) <0.001 version, scale total 14.28)

No of patients 69 67 136

Data for the DTSQ change version is collected only at the end of period 2. For CGM therapy column it is showing the change in satisfaction from conventional therapy to CGM therapy and for conventional therapy column it is showing the change from CGM therapy to conventional therapy

WHO-5 Well being Index

CGM, Mean (95% Conventional Therapy, Least Square Means or Mean P value CI) Mean for Difference:

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 231 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(95% CI) CGM−Conventional

Treatment (95% CI)

WHO-5 Well being 66.13 (62.94 to 62.74 (60.18 to 65.31) 3.54(0.61 to 6.48) 0.02 Index 69.32)

No of patients 139 140

Hypoglycaemia Confidence Questionnaire (HCQ) scale total

CGM, Mean (95% Conventional Therapy, Least Square Means or Mean P value CI) Mean for Difference: CGM−Conventional (95% CI) Treatment (95% CI)

HCQ, scale total 3.4 (3.32 to 3.47) 3.27 (3.18 to 3.35) 0.12(0.05 to 0.19) <0.001

No of patients 137 137 135

Hypoglycaemia fear Scale Behavior/Avoidance

CGM, Mean Conventional Least Square Means or Mean P value (95% CI) Therapy, Mean for Difference:

(95% CI) CGM−Conventional Treatment (95% CI)

Hypoglycemic Fear Scale Be- 1.93 (1.83 to 1.91 (1.81 to 0.03 (−0.05 to 0.10) 0.45 havior/Avoidance 2.03) 2.00)

N of patients 140 140

Incidence of diabetic ketoacidosis Ketoacidosis was not reported during the study

Incidence of hyperosmolar, hyperglycaemic coma NR

Resource utilization related to DM NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 232 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of visits to emergency room NR

Number of visits to primary care NR

Number of visits to specialists NR

Number of hospitalizations

CGM (Dexcom G4®) Conventional

(n=156) therapy

(n=151)

Events Patients Events Patients

with with

Events Events

n (%) n (%)

Hospitalisation 5 1 (0.7%)

Number of daily finger-sticks tests

CGM group Conventional therapy

Self measuremnts of blood 2.75 (1.39) 3.66 (2.30)

glucose , mean (SD)

Number of calibration NR

Need (Yes, with number or No) of re-calibration NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 233 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Compliance/adherence

Variable CGM (Dexcom G4®)

(n=161)

Patients with CGM Data available in Study

No 13 (8.1%)

Yes 148 (91.9%)

Total Adherence in Study (while Using CGM 87.8 (13.4)

92.8 (16.7; 98.2)

n=148

Adherence before Week 2 (CGM used in Period 1) 91.9 (11.7)

or Week 43 (CGM used in Period 2) 95.1 (0.2; 99.8)

n=144

Adherence before Week 4 (CGM used in Period 1) 91.8 (13.0)

or Week 47 (CGM used in Period 2) 95.9 (1.5; 98.9)

n=144

Adherence before Week 13 (CGM used in Period 1) 88.9 (14.5)

or Week 56 (CGM used in Period 2) 94.6 (19.7; 105.1)

n=142

Adherence before Week 26 (CGM used in Period 1) 86.5 (16.7)

or Week 69 (CGM used in Period 2) 92.6 (6.8; 98.6)

n=140

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 234 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

For categorical variables n (%) is presented.

For continuous variables Mean (SD) / Median (Min; Max) n= is presented.

Only patients using CGM for at least one day during last 30 days before a specific

visit are included in this summary.

Adherence to CGM sensor usage is defined as percent of total time on CGM in-between the visits and in total.

The total time in-between the visits was set to maximum of 30 days.

Percentage of time using CGM (range) 87.8% during CGM treatment periods ( 86.5%-91.9%)

Number of sensor scans per day (in FGM system) NR

Safety results n (%) 95% CI

Any AEs

CGM group Conventional therapy (n=156) (n=151)

No of patients 77 67

(49.4%) (44.4%)

No of events 137 122

Serious AE (SAE)

CGM group Conventional therapy

No of patients 7 (4.5%) 3 (2.0%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 235 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

No of events 9 9

Most frequent AEs (by arms)

CGM (Dexcom G4®) Conventional

(n=156) therapy

(n=151)

Events Patients Events Patients

with with

Events Events

n (%) n (%)

Anaemia 1 1 (0.7%)

Palpitations 1 1 (0.6%)

Vertigo 1 1 (0.6%)

Cataract 1 1 (0.6%)

Eye inflammation 1 1 (0.6%)

Macular degeneration 1 1 (0.7%)

Macular oedema 1 1 (0.6%)

Retinal detachment 1 1 (0.6%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 236 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Rethinopaty 1 1 (0.6%)

Abdominal pain upper 1 1 (0.6%) 1 1 (0.7%)

Diarrhoea 2 2 (1.3%)

Dyspepsia 1 1 (0.7%)

Gastritis 1 1 (0.6%)

Toothache 1 1 (0.6%)

Vomiting 2 2 (1.3%)

Application site pruritus 1 1 (0.6%) 1 1 (0.7%)

Application site rash 2 1 (0.6%)

Chest pain 4 2 (1.3%)

Device issue 1 1 (0.6%)

Fatigue 1 1 (0.6%) 2 2 (1.3%)

Hernia 1 1 (0.7%)

Inflammation 1 1 (0.7%)

Oedema peripheral 1 1 (0.7%)

Pyrexia 2 2 (1.3%) 2 2 (1.3%)

Thirst 1 1 (0.6%)

Allergy to arthropod bite 1 1 (0.6%)

Seasonal allergy 1 1 (0.6%)

Bronchitis 1 1 (0.6%)

Clostridial infection 1 1 (0.6%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 237 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Cystitis 1 1 (0.7%)

Ear infection 2 2 (1.3%)

Gastroenteritis 4 4(2.6%) 3 3 (2.0%)

Gastroenteritis viral 1 1 (0.7%)

Hand-foot-and-mouth 1 1 (0.7%) disease

Infection 2 2 (1.3%)

Influenza 3 3 (1.9%) 1 1 (0.7%)

Localised infection 1 1 (0.6%) 1 1 (0.7%)

Nasopharyngitis 34 29 (18.6%) 46 37 (24.5%)

Oral candidiasis 1 1 (0.6%)

Oral herpes 1 1 (0.6%)

Otitis externa 1 1 (0.6%) 1 1 (0.7%)

Paronychia 1 1 (0.7%)

Pneumonia 2 2 (1.3%) 1 1 (0.7%)

Post procedural infection 1 1 (0.6%)

Tonsillitis 2 2 (1.3%)

Upper respiratory tract 2 2 (1.3%) infection

Urinary tract infection 8 6 (3.8%) 3 3 (2.0%)

Vaginal infection 1 1 (0.7%)

Vulvovaginal candidiasis 1 1 (0.7%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 238 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Fall 1 1 (0.6%) 1 1 (0.7%)

Hand fracture 1 1 (0.7%)

Intercepted drug dispens- 1 1 (0.6%) ing error

Ligament sprain 1 1 (0.6%)

Limb injury 1 1 (0.6%)

Medication error 1 1 (0.7%)

Skeletal injury 1 1 (0.7%)

Hypoglycaemia 1 1 (0.6%) 5 5(3.3%)

Arthralgia 1 1 (0.6%) 2 2 (1.3%)

Back pain 1 1 (0.6%) 2 2 (1.3%)

Bursitis 1 1 (0.6%)

Exostosis 1 1 (0.6%)

Groin pain 1 1 (0.7%)

Intervertebral disc protru- 1 1 (0.7%) sion

Myalgia 1 1 (0.7%)

Myositis 1 1 (0.7%)

Neck pain 1 1 (0.7%)

Pain in extremity 1 1 (0.6%) 1 1 (0.7%)

Tendonitis 1 1 (0.7%)

Breast cancer 3 2 (1.3%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 239 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Prostate cancer 1 1 (0.7%)

Burning sensation 1 1 (0.7%)

Headache 1 1 (0.6%)

Sciatica 1 1 (0.7%)

Sensory disturbance 1 1 (0.6%)

Attention defi- 1 1 (0.6%) cit/hyperactivity disorder

Depression 2 2 (1.3%) 1 1 (0.7%)

Stress 1 1 (0.6%)

Benign prostatic hyperplas- 1 1 (0.7%) ia

Asthma 1 1 (0.6%)

Cough 1 1 (0.6%) 1 1 (0.7%)

Dyspnoea 1 1 (0.6%)

Nasal congestion 1 1 (0.6%)

Oropharyngeal pain 1 1 (0.6%) 2 2 (1.3%)

Pleurisy 1 1 (0.6%)

Dermatitis contact 1 1 (0.6%) 1 1 (0.7%)

Drug eruption 1 1 (0.6%)

Eczema 1 1 (0.6%)

Hidradenitis 1 1 (0.6%)

Pruritus 1 1 (0.6%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 240 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Rash 1 1 (0.7%)

Skin reaction 1 1 (0.6%)

Skin ulcer 8 3 (1.9%) 2 1 (0.7%)

Abstains from alcohol 1 1 (0.7%)

Failed examinations 1 1 (0.6%) 1 1 (0.7%)

Eye laser surgery 4 2 (1.3%)

Hospitalisation 5 1 (0.7%)

Mass excision 1 1 (0.6%)

Spinal decompression 1 1 (0.7%)

Deep vein thrombosis 1 1 (0.6%)

Hypertension 1 1 (0.7%)

Hypotension 1 1 (0.6%)

Thrombophlebitis 1 1 (0.6%)

Most frequent SEAs (by arms)

® CGM (Dexcom G4 ) Conventional

(n=156) therapy

(n=151)

Events Patients Events Patients

with with

Events Events

n (%) n (%)

Retinal detachment 1 1(0.6%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 241 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Diarrhoea 1 1(0.7%)

Chest pain 3 2 (1.3%)

Pneumonia 1 1(0.6%)

Post procedural infection 1 1(0.6%)

Breast cancer 3 2 (1.3%)

Prostate cancer 1 1(0.7%)

Depression 1 1(0.7%)

Abstains from alcohol 1 1 (0.7%)

Hospitalisation 5 1 (0.7%)

Death as SAE 1 died of prostate cancer

Withdrawals due AEs One patient in the CGM discontinued because of an allergic reaction to the sensor

Costs (only for national assessment) NR

Author Disclosure (Conflict of interest) All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

Dr Lind reports receipt of grants from AstraZeneca, Dexcom, and Novo Nordisk; consulting and receipt of honoraria from Novo Nordisk and Rubin Medical; and lecturing for Eli Lilly, AstraZeneca, Novo Nordisk, Medtronic, and Rubin Medical.

Dr Polonsky reports consulting for Dexcom and Abbott Diabetes Care.

Dr Hirsch reports consulting for Abbott Diabetes Care, Roche, and Intarcia.

Dr Heise reports receipt of grants from Adocia, Becton Dickinson, AstraZeneca, Biocon, Boehringer Ingelheim, Dance Pharmaceuticals, Eli Lilly, Grünenthal, Gulf Pharmaceuticals, Johnson & Johnson, Marvel, Medimmune, Medtronic, Mylan, Novartis, Novo Nordisk, Roche Diagnostics, Sanofi, Senseonics, and Zealand Pharma. He also reports receipt of personal fees from Eli Lilly, Mylan, and Novo Nordisk.

Dr Bolinder reports serving on advisory boards for Abbott Diabetes Care, Insulet, Integrity Applications, Novo Nordisk, and

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 242 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Sanofi; lecturing for Abbott Diabetes Care, AstraZeneca, Novo Nordisk, and Sanofi.

Dr Hellman reports served on advisory boards for Sanofi, Eli Lilly, Merck, Jensen Cilag, Novo Nordisk, AstraZeneca, Dex- com, and Abbott; lecturing for Sanofi, Boehringer Ingelheim, Eli Lilly, Merck, Novo Nordisk and AstraZeneca.

No other disclosures were reported.

RR: relative risk ITT: intention to treat PP: per protocol NR: Not reported

Risk of Bias

Study (Author, year): Lind et al 2017 GOLD Trial

Judgement (Low, Unclear, High) Support for judgement

® Random sequence generation (Selection bias) Low Patients were randomized 1:1 into the first treatment period to CGM using the Dexcom G 4 PLATINUM stand-alone system or conventional therapy. Randomization was performed by a centralized web-based program stratifying patients by site according to a predefined sequence; random block size varied between 1 + 1 and 2 + 2. The order of arm assignment position within each one block was random. The study used an externally managed web- based randomisation tool. Arm assignment was revealed one subject at a time upon completing randomisation of each subject. There were no means to preview the block schema or the next position within a block.

Allocation concealment (Selection bias) Low See above

Blinding of participants (Performance bias) High unmasked

Blinding of personnel (Performance bias) High unmasked

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) Unclear The full analysis set consisted of all randomized patients who had at least 1 follow-up measurement in each treatment period. The safety analysis consisted of all randomized patients who received treatment (CGM or conventional therapy) at any time with patients assigned to

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 243 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

treatment administered but not randomized treatment. The last observation carried forward principle was applied for any missing efficacy measurements from the last weeks of each treatment period. For the primary efficacy outcome HbA1c, full analysis set population, the LOCF imputation was done for 2 (2.9%) patients at the end of CGM therapy and 3 (4.1%) at the end of conventional therapy. Some secondary endpoints were not tested, and descriptive data for these variables are shown. There were19 patients (11.8%) excluded from the full analysis set population for lack of follow-up data in the second treatment period. Generally, in a parallel-group study, this can lead to an imbalance between groups. In the current study, patients served as their own controls and thus no such problem existed. It has therefore been proposed that the full analysis set population should be used in crossover studies as the main analysis.

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear Funding and COI

Author, year, reference Beck et al. 2017

Beck R, Riddlesworth T D, Ruedy K, Ahmann A, Haller S, Kruger D et al. Ann Intern Med. 2017; 167:365-374. doi:10.7326/M16-2855

Study title/objectives Continuous Glucose Monitoring Versus Usual Care in Patients With Type 2 Diabetes Receiving Multiple Daily Insulin Injections A Randomized Trial

Objective:To determine the effectiveness of CGM in adults with type 2 diabetes receiving multiple daily injections of insulin

Study characteristics

Study design Randomized, open label, parallel group clinical trial

Study Registration number NCT02282397

Country of recruitment North America (United States and Canada)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 244 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Centre (single or multicentre) Multicentre (25 endocrinology practice)

Ethics Committee Approval The protocol and Health Insurance Portability and Accountability Act– compliant informed consent forms were approved by 1 central and multiple local institutional review boards

Sponsor Dexcom

Study period (study start, study end) October 2014 – March 2016

Duration of follow-up 6 months (24 weeks)

Inclusion criteria Age ≥25 y

Diagnosis of type 2 diabetes

Followed regularly by a physician or diabetes educator for diabetes management, with ≥2 office visits in last year as documented by clinical history

Use of multiple daily injections of insulin for ≥12 mo before study entry

Suboptimal glycemic control, defined as persistent hyperglycaemia, confirmed initially by historical or local laboratory (POC or site's laboratory) HbA1c level of ≥7.7% to ≤10%, then followed with a confirmatory result by central laboratory of ≥7.5% to ≤10%

Desire to lower HbA1c level, such as a goal of 7%

Stable control of diabetes, as determined by investigator assessment

Stable diabetes medication regimen for 3 mo before study entry

Stable weight maintained for 3 mo before study entry, per investigator's assessment, and not planning any structured weight reduction interventions, such as prescription weight loss medications, bariatric surgery, or protein-sparing modified fast during the study

Willingness to use a CGM device

Willingness to avoid use of acetaminophen medications throughout the study

Currently performing self-monitoring blood glucose testing (by history) an average of ≥2 times per day

Ability to speak, read, and write English

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 245 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Exclusion criteria Use of personal real-time CGM ≤3 mo before study entry (professional CGM use, blinded or unblinded, is acceptable)

Use of CSII ≤3 mo before study entry (including patch pumps)

Plan to use personal CGM device and/or pump during study

Addition of any new oral or injectable hypoglycemic agents (including GLP-1 analogues, pramlintide, and SGLT-2 inhibitors) <3 mo before study entry.

(Use of these agents does not affect eligibility if used ≥3 mo before study entry.) For GLP-1 medications, must be on stable dose and the GLP-1 medication will be maintained throughout the study. Note: These agents should not be added or modified during course of the study. If use of this class medication is planned, the patient is not eligible

Use of premixed insulin (e.g., 70/30 or 50/50) ≤6 mo before study entry

Current or anticipated short-term uses of glucocorticoids (oral, injectable, or intravenous) that will affect glycemic control and HbA1c levels, such as frequent steroid bursts required for inflammatory arthritis or inflammatory bowel disease, recurrent lumbar epidural steroid injections, etc. (Long-term stable glucocorticoid doses are allowed, such as when used for the treatment of rheumatoid arthritis or Addison disease.)

Pregnancy (as demonstrated by a positive test result) at time of screening or plan to become pregnant during study

Medical conditions that, per investigator determination, make it inappropriate or unsafe to target an HbA1c level of <7%; conditions may include but are

not limited to unstable recent cardiovascular disease, recent myocardial infarction, significant heart failure, ventricular rhythm disturbances, recent transient ischemic attack or cerebrovascular accident, significant malignancy, other conditions resulting in physical or cognitive decline, or recurrent severe hypoglycaemia

History of visual impairment that would hinder patient's ability to participate in the study and perform all study procedures safely, as determined by investigatorHistory of psychiatric, psychological, or psychosocial issues that could limit adherence to required study tasks

Renal disease, defined as estimated glomerular filtration rate <45 mL/min/1.73 m2

Extensive skin changes/disease that preclude wearing the sensor on normal skin (e.g., extensive psoriasis, recent burns or severe sunburn, extensive eczema, extensive scarring, extensive tattoos, or dermatitis herpetiformis)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 246 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Known allergy to medical-grade adhesives

Current participation in another investigational study (must have completed any previous studies ≥30 d before being enrolled in this study)

Hospitalization or emergency department visit ≤6 mo before screening resulting in a primary diagnosis of uncontrolled diabetes

Currently abusing illicit drugs, alcohol, or prescription drugs

Any condition, per investigator assessment, that could impact reliability of the HbA1c measurement, such as (but not limited to) hemoglobinopathy, hemolytic anemia, chronic liver disease, chronic gastrointestinal blood loss, red blood cell transfusion, or erythropoietin administration < 3 months before screening

Patient characteristics

Age of patients, Mean, (SD), [range] y 60 (10) [35-79]

Baseline CGM group Control

Age, Mean, (SD), y 60 (11) 60 (9)

Sex Baseline CGM Control

group

No, (%) Female 49 (62) 40 (51)

BMI CGM group Control

BMI mean, (SD), 35 (8) 37 (7) 2 kg/m

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 247 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Diagnosis (Type I, II or gestational) Type II

Comorbidities (i.e. obesity…) Preexisting condi- CGM group Control

tions

Hypertension 61 (77) 65 (82)

Hyperlipidemia 54 (68) 50 (63)

Myocardial infrac- 4 (5) 3 (4) tion

Stroke 1 (1) 1 (1)

Time since diagnosis with DM, Median diabetes duration 17 y (interquartile range 11 to 23 years)

Baseline CGM group Control

Median diabetes duration 17 (11-23) 18 (12-23)

(Interquartile range), y

Insulin treatment (CSII or MDII) MDII

Pregnancy (yes, no) No

Exclusion criteria - Pregnancy (as demonstrated by a positive test result) at time of screening or plan to become pregnant during study

Special subgroup pf patient (i.e., hypoglycaemia fear…) NR

Intervention

Type of medical device (CGM or FGM) CGM

Name (Description) of medical device Dexcom G4 Platinum

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 248 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device Contour Next USB meter (Ascensia Diabetes care)

Sensor integrated (Yes, No) NR

Sensor augmented (or enabled) insulin pump systems compatible (con- NR nected) with specific CGM systems (Yes, No)

Outcomes

Primary Change in HbA1c level from baseline to 24 weeks

Secondary Proportions of participants with HbA1c levels below 7.0%, HbA1c levels below 7.5%,

Relative reduction of at least 10%, Reduction of at least 1%, Reduction of at least 1% or HbA1c level below 7.0%,

Length of time per day the glucose concentration was hypoglycemic (<3.89, <3.33, and <2.78 mmol/L [<70, <60, and <50 mg/dL]), hyperglycemic (>9.99, >13.88, and >16.65 mmol/L [>180, >250, and >300 mg/dL]), and in the target range (3.89 to 9.99 mmol/L [70 to 180 mg/dL]).

Glucose variability

Scores on the Clarke Hypoglycaemia Unawareness Survey ,

2 general quality-of-life measures (5-level EuroQol-5D and 5-item World Health Organization Well-Being Index)

3 diabetes-specific quality-of-life measures (Hypoglycaemia Fear Survey, Diabetes Distress

Scale, and Hypoglycemic Confidence Scale)

The CGM group's satisfaction

Insulin use,

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 249 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Body weight

Frequency of blood glucose meter testing.

All device or study-related adverse events

Severe hypoglycaemia (defined as an event that required assistance from another person to administer carbohydrates or other resuscitative action),

Diabetic ketoacidosis or severe hyperglycaemia if treatment was received at a health care

Facility

Serious adverse events regardless of causality.

Subgroup analysis Preplanned exploratory subgroup analyses were defined on the basis of baseline HbA1c level, age,CGM-measured hypoglycaemia, frequency of blood glucose meter testing, education level, diabetes numeracy, Hypoglycaemia Una- wareness score, and Hypoglycaemia Fear score

Flow of patients

No of patients enrolled 190

No of randomized 158

Allocated per arms

CGM group Control

No of patients 79 79

Received int. per arms

CGM group Control

No of patients 79 79

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 250 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Lost to follow-up per arms

CGM group Control

No of patients 1 1

No of analysed per arm

CGM group Control

No of patients 79 79

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) ITT

Results

Effectiveness results n (%) 95% CI

Mortality NR

Change in HbA1c

Change in 12 weeks 24 weeks HbA1c,

mean (95%CI),%

CGM Control Adjusted differ- CGM Control Adjusted ence, (95% CI); difference, N=77 N=75 N=79 N=79 P value (95% CI); P value

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 251 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Change in -1.1 -0.6 -0.3 -0.8 -0.5 -0.3 HbA1c (-1.2 to -0.8) (-0.8 to -0.4) (-0.6 to -0.1); (-1.0 to -0.7) (-0.7 to - (-0.5 to 0.0) ; from base- 0.005 0.3) 0.022 line

Other HbA1c outcomes

12 weeks 24 weeks

CGM Control Adjusted differ- CGM Control Adjusted ence, (95% CI); difference, N=77 N=75 N=79 N=79 P value (95% CI); P value

Mean HbA1c 7.5 (7.4 to 7.7) 7.9 (7.7 to 8.1) - level (95%

CI),%

HbA1c level 17 (22) 9 (12) 10%(-2% to 11 (14) 9 (12) 3%(-9% to <7% 23%);0.26 14%);0.88 N(%),N(%)

HbA1c level 35 (45) 22 (29) 17%(-3% to 27 (35) 21 (28) 8%(-11% to <7.5%, N(%) 37%);0.054 26%);0.63

Relative 44 (57) 26 (35) 25%(3% to 40 (52) 24 (32) 22%(0% to reduction in 46%);0.016 42%) HbA1c ;0.028 ≥10%, N(%)

Reduction in 40 (52) 25 (33) 20%(-1% to 30 (39) 21 (28) 12%(-7% to HbA1c level 41%);0.044 30%);0.21 ≥1%, N(%)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 252 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Reduction in 41 (53) 25 (33) 22%(0% to 33 (43) 22 (29) 15%(-5% to HbA1c level 43%);0.034 34%) ≥1% or ;0.146 HbA1c level <7%, N(%)

Reduction in 61 (79) 38 (51) 31%(5% to 56 (73) 37 (49) 26%(0% to HbA1c level 57%);0.002 50%) ≥ 0.5 ;0.007 %, N(%)

Incidence of hypoglycaemia NR level 1 hypoglycaemia NR level 2 hypoglycaemia NR level 3 hypoglycaemia NR

Incidence of hyperglycaemia NR

Time spent in range, median (interquartile ranges), min

CGM group Control group

median Baseline 12 wk 24 wk Baseline 12 wk 24 wk (interquar- N=79 N=77 N=74 N=78 N=74 N=72 tile ranges), min

Time per 802 937 882 794 822 836 day in (604-974) (664-1083) (647-1077) (665-976) (537-1025) (551-965 range of

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 253 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

70-180 mg/dL, min

Time spent in hypoglycaemia, median (interquartile ranges), min

CGM group Control group

Median Baseline 12 wk 24 wk Baseline 12 wk 24 wk (interquar- N=79 N=77 N=74 N=78 N=74 N=72 tile ranges), min

Time per 11 9 4 12 11 12 day < 70 (1-33) (1-25) (0-17) (3-39) (0-37) (0-34) mg/dL

Time per 3 1 0 4 1 2 day < 60 (0-15) (0-7) (0-6) (0-17) (0-12) (0-12) mg/dL

Time per 0 0 0 0 0 0 day < 50 (1-33) (0-1) (0-1) (0-7) (0-3) (0-5) mg/dL

Time spent in hyperglycaemia

CGM group Control group

median Baseline 12 wk 24 wk Baseline 12 wk 24 wk (interquar- N=79 N=77 N=74 N=78 N=74 N=72 tile ranges), min

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 254 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time per 612 501 549 607 560 571 day > 180 (411-809) (323-746) (353-789) (392-775) (382-818) (422-883) mg/dL

Time per 150 100 105 154 137 118 day > 250 (68-265) (37-180) (37-246) (66-281) (53-251) (48-288) mg/dL

Time per 33 19 23 42 33 18 day> 300 (9-77) (0-56) (0-66) (9-96) (1-95) (0-83) mg/dL

Glucose variability:coefficient of variation, %

CGM group Control group

Baseline 12 wk 24 wk Baseline 12 wk 24 wk

N=79 N=77 N=74 N=78 N=74 N=72

Glucose 31 30 30 32 (27-37) 30 29 variability: (27-38) (26-34) (26-33) (25-37) (25-36) coefficient of variation, %

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 255 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Mean glucose concentration, mg/dL

CGM group Control group

Baseline 12 wk 24 wk Baseline 12 wk 24 wk

N=79 N=77 N=74 N=78 N=74 N=72

Mean glu- 177 166 171 175 (155- 172 171 cose con- 191) (154-191) (149-187) (149-195) (155-199) (156-199) centration, mg/dL

Quality of life

Baseline 24wk

CGM Group Control Group CGM Group Control Group

N=79 N=79 N=77 N=73

EQ-5D-5L overall 0.82 (0.15) 0.82 (0.14) 0.82 (0.14) 0.82 (0.16) index

WHO-5 total score 16 (4) 17 (4) 16 (5) 17 (4)

Diabetes specific measures

Diabetes Distress 1.9 (0.8) 2 (0.8) 1.8 (0.9) 1.8 (0.6) Scale overall mean score

Emotional burden 2.3 (1.2) 2.3 (1.1) 2.2 (1.2) 2.1 (1.0)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 256 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

mean score

Clinician related 1.3 (0.6) 1.3 (0.8) 1.3 (0.9) 1.1 (0.3) distress mean score

Regimen related 2.2 (0.9) 2.4 (1.0) 2.0 (0.9) 2.1 (0.9) distress mean score

Diabetes related 1.8 (1.0) 2.0 (1.2) 1.7 (1.1) 1.7 (0.8) interpersonal distress mean score

Hypoglycaemia 0.8 (0.7) 0.8 (0.6) 0.8 (0.6) 0.7 (0.5) Fear Survey worry subscale, mean score

Hypoglycemic 3.2 (0.7) 3.4 (0.6) 3.3 (0.6) 3.4 (0.6) Confidence Scale mean score

Patient satisfaction

CGM group

Patient satisfaction with use of CGM mean (SD)

CGM Satisfaction Scale (score range 1 4.3 (0.4) to 5)

Benefit subscale 4.4 (0.5)

Hassles subscale 1.8 (0.5)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 257 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Hypoglycaemia fear NR

Hypoglycaemia Unawereness

Baseline 24wk

CGM Group Control Group CGM Group Control Group

N=79 N=79 N=77 N=73

Mean Clarke Hypo- 1.8 (1.4) 1.6 (1.3) 2.0 (1.5) 1.7 (1.4) glycaemia

Unawareness Total Score (SD)

Mean Change in - - 0.1 (1.5) 0.2 (1.7) Clarke Hypoglycae- mia

Unawareness

Total Score from baseline (SD)

Incidence of diabetic ketoacidosis Did not occur in either group

Incidence of hyperosmolar, hyperglycaemic coma NR

Resource utilization related to DM NR

Number of visits to emergency room NR

Number of visits to primary care NR

Number of visits to specialists NR

Number of hospitalizations CGM Group – 2 partcipants (chest pain, fully recovered), considered unrelated to CGM use

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 258 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of daily finger-sticks tests

Frequencies of BGSM, (SD) CGM Control

Baseline period of blinded CGM (2 4.1 (1.1) 4.0 (1.2) weeks)

At 24 weeks 2.9 (1.1) 3.8 (1.5) P<0.001

Number of calibration NR

Need (Yes, with number or No) of re-calibration NR

Compliance/adherence NR

Percentage of time using CGM

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 259 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Average CGM use per week Week 4 Visit Week 12 Visit Week 24 Visit Overall (N=78 ) (N=77) (N=76 ) (N=77 )

Median (interquartile 7.0 (7.0-7.0) 7.0 (7.0-7.0) 7.0 (7.0-7.0) 7.0 (6.8-7.0) range),d

Mean (SD), d 6.9 (0.4) 6.7 (1.0) 6.7 (1.0) 6.7 (0.9)

Zero use, n (%) 0 (0) 1 (1) 1 (1) 1 (1)

2-<3 days/week, n (%) 0 (0) 1 (1) 1 (1) 0 (0)

3-<4 days/week, n (%) 1 (1) 0 (0) 0 (0) 0 (0)

4-<5days/week, n (%) 0 (0) 2 (3) 1 (1) 1 (1)

5-<6 days/week, n (%) 1 (1) 1 (1) 2 (3) 5 (6)

6-<7 days/week, n (%) 5 (6) 5 (6) 7 (9) 16 (21)

7 days/week, n (%) 71 (91) 65 (84) 64 (84) 54 (70)

<6 days/week, n (%) 2 (3) 5 (6) 5 (7) 7 (9)

≥6 days/week, n (%) 76 (97) 72 (94) 71 (93) 70 (91)

% of Possible Readings, 97 97(92-99) 96 96 (92-98) median (interquartile (95-99) (91-98) range),%

% of Possible Readings , 95 (7) 92 (16) 91 (16) 92 (13) mean (SD)

Number of sensor scans per day (in FGM system) NR

Safety results n (%) 95% CI

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 260 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Any AEs NR

Serious AE (SAE)

CGM Control

Diabetic ketoacidosis 0 No serious adverse events Severe hypoglycaemia 0

Myocardial infraction 1

Chest pain 1

Most frequent AEs (by arms) NR

Most frequent SEAs (by arms) NR

Death as SAE 1 in CGM group – Myocardial infraction

Withdrawals due AEs NR

Costs (only for national assessment) Reported separetly

Author Disclosure (Conflict of interest) Dr. Beck reports grants from Dexcom during the conduct of the study and other support from Dexcom and Abbott Dia- betes Care outside the submitted work. Ms. Ruedy reports grants from Dexcom during the conduct of the study and other fees from Dexcom and Abbott Diabetes Care outside the submitted work. Dr. Ahmann reports grants and personal fees from Dexcom during the conduct of the study; grants from Novo Nordisk, Sanofi, Lexicon, and Medtronic outside the submitted work; and personal fees from Novo Nordisk, Sanofi, Eli Lilly, and Janssen outside the submitted work. Ms. Kruger reports grants from Henry Ford Health System during the conduct of the study; grants from Novo Nordisk, Eli Lilly, Dexcom, and Abbott Diabetes Care outside the submitted work; personal fees from Novo Nordisk, Janssen, Eli Lilly, Boehringer Ingelheim, Sanofi, Dexcom, Abbott Diabetes Care, Intarsia, and AstraZeneca outside the submitted work; and holding Dexcom stock. Dr. McGill reports personal fees from Boehringer Ingelheim, Dexcom, Dynavax, Janssen, Intarcia, Merck, Novo Nordisk, and Valeritas and grants from Dexcom, AstraZeneca/Bristol-Myers Squibb, Novartis, and Lexicon outside the submitted work. Dr. Polonsky reports personal fees from Dexcom during the conduct

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 261 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

of the study and from Dexcom and Abbott Diabetes Care outside the submitted work. Dr. Price reports being a Dexcom employee and a shareholder of Dexcom stock. Dr. Aronson reports research support from Merck, Boehringer Ingel- heim, Regeneron, Abbott Diabetes Care, Quintiles, ICON, GlaxoSmithKline, Medpace, Novo Nordisk, Janssen, Sanofi, Bristol-Myers Squibb, AstraZeneca, Becton Dickinson, Eli Lilly, Amgen, and Takeda and personal fees from Novo Nordisk, Janssen, Sanofi, AstraZeneca, Becton Dickinson, Eli Lilly, and Amgen outside the submitted work. Dr. Kollman reports grants from Dexcom during the conduct of the study. Dr. Bergenstal reports grants and other fees from Abbott Diabetes Care, Becton Dickinson, Boehringer Ingelheim, Bristol-Myers Squibb/AstraZeneca, Calibra Medical, Eli Lilly, Hygieia, Johnson & Johnson, Medtronic, Novo Nordisk, Roche, Sanofi, Takeda, and Dexcom during the conduct of the study. Dr. Bergenstal is employed by HealthPartners Institute/Park Nicollet Health Services and has contracts with the listed companies for his services as a research investigator or consultant; no personal income from any of these services goes to Dr. Bergenstal. Dr. Bergenstal also reports holding stock in Merck. Authors not named here have disclosed no conflicts of interest.

RR: relative risk ITT: intention to treat PP: per protocol NR: Not reported

Risk of Bias

Study (Author, year): Beck et al 2017 T2 DM Trial NCT02282397 The protocol also included a type 1 diabetes cohort in a parallel trial and subsequent second trial. The protocol, was identical to that of a parallel trial in patients with type 1 diabetes (DIAMOND Trial).

Judgement (Low, Unclear, Support for judgement High)

Random sequence generation (Selection bias) Low Randomly assigned by a computer-generated sequence to either the CGM or control group in a 1:1 ratio, using a permuted block design (random block sizes of 2 and 4) stratified by HbA1c level (<8.5% and ≥8.5%). The protocol, was identical to that of a parallel trial in patients with type 1 diabetes (DIAMOND Trial).

Allocation concealment (Selection bias) Unclear Not reported

Blinding of participants (Performance bias) High Unmasked

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 262 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Blinding of personnel (Performance bias) High Unmasked

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) low Intention-to-treat principle; The 24-week primary outcome visit was completed by 77 participants (97%) in the CGM group and 75 (95%) in the control group

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear Funding and CoI ; With broad eligibility criteria and participation by both community-based and academic sites, the trial results should be generalizable to most patients with type 2 diabetes who are 35 to 79 years of age, have HbA1c levels of 7.5% to 9.9%, and are being treated with multiple daily injections of insulin; a limitation of the trial is that follow-up was only 6 months.

Author, year, reference Heinemann L, Freckmann G, Ehrmann D, Faber-Heinemann G, Guerra S, Waldenmaier D, et al. Real-time continuous glucose monitoring in adults with type 1 diabetes and impaired hypoglycaemia awareness or severe hypoglycaemia treated with multiple daily insulin injections (HypoDE): a multicentre, randomised controlled trial. The Lancet. 2018.

Study title/objectives Real-Time Continuous Glucose Monitoring (RT-CGM) in Patients With Type 1 Diabetes at High Risk for Low Glucose Values Using Multiple Daily Injections (MDI) in Germany (HYPODE-STUDY)

Study characteristics

Study design multicentre, open-label, parallel, randomised controlled trial with a 6-month study period

eligible study participants were randomly assigned to one of two groups:

rtCGM system use (rtCGM group) or continued use of SMBG (control group).

Randomisation was done centrally at the study coordinating centre by staff who were not involved with recruitment or treatment of study participants. A randomisation sequence was generated with SYSTAT 12.0 with a 1:1 allocation; the study centre was a stratifying variable. Randomisation was done block-wise per site (four participants per block). Each study site received sealed envelopes with the respective group allocation.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 263 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

After successful completion of the baseline phase, the respective envelope was opened. Study site personnel informed participants about their group allocation.

Because of the nature of the intervention, masking of study participants and study personnel was not possible.

Study Registration number NCT02671968

Country of recruitment Germany

Centre (single or multicentre) 12 specialised diabetes practices in Germany.

All sites had experience of conducting clinical trials and of rtCGM usage.

Ethics Committee Approval The clinical study protocol was approved by the ethics committee of Landesaerztekammer Baden Wuerttemberg, Stuttgart, Germany, and the respective local ethics committees.

Sponsor Science Consulting in Diabetes

Dexcom, Inc. a Delaware corporation, USA

Institut für Diabetes-Technologie Forschungs- und Entwicklungsgesellschaft mbH an der Universität Ulm, Germany

Study period (study start, study end) March 4, 2016, and Jan 12, 2017

Duration of follow-up (days) Baseline phase 4 weeks, Therapy phase 22 weeks + 4 weeks follow up phase

Inclusion criteria Study participants were eligible for inclusion if:

they had type 1 diabetes for 1 year or more and

problematic hypoglycaemia, which was defined as having had at least one severe hypoglycaemia event requiring third- party assistance for recovery in the previous year, or having impaired hypoglycaemia awareness as defined by a total score of 4 or more in the hypoglycaemia unawareness questionnaire developed by Clarke and colleagues.

Additional inclusion criteria were treatment with MDI (prandial insulin at each major meal and at least one dose of basal insulin),

age 18 years or older,

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 264 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

and screening

HbA1c 75·0 mmol/mol or lower (≤9·0%).

Exclusion criteria Exclusion criteria were;

treatment with CSII therapy,

use of the rtCGM system or another rtCGM device in the previous 3 months,

and pregnancy.

Patient characteristics

Age of patients Age(years) 18 and older

Control group 47.3 (11.7)

rtCGM group 45.8 (12.0)

Sex Both

Women 25 (34%)

Men 35 (47%)

BMI Control group 26.0 (4.6)

rtCGM group 26.1 (6.7)

Diagnosis (Type I, II or gestational) Type I

Comorbidities (i.e. obesity…)

Time since diagnosis with DM Control group 21.6 (13.9)

rtCGM group 20.9 (14.0)

Insulin treatment (CSII or MDII) MDII

Pregnancy (yes, no) No

Special subgroup pf patient (i.e., hypoglycaemia fear…) The frequency of severe hypoglycaemia events was defined as the number of hypoglycaemic events requiring third- party assistance to administer carbohydrate, glucagon, or intravenous glucose injections during the therapy and follow-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 265 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

up phases.

Severe hypoglycaemia was further divided into two additional categories:

Events requiring medical assistance to inject glucagon or glucose or associated with hospital admission; and

events requiring third-party assistance without medical assistance.

Intervention

Type of medical device (CGM or FGM) rtCGM

® Name (Description) of medical device Dexcom G5 Mobile system; Dexcom, San Diego, CA,USA

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device

Sensor integrated (Yes, No)

Sensor augmented (or enabled) insulin pump systems compatible (connected) with specific CGM systems (Yes, No)

Outcomes

Primary Number of hypoglycaemic events measured by rtCGM during the follow-up phase compared with baseline.

A hypoglycaemic event derived from rtCGM was defined as glucose values of 3·0 mmol/L (≤54 mg/dL) or lower for at least 20 min, preceded by a minimum of 30 min with glucose values greater than 3·0 mmol/L (>54 mg/dL).

Secondary Secondary outcomes were:

changes in nocturnal hypoglycaemic events (0000 h to 0600 h);

percentage and duration of glucose readings derived from continuous glucose monitoring per day

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 266 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

in different glucose ranges (≤3·0 mmol/L [≤54 mg/dL], ≤3·9 mmol/L [≤70 mg/dL],

>3·9 mmol/L to ≤10·0 mmol/L [>70 mg/dL to 180 mg/dL], and >10·0 mmol/L [>180 mg/dL]), and

percentage of blood glucose readings based on SMBG measurements in these different glucose ranges.

Impaired hypoglycaemia awareness assessed with the hypoglycaemia unawareness questionnaire;

diabetes distress assessed with the Diabetes Distress Scale for type 1diabetes (T1-DDS);

fear of hypoglycaemia assessed with the Hypoglycaemia Fear Survey;

self-reported health status assessed with the European Quality of Life 5 Dimensions questionnaire (EQ-5D);

and satisfaction with glucose measurement assessed with the Glucose Monitoring Satisfaction Survey

Flow of patients

No of patients enrolled 170

No of randomized 149

Allocated per arms 75 assigned to rtCGM group / 74 assigned to control group

Received int. per arms 75 participants interventional group

74 participants in control group

Lost to follow-up per arms 8 discontinued from control group

No of analysed per arm 75 participants completed baseline and follow-up

phases

75 included in intention-to-treat analysis

66 participants completed baseline and follow-up phases

74 included in intention-to-treat analysis

Statistical analysis

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 267 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ITT, modified ITT, Per protocol; other (specify) ITT

Results

Effectiveness results n (%) 95% CI

Mortality

Change in HbA1c Mean baseline HbA1c was 58.5 mmol/mol (7.5%) for all study participants

HbA1c values remained stable in both groups, with only a marginal between-group difference.

Reductions in glycaemic variability were observed in rtCGM group participants but not in control group participants

Incidence of hypoglycaemia The mean number of hypoglycaemic events per 28 days defined by rtCGM was reduced from 10·8 (SD 10·0) to 3·5 (4·7) among rtCGM group participants and from 14·4 (12·4) to 13·7 (11·6) among control group participants.

Incidence of hypoglycaemic events decreased by 72% for rtCGM participants (IRR 0·28, 95% CI 0·20–0·39, p<0·0001;

the mean number of hypoglycaemic events dropped from 10.4 (SD 9.6) to 3.4 (4.5) in the rtCGM group but remained relatively unchanged in the control group (13.2 [11.4] to 13.2 [10.9]) with an IRR of 0.27 (95% CI 0.20–0.38; p<0.0001).

25 (33·3%) of 75 rtCGM group participants had no hypoglycaemic events during the follow-up phase compared with five (7·6%) of 66 control group participants. This difference corresponds to an odds ratio of 6·1 (95% CI 2.2–17.1; p=0·0006) for avoidance of hypoglycaemia in the rtCGM group compared with the control group.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 268 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The number of nocturnal hypoglycaemic events

Was significantly reduced in the rtCGM group, but not in the control group level 1 hypoglycaemia level 2 hypoglycaemia The percentages of glucose values 3.0 mmol/L or lower

and 3.9 mmol/L or lower

were reduced in the rtCGM group compared with the control group.

The LBGI, as a risk indicator for severe hypoglycaemia, was also reduced in the rtCGM group, whereas it remained relatively unchanged in the control group level 3 hypoglycaemia

Incidence of hyperglycaemia The percentage of hyperglycaemic glucose values was increased slightly in both

study groups but with no significant between-group differences

Time spent in range The time in range increased by 0.7 percentage points in the rtCGM group, whereas the

control group showed a reduction by 2.6 percentage points (p=0.0513)

Time spent in hypoglycaemia

Time spent in hyperglycaemia

Quality of life The diabetes distress total score was also reduced in both groups.

A significant between-group effect was observed only for the hypoglycaemia distress subscale score of the T1-DDS (appendix). Self-reported health status, measured by the EQ-5D questionnaire, showed no significant difference between both groups.

Patient satisfaction The glucose monitoring satisfaction score showed that participants in the rtCGM group were more satisfied with their method of glucose monitoring than were those in the control group.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 269 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Hypoglycaemia fear At study end, fear of hypoglycaemia was lowered in both groups (between group

difference p=0.067).

Incidence of diabetic ketoacidosis

Incidence of hyperosmolar, hyperglycaemic coma

Resource utilization related to DM

Number of visits to emergency room

Number of visits to primary care

Number of visits to specialists

Number of hospitalizations

Number of daily finger-sticks tests The mean frequency of daily SMBG was significantly lower in the rtCGM group than

in the SMBG groups (3·7 [SD 1·9] vs 6·0 [1·3], p<0·0001).

The frequency of SMBG was substantially lower in the rtCGM group than in the control group during the outcome phase.

Therefore, the SMBG frequency during the follow-up phase was adjusted. The adjusted SMBG

data are consistent with those from the rtCGM group

Number of calibration

Need (Yes, with number or No) of re-calibration

Compliance/adherence

Percentage of time using CGM Among rtCGM participants, the average percentage of sensor wear time was 90·7% of study days assessed (first 4 weeks subsequent to randomisation, 30 days before 12-week visit, and 30 days before 26-week visit).

Assessment of adherence during the total therapy phase and the follow-up phase was not possible because rtCGM data were overwritten after 30 days with the most recent data.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 270 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of sensor scans per day (in FGM system)

Safety results n (%) 95% CI

Any AEs 63 severe hypoglycaemia events were observed during the therapy and follow-up phases: 24 in the rtCGM group and 39 in the control group.

The incidence of all severe hypoglycaemia events among control group participants during follow-up was approximately twice the incidence seen in the rtCGM group (1·18 [SD 3·46]

vs 0·64 [1·92] events per patient-year; IRR 0·36 [95% CI 0·15–0·88], p=0·0247; appendix)

Severe hypoglycaemia events requiring third-party assistance without medical assistance for recovery were also less frequent in the rtCGM group than in the control group (19 vs 36 events), with a similar difference in incidence (0·51 [SD 1·75] vs

1·09 [3·41] events per patient-year; IRR 0·26 [95% CI 0·10–0·69], p=0·0071).

Of the eight severe hypoglycaemia episodes requiring medical assistance

for recovery:

five occurred in rtCGM group participants

and three in control group participants (0·13 vs 0·09 events per patient-year; IRR 1·60 [95% CI 0·30–8·49], p=0·59).

Serious AE (SAE) 18 serious adverse events were reported for 15 participants:

7 events occurred in the control group (two severe episodes of hypoglycaemia, one kidney transplantation, one myo-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 271 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

cardial infarction, two colon polyps, and one seizure) and

10 occurred in the rtCGM group (four episodes of severe hypoglycaemia, two diabetic

foot ulcers, one allergic reaction following a wasp sting, two fractures, and one kidney tumour removal).

One serious adverse event occurred before randomisation (whiplash after a car accident).

No event was considered to be related to the investigational device.

Most frequent AEs (by arms)

Most frequent SEAs (by arms)

Death as SAE

Withdrawals due AEs

Costs (only for national assessment)

Author Disclosure (Conflict of interest) LH reports grants from Dexcom Inc during the conduct of the study;

and personal fees from Roche Diagnostics, Integrity Ltd, Medtronic,

and Sanofi outside the submitted work. LH owns shares of Profil

Institut fur Stoffwechselforschung GmbH (Neuss, Germany) and

ProSciento (San Diego, CA USA). GF reports grants from Dexcom Inc

during the conduct of the study; personal fees from Abbott,

Berlin-Chemie, Becton-Dickinson, Dexcom Inc, LifeScan, Menarini,

Novo Nordisk, Sanofi, Sensile, and Ypsomed; and grants and personal

fees from Ascensia and Roche outside the submitted work.

GF-H reports grants from Dexcom Inc during the conduct of the study.

SG reports personal fees from Dexcom Inc during the conduct of the

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 272 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

study. Additionally, SG has two patents (US 61/257,288 and US

61/551,773) licensed to Dexcom Inc. DE reports grants from Dexcom

Inc during the conduct of the study and personal fees from

Berlin-Chemie AG outside the submitted work. DW reports grants from

Dexcom Inc during the conduct of the study. NH reports grants from

Dexcom Inc during the conduct of the study; personal fees from Novo

Nordisk; and grants and personal fees from Abbott, Berlin Chemie,

Ypsomed, and Roche outside the submitted work.

RR: relative risk ITT: intention to treat PP: per protocol

Baseline phase Follow-up phase Adjusted between – p value* group differences(95 %CI)

Control group rtCGM group Control group rtCGM group

n=66 n=75 n=75

Mean duration of rtCGM 26.4(1.7) 27.0(1.5) 27.0(1.8) 27.7(1.5) 0.02(-0.49 to 0.54) 0.9233 wearing during baseline and follow-up phases, days

Primary outcome, low glucose events≤3.0 mmol/L

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 273 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Mean nuber of hypogli- 14.4 (12.4) 10.8 (10.0) 13.7 ( 11.6) 3.5 ( 4.7) 0.28 (0.20 to 0.39)+ <0.001# caemic events per 28 days

Secondary outcomes, rtCGM characteristic

Mean number of noctur- 2.4(2.6) 2.3 ( 2.4) 2.7 ( 2.8) 1.0 (1.0) 0.35 (0.22 to 0.56)+ <0.001# nal hypoglycaemic events per 28 days

Mean rtCGM glucose, 8.7(1.5) 9.0(1.6) 8.9(1.5) 9.5(1.6) 0.28 (-0.05 to 0.62)+ 0.0982 mmol/L

Median percentage of 6.9%(3.6 to 12.3) 5.0%(2.7 to 9.0) 6.4%(3.7 to 12.0) 1.6%(0.9 to 3.7) - <0.0001 rtCGM values

≤3-9 mmol/L

Median percentage of 2.7%(1.0 to 5.7) 1.7%(0.7 to 3.8) 2.5%(1.0 to 6.1) 0.3%(1.0 to 0.9) - <0.0001 rtCGM values≤3.0mmol/L

Median percentage of 59.1%(13.3) 57.8%(15.4) 56.5%(12.2) 58.5%(17.7) 3.1(0.0 to 6.2) 0.0535 rtCGM values

>3-9 mmol/L and

≤ 10.0 mmol/L

Median percentage of 32.8%(15.5) 35.4%(17.5) 35.3%(15.2) 38.8%(18.7) 1.3(-2.3 to 4.9) 0.4681 rtCGM values >10.0mmol/L

Mean duration of 99.5(52.3 to 178.1) 70.9(38.8 to 130.2) 92.2(51.8 to 172.6) 23.9(12.9 to 54.5) - <0.0001 rtCGMvalues ≤3-9

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 274 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

mmol/L per day, min

Mean duration of 36.3(13.1 to 79.7) 24.1(8.9 to 51.0) 32.9(13.1 to 83.9) 3.8(1.1 to 11.9) - <0.0001 rtCGM≤3.0 mmol/L per day, min

Mean duration of rtCGM 851.0(191.7) 831.9(221.5) 814.2(176.0) 842.9(225.2) 44.9(-0.3 to 90.0) 0.0513 values>3-9 mmol/L and ≤10.0 mmol/L per day, min

Mean duration of rtCGM 471.7(223.1) 509.8(252.2) 509.1(219.1) 558.6(268.4) -18.7(-70.3 to 32.9) 0.4744 values>10.0 mmol/L per day, min

Mean rtCGM variability, 40.5%(70.0) 39.3%(7.6) 41.1%(6.9) 34.1%(5.6) 6.2(5.0 to 7.5) <0.0001 coefficient of variation

Median low blood glu- 1.60(0.88 to 2.92) 1.26(0.70 to 2.15) 1.53(0.84 to 2.97) 0.52(0.25 to 0.98) - <0.0001 cose index(rtCGM-LBGI)

Secondary outcomes, SMBG characteristic

Mean SMBG glucose, 8.8(1.6) 9.3(1.7) 9.1(1.6) 9.7(1.8) -0.23(-0.62 to 0.15) 0.2385 mmol/L

Mean number of SMBG 6.4(1.7) 6.8(2.5) 6.0(1.3) 3.7(1.9) -2.5(-3.0 to 2.1) <0.0001 test per day

Median percentage of 9.0%(5.8 to 14.4) 7.6%(4.1 to 11.5) 8.6%(4.8 to 11.7) 2.6%(1.0 to 6.2) - <0.00015 SMBG values

≤3.9 mmol/L

Median percentage of 2.9%(1.0 to 7.2) 2.4%(0.6 to 4.8) 2.6%(1.0 to 4.9) 0.0(0.0 to 1.6) - <0.00015 SMBG values ≤3.0mmol

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 275 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Median percentage of 55.5%(13.5) 53.9%(14.5) 53.6%(12.7) 54.4%(16.6) 3.4(-1.0 to 7.9) 0.12515 SMBG values≤3.0mmol/L and ≤10.0 mmol/L

Median percentage of 33.9%(18.9) 37.5%(16.3) 37.2%(15.2) 41.4%(18.3) 0.2(-4.5 to 4.9) 0.94225 SMBG values>10.0 mmol/L

Mean SMBG variability, 43.7%(6.8) 43.0%(9.7) 43.9%(7.4) 37.8%(7.2) 5.7%(3.4 to 8.0) <0.00015 coefficient of variation

Median low blood glu- 1.85(1.20 to 3.24) 1.58(0.90 to 2.45) 1.75(1.11 to 2.71) 0.61(0.28 to 1.45) - <0.00015 cose index(SMBG-LBGI)

Secondary outcomes, glycaemic control

Mean Hb A1C % 7.4 % (1.0) 7.6 % (1.0) 7.3 % (0.9) 743 % (0.8) 0.03(-0.12 to 0.19) 0.6653

Mean Hb A1Cmmol/mol 57.1 (10.7) 59.3 (10.9) 55.8 (9.6) 57.0 (9.1) 0.37(-2.07 to 1.32) 0.6653 p values are based on covariance analysiswith group allocation as independent factor and baseline values as covariates, and p values for data with skewed distributions are based on covariance analysis using van der Waerden scores. 'Incidence rate ratio adjusted for baseline (reference category-control group), values are based on negative binominal regression analysis (model fit: Pear- son 12=0.92). Adjusted for baseline and frequency of 5BMG during follw•up phase

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 276 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Risk of Bias

Study (Author, year): Heinemann et al HypoDE study, NCT02671968 2018

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Randomisation was done centrally at the study coordinating centre by staff who were not involved

with recruitment or treatment of study participants. A randomisation sequence was generated with SYSTAT 12.0 with a 1:1 allocation; the study centre was a stratifying variable. Randomisation was done block-wise per site (four participants per block).

Allocation concealment (Selection bias) Low Each study site received sealed envelopes with the respective group allocation. After successful completion of the baseline phase, the respective envelope was opened. Participants were

required to wear their rtCGM device 85% of the time during the baseline phase to continue in the study. This requirement might have resulted in selection bias, which could potentially limit the generalisability of our findings to all high-risk individuals with type 1 diabetes.

Blinding of participants (Performance bias) High Study site personnel informed participants about their group allocation. Because of the nature of the intervention, masking of study participants and study personnel was not possible.

Blinding of personnel (Performance bias) High See above

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) Low Intention-to-treat analysis was based on all randomised participants. For the intention-to-treat

analysis, missing values were replaced with multiple imputation technique.

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear Funding; The use of SMBG data to assess the effect of rtCGM on glycaemic outcomes could also be problematic since the control group might have tested blood glucose several times during one hypoglycaemic event. This repeated testing might have biased the effect of SMBG on hypoglycaemia-related outcomes. Additionally, the frequency of SMBG was substantially different during the follow-up period between the groups, which necessitated the use of a post-randomisation covariate. The absence of adjustment for multiplicity for secondary outcomes can be regarded as another limitation.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 277 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

rtCGM MDII or CSII mixed patients EVIDENCE TABLES

MDII and CSII patients Time spent in hypoglycaemia (< 63 mg/dL) during the 26-week study period; Time Clinical effectiveness and safety Battelino RCT T1 MDII or CGM (FreeStyle 2011 CSII Navigator, Abbott spent in hyperglycaemia (>180 mg/dL or >250 mg/dL); The number of hypoglyce- domains mic excursions (< 55 and < 63 mg/dL) per day; The number of hypoglycemic ex- 24 weeks Diabetes Care) vs EU and 120/116 SMBG cursions (< 55 and < 63 mg/dL) during the night period of 0000–0600 h; The risk Israel associated with glucose concentration outside the recommended range; AEs 10-65 y NCT0084360 (stratified 10- 9 17 y; 18-65 y) Clinical effectiveness and safety Mauras 2012 RCT T1 MDII or CGM (FreeStyle Decrease in HbA1c≥0.5% with no severe hypoglycaemic events; severe CSII Navigator, Abbott hypoglycaemia; sensor use; biochemical hypoglycaemia; measures of variability; domains USA 26 weeks Diabetes Care or parental quality of life; patient satisfaction, hypoglycaemia fear; AEs 146/146 for patients on NCT0076052 Medtronic 6 Children 4-10 y Paradigm™ system -Medtronic MiniMed™ MiniLink REAL- time transmiter) vs SMBG Clinical effectiveness and safety Riveline RCT T1 MDII or CGM (FreeStyle HbA1C, glucose stability, hypoglycaemia, ketoacidosis, sensor use, QoL, patient 2012 CSII Navigator, Abbott satisfaction domains 12 months Diabetes Care) vs France Group 1 SMBG n=62/Control NCT0072644 n=61 0 8-60 y

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 278 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Van Beers RCT T1 CSII and Paradigm™ Veo Percentage of time that patients spent in normoglycaemia; Time spent in Clinical effectiveness and safety 2016 MDII normoglycaemia; Severe hypoglycaemia (defined as a hypoglycaemic event domains system used sole- 16 weeks requiring third-party assistance); The percentage of time patients spent in a Netherlands 52/52 ly as a monitor hypoglycaemic state (blood glucose ≤3·9 mmol/L) and a hyperglycaemic state (>10·0 mmol/L);Average daily area under the curve (AUC) of 3·9 mmol/L or less IN CONTROL 18-75 y with a MiniLink™ (expressed as mmol/L min);Frequency (episodes per week) and duration (min per trial transmitter (Med- episode) of CGM-derived hypo glycaemic episodes (≥three sequential sensor Impaired values ≤3·9 mmol/L);Frequency (episode per night) and duration of CGM derived NCT0178790 hypoglycaemi tronic), and the hypoglycaemic episodes at night-time (00 00– 06 00 h); Within-day and between- 3 a awareness Enlite™ glucose day glucose variability (calculated as within-day SD of glucose concentration, coefficient of variation, mean absolute change in glucose concentration, mean of CSII=23 sensor / CSII- daily differences, and continuous overall net glycaemic action);Baseline and 16- MDII=29 treated patients week HbA1c measurements; Self-reported hypoglycaemia awareness (based on Gold and Clarke methods);Diabetes-specific measures of quality of life (PAID-5, continued using HFS, CIDS, EQ5D, and WHO-5), Satisfaction with use of CGM by the CGM-SAT their own pump questionnaire; Frequency of CGM-derived hypoglycaemic episodes with cutoff s of less than 3·5 mmol/L and less than 2·8 mmol/L; AEs for insulin treatment with SMBG

Author, year, reference Battelino et al. 2011.

Battelino T, Phillip M, Bratina N, Nimri R, Oskarsson P, Bolinder J. Effect of Continuous Glucose Monitoring on Hypogly- caemia in Type 1 Diabetes. Diabetes Care. 2011;34(4):795-800. doi:10.2337/dc10-1989.

Study title/objectives Effect of continuous glucose monitoring on hypoglycaemia in type 1 diabetes

Objective: To assess the impact of continuous glucose monitoring on hypoglycaemia in

people with type 1 diabetes.

Study characteristics

Study design Randomized, controlled, multicenter clinical trial

Study Registration number Clinical trial reg. no. NCT00843609, clinicaltrials.gov.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 279 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Country of recruitment -

Centre (single or multicentre) Multicenter (3 centres)

Ethics Committee Approval Approved by the institutional or national medical ethics committees from all three centers included.

The conduct of the study was consistent with the Good Clinical Practice provisions of the Declaration of Helsinki with all amendments and local regulatory requirements.

A written informed consent was obtained fromall participants and parents of minors (under 18 years of age, who signed an assent) before enrollment.

Sponsor This study was supported by Abbott Diabetes Care. T.B. was supported in part by the Slovenian National Research Agen- cy Grants J3-9663, J3-2412, and P3-0343.

Study period (study start, study end) October 2008 to May 2009 (1-month run-in period + 6-month intervention period)

Duration of follow-up (days) Follow-up visits : identical for both groups. • 2–6 days after randomization (confirmation of data recording and to replace the subcutaneous sensor under supervision) • Further visits : at days 60, 120, and 180 (±7 days). • Type 1 diabetes diagnosed for more than 1 year, Inclusion criteria • Reasonable metabolic control assessing carbohydrate intake and self-adjusting insulin, • HbA1c level <7.5% • Intensive insulin treatment with either an insulin pump or multiple daily injections • Not using a real-time continuous glucose monitoring device for at least 4weeks

Exclusion criteria -

Patient characteristics Control group (n=58) Continuous monitoring group (n=62)

Age of patients (years; means ± SD) 26.0 ± 14.6 25.7 ± 14.1

Paediatric patients (number, (%)) 26 (45) 27 (44)

Sex (number, (%)) Female: 33% Female: 42%

BMI (kg/m2) (means ± SD) 22.0 ± 3.8 22.4 ± 3.8

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 280 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Diagnosis (Type I, II or gestational) Type 1 diabetes Type 1 diabetes

Comorbidities (i.e. obesity…) - -

Time since diagnosis with DM (years; means ± SD) 11.4 ± 11.4 11.6 ± 11.3

Insulin treatment (CSII or MDII) (number (%)) • Pump 34 (59%) 47 (76%) • MDI 24 (41%) 15 (24%)

Pregnancy (yes, no) - -

Special subgroup pf patient (i.e., hypoglycaemia fear…) - -

Intervention

Type of medical device (CGM or FGM) Real-time CGM

Name (Description) of medical device FreeStyle Navigator (Abbott Diabetes Care), a continuous glucose monitoring system that measures glucose in interstitial fluid.

Adjunctive or non-adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG + masked continuous glucose monitor es)

Name (Description) of medical device FreeStyle blood glucose meter (Abbott Diabetes Care, Alameda, CA) and FreeStyle test strips and a masked continuous glucose monitor FreeStyle Navigator (Abbott Diabetes Care)

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible (con- - nected) with specific CGM systems (Yes, No)

Outcomes

Primary Time spent in hypoglycaemia (< 63 mg/dL) during the 26-week study period

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 281 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Secondary Time spent in hyperglycaemia (>180 mg/dL or >250 mg/dL)

Time spent in target range (70 to 180 mg/dL or 90 to 180 mg/dL)

The number of hypoglycemic excursions (< 55 and < 63 mg/dL) per day

*An excursion was defined as all consecutive recordings outside the boundary covering at least 10 min. The duration of an excursion was defined as the elapsed time from first excursion to the first reading indicating return inside the excursion boundary.

The number of hypoglycemic excursions (< 55 and < 63 mg/dL) during the night period of 0000–0600 h

The risk associated with glucose concentration outside the recommended range

Flow of patients

No of patients enrolled 122

No of randomized 120

Allocated per arms Control group: 58 Continous Monitoring Group: 62

Received int. per arms Control group: 57 Continous Monitoring Group:62

Lost to follow-up per arms Control group: 9 Continous Monitoring Group:9

No of analysed per arm Control group: 54 Continous Monitoring Group: 62

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) All analyses: performed according to the intent-to-treat principle.

Results

Effectiveness results n (%) 95% CI

Mortality -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 282 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Change in HbA1c Continuous Monitoring Group Control group Difference in means • HbA1c at 6 months, adjusted for baseline HbA1c (visit 2), center 6.69% 6.95% -0.27, (95% CI -0.47 to -0.07, P = 0.008) and age-group (mean) • Mean HbA1c at 6 months, adjusted for baseline HbA1c, center Reduced by 0.39 (6.72 vs. 7.11) and age-group in pump users • Mean HbA1c at 6 months, adjusted for baseline HbA1c, center Reduced by 0.06 (6.70 vs. 6.65) and age-group in MDI patients • Adjusted mean HbA1c in (10–17 years of paediatric subjects Reduced by 0.23 (6.92 vs. 7.15) age) • Adjusted mean HbA1c in (18–65 years of age). adults Reduced by 0.31 (6.51 vs. 6.83)

Time spent in hypoglycaemia Control group Continuous Monitor- Ratio of 95% CI for P ing Group means ratio of means • Hours per day in hypoglycaemia < 63 mg/dL 0.97 ± 1.55 0.48 ±0.57 0.49 0.26–0.76 0.03 • Median (interquartile range) 0.54 (0.23–1.31) 0.26 (0.14–0.54) • Number of hypoglycemic excursions per day 0.76 ± 0.94 0.53 ± 0.60 0.70 0.43–1.03 0.08 • <63 mg/dL • Integrated glucose excursion index (area under 11.1 ± 14.2 5.4 ± 7.6 0.49 0.29–0.79 0.02 • the curve) <63 mg/dL • Hours per day in hypoglycaemia <55 mg/dL 0.41 ± 0.48 0.22 ± 0.34 0.55 0.34–0.91 0.05 • Number of hypoglycemic excursions per day 0.37 ± 0.40 0.28 ± 0.54 0.76 0.47–1.43 0.07 • <55 mg/dL • Hours per day in hypoglycaemia <70 mg/dL 1.60 ± 2.02 0.91 ± 0.81 0.57 0.36–0.80 0.01 • Low blood glucose index 1.74 ± 1.62 1.18 ± 0.82 0.68 0.49–0.89 0.02 • Hours per day in hypoglycaemia < 63 mg/dL – pump users 0.81 0.48 • Hours per day in hypoglycaemia < 63 mg/dL - MDI 1.20 0.49 • Hours per day in hypoglycaemia < 63 mg/dL – paediatric pa- 0.65 0.34 tients (10-17 years) • Hours per day in hypoglycaemia < 63 mg/dL – adults (18-65 1.27 0.59

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 283 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

years)

Post hoc per protocol analysis, where only patients that wore the sensor for >20 days (corresponding to one third of the required time in the control group) (time spent in hypoglycaemia below 63 mg/dL) • paediatric patients (44 of 53 patients) Primary outcome was reduced by 64% (P < 0.001) • adult patients (53 of 63 patients) Primary outcome was reduced by 50% (P = 0.02)

Incidence of hypoglycaemia - level 1 hypoglycaemia - level 2 hypoglycaemia - level 3 hypoglycaemia -

Time spent in range (normoglycaemia) Control group Continuous Monitor- Ratio of 95% CI for P ing Group means ratio of means • 90 to 180 mg/dL 13.5 ± 3.1 15.1 ± 2.7 1.12 1.04–1.21 0.003 • 70 to 180 mg/dL 16.0 ± 3.4 17.6 ± 3.2 1.10 1.02–1.18 0.009

Time spent in hyperglycaemia Control group Continuous Monitor- Ratio of 95% CI for P ing Group means ratio of means • > 180 mg/dL 6.4 ± 3.4 5.5 ± 3.2 0.86 0.71–1.06 0.08 • > 250 mg/dL 1.66 ± 1.53 1.14 ± 1.46 0.69 0.48–1.07 0.06

High blood glucose index 6.0 ± 3.2 5.1 ± 3.1 0.85 0.70–1.05 0.05

Quality of life -

Patient satisfaction -

Hypoglycaemia fear -

Incidence of diabetic ketoacidosis 1 (mild diabetic ketoacidosis, patient in the continuous monitoring group, due to the patient

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 284 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

disconnecting his or her insulin pump)

Incidence of hyperosmolar, hyperglycaemic coma -

Resource utilization related to DM -

Number of visits to emergency room -

Number of visits to primary care -

Number of visits to specialists -

Number of hospitalizations -

Number of daily finger-sticks tests -

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Compliance/adherence -

Sensor wear (during 6 months period) Total Sensor Wear (Days) Sensor Wear (Days / Week)

Control Group Continuous Monitoring Group Control Group Continuous Monitoring Group • Mean 40 136 4.7 5.6 • SD 23 52 2.2 1.4 • Minimum 0 3 0.0 0.6 • 10th Percentile 1 45 1.4 3.6 • 25th Percentile 21 126 3.0 5.1 • Median 47 157 5.6 6.1 • Median sensor wear – paediatric patients 5.6 6.1 • Median sensor wear – adult patients 4.9 6.1 • 75th Percentile 61 172 6.6 6.6

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 285 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• 90th Percentile 66 179 7.1 6.8 • Maximum 71 187 7.4 6.9 • Number of subjects 58 62 57 62

Percentage of time using CGM -

Number of sensor scans per day (in FGM system) -

Safety results n (%) 95% CI

Any AEs 4

Serious AE (SAE) 4 (none related to the study or device); Control group: 1, Continuous monitoring group: 3

Most frequent AEs (by arms) See below (all AE are only SEA)

Most frequent SEAs (by arms) Control group Continuous monitoring group • Prolonged cephalea • Mild diabetic ketoacidosis (DKA) • Myasthenia gravis • Laparoscopic repair of the spermatic • vein

Death as SAE -

Withdrawals due AEs -

Costs (only for national assessment)

Author Disclosure (Conflict of interest)

General Abbott Diabetes Care provided funding, device-related training, and analytical support.

Abbott Diabetes Care was permitted to review the manuscript and suggest changes, but the final decision on content and submission of the manuscript was exclusively retained by the authors,who take responsibility for the accuracy and integrity

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 286 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

of the data and analyses. This study was an investigator-initiated trial. The study and protocol were designed by the investigators. The manuscript was prepared by the investigators.

TADEJ BATTELINO, MD, PHD T.B.’s institution received research grant support, with receipt of travel and accommodation expenses in some cases, from Abbott, Medtronic, Novo Nordisk, and Diamyd. T.B. is on the speaker’s bureaux of Eli Lilly, Novo Nordisk, Bayer, and

Medtronic and is a member of scientific advisory boards for Bayer, Life Scan, and Medtronic.

MOSHE PHILLIP, MD M.P.’s institution received research grant support, with receipt of travel and accommodation expenses in some cases, from Medtronic and Dexcom. M.P. is a consultant for Animas, Medtronic, and Bayer and is a member of scientific advisory boards for CGM3, D-Medical, and Physical Logic.

NATASA BRATINA, MD, PHD N.B. is on the speaker’s bureau of Medtronic.

JAN BOLINDER, MD, PHD J.B. is on the speaker’s bureaux of Abbott, Medtronic, and Sanofi-Aventis and is a member of scientific advisory boards for AstraZeneca, Medtronic, and Merck, Sharp & Dohme.

. RR: relative risk

ITT: intention to treat

PP: per protocol

Risk of Bias

Study (Author, year): Battelino 2011

Judgement (Low, Unclear, High) Support for judgement

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 287 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Random sequence generation (Selection bias) Low Quote: “The randomization sequence was computer generated”

Allocation concealment (Selection bias) Unclear Quote: “allocations were concealed using envelopes”

Comment: it is unclear whether envelopes were sequentially numbered opaque and sealed

Blinding of participants and personnel (Performance bias) High Study not blinded

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) Low Attrition was low: 10 (17%) in control group, 9 (15%) in the CGM group

Selective reporting (Reporting bias) High The study protocol was registered (https://clinicaltrials.gov/ct2/show/NCT00843609), and by comparing results with registered protocol, it is obvious that multiple secondary outcomes that were planned in the protocol were not reported

Other source of bias (Other bias) Low Other sources of bias not found

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 288 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, year, reference Riveline et.al 2012.

Riveline J-P, Schaepelynck P, Chaillous L, et al. Assessment of Patient-Led or Physician-Driven Continuous Glucose Monitoring in Patients With Poorly Controlled Type 1 Diabetes Using Basal-Bolus Insulin Regimens: A 1-year multicenter study. Diabetes Care. 2012; 35(5):965-971. doi:10.2337/dc11-2021.

Study title/objectives Assessment of Patient-Led or Physician-Driven Continuous Glucose Monitoring in Patients With Poorly Controlled Type 1 Diabetes Using Basal-Bolus Insulin Regimens: A 1-year multicenter study

Objective: The benefits of real-time continuous glucose monitoring (CGM) have been demonstrated in patients with type 1 diabetes. Aim was to compare the effect of two modes of use of CGM, patient led or physician driven, for 1 year in subjects with poorly controlled type 1 diabetes.

The secondary objectives were evaluation of changes in glucose variability, frequency of mild and/or severe hypoglycemic events, changes in QoL, efficacy of CGM in patients treated by CSII and MDI, and efficacy of CGM in optimally educated patients versus nonoptimally educated patients.

Study characteristics

Study design Randomized multicenter controlled study

Study Registration number Clinical trial reg. no. NCT00726440, clinicaltrials.gov.

Country of recruitment -

Centre (single or multicentre) Multicenter (19 diabetes care centers)

Ethics Committee Approval The trial was approved by the ethics committee of the Paris VIMedical Faculty.

All patients (or the parents of minors) had read the patient information sheet and signed informed consent forms before enrolment.

Sponsor This study was supported by the Association Française des Diabétiques and the Leon Fredericq Foundation of the University of Liège (for the Belgian part of this trial).

FreeStyle Navigators, home glucose meters, and test strips used for the study were provided by Abbott Diabetes Care.

Study period (study start, study end) May 2008 - June 2009

Duration of follow-up (days) Visits were scheduled 20 days after randomization and at 3, 6, 9, and 12 months. ● Age between 8 and 60 years Inclusion criteria ● Type 1 diabetes diagnosed 12 months earlier ● Treatment by insulin analogs using either CSII or MDI ● HbA1c level ≥ 8.0%

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 289 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

● SMBG performed at least twice daily ● Required wearing of CGM during a 10-day test period to confirm patient ability and willingness to use CGM

Exclusion criteria Not specified

* The per-protocol population excluded 3 of full analysis set patients on account of major protocol deviations: switch from MDI to CSII in 2 participants and pregnancy in 1 participant

Patient characteristics Group 1 (patient-led Group 2 (physician pre- Group 3 (conventional All patients scribed SMBG [control group]) use of CGM) (n = 62) (n = 178) CGM) (n = 55) (n = 61)

Age of patients (years) 37.5 ± 13.4 33.5 ± 13.3 37.8 ± 13.9 36.4 ± 13.6

Sex (Male patients, n (%)) 31 (50.0%) 25 (45.5%) 39 (63.9%) 95 (53.4%)

BMI (kg/m2) 24.1 ± 3.9 24.7 ± 3.2 25.3 ± 3.6 24.7 ± 3.6

Diagnosis (Type I, II or gestational) Type I Type I Type I Type I

Comorbidities (i.e. obesity…) - - - -

Time since diagnosis with DM (years) 16.4 ± 9.1 15.4 ± 8.9 18.8 ± 10.6 16.9 ± 9.6

Insulin treatment (CSII or MDII)

● CSII 30 (48.4) 27 (49.1) 37 (60.7) 94 (52.8)

● MDI 32 (51.6) 28 (50.9) 24 (39.3) 84 (47.2)

Pregnancy (yes, no) No No No No

Special subgroup of patient (i.e., hypoglycaemia fear…) - - - -

Intervention

Type of medical device (CGM or FGM) CGM (Group 1: patient-led use; Group 2: physician-prescribed)

Name (Description) of medical device FreeStyle Navigator glucose needle-type sensor system (Abbott Diabetes Care, Alameda, CA)

Adjunctive or non-adjunctive Adjunctive (confirmation of glucose values using the meter included in the Navigator device before making therapeutic decisions)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 290 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device Home glucose meters and test strips (Abbott Diabetes Care, Alameda, CA)

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible (con- - nected) with specific CGM systems (Yes, No)

Outcomes

Primary Reduction in HbA1c at 12 months versus baseline

Secondary SD of BG calculated from the 8-point BG profiles

Number of mild and severe hypoglycemic events

Number of ketoacidosis episodes

Change in QoL scores from baseline to 12 months

Daily insulin requirements

Relationship between compliance (i.e., mean number of sensors used per month) and ΔHbA1c

Flow of patients

No of patients enrolled 197

No of randomized 178

Allocated per arms Group 1: 62 ; Group 2: 55; Group 3 (control): 61

Received int. per arms -

Lost to follow-up per arms 3 excluded patients (groups not specified)

The per-protocol population excluded 3 full analysis set patients on account of major protocol deviations: switch from

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 291 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

MDI to CSII in 2 participants and pregnancy in 1 participant.

No of analysed per arm Group 1: 62 ; Group 2: 55; Group 3 (control): 61

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) Per protocol

Results

Effectiveness results n (%) 95% CI

Change in HbA1c Group 1 Group 2 Group 3 Group 1+ Group 2 P value

(patient) (physician) (control) Vs 3

(n = 62) (n = 55) (n = 61)

Group

1 vs 3

2 vs.3

1+2 vs3

Change in HbA1c (from baseline to 12 months) (-0.50% [95% CI - (-0.45% [-0.66 to (0.02% [-0.18 (-0.48% [-0.63 to - P = 0.0006 0.70 to -0.29]) 0.33]) -0.24]) to 0.23] P = 0.0018

P<0.0001

Relative HbA1c reduction >10% versus baseline 29.0 % 25.5% 11.5% - P=0.044

P=0.036

Percentage of patients achieving the HbA1c target 9.7% 14.6% 1.6% - P=0.025

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 292 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

of < 7.5% P=0.026

Change in HbA1c:

● intergroup HbA1c in patients on CSII [Groups 1+2 (n=60) ver- -0.67% (-1.01 to -0.33; P = 0.0001) sus Group 3 (n = 24)] ● intergroup HbA1c in patients on MDI [Groups 1+2 (n=56) versus -0.28% (-0.67 to 0.10; NS) Group 3 (n = 37)] Changes in basal or prandial insulin doses Ranged from -2 to 2 units/day during the study period

P value

Group 1+ Group Group Group 1 Group 2 Group 3 2 1 vs 3 (n = 62) (n = 55) (n = 61) Vs 3 2 vs.3

1 vs2

Glycemic stability (Δ SD in BG (mg/dL; SD of 8-point daily blood glucose -9.3 [-19.0 to 0.4] -15.7 [-26.8 to - 4.6] -0.6 [-8.9 to 7.7] 0.018 0.183 profile (mg/dL), values are means (95%CIs)) between month 12 and month 0 0.049

0.393

Incidence of hypoglycaemia Group 1 Group 2 Group 3

Level 3 hypoglycaemia (Severe hypoglycaemia) (n,%) (n = 62) (n = 55) (n = 61)

● 0 events 47 (75.8) 50 (90.9) 55 (90.2)

● 1 8 (12.9) 4 (7.3) 3 (4.9)

● 2 4 (6.5) 0 (0.0) 0 (0.0)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 293 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

● 3 1 (1.6) 1 (1.8) 2 (3.3)

● 4 1 (1.6) 0 (0.0) 0 (0.0)

● 6 0 (0.0) 0 (0.0) 1 (1.6)

● 7 1 (1.6) 0 (0.0) 0 (0.0)

Patients with ≥ 1 event, n (%) P value

Group 1+ Group 2 Group

Group 1 Group 2 Group 3 Vs 3 1 vs 3

(n=62) (n=55) (n=61) 2 vs.3

1 vs2

15 (24.2 5 (9.19) 6 (9.8) 0.1682 0.9962

0.2153

0.9962

Patients with diabetic ketoacidosis, n (%) Group 1 (n=62) Group 2 (n=55) Group 3 (n = 61)

● 0 60 (96.8) 53 (96.4) 59 (96.7)

● 1 1 (1.6) 2 (3.6) 1 (1.6)

● 2 0 (0.0) 0 (0.0) 1 (1.6)

● 8 1 (1.6) 0 (0.0) 0 (0.0)

Compliance/adherence (Sensor prescription and real use) Group 1 (n=62) Group 2 n=55

● Actual sensor use per month 3.42 per month (median [Q1–Q3] [2.20–3.91]) 2.25 per month (1.27–2.99)

● Percentage of time using CGM (Actual sensor use in relation to 57 ± 20% of the prescribed time 65 ± 29% of the prescribed time the prescribed time)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 294 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

● Correlation between sensor use and HbA1c levels No correlation (P = 0.117, R2 = 0.0449) Negatively correlated (P = 0.026, R2 = 0.1050)

● Total sensor consumtion during 1 year Significantly lower (34%) in group 2 than in group 1 (P = 0.001)

Number of SMBG per week Significantly decreased in both CGM groups versus the control group (-9 ± 12 vs. 1 ± 12, P < 0.0001).

(*Baseline Daily home glucose meter readings (n/week): Group 1= 29.2 ± 15.4; Group 2= 25.1 ±13.7; Group 3 (control)= 30.0 ± 14.8)

Training: In groups 1 and 2, 47.6% of patients had received optimal training. These optimally trained patients exhibited greater improvement in HbA1c than the others.

This difference remained significant after adjustment for compliance with CGM use (ΔHbA1c: -0.71 ± 0.81 vs. -0.30 ± 0.81, P = 0.033).

Quality of life (at year 1) Group 1+2 (n=117) Group 3 (n = 61) P value

● Physical component (SF-36 questionnaire) 1.47 ± 6.52 -2.48 ± 6.52 P = 0.0042

● Mental component 0.65 ± 10.55 -1.03 ± 10.62 NS

● Global DqoL Did not differ significantly between the groups

Group 1+2 (n=117) Group 3 (n=61) P value

Patient satisfaction (assessed by one scale of the DQoL) 2.83 ± 12.61 -2.12 ± 12.61 P = 0.0447

Mortality -

Incidence of hyperglycaemia -

Time spent in range -

Time spent in hypoglycaemia -

Time spent in hyperglycaemia -

Hypoglycaemia fear -

Incidence of hyperosmolar, hyperglycaemic coma -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 295 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Resource utilization related to DM -

Number of visits to emergency room -

Number of visits to primary care -

Number of visits to specialists -

Number of hospitalizations -

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Number of sensor scans per day (in FGM system) -

Safety results n (%) 95% CI

Any Aes -

Serious AE (SAE) -

Most frequent AEs (by arms) -

Most frequent SEAs (by arms) -

Death as SAE -

Withdrawals due Aes -

Costs (only for national assessment)

Author Disclosure (Conflict of interest)

JEAN-PIERRE RIVELINE, MD J.-P.R. participates in advisory boards and as a consultant for Abbott Diabetes Care, LifeScan, sanofiaventis, and Eli Lilly and has received honoraria and payment for presentations, travel, and accommodation expenses from Abbott Diabetes Care and Novo Nordisk.

ERIC RENARD, MD, PHD E.R. is a consultant for Abbott Diabetes Care and has received honoraria and payment for the development of educa-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 296 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

tional presentations, travel, and accommodation expenses from Abbott Diabetes Care.

ALFRED PENFORNIS, MD, PHD A.P. is a member of the boards of Boehringer Ingelheim, Merck-Serono, Novartis, Novo Nordisk, and sanofi-aventis; has received payment for the development of educational presentations including speakers’ office services from Abbott, Merck Sharp & Dohme, Novartis, Novo Nordisk, and sanofi-aventis; and has had travel and accommodation expenses covered or reimbursed by Abbott, Boehringer Ingelheim, Merck Sharp & Dohme, and sanofi-aventis.

NADIA TUBIANA-RUFI, MD N.T.-R. is a member of a scientific advisory panel.

PIERRE-YVES BENHAMOU, MD, PHD P.-Y.B. participates in advisory boards and as a consultant for LifeScan, sanofi-aventis, and Eli Lilly and has received fees for speaking in symposia organized by Abbott Diabetes Care, Medtronic, Eli Lilly, Roche Diagnostics, and

Novo Nordisk.

HÉLÈNE HANAIRE, MD H.H. participates in advisory boards and as a consultant for Abbott Diabetes Care, Medtronic, sanofi-aventis, Eli Lilly, and Novo Nordisk and has received fees for speaking in symposia organized by Abbott Diabetes Care, Medtronic, Eli Lilly, and LifeScan RR: relative risk ITT: intention to treat PP: per protocol

Risk of Bias

Study (Author, year): Riveline 2012

Judgement (Low, Unclear, Support for judgement High)

Random sequence generation (Selection bias) Unclear It is only indicated that participants were randomized, but method not reported

Allocation concealment (Selection bias) Unclear Not reported

Blinding of participants (Performance bias) High Quote: Because of the nature of the treatment, the study was not blinded.

Blinding of personnel (Performance bias) High Quote: Because of the nature of the treatment, the study was not blinded.

Blinding of outcome assessment (Detection bias) Unclear Not reported

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 297 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Incomplete outcome data (Attrition bias) Unclear The authors indicate that 197 participants were randomized an later full analysis set – FAS- included 178 participants, namely 62 in group 1, 55 in group 2, 61 in group 3. However, FAS was defined as all randomized patients having at least

one postbaselineHbA1c available. So it is unclear how big was actually attrition, and how many participants were lost from each group.

Selective reporting (Reporting bias) Unclear Protocol registration not reported

Other source of bias (Other bias) Low Other sources of bias not found

Author, year, reference Mauras et al 2012

Mauras et al. A Randomized Clinical Trial to Assess the Efficacy and Safety of Real-Time Continuous GlucoseMonitor- ing in the Management of Type 1 Diabetes in Young Children Aged 4 to <10 Years. DIABETES CARE, VOLUME 35, FEBRUARY 2012.

Study title/objectives Randomized Trial to Assess Efficacy and Safety of Continuous Glucose Monitoring in Children 4-<10 Years With T1DM

Study characteristics

Study design RCT, phase 3, open label

Participants meeting these criteria were randomly assigned to either the CGM group or the usual care control group, using a permuted-blocks design stratified by clinical center.

Study Registration number NCT00760526

Country of recruitment USA

Centre (single or multicentre) 5 centres in USA

Ethics Committee Approval The protocol was approved by the institutional review boards of the five participating sites.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 298 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Sponsor Jaeb Center for Health Research

Study period (study start, study end) Between January 2009 and December 2010 was randomization

September 2010-June 2011 Clinical trial.gov

Duration of follow-up (days)

Inclusion criteria Study participants had a clinical diagnosis of type 1 diabetes and were using daily insulin therapy for at least 12months.

Eligibility criteria included age 4.0 to <10.0 years, HbA1c ≥7.0%, and basal bolus therapy using either an insulin pump or at least three multiple daily injections (MDIs) of insulin for the prior

3 months with no plans to switch the insulin modality within the next 6 months.

Exclusion criteria Exclusion criteria included 1) diagnosis of diabetes prior to 6 months of age; 2) use

of a medication that could affect glycemic control, the performance of the CGM sensor,

or completion of any aspect of the protocol; and 3) use of CGM during the prior 6 months.

Patient characteristics

No of patient 146

Age of patients 4-9 year

mean age 7.5 ± 1.7 years

Sex Both

BMI

Diagnosis (Type I, II or gestational) Type 1 diabetes

Comorbidities (i.e. obesity…)

Time since diagnosis with DM median diabetes duration 3.5 years

Insulin treatment (CSII or MDII) 52 MDII

94 CSII

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 299 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Pregnancy (yes, no)

Special subgroup of patient (i.e., hypoglycaemia fear…)

Intervention

Type of medical device (CGM or FGM) CGM

Name (Description) of medical device FreeStyle Navigator (Abbott Diabetes Care,Inc.)

Medtronic Paradigm™ System

Participants randomized to the CGM (treatment) group were provided with an unblinded CGM device, sensors, and a FreeStyle Flash blood glucose meter and test strips.

A Free- Style Navigator was provided unless the participant was already using a Medtronic Paradigm™ insulin pump, in which case a MiniMed MiniLink™ REAL-Time Transmitter could be used.

Among the 74 participants in the CGM group, 10 (14%) were provided with a

Paradigm™ CGM device and 64 (86%) with a Navigator CGM device.

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es) Standard Glucose Monitoring With Home Glucose Meter

Name (Description) of medical device FreeStyle meter

FreeStyle Flash (Abbott Diabetes Care, Inc., Alameda, CA)blood glucose meter and test strips

Sensor integrated (Yes, No)

Sensor augmented (or enabled) insulin pump systems compatible (con-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 300 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

nected) with specific CGM systems (Yes, No)

Outcomes

Primary The primary outcome of a decrease from randomization to 26 weeks in HbA1c ≥0.5% with no severe hypoglycemic events occurred in 13 of 69 (19%) participants in the CGM group and 19 of 68 (28%) in the control group (P = 0.17).

Secondary Clinical Trials.gov

• Severe hypoglycaemia [ Time Frame: 26 weeks ] • Percentage of sensors values in range (71 mg/dL to 180 mg/dL) [ Time Frame: 26 weeks ] • Biochemical hypoglycaemia (percentage of sensor values

• Parental quality of life measures [ Time Frame: 26 weeks ]

Flow of patients

No of patients enrolled

No of randomized the trial randomized 146 children

Allocated per arms 74 intervention; 72 control

Received int. per arms Intervention 69; control 68

Lost to follow-up per arms Intervention 4; control 4

No of analysed per arm

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) ITT

Results

Effectiveness results

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 301 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

n (%) 95% CI

Mortality

Change in HbA1c Mean change in HbA1c was 20.1% in each group (P = 0.79).

There was no correlation between CGMuse and change in HbA1c (rs = 20.09, P = 0.44).

CGM in 4- to 9-year-olds did not improve glycemic control despite a high degree of parental satisfaction with CGM.

Mean change in HbA1c was similar between groups (-0.16 ±0.6 in each group, P = 0.79)

There was no association between change in HbA1c from baseline to 26 weeks

and the overall amount of CGM sensor wear during the entire 26 weeks (Spearman

rs = 20.09, P = 0.44) or during month 6 (Spearman rs = 20.11, P = 0.37).

However, the 28 participants who wore a sensor ≥6 days/week during month 6 tended to have a slightly greater reduction in HbA1c compared with the 41 participants who wore a sensor less frequently

Incidence of hypoglycaemia Severe hypoglycaemia rates were similarly low in both groups.

Three participants (4%, three total events) in the CGM group and five participants (7%, six total events) in the control group experienced at least one severe hypoglycemic event, with no significant differences comparing treatment groups (incidence rate = 8.6 and 17.6 per 100 person-years, respectively; P = 0.80) level 1 hypoglycaemia level 2 hypoglycaemia At 26 weeks, both groups had CGM glucose values≤ 60 mg/dL for<1% of the day. level 3 hypoglycaemia

Incidence of hyperglycaemia

Time spent in range

Time spent in hypoglycaemia

Time spent in hyperglycaemia Participants in both groups had glucose values >250 mg/dL (13.9 mmol/L)for >20% of the day.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 302 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Quality of life At 26 weeks, there were no significant differences between treatment groups on

the Hypoglycaemia Fear or the PAID questionnaire survey scores (Table 2 and Supplementary

Tables A3 and A6). However, scores on the Blood Glucose Monitoring System Rating Scale were indicative of fewer problems/concerns perceived by the CGM group compared with the

control group (Table 2 and Supplementary Table A5).

On the CGM Satisfaction Scale at 26 weeks (Table 3 and Supplementary Table A4), parents generally reported a high degree of satisfaction with CGM, with an average item score of 3.9 and 86% of scores ≥3.5 (on a 5-point Likert-type scale, with 3 being neutral). Mean item

scores were more favourable than neutral (>3.0) on all 43 items.

Scores on the Benefits of CGM subscale tended to be slightly higher than scores on the Lack

of Hassles of CGM subscale (mean 4.1 ± 0.4 vs. 3.9 ± 0.6, respectively). It is particularly noteworthy that >90% of parents responded that use of CGM makes adjusting insulin easier, shows patterns in blood glucose not seen before, and makes them feel safer knowing that

they will be warned about low blood glucose before it happens.

No one responded that he or she would not recommend CGM for other children with type 1 diabetes.

Patient satisfaction CGM wear was well tolerated, and parental satisfaction with CGM was high.

Hypoglycaemia fear However, parental fear of hypoglycaemia was not reduced.

Incidence of diabetic ketoacidosis

Incidence of hyperosmolar, hyperglycaemic coma

Resource utilization related to DM

Number of visits to emergency room

Number of visits to primary care

Number of visits to specialists

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 303 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of hospitalizations

Number of daily finger-sticks tests

Number of calibration

Need (Yes, with number or No) of re-calibration

Compliance/adherence

Percentage of time using CGM The amount of CGM sensor wear decreased during the 26 weeks of the study (P < 0.001) (Fig. 1), with only 41% averaging at least 6 days/week of wear in month 6.

The amount of sensor wear in month 6 did not vary with age overall (r =-0.07) and was not associated with baseline HbA1c (r = -0.02)

There was no association between change in HbA1c from baseline to 26 weeks and the overall amount of CGM sensor wear during the entire 26 weeks (Spearman rs = -0.09, P = 0.44) or during month 6 (Spearman rs = -0.11, P = 0.37).

However, the 28 participants who wore a sensor ≥6 days/week during month 6 tended to have a slightly greater reduction in HbA1c compared with the 41 participants who wore a sensor less frequently(mean change from baseline to 26 weeks -0.3 ± 0.7% vs. 0.0 ± 0.5%, P = 0.01; 25 vs. 15% with a reduction in HbA1c ≥0.5% without a severe hypoglycaemia event, P = 0.33).

Among those wearing a sensor ≥6 days/week in month 6, the median percentage of glucose values in the target range of 71–180 mg/dL (3.9-4.4 mmol/L)was 51%, with 38% >200 (11.1mmol/L) mg/dL, 16% >250 mg/dL, and 0.3% ≤60 (3.3 mmol/L) mg/dL compared with 43, 44, 23, and 0.4%, respectively, in those wearing a sensor less frequently

Number of sensor scans per day (in FGM system)

Safety results n (%) 95% CI

Any AEs There were no cases of diabetic ketoacidosis and no serious adverse events attributable to

the study interventions, including no serious skin reactions.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 304 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Serious AE (SAE)

Most frequent AEs (by arms)

Most frequent SEAs (by arms)

Death as SAE

Withdrawals due AEs

Costs (only for national assessment)

Author Disclosure (Conflict of interest) Authors: NELLY MAURAS, ROY BECK, DONGYUAN XING, KATRINA RUEDY, BRUCE BUCKINGHAM, MICHAEL TANSEY, NEIL H WHITE, STUART A. WEINZIMER, WILLIAM TAMBORLANE, CRAIG KOLLMAN.

B.B. serves on theMedtronic MiniMedMedical Advisory Board and provides research support to Medtronic MiniMed and Abbott Diabetes Care.

S.A.W. received an honorarium for serving as a consultant to Medtronic and serves as an advisor and received honoraria for Animas/LifeScan.

W.T. serves as a consultant/advisor toMedtronic and is paid a fee based on hours ofwork. C.K. has served as a paid consultant to Diabetes Technology Management, which was hired by Medtronic MiniMed to form a Veo advisory board to formulate a consensus statement on the design and analysis of a trial to evaluate the Veo LGS system in January 2011 for a fee of less than $10,000. Abbott Diabetes Care provided the FreeStyle Navigator and the FreeStyle blood

glucose meters and test strips. Medtronic Mini-Med provided the Paradigm™ MiniLink transmitters and Sof-sensors at a discounted price.

No other potential conflicts of interest relevant to this article were reported.

RR: relative risk ITT: intention to treat PP: per protocol

Risk of Bias

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 305 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Study (Author, year): Mauras 2012

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Unclear Quote: Participants meeting these criteria were randomly assigned to either

the CGM group or the usual care control group, using a permuted-blocks design stratified by clinical center.

Comment: although some characteristics of randomization were reported – such as stratification and using permuted blocks – the actual method of randomization was not reported

Allocation concealment (Selection bias) Unclear Not reported

Blinding of participants (Performance bias) High Quote 1: Participants randomized to the CGM group were provided with an unblinded CGMdevice, sensors, and FreeStyle Flash

Blinding of personnel (Performance bias) Unclear Not reported

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) Unclear Attrition not clearly reported. Below tables and figure there are notes that certain result excludes certain number of participants who dropped out at some stages of the trial, but there is no clear and detailed information about overall attrition in each group

Selective reporting (Reporting bias) Unclear Quote: The study is listed on www.clinicaltrials.gov (NCT00760526).

Protocol is available on this link: https://clinicaltrials.gov/ct2/show/NCT00760526

By comparing outcomes planned in the protocol and outcomes reported in the manuscript, it is clear that certain outcomes – such as MAGE – were not reported

Other source of bias (Other bias) Low Other sources of bias not found

Author, year, reference Van Beers, 2016

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 306 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Van Beers A J C, Hans DeVries H J, Kleijer J S, Smits M M, Geelhoed-Duijvestijn H P, Mark H H Kramer H H M et al. Continuous glucose monitoring for patients with type 1 diabetes and impaired awareness of hypoglycaemia (IN CONTROL): a randomised, open-label, crossover trial. Lancet Diabetes Endocrinol 2016

doi: 10.1016/S2213-8587(16)30193-0

Study title/objectives Continuous glucose monitoring for patients with type 1 diabetes and impaired awareness of hypoglycaemia (IN CON- TROL): a randomised, open-label, crossover trial

Study characteristics

Study design Randomised, open-label, crossover trial

Study Registration number NCT01787903

Country of recruitment Netherlands

Centre (single or multicentre) Multicentre (two medical centres)

Ethics Committee Approval Ethical approval was granted by the medical ethical committee of the VU University Medical Center

Sponsor Eli Lilly and Sanofi

Study period (study start, study end) February 2013 – April 2016

Duration of follow-up 28 weeks (16 weeks intervention period + 12 weeks of washout period before cross over)

6 week run in period

Inclusion criteria Type 1 diabetes (based on American Diabetes Association [ADA] criteria)

Aged 18–75 years

Treated with either continuous subcutaneous insulin infusion (CSII) or multiple daily insulin injections (MDI)

Undertaking at least three SMBG measurements per day,

Impaired awareness of hypoglycaemia as defined by Gold criteria (ie, with a Gold score ≥4)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 307 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

After 6 week run in phase patients were eligible for randomisation if the maximum number of sensor values per day (288) for at least 4 days per week had been obtained, three to four valid calibrations per day had been done, and a daily mean absolute diff erence less than 18% (in case of a difference between the highest and the lowest calibration value <5·6 mmol/L) or a daily mean absolute difference less than 28% (in case of a diff erence between the highest and the lowest calibration value ≥5·6 mmol/L) was noted.

Exclusion criteria History of renal, liver, or heart disease

Current untreated proliferative diabetic retinopathy

Current malignancy

Current use of non-selective β blockers

Current psychiatric disorders

Current substance abuse or alcohol abuse

Pregnancy

Current use of CGM other than for a short period (3 consecutive months)

Hearing or vision impairment that could hinder perception of the glucose display and alarms, Poor command of the Dutch language

Any disorder that precluded full understanding of the purpose and instructions of the study Participation in another clinical study

Any known or suspected allergy to trial-related products

Patient characteristics

Age of patients ITT population Baseline characteristics

Age, mean (SD), y 48.6 (11.6)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 308 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Sex ITT population Baseline characteristics

Female, n (%) 24 (46%)

BMI ITT population Baseline characteristics

BMI, mean (SD), kg/m2 25 (3.8)

Diagnosis (Type I, II or gestational) Type I

Comorbidities (i.e. obesity…) ITT population Baseline character- N (%)

istics

Rethinopathy 24 (46%)

Peripheral neurophaty 14 (27%)

Microalbuminuria 8 (15%)

Time since diagnosis with DM

ITT population Baseline characteristics

Diabetes duration, median (IQR) 30.5 (18.5-40.8)

Impaired awareness of hypoglycaemia ITT population Baseline character- N (%) istics

Impaired awareness of hypogly- 45 (87%)

caemia (Gold and Clarke

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 309 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ITT population Baseline character- Mean (SD)

istics

Gold score 5.4 (0.7)

Clark score 5.0 (1.3)

Insulin treatment (CSII or MDII) CSII or MDII

Pregnancy (yes, no) No

Special subgroup pf patient (i.e., hypoglycaemia fear…) Patients with impaired awareness of hypoglycaemia

Intervention

Type of medical device (CGM or FGM) CGM

Name (Description) of medical device The CGM system used during the intervention phase consisted of the Paradigm™ Veo system used solely as a monitor with a MiniLink™ transmitter (Medtronic), and the Enlite™ glucose sensor. CSII-treated patients continued using their own pump for insulin treatment.

Adjunctive or non-adjunctive Adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device NR

Sensor integrated (Yes, No) NR

Sensor augmented (or enabled) insulin pump systems compatible (con- NR nected) with specific CGM systems (Yes, No)

Outcomes

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 310 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Primary Mean difference in the percentage of time that patients spent in normoglycaemia (4·0–10·0 mmol/L) between CGM and SMBG calculated over the total intervention periods

Secondary Time spent in normoglycaemia each month to show an effect over time,

Severe hypoglycaemia (defined as a hypoglycaemic event requiring third-party assistance),

The percentage of time patients spent in a hypoglycaemic state (blood glucose ≤3·9 mmol/L) and a hyperglycaemic state (>10·0 mmol/L)

Average daily area under the curve (AUC) of 3·9 mmol/L or less (expressed as mmol/L min),

Frequency (episodes per week) and duration (min per episode) of CGM-derived hypo glycaemic episodes (≥three sequential sensor values ≤3·9 mmol/L)

Frequency (episode per night) and duration of CGM derived hypoglycaemic episodes at night-time (00 00– 06 00 h),

Within-day and between-day glucose variability (calculated as within-day SD of glucose concentration, coefficient of variation, mean absolute change in glucose concentration, mean of daily diff erences, and continuous overall net glycaemic action)

Baseline and 16-week HbA1c measurements

Self-reported hypoglycaemia awareness (based on Gold and Clarke methods),

Diabetes-specific measures of quality of life (PAID-5, HFS, CIDS, EQ5D, and WHO-5), Satisfaction with use of CGM by the CGM-SAT questionnaire.

Frequency of CGM-derived hypoglycaemic episodes with cutoff s of less than 3·5 mmol/L and less than 2·8 mmol/L.

Flow of patients

No of patients enrolled 52

No of randomized 52

Allocated per arms 26 assigned to initial CGM 25 assigned to initial SMBG

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 311 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

23 crossover to SMBG 24 crossover to CGM

Received int. per arms NR

Lost to follow-up per arms assigned to initial CGM assigned to initial SMBG

3 discontinued treatment 3 discontinued treatment

3 withdrew consent 3 withdrew consent

No of analysed per arm assigned to initial CGM assigned to initial SMBG

26 26

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) ITT

Results

Effectiveness results n (%) 95% CI

Mortality NR

Change in HbA1c NR

HbA1c

CGM phase SMBG phase Mean difference P value (95% CI)

Value at study endpoint 7.3% 7.3% 0% 0.812 (%) (7.1 to 7.5) (7.1 to 7.5) (–0.1 to 0.2)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 312 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(measured after 16 weeks intervention period)

Value at study endpoint 56.0 56.3 0.2 0.812 (mmol/mol) (53.9 to 58.1) (54.1 to 58.3) (–1.4 to 1.9) (measured after 16 weeks intervention period)

Change from baseline (%) –0.1% –0.1% –0.1% 0.449

(–0.2 to 0.1) (–0.2 to 0.0) (–0.2 to 0.1)

Change from baseline –0.5 –1.3 –0.8 0.449 (mmol/mol) (–1.9 to 0.9) (–2.7 to 0.1) (–2.8 to 1.2)

Mean glucose concentration (mmol/L)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Mean glucose concentration 8.3 (8.0–8.6) 8.7 (8.4–9.0) –0.4 (–0.6 to –0.2) 0.001 (mmol/L)

Within-day SD of glucose concentration (mmol/L) Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Within-day SD of glucose concen- 2.8 (2.7–2.9) 3.3 (3.1–3.4) –0.5 (–0.6 to –0.4) <0.0001 tration (mmol/L)

Coefficient of variation of glucose concentration

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Overall 39.5 46.3 –6.7 <0.0001

(38.2–40.8) (44·9–47.6) (–8.0 to –5.5)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 313 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Within day 33.5 38.0 –4.5 <0.0001

(32.4–34.6) (36.9–39.1) (–5.5 to –3.6)

Between days 18.4 23.1 –4.7 <0.0001

(17.5–19.4) (22.2–24.1) (–5.9 to –3.5)

Mean absolute glucose change, (mmol/L per h)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Mean absolute glucose change 1.7 (1.7–1.8) 1.8 (1.7–1.9) –0.1 0.049

(–0.1 to –0.0)

Mean of daily difference, (mmol/L)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Mean of daily difference 3.3 (3.1–3.5) 4.2 (4.0–4.4) –0.9 (–1.1 to 0.7) <0.0001

Continuous overall net glycaemic action at 1 h intervals, (mmol/L)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Continuous overall net glycaemic 1.7 (1.6–1.8) 1.8 (1.7–1.9) –0.1 (–0.2 to –0.0) 0.002 action at 1 h intervals

Incidence of hypoglycaemia NR level 1 hypoglycaemia NR level 2 hypoglycaemia NR level 3 hypoglycaemia NR

Severe hypoglycaemia

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 314 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

CGM phase SMBG phase Odds ratio P value

(95% CI)

Number of severe hypogly- 14 34 0.033 caemic events

Patients with ≥1 severe hypo- 10 (19%) 18 (35%) 0.48 (0.2 to 0.062 glycaemic event, n(%) 1.0)

Both CGM phase and

SMBG phase

Number of severe hypoglycaemic 4 events resulting in seizure or coma

Severe hypoglycaemic event resulted 1 in the patient being admitted to the hospital

Self-reported hypoglycaemia awareness

Self-reported hypoglycaemia aware- CGM phase SMBG phase Mean difference (95% P value ness CI)

Gold score at study endpoint 4.6 5.0 –0.4 (–0.7 to 0.0) 0.035

(measured after 16 weeks interven- (4.3 to 5.0) (4.6 to 5.4) tion period)

Change in Gold score from baseline –0.5 –0.1 –0.4 (–0.8 to 0.0) 0.076

(–0.8 to –0.1) (–0.4 to 0.2)

Clarke score at study endpoint 4.4 4.4 0.0 (–0.4 to 0.4) 0.953

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 315 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(measured after 16 weeks interven- (3.9 to 4.8) (3.9 to 4.8) tion period)

Change in Clarke score from base- –0.1 –0.4 –0.3 (–0.9 to 0.2) 0.216 line (–0.5 to 0.3) (–0.8 to 0.0)

Incidence of hyperglycaemia NR

Time spent in range

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Percentage of time spent with 65.0% 55.4% 9.6% (8.0 to 11.2) <0.0001 glucose concentration 4.0–10 (62.8–67.3) (53.1–57.7) mmol/L

Time spent with glucose concen- 15.6 13·3 2.3 (1.9 to 2.7) <0.0001 tration 4.0–10 mmol/L (h per day) (15.1–16.2) (12·7–13·8)

Time spent in hypoglycaemia

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Percentage of time spent with 6.8% 11.4% –4.7% <0.0001 glucose concentration ≤ 3.9 (5.2–8.3) (9.9–13.0) (–5.9 to –3.4) mmol/L

Time spent with glucose concen- 1.6 (1.3–2.0) 2.7 (2.4–3.1) –1.1 <0.0001 tration ≤ 3.9 mmol/L (h per day) (–1.4 to –0.8)

CGM-derived hypoglycaemic events (events per week)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 316 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

CGM-derived hypoglycaemic 10.1 (8.7–11.4) 11.1 (9.8–12.5) –1.1 (–2.1 to –0.1) 0·028 events (events per week)

Duration of CGM-derived hypoglycaemic events (min per event)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Duration of CGM-derived hypogly- 60.7 (54.9–66.4) 98.5 (92.6– –37.8 (–44.6 to –30.9 <0.0001

caemic events (min per event) 104.3)

AUC ≤3.9 mmol/L per 24 h (mmol/L per min)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

AUC ≤3.9 mmol/L per 24 h 62.9 (45.1–80.7) 115.8 (97.8– –52.9 (–68.1 to –37. <0.0001 (mmol/L per min) 133.8) 7)

Nocturnal hypoglycaemia (00 00–06 00 h)

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value CI)

Percentage of time spent with 7.6% (5.3–9.8) 13.3% (11.0– –5.7% (–8.2 to –3.2) <0.0001 glucose concentration ≤3·9 mmol/L 15.5)

CGM-derived hypoglycaemic 0.26 (0.21–0.31) 0.33 (0.28–0.38) –0.07 (–0.11 to –0.02) 0·003 events per night

Duration of CGM-derived hypogly- 78.7 (69.3–88.1) 131.4 (121.9– –52.7 (–62.7 to –42.7) <0·0001 caemic events at night (min per 140.9)

event)

Time spent in hyperglycaemia

Mean (95% CI) CGM phase SMBG phase Mean difference (95% P value

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 317 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

CI)

Percentage of time spent with 28.2% 33.2% (30.0– –5.0% (–6.9 to –3.1) <0.0001 glucose concentration ≥ 10 mmol/L 36.3) (25.1–31.3)

Time spent with glucose concen- 6.8 (6.0–7.5) 8.0 (7.2–8.7) –1.2 (–1.6 to –0.7) <0.0001 tration ≥ 10 mmol/L (h per day)

Quality of life No between group differences were noted in quality of life from scores on the HFS Behaviour subscale, PAID-5, CIDS, EQ5D, or WHO-5 between the CGM and SMBG phases (data not shown). Scores on the HFS Worry subscale, transformed to a 0–100 scale, were lower after the CGM phase compared with the SMBG phase (32.5 vs 38.9; mean difference 6.4, 95% CI 1.4–11.4; p=0.014). CGM-SAT scores after the CGM phase were higher than neutral (3.0 on a 5.0 scale), with a mean score of 3.8 (SD 0.6).

Patient satisfaction Satisfaction with use of CGM was assessed by the CGM-SAT questionnaire.

CGM-SAT scores after the CGM phase were higher than neutral (3.0 on a 5.0 scale), with a mean score of 3.8 (SD 0.6).

Hypoglycaemia fear NR

Incidence of diabetic ketoacidosis No ketoacidosis occurred during the trial

Incidence of hyperosmolar, hyperglycaemic coma NR

Resource utilization related to DM NR

Number of visits to emergency room NR

Number of visits to primary care NR

Number of visits to specialists NR

Number of hospitalizations 1

One hospital admission for erysipelas (not at the CGM insertion site) occurred during the CGM phase.

Number of daily finger-sticks tests ITT population Baseline characteristics

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 318 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Self-reported daily home glucose-meter readings 5.0 (4.0-6.0) (number per day)

Number of calibration NR

Need (Yes, with number or No) of re-calibration NR

Compliance/adherence

Percentage of time using CGM Sensor use during CGM period 89.4% (IQR 80.8 – 95.5)

Number of sensor scans per day (in FGM system) NR

Safety results n (%) 95% CI

Any AEs NR

Serious AE (SAE) 5

Five serious adverse events other than severe hypoglycaemia occurred during the trial, but none were deemed related to the study intervention

In the washout phase, one event each of anaphylactic reaction to eye drops, cerebral contusion, rupture of the Achilles tendon, and rupture of the biceps tendon occurred.

Most frequent AEs (by arms) NR

Mild to Moderate adverse events

CGM phase SMBG phase Wash out phase

Mild to Moderate adverse events, 11 16 2 No

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 319 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Mild to Moderate adverse events, CGM phase SMBG phase Wash out phase No

Related to the musculoskeletal 2 6 0

System

Urinary tract infection 2 0 0

Dermal infection 2 1 1

Gastrointestinal infection 1 3 0

Dermal burn 1 0 0

Fever for less than 1 week 2 3

Excision of lipoma 1 0 0

Dyspnoea 0 0 1

Peridontitis 0 1 0

Infection of 1

the upper respiratory tract

Glaucoma 0 1 0

The mild to moderate adverse events were deemed unrelated to the study intervention.

Most frequent SEAs (by arms) NR

Death as SAE NR

Withdrawals due AEs No adverse events resulted in discontinuation of the study

Costs (only for national assessment) NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 320 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author Disclosure (Conflict of interest) Through MHHK and MD, the VU University Medical Center (Amsterdam, Netherlands) received research grants from AstraZeneca, Boehringer Ingelheim, Novo Nordisk, and Sanofi . JHD (or his institution) has received research support and speaker’s fees from Abbott, Dexcom, Medtronic, Roche Diabetes Care and Seneonics. PHG-D has received speaker fees from Abbott and Medtronic and is a member of the advisory board of Medtronic. All other authors declare no competing interests.

RR: relative risk ITT: intention to treat PP: per protocol NR: not reported

Risk of Bias

Study (Author, year): Van Beers 2016

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Quote: “we randomly assigned patients (1:1) using a computer-generated allocation sequence”

Allocation concealment (Selection bias) Low The allocation sequence (CGM–SMBG or SMBG–CGM) was generated by the institutional trial pharmacist, and masked to the physicians (by use of sealed envelopes) at the time of randomisation (ensuring low risk of allocation bias)

Blinding of participants and personnel (Performance bias) High After randomisation, the sequence was no longer masked for both study physicians (who also assessed outcomes and analysed the data) and patients.

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) Low Quote: “six patients (12%) withdrew early: two discontinued after the CGM period because of motivational issues, one had personal circumstances necessitating discontinuation, two withdrew because they could not upload the masked CGM device, and one withdrew because of poor adherence to CGM (baseline character-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 321 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

istics presented in table 1). 18 patients (35%) of 52 practised the technique of carbohydrate counting. All 52 randomly assigned patients were included in the primary analysis.

Selective reporting (Reporting bias) Low Study protocol published: Cornelis A.J. van Beers, Susanne J. Kleijer, Erik H. Serné, Petronella H. Geel- hoed-Duijvestijn, Frank J. Snoek, Mark H.H. Kramer, Michaela Diamant

BMC Endocr Disord. 2015; 15: 42. Published online 2015 Aug 21. doi: 10.1186/s12902-015-0040-3

Some outcomes planned in protocol were not reported in the manuscript:

Quote: “outcomes specified in our protocol (qualitative analysis of experience with CGM and function of autonomic nervous system) are beyond the scope of this report and will be reported elsewhere”.

Other source of bias (Other bias) Low Other sources of bias not found

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 322 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

rtCGM CSII patients EVIDENCE TABLES

Table : Main characteristics of studies included

Author and year Study Number of patients Intervention (s)/Control Main Included in clinical or study type/T1 or randomised/analysed endpoints effectiveness and/ or name/MDII or T2 DM or safety domain CSII or both both/Study period

RCTs

rtCGM

CSII patients

Ly 2013 RCT T1 CSII Medtronic Paradigm™ Veo/CSII with SMBG Change in HbA1c levels, % Clinical effectiveness and (95% CI); Number of people safety domains CSII + CGM + Suspend: sensor-integrated pump (Medtron- Australia 95/95 with hypoglycaemic events; 24 weeks ic Paradigm™ Veo System, Medtronic MiniMed™) with Hypoglycaemic incidence rate; ACTRN126100 Mixed population 4-50 y automated insulin suspension n=46; CSII + SMBG: contin- HUS 00024044 (70% children <18 y) ue using their insulin pump n=49 Impaired hypoglycaemia awareness

Author, year, reference Ly et al. 2013.

Ly T T, Nicholas A J, Retterath A, Mun Lim E, Davis A E and Jones W T. Effect of Sensor-Augmented Insulin Pump Therapy and Automated Insulin Suspension vs Standard Insulin Pump Therapy on Hypoglycaemia in Patients With Type 1 Diabetes A Randomized Clinical Trial. JAMA. 2013; 310(12):1240-1247. doi:10.1001/jama.2013.277818

Study title/objectives Effect of Sensor-Augmented Insulin Pump Therapy and Automated Insulin Suspension vs Standard Insulin Pump Therapy on Hypoglycaemia in Patients With Type 1 Diabetes

A Randomized Clinical Trial

Objective: To determine the incidence of severe and moderate hypoglycaemia with sensor-augmented pump with low-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 323 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

glucose suspension compared with standard insulin pump

therapy

Study characteristics

Study design Randomized clinical trial

Study Registration number anzctr.org.au Identifier: ACTRN12610000024044

Country of recruitment Australia

Centre (single or multicentre) Multicentre (tertiary adult and paediatric hospitals in Western Australia)

Ethics Committee Approval The ethics committees of participating institutions approved the protocol, and participants gave written informed consent

Sponsor This study was partly funded by the Juvenile Diabetes Research Foundation. Insulin pumps and glucose sensors were provided by Medtronic via an unrestricted grant.

Study period (study start, study end) Recruitment: Between December 2009 and January 2012

Duration of follow-up 6 months

Inclusion criteria Patients aged 4 to 50 years with type 1 diabetes receiving insulin pump therapy, having been diagnosed with diabetes for at least a year, being treated with an insulin pump for at least 6 months, having a glycated haemoglobin level of 8.5%or lower, and having impaired awareness of hypoglycaemia. Hypoglycaemia unawareness score (HUS) was determined with the modified Clarke questionnaire with a minimum score of 4, suggestive of impaired hypoglycaemia awareness.

Exclusion criteria Use of sensor augmented insulin pump therapy

Pregnancy

Patient characteristics

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 324 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Age of patients , mean (SD), y 18.6 (11.8) y

Baseline Standard Insulin Sensor-Augmented Pump With Low-

Pump Glucose Suspension

(n = 49) (n = 46)

Age, Mean, (SD)[ range], y 19.7 (12.9)[5.4-48.6] 17.4 (10.6) [5.1-45.7]

No (%) of Participants

Age group,y Insulin Pump Sensor-Augmented Pump With Low- Glucose Suspension (n = 49) (n = 46)

4-<7 2 (4.1) 2 (4.3)

7-<12 14 (28.6) 13 (28.3)

12-<18 18 (36.7) 16 (34.8)

18-50 15 (30.6) 15 (32.6)

Sex Baseline No (%) of Participants

Insulin Pump Sensor-Augmented Pump With Low-

Glucose Suspension (n = 49)

(n = 46)

Female 28 (57.1) 20 (43.5)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 325 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

BMI NR

Diagnosis (Type I, II or gestational) Type I

Comorbidities (i.e. obesity…) NR

Time since diagnosis with DM, mean (SD), y 11.0 (8.9) years

Baseline

Insulin Pump Sensor-Augmented Pump With Low-

Glucose Suspension (n = 49)

(n = 46)

Duration of diabetes, Mean (SD), y 12.1 (10.0) 9.8 (7.4)

Duration of pump therapy, Mean (SD), y 4.1 (3.4) y

Baseline Insulin Pump Sensor-Augmented Pump With Low-

Glucose Suspension (n = 49)

(n = 46)

Duration of pump therapy, Mean 4.4 (3.4) 3.8 (3.3) (SD), y

Insulin/kg, mean (SD), U/kg

Baseline Insulin Pump Sensor-Augmented Pump With Low-

Glucose Suspension (n = 49)

(n = 46)

Insulin/kg, mean (SD), U/kg 0.76 (0.23) 0.83 (0.17)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 326 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Insulin treatment (CSII or MDII) CSII

Pregnancy (yes, no) No (Exclusion criteria included pregnancy)

Special subgroup pf patient (i.e., hypoglycaemia fear…) Hypoglycaemia unawareness

Subgroup analysis In a subgroup, counterregulatory hormone responses to hypoglycaemia were assessed using the hypoglycaemic clamp technique

Intervention

Sensor augmented (or enabled) insulin pump systems compatible (connected) with specific CGM systems (Yes, No)

Sensor integrated (Yes, No) Yes

Sensor-augmented pump (Medtronic Paradigm™ Veo System, Medtronic Minimed™) with automated insulin suspension

Type of medical device (CGM or FGM) CGM

Name (Description) of medical device NR

Adjunctive or non-adjunctive Adjunctive

Comparator

Sensor augmented (or enabled) insulin pump systems compatible (con- No nected) with specific CGM systems (Yes, No)

Sensor integrated (Yes, No) No

Patients continued to use their own pumps with SMBG

Outcomes

Primary The primary outcome was the combined incidence of severe and moderate hypoglycaemia

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 327 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Severe hypoglycaemia was defined as a hypoglycaemic seizure or coma.

Moderate hypoglycaemia was defined as a hypoglycaemic event requiring assistance from another person.

Secondary Glycated haemoglobin

Average percentage of time spent in the hypoglycaemic range during day and night

Hypoglycaemia unawareness score

Flow of patients

No of patients enrolled 271 Patients screened for impaired hypoglycaemia awareness

100 Assessed for eligibility

No of randomized 95

Allocated per arms

Standard Insulin Sensor-Augmented Pump With Low- Pump Glucose Suspension

No of patients 49 46

Received int. per arms NR

Withdrew consent Standard Insulin Pump Sensor-Augmented Pump With Low- Glucose Suspension

No of patients 4 Withdrew consent 4 Withdrew consent

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 328 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

1 Moved away 1 Moved away

2 Discontinued 3 Discontinued

intervention intervention

1 Dissatisfied with

randomization

Lost to follow-up per arms

Standard Insulin Sensor-Augmented Pump With Low- Pump Glucose Suspension

No of patients 0 1

No of analysed per arm

Standard Insulin Sensor-Augmented Pump With Low- Pump Glucose Suspension

No of patients 49 46

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) ITT. Intention-to-treat population was defined as all patients who were randomized and had at least 1 visit after baseline. P values <.05 were considered statistically significant; 2-sided P values are reported.

To analyse the change in moderate and severe hypoglycaemia from baseline to end point and to account for a large proportion of 0 events as well as baseline differences, a 0-inflated

Poisson model was implemented using PROC GENMOD.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 329 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Results

Effectiveness results n (%) 95% CI

Mortality NR

Glycated hemoglobin, % Glycated hemoglobin, % Insulin Pump Sensor-Augmented Pump

With Low-Glucose Sus- (n = 49) pension

(n = 46)

Baseline, mean (95% CI) 7.4 (7.2 to 7.6) 7.6 (7.4 to 7.9)

End point, mean (95% CI) 7.4 (7.2 to 7.7) 7.5 (7.3 to 7.7)

Change, least square mean −0.06 (−0.2 to 0.09) −0.1 (−0.3 to 0.03) (95% CI)

P value 0.42 0.11

Least square mean differ- 0.07 (−0.2 to 0.3) ence (95% CI)

P value 0.55

Sum of Severe and Moderate Hypoglycaemia Sum of Severe and Moder- Insulin Pump Sensor-Augmented ate Hypoglycaemia Pump With Low- (n = 49) Glucose Suspension

(n = 46)

Baseline Rate per 100 patient- 20.7 (13.8 to 30) 129.6 (111.1 to 150.3)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 330 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

months (95% CI)

No. of events (total No. of 28 (45) 175 (45) patients)

End point 6-Month rate per 100 11.9 (6.8 to 19.3) 28.4 (19.8 to 39.6) patient-months (95% CI)

No. of events (total No. of 13 (45) 35 (41) patients)

Incidence rate per 100 34.2 (22.0 to 53.3) 9.5 (5.2 to 17.4) patient-months (95% CI)

Patients modelled 45 41

Incidence rate ratio per 100 3.6 (1.7 to 7.5) patient-months (95% CI)

P value <0.001

Severe Hypoglycaemia Severe Hypoglycaemia Insulin Pump Sensor-Augmented Pump With Low- (n = 49) Glucose Suspension

(n = 46)

Baseline Rate per 100 patient- 2.1 (0.8 to 4.6) 1.8 (0.6 to 4.3) months (95% CI)

No. of events (total No. of 6 (49) 5 (46) patients)

End point 6-Month rate per 100 2.2 (0.5 to 6.5) 0 (0 to 2.4)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 331 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

patient-months (95% CI)

No. of events (total No. of 6 (45) 0 (41) patients)

Incidence rate difference 1.5 (0.3 to 2.7) from baseline to end point

(95% CI)

0.02 P value

Moderate hypoglycaemia Moderate Hypoglycaemia Insulin Pump Sensor-Augmented Pump With Low- (n = 49) Glucose Suspension

(n = 46)

Baseline Rate per 100 patient- 20 (13.2 to 29.1) 128.1 (109.8 to 148.7) months (95% CI)

No. of events (total No. of 27 (45) 173 (45) patients)

End point 6-Month rate per 100 9.6 (5.1 to 16.5) 28.5 (19.8 to 39.6) patient-months (95% CI)

No. of events (total No. of 13 (45) 35 (41) patients)

Incidence rate per 100 26.3 (15.4 to 45.0) 9.6 (5.1 to 18.1) patient-months (95% CI)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 332 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Patients modeled 45 41

Incidence rate ratio per 100 2.7 (1.2 to 6.1) patient-months (95% CI)

P value

0.01

Sensitivity analysis Severe and Moderate Insulin Pump Sensor-Augmented For patients younger than 12 years Hypoglycaemia Pump With Low- (n = 15) Glucose Suspension

(n = 15)

Baseline Rate per 100 patient- 42.2 (95%CI, 25.4- 302.2 months 65.9) (95%CI, 253.6-357.5)

No. of events (total No. of 19 136 patients)

End point 6-Month rate per 100 17.8 (95%CI, 7.7-35.0) 64.4 (95%CI, patient-months 43.2-92.6)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 333 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

No. of events (total No. of 8 29 patients)

Adjusted incidence rate 5.5 (95% CI, 2.0-15.7; P < 0.001) favouring the low-glucose suspension group ratio

Incidence of hypoglycaemia level 1 hypoglycaemia NR level 2 hypoglycaemia NR level 3 hypoglycaemia NR

Incidence of hyperglycaemia NR

Time spent in range

Average Percentage of Hours Spent in Hypoglycemic Range Median (Interquartile Range)

Average time of glu- Time point Insulin Pump Sensor-Augmented P Value cose levels <70 Pump With Low-Glucose (n = 49) mg/dLb Suspension

(n = 46)

Day Baseline 8.3 5.7 (2.8-8.2) 0.15

(2.8-13.0)

End point 6.9 4.1 (2.6-7.6) 0.02

(3.9-10.6)

Night Baseline 11.1 7.3 (2.4-16.4) 0.20

(3.1-21.3)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 334 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Endpoint 11.8 4.4 (2.1-8.8) <0.001

(6.4-16.2)

Median (Interquartile Range)

Average time of glu- Time point Insulin Pump Sensor-Augmented P Value cose levels <60 Pump With Low-Glucose (n = 49) mg/dLb Suspension

(n = 46)

Day Baseline 3.2 (0.7-9.0) 2.4 (0.4-4.4) 0.15

End point 3.3 (1.6-5.9) 1.5 (0.9-3.7) 0.01

Night Baseline 4.8 (0-12.9) 2.3 (0-9.5) 0.38

Endpoint 6.2 (4.2-9.9) 2.4 (0.4-5.3) <0.001

A retrospective continuous glucose-monitoring device (Medtronic iPro2, Medtronic Minimed) was inserted at baseline and at each visit during the intervention period to allow measurement of glucose values over 6 days to determine the amount of time that participants were in the hypoglycemic range

Time spent in hypoglycaemia NR

Time spent in hyperglycaemia NR

Quality of life NR

Patient satisfaction NR

Hypoglycaemia fear NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 335 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Hypoglycaemia unawareness Hypoglycaemia una- Insulin Pump Sensor-Augmented Pump With

wareness score Low-Glucose Suspension (n = 49)

(n = 46)

Baseline 6.4 (95%CI, 5.9-6.8) 5.9 (95% CI, 5.5-6.4)

End point 5.1 (95%CI, 4.5-5.6) 4.7 (95% CI, 4.0-5.1)

P value P <0 .001 P < 0.001

Square mean difference 95% CI, −0.2; −0.9 to 0.5; low-glucose suspension P = 0.58 pump only

Incidence of diabetic ketoacidosis None of the participants experienced diabetic ketoacidosis or hyperglycaemia with ketosis during the intervention

Incidence of hyperosmolar, hyperglycaemic coma NR

Resource utilization related to DM NR

Number of visits to emergency room NR

Number of visits to primary care NR

Number of visits to specialists NR

Number of hospitalizations NR

Number of daily finger-sticks tests NR

Number of calibration NR

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 336 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Need (Yes, with number or No) of re-calibration NR

Compliance/adherence NR

Percentage of time using CGM NR

Sensor use Sensor use over 6 month, Sensor-Augmented Pump With

median Low-Glucose Suspension

(n = 46)

All age groups 68%

Younger than 12 years 71%

12-18 years 54%

>18 years 81%

Number of sensor scans per day (in FGM system) NR

Hypoglycaemic Clamp Studies Epinephrine response to Insulin Pump Sensor-Augmented Pump With

hypoglycaemia Low-Glucose Suspension (n = 49)

(n = 46) 15 patients Before intervention period 113 pg/mL 220 pg/mL included (95% CI, 101-124 pg/mL) (95% CI, 192-248 pg/mL) (between 12

After the intervention (at 123 pg/mL 148 pg/mL and 26 years) the end of 6 months) (95% CI, 110-137 pg/mL) (95% CI, 134-168) before and after the P value 0.74 0.26 intervention period

mean [SD] -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 337 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

age, 16.4 [3.2] years;

diabetes duration 9.5 [4.3] years;

insulin dose, 0.81 [0.15] units/kg;

hypoglycaemia unawareness score, 5.7 [1.2]

Intravenous insulin is infused at 80 mU/m2 per minute and 20% glucose is infused simultaneously to maintain blood glucose levels at 100 mg/dL for 60 minutes (euglycaemia) before reducing over 30 minutes to 50 mg/dL for 40 minutes (hypoglycaemia). Samples for epinephrine were taken every 10 to 30 minutes during the baseline euglycaemia and hypoglycaemia phases. Three samples were obtained in each phase. Epinephrine response was calculated as the mean difference in epinephrine concentration obtained during hypoglycaemia compared with baseline euglycaemia.

Safety results Device failure occurred on 2 occasions and was corrected with replacement of the sensor transmitter (Minilink™). n (%) 95% CI

Any AEs NR

Serious AE (SAE) NR

Most frequent AEs (by arms) NR

Most frequent SEAs (by arms) NR

Death as SAE NR

Withdrawals due AEs NR

Costs (only for national assessment)

Author Disclosure (Conflict of interest) All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Ly reported that he was supported by the Juvenile Diabetes Research Foundation postdoctoral fellowship and received travel support from Medtronic. Dr Jones reported receiving honoraria for scientific lectures and travel reimbursement from Medtronic, sano-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 338 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

fi-aventis, Eli Lilly, and Novo-Nordisk. No other financial disclosures were

reported.

RR: relative risk ITT: intention to treat PP: per protocol NR: not reported

Risk of Bias

Study (Author, year): Ly 2013

Judgement (Low, Unclear, High) Support for judgement

Random sequence generation (Selection bias) Low Quote: Randomization was computer-generated

Allocation concealment (Selection bias) Unclear Not reported

Blinding of participants (Performance bias) High Quote: it was not possible to blind the patients to the intervention

Blinding of personnel (Performance bias) High Quote: it was not possible to blind the patients to the intervention

Blinding of outcome assessment (Detection bias) Unclear Not reported

Incomplete outcome data (Attrition bias) Low Less than 10 percent attrition in both groups; reasons for attrition reported

Selective reporting (Reporting bias) Unclear Trial registration: anzctr.org.au Identifier: ACTRN12610000024044

The protocol is available on this link: https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?ACTRN=ACTRN12610000024044

Type of registration is indicated as retrospectively registered. Therefore, the risk of selective reporting is automatically unclear because we do not know what the authors prospectively planned for this study.

Other source of bias (Other bias) Low Other sources of bias not found

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 339 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

rtCGM vs FGM EVIDENCE TABLES

Main characteristics of studies included

CGM vs FGM

Clinical effectiveness and safety domains Reddy 2018 RCT T1 40/40 CGM vs FGM Change in time spent in hypoglycaemia (< 3.3 mmol/l); time spent in hypo, eugycaemia, hyper and in target; Low blood UK ≥ 18 years Dexcom G5® vs glucose index; Severe hypoglycaemia; Hypoglycaemia risk; 8 weeks HbA1c levels; Gold Score; Hypoglycaemia fear (HFS-II); NCT030282 Abbott Freestyle Impaired ® Diabetes-related emotional distress 20 hypoglycaemia Libre awareness

Author, year, reference Reddy et al. 2018.

Reddy M, Jugnee N, El Laboudi A, et al. A randomized controlled pilot study of continuous glucose monitoring and flash glucose monitoring in people with Type 1 diabetes and impaired awareness of hypoglycaemia. Diabet. Med. 2018; 35, 483–490

https://doi.org/10.1111/dme.13561

Study title/objectives A randomized controlled pilot study of continuous glucose monitoring and flash glucose monitoring in people with Type 1 diabetes and impaired awareness of hypoglycaemia.

Objectives: To assess the impact of CGM and flash glucose monitoring in a high-risk group of people with Type 1 diabetes.

This is the first head-to-head glucose monitoring study comparing continuous glucose monitoring (CGM) and flash glucose monitoring.

Study characteristics

Study design Randomized, non-masked parallel group study

Study Registration number Clinical Trial Registry No: NCT03028220

Country of recruitment United Kingdom (UK)

Centre (single or multicentre) Single (a single specialist site)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 340 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Ethics Committee Approval Ethical approval was obtained from the National Health Service (NHS) Research Ethics Committee.

All participants gave written informed consent.

Sponsor Dexcom funded the investigator-initiated study and provided materials. The study was sponsored by

Imperial College London. This paper presents independent research funded by Dexcom and supported by the NIHR CRF and BRC at Imperial College Healthcare NHS Trust. The views expressed are those of the authors and not necessarily those of Dexcom, the NHS, the NIHR or the Department of Health.

Study period (study start, study end) 22 January 2016 and 7 December 2016 (recruitment period + two-week run-in phase + treatment period of 8 weeks)

Duration of follow-up (days) 8 weeks

(2 weeks after randomization: a telephone visit; 4 weeks after randomization; 8 weeks after randomization) • Age ≥ 18 years. Inclusion criteria • Type 1 diabetes for > 3 years • A severe hypoglycaemic event recorded in the last 12 months requiring third-party assistance or had a Gold score of ≥ 4 • Participants with severe hypoglycaemia and a Gold score of < 4 may not have impaired hypoglycaemia awareness; however, severe hypoglycaemia is associated with impaired awareness of hypoglycaemia and they have therefore been included in this high-risk study population. • An intensified multiple-dose insulin injection regimen for over 6 months • Diagnosis of Type 1 diabetes confirmed based on clinical features and a fasting c-peptide < 200 pmol/l • CGM or flash glucose monitoring usage within the last 6 months (except short periods of diagnostic blinded use un- Exclusion criteria der clinic supervision) • Regular usage of paracetamol • Pregnancy or participants planning pregnancy or breastfeeding • Enrollement in other clinical trials • Active malignancy or participants under investigation for malignancy. • Severe visual impairment. • Rreduced manual dexterity.

Patient characteristics CGM (n = 20) Flash glucose monitoring All participants (n = 40)

(n = 20)

Age of patients (years) (median (IQR)) 50.5 (45.0–64.5) 48.5 (34.0–63.0) 49.5 (37.5–63.5)

Sex (male : female) 12 : 8 12 : 8 24 : 16

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 341 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

BMI - - -

Diagnosis (Type I, II or gestational) Type 1 Type 1 Type 1

Comorbidities (i.e. obesity…) - - -

Time since diagnosis with DM (years) (median (IQR)) 30.0 (25.0–36.0) 28.0 (16.5–36.5) 30.0 (21.0–36.5)

Insulin treatment (CSII or MDII) MDI MDI MDI

Pregnancy (yes, no) No No No

Special subgroup pf patient (i.e., hypoglycaemia fear…) - - -

Intervention

Type of medical device (CGM or FGM) CGM

Name (Description) of medical device Dexcom G5®

Adjunctive or non-adjunctive Used nonadjunctively (without capillary blood glucose verification before making a treatment decision)

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical FGM devices)

Name (Description) of medical device Abbott Freestyle Libre® flash glucose monitoring system

Adjunctive or non-adjunctive Non-adjunctive (without capillary blood glucose verification before making a treatment decision)

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible - (connected) with specific CGM systems (Yes, No)

Outcomes

Primary Change in time spent in hypoglycaemia (< 3.3 mmol/l) from baseline to endpoint with CGM vs. flash glucose monitoring

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 342 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Secondary Percentage of time spent in hypoglycaemia < 2.8, < 3.5 and < 3.9 mmol/l

Percentage of time in euglycaemia (3.9–7.8 mmol/l)

Percentage of time spent in target (3.9–10 mmol/l)

Percentage of time spent in hyperglycaemia > 7.8, > 10 and > 15 mmol/l

Low blood glucose index (LBGI, a measure of hypoglycaemia risk derived from continuous glucose data)

Severe hypoglycaemia (requiring third-party assistance to treat)

Hypoglycaemia risk

HbA1c levels

Gold Score

Hypoglycaemia fear (HFS-II)

Diabetes-related emotional distress (PAID questionnaire)

Flow of patients

No of patients enrolled 47

No of randomized 40

Allocated per arms CGM group: 20 FGM group: 20

Received int. per arms CGM group: 20 FGM group: 20

Lost to follow-up per arms CGM group: 0 FGM group: 0

No of analysed per arm CGM group: 20 FGM group: 20

*For outcomes derived from CGM data n = 19 were analysed in the CGM group due to loss of the 8-week CGM data for one participant resulting from uploading error.

Statistical analysis

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 343 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ITT, modified ITT, Per protocol; other (specify) Intention to treat.

Results

Effectiveness results n (%) 95% CI

Percentage time within defined glucose range (median percentage time (and IQR))

Continuous glucose monitoring Flash glucose monitoring Median change from baseline

(n = 19) (n = 20) (95% CI)

Median (IQR) Median (IQR)

Baseline Endpoint Baseline Endpoint Continuous glucose Flash glucose P value

(−2 to 0 weeks) (4 to 8 weeks) (−2 to 0 weeks) (4 to 8 weeks) monitoring monitoring

< 2.8 mmol/l 2.3 (0.6–10.7) 0.9 (0.2–1.8) 4.1 (2.5–5.9) 3.8 (3.0–6.4) −1.2 (−4.3 to −0.5) 1.3 (−1.0 to 2.4) 0.003

< 3.3mmol/l 4.5 (1.9–14.1) 2.4 (1.0–5.1) 6.7 (4.8–9.5) 6.8 (4.8–11.7) −3.0 (−5.0 to −0.3) 1.3 (−1.4 to 3.6) 0.006

< 3.5 mmol/l 5.5 (3.1–15.7) 3.5 (1.8–6.3) 8.0 (5.7–10.7) 8.2 (6.0–13.2) −2.8 (−4.7 to −0.3) 2.0 (−1.0 to 4.7) 0.004

< 3.9 mmol/l 8.8 (5.7–19.5) 6.2 (3.1–10.2) 11.9 (8.8–13.7) 11.0 (8.2– 17.0) −2.7 (−6.1 to −0.1) 0.6 (−2.1 to 5.4) 0.01

> 7.8 mmol/l 48.8 (40.8– 70.0) 49.0 (36.6–58.1) 50.3 (43.9–58.6) 47.1 (37.4– 53.5) −3.4 (−10.5 to 1.4) −5.9 (−15.0 to 5.6) 0.57

>10.0 mmol/L 33.3 (25.2–49.9) 26.7 (16.9–37.4) 35.0 (21.9–38.7) 28.0 (18.0– 32.1) −8.6 (−13.0 to−1.1) −7.0 (−16.9 to 1.7) 0.71

>15 mmol/l 10.0 (1.6–20.4) 4.2 (1.2–9.7) 5.9 (2.7–9.2) 2.6 (1.2–5.1) −4.9 (−8.6 to −0.7) −3.1 (−5.3 to −0.4) 0.48

3.9-7.8 mmol/l 31.7 (24.1–43.8) 43.7 (34.7–52.3) 34.8 (30.2–44.1) 40.4 (34.7– 45.3) 10.6 (3.3 to 14.4) 5.9 (−2.4 to 9.0) 0.15

3.9 -10 mmol/l 50.2 (40.8– 66.5) 65.9 (53.–74.8) 54.1 (47.5–64.5) 60.0 (54.5– 67.8) 12.7 (7.2 to 15.8) 5.3 (1.1 to 11.7) 0.05

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 344 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Percentage time within defined glucose range (median percentage time (and IQR)) overnight (22.00–07.00)

Continuous glucose monitoring Flash glucose monitoring Median change from baseline (n = 19) (n = 20) (95% CI) Median (IQR) Median (IQR) Baseline Endpoint Baseline Endpoint Continuous Flash glucose P value (−2 to 0 weeks) (4 to 8 weeks) (−2 to 0 weeks) (4 to 8 weeks) glucose monitoring monitoring

< 2.8 mmol/l 4.1 (0.5–13.1) 0.5 (0.0–2.3) 5.1 (1.7–8.2) 6.1 (3.6–10.8) −2.7 (−6.1 to−0.5) 1.2 (−1.5 to 4.8) 0.001

< 3.3mmol/l 6.1 (2.9–17.3) 1.4 (0.4–5.7) 8.3 (2.7–12.2) 8.8 (6.0–16.7) −4.4 (−6.9 to 0.0) 1.6 (−1.6 to 6.0) < 0.001

< 3.5 mmol/l 7.0 (3.9–18.7) 2.7 (0.6–7.1) 9.5 (4.0–13.6) 10.2 (7.7– 18.9) −5.2 (−6.4 to 0.0) 2.2 (−0.9 to 6.6) 0.001

< 3.9 mmol/l 9.6 (5.2–20.7) 5.5 (1.5–10.5) 13.0 (6.7– 17.1) 12.6 (10.1– 22.0) −4.8 (−9.5 to −0.7) 3.1 (−2.7 to 6.8) 0.004

> 7.8 mmol/l 51.9 (36.9– 68.9) 52.4 (35.5– 63.3) 49.4 (34.4– 64.6) 43.4 (30.5– 60.7) −1.9 (−11.1 to 9.4) −7.0 (−12.7 to 3.4) 0.41

>10.0 mmol/l 33.8 (13.5– 53.1) 26.7 (11.2– 44.6) 30.0 (16.6– 44.9) 24.0 (13.5– 32.5) −4.4 (−15.4 to 9.5) −9.9 (−15.7 to −4.3) 0.36

>15 mmol/l 8.5 (1.0–13.8) 5.1 (0.5–8.3) 5.4 (2.1–9.8) 1.1 (0.7–4.6) −4.1 (−6.1 to 0.0) −2.9 (−6.1 to −1.4) 0.70

3.9-7.8 mmol/l 31.8 (21.8– 46.6) 42.8 (29.2– 49.5) 37.6 (25.2– 46.7) 41.4 (30.5– 46.6) 13.0 (−4.1 to 19.6) 4.1 (−1.0 to 11.0) 0.16

3.9 -10 mmol/l 47.8 (39.2– 65.9) 62.6 (51.7– 72.7) 53.9 (42.3– 67.5) 59.5 (52.1–64.2) 14.1 (−1.5 to 23.7) 5.2 (0.7 to 11.6) 0.20

Net effect of CGM relative to flash glucose monitoring (median) Reduction of 4.3% in percentage time < 3.3 mmol/l

Continuous glucose monitoring Flash glucose monitoring Median change from baseline (n = 19) (n = 20) (95% CI) Median (IQR) Median (IQR) Baseline Endpoint Baseline Endpoint Continuous glucose Flash glucose P - value (at 8 weeks) (at 8 weeks) monitoring monitoring

LBGI 7.0 (5.4–12.3) 5.3 (3.2–6.3) 8.5 (5.9–9.8) 9.1 (7.2–10.7) −3.5 (−4.9 to −0.9) 0.9 (−0.2 to 3.0) < 0.001

Gold score 5 (5–6) 4.5 (3.0 - 5.0) 5 (4–5) 5.0 (3.5–6.0) 0.0 (−1.0 to 0.0) 0.0 (−0.8 to 0.0) 0.23

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 345 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of participats with 18/20 (90%) 12/20 (60%) 17/20 (85%) 12/20 (60%) Gold score ≥ 4

HbA1c (mmol/mol) 57 (49–62) 54 (45–61) 55 (48–65) 51 (48–59) −1.5 (−8.6 to −1.0) −4.5 (−5.8 to 0.0) 0.91

HbA1c (%) 7.4 (6.6–7.8) 7.1 (6.3–7.7) 7.2 (6.5–8.1) 6.8 (6.5–7.5) −0.15 (−0.8 to −0.05) −0.35 (−0.6 to 0.0)

HFS total score 59.5 (37.0– 78.0) 49.5 (28.0–74.0) 42.5 (32.0– 56.5) 42.0 (28.5–65.5) −6.5 (−10.8 to −2.2) −2.0 (−3.8 to 2.8) 0.02

HFS-Behaviour subscore 21.0 (13.5– 31.0) 20.0 (10.5–26.0) 17.5 (12.5– 24.5) 15.0 (11.5–25.5) −2.0 (−3.8 to −0.1) −0.5 (−3.0 to 1.8) 0.36

HFS- Worry subscore 40.5 (24.0– 52.5) 30.0 (17.5–44.0) 27.5 (18.0–34.5) 31.0 (15.5–46.0) −4.5 (−7.8 to −0.1) 0.5 (−3.0 to 2.8) 0.02

PAID score 31.0 (13.5– 45.5) 28.5 (17.5–43.0) 19.0 (14.0– 46.0) 22.0 (11.5–40.0) −1.0 (−5.7 to 4.8) −1 (−5.0 to 2.0) 0.82

Mortality -

Incidence of hypoglycaemia No episodes of severe hypoglycaemia were reported during the 8-week intervention phase in either group. level 1 hypoglycaemia - level 2 hypoglycaemia - level 3 hypoglycaemia -

Incidence of hyperglycaemia -

Quality of life -

Patient satisfaction -

Incidence of diabetic ketoacidosis -

Incidence of hyperosmolar, hyperglycaemic coma -

Resource utilization related to DM -

Number of visits to emergency room -

Number of visits to primary care -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 346 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of visits to specialists -

Number of hospitalizations -

Number of daily finger-sticks tests -

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Compliance/adherence -

Percentage of time using CGM -

Number of sensor scans per day (in FGM system) -

Safety results - n (%) 95% CI

Any AEs -

Serious AE (SAE) -

Most frequent AEs (by arms) -

Most frequent SEAs (by arms) -

Death as SAE -

Withdrawals due AEs -

Costs (only for national assessment) -

Author Disclosure (Conflict of interest)

Oliver N NO has received honoraria for speaking and advisory board participation from Abbott Diabetes, Dexcom, Medtronic Diabetes and Roche Diabetes.

RR: relative risk ITT: intention to treat PP: per protocol

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 347 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Risk of Bias

Study (Author, year): Reddy et al. 2018 NCT03028220

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Participants were randomly assigned to CGM (Dexcom G5) or flash glucose monitoring (Abbott Freestyle Libre®) in a 1 :1 ratio using an online randomization tool (www.sealedenvelope.com).

Randomization was stratified by HbA1c (< 58 mmol/mol and ≥ 58 mmol/mol).

Allocation concealment (Selection bias) Unclear Not written

Blinding of participants (Performance bias) High A randomized, non-masked parallel group study

Blinding of personnel (Performance bias) High A randomized, non-masked parallel group study

Blinding of outcome assessment (Detection bias) Unclear A randomized, non-masked parallel group study, not decribed

Incomplete outcome data (Attrition bias) Low Intention to treat analysis, all 40 randomized participants completed the intervention period

Selective reporting (Reporting bias) High Bias due to selective outcome reporting: AEs not mentioned as aim or outcome nor reported; not pre-specified primary or secondary endpoints; so the study does not report the results for a key outcome that could reasonably be expected for a study of its nature.

Other source of bias (Other bias) High Small sample size (40 participants in total); no significant differences in baseline characteristics between the groups, short follow uo (8 weeks); CGM and FGM data, where accuracy may not be equivalent, so glucose outcomes may not be directly comparable; stratification at randomization was based on HbA1c alone and does not consider other factors such as age, gender and diabetes duration; the inclusion of participants with severe hypoglycaemia and a Gold score of < 4 makes the study population heterogeneous as those five participants with a Gold score of < 4 may not have impaired awareness of hypoglycaemia; but these participants belong to a high-risk population and were randomized in an equal distribution (two in the CGM group and five in the flash glucose monitoring group); A new consensus for reporting hypoglycaemia in studies as < 3.0 mmol/l was recently recommended by The International Hypoglycaemia Study Group [25], but this was not the case at the time of study design. The percentage time spent at glucose < 3.0 mmol/l was therefore not a predetermined study outcome in this study, but when analysed post hoc the baseline vs. endpoint values were (3.1 vs. 1.5) and (4.7 vs. 5.0) in the CGM group and flash glucose monitoring group respectively and there was a significant difference in median change from baseline between groups (P = 0.004), suggesting benefit with CGM.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 348 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

FGM vs SMBG EVIDENCE TABLES

Main characteristics of studies included

FGM

MDII and CSII patients Bolinder RCT T1 MDII or FGM vs SMBG Time spent in hypoglycaemia (<3.9 mmol/L [<70 mg/dL]) for the 14 days preceeding the end of Clinical effectiveness and safety domains 2016 CSII the 6 month study period (days 194–208); time spent in hyperglycaemia; time spent in range; Impact trial 24 weeks Abbott HbA1c; System utilisation; Number of sensor scans per day; Frequency of glucose finger- 241/239 Freestyle sticks tests; Emergency room visits or admissions and non-protocol related additional clinic EU Libre® time; Patient-recorded outcome measures; Adverse events and sensor insertion-site ≥ 18 years symptoms NCT022326 98 Haak 2017 RCT T2 MDII or FGM vs SMBG Difference in HbA1c at 6 months; Prespecified secondary endpoints were subgroup analyses Clinical effectiveness and safety domains Replace CSII by age (less than and 65 years or older), sensor-derived glycemic measures from baseline to trial Abbott days 194–208 (Sensor-derived glycemic measures comprised number and duration of 24 weeks 224/201 Freestyle hypoglycemic events (<3.9 mmol/L[70 mg/dL], and <3.1 mmol/L [55 mg/dL]); time in range EU Libre® (3.9–10.0 mmol/L [70–180 mg/dL]), number and duration of hyperglycemic events ([10.0 ≥ 18 years mmol/L [180 mg/dL], and [13.3mmol/L [240 mg/dL]), mean glucose, and glucose variability NCT020821 measures) frequency of glucose finger-sticks and sensor scans per day during the study 84 period,system utilization for days 15–208 (defined as the percentage of data collected, assuming continuous device wear), and change in total daily dose of insulin, body mass index (BMI),weight, and participant questionnaire responses. Secondary endpoints reported in the clinical study report and not here, include change in HbA1c from baseline to day 105, proportion of participants with reduction in HbA1c of ≥5.5 mmol/mol (0.5%) from baseline, or achieving HbA1c ≤ 58 mmol/mol (7.5%), post-prandial hyperglycaemia, blood pressure, lipid levels, HCP questionnaire responses, emergency room visits, hospital admissions, additional clinic time, lancet use and non-insulin medication use. AEs

MDII patients

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 349 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Oskarsson RCT T1 MDII FGM vs SMBG Mean time in hypoglycaemia; sensor scanning frequency; number of self-monitored blood Clinical effectiveness and safety domains 2018 glucose tests; treatment satisfaction; perception of hypo/hyperglycaemia; AEs 167/161 Abbott MDII 24 weeks Freestyle ≥ 18 years subgroup Libre® of Impact trial EU NCT022326 98

Author, year, reference Bolinder et al. 2016

Bolinder J, Antuna R, Geelhoed-Duijvestijn P, Kröger J, Weitgasser R. Novel glucose-sensing technology and hypoglycaemia in type 1 diabetes: a multicentre, non-masked, randomised controlled trial. Lancet. 2016 Nov 5; 388(10057):2254- 2263. Epub 2016 Sep 12. doi: 10.1016/S0140-6736(16)31535-5 http://dx.doi.org/10.1016/S0140-6736(16)31535-5

Study title/objectives Novel glucose-sensing technology and hypoglycaemia in type 1 diabetes: a multicentre, non-masked, randomised controlled trial

Objectives: to assess whether a factory-calibrated, sensor-based, fl ash glucose-monitoring system compared with self- monitored glucose testing reduces exposure to hypoglycaemia in patients with type 1 diabetes.

Study characteristics

Study design Prospective, non-masked, randomised controlled study

Study Registration number ClinicalTrials.gov, number NCT02232698

Country of recruitment Sweden, Austria, Germany, Spain, Netherlands

Centre (single or multicentre) Multicentre [23 European diabetes centres (three in Sweden, six in Austria, fi ve in Germany, three in Spain, and six in the Netherlands)]

Ethics Committee Approval Approval was given by the appropriate competent authority in each country. All participating centres gave ethics approval

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 350 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

before the study.

Participants gave written informed consent.

Sponsor Abbott Diabetes Care sponsored the study. Abbott Diabetes Care provided study devices and all materials.

The sponsor designed the study protocol in collaboration with the principal investigator in each country and provided all the study materials. The sponsor was involved in collecting data and reporting results, but wasnot involved in the authors’ interpretation or in writing text. The sponsor also funded medical writing services and gave approval to submit for publication. The corresponding author had full access to all the data in the study and, together with all authors, had final responsibility for the decision to submit for publication.

Study period (study start, study end) Sept 4, 2014 - Feb 12, 2015

Duration of follow-up (days) 3 months and 6 months • Participants aged 18 years or older Inclusion criteria • Diagnosis: type 1 diabetes for 5 years or longer • On current insulin regimen for at least 3 months before study entry • Screening HbA1c concentration of 58 mmol/mol (7.5%) or lower • Reported self-monitoring of blood glucose levels on a regular basis (equivalent to ≥3 times a day) for 2 months or more before study entry • Considered by the investigator to be technically capable of using the flash sensor-based glucose monitoring system • Current diagnosis of hypoglycaemia unawareness Exclusion criteria • Diabetic ketoacidosis or myocardial infarction in the preceding 6 months • Known allergy to medical-grade adhesives • Usage of continuous glucose monitoring within the preceding 4 months • Current use of sensor-augmented pump therapy • Pregnancy or planning pregnancy • Receiving oral steroid therapy for any disorders.

Patient characteristics Intervention (n=119) Control (n=120)

Age of patients (years) 42 (33–51) 45 (33–57)

Sex Men: 77 (65%); Female: 42 (35%) Men: 59(49%); Female: 61 (51%)

BMI (kg/m2) 25.2 (±3.6) 24.8 (±3.5)

Diagnosis (Type I, II or gestational) Type I Type I

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 351 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Comorbidities (i.e. obesity…) - -

Time since diagnosis with DM (years) 20 (13–27) 20 (12–32)

Insulin treatment (CSII or MDII) • Multiple daily injections 81 (68%) 80 (67%) • Continuous subcutaneous insulin infusion 38 (32%) 40 (33%)

Pregnancy (yes, no) No No

Special subgroup pf patient (i.e., hypoglycaemia fear…) - -

Intervention

Type of medical device (CGM or FGM) FGM (supported by SMBG)

Name (Description) of medical device FreeStyle Libre®; sensor-based fl ash glucose monitoring system (Abbott Diabetes Care, Witney,Oxon, UK).

Glucose management was supported by self-monitoring of blood glucose, using the strip port built into the reader and compatible test strips (Abbott Diabetes Care, Witney, Oxon, UK).

Adjunctive or non-adjunctive Non-adjuctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device FreeStyle Lite meter and test strips (Abbott Diabetes Care, Witney, Oxon, UK) for self-monitoring of glucose concentrations

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible (con- - nected) withspecific CGM systems (Yes, No)

Outcomes

Primary Time spent in hypoglycaemia (<3.9 mmol/L [<70 mg/dL]) for the 14 days preceeding the end of the 6 month study period (days 194–208).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 352 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Secondary Sensor-derived glycaemic measures at days 194–208: • number and duration of hypoglycaemic episodes (sensor glucose <3.9 mmol/L in 24 h, by day [0600–2300 h], and night [2300–0600 h]; <3.1 mmol/L in 24 h, and <2.2 mmol/L in 24 h [<70 mg/dL, <55 mg/dL, and <40 mg/dL, respectively]; an episode was defi ned as at least two consecutive readings, at 15 min intervals, outside the predefined glucose range, the end of an episode was one reading at or higher than the threshold) • time with glucose in range 3.9–10.0 mmol/L (70–180 mg/dL) • number and duration of hyperglycaemic episodes (>10.0 mmol/L and >13.3 mmol/L [>180 mg/dL and >240 mg/dL, respectively]) • glucose variability measurements

Day 208 HbA1c concentrations

Change in total daily dose of insulin from day 1 to day 208

System utilisation for days 15–208 (defi ned as the percentage of data collected, assuming continuous device wear)

Frequency of glucose finger-sticks tests

Number of sensor scans per day

Additional outcomes Proportion of participants who achieve time spent in hypoglycaemia (<3.9 mmol/L; <70 mg/dL) ≤1 h/day

Number of events of symptomatic hypoglycaemia

Post prandial hyperglycaemia (>10.0 mmol/L, 180 mg/dL)

Prandial to basal insulin ratio

Number of participants changing from once daily to twice daily basal insulin

Body weight and body-mass index (BMI)

Fasting cholesterol and triglycerides

Blood pressure

Emergency room visits or admissions and non-protocol related additional clinic time

Medication usage (non-insulin related, including glucagon, self-reported from event diary)

Patient-recorded outcome measures:

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 353 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• HFS • DTSQ • DDS • DQoL

Adverse events and sensor insertion-site symptoms

Number of episodes of diabetic ketoacidosis

Number of severe hypoglycaemia events (requiring third-party assistance)

Flow of patients

No of patients enrolled 328

No of randomized 241

Allocated per arms Intervention group: 120 Control group: 121

Received int. per arms Intervention group: 119 (1 excluded due to pregnancy) Control group:120 (1 excluded due to pregnancy)

Lost to follow-up per arms Intervention group: 9 withdrew or were excluded Control group: 19 withdrew or were excluded • 1 met exclusion criteria • 4 due to non-compliance with study device • 7 had device-associated symptoms • 1 met exclusion criteria • 1 due to non-compliance with study device • 3 because allocated to control group • 11 for other reasons

No of analysed per arm Intervention group: 110 Control group:101

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) Per protocol

Results

Effectiveness results n (%) 95% CI

Baseline Study end Difference in adjusted Difference in interven- P value means in intervention vs tion vs control (%) Data in parentheses are SDs, apart control from when given with adjusted means

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 354 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

where they are SEs.. *Post-hoc end- Intervention Control Intervention Control point. (n=119) (n=119) (n=119) (n=119)

HbA1c (mmol/mol) 50.7 (5.7) 50.6 (7.0) 52.4 (7.2) 52.4 (7.2) 0.0 (0.65) NA 0.9543

HbA1c (%) 6.79 (0.52) 6.78 (0.64) 6.94 (0.65) 6.95 (0.66) 0.00 (0.059) NA 0.9556

Time withglucose 3.9–10.0 mmol/L 15.0 (2.5) 14.8 (2.8) 15.8 (2.9) 14.6 (2.9) 1.0 (0.30) NA 0.0006 (70–180 mg/dL) in h

Glucose<3.9 mmol/L (70 mg/dL) within 24 h • Events 1.81 (0.90) 1.67 (0.80) 1.32 (0.81) 1.69 (0.83) −0.45 (0.089) −25.8% <0.0001 • Time in h 3.38 (2.31) 3.44 (2.62) 2.03 (1.93) 3.27 (2.58) −1.24 (0.239) −38.0% <0.0001 • AUC, area under the curve 53.42 (43.46) 58.34 (57.22) 28.58 (31.15) 54.67 (60.08) –25.14 (5.32) –46.7 <0.0001 (h×mg/dL)

Glucose<3.9 mmol/L (70 mg/dL) at night (2300–0600 h) within 7 h • Events 0.47 (0.32) 0.46 (0.29) 0.27 (0.23) 0.40 (0.29) −0.14 (0.029) −33.2% <0.0001 • Time in h 1.32 (1.07) 1.48 (1.29) 0.68 (0.97) 1.23 (1.10) −0.47 (0.118) −39.8% <0.0001

Glucose<3.1 mmol/L (55 mg/dL) within 24 h • Events 0.96 (0.65) 0.92 (0.73) 0.56 (0.55) 0.92 (0.74) −0.38 (0.074) −41.3% <0.0001 • Time in h 1.59 (1.42) 1.77 (1.86) 0.80 (0.96) 1.65 (1.97) −0.82 (0.175) −50.3% <0.0001 • AUC (h×mg/dL) 16.04 (17.46) 18.94 (23.22) 7.59 (10.25) 17.69 (26.34) −9.67 (2.29) −56.1% <0.0001

Glucose<3.1 mmol/L (55 mg/dL) at night (2300–0600 h) within 7 h • Events 0.34 (0.27) 0.36 (0.34) 0.19 (0.24) 0.30 (0.28) −0.11 (0.03) −34.9% 0.0005

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 355 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• Time in h 0.62 (0.60) 0.75 (0.83) 0.31 (0.43) 0.66 (0.080) −0.32 (0.07) −48.9% <0.0001

Glucose<2.5 mmol/L (45 mg/dL) within 24 h* • Events 0.56 (0.52) 0.59 (0.60) 0.29 (0.36) 0.56 (0.59) −0·26 (0.06) −48.5% <0.0001 • Time in h 0.85 (1.03) 1.04 (1.36) 0.38 (0.58) 0.96 (1.57) −0.55 (0.14) −59.5% <0.0001 • AUC (h×mg/dL) 3.99 (5.36) 5.00 (7.10) 1.74 (2.91) 4.73 (8.66) –2.88 (0.75) –63.1 0.0002

Glucose<2.5 mmol/L (45 mg/dL) at night (2300–0600 h) within 7 h* • Events 0.23 (0.23) 0.27 (0.31) 0.11 (0.16) 0.21 (0.22) −0.09 (0.02) −44.9% <0.0001 • Time in h 0.36 (0.44) 0.48 (0.66) 0.15 (0.25) 0.43 (0.65) −0.25 (0.06) −60.4% <0.0001

Glucose <2.2 mmol/L (40 mg/dL) within 24 h • Events 0.39 (0.43) 0.44 (0.51) 0.19 (0.29) 0.43 (0.55) −0.22 (0.050) −55.0% <0.0001 • Time in h 0.59 (0.85) 0.75 (1.11) 0.26 (0.47) 0.73 (1.41) −0.46 (0.122) −65.3% 0.0003

Glucose>13.3 mmol/L (240 mg/dL) within 24 h • Time in h 1.85 (1.44) 1.91 (1.70) 1.67 (1.36) 2.06 (1.61) −0.37 (0.163) −19.1% 0.0247

Glucose variability • BGRI (lood glucose risk in- 8.2 (2.3) 8.3 (2.7) 7.3 (2.4) 8.4 (2.6) −0.9 (0.26) ·· 0.0004 dex) • CV (coefficient of variation) 43.0 (7.0) 42.5 (6.6) 37.6 (5.7) 41.8 (6.8) −4.4 (0.62) ·· <0.0001 glucose (%) • LBGI (low blood glucose in- 2.7 (1.5) 2.7 (1.7) 1.8 (1.4) 2.6 (1.7) −0.8 (0.16) ·· <0.0001 dex) • MAGE(mean amplitude of 142 (29) 144 (31) 132 (27) 141 (31) −8 (3.0) ·· 0.0055 glycaemic excursions)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 356 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(mg/dL; average) • Mean glucose (mg/dL) 141 (19) 142 (23) 146 (20) 143 (23) 3 (2.3) ·· 0.1479 • Standard deviationofglu- 60.6 (12.6) 60.1 (12.9) 55.0 (10.9) 59.7 (13.8) −5.0 (1.16) ·· <0.0001 cose(mg/dL)

CONGA (continuous overall net glycaemic action) • 2 h (mg/dL) 56 (13) 56 (14) 49 (12) 58 (13) −9 (1.3) ·· <0.0001 • 6 h (mg/dL) 71 (25) 69 (26) 61 (25) 72 (28) −12 (3.4) ·· 0.0004

Mortality -

Quality of life Neither group in the full analysis set was significantly favoured (−0.08 [0.039]; p=0.0524) but

was significantly improved in the per-protocol set

Patient satisfaction Patient satisfaction with treatment was signifycantly improved for intervention compared with control (adjusted between- group difference −0.24 [SE 0.049]; p<0.0001).

Total treatment satisfaction Significantly improved in the intervention group compared with the control group (6.1 [0.84]; p<0·0001)

Perceived frequency of hyperglycaemia Significantly improved in the intervention group compared with the control group(−1.0 [0.22]; p<0·0001)

Diabetes distress No differences between the study groups (−0.03 [SE 0.089]; p=0.7634)

Hypoglycaemia fear behaviour No differences between the study groups (0.0 [0.72]; p=0.9834)

Worry scores No differences between the study groups (−1.2 [1.48]; p=0.4154)

Incidence of diabetic ketoacidosis -

Incidence of hyperosmolar, hyperglycaemic coma -

Resource utilization related to DM (System utilisation, defined as the Intervention group (n=112) = 92.8% (SD 7.3) percentage of data collected, assuming continuous device wear for 6 months by the intervention group)

Number of visits to emergency room -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 357 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of visits to primary care -

Number of visits to specialists -

Number of hospitalizations -

Number of daily finger-sticks tests (mean number of self-monitored blood Intervention group Control group glucose tests performed per day) Baseline: 5.5 (SD 2.0) tests per day Baseline: 5.8 (SD 1.7) tests per day

Treatment phase of the trial: 0.5 (0.7) tests (clinically At 6 months: 5.6 (2.2) tests per day equates to one self-monitoring of blood glucose test every 2–5 days)

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Compliance/adherence -

Percentage of time using CGM -

Number of sensor scans per day (in FGM system) (mean) Intervention group only: 15.1 (SD 6.9)

Change in total insulin dose (mean) Intervention group Control group P value • MDI −2.7 units (SD 7.3) −3.0 units (6.4) p=0.7973 • CSII −0.5 units (SD 5.8) −0.7 units(3.4) p=0.5860

Total daily doses of insulin No differences between the study groups

Bolus/basal insulin ratios No differences between the study groups

Safety results Intervention group Control group n (%) 95% CI (full analysis set and two participants that became preg- (n=120) (n=121) nant.)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 358 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Any AEs (number) 138 138

Serious AE (SAE) (number) 5 5

Most frequent AEs (by arms) - -

Most frequent SEAs(by arms) - -

Death as SAE - -

Withdrawals due AEs 6 (5%) 1 (<1%)‡

‡Due to severe hypoglycaemia.

Participants with adverse or serious adverse events 63 (53%) 61 (50%)

Participants with serious adverse events 5 (4%) 4 (3%)

Participants with hypoglycaemic serious adverse events* 2 (2%) 3 (2%)

*A hypoglycaemic serious adverse event was reported during the baseline phase

Number of hypoglycaemic serious adverse events* 2 4

*A hypoglycaemic serious adverse event was reported during the baseline phase

Participants with hypoglycaemic adverse events 0 2 (2%)

Number of hypoglycaemic adverse events 0 3

Participants with device-related adverse events† 10 (8%) 0

†Device-related adverse events were all related to wearing the sensor: four participants with allergy (one severe, three moderate); one with itching (mild); one with rash (mild); four with insertion-site symptom (severe); two with erythema (one severe, one mild); and one with oedema (moderate); all resolved

Number of device-related adverse events 13 0

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 359 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Sensor insertion-site signs and symptoms Total number (both groups): 248

Participants with sensor insertion-site signs and symptoms 65 participants across both groups • pain (38) Number of signs expected due to sensor insertion • bleeding (25) • oedema (8) • induration (5) • bruising (5) • erythema (85) Number of signs associated with sensor wear • itching (51) • rash (31)

Costs (only for national assessment) -

Author Disclosure (Conflict of interest)

Jan Bolinder JB has received honoraria for consulting or lecture fees from Abbott Diabetes Care, AstraZeneca, Insulet Corporation, Integrity Applications, and Sanofi -Aventis.

Ramiro Antuna RA has received consulting and speaking honoraria from Abbot Diabetes Care.

Petronella Geelhoed-Duijvestijn PG-D has received lecture honoraria, and serves on advisory boards for Abbott Diabetes Care, Medtronic, and Novo Nordisk.

Jens Kröge JK has received lecture honoraria from Abbott Diabetes Care, AstraZeneca, Bayer Vital, Boehringer Ingelheim, Boehringer- Mannhein, GlaxoSmithKline, Medtronic, Merck, Sharp & Dohme, Novo Nordisk, Lilly, Roche, and Sanofi -Aventis.

JK serves on advisory boards for Abbott Diabetes Care, AstraZeneca, Merck, Sharp & Dohme, Novo-Nordisk and Lilly.

Raimund Weitgasser RW received lecture honoraria and serves on advisory boards for Abbott Diabetes Care, Allergan, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Janssen Cilag, Medtronic, Merck, Sharp & Dohme, Novartis, Novo Nordisk, Pfzer, Roche Diagnostics, Sanofi , Schulke, Servier, and Takeda, and has received unrestricted study grants from Eli Lilly, Medtronic, Novo Nordisk, and Sanofi.

RR: relative risk ITT: intention to treat PP: per protocol

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 360 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Risk of Bias

Study (Author, year): Bolinder et al. 2016 NCT02232698

Judgement (Low, Support for judgement Unclear, High)

by central interactive web response system (IWRS) using the biased-coin minimisation method; study centre and type of insulin administration were prognostic factors.

Allocation concealment (Selection bias) Low Please see above

Blinding of participants (Performance bias) High Participants, investigators, and study staff were not masked to group allocation

Blinding of personnel (Performance bias) High Participants, investigators, and study staff were not masked to group allocation

Blinding of outcome assessment (Detection bias) Low All sensor glucose data were blinded for both participants and investigators.

Incomplete outcome data (Attrition bias) Low The primary endpoint and all secondary endpoints were assessed in the full analysis set, which included all randomised participants apart from those who had a positive

pregnancy test during the study period. Safety outcomes were analysed in all participants who were enrolled. Missing values were imputed by last observation carried forward. The full analysis set included 239 randomised participants; one woman from each group was excluded due to pregnancy.

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear Funding; In absolute terms, there were fewer serious adverse events and adverse events associated with hypoglycaemia in the intervention (two) than in the control group (seven). It should be noted, however, the study was not powered to detect any statistically significant differences in the incidence of adverse events associated with hypoglycaemia. There are a number of study limitations that might affect the generalisability of our findings. For individuals diagnosed with severe hypoglycaemia unawareness, this technology might not be ideal and predictive or low-threshold glucose insulin-suspend technology might be preferable. Inclu- sion criteria of well controlled diabetes (HbA1c <7・5%) implies that participants were highly motivated and successful in their self-management compared with other populations; although a concern for this group is

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 361 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

susceptibility to hypoglycaemia. The relative proportion of continuous insulin infusion users in the trial was higher than usually seen in most European type 1 diabetes populations, and only adults were enrolled. Fu- ture studies are needed to assess the effectiveness of this novel glucose monitoring system in younger age groups in addition to less well controlled and less motivated people with type 1 diabetes. All participants experienced periods of sensor wear; consequently, the intervention was not masked to participants, investigators, and study staff . As such, treatment decisions and assessment were based on the same sensor glucose values. This is a common limitation in glucose technology studies and it is recognised that there is no practical alternative to this approach. The trial took place over a period of 6 months and therefore there are limitations around expected compliance to device use over a longer period. No adjustment was made for multiple testing of secondary endpoints. Many of the endpoints, particularly those derived from sensor glucose values, are highly inter-related and should not be considered in isolation.

Author, year, reference Haak T, Hanaire H, Ajjan R, Hermanns N, Riveline JP, Rayman G. Flash Glucose-Sensing Technology as a Replacement for Blood Glucose Monitoring for the Management of Insulin-Treated Type 2 Diabetes: a Multicenter, Open-Label Randomized Controlled Trial. Diabetes therapy : research, treatment and education of diabetes and related disorders. 2017;8(1):55-73.

Study title/objectives An Evaluation of a Novel Glucose Sensing Technology in Type 2 Diabetes (REPLACE)

Study characteristics

Study design 6-month, prospective, open-label, non-masked, two-arm randomized controlled study

Randomization – interactive web response system (IWRS) using biased-coin

minimization, with study centre and insulin administration as prognostic factors

Study Registration number ClinicalTrials.gov NCT02082184

Country of recruitment France(8 sites), Germany(10), United Kingdom(8)

Centre (single or multicentre) 26 European diabetes centres

Ethics Committee Approval Approval was given by the appropriate competent authorities in each country

Sponsor Abbott Diabetes Care

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 362 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Study period (study start, study end) March 2014- December 2015

Duration of follow-up (days)

Inclusion criteria We enrolled participants aged 18 years or older with

type 2 diabetes treated with insulin for at least 6 months and

on their current regimen (prandial only or prandial and basal intensive insulin therapy or CSII therapy) for 3 months or more,

an HbA1c level 58–108 mmol/mol (7.5–12.0%),

self-reported regular blood glucose testing (more than 10/week for at least 2 months prior to study entry), and

were considered by the investigator to be technically capable of using the flash sensor-based glucose monitoring system.

Exclusion criteria Participants were not included if:

they had any other insulin regimen to that described above;

a total daily dose of insulin ≥1.75 units/kg on study entry;

had severe hypoglycaemia (requiring third-party assistance),

diabetic ketoacidosis, or

hyperosmolar-hyperglycaemic state in the preceding 6 months;

known allergy to medical-grade adhesives;

used continuous glucose monitoring within the previous 4 months;

were pregnant or planning pregnancy;

were receiving steroid therapy for any condition;

or were considered by the investigator to be unsuitable to participate

Patient characteristics

Age of patients 18 years or older / Intervention (age) 59.0 ± 9.9 (33, 81) Control ( age) 59.5 ± 11.0 (22, 80)

Sex Both

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 363 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

BMI Intervention 33.1 ± 6.2 (18.8, 54.1) Control 33.3 ± 5.5 (23.7, 52.4)

There were no changes in body weight (p = 0.2496) or BMI (p = 0.2668) from baseline

for either group.

Diagnosis (Type I, II or gestational) Type 2 diabetes

Comorbidities (i.e. obesity…)

Time since diagnosis with DM Interventional 17 ± 8 (2, 43) Control 18 ± 8 (4, 37)

Insulin treatment (CSII or MDII) MDII or CSII

95% of participants used an insulin pen device or syringe for intensive insulin therapy, with the remainder (5%) on CSII

Pregnancy (yes, no) No

Special subgroup pf patient (i.e., hypoglycaemia fear…) Please see secondary outcomes.

Intervention

Type of medical device (CGM or FGM) FGM

Name (Description) of medical device Sensor-based flash glucose monitoring system (FreeStyle Libre®; Abbott Diabetes Care, Witney, UK).

The small, single-use, factory-calibrated, on-body sensor utilizes wired enzyme technology (osmium mediator and glucose oxidase enzyme co-immobilized on an electrochemical sensor)

to continuously monitor interstitial glucose levels.

The sensor is worn on the back of the arm for up to 14 days and automatically stores

glucose data every 15 min. A real-time glucose level may be obtained as often as every minute

by scanning the sensor with the reader. A glucose trend arrow (indicating rate and

direction of change in glucose levels) and a graphical trace of glucose values for the

previous 8-h period are also displayed on the screen. Data are transferred by radio frequency

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 364 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

identification (RFID) from the sensor to the reader memory which stores historical sensor

data for 90 days.

Adjunctive or non-adjunctive Non-adjunctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devic- SMBG es)

Name (Description) of medical device Standard blood glucose device (Abbott Diabetes Care, Witney,UK)

Sensor integrated (Yes, No)

Sensor augmented (or enabled) insulin pump systems compatible (connected) with specific CGM systems (Yes, No)

Outcomes

Primary Primary outcome was difference in HbA1c at 6 months in the full analysis set

Secondary Prespecified secondary endpoints were subgroup analyses

by age (less than and 65 years or older),

sensor-derived glycemic measures from baseline to days 194–208 (Sensor-derived glycemic measures comprised number and duration of hypoglycemic events (<3.9 mmol/L

[70 mg/dL], and <3.1 mmol/L [55 mg/dL]); time in range (3.9–10.0 mmol/L [70–180 mg/

dL]), number and duration of hyperglycemic events ([10.0 mmol/L [180 mg/dL], and [13.3

mmol/L [240 mg/dL]), mean glucose, and glucose variability measures)

frequency of glucose finger-sticks and sensor scans per day during the study period,

system utilization for days 15–208 (defined as the percentage of data collected, assuming

continuous device wear), and

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 365 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

change in total daily dose of insulin,

body mass index (BMI),

weight,

and participant questionnaire responses.

Secondary endpoints reported in the clinical study report and not here, include

change in HbA1c from baseline to day 105,

proportion of participants with reduction in HbA1c of ≥5.5 mmol/mol (0.5%) from baseline, or

achieving HbA1c ≤ 58 mmol/mol (7.5%),

post-prandial hyperglycaemia,

blood pressure,

lipid levels,

HCP questionnaire responses,

emergency room visits,

hospital admissions,

additional clinic time,

lancet use and

non-insulin medication use.

Flow of patients

No of patients enrolled 302

No of randomized 224

Allocated per arms Interventional 149 / Control 75

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 366 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

withdrawal after randomization : Interventional 10 / Control

Received int. per arms 2 from control group

Lost to follow-up per arms Interventional 10 / Control 13 withdrew

No of analysed per arm Interventional 139 / Control 62 (on day 208)

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) This study was powered at 90% to detect a difference of 3.8 mmol/mol (0.35%) in HbA1c

between the intervention and control group at 6 months with a 5% significance level as per

guidance of the Food and Drug Administration and assuming SD for the change of 0.65.

Analysis of covariance was used to adjust for chance imbalances in baseline measurements between the treatment groups, adjusted means were then used to compare differences between the groups for the 6-month endpoints.

Glucose variability measured as coefficient of variation (CV) reduced by 2.26 ± 0.71%

mean ± SE for intervention participants compared with controls (p = 0.0017).

LBGI reduced by 0.3 ± 0.11 mean ± SE for intervention participants compared with

controls (p = 0.0029).

CONGA was reduced for intervention compared with controls by 3 ± 1.3 mg/dL mean ± SE at 2 h time interval (p = 0.0385), by 5 ± 2.2 at 4 h (p = 0.0133), and by 8 ± 3.0 at 6 h (p = 0.0046).

Changes in questionnaire responses were considered using analysis of covariance on baseline values and study center to compare scores from intervention with control group participants.

Confidence intervals were calculated for the group least-square mean of each measure and

the difference between group least-square means.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 367 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Results

Effectiveness results n (%) 95% CI

Mortality

Table 2 Glycemic and glucose variability measures

G!ycemnic measure Baseline mean (SD) Study end mean (SD) Difference in Difference p value adjusted Intervention Control Intervention Control (n = 149) (n = 75) (n = 149) (n = means in intervention 75) intervention vs control vs control (SE) (%) HbAl c (mmol/mol) 71.0 (11.1) 72.1 (10.7) 68.0 (9.0) 67.7 (12.4) 0.3 (1.25) N/A 0.8259 HbAl c (%) 8.65 (1.01) 8.75 (0.98) 8.37 (0.83) 8.34 (1.14) 0.03 (0.114) N/A 0.8222

Time with glucose 3.9-10.0 mmol/L (70-180 mg/dL) (h) 13.9 (4.5) 13.5 (5.2) 13.6 (4.6) 13.2 (4.9) 0.2 (0.58) 1.1 0.7925

Glucose <3.9 mmol/L (70 mg/dL) within 24 h Events 0.64 (0.63) 0.63 (0.66) 0,38 (0,45) 0,53 (0,59) -0.16 (0.065) -27,7 0,0164 Time (h) 1.30 (1.78) 1.08 (1.58) 0.59 (0.82) 0.99 (1.29) -0.47 (0.134) -43.1 0.0006 AUC (11 x mg/dL) 20.15 (35.21) 14.05 (26.35) 7.23 (12.35) 13.59 (22.31) -7.80 (2.20) -51.1 0.0005 Glucose <3.9 mmol/L (70 mg/dL) at night (23.00-06.00) within 7 h Events 0.25 (0.28) 0.27 (0.32) 0.14 (0.20) 0.27 (0.33) -0.12 (0.03) -44.9 0.0003 Time (h) 0.55 (0.84) 0,49 (0.71) 0,23 (0,43) 0.51 (0,72) -0.29 (0.08) -54.3 0.0001 Glucose <3.1 mmol/L (55 mg/dL) within 24 h Events 034 (0.50) 0.27 (0.44) 0.14 (0.24) 0.24 (0.36) -0.12 (0.037) -44.3 0.0017 Time (h) 0,59 (1.13) 0.38 (0.83) 0.19 (0,37) 0.37 (0,69) -0.22 (0.068) -53,1 0,0014

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 368 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

AUG (h x mg/dL) 6,02 (13,23) 3.40 (9.16) 1.64 (3,85) 3,66 (7,97) -2,51 (0.76) -60,3 0.0012 Glucose <3.1 mmol/L (55 mg/dL) at night (23.00-06.00) within 7 h Events 0.15 (0.23) 0.13 (0.20) 0.06 (0.13) 0.13 (0.21) -0.07 (0.02) -53.0 0.0012 Time (h) 0.27 (0.58) 0.18 (0.35) 0.09 (0.22) 0.19 (0.40) -0.12 (0.04) -58.1 0.0032 Glucose <2.5 mmol/L (45 mg/dL) within 24 h Events 0.19 (0.37) 0,13 (0.34) 0.06 (0.13) 0,11 (0,25) -0,06 (0,02) -48.8 0.0098 Time (h) 0.32 (0.74) 0,17 (0.54) 0,08 (0.21) 0,19 (0,45) -0,14 (0,04) -64.1 0.0013 AUC (h x mg/dL) 1.52 (3.77) 0.77 (2.63) 0.35 (1.11) 0.93 (2.23) -0.70 (0.22) -66.7 0.0015

Glycemic measure Baseline mean (SD) Study end mean (SD) Difference in Difference p value

Adjusted in intervention vs con- Intervention Control Intervention Control means in in- trol (%) (n = 149) (n = 149) (n = 75) tervention = 75) vs control (SE) Glucose <2,5 mmol/L (45 nigkiL) at night (23.00-06.00) within 7 h Events 0.08 (0.17) 0.06 (0.14) 0.03 (0.08) 0.07 (0.16) -0.04 (0.02) -57.8 0.0086 Time (h) 0.16 (0.42) 0.08 (0.23) 0.04 (0.12) 0.11 (0.28) -0.08 (0.03) -68.3 0.0041 Glucose <2.2 mmol/L (40 mg/dL) within 24 h Events 0.13 (0.30) 0.10 (0.30) 0.05 (0.13) 0.09 (0.22) -0.05 (0.02) -52.6 0.0199 Time (h) 0.22 (0.57) 0.12 (0.43) 0.05 (0.17) 0.14 (0.34) -0.10 (0.03) -66.7 0.0020 Time with glucose >10.0 mmol/L (180 mg/dL) (h) 8.8 (5.0) 9.4 (5.8) 9.8 (4.8) 9.8 (5.4) 0.3 (0.63) 3.5 0.5970 Time with glucose >13,3 mmol/L (240 mg/dL) (h) 3.1 (3.3) 3.9 (4.5) 3.5 (3.7) 3.9 (4.2) 0.1 (0.46) 2.1 0.8729 Glucose variability BGRI 9.5 (5.1) 10.4 (6.7) 9.9 (5.6) 10.5 (6.1) 0.0 (0.70) N/A 0.9431 CV glucose (%) 34.1 (7.2) 33.1 (6.7) 31.4 (6.2) 33.0 (8.0) -2.26 (0.71) N/A 0.0017 LBGI 1.1 (1.3) 1.0 (1.2) 0.6 (0.7) 0.9 (1.0) -0.3 (0.11) N/A 0.0029 MAGE (mg/dL; average) 128 (29) 131 (33) 125 (29) 131 (33) -4 (3.3) N/A 0.1909 Mean glucose (mg/dL) 165 (34) 171 (43) 174 (33) 174 (38) 3 (4.3) N/A 0.4236 Standard deviation of glucose (mg/dL) 56 (14) 56 (15) 54 (13) 56 (15) -1.67 (1.45) N/A 0.2538

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 369 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

CONGA 2 h (mg/dL) 49 (11) 50 (14) 47 (12) 51 (11) -3 (1.3) N/A 0.0385 CONGA 4 h (mg/dL) 61 (16) 61 (19) 57 (18) 64 (17) -5 (2.2) N/A 0.0133 CONGA 6 h (mg/dL) 63 (21) 62 (22) 58 (23) 65 (23) -8 (3.0) N/A 0.0046

AUC area under curve, BGRI blood glucose risk index, CV coefficient of variation, LBGI low blood glucose index, MAGE mean amplitude of glycemic excursions, CONGA continuous overall net glycemic action x hours

Quality of life All participants completed quality of life and patient-reported outcome questionnaires

prior to other study activities on day 1 and on day 194.

Patient-reported outcome and quality of life (QoL) measures were assessed using validated questionnaires: Diabetes Dis- tress Scale (DDS), Diabetes Quality of Life (DQoL), and Diabetes Treatment Satisfaction (DTSQs and DTSQc).

Patient satisfaction Treatment satisfaction was higher in intervention compared with controls (DTSQ

13.1 ± 0.50 (mean ± SE) and 9.0 ± 0.72, respectively; p\0.0001).

Total treatment satisfaction score for DTSQ (status versus change) was significantly

improved for intervention group participants (13.1 ± 0.50, mean ± SE) compared with

controls (9.0 ± 0.72), p\0.0001.

Satisfaction with treatment results using DQoL demonstrated significant improvement for the

intervention group (-0.2 ± 0.04, mean ± SE) versus the control group (0.0 ± 0.06),

p = 0.0259, for this element of the questionnaire.

There were no other significant differences observed in other aspects of DTSQ

and DQoL or for the DDS scales

User questionnaire results showed intervention participants agreed with positive

aspects of the device including use, comfort, and utilization of sensor glucose information

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 370 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Hypoglycaemia fear

Incidence of diabetic ketoacidosis 0

Incidence of hyperosmolar, hyperglycaemic coma 0

Resource utilization related to DM

Number of visits to emergency room

Number of visits to primary care

Number of visits to specialists

Number of hospitalizations

Number of daily finger-sticks tests Self-monitoring blood glucose frequency for intervention participants fell from 3.8 ± 1.4 tests/day mean ± SD (3.8 tests/day median) at baseline to 0.5 ± 1.1 (0.1 median) from the first unblinded sensor wear with full access to sensor glucose data (day 15–31), reducing further to 0.4 ± 1.0 tests/day (0.0 median) by study end (day 208). The overall blood glucose monitoring rate over 6 months was 0.3 ± 0.7, median 0.1

Self-monitoring of blood glucose frequency for control participants was 3.9 ± 1.5 test/day

(median 3.9) at baseline and this rate was maintained until study end [3.8 ± 1.9 (median

3.9), Fig. 3].

Control group participants <65 years performed less blood glucose monitoring tests (2.78 ± 1.08 test/day) than those ≥65 years (3.46 ± 0.94), p = 0.0247.

Number of calibration Factory calibrated

Need (Yes, with number or No) of re-calibration

Compliance/adherence

Percentage of time using CGM

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 371 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of sensor scans per day (in FGM system) During the treatment phase (day 15 onwards) average sensor-scanning frequency was 8.3 ± 4.4 (mean ± SD) times/day (median 6.8), i.e., double the frequency of blood glucose testing (Fig. 3). There was no significant difference in the number of scans performed by those \65 years and C65 years of age [8.1 ± 4.6 (median 6.8) and 8.5 ± 4.1 (median 6.9), respectively, p = 0.6627

Device use for the intervention group (n = 138) was 88.7 ± 9.2% (defined as the percentage of data collected, assuming continuous device wear for 6 months).

Safety results n (%) 95% CI

Any AEs Participants (%) with adverse or serious adverse events: Intervention 114 (77%) /Control 47 (63%) Number of adverse events (excluding serious events) Intervention 316/ Control 157

Participants (%) with hypoglycemic adverse events: Interventional 10 (7%)/ Control 7 (9%) 57 hypoglycaemia adverse events by 10 (7%) intervention and seven (9%) control participants No hypoglycemic adverse events were associated with the device.

Six (4.0%) intervention participants reported nine device-related adverse events (two severe, six moderate, and one mild). These were sensor-adhesive reactions, primarily treated with topical preparations. All were resolved at study exit.

In total, serious adverse or adverse events (n = 515) were experienced by 114 (76.5%) intervention and 47 (62.7%) control participants. There were no serious adverse events related to the device or study procedure. Three participants (one intervention, two controls) experienced an adverse event leading to withdrawal from the study; none were associated with the device.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 372 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Serious AE (SAE) Participants (%) with serious adverse events: Interventional 16 (11%) /Control 12 (16%) Number of serious adverse events: Interventional 20 / Control 22

2 serious adverse events for sensor-wear reactions

Four hypoglycaemia serious adverse events were experienced by four participants (three intervention and one control) None of the severe hypoglycemic episodes were associated with the device.

Forty-two serious events [16 (10.7%) intervention participants, 12 (16.0%) controls] were not device-related.

Most frequent AEs (by arms) There were 158 anticipated sensor insertion site symptoms observed for 41 (27.5%) intervention and 9 (12.0%) control participants. These symptoms were primarily

(63%) due to the sensor adhesive (erythema, itching, and rash) and resolved without medical

intervention.

Most frequent SEAs (by arms)

Death as SAE

Withdrawals due AEs

Costs (only for national assessment)

Author Disclosure (Conflict of interest) Thomas Haak reports personal fees from Abbott Diabetes Care outside the submitted work. Gerry Rayman reports personal fees from Abbott Diabetes Care outside the submitted work. Helene Hanaire reports personal fees from Abbott Diabetes Care and Medtronic, and grants from Johnson and Johnson outside the submitted work.

Ramzi Ajjan reports other funding from Abbott Diabetes Care during the conduct of the study

and personal fees from Abbott Diabetes Care outside the submitted work. Norbert Hermanns

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 373 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

reports grants and personal fees from Abbott Diabetes Care Germany, grants from Dexcom,

grants and personal fees from Berlin-Chemie, grants from Ypsomed, personal fees and

non-financial support from Novo Nordisk, and grants from Lilly International, outside the

submitted work.

Jean-Pierre Riveline reports grants outside the submitted work.

RR: relative risk ITT: intention to treat PP: per protocol

Table 1 from Supplement Haak 2017 : HbA1C, Time in range, hypoglycaemic, and hyperglycaemic measures for patients <65 years

Baseline Mean (SD) Study End Mean (SD) Difference in adjusted Difference in inter- Glycemic Measure means in intervention vention vs control p-value Intervention Control Intervention Control vs control (SE) (%) (n=95) (n=47) (n=95) (n=47)

HbA1c (mmol/mol) 72.8 (11.5) 74.1 (11.7) 68.1 (9.6) 70.5 (13.5) -3.6 (1.63) N/A 0.0300

HbA1c (%) 8.81 (1.06) 8.93 (1.08) 8.38 (0.88) 8.60 (1.24) -0.33 (0.149) N/A 0.0301

Time with glucose 3.9-10.0 mmol/L (70-180 13.3 (4.7) 12.7 (5.3) 13.3 (4.8) 12.7 (5.1) 0.3 (0.76) N/A 0.6777 mg/dL) (h)

Glucose <3.9 mmol/L (70 mg/dL) within 24 h

Events 0.58 (0.59) 0.55 (0.59) 0.39 (0.50) 0.50 (0.50) -0.12 (0.083) -22.1 0.1516

Time (h) 1.17 (1.60) 0.98 (1.41) 0.64 (0.95) 0.96 (1.08) -0.37 (0.168) -35.4 0.0279

AUC (h x mg/dL) 17.88 (29.57) 12.34 (19.94) 8.27 (14.65) 12.80 (17.13) -5.76 (2.64) -39.6 0.0305

Glucose <3.1 mmol/L (55 mg/dL)

Events 0.31 (0.45) 0.22 (0.31) 0.15 (0.27) 0.24 (0.30) -0.11 (0.048) -40.5 0.0221

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 374 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time (h) 0.52 (0.92) 0.33 (0.60) 0.22 (0.45) 0.34 (0.52) -0.15 (0.081) -38.2 0.0604

AUC (h x mg/dL) 5.19 (10.28) 2.83 (5.38) 2.01 (4.65) 3.27 (6.05) -1.70 (0.90) -43.7 0.0600

Glucose <2.5 mmol/L (45 mg/dL) within 24 h

Events 0.16 (0.30) 0.10 (0.18) 0.07 (0.16) 0.10 (0.17) -0.04 (0.03) -33.8 0.1842

Time (h) 0.27 (0.56) 0.13 (0.27) 0.10 (0.25) 0.17 (0.34) -0.10 (0.05) -48.3 0.0570

AUC (h x mg/dL) 1.30 (2.82) 0.57 (1.24) 0.44 (1.35) 0.82 (1.80) -0.50 (0.27) -50.4 0.0660

Glucose <2.2 mmol/L (40 mg/dL) within 24 h

Events 0.11 (0.22) 0.07 (0.14) 0.05 (0.15) 0.07 (0.15) -0.03 (0.03) -32.1 0.3309

Time (h) 0.19 (0.42) 0.08 (0.18) 0.06 (0.21) 0.12 (0.28) -0.08 (0.04) -51.1 0.0718

Glucose >10.0 mmol/L (180 mg/dL) within 24 h

Time (h) 9.5 (5.3) 10.3 (5.9) 10.1 (5.1) 10.3 (5.6) 0.1 (0.83) 0.5 0.9467

Glucose >13.3 mmol/L (240 mg/dL) within 24 h

Time (h) 3.5 (3.5) 4.4 (4.9) 3.7 (3.9) 4.2 (4.3) -0.1 (0.63) -1.9 0.9063

Glucose >16.7 mmol/L (300 mg/dL) within 24 h

Time (h) 0.99 (1.80) 1.53 (2.59) 1.10 (1.99) 1.33 (2.25) -0.06 (0.34) -4.9 0.8651

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 375 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table 2 from Supplement Haak 2017: HbA1C, Time in range, hypoglycaemic, and hyperglycaemic measures for patients ≥65 years

HbA1c (mmol/mol) 67.9 (9.5) 68.7 (7.7) 67.9 (8.0) 62.9 (8.6) 4.8 (1.76) N/A 0.0081

HbA1c (%) 8.36 (0.86) 8.44 (0.71) 8.36 (0.73) 7.90 (0.79) 0.44 (0.161) N/A 0.0081

Time with glucose 3.9-10.0 mmol/L (70-180 14.9 (4.1) 14.9 (4.8) 14.2 (4.2) 14.0 (4.5) 0.3 (0.89) N/A 0.7476 mg/dL) (h)

Glucose <3.9 mmol/L (70 mg/dL) within 24 h

Events 0.76 (0.68) 0.75 (0.76) 0.35 (0.35) 0.56 (0.71) -0.21 (0.106) -36.7 0.0513

Time (h) 1.53 (2.06) 1.26 (1.83) 0.49 (0.52) 1.03 (1.61) -0.60 (0.220) -55.9 0.0083

AUC (h x mg/dL) 24.14 (43.43) 16.92 (34.81) 5.40 (6.29) 14.91 (29.36) -10.86 (3.85) -70.9 0.0061

Glucose <3.1 mmol/L (55 mg/dL) within 24 h

Events 0.39 (0.57) 0.35 (0.60) 0.12 (0.14) 0.23 (0.46) -0.12 (0.058) -53.5 0.0357

Time (h) 0.72 (1.42) 0.46 (1.13) 0.13 (0.18) 0.41 (0.92) -0.33 (0.120) -78.0 0.0075

AUC (h x mg/dL) 7.49 (17.26) 4.36 (13.39) 0.99 (1.54) 4.31 (10.54) -3.84 (1.36) -88.8 0.0060

Glucose <2.5 mmol/L (45 mg/dL) within 24 h

Events 0.23 (0.47) 0.19 (0.51) 0.05 (0.08) 0.14 (0.36) -0.10 (0.04) -77.0 0.0225

Time (h) 0.40 (0.99) 0.24 (0.82) 0.05 (0.08) 0.23 (0.59) -0.21 (0.07) -93.2 0.0066

AUC (h x mg/dL) 1.90 (5.03) 1.11 (4.02) 0.19 (0.37) 1.11 (2.85) -1.04 (0.37) -96.6 0.0059

Glucose <2.2 mmol/L (40 mg/dL) within 24 h

Events 0.17 (0.42) 0.15 (0.47) 0.03 (0.06) 0.12 (0.30) -0.09 (0.04) -83.1 0.0189

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 376 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time (h) 0.29 (0.77) 0.18 (0.67) 0.03 (0.06) 0.17 (0.43) -0.15 (0.06) -95.9 0.0072

Glucose >10.0 mmol/L (180 mg/dL) within 24 h

Time (h) 7.6 (4.3) 7.8 (5.3) 9.3 (4.4) 9.0 (4.9) 0.4 (0.98) 4.2 0.7131

Glucose >13.3 mmol/L (240 mg/dL) within 24 h

Time (h) 2.4 (2.6) 3.0 (3.8) 3.2 (3.2) 3.4 (3.9) 0.1 (0.71) 3.7 0.8791

Glucose >16.7 mmol/L (300 mg/dL) within 24 h

Time (h) 0.52 (1.15) 0.85 (1.56) 0.86 (1.84) 0.81 (1.39) 0.25 (0.36) 39.8 0.4806 Analysis based on ANCOVA model with treatment group and study centre (pooled per country for <5 subjects/centre) as fixed effects and the baseline measurement as covariate. n=number of subjects with non-missing 14-day baseline and 14-day treatment-phase values; NA=not applicable; SD=standard deviation; SE=standard error.

Table 3 from Supplement Haak 2017: Hypoglycaemic measures

Baseline Mean (SD) Study End Mean (SD) Difference in adjusted Difference in inter- Glycemic Measure means in intervention vention vs control p-value Intervention (n=149) Control (n=75) Intervention (n=149) Control (n=75) vs control (SE) (%)

Glucose Events 0.64 (0.63) 0.63 (0.66) 0.38 (0.45) 0.53 (0.59) -0.16 (0.065) -27.7 0.0164 <3.9 mmol/L (70 Time (h) 1.30 (1.78) 1.08 (1.58) 0.59 (0.82) 0.99 (1.29) -0.47 (0.134) -43.1 0.0006 mg/dL)

within 24 h AUC (h x mg/dL) 20.15 (35.21) 14.05 (26.35) 7.23 (12.35) 13.59 (22.31) -7.80 (2.20) -51.1 0.0005

Glucose Events 0.49 (0.54) 0.44 (0.51) 0.29 (0.37) 0.34 (0.44) -0.08 (0.05) -21.6 0.1179 <3.9 mmol/L (70 mg/dL) Time (h) 0.74 (1.08) 0.59 (0.99) 0.35 (0.49) 0.47 (0.75) -0.16 (0.08) -30.7 0.0374 during day (06.00- 23.00) AUC (h x mg/dL) 10.81 (20.21) 7.42 (16.11) 4.07 (7.00) 6.26 (12.58) -2.98 (1.21) -41.1 0.0148

Glucose Events 0.25 (0.28) 0.27 (0.32) 0.14 (0.20) 0.27 (0.33) -0.12 (0.03) -44.9 0.0003 <3.9 mmol/L (70 mg/dL) Time (h) 0.55 (0.84) 0.49 (0.71) 0.23 (0.43) 0.51 (0.72) -0.29 (0.08) -54.3 0.0001

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 377 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

at night (23.00- AUC (h x mg/dL) 9.37 (18.08) 6.60 (11.45) 3.22 (6.95) 7.27 (13.02) -4.64 (1.33) -58.6 0.0006 06.00)

Glucose Events 0.34 (0.50) 0.27 (0.44) 0.14 (0.24) 0.24 (0.36) -0.12 (0.037) -44.3 0.0017 <3.1 mmol/L (55 Time (h) 0.59 (1.13) 0.38 (0.83) 0.19 (0.37) 0.37 (0.69) -0.22 (0.068) -53.1 0.0014 mg/dL)

within 24 h AUC (h x mg/dL) 6.02 (13.23) 3.40 (9.16) 1.64 (3.85) 3.66 (7.97) -2.51 (0.76) -60.3 0.0012

Glucose Events 0.23 (0.39) 0.17 (0.34) 0.10 (0.18) 0.14 (0.28) -0.07 (0.03) -39.7 0.0205 <3.1 mmol/L (55 mg/dL) Time (h) 0.31 (0.66) 0.20 (0.50) 0.10 (0.22) 0.17 (0.39) -0.09 (0.04) -44.3 0.0218 during day (06.00- 23.00) AUC (h x mg/dL) 3.07 (7.19) 1.80 (5.34) 0.85 (2.20) 1.60 (4.26) -0.95 (0.41) -50.5 0.0209

Glucose Events 0.15 (0.23) 0.13 (0.20) 0.06 (0.13) 0.13 (0.21) -0.07 (0.02) -53.0 0.0012 <3.1 mmol/L (55 mg/dL) Time (h) 0.27 (0.58) 0.18 (0.35) 0.09 (0.22) 0.19 (0.40) -0.12 (0.04) -58.1 0.0032 at night (23.00- 06.00) AUC (h x mg/dL) 2.98 (7.39) 1.58 (4.02) 0.80 (2.17) 2.02 (4.96) -1.46 (0.48) -65.6 0.0029

Glucose Events 0.19 (0.37) 0.13 (0.34) 0.06 (0.13) 0.11 (0.25) -0.06 (0.02) -48.8 0.0098 <2.5 mmol/L (45 Time (h) 0.32 (0.74) 0.17 (0.54) 0.08 (0.21) 0.19 (0.45) -0.14 (0.04) -64.1 0.0013 mg/dL)

within 24 h AUC (h x mg/dL) 1.52 (3.77) 0.77 (2.63) 0.35 (1.11) 0.93 (2.23) -0.70 (0.22) -66.7 0.0015

Glucose Events 0.13 (0.27) 0.08 (0.23) 0.04 (0.12) 0.06 (0.18) -0.03 (0.02) -40.3 0.1037 <2.5 mmol/L (45 mg/dL) Time (h) 0.16 (0.40) 0.09 (0.31) 0.04 (0.12) 0.08 (0.24) -0.05 (0.02) -55.7 0.0184 during day (06.00- 23.00) AUC (h x mg/dL) 0.73 (1.91) 0.42 (1.46) 0.17 (0.64) 0.39 (1.18) -0.26 (0.12) -56.7 0.0296

Glucose Events 0.08 (0.17) 0.06 (0.14) 0.03 (0.08) 0.07 (0.16) -0.04 (0.02) -57.8 0.0086 <2.5 mmol/L (45 mg/dL) Time (h) 0.16 (0.42) 0.08 (0.23) 0.04 (0.12) 0.11 (0.28) -0.08 (0.03) -68.3 0.0041 at night (23.00- 06.00) AUC (h x mg/dL) 0.80 (2.27) 0.35 (1.22) 0.18 (0.58) 0.53 (1.48) -0.42 (0.14) -71.7 0.0032

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 378 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

-0.05 (0.02) Glucose Events 0.13 (0.30) 0.10 (0.30) 0.05 (0.13) 0.09 (0.22) -52.6 0.0199 <2.2 mmol/L (40 -0.10 (0.03) mg/dL) Time (h) 0.22 (0.57) 0.12 (0.43) 0.05 (0.17) 0.14 (0.34) -66.7 0.0020

Glucose -0.03 (0.02) Events 0.12 (0.29) 0.09 (0.30) 0.04 (0.14) 0.06 (0.18) -36.6 0.2068 <2.2 mmol/L (40 mg/dL) -0.05 (0.03) during day (06.00- Time (h) 0.15 (0.40) 0.09 (0.34) 0.04 (0.15) 0.08 (0.25) -52.8 0.0550 23.00)

Glucose Events 0.23 (0.53) 0.16 (0.43) 0.07 (0.24) 0.19 (0.51) -0.13 (0.05) -66.3 0.0098 <2.2 mmol/L (40 mg/dL)

at night (23.00- Time (h) 0.41 (1.20) 0.18 (0.69) 0.09 (0.29) 0.27 (0.79) -0.22 (0.07) -73.5 0.0038 06.00)

*Analysis based on ANCOVA model with treatment group and study centre (pooled per country for <5 subjects/centre) as fixed effects and the baseline measurement as covariate. n=number of subjects with non-missing 14-day baseline and 14-day treatment-phase values; NA=not applicable; SD=standard deviation; SE=standard error.

Risk of Bias

Study (Author, year): Haak 2017 NCT02082184

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Participants were centrally randomized in a 2:1 ratio to sensor-based flash glucose monitoring (intervention group) or to self-monitoring of blood glucose (control group) by an interactive

web response system (IWRS) using biased-coin minimization, with study center and insulin administration as prognostic factors. The intention of a 2:1 randomization ratio was to ensure a sufficient number of participants in the intervention arm to complete an additional 6-month, open-access study phase.

Allocation concealment (Selection bias) Low See above

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 379 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Blinding of participants (Performance bias) High Participants, investigators, and study staff were not masked to group allocation.

Blinding of personnel (Performance bias) High Participants, investigators, and study staff were not masked to group allocation.

Blinding of outcome assessment (Detection bias) High Participants, investigators, and study staff were not masked to group allocation.

Incomplete outcome data (Attrition bias) Low Missing values for the primary endpoint were imputed using the last observation carried forward (LOCF) approach. Results presented are for the full analysis set, which included all randomized participants since there were no pregnancies. 10 patients withdrew in the intervention group and 17 in the control group.

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear Funding; Having restrictive protocols for treatment changes

would have made general applicability of data uncertain. Inclusion of only adults with intensive insulin therapy performing regular

glucose testing means future studies to assess the effectiveness of this novel glucose- sensing technology in younger, less concordant,

individuals with type 2 diabetes are needed. Had there been an insulin treatment algorithm and inclusion of participants with less regular blood glucose testing, the similar decline in HbA1c observed in both groups during the short period of this study may have been different. Common to glucose technology studies, our intervention was non- masked to subjects as sensor wear was experienced by all with assessment and some treatment decisions

based on the same sensor glucose values. No adjustment was made for multiple testing of secondary endpoints. Many of the endpoints,

particularly those derived from sensor glucose values, are highly inter-related and should not be considered in isolation.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 380 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author, year, reference Oskarsson et al. 2018

Oskarsson P, Antuna R., Geelhoed-Duijvestijn P. et al. Impact of flash glucose monitoring on hypoglycaemia in adults with type 1 diabetes managed with multiple daily injection therapy:

a pre-specified subgroup analysis of the IMPACT randomised controlled trial. Diabetologia (2018) 61: 539. https://doi.org/10.1007/s00125-017-4527-5

Study title/objectives Impact of flash glucose monitoring on hypoglycaemia in adults with type 1 diabetes managed with multiple daily injection therapy: a pre-specified subgroup analysis of the IMPACT randomised controlled trial.

Objectives:. In this pre-specified subgroup analysis (individuals with type 1 diabetes using multiple daily injection (MDI) therapy) of the Novel Glucose-Sensing Technology and Hypoglycaemia in Type 1 Diabetes: a Multicentre, Non-masked, Randomised Controlled Trial’ (IMPACT), the impact of flash glucose technology on hypoglycaemia compared with capillary glucose monitoring was assessed.

Study characteristics

Study design Prospective, non-masked, randomised controlled study

Study Registration number ClinicalTrials.gov, number NCT02232698

Country of recruitment Sweden, Austria, Germany, Spain, Netherlands

Centre (single or multicentre) Multicentre [23 European diabetes centres (three in Sweden, six in Austria, fi ve in Germany, three in Spain, and six in the Netherlands)]

Ethics Committee Approval Approval was given by the appropriate competent authority in each country. All participating centres gave ethics approval before the study.

Participants gave written informed consent.

Sponsor Abbott Diabetes Care sponsored the study. Abbott Diabetes Care provided study devices and all materials.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 381 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

The sponsor designed the study protocol in collaboration with the principal investigator in each country and provided all the study materials. The sponsor was involved in collecting data and reporting results, but wasnot involved in the authors’ interpretation or in writing text. The sponsor also funded medical writing services and gave approval to submit for publication. The corresponding author had full access to all the data in the study and, together with all authors, had final responsibility for the decision to submit for publication.

Study period (study start, study end) Sept 4, 2014 - Feb 12, 2015

Duration of follow-up (days) 3 months and 6 months • Participants aged 18 years or older Inclusion criteria • Diagnosis: type 1 diabetes for 5 years or longer • On current insulin regimen for at least 3 months before study entry • Screening HbA1c concentration of 58 mmol/mol (7.5%) or lower • Reported self-monitoring of blood glucose levels on a regular basis (equivalent to ≥3 times a day) for 2 months or more before study entry • Considered by the investigator to be technically capable of using the flash sensor-based glucose monitoring system • Current diagnosis of hypoglycaemia unawareness Exclusion criteria • Diabetic ketoacidosis or myocardial infarction in the preceding 6 months • Known allergy to medical-grade adhesives • Usage of continuous glucose monitoring within the preceding 4 months • Current use of sensor-augmented pump therapy • Pregnancy or planning pregnancy • Receiving oral steroid therapy for any disorders.

Patient characteristics Intervention (n=81) Control (n=80)

Age of patients (years) 42 (32–53) 44 (34–53)

Sex Men: 56 (69%) Men: 47(59%)

BMI (kg/m2) 25.1 ± 3.9 25.1 ± 3.7

Diagnosis (Type I, II or gestational) Type I Type I

Comorbidities (i.e. obesity…) - -

Time since diagnosis with DM (years) 19 (14–25) 19 (11–31)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 382 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Insulin treatment (CSII or MDII) • Multiple daily injections 81 80

Pregnancy (yes, no) No No

Special subgroup pf patient (i.e., hypoglycaemia fear…) - -

Intervention

Type of medical device (CGM or FGM) FGM (supported by SMBG)

Name (Description) of medical device FreeStyle Libre®; sensor-based fl ash glucose monitoring system (Abbott Diabetes Care, Witney, Oxon, UK).

Glucose management was supported by self-monitoring of blood glucose, using the strip port built into the reader and compatible test strips (Abbott Diabetes Care, Witney, Oxon, UK).

Adjunctive or non-adjunctive Non-adjuctive

Comparator

Type of comparator (SMBG or other CGM or FGM or other medical devices) SMBG

Name (Description) of medical device FreeStyle Lite meter and test strips (Abbott Diabetes Care, Witney, Oxon, UK) for self-monitoring of glucose concentrations

Sensor integrated (Yes, No) -

Sensor augmented (or enabled) insulin pump systems compatible (connect- - ed) withspecific CGM systems (Yes, No)

Outcomes

Primary Time spent in hypoglycaemia (<3.9 mmol/L [<70 mg/dL]) for the 14 days preceeding the end of the 6 month study period (days 194–208).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 383 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

mg/dL, respectively]; an episode was defi ned as at least two consecutive readings, at 15 min intervals, outside the predefined glucose range, the end of an episode was one reading at or higher than the threshold) • time with glucose in range 3.9–10.0 mmol/L (70–180 mg/dL) • number and duration of hyperglycaemic episodes (>10.0 mmol/L and >13.3 mmol/L [>180 mg/dL and >240 mg/dL, respectively]) • glucose variability measurements

Day 208 HbA1c concentrations

Change in total daily dose of insulin from day 1 to day 208

System utilisation for days 15–208 (defi ned as the percentage of data collected, assuming continuous device wear)

Frequency of glucose finger-sticks tests

Number of sensor scans per day

Additional outcomes Proportion of participants who achieve time spent in hypoglycaemia (<3.9 mmol/L; <70 mg/dL) ≤1 h/day

Number of events of symptomatic hypoglycaemia

Post prandial hyperglycaemia (>10.0 mmol/L, 180 mg/dL)

Prandial to basal insulin ratio

Number of participants changing from once daily to twice daily basal insulin

Body weight and body-mass index (BMI)

Fasting cholesterol and triglycerides

Blood pressure

Emergency room visits or admissions and non-protocol related additional clinic time

Medication usage (non-insulin related, including glucagon, self-reported from event diary)

Patient-recorded outcome measures: • HFS • DTSQ • DDS • DQoL

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 384 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Adverse events and sensor insertion-site symptoms

Number of episodes of diabetic ketoacidosis

Number of severe hypoglycaemia events (requiring third-party assistance)

Flow of patients IMPACT STUDY: SUBGROUP ANALYSIS (MDI patients only)

No of patients enrolled 328

No of randomized 167

Allocated per arms Intervention group: 82 Control group: 81

Received int. per arms Intervention group: 81 (1 excluded due to pregnancy) Control group:80 (1 excluded due to pregnancy)

Lost to follow-up per arms Intervention group: 6 withdrew or were excluded Control group: 11 withdrew or were excluded • 5 had device-associated symptoms • 2 owing to non-compliance with study device • 1 owing to non-compliance with study device • 1 met exclusion criteria • 2 on allocation to control group • 6 for other reasons

No of analysed per arm Intervention group: 75 Control group:69

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) Per protocol

Results

Effectiveness results n (%) 95% CI

Baseline Study end Difference in adjusted Difference in interven- P value tion vs control (%) Data in parentheses are SDs, apart Intervention Control Intervention Control means in intervention vs from when given with adjusted means control (95% CI) (n=81) (n=79) (n=81) (n=79) where they are SEs.. *Post-hoc endpoint.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 385 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

HbA1c (mmol/mol) 50.8 (4.8) 49.9 (7.5) 53.0 (6.5) 52.0 (7.5) 0.3 (−1.4, 2.0) NA 0.77

HbA1c (%) 6.80 (0.44) 6.71 (0.69) 7.00 (0.60) 6.91 (0.69) 0.02 (−0.13, 0.18) NA 0.77

Time withglucose 3.9–10.0 mmol/L 15.0 (2.6) 14.3 (2.9) 15.7 (2.8) 14.3 (3.0) 0.9 (0.2, 1.7) 6.5 0.011 (70–180 mg/dL) in h

Glucose<3.9 mmol/L (70 mg/dL) within 24 h • Events 1.80 (0.80) 1.72 (0.75) 1.23 (0.69) 1.78 (0.78) −0.59 (−0.78, −0.40) −32.8 <0.0001 • Time in h 3.44 (2.10) 3.73 (2.72) 1.86 (1.36) 3.66 (2.79) −1.65 (−2.21, −1.09) −46.0 <0.0001 • AUC, area under the curve 3.17 (2.57) 3.60 (3.38) 1.48 (1.49) 3.56 (3.79) −1.87 (−2.63, −1.10) −54.1 <0.0001 (h×mg/dL)

Glucose<3.9 mmol/L (70 mg/dL) at night (2300–0600 h) within 7 h • Events 0.57 (0.34) 0.61 (0.38) 0.30 (0.26) 0.54 (0.33) −0.22 (−0.30, −0.14) −41.7 <0.0001 • Time in h 1.20 (0.89) 1.41 (1.12) 0.61 (0.64) 1.28 (1.09) −0.57 (−0.81, −0.34) −46.6 <0.0001

Glucose<3.1 mmol/L (55 mg/dL) within 24 h • Events 1.01 (0.65) 1.00 (0.69) 0.50 (0.48) 1.04 (0.76) −0.55 (−0.71, −0.38) −52.2 <0.0001 • Time in h 1.75 (1.53) 1.99 (1.97) 0.75 (0.88) 1.97 (2.24) −1.10 (−1.55, −0.65) −57.7 <0.0001 • AUC (h×mg/dL) 1.00 (1.07) 1.20 (1.39) 0.40 (0.58) 1.20 (1.71) −0.71 (−1.06, −0.36) −61.1 0.0001

Glucose<3.1 mmol/L (55 mg/dL) at night (2300–0600 h) within 7 h • Events 0.37 (0.27) 0.41 (0.34) 0.16 (0.18) 0.33 (0.27) −0.16 (−0.22, −0.09) −47.6 <0.0001 • Time in h 0.67 (0.62) 0.85 (0.85) 0.28 (0.37) 0.76 (0.86) −0.39 (−0.57, −0.21) −54.4 <0.0001

Glucose<2.5 mmol/L (45 mg/dL) within

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 386 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

24 h* • Events 0.61 (0.55) 0.63 (0.59) 0.28 (0.36) 0.65 (0.66) −0.37 (−0.50, −0.23) −56.4 <0.0001 • Time in h 0.97 (1.15) 1.19 (1.48) 0.38 (0.62) 1.20 (1.84) −0.72 (−1.11, −0.34) −62.6 0.0003 • AUC (h×mg/dL) 0.26 (0.34) 0.32 (0.44) 0.10 (0.18) 0.33 (0.57) −0.21 (−0.33, −0.09) −64.8 0.0008

Glucose<2.5 mmol/L (45 mg/dL) at night (2300–0600 h) within 7 h* • Events 0.26 (0.25) 0.30 (0.32) 0.10 (0.15) 0.23 (0.24) −0.12 (−0.17, −0.06) −51.3 <0.0001 • Time in h 0.40 (0.46) 0.56 (0.69) 0.15 (0.25) 0.50 (0.73) −0.28 (−0.44, −0.13) −60.8 0.0003

Glucose <2.2 mmol/L (40 mg/dL) within 24 h • Events 0.44 (0.48) 0.49 (0.52) 0.20 (0.32) 0.52 (0.63) −0.30 (−0.43, −0.17) −58.6 <0.0001 • Time in h 0.69 (0.97) 0.88 (1.24) 0.27 (0.53) 0.94 (1.66) −0.59 (−0.94, −0.24) −65.6 0.0012

Duration (h) at hyperglycaemic glucose level within 24 h period • >10.0 mmol/l 5.6 (2.4) 6.0 (3.3) 6.4 (3.0) 6.0 (3.3) 0.7 (−0.1, 1.4) 11.1 0.10 • >13.3 mmol/l 1.77 (1.36) 2.05 (1.86) 1.78 (1.41) 2.10 (1.62) −0.19 (−0.58, 0.21) −9.2 0.36 • >16.7 mmol/l 0.44 (0.50) 0.57 (0.77) 0.37 (0.47) 0.47 (0.57) −0.06 (−0.21, 0.09) −13.1 0.45

Glucose variability • BGRI (lood glucose risk in- 8.1 (2.3) 8.7 (2.9) 7.4 (2.5) 8.6 (2.7) −0.8 (−1.4, −0.1) −9.4 0.017 dex) • CV (coefficient of variation) 43.2 (6.6) 43.4 (6.5) 37.8 (5.6) 42.6 (6.8) −4.7 (−6.2, −3.2) −11.1 <0.0001 glucose (%) • LBGI (low blood glucose in- 2.70 (1.35) 2.87 (1.76) 1.61 (0.93) 2.77 (1.73) −1.07 (−1.42, −0.72) −39.3 <0.0001 dex) • MAGE(mean amplitude of 7.9 (1.5) 8.2 (1.8) 7.5 (1.4) 8.0 (1.8) −0.31 (−0.72, 0.11) −3.9 0.14 glycaemic excursions)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 387 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(mg/dL; average) • Mean glucose (mg/dL) 7.8 (1.0) 7.9 (1.4) 8.2 (1.1) 7.9 (1.3) 0.38 (0.08, 0.68) 4.9 0.014 • Standard deviation of glu- 3.36 (0.63) 3.41 (0.76) 3.10 (0.58) 3.36 (0.78) −0.23 (−0.39, −0.07) −6.9 0.0051 cose(mg/dL)

CONGA (continuous overall net glycaemic action) • 2 h (mg/dL) 3.2 (0.7) 3.2 (0.8) 2.8 (0.7) 3.3 (0.8) −0.48 (−0.66, −0.30) −14.8 <0.0001 • 6 h (mg/dL) 4.0 (1.5) 4.0 (1.5) 3.7 (1.4) 4.1 (1.6) −0.39 (−0.85, 0.06) −9.7 0.089

Mean number of glucose readings used in the primary endpoint analysis Baseline Study End

Intervention Control Intervention Control

(n = 81) (n = 79) (n = 81) (n = 79)

Overall 24 h 1265 1269 1364 1172

Daytime 901 902 968 827

Night-time 364 367 396 345

Mortality -

Quality of life Questionnaire item Baseline Study end Difference in adjusted p value means in intervention Patient satisfaction (mean (SD) unless otherwise stated) Intervention Control Intervention Control and control (95% CI) * Participants were requested to (n = 78) (n = 70) (n = 78) (n = 70) complete questionnaires at 6 months. DTSQ Hence, data from n = 5 participants who did not complete the study (n = 4 Total treatment satisfaction score 28.3 (4.7) 27.7 (5.3) 13.3 (5.4) 6.8 (6.2) 6.4 (4.4, 8.4) <0.0001 from the intervention group and n = 1 Perceived frequency of hypoglycaemia 2.3 (1.2) 2.6 (1.4) −0.4 (1.6) 0.2 (1.1) −0.6 (−1.1, −0.2) 0.010 from the control group) are included in Perceived frequency of hyperglycaemia 2.5 (1.3) 2.8 (1.4) −0.6 (1.7) 0.6 (1.2) −1.2 (−1.7, −0.7) <0.0001 the analysis. Questionnaire data was not available for n = 1 individual in the DQoL

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 388 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

intervention group who did complete Total core scale score 1.9 (0.3) 2.1 (0.5) 1.9 (0.4) 2.1 (0.5) −0.1 (−0.2, 0.0) 0.15 the study Satisfaction with treatment 2.0 (0.5) 2.1 (0.5) 1.8 (0.5) 2.2 (0.5) −0.3 (−0.4, −0.1) <0.0001

Social worry 1.6 (0.5) 1.9 (0.7) 1.7 (0.6) 1.9 (0.6) 0.0 (−0.1, 0.2) 0.74

Diabetes worry 2.0 (0.6) 2.1 (0.7) 1.9 (0.6) 2.1 (0.7) −0.1 (−0.3, 0.1) 0.23

Impact of treatment 2.0 (0.3) 2.2 (0.4) 2.0 (0.3) 2.2 (0.4) −0.0 (−0.1, 0.1) 0.82

DDS

Total DDS score 1.9 (1.0) 2.1 (1.1) 1.9 (1.0) 2.0 (1.0) 0.0 (−0.2, 0.2) 0.98

Emotional burden subscore 2.0 (1.0) 2.3 (1.2) 2.0 (1.1) 2.2 (1.1) −0.0 (−0.3, 0.2) 0.77

Physician distress subscore 1.8 (1.2) 1.9 (1.3) 1.8 (1.4) 1.8 (1.2) 0.1 (−0.2, 0.4) 0.45

Regimen distress subscore 2.1 (1.1) 2.2 (1.2) 2.0 (1.1) 2.1 (1.1) −0.0 (−0.3, 0.2) 0.71

Interpersonal distress subscore 1.6 (0.8) 2.0 (1.3) 1.6 (1.0) 1.8 (1.2) 0.0 (−0.2, 0.3) 0.74

HFS

Behavioural subscale 11.9 (6.4) 12.7 (7.3) 13.4 (5.6) 14.2 (7.3) −0.3 (−2.0, 1.4) 0.76

Worry subscale 15.0 (10.1) 19.0 (14.0) 14.9 (11.8) 18.4 (13.5) −1.0 (−4.6, 2.6) 0.59

Incidence of diabetic ketoacidosis -

Incidence of hyperosmolar, hyperglycaemic coma -

Resource utilization related to DM (System utilisation, defined as the percentage of Intervention group (n = 76, including one participant that withdrew at 6 months): 92.5 ± 8.1% (median 95.0%) data collected, assuming continuous device wear for 6 months by the intervention group)

Number of visits to emergency room -

Number of visits to primary care -

Number of visits to specialists -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 389 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of hospitalizations -

Number of daily finger-sticks tests (mean number of self-monitored blood glucose Intervention group Control group tests performed per day) • baseline (days 1–15): 5.5 ± 2.0 (mean ± SD) • baseline: 5.6 ± 1.9 (median 5.2) (median 5.4) • final phase: 5.5 ± 2.6 [median 5.1] • final phase 0.5 ± 1.0 (median 0.1)

Number of calibration -

Need (Yes, with number or No) of re-calibration -

Compliance/adherence -

Percentage of time using CGM -

Median sensor-wear duration 13.4 days (mean ± SE, 10.0 ± 0.13 days)

Number of sensor scans per day (in FGM system) (mean) Intervention group only: over 18 per day

Scaning frequency 14.7 ± 10.7 (mean ± SD [median 12.3]) per day in the final phase (days 194–208)

Change in total daily insulin dose (mean) Intervention group Control group P value • MDI −2.7 (SD 7.3) units −3.0 (SD 6.4) units 0.80

Total daily doses of insulin -

Body weight and BMI At the end of the study, weight (p = 0.34) and BMI (p= 0.32) were comparable between the groups.

Bolus/basal insulin ratios There was no change in basal/bolus insulin ratio for either group.

Safety results Intervention group Control group n (%) 95% CI (full analysis set and two participants that became pregnant.) (n=82) (n=81)

Any AEs (number) 92 86

Serious AE (SAE) (number) 4 5

Most frequent AEs (by arms) - -

Most frequent SEAs(by arms) - -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 390 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Death as SAE - -

Withdrawals due AEs 4 (5%) 1 (1%)

‡Due to severe hypoglycaemia.

Participants with adverse or serious adverse events 43 (52%) 42 (52%)

Participants with serious adverse events 4 (5%) 4 (5%)

Participants with hypoglycaemic serious adverse events* 1 (1%) 3 (4%)

*A hypoglycaemic serious adverse event was reported during the baseline phase

Number of hypoglycaemic serious adverse events* 1 4

*A hypoglycaemic serious adverse event was reported during the baseline phase

Participants with hypoglycaemic adverse events 0 1 (1%)

Number of hypoglycaemic adverse events 0 2

Participants with device-related adverse events† 6 (7%) 0

†Device-related adverse events were all related to wearing the sensor: one participant with allergy (moderate); one with itching (mild); one with rash (mild); two with insertion-site symptom (four severe; one participant had three events and one participant had one event) and one with erythema (severe). Four intervention group participants withdrew owing to adverse events (two severe, two mild), which were primarily itching, redness and erythema

Number of device-related adverse events 8 0

Sensor insertion-site signs and symptoms Total number (both groups): 144

Participants with sensor insertion-site signs and symptoms 34 participants across both groups • pain (14) Number of signs expected due to sensor insertion • bleeding (9) • oedema (3) • induration (3)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 391 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• erythema (23) Number of signs associated with sensor wear • itching (14) • rash (8)

Costs (only for national assessment) -

Author Disclosure (Conflict of interest)

Ramiro Antuna RA has received honoraria for lecture fees and served on advisory boards for Abbott Diabetes Care.

Petronella Geelhoed-Duijvestijn PG-D has received lecture honoraria from Abbott Diabetes Care and Medtronic and served on advisory boards for Medtronic.

Jens Kröge JK has received personal fees from Abbott Diabetes Care, personal fees from Abbott Diabetes Care during the conduct of this study, personal fees from Lilly, Novo Nordisk, Berlin Chemie, Medtronic, Sanofi,MSD and Astra- Zeneca.

Raimund Weitgasser RW has received lecture honoraria and serves on advisory boards for Abbott Diabetes Care,

Boehringer-Ingelheim, Eli Lilly, Merck Sharp & Dohme, Novo Nordisk, Roche Diabetes Care, Sanofi and Servier, and has received unrestricted study grants from Eli Lilly, Novo Nordisk and Sanofi.

Jan Bolinder JB has received honoraria for consulting or lecture fees from Abbott Diabetes Care, AstraZeneca, Insulet Corpo- ration, Integrity ApplicationsNovo Nordisk and Sanofi-Aventis.

Per Oskarsson PO declares that there is no duality of interest associated with their contribution to this manuscript.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 392 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Risk of Bias

Study (Author, year): Oskarsson 2019 Prespecified subgroup analysis of IMPACT Trial, NCT02232698, MDII patients

Judgement (Low, Support for judgement Unclear, High)

Random sequence generation (Selection bias) Low Participants were randomly assigned to fl ash sensorbased glucose monitoring (intervention group) or to selfmonitoring of blood glucose (control group) in a 1:1 ratio by central interactive web response system (IWRS) using the biased-coin minimisation method; study centre and type of insulin administration were prognostic factors.

Allocation concealment (Selection bias) Low Please see above

Blinding of participants (Performance bias) High Participants, investigators, and study staff were not masked to group allocation

Blinding of personnel (Performance bias) High Participants, investigators, and study staff were not masked to group allocation

Blinding of outcome assessment (Detection bias) Low All sensor glucose data were blinded for both participants and investigators.

Incomplete outcome data (Attrition bias) Low The primary endpoint and all secondary endpoints were assessed in the full analysis set, which included all randomised participants apart from those who had a positive pregnancy test during the study period. Safety outcomes were analysed in all participants who were enrolled. Missing values were imputed by last observation carried forward. The full analysis set included 167 MDII users, randomised to intervention (n=82, 75 completed) and control group (n=81, 69 completed); one woman from each group was excluded due to pregnancy.

Selective reporting (Reporting bias) Low No selective reporting

Other source of bias (Other bias) Unclear Funding; COI; Many of the endpoints, particularly those derived from sensor glucose values, are highly inter-related and should not be considered in isolation, as no adjustment was made for multiple testing of secondary endpoints. This pre-specified subgroup was limited to adults with well-controlled type 1 diabetes managed with MDI therapy, which may suggest these participants were more motivated or committed to self-management than other MDI populations; neither participants nor investigators were blinded to the intervention, which may have influenced the treatment effect

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 393 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

nRCT for SAF EVIDENCE TABLES

nRCTs

Bailey 2015 nRCT T1 MDII or CSII FGM Safety domain USA 14 days 75/72 Abbott Freestyle Libre® NCT02073058 Prospective, single-arm, clinical study 18-71 y Edge 2016 nRCT T1 MDII or CSII FGM Safety domain UK 14 days 89/89 Abbott Freestyle Libre® NCT02388815 Prospective, single-arm, clinical study 4-17 y Haak 2017 RCT T2 MDII or CSII 139/125 FGM Safety domain EU ≥ 18 years Abbott Freestyle Libre® 12 months NCT02082184 Single arm results at 12 month of Replace trial (participants in the interventional group continued into the further 6 months open-access phase)

Study author/Year Bailey et al. 2015 Bailey T, Bode BW, Christiansen MP, Klaff LJ, Alva S. The Performance and Usability of a Factory- Calibrated Flash Glucose Monitoring System. Diabetes Technology & Therapeutics. 2015;17(11):787-794. doi:10.1089/dia.2014.0378. Design Prospective, single-arm, clinical study Objective: The objective of this article is to describe the performance and usability of the System by comparing its scanned sensor results with capillary BG values measured using the built-in BG meter. Performance against

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 394 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

the venous blood reference is provided for comparison. Country U.S. No of centres 4 clinical sites No of patients 72 (of 75 study participants) included in the evaluation Diagnosis Type 1 or Type 2 diabetes Follow-up period 14 days (first in-clinic visit between Day 1 and 3, second in-clinic visit was between Day 4 and 9, and the third in-clinic visit between Day 10 and 14) Inclusion criteria Type 1 or Type 2 diabetes Exclusion criteria -

Patient characteristics Mean ± SD Median Range Age of patients (years) 46.4 ± 15.1 48.5 18–71 Sex - - - BMI (kg/m2) 28.3 ± 5.3 27.4 18.7–47.2 Diagnosis (Type I, II or gestational) Type I or Type II Comorbidities (i.e. obesity…) - Time since diagnosis with DM (years) 23.0 – 13.1 22.3 2.4–50.6 Insulin treatment (CSII or MDII) CSII and MDI Pregnancy (yes, no) - Special subgroup pf patient (i.e., hypogly- - caemia fear…) Number of included patients 75 (Only 1 group of patients-single arm study) Number of of analysed per arm 72 (3 participants exited after Visit 1 (two could not comply with study visits, and one had non–study-related severe hypoglycaemia prior to sensor insertion with unknown complications)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 395 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Intervention Type of medical device (CGM or FGM) FGM Name (Description) of medical device FreeStyle Libre® Flash glucose monitoring system (Abbott Diabetes Care, Alameda, CA) Adjunctive or non-adjunctive Non-adjuctive

Comparator Type of comparator (SMBG or other CGM or SMBG FGM or other medical devices) Name (Description) of medical device FreeStyle Precision BG meter (built-in FreeStyle Libre® Reader ) Performance against the venous blood reference is provided for comparison. (YSI analyzer (Yellow Springs Instrument, Yellow Springs, OH) Sensor integrated (Yes, No) - Sensor augmented (or enabled) insulin pump - systems compatible (connected) with specific CGM systems (Yes, No)

Accuracy Pairinig of results: Real-time glucose level results: available for 99.2% (25,834/26,045) of sensor scans BG levels results: 13,195 YSI reference results: 12,172 Twenty-eight pairs were excluded because the reference glucose result was beyond the BG system’s dynamic range (20–500 mg/dL), and 114 pairs were excluded because the sensor result was beyond the Sys- tem’s dynamic range (40– 500 mg/dL).

Metrics for accuracy

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 396 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

MARD mean absolute relative difference See below

MAD mean absolute difference (deviation) See below

Days 1–14 Days 2–14 Day (7 Night Without hypogly- a.m.– (11 caemia/ rapid 11p.m) p.m-7 change a.m) Reference, glucose range, All Home Clinic All Home Clinic All Clin- parameter ic Capillary, BG < 100 mg/dL • MAD (mg/dL) 11.3 11.3 11.8 11.0 11.0 10.6 11.5 10.4 10.3 8.4 • N 2,153 1,946 207 1,985 1,813 172 1,787 366 905 97 Capillary, BG ≥ 100 mg/dL • MARD (%) 10.7 10.5 11.8 10.2 10.2 10.4 10.8 9.9 10.2 11.1 • N 11,042 9,642 1,400 9,987 8,878 1,109 9,717 1,325 9,341 1,23 6 All • % 11.4 11.3 12.1 11.0 11.0 10.7 11.5 10.8 10.4 11.0 • N 13,195 11,588 1,607 11,972 1,0691 1,281 11,504 1,691 10,246 1,33 3 Venous, YSI < 100 mg/dL • MAD (mg/dL) — — 13.4 — — 12.6 — — — 10.2 • N — — 1,475 — — 1,327 — — — 620

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 397 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Venous, YSI ≥ 100 mg/dL • MARD (%) — — 11.4 — — 10.3 — — — 10.9 • N — — 10,697 — — 8,789 — — — 9,72 2 All • % — — 12.0 — — 11.0 — — — 11.0 • N — — 12,172 — — 10,116 — — — 1,03 42 • MARD on days 2,7,14 Day 2 Day 7 Day 14 11.9% 10.9% 10.8% Median ARD (absolute relative difference) - PARD (paired or precision absolute relative - difference) Correlation between FreeStyle Libr®e sen- Highly correlated. Regression analysis produced a slope of 1.02, an intercept of -6.4 mg/dL, and a correlation sor results and capillary BG coefficient of 0.95 (range, 23–498 mg/dL). There were no statistically significant differences in sensor sensitivity (i.e., slope) between insertion sites on either the right or left arm (P = 0.5542).

Accuracy assessment methods (Surveil- lance) FreeStyle Libre® sensor readings vs capillary blood glucose reference values using the FreeStyle Precision meter built into the Reader Clarke Error Grid Analysis (EGA) Percentage of results in zone A of EGA 85.5

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 398 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Percentage of results in zones A and B of EGA 99.0 Consensus Error Grid (CEG) (also known as Percentage of results in zone A of CEG 86.7 the Parkes error grid) Percentage of results in zones A and B of CEG 99.7

The percentage of readings within Consensus Day 2 Day 7 Day 14 Error Grid Zone A (BG reference) 88.4% 89.2% 85.2% Sensor accuracy was not affected by factors such as body mass index (BMI), age, type of diabetes, clinical site, insulin administration, or hemoglobin A1c, as the percentages of readings within Consensus Error Grid Zone A were similar Clarke and Parkes Error Grid Analysis (EGA) Surveillance Error Grid methodology (CG- EGA) Continuous Glucose-Error Grid Analysis (CG- 96.5% (11,232/11,640) of the data categorized as clinically accurate, and a further 2.4% (274/11,640) classi- EGA) vs venous reference fied as benign errors Percentage of FreeStyle Libre® sensor results 86.2% within ± 15 mg/dl ± 20 or BG reference and venous reference Percentage of FreeStyle Libre® sensor results 82.8% within ± 15 mg/dl ± 20 % of venous reference

Safety outcomes Adverse events n (%) 95% CI Any AEs 202 skin issues observed in site exams (n of patients=72) Serious AE (SAE) 1 severe hypoglycaemia prior to sensor insertion - not related to the study or device

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 399 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Most frequent AEs (by arms) Moderate to severe incidences: • moderate to severe itching 0.5% of the time, • moderate erythema 4.0% of the time • 98.6% of the insertions with estimated pain rating of ≤ 2. • Rate of mild incidences • < 9% for any individual category of skin issues mentioned above, including oedema, rash, induration, bruising, bleeding, and others. Most frequent SEAs (by arms) - Death as SAE - Withdrawals due AEs - User acceptability outcome Rates of the experience with the system on a Day 1, favourable ratings were reported by most respondents (≥97.2%; n = 72) for seven of seven subjective scale of 0 (strongly agree) to 4 (strongly disa- statements. gree) (Day 1 and day 15) Day 15, the majority of respondents (≥ 94.4%; n = 72) reported favourable ratings for nine of nine subjective Day 1: 7 categories related to ease of use, statements. pain compared with finger-stick, adequacy of In total, 98.6% (142/144; both arms of 72 respondents) of responses were favourable ratings of ≤ 2 for the packaging information, and pain or bleeding statement ‘‘The amount of pain I felt when applying the Sensor was acceptable.’’ A favourable response of ≤ when applying the sensor to either arm 2 was also reported by 99.3% (143/144) of responses for statement ‘‘The amount of bleeding I experienced Day 15: score card included nine statements when applying the Sensor was acceptable.’’ about the sensor relative to comfort of wear, ease of wear, whether the sensor got in the way of daily activities, pain compared with finger-stick, easier than finger-stick, and erythema or oedema after removal of sensor from either arm

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 400 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author Disclosure (Conflict of interest) Shridhara Alva, PhD S.A. is an employee of Abbott Diabetes Care. M.P.C., Timothy Bailey, MD, FACE, FACP L.J.K., T.B., and B.W.B. were the study investigators and declare that no competing financial interests exist. Bruce W. Bode, MD, FACE Leslie J. Klaff, MD

Study author/Year Edge et al. 2016 Edge J, Acerini C, Campbell F, et al An alternative sensor-based method for glucose monitoring in children and young people with diabetes Archives of Disease in Childhood 2017;102:543-549. Design Prospective, single arm study Objective To determine accuracy, safety and acceptability of the FreeStyle Libre® Flash Glucose Monitoring System in the paediatric population. This study is the first to report accuracy, safety and acceptability of the FreeStyle Libre® System in children and young people. Country UK No of centres 9 diabetes centres No of patients 89 Diagnosis Type I diabetes Follow-up period 14 days (3 visits: Visit 1 on day 1, visit 2 on days 5, 6, 7 or 8 and final visit on days 12, 13, 14 or 15) Inclusion criteria • Children and young people aged 4–17 years, • Type 1 diabetes (T1D) or Type 2 diabetes • Treatement with multiple daily injections (MDI) of insulin or continuous subcutaneous insulin infusion (CSII)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 401 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• Currently testing BG at least two times per day Exclusion criteria • Concomitant disease/condition that may compromise patient safety • Other labelling prohibitions • Currently using a CGM device • Known/suspected allergy to medical grade adhesives.

Patient characteristics Age of patients mean age: 10.4 years • 24 participants aged 4–7 years • 39 aged 8–12 years • 26 aged 13–17 years. Sex Male: 50.6% BMI - Diagnosis (Type I, II or gestational) Type I (all patients) Comorbidities (i.e. obesity…) - Time since diagnosis with DM (years) 4.0 ± 2.8 (0.3, 12.2) (mean±SD (min, max) Insulin treatment (CSII or MDII) CSII: 56% ; MDI: 44% Pregnancy (yes, no) No Special subgroup pf patient (i.e., hypo- - glycaemia fear…)

Intervention Type of medical device (CGM or FGM) FGM

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 402 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Name (Description) of medical device FreeStyle Libre® System (Abbott Diabetes Care) Adjunctive or non-adjunctive Non-adjunctive

Comparator Type of comparator (SMBG or other SMBG CGM or FGM or other medical devices) Name (Description) of medical device Capillary blood glucose (BG) measurements using the BG strip-port on the reader (FreeStyle Optium test strips, Abbott Diabetes Care) Sensor integrated (Yes, No) - Sensor augmented (or enabled) insulin - pump systems compatible (connected) with specific CGM systems (Yes, No)

Accuracy Participants included in the safety analyses: n=87 ((1 withdrawal prior to obtaining sensor data, 1 protocol deviation led to no reader BG data) For assessment of sensor accuracy against capillary BG there were 5493 paired sensor-BG results. Metrics for accuracy MARD mean absolute relative differ- • Overall : 13.9%, ence • At higher glucose concentrations, BG 5.55–10.0 mmol/L (n=2090): 13.5% • At BG>10.0 mmol/L (n=1935):10.6% Median ARD (absolute relative differ- 10.4%, ence) MAD mean absolute difference (devia- 0.75 mmol/L for paired results at lower glucose concentrations, with BG<5.55 mmol/L (n=1468) tion) MRD (mean relative difference) 1.7% PARD (paired or precision absolute -

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 403 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

relative difference) Percentage within 1.1/20 82.0%. Accuracy assessment methods (Surveillance) Clarke Error Grid Analysis (EGA) Consensus Error Grid (CEG) (also • 83.8% of results in zone A known as the Parkes error grid) • 99.4% of results in zones A and B • 0.6% of results in zone C

There were no statistically significant differences in accuracy detected (ANOVA) for sex (p=0.951), method of insulin administration (p=0.640), sensor lot (p=0.135) or daytime versus night-time use (p=0.909), as the percentage of results within CEG zone A were similar for the subgroups. Accuracy was also not significantly influenced by age or body weight—the regression slopes (for percentage within CEG zone A vs participant age and weight) were not significantly different from zero (p=0.133 and p=0.284, respectively). Clarke and Parkes Error Grid Analysis - (EGA) Surveillance Error Grid methodology - (CG-EGA) Continuous Glucose-Error Grid Analysis - (CG-EGA) Sensor results vs BG results— Deming Sensor results were in good agreement with BG results— Deming regression: slope=1.03, intercept=−0.23 regression mmol/L, correlation coefficient=0.95. Other

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 404 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Trend (sensor results higher/ lower than BG when glucose is decreasing/increasing) was not evident with the FreeStyle Libre® System), demonstrating that there is little ‘lag’ effect. The sensor detected hypoglycaemia (when capillary BG <3.9 mmol/L) on 70% (438/622) of occasions, increasing to 84% when pending alerts (sensor results within 10% of the hypoglycaemic threshold) were included. For the 30% of occasions where hypoglycaemia measured in capillary testing was not detected by the sensor, further analysis of those results for clinical significance using the CEG indicates that 164 were in zones A and B (clinically acceptable) and 20 were in zone C (altered clinical action—likely to affect clinical outcome). The sensor detected hyperglycaemia (when BG >13.3 mmol/L) on 85% of occasions, increasing to 94% when pending alerts were included (n=999).

Safety outcomes Adverse events Participants included in the safety analyses: n=89 n (%) 95% CI Any AEs 5 (in 5 (6%) participants aged 6, 9, 10, 12 and 15 years) Anticipated AEs associated with sensor application or insertion sites: site exams performed for all sensor insertions Serious AE (SAE) 1 (One participant had a serious AE that was not related to the study or device (pain and lack of feeling in leg)) Most frequent AEs (by arms) Abrasion on sensor removal (n=2); allergic reaction, blister, pink mark/scabbing; 4 mild, one 4 moderate, all were resolved at study completion. Anticipated AEs associated with sensor application or insertion sites: moderate erythema was observed on 11.6% of occasions, mild erythema and pain 13.6% and 4.1%, respectively, and mild instances of bleeding, bruising, itching and oedema were each on <3% of occasions. Most frequent SEAs (by arms) - Death as SAE - Withdrawals due AEs - User acceptability outcome

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 405 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Rates of the experience with the system Girls (n=41-44) Boys (n=40-45) on a scale of 0 (strongly agree) to 4 (% answerd (% answerd (strongly disagree) favourably) favourably) After sensor It did not hurt when the sensor was put on 84.1 % 84.4% applied It was easy to put the sensor on 90.9 % 93.3 % After sensor I did not mind wearing the sensor on my arm 93.0% 97.7% wear It was comfortable to wear the sensor 88.4% 86.0% It was easy to scan the sensor 100% 100% It was more comfortable 92.9% 95.0% It was less painful 90.5% 85.0% It was more private 83.3% 87.5% Comparison It was quicker to check my blood glucose 100% 95.0% to finger prick It was easier to use 97.6% 97.5% testing It didi not get in the way of my daily activities 88.1% 85.0% I liked how my glucose readings were shown on the 87.8% 90.0% screen It gave me more information than my current glucose 71.4% 67.5% meter to take care of my diabetes It helped me to understand how my daily activities 66.7% 70.0% changed my glucose levels It made me feel more interested in taking care of my dia- 73.8% 72.5% betes I woul recommend it to someone else with diabetes 95.2% 97.5%

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 406 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Author Disclosure (Conflict of inter- All authors’ institutions received financial support from Abbott Diabetes Care to conduct the clinical trial. The study est) sponsor (Abbott Diabetes Care) designed the study protocol in collaboration with the principal investigator and provided all study materials. The sponsor was involved in collecting data and reporting results. The sponsor also gave approval to submit for publication. The corresponding author had full access to all the data in the study and, together with all authors, had final responsibility for the decision to submit for publication Tabitha Randell TR received honoraria/fees for consulting from Abbott Diabetes Care outside of the submitted work.

Author, year, reference Haak 2017 12 months results in one arm (participants in the interventional group continued into the further 6 months open-access phase)

Haak T, Hanaire H, Ajjan R, Hermanns N, Riveline JP, Rayman G. Use of Flash Glucose-Sensing Technology for 12 months as a Replacement for Blood Glucose Monitoring in Insulin-treated Type 2 Diabetes. Diabetes therapy : research, treatment and education of diabetes and related disorders. 2017;8(3):573-86.

Study title/objectives An Evaluation of a Novel Glucose Sensing Technology in Type 2 Diabetes (REPLACE)

Study characteristics

Study design 6-month, prospective, open-label, non-masked, randomized controlled study

Randomization was centrally, using biased-coin minimization.

For the 6-month treatment phase (post-randomization), the sensor-based glucose monitoring system was un-blinded for the participants in the intervention group so that they could continuously use sensor glucose data for self- management, including insulin dose decisions, in accordance with the product labeling.

Study Registration number ClinicalTrials.gov, NCT02082184

Country of recruitment France(8 sites), Germany(10), United Kingdom(8)

Centre (single or multicentre) 26 European diabetes centers

Ethics Committee Approval Approval was given by the appropriate competent authorities in each country.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 407 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Sponsor Abbott Diabetes Care

Study period (study start, study end) 6 March 2014-December 2015

Duration of follow-up (days) After 6 months treatment phase, participants in the interventional group continued into the further 6 months open- access phase

Review of glucose reports with clinician at 3-month intervals(beginning and day 284)

Inclusion criteria Any potentially eligible patient from the general diabetes population

if they were ≥18 years of age with type 2 diabetes

treated with insulin for at least 6 months

on their current regimen (prandial only or prandial and basal multi-dose-insulin therapy or CSII therapy) for ≥3 months;

had an HbA1c level of 58–108 mmol/mol (7.5–12.0%);

had self-reported regular blood glucose testing data (more than 10/week for at least 2 months prior to study entry);

were considered by the investigator to be technically capable of using the flash sensor-based glucose monitoring system.

ClinicalTrials.gov 1. Has Type 2 diabetes on insulin therapy for ≥ 6 months and on their current regimen for ≥3 months prior to study entry. 2. Their insulin management must be one of the following; 1. an injection regimen of prandial insulin at least once daily, 2. or, prandial insulin at least once daily plus basal insulin at least once daily, 3. or, continuous subcutaneous insulin infusion (CSII) with no plans to change during the study. 3. HbA1c result ≥7.5% (58 mmol/mol) and ≤12.0% (108 mmol/mol) on entry to the study. 4. Reports self-testing of blood glucose levels on a regular basis equivalent to a minimum of 10 tests per week, for at least 2 months prior to study entry. 5. In the investigator's opinion the subject is considered technically capable of using the Abbott Sensor Based Glucose Monitoring System. 6. In the Investigator's opinion the subject is proactive and therefore willing to modify their diabetes management 7. Aged 18 years or over.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 408 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Exclusion criteria Participants were not included for the following reasons:

if they had any other insulin regimen to that described above;

had a total daily dose of insulin ≥1.75 U/kg on study entry;

had severe hypoglycaemia (requiring third-party assistance), diabetic ketoacidosis or hyperosmolar–hyperglycemic state in the preceding 6 months;

had a known allergy to medical-grade adhesives;

used continuous glucose monitoring within the previous 4 months;

were pregnant or planning pregnancy;

were receiving steroid therapy for any condition;

were considered by the investigator to be unsuitable to participate.

ClinicalTrials.gov 1. Insulin regimen consists entirely of basal or includes bi-phasic insulin. 2. Subject is currently prescribed animal insulin. 3. Subject is currently prescribed steroid therapy or is likely to require steroid therapy for any acute or chronic condition during the study. 4. Has known allergy to medical grade adhesives. 5. Currently participating in another device or drug study that could affect glucose measurements or glucose management. 6. Currently using a Continuous Glucose Monitoring (CGM) device or has used one within the previous 4 months. 7. Is planning to use a CGM device at any time during the study. 8. Total daily dose of insulin (TDD) is >1.75iu/kg at entry to the study. 9. A female subject who is pregnant or planning to become pregnant within the study duration. 10. Currently receiving dialysis treatment or planning to receive dialysis during the study. 11. Has experienced an acute myocardial infarction within previous 6 months. 12. Has a concomitant disease or condition that may compromise patient safety, including unstable coronary heart disease, cystic fibrosis, serious psychiatric disorder, or any other uncontrolled medical condition.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 409 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

13. Has a pacemaker or any other neuro stimulators. 14. Has experienced any episode of severe hypoglycaemia, requiring third party assistance and/or admission to hospital, in the previous 6 months. 15. Has experienced any episode of diabetic ketoacidosis (DKA) or hyperosmolar hyperglycaemic state (HHS) in the previous 6 months. 16. In the investigator's opinion, the subject is considered as unsuitable for inclusion in the study for any other reason.

Patient characteristics

Age of patients 18 Years and older

Baseline characteristics in Open-access phase intervention participants

59.3 ± 9.6

Sex Both(all)

BMI Included in baseline characteristic but not specified (in open access phase intervention participants 33.1 ± 6.0

Diagnosis (Type I, II or gestational) Type 2 Diabetes

Comorbidities (i.e. obesity…)

Time since diagnosis with DM 17 ± 8 years (Duration of diabetes (years) in open access phase intervention participants)

Insulin treatment (CSII or MDII) MDII or CSII (CSII units (n = 5))

Pregnancy (yes, no) No

Special subgroup pf patient (i.e., hypoglycaemia fear…)

Intervention

Type of medical device (CGM or FGM) • FGM

Sensor Based Glucose Monitoring System

Subjects will wear the Abbott Sensor Based Glucose Monitoring System (unmasked) for 6 months to monitor their glucose levels. Post completion of the 6 month intervention, subjects participating in this arm of the study will be given a

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 410 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

further 6 month period of open access to the device.

All subjects will wear a masked Abbott Sensor Based Glucose Monitoring System, for 14 days prior to randomisation. • FreeStyle Libre®, Abbot Diabetes care, Witney, UK) Name (Description) of medical device On-body sensor utilizes wired enzyme technology (osmium mediator and glucose oxidase enzyme co-immobilized on electrochemical sensor) to continuously monitor interstitial glucose levels. The sensor is worn on the back of the arm for up to 14 days and automatically stores glucose data every 15 min.

A glucose trend arrow (indicating rate and direction of change in glucose levels) and a graphical trace of glucose values for the previous 8-h period is also displayed on the

screen. Data are transferred wirelessly by radio frequency identification from the sensor to the reader memory which stores historical sensor data for 90 days.

Factory-calibrated.

Adjunctive or non-adjunctive non-adjunctive

Comparator • SMBG

Standard Blood Glucose Monitoring

Subjects will use an Abbott Blood Glucose Monitoring System (standard blood glucose meter) for 6 months to monitor their glucose levels. A 14-day masked wear of the Abbott Sensor Based Glucose Monitoring System is included for these subjects at the 6 month time point, to collect glycaemic variability data for comparison to the intervention group of the study.

All subjects will wear a masked Abbott Sensor Based Glucose Monitoring System, for 14 days prior to randomisation.

Type of comparator (SMBG or other CGM or FGM or other medical devices) SMBG

Name (Description) of medical device Abbott Blood Glucose Monitoring System (standard blood glucose meter)

Sensor integrated (Yes, No) No

Sensor augmented (or enabled) insulin pump systems compatible (con- No nected) with specific CGM systems (Yes, No)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 411 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Outcomes

Primary The primary outcomes were changes in sensor-derived glycemic measures between baseline

and 12 months post-baseline. The sensor-derived glycemic measures were number and duration of hypoglycemic events [glucose \3.9 mmol/L (70 mg/dL)] and number and

duration of hyperglycemic events [glucose [13.3 mmol/L (240 mg/dL)].

Secondary Pre-specified secondary endpoints included:

sensor-derived glycemic measures between

baseline and 12 months post-baseline;

frequency of glucose finger-sticks and sensor scans per day during the study period;

and total daily dose of insulin.

Sensor-derived glycemic measures included:

number and duration of hypoglycemic events [glucose\3.1 mmol/L (55 mg/dL)];

time in glucose range 3.9–10.0 mmol/L (70–180 mg/dL),

number and duration of hyperglycemic events [glucose [10.0 mmol/L (180 mg/dL)];

mean and standard deviation (SD) glucose.

An event was defined as at least two consecutive readings, at 15-min intervals, outside the predefined glucose range (the end of an episode was 1 reading at or inside the predefined range).

Safety endpoints

incorporated all adverse events, including severe hypoglycaemia (requiring third-party assistance), hypoglycemic events and sensor insertion or sensor wear-related symptoms, diabetic ketoacidosis or hyperosmolar hyperglycemic state episodes and cardiac events.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 412 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ClinicalTrials.gov:

• Time in Range [ Time Frame: Baseline and Days 194 to 208 ]

Difference in time in range 70 -180 mg/dL between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15 time in range).

• Change in Time in Range at Day 45 [ Time Frame: Baseline and Days 31 to 45 ]

Change is defined as time in range 70 - 180 mg/dL in days 31 -45 minus time in range at baseline (days 1- 15) in the intervention group

• Decrease in HbA1c by greater than or equal to 0.5% [ Time Frame: Baseline and Day 194 ]

Difference in the proportion of subjects with a greater than or equal to 0.5% reduction in HbA1c from baseline at day 194 between intervention and control group

• 6 Month HbA1c Level ≤7.5% [ Time Frame: Day 194 ]

Difference in the proportion of subjects that achieve HbA1c level ≤7.5% at day 194 between intervention and control group

• HbA1c at 3 Months [ Time Frame: Baseline and 105 Days ]

Difference in HbA1c between intervention and control group at day 105 adjusting for baseline HbA1c at day 1

• Mean Glucose [ Time Frame: Baseline and Days 194 to 208 ]

Difference in mean glucose between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15). mean glucose

• Glucose variability [ Time Frame: Baseline and Days 194 to 208 ]

Difference in continuous overall net glycaemic action (CONGA) (1, 2 & 4 hour), standard deviation (SD) rate of change, (mean amplitude of glycaemic excursions) MAGE, mean of daily differences (MODD) , hign blood glucose index (HBGI) , low blood glucose index (LBGI) post prandial glucose peak, time to peak, excursion and area under the curve (AUC) >180mg/dL between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15). Post prandial glucose peak, time to peak, excursion and AUC>180mg/dL will be calculated during the 3 hours after each logged meal, for each subject and timeframe. Glucose excursion is change from meal time to peak glucose.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 413 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• Time spent and frequency < 70 mg/dL and <55mg/dL [ Time Frame: Baseline and Days 194 to 208 ]

Difference in time and frequency < 70 mg/dL and <55mg/dL(hours per day) between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15).

• Time spent and frequency > 180 mg/dL and > 240mg/dL [ Time Frame: Baseline and Days 194 to 208 ]

Difference in time >180 mg/dL and >240mg/dL (hours per day) between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15).

• Number of episodes of diabetic ketoacidosis (DKA), Hyperosmolar hyperglycaemic state (HHS), and severe hypoglycaemia. [ Time Frame: Baseline and Days 15 to 208 ]

The number of episodes of DKA, HHS and severe hypoglycaemia.will be summarised in the intervention and control group in each timeframe.

• Total daily dose of insulin (TDD) [ Time Frame: Baseline and Days 194 to 208 ]

Difference in TDD between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15) TDD. TDD calculated from Reader uploads.

• Body weight, BMI, blood pressure and lipids at 6 Months [ Time Frame: Baseline (day 1) and Day 194 ]

Difference in body weight, BMI blood pressure and lipids between intervention and control group assessed at day 194 adjusting for baseline (day 1).

• User Questionnaire (Subject & HCP) at 6 Months [ Time Frame: Day 208 ]

User questionnaire responses will be tabulated as number and percentage of subjects with each response. HCP questionnaire responses will be tabulated as number and percentage of HCPs with each response

• Quality of Life [ Time Frame: Baseline and Day 194 ]

Includes diabetes treatment satisfaction questionnaire (DTSQ), diabetes distress scale(DDS) & diabetes quality of life (DQoL) Difference in each score between intervention and control group assessed at day 194 adjusting for the corresponding score at baseline (day 1).

• Clinic time and medication usage [ Time Frame: Baseline (days 1 -15) and Days 15 to 208 ]

The number of emergency room visits, admissions, additional non protocol related clinic time and medication usage will be summarised in the intervention and control group in each timeframe (days 1-15 and days 15 -208).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 414 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

• Number of Glucose Measurements Performed [ Time Frame: Baseline and Days 194 to 208 ]

Difference in frequency (number per day) of blood glucose finger stick tests between intervention and control group assessed in days 194 to 208 adjusting for baseline (days 1 to 15) frequency of finger stick tests performed. The frequency of sensor scans performed will be summarised in the intervention group in days 194 to 208.

• Adverse Events [ Time Frame: Days -30 to 208 ]

The number of adverse events in days -30 to 208 will be tabulated as the number and percentage of subjects experi- encing adverse events in the intervention and control groups, classified by seriousness, severity, relationship to the device and relationship to the study. Glycaemic adverse events will be classed separately.

• Adverse Events Open Access Phase [ Time Frame: Days 208 to 388 ]

The number of adverse events in days 208 to 388 will be tabulated as the number and percentage of subjects experi- encing adverse events in the intervention and control groups, classified by seriousness, severity, relationship to the device and relationship to the study. Glycaemic adverse events will be classed separately.

• Adverse Events at 1 Month [ Time Frame: Days -30 to 45 ]

The number of adverse events in days -30 to 45 will be tabulated as the number and percentage of subjects experienc- ing adverse events in the intervention groups, classified by seriousness, severity, relationship to the device and relation to the study, Glycaemic adverse events will be classed separately.

Flow of patients

No of patients enrolled treatment phase 302 (March 13 and October 15)

No of randomized 224 subjects were randomised 1:2

Allocated per arms 149 intervention group : 75 control group

Received int. per arms 149 Free Style Libre® Flash Glucose Monitor : 75 SMBG

Lost to follow-up per arms 13 intervention/ sensor : 10 control/ standard

No of analysed per arm

Open access phase 139 interventional arm (further in open access)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 415 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

All 139 (100%) intervention participants completing the treatment phase continued into the

open-access phase, of whom 125 completed the open-access phase

No completed in open access 125 (14 discontinuited)

Included in primary endpoint ITT analysis 108

Statistical analysis

ITT, modified ITT, Per protocol; other (specify) ITT (Trial profile ITT)

Differences between post-baseline and baseline measurements were evaluated using a paired

t test. Sensor-derived glycemic endpoint values were excluded from the analysis if <72 h of

sensor results were available from the final 14-day sensor wear (days 374–388). Confidence

intervals were calculated for the mean difference from baseline. The results presented here are for the full analysis set.

Results

Effectiveness results n (%) 95% CI

Sensor-derived glycemic endpoint values were included

for 108 participants who had ≥72 h of sensor results from the three (baseline and 6 and 12 months post-baseline) 14- day sensor wear periods.

Mortality

Change in HbA1c HbA1c level of 58–108 mmol/mol (7.5–12.0%) at baseline

8.72 ± 0.96 [7.5, 11.6] declared in open access phase participants

Incidence of hypoglycaemia level 1 hypoglycaemia At 12 months (end of open-access period), time in hypoglycaemia [sensor glucose <3.9 mmol/L

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 416 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

(70 mg/dL)] was reduced by 50% compared with baseline [-0.70 ± 1.85/24 h (mean ± standard

deviation); p = 0.0002].

Nocturnal hypoglycaemia [glucose <3.9 mmol/L (70 mg/dL), 2300–0600 hours] was reduced by 52% (-0.31 ± 0.84 h per 7 h) at 12 months post-baseline compared with baseline (p = 0.0002)

Daytime hypoglycaemia [glucose<3.9 mmol/L (\70 mg/dL), 0600–2300 hours] was reduced by

48% (-0.38 ± 1.18 h per 17 h) at 12 months post-baseline comparedwith baseline p = 0.0011

The frequency of events with glucose at <3.9 mmol/L (70 mg/dL) was reduced by 41%

(-0.27 ± 0.67, mean ± SD) at 12 months compared with baseline (p<0.0001). The frequency

of events with glucose at<3.1 mmol/L (55 mg/ dL) was reduced by 56% (-0.20 ± 0.49,

p\0.0001), and that of events with glucose at <2.5 mmol/L (45 mg/dL) by 62% (-0.13 ± 0.35)

compared with baseline (p = 0.0002).

A difference for area under the curve of 58% (-12.73 ± 34.53 h/day 9 mg/dL, mean ± SD)

for sensor glucose level of<3.9 mmol/L (70 mg/dL) was observed at 12 months compared with

baseline (p = 0.0002). level 2 hypoglycaemia For sensor glucose levels of <3.1 mmol/L (55 mg/dL) and <2.5 mmol/L (45 mg/dL), the area under the curve was reduced by 65% (-4.28 ± 12.76 h/day 9 mg/dL, p = 0.0007) and by 69% (-1.12 ± 3.67 h/day 9 mg/dL p = 0.0021), respectively. level 3 hypoglycaemia

Incidence of hyperglycaemia

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 417 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time spent in range There was also no difference in time in glucose range 3.9–10.0 mmol/L (70–180 mg/dL)]

between baseline and 12 months post-baseline (p = 0.8519) or change in glycemic variability

[p = 0.1324;

The mean glucose level increased from 9.1 ± 1.8 to 9.4 ± 1.5 mmol/L (p = 0.0409).

Time spent in hypoglycaemia Significant reductions in all sensor measures of time spent in hypoglycaemia [glucose

<3.9 mmol/L (70 mg/dL),<3.1 mmol/L (55 mg/dL) and <2.5 mmol/L (45 mg/dL)], number of

events and area under the curve were observed for participants at the end of the open-access

phase (12 months) compared with the baseline phase.

Time in hypoglycaemia [glucose<3.9 mmol/L (70 mg/dL)] was reduced by 50% (-0.70 ± 1.85 h/day; mean ± SD) at 12 months post-baseline compared with baseline(p = 0.0002).

The frequency of events with glucose at <3.9 mmol/L (70 mg/dL) was reduced by 41%

(-0.27 ± 0.67, mean ± SD) at 12 months compared with baseline (p<0.0001).

Time in hypoglycaemia [glucose<3.1 mmol/L (55 mg/dL)] was reduced by 62% (-0.40 ± 1.09 h/day) at 12 months post- baseline compared with baseline (p = 0.0002).

The frequency of events with glucose at<3.1 mmol/L (55 mg/dL) was reduced by 56% (-0.20 ± 0.49, p<0.0001)

Time in hypoglycaemia [glucose<2.5 mmol/L (45 mg/dL)] was reduced by 67% (-0.23 ± 0.73 h/day) at 12 months post- baseline compared with baseline (p = 0.0013).

The frequency of events with glucose at <2.5 mmol/L (45 mg/dL) by 62% (-0.13 ± 0.35)

compared with baseline (p = 0.0002).

Time spent in hyperglycaemia At 12 months post-baseline there was no difference in time in hyperglycaemia [[10.0 mmol/L (180 mg/dL), [13.3 mmol/L (240 mg/dL), and [16.7 mmol/L (300 mg/dL)] compared with baseline (p = 0.1981, p = 0.9533, and p = 0.8349, respectively.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 418 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

For those participants aged <65 years (n = 60; 56%) and those aged ≥65 years (n = 48;

44%), significant reductions in time spent in hypoglycaemia [<3.9 mmol/L (70 mg/dL),

<3.1 mmol/L (55 mg/dL) and <2.5 mmol/L (45 mg/dL)] were observed at the end of the

open-access phase (12 months) compared with the baseline phase.

Quality of life Measured(see secondary outcomes) but non declared

Patient satisfaction

Hypoglycaemia fear

Incidence of diabetic ketoacidosis 0

Incidence of hyperosmolar, hyperglycaemic coma 0

Resource utilization related to DM

Number of visits to emergency room

Number of visits to primary care

Number of visits to specialists

Number of hospitalizations

Number of daily finger-sticks tests For the participants who continued into the open-access phase, mean SMBG frequency was

3.9 ± 1.2 (SD) tests/day (median 3.9 tests/day) at baseline, falling to a mean of 0.6 ± 1.2 tests/day (median 0.1) when participants first had full access to sensor glucose data

(days 15–31, treatment phase).

The mean overall blood glucose monitoring rate for the 6-month treatment phase was 0.3 ± 0.7 tests/day (median 0.1), further reducing to 0.2 ± 0.6 tests/day (median 0.0) during the

open-access phase

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 419 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Number of calibration Factory calibrated.

Need (Yes, with number or No) of re-calibration No

Compliance/adherence

Percentage of time using CGM

Number of sensor scans per day (in FGM system) Average sensor-scanning frequency was 7.1 ± 3.5 times/day (median 5.7) during the

open-access phase compared with 8.4 ± 4.6 during the 6-month treatment phase (median 6.8

times/day)

There was no correlation between increased frequency of sensor scanning and reduction in time in hypoglycaemia <3.9mmol/L (70 mg/dL) or hyperglycaemia 13.3 mmol/L (240 mg/dL)] between the baseline phase and end of the open- access phase (12months).

Mean device use (defined as the percentage of data collected, assuming continuous

device wear) was 83.6% ± 13.8 (median 88.3%)between 6 and 12months and 88.7 ± 9.2% (median 90.7%) in the treatment phase.

Safety results Open-access Phase Participants (N= 139) n (%) 95% CI

Any AEs Serious adverse or adverse events (n = 135) were experienced by 60 (43%) of 139 participants.

There were nine occurrences of a serious adverse event, none of which were related to the device, study procedure, or to hypoglycaemia.

9 (6.5%)participants reported 16 device-related adverse events; four severe, nine moderate and three mild. These were all sensor-adhesive or site reactions, primarily treated with topical preparations and all were resolved.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 420 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Serious AE (SAE) Participants with serious adverse events 7 (5.0%) (Table 3)

Number of serious adverse events 9 (This number includes seven serious adverse events reported in the 6-month treatment phase results) (Table 3)

Most frequent AEs (by arms) 9 (6.5%) participants reported 16 instances of device-related adverse events (e.g. infection, allergy) and

28 participants (20.1%) experienced 134 occurrences of anticipated skin symptoms/sensor-insertion events expected with device use (e.g. erythema, itching and rash).

There were 134 anticipated sensor insertion site symptoms observed for 28 (20%) participants.

These symptoms were primarily (n = 117; 87%) due to the sensor wear (erythema, itching and

rash) and most were resolved without medical intervention; 63 were mild in nature, 67 were

moderate and four were severe.

Most frequent SEAs (by arms)

Death as SAE

Withdrawals due AEs 5 (3.6%) Table 3.

Costs (only for national assessment)

Author Disclosure (Conflict of interest) Thomas Haak reports personal fees from Abbott Diabetes Care outside the submitted work; Gerry Rayman reports personal fees from Abbott Diabetes Care outside the submitted work; Helene Hanaire reports personal fees from Ab- bott Diabetes Care and Medtronic, and grants from Johnson and Johnson outside the submitted work;

Ramzi Ajjan reports other funding from Abbott Diabetes Care during the conduct of the study and personal fees from Abbott Diabetes Care outside the submitted work; Norbert Hermanns reports grants and personal fees from Abbott Diabetes Care Germany, grants from Dexcom, grants and personal fees from Berlin-Chemie, grants from Ypsomed, personal fees and non-financial support from Novo Nordisk, and grants from Lilly International, outside the submitted work; and

Jean-Pierre Riveline reports grants outside the submitted work.

RR: relative risk ITT: intention to treat PP: per protocol

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 421 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

List of ongoing and planned studies

Table A2: List of ongoing studies with rtCGM or FGM medical devices

Study Identifier Estimated Study type Number Intervention Comparator Patient population Endpoints completion date of patients NCT03522870 December 30, 2019 RCT 76 FGM Freestyle SMBG DM Type 1, adults Difference in HbA1c at week 12 and Libre® week26 adjusted for baseline(week-2); Difference in time in hypoglycaemia (<70mg/dL,<54mg/dL and<40mg/dL) assessed in week 12-14 and week 26- 28 adjusted for baseline(week-2 to 0) NCT03369899 July 2018 nRCT 100 FGM Freestyle DM Type 1 or 2, 4-17 years Performance and AEs Libre® NCT03175315 March 2018 RCT 216 FGM No intervention: DM Type 1 or 2, 16-75 years Primary outcome: difference in glycemic Waiting list control between baseline and the 6- month follow-up. Secondary outcome: time-in-range, frequency and duration of hypo- and hyperglycemic episodes, diabetes-related distress, depressive symptoms, health-related quality of life, diabetes self-efficacy, self-care, behavior, and hypoglycaemia awareness. NCT02898714 August 2019 nRCT 1100 FGM DM, older than 4 years QoL, Change in HbA1c NCT03448380 March 2021 nRCT 920 FGM Freestyle DM Type 1 or 2, 18 and older Safety Libre® NCT03448367 February 2021 nRCT 400 FGM Freestyle DM Type 1 or 2, 4-17 years Safety Libre® NCT03502174 August 2018 nRCT 85 FGM Freestyle DM Type 1 or 2, 6-17 years Performance and AEs Libre® NCT02776007 December 2018 RCT 60 FGM Freestyle SMBG DM Type 1, 12-17 years Diabetes Treatment satisfaction Libre® questionnaire, Libre-user evaluation quiestionaire, HbA1c NCT03048227 December 2019 RCT 28 Freestyle SMBG DM Type 1, 18-70 years Time spent in range, HbA1c, time in Navigator II®, hypo, time spent in hyper, QoL, AEs Abbott

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 422 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Study Identifier Estimated Study type Number Intervention Comparator Patient population Endpoints completion date of patients NCT03249974 July 2018 nRCT 20 Enlite™ sensor DM Type 1, 5-18 years, Performance, patient satisfation and communicating treated with an insulin pump AEs and wearing a FreeStyle Flash with Minimed™ ® 640G pump Libre NCT03445065 November 2018 RCT 324 Eversense® XL SMBG or FGM DM Type 1 or 2, 18 and older % HbA1c, Time spent in Hypoglycaemia, AEs NCT03445377 October 2019 RCT 30 rtCGM, Dexom SMBG DM Type 1, 16 to 24 y Time spent in range, time in hypo, time G5® spent in hyper, HbA1c, NCT03340831 May 2020 nRCT 1388 rtCGM, Dexom SMBG DM Type 1 or 2, 2 y or older Change in hypoglycaemic events, G5™ change in HbA1c, Satisfaction NCT02556554 September 2018 nRCT 40 rtCGM, Dexom Routine care DM Type 1, pregnancy Change in glucose variability, change in G5™ or G4™; behaviour, change in HbA1c, maternal Dexom G5™ or and foetal outcome G4™ with Share

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 423 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Risk of bias and GRADE tables

Please note that RoB assessment tables of individual studies included for clinical effectiveness and safety can be found in section “Evidence tables of individual studies included for clinical effectiveness and safety”. In this chapter you can find summarized RoB assessment tables as well as GRADE assessment tables.

RoB:

Figure A1: Risk of Bias at study level, with summary, for all RCTs included in this assessment [13,24-38]

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 424 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A3: Risk of bias – study level (RCTs)

Blinding of c- n- s- -

– study

t- term, casue -

Trial

- outcome

term)

Random sequence generation Allocation concealment Participants Medicinal perso nel Outcome asses ment (patient reported ou comes, all mortality) Incomplete outcome data (short long Selective reporting otherNo aspects a cording to risk of bias Risk of bias level

DIAMOND Beck Low Unclear High High Unclear Low Low Unclear High 2017

Ruedy 2017 Low Unclear High High Unclear Low Low Unclear High A subset analysis of the DIAMOND trial

GOLD Lind Low Unclear High High Unclear Low Low Unclear High 2017

Beck 2017 Low Unclear High High Unclear Low Low Unclear High

HypoDE Low Low High High Unclear Low Low Unclear High Heinemann 2018

Battelino Low Unclear High High Unclear Low High Low High 2011

Mauras 2012 Unclear Unclear High Unclear Unclear Unclear Unclear Low High

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 425 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Riveline Unclear Unclear High High Unclear Unclear Unclear Low High 2012

IN Low Low High High Unclear Low Low Low High CONTROL Van Beers 2016

Ly 2013 Low Unclear High High Unclear Low Unclear Low High

Reddy 2018 Low Unclear High High Unclear Low High High High

IMPACT Low Low High High Low Low Low Unclear High Bolinder 2016

REPLACE Low Low High High High Low Low Unclear High Haak 2017

Oskarsson Low Low High High Low Low Low Unclear High 2018 MDII subgroup of Impact trial

Comments: Specific details can be found in Table A , Appendix 1

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 426 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A4: Risk of bias – outcome level (randomised controlled trials) rtCGM vs SMBG

Blinding – outcome ITT principle adequately Selective outcome No other aspects Risk of bias – outcome Outcome and trial assessors realised reporting unlikely according to risk of bias level

HbA1c changes Beck 2017 DIAMOND Unclear Low Low Unclear Unclear

Ruedy 2017 Unclear Low Low Unclear Unclear

Lind 2017 GOLD Unclear Low Low Unclear Unclear

Beck 2017 (T2DM) Unclear Low Low Unclear Unclear

Heinemann 2018 Unclear Low Low Unclear Unclear van Beers 2016 Unclear Low Low Low Unclear

Ly 2013 Unclear Low Low Low Unclear

Time spent in target glycaemic range Beck 2017 DIAMOND Unclear Low Low Unclear Unclear

Ruedy 2017 Unclear Low Low Unclear Unclear

Lind 2017 GOLD Unclear Low Low Unclear Unclear

Back 2017 (T2DM) Unclear Low Low Unclear Unclear

Heinemann 2018 Unclear Low Low Unclear Unclear van Beers 2016 Unclear Low Low Low Unclear

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 427 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Blinding – outcome ITT principle adequately Selective outcome No other aspects Risk of bias – outcome Outcome and trial assessors realised reporting unlikely according to risk of bias level

Time spent in hypoglycaemic range Beck 2017 DIAMOND Unclear Low Low Unclear Unclear

Ruedy 2017 Unclear Low Low Unclear Unclear

Lind 2017 GOLD Unclear Low Low Unclear Unclear

Back 2017 (T2DM) Unclear Low Low Unclear Unclear

Heinemann 2018 Unclear Low Low Unclear Unclear van Beers 2016 Unclear Low Low Low Unclear

Ly 2013 Unclear Low Low Low Unclear

Hypoglycaemic events DIAMOND [Riddlesworth 2017] Unclear Low Low Unclear Unclear

Heinemann 2018 Unclear Low Low Unclear Unclear van Beers 2016 Unclear Low Low Low Unclear

Ly 2013 Unclear Low Low Low Unclear

Adverse events Device related, local

Lind 2017 Unclear Low Low Unclear Unclear

Ly 2013 Unclear Low Low Low Unclear

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 428 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Blinding – outcome ITT principle adequately Selective outcome No other aspects Risk of bias – outcome Outcome and trial assessors realised reporting unlikely according to risk of bias level

Patient-reported outcomes (QoL and patient satisfaction) DIAMOND Beck 2017 Unclear Low Low Unclear Unclear

DIAMOND Polonsky 2017 Unclear Low Low Unclear Unclear

Ruedy 2017 Unclear Low Low Unclear Unclear

Lind 2017 Unclear Low Low Unclear Unclear

Heinemann 2018 Unclear Low Low Unclear Unclear

Beck 2017 T2DM Unclear Low Low Unclear Unclear

Van Beers et al 2016 Unclear Low Low Low Unclear

Mauras 2012 Unclear Unclear Unclear Low Unclear

Riveline 2012 Unclear Unclear Unclear Low Unclear

Ly 2013 Unclear Low Low Low Unclear

Abbreviations:

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 429 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A5: Risk of bias – outcome level (randomised controlled trials) FGM vs CGM and FGM vs SMBG

Blinding – outcome ITT principle adequately Selective outcome No other aspects Risk of bias – outcome Outcome and trial assessors realised reporting unlikely according to risk of bias level

HbA1c changes Reddy 2017 Unclear Low High High High

Bolinder 2016 Low Low Low Unclear Unclear

Haak 2017 High Low Low Unclear High

Time spent in target glycaemic range

Reddy 2017 Unclear Low High High High

Bolinder 2016 Low Low Low Unclear Unclear

Haak 2017 High Low Low Unclear High

Time spent in hypoglycaemic range

Reddy 2017 Unclear Low High High High

Bolinder 2016 Low Low Low Unclear Unclear

Haak 2017 High Low Low Unclear High

Oskarsson 2018 Low Low Low Unclear Unclear

Hypoglycaemic events

Reddy 2017 Unclear Low High High High

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 430 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Blinding – outcome ITT principle adequately Selective outcome No other aspects Risk of bias – outcome Outcome and trial assessors realised reporting unlikely according to risk of bias level

Bolinder 2016 Low Low Low Unclear Unclear

Haak 2017 High Low Low Unclear High

Oskarsson 2018 Low Low Low Unclear Unclear

Adverse events Device related, local

Bolinder 2016 Low Low Low Unclear Unclear

Haak 2017 High Low Low Unclear High

Oskarsson 2018 Low Low Low Unclear Unclear

Patient-reported outcomes (QoL and patient satisfaction)

Reddy 2017 Unclear Low High High High

Bolinder 2016 Low Low Low Unclear Unclear

Haak 2017 High Low Low Unclear High

Oskarsson 2018 Low Low Low Unclear Unclear

Abbreviations: GRADE Assessment

Template for GRADE assessment (e.g., using GRADEproGDT); see http://gdt.guidelinedevelopment.org/app/ and http://gdt.guidelinedevelopment.org/app/handbook/handbook.html

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 431 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A6: rtCGM compared with SMBG for different outcomes

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

Patients on MDII, Changes in HbA1C, Changes in HbA1c from baseline to end of study, % (Beck 2017 and Lind 2017) (follow up: range 24 weeks to 26 weeks)

a 2 randomised serious not serious not serious not serious none 247 195 - MD 0.48 ◯ trials lower MODERATE (0.55 lower ⨁⨁⨁ to 0.41 lower)

Patients on MDII, Changes in HbA1C, HbA1C from Baseline to End of Study, % (Beck 2017 - T2DM) (follow up: 24 weeks)

a 1 randomised serious not serious not serious not serious none 105 53 - MD 0.6 ◯ trials lower MODERATE (0.84 lower ⨁⨁⨁ to 0.3 lower)

Patients on MDII, Changes in HbA1C, HbA1c <7% (Beck 2017- T1DM) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 103 53 - mean 15 ◯◯ trials higher LOW (0 to 30 ⨁⨁ higher)

Patients on MDII, Changes in HbA1C, Reduction from baseline at 24 weeks (Ruedy 2017) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 61 53 - mean 0.4 ◯◯ trials lower LOW (0.42 lower ⨁⨁ to 0.38 lower)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 432 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

Patients on MDII, Changes in HbA1C, HbA1c <7% (Beck 2017- T2DM) (follow up: 24 weeks)

a b 1 randomised very serious not serious not serious very serious none 11/79 (13.9%) 9/79 (11.4%) 3 -- per 100 ◯◯◯ trials (-9 to 14) (from -- to -- VERY LOW ) ⨁

Patients on MDII, Changes in HbA1C from Baseline to End of Study, % (Lind 2017 ) (follow up: 26 weeks) (follow up: 26 weeks)

1 randomised very serious a not serious not serious not serious 142 142 - mean 0.47 - trials lower (0.53 lower to 0.41 lower)

Patients on MDII, Changes in HbA1C from baseline to end of study, % (Beck 2017 - T2DM) (follow up: 24 weeks) (follow up: 24 weeks)

1 randomised very serious a not serious not serious very serious b - mean 0.3 - trials lower (0.5 lower to 0 )

Patients on MDII, Changes in HbA1C, Mean HbA1c, % (Heinemann 2018) (follow up: 26 weeks)

a b 1 randomised very serious not serious not serious very serious none 75 66 - 0.03 higher ◯◯◯ trials (0.12 lower VERY LOW to 0.19 ⨁ higher)

Patients on MDII and CSII, Changes in HbA1C, Changes in HbA1C from Baseline to End of Study, % (van Beers 2016) (follow up: 16 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 433 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a c 1 randomised very serious not serious not serious very serious none 26 26 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, Mean (adjusted) HbA1c, % (Battelino 2011) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 62 54 - mean 0.27 ◯◯ trials lower LOW (0.47 lower ⨁⨁ to 0.07 lower)

Patients on MDII and CSII, Changes in HbA1C from Baseline to End of Study, % (Mauras 2012) (follow up: 26 weeks)

a d d 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, Changes in HbA1C from Baseline to End of Study, % (Riveline 2012) (follow up: 12 weeks)

a c 1 randomised very serious not serious not serious very serious none 62 61 - mean 0.52 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients CSII, HbA1C from Baseline to End of Study, % (Ly 2013) (follow up: 24 weeks)

a b 1 randomised very serious not serious not serious very serious none 46 49 - mean 0.07 ◯◯◯ trials higher VERY LOW (0.2 lower to ⨁ 0.3 higher)

Patients on MDII, Time spent in the target glycemic range, Mean minutes per day in range 70–180 mg/dL (3.9-10.0 mmol/l) (Beck 2017) (follow up: 24 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 434 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a e 1 randomised very serious not serious not serious very serious none 105 53 - mean 77 ◯◯◯ trials higher VERY LOW (6 higher to ⨁ 147 higher)

Patients on MDII, Time spent in the target glycemic range, Mean (SD) minutes per day in range 70–180 mg/dL (3.9-10.0 mmol/l) (Ruedy 2017) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none 58 50 - mean 114 ◯◯◯ trials higher VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time spent in the target glycemic range, Mean (SD) % time spent in normoglycaemia (3.9-10.0 mmol/l) (Lind 2017) (follow up: 26 weeks)

a c 1 randomised very serious not serious not serious very serious none 123 125 - mean 2.47 ◯◯◯ trials higher VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time spent in the target glycemic range, Median minutes per day in range 70–180 mg/dL (3.9-10.0 mmol/l) (Beck 2017 T2DM) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious not serious none 74 72 - mean 38 ◯◯ trials higher LOW (0 to 0 ) ⨁⨁

Patients on MDII, Time spent in the target glycemic range, Mean (SD) duration per day, min (3.9-10.0 mmol/l) (Heinemann 2018) (follow up: 26 weeks)

a b 1 randomised very serious not serious not serious very serious none 75 66 - mean 44.9 ◯◯◯ trials higher VERY LOW (0.3 lower to ⨁ 90 higher)

Patients on MDII and CSII, Time spent in the target glycemic range, Mean % time spent in normoglycaemia (4.0–10.0 mmol/L) (van Beers 2016) (follow up: 16 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 435 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a 1 randomised very serious not serious not serious not serious none 26 26 - mean 9.6 ◯◯ trials higher LOW (8 higher to ⨁⨁ 11.2 higher)

Patients on MDII and CSII, Time spent in the target glycemic range, Mean (95% CI) min per day (van Beers 2016) (follow up: 16 weeks)

a 1 randomised very serious not serious not serious not serious none 26 26 - mean 138 ◯◯ trials higher LOW (114 higher ⨁⨁ to 162 higher)

Patients on MDII and CSII, Time spent in the target glycemic range, Mean (SD) hours per day in range 70–180 mg/dL (3.9-10.0 mmol/l) (Battelino 2011) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 62 54 - mean 1.1 ◯◯ trials higher LOW (1.02 higher ⨁⨁ to 1.18 higher)

Patients on MDII and CSII, Time spent in the target glycemic range, CGM glucose value (mg/dl) (% median) 71-–180 mg/dL (Mauras 2012) (follow up: 26 weeks)

a d 1 randomised very serious not serious not serious very serious none 62 67 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range ), Median minutes per day (<70 mg/dL) (Beck 2017 T1DM) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none 105 53 - median 37 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 436 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range), Median minutes per day (<60 mg/dL) (Beck 2017 T1DM) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none 105 53 - median 20 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range), Median minutes per day (<50 mg/dL), (Beck 2017 T1DM) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none 105 53 - median 14 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range), Median minutes per day (<60 mg/dL) (Ruedy 2017) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none 58 50 - 1 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range), Mean % time spent in hypoglycaemia (<70 mg/dL or 3.9 mmol/l) (Lind 2017 - T1DM) (follow up: 26 weeks)

a c 1 randomised very serious not serious not serious very serious none 69 73 - median 2 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Mean % time spent in hypoglycaemia (<70 mg/dL or 3.9 mmol/l) (Lind 2017 - T1DM) (follow up: 26 weeks)

a c 1 randomised very serious not serious not serious very serious none 123 125 - median 2 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range ), Time spent in hypoglycaemia, median (interquartile ranges), min (<54 mg/dL or 3.9 mmol/l) (Beck 2017 - T2DM) (follow up: 24 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 437 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a c 1 randomised very serious not serious not serious very serious none 74 72 - median 8 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range ), Time spent in hypoglycaemia, median (interquartile ranges), min (<50 mg/dL ) (Beck 2017 - T2DM) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none 74 72 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Median duration ≤3.9 mmol/l per day, min (Heinemann 2018) (follow up: 26 weeks)

a c 1 randomised very serious not serious not serious very serious none 75 66 - median 68.3 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Median duration ≤3.0 mmol/l per day, min (Heinemann 2018) (follow up: 26 weeks)

a c 1 randomised very serious not serious not serious very serious none 75 66 - median 29.1 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Median percentage ≤3.9 mmol/l (Heinemann 2018) (follow up: 26 weeks)

a c 1 randomised very serious not serious not serious very serious none 75 66 - median 4.8 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Median percentage ≤3.0 mmol/l (Heinemann 2018) (follow up: 26 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 438 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a c 1 randomised very serious not serious not serious very serious none 75 66 - median 2.2 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII and SCII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Mean hours per day in hypoglycaemia (≤ 3.9 mmol/L) (van Beers 2016) (follow up: 16 weeks)

a 1 randomised very serious not serious not serious not serious none 26 26 - mean 1.1 ◯◯ trials lower LOW (1.4 lower to ⨁⨁ 0.8 lower)

Patients on MDII and SCII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), Mean Hours per day in hypoglycaemia <63 mg/dL (Battelino 2011) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 62 54 - mean 0.49 ◯◯ trials higher LOW (0.26 higher ⨁⨁ to 0.76 higher)

Patients on MDII and SCII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), CGM glucose valuae (mg/dl) (% median) ≤70 mg/dL (Mauras 2012) (follow up: 26 weeks)

a d 1 randomised very serious not serious not serious very serious none 62 67 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and SCII, Time Spent Outside of Target Glycemic Range (in hypoglycemic range and in hyperglycemic range), CGM glucose valuae (mg/dl) (% median) ≤60 mg/dL (Mauras 2012) (follow up: 26 weeks)

a d 1 randomised very serious not serious not serious very serious none 62 67 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Number of events per 2 weeks in median (IQR) (Riddlesworth 2017)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 439 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

1 randomised very serious a not serious not serious 103 53 - 0 - trials (0 to 0 )

Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Number of events per 2 weeks in N (%) (Riddlesworth 2017)

1 randomised very serious a not serious not serious not estimable - trials

Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Event rate per 24 h, Median (IQR) (Riddlesworth 2017)

a 1 randomised very serious not serious not serious not serious none - 0 ◯◯ trials (0 to 0 ) LOW ⨁⨁ Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Change in event rate per 24 h, Median (IQR) (Riddlesworth 2017)

a c 1 randomised very serious not serious not serious very serious none - median 0.08 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Mean number of hypoglycaemic events per 28 days (Heinemann 2018) (follow up: 26 weeks)

a 1 randomised very serious not serious not serious not serious none 75 66 - mean 0.28 ◯◯ trials higher LOW (0.2 higher ⨁⨁ to 0.39 higher)

Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Mean number of hypoglycaemic events (Heinemann 2018) (follow up: 26 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 440 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a 1 randomised very serious not serious not serious not serious none 75 66 - mean 0.28 ◯◯ trials higher LOW (0.2 higher ⨁⨁ to 0.39 higher)

Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Incidence of hypoglycaemic events (Heinemann 2018) (follow up: 26 weeks)

a 1 randomised very serious not serious not serious not serious none -/75 -/66 RR 0.28 0 fewer per ◯◯ trials (0.20 to 0.39) 1,000 LOW (from 0 ⨁⨁ fewer to 0 fewer)

Patients on MDII, Results for hypoglycaemia and severe hypoglycaemia, Mean number of nocturnal hypoglycaemic events per 28 days (Heinemann 2018) (follow up: 26 weeks)

a 1 randomised very serious not serious not serious not serious none 75 66 - mean 0.35 ◯◯ trials higher LOW (0.22 higher ⨁⨁ to 0.56 higher)

Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, Number of severe hypoglycemic events (van Beers 2016) (follow up: 16 weeks)

a c 1 randomised very serious not serious not serious not serious none 26 26 - mean 20 ◯◯ trials lower LOW (0 to 0 ) ⨁⨁

Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, Patients with more than one severe hypoglycaemic event (van Beers 2016) (follow up: 16 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 441 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a 1 randomised very serious not serious not serious not serious none 10/10 (100.0%) 18/18 (100.0%) OR 0.48 0 fewer per ◯◯ trials (0.20 to 1.00) 1,000 LOW (from 0 ⨁⨁ fewer to 0 fewer)

Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, The number of hypoglycaemic excursions (Battelino 2011) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 62 54 - mean 0.76 ◯◯ trials higher LOW (0.47 higher ⨁⨁ to 1.43 higher)

Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, Glycaemia <63 mg/dL per day (Battelino 2011) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none 62 54 - mean 0.7 ◯◯ trials higher LOW (0.43 higher ⨁⨁ to 1.03 higher)

Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, Severe hypoglycemic events, Subjects with at least one event (Mauras 2012) (follow up: 26 weeks)

a d 1 randomised very serious not serious not serious very serious none 3/- 5/- not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, Incidence rate (per 100 person-years) (Mauras 2012) (follow up: 26 weeks)

a d 1 randomised very serious not serious not serious very serious none 8.6/- 17.6/- not estimable ◯◯◯ trials VERY LOW ⨁

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 442 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

Patients on MDII and CSII, Results for hypoglycaemia and severe hypoglycaemia, Severe hypoglycemic events, Subjects with at least one event (Riveline 2012) (follow up: 12 months)

a d 1 randomised very serious not serious not serious very serious none 15/- 6/- not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Severe hypoglycaemia including seizure of coma, 6-months rate per 100 patient-months (Ly 2013) (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious very serious none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Number of severe hypoglycemic events (Ly 2013) (follow up: 24 weeks)

a d 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Incidence rate difference from baseline to end point (95% CI) (Ly 2013) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none RR 1.5 2 fewer per ◯◯ trials (0.3 to 2.7) 1,000 LOW (from 0 ⨁⨁ fewer to 3 fewer)

Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Moderate hypoglycaemia, 6-month rate per 100 patient-months (Ly 2013) (follow up: 24 weeks)

a d 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Number of moderate hypoglycaemia events (Ly 2013) (follow up: 24 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 443 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a d 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Incidence rate per 100 patient-months (Ly 2013) (follow up: 24 weeks)

a b 1 randomised very serious not serious not serious very serious none RR 2.7 3 fewer per ◯◯◯ trials (1.2 to 6.1) 1,000 VERY LOW (from 1 ⨁ fewer to 6 fewer)

Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Sum of severe and moderate hypoglycaemia (Ly 2013) (follow up: 24 weeks)

a f 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, 6-month rate per 100 patient-months (Ly 2013)

a d 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Incidence rate per 100 patients-months (Ly 2013) (follow up: 24 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 444 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a 1 randomised very serious not serious not serious not serious none RR 3.6 4 fewer per ◯◯ trials (1.7 to 7.5) 1,000 LOW (from 2 ⨁⨁ fewer to 8 fewer)

Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Sensitivity analysis for patients younger than 12 years, 6-month rate per 100 patient-months (95% CI) (Ly 2013) (follow up: 24 weeks)

a d 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Sensitivity analysis for patients younger than 12 years, Number of events (Ly 2013) (follow up: 24 weeks)

a 1 randomised very serious not serious not serious not serious none RR 5.5 6 fewer per ◯◯ trials (2.0 to 15.7) 1,000 LOW (from 2 ⨁⨁ fewer to 16 fewer)

Patients on CSII, Results for hypoglycaemia and severe hypoglycaemia, Sensitivity analysis for patients younger than 12 years, Sum of moderate and severe outcomes, (Ly 2013) (follow up: 24 weeks)

a f 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, Change in Clarke score from baseline (van Beers 2016) (follow up: 16 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 445 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a b 1 randomised very serious not serious not serious very serious none 26 26 - mean 0.3 ◯◯◯ trials lower VERY LOW (0.9 lower to ⨁ 0.2 higher)

Patients on MDII and CSII, QoL and Satisfaction, QoL In quality of life from scores on the HFS Behaviour subscale, PAID-5, CIDS, EQ5D, or WHO-5 between the CGM and SMBG phases (van Beers 2016) (follow up: 16 weeks)

a f 1 randomised very serious not serious not serious very serious none 26 26 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, Scores on the HFS Worry subscale (van Beers 2016) (follow up: 16 weeks)

a 1 randomised very serious not serious not serious not serious none 26 26 - mean 6.4 ◯◯ trials higher LOW (1.4 higher ⨁⨁ to 11.4 higher)

Patients on MDII and CSII, QoL and Satisfaction, CGM satisfaction survey, mean score (van Beers 2016) (follow up: 16 weeks)

a f 1 randomised very serious not serious not serious very serious none 26 26 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, QoL Hypoglycaemia fear (Mauras 2012) (follow up: 26 weeks)

a d 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, QoL PAID questionaire (Mauras 2012) (follow up: 26 weeks)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 446 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

a d 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, CGM Satisfaction Scale (Mauras 2012) (follow up: 26 weeks)

a f 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, Overall (Mauras 2012) (follow up: 26 weeks)

a f 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, Benefits of CGM subscale (Mauras 2012) (follow up: 26 weeks)

a f 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL and Satisfaction, Lack of Hassles of CGM subscale (Mauras 2012) (follow up: 26 weeks)

a f 1 randomised very serious not serious not serious very serious none 69 68 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on CSII, Average-median % hours spent in hypoglycaemia, day (<60 mg/dL); Ly, 2013 (follow up: 24 weeks)

a c 1 randomised very serious not serious not serious serious none 46 49 - median 1.8 ◯◯◯ trials lower VERY LOW (0 to 0 ) ⨁

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 447 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM SMBG studies design (95% CI) (95% CI)

Patients on CSII, Average-median % hours spent in hypoglycaemia, night (<60 mg/dL); Ly, 2013 (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c 46 49 - 3.8 lower - trials (0 to 0 )

CI: Confidence interval; MD: Mean difference; RR: Risk ratio; OR: Odds ratio

Explanations a. At least domain of RoB tool had high risk of bias in all studies b. Confidence interval includes both positive and negative values c. No data available for confidence interval d. Only p value available; no data for main effect and confidence interval e. Wide confidence interval f. No numerical data reported for this outcome

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 448 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A7: rtCGM compared with SMBG for different outcomes (continued)

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM FGM studies design (95% CI) (95% CI)

Patients on MDII, CGM satisfac-tion survey, mean score, Beck et al, 2017

a c c 1 randomised very serious serious not serious serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII, CGM satisfaction survey, mean score, Change in Clarke Hypo-glycaemia Unawareness Questionnaire, Beck et al, 2017

a d 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII, CGM satisfaction survey, mean score, Ruedy et al, 2017

a c 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII, CGM satisfaction survey, mean score, QOL, Polonsky 2017

a c 1 randomised very serious not serious not serious very serious none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII, WHO-5 (mean±SD), QOL, Polonsky 2017

a d 1 randomised very serious not serious not serious very serious none - MD 1.26 ◯◯◯ trials lower VERY LOW (5.42 lower ⨁ to 2.91 higher)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 449 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM FGM studies design (95% CI) (95% CI)

Patients on MDII, EQ-5D-5L (mean±SD), QOL, Polonsky 2017

a d 1 randomised very serious not serious not serious serious none - MD 0 ◯◯◯ trials (0.03 lower VERY LOW to 0.03 ⨁ higher)

Patients on MDII, Diabetes dis-tress (DDS) (mean±SD), total, QOL, Polonsky 2017

a 1 randomised very serious not serious not serious not serious none - MD 0.22 ◯◯ trials higher LOW (0.08 higher ⨁⨁ to 0.36 higher)

Patients on MDII, Hypoglycaemia fear (worry subscale of HFS-II) (mean±SD), total, QOL, Polonsky 2017

a 1 randomised very serious not serious not serious not serious none - MD 3.17 ◯◯ trials higher LOW (0.19 higher ⨁⨁ to 6.14 higher)

Patients on MDII, DTSQ status version, scale total, Lind 2017

a 1 randomised very serious not serious not serious not serious none - MD 3.43 ◯◯ trials higher LOW (2.31 higher ⨁⨁ to 4.54 higher)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 450 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM FGM studies design (95% CI) (95% CI)

Patients on MDII, WHO-5, EQ-5D-5L, Diabetes dis-tress (DDS), Hypoglycaemia fear (worry subscale of HFS-II) , Patient satisfac-tion, Hypoglycaemia Unawereness Mean Change in Clarke Hypo-glycaemia Unawareness Total Score from baseline (SD) Beck, 2017

a c c 1 randomised very serious serious not serious serious none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII, QoL, Patient satisfaction, Hypoglycaemia fear, Heineman 2018

a c c 1 randomised very serious serious not serious serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII and CSII, Change in Clarke score from baseline, van Beers, 2016

1 randomised very serious a not serious not serious serious d - MD 0.3 - trials lower (0.9 lower to 0.2 higher)

Patients on MDII and CSII, QoL, van Beers, 2016

a c 1 randomised very serious not serious not serious very serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on MDII and CSII, Scores on the HFS Worry subscale, van Beers, 2016

a 1 randomised very serious not serious not serious not serious none - MD 6.4 ◯◯ trials higher LOW (1.4 higher ⨁⨁ to 11.4 higher)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 451 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM FGM studies design (95% CI) (95% CI)

Patients on MDII and CSII, CGM satisfaction survey, mean score , van Beers, 2016

a c 1 randomised very serious not serious not serious serious none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, QoL Hypoglycaemia fear, QoL PAID questionaire, CGM Satisfaction Scale, Overall, Benefits of CGM subscale, Lack of Hassles of CGM subscale, Mauras 2012

a c b,c 1 randomised very serious serious not serious serious none not estimable ◯◯◯ trials VERY LOW ⨁ Patients on CSII, Modified Clarke questionnaire for hypoglycaemia unawareness, Ly 2013

a 1 randomised very serious not serious not serious not serious none - MD 0.2 ◯◯ trials lower LOW (0.9 lower to ⨁⨁ 0.5 higher)

CI: Confidence interval; MD: Mean difference

Explanations a. At least one domain judged with high risk of bias b. Only p values available c. No numerical data provided for effect measures d. CI values include both positive and negative values

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 452 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A8: rtCGM compared with FGM

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM FGM studies design (95% CI) (95% CI)

Patients on MDII and CSII, median change in HbA1c (Reddy, 2017) (follow up: 8 weeks)

a b 1 randomised very serious serious not serious not serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, median change in HbA1c % (Reddy, 2017) (follow up: 8 weeks) (follow up: 8 weeks)

a b 1 randomised very serious serious not serious not serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Percentage time within defined glucose range 3.9-7.8 mmol/l (Reddy, 2017) (follow up: 8 weeks)

a b 1 randomised very serious serious not serious serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Percentage time within defined glucose range 3.9-01 mmol/l (Reddy, 2017) (follow up: 8 weeks)

a b 1 randomised very serious serious not serious not serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Percentage time within defined glucose range overnight ; 3.9-7.8 mmol/l (22.00–07.00) (Reddy 2017) (follow up: 8 weeks)

a b 1 randomised very serious serious not serious not serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 453 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Study Relative Absolute Risk of bias Inconsistency Indirectness Imprecision Other considerations rtCGM FGM studies design (95% CI) (95% CI)

Percentage time within defined glucose range overnight ; 3.9-10 mmol/l (22.00–07.00) (Reddy 2017) (follow up: 8 weeks)

a b 1 randomised very serious serious not serious not serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Percentage time within defined glucose range (hypoglycaemia), <2.8 mmol/l, 3.3 mmol/l, 3.5mmol/l, 3.9mmol/l; Reddy, 2017 (follow up: 8 weeks)

a b 1 randomised very serious not serious not serious very serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Percentage time within defined glucose range (hypoglycaemia) overnight (22.00–07.00), <2.8 mmol/l, 3.3 mmol/l, 3.5mmol/l, 3.9mmol/l; Reddy, 2017 (follow up: 8 weeks)

a b 1 randomised very serious not serious not serious serious none 19 20 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Gold score, HFS total score, HFS-Behaviour subscore, HFS- Worry subscore, PAID score, Reddy 2018 (follow up: 8 weeks)

a c c 1 randomised very serious serious not serious serious none not estimable ◯◯◯ trials VERY LOW ⨁ CI: Confidence interval; MD: Mean difference

Explanations a. At least one domain judged with high risk of bias b. Only p values available c. No numerical data provided for effect measures d. CI values include both positive and negative values

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 454 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Table A9: FGM compared with SMBG

Certainty assessment № of patients Effect

Certainty Importance № of Relative Absolute Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations FGM SMBG studies (95% CI) (95% CI)

Patients on MDII and CSII, Time with glucose 3.9–10.0 mmol/L (70–180 mg/dL) in h (Bolinder, 2016) (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious b none 119 119 - MD 1 higher ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patients on MDII and CSII, Time with glucose 3.9–10.0 mmol/L (70–180 mg/dL) in h, (Haak, 2017) (follow-up 24 weeks)

1 randomised very serious a not serious not serious serious c none 149 75 - MD 0.2 higher ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Time spent in hypoglycaemia, Glucose <3.9 mmol/L (70 mg/dL) within 24 h, Events, Bolinder, 2016 (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c,d none 119 119 - MD 25.8 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Time spent in hypoglycaemia, Glucose <3.9 mmol/L (70 mg/dL) within 24 h, Time in h, Bolinder, 2016 (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c,d none 119 119 - MD 38 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Time spent in hypoglycaemia, Glucose <3.9 mmol/L (70 mg/dL) within 24 h, AUC, Bolinder, 2016 (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c,d none 119 119 - MD 46.7 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Time spent in hypoglycaemia, Glucose <3.9 mmol/L (70 mg/dL) within 24 h, Events, Haak, 2017 (follow up: 1)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 455 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Relative Absolute Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations FGM SMBG studies (95% CI) (95% CI)

1 randomised very serious a not serious not serious serious c,d none 149 75 - MD 27.7 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Time spent in hypoglycaemia, Glucose <3.9 mmol/L (70 mg/dL) within 24 h, Time in h, Haak, 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - MD 43.1 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Time spent in hypoglycaemia, Glucose <3.9 mmol/L (70 mg/dL) within 24 h, AUC, Haak, 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - MD 51.1 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.9 mmol/L (70 mg/dL) at night (2300–0600 h) within 7 h, Events, Bolinder 2016 (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c,d none 119 119 - MD 33.2 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.9 mmol/L (70 mg/dL) at night (2300–0600 h) within 7 h, Time in h, Bolinder 2016 (follow up: 24 weeks) (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c,d none 119 119 - MD 39.8 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.9 mmol/L (70 mg/dL) at night (2300–0600 h) within 7 h, Events, Haak 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - MD 44.9 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.9 mmol/L (70 mg/dL) at night (2300–0600 h) within 7 h, Time in h, Haak 2017

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 456 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Relative Absolute Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations FGM SMBG studies (95% CI) (95% CI)

1 randomised very serious a not serious not serious serious c,d none 149 75 - MD 54.3 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.1 mmol/L (55 mg/dL) within 24 h, Events, Time in h, AUC; Bolinder 2016

1 randomised very serious a not serious not serious serious c,d none 119 119 - MD 41.3 lower ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.1 mmol/L (55 mg/dL) within 24 h, Events, Time in h, AUC; Haak 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.1 mmol/L (55 mg/dL) at night (2300–0600 h) within 7 h; Events, Time in h, AUC; Bolinder 2016

1 randomised very serious a not serious not serious serious c,d none 119 119 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <3.1 mmol/L (55 mg/dL) at night (2300–0600 h) within 7 h; Events, Time in h; Haak 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose<2.5 mmol/L (45 mg/dL) within 24 h*, Events, Time in h, AUC; Bolinder 2016

1 randomised very serious a not serious not serious serious c,d none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose<2.5 mmol/L (45 mg/dL) within 24 h*, Events, Time in h, AUC; Haak 2017

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 457 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Relative Absolute Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations FGM SMBG studies (95% CI) (95% CI)

1 randomised very serious a not serious not serious serious c,d none 149 75 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <2.5 mmol/L (45 mg/dL) at night (2300–0600 h) within 7 h*, Events, Time in h; AUC; Bolinder 2016

1 randomised very serious a not serious not serious serious c,d none 119 119 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <2.5 mmol/L (45 mg/dL) at night (2300–0600 h) within 7 h*, Events, Time in h; Haak 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <2.2 mmol/L (40 mg/dL) within 24 h, Events, Time in h; Bolinder 2016

1 randomised very serious a not serious not serious serious c,d none 119 119 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Glucose <2.2 mmol/L (40 mg/dL) within 24 h, Events, Time in h; Haak 2017

1 randomised very serious a not serious not serious serious c,d none 149 75 - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Quality of life, Patient satisfaction, Total treatment satisfaction, Perceived frequency of hyperglycaemia, Diabetes distress, Hypoglycaemia fear behaviour, Worry scores, Bolinder 2016 IMPACT TRIAL

1 randomised very serious a not serious not serious very serious b,c none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Patient satisfaction, Haak 2017

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 458 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Relative Absolute Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations FGM SMBG studies (95% CI) (95% CI)

1 randomised very serious a not serious not serious very serious b,c none - 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ Quality of life Patient satisfaction, Total treatment satisfaction score , Oskarsson 2018

1 randomised very serious a not serious not serious not serious none 78 70 - MD 6.4 higher ◯◯ trials (4.4 higher to LOW 8.4 higher) ⨁⨁

Perceived frequency of hypoglycaemia , Oskarsson 2018

1 randomised very serious a not serious not serious not serious none 78 70 - MD 0.6 lower ◯◯ trials (1.1 lower to LOW 0.2 lower) ⨁⨁

Perceived frequency of hyperglycaemia , Satisfaction with treatment , Oskarsson 2018

1 randomised very serious a not serious not serious not serious none - MD 1.2 lower ◯◯ trials (1.7 lower to LOW 0.7 lower) ⨁⨁

Total core scale score , Social worry , Diabetes worry , Impact of treatment , Total DDS score , Emotional burden subscore , Physician distress subscore , Regimen distress subscore , Interpersonal distress subscore , Behavioural subscale , Worry subscale, Oskarsson 2018

1 randomised very serious a not serious not serious serious e none - MD 0.1 lower ◯◯◯ trials (0.2 lower to 0 VERY LOW ) ⨁

MDII and CSII treatment, Changes in HbA1c, (mmol/mol) (Bollinder 2016; T1 DM) (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c none 119 119 - MD 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 459 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with diabetes DM with insulin

Certainty assessment № of patients Effect

Certainty Importance № of Relative Absolute Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations FGM SMBG studies (95% CI) (95% CI)

MDII and CSII treatment, Changes in HbA1c, (%) (Bollinder 2016; T1 DM) (follow up: 24 weeks) (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c none 119 119 - MD 0 ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ MDII and CSII treatment, Changes in HbA1c, (mmol/mol) (Haak 2016; T2 DM) (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c none 149 75 - MD 0.3 higher ◯◯◯ trials (0 to 0 ) VERY LOW ⨁ MDII and CSII treatment, Changes in HbA1c, (%) (Haak 2016; T2 DM) (follow up: 24 weeks)

1 randomised very serious a not serious not serious serious c none 149 75 - MD 0.03 ◯◯◯ trials higher VERY LOW (0 to 0 ) ⨁ CI: Confidence interval; MD: Mean difference

Explanations a. At least one domain judged with high risk of bias. b. Only P value available c. No CI available d. No data for Standard deviation e. CI oncludes both positive and positive (0) value

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 460 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Applicability tables

Table A10: Summary table characterising the applicability of a body of studies

Domain Description of applicability of evidence Population The population (mainly adults) included in RCTs is representative of patients usually included in such clinical trials (T1 and T2 DM patients). Majority were T1 DM patients and their results should not be appied to T2 DM patients. Baseline characteristics show that the studies included similar groups of patients. There were a small number of paediatric patients only (1 RCTs) and no RCTs in pregnancy related to medical devices under assessment. The studies varied by patient age and DM type as well as by inclusion criteria. Intervention The studies varied by rtCGM medical devices or FGM. Majority of RCTs are related to different rtCGM medical devices in comparisons to SMBG. Only one head-to-head study are published comparing the rtCGM and FGM medical devices. Only 2 RCTs are published related to one FGM medical device, in comparisons to SMBG, one in T1 and one in T2 DM patients. Comparators Most commonly used CE marked approved comparators – SMBG medical devices – were used in majority of RCTs included in this assessment. Only one head-to-head study are published, as already mentioned above. Outcomes The choice of outcomes is representative (HbA1c changes from baseline to the end of the study; time spent in the target glycemic range; time spent in hypoglycaemia; time spent in hyperglycaemia; hypoglycaemia and severe hypoglycaemia events; patient-reported outcome, quality of life (QoL) measures and user satisfaction), according the clinical guidelines and majority of RCTs reported these outcomes. But outcomes measures varied a lot among studies resulting in heterogeneity which not allowed MA on these outcomes (MA was done only on one outcome, HbA1c change from baseline to the end of the study, pooling the data from 2 RCTs DIAMOND and GOLD (25, 29). Mortality was not specified as outcome or reported in any of RCTs included in this assessment. Hypoglycaemia was reported differently between studies as well as measures for user satisfaction and QoL. Pa- tients were followed up from 8 weeks to 12 months; different time period of follow-up also do not allowed pooling of results related to different outcomes. Setting RCTs included patients worldwide in outpatient setting which is representative for the expected use.

Mean and standard deviation estimation for Forest plots visualization Due to (often large) heterogeneities between populations (e.g. age, type of insulin regimens, hypoglycaemia unawareness etc.), varieties of devices assessed and different outcomes measured, meta-analyses were not suited/justified, except for one case that included two studies (CGM vs SMBG on change in HbA1c %). Hence, to help in capturing trends of estimates along different studies, we have produced F-plots without pooling data.

As only means (SD) and MD (SE or 95% CI) values can be used in the RevMan tool generating F-plots, we made estimations of means, SDs and 95% CIs using the approximation method as described by Wan et al. (18)

Moreover, some of the studies reported percentage time in various glycemic ranges instead of actual time (minutes) spent in the range. In these cases, we calculated number of minutes assuming that 100% corresponded to 24 hours, i.e. 1,440 minutes.

Notably, as all final estimates of effect reported in each study, i.e. MDs but mostly differences in adjusted means MD, were directly plotted into RevMan, some of the intermediary data (e.g. means of each arms) varied to some extent from the value indicated in the articles, as RevMan is using different aligorithms and ormulas.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 461 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

F-plots were exclusively generated for visualisation purposes showing trends, In the few cases baseline values were not reported, we assumed baseline values to be equal.

Finally, although F-plots could basically have been generated for all outcomes, we considered that only the ones presented here would be useful to show trends of effect, as for other outcomes, estimates were either indicating the same (just using another unit) or estimates could not be gathered in one and same F-plot, thus not helpful in indicating any overall trend among the different studies).

Below are presented estimates of means and standard deviations for outcomes that were originally reported using nonparametric summary statistics. The method used to estimate the means and standard deviations is described by Wan et al. 2014 (18), which develops on previous methods described.

Note that while the method used was developed for use in meta-analysis, the use of means and standard deviations that have been estimated from nonparametric summary statistics will intro- duce some degree of error into a meta-analysis. In particular, authors typically report nonparametric summary statistics because they think parametric summaries (susually the mean and standard deviation) might be misleading. If the purpose of the meta-analysis is to accurately estimate an effect, then some caution is warranted. However, if the purpose of the meta-analysis is to test a hypothesis (i.e., is the effect of the intervention statistically significantly different from control), and the effect size is large (e.g., as concluded by the opriginal studies), then less caution is warranted.

Outcome: time spent in normoglycemic range per day (minutes) rtCGM vs SMBG (at follow-up)

Estimated Estimated Estimated Study (arm) Median RL RU n Mean SD 95% CI Beck (T2) 2017 (MDII) 882 647 1077 74 651.5 89.9 (631, 672) (Intervention) Beck (T2) 2017 (MDII) 836 551 965 72 588.0 87.0 (568, 608) (Control) Outcome: time spent in hypoglycemic range per day (minutes) < 3,9 mmol/l or 70 mg/dl rtCGM vs SMBG (at follow-up)

Estimated Estimated Estimated Study (arm) Median LQ UQ n Mean SD 95% CI Beck (T1) 2017 (MDII) 43.0 27.0 69.0 105 46.3 31.6 (40.3, 52.4) (Intervention) Beck (T1) 2017 (MDII) 80.0 36.0 111.0 53 75.7 57.2 (60.3, 91.1) (Control) Beck (T2) 2017 (MDII) 4.0 0.0 17.0 74 7.0 12.9 (4.07, 9.93) (Intervention) Beck (T2) 2017 (MDII) 12.0 0.0 34.0 72 15.3 25.7 (9.39, 21.3) (Control) Heinemann (T1) 2018 23.9 12.9 54.5 75 30.4 31.4 (23.3, 37.5) (MDII) (Intervention) Heinemann (T1) 2018 92.2 51.8 172.6 66 105.5 91.6 (83.4, 128) (MDII) (Control)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 462 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Outcome: time spent in hypoglycemic range per day (minutes) < 3,0-3,3 mmol/l or 50-60 mg/dl rtCGM vs SMBG (at follow-up)

Estimated Estimated Estimated Study (arm) Median LQ UQ n Mean SD 95% CI Beck (T1) 2017 (MDII) 20.0 9.0 30.0 105 19.7 15.8 (16.6, 22.7) (Intervention) Beck (T1) 2017 (MDII) 40.0 16.0 68.0 53 41.3 39.6 (30.7, 52) (Control) Ruedy (T1+T2) 2017 3.0 0.0 15.0 58 6.0 11.4 (3.07, 8.93) (MDII) (Intervention) Ruedy (T1+T2) 2017 4.0 0.0 24.0 50 9.3 18.3 (4.26, 14.4) (MDII) (Control) Heinemann (T1) 2018 3.8 1.1 11.9 75 5.6 8.2 (3.75, 7.45) (MDII) (Intervention) Heinemann (T1) 2018 32.9 13.1 83.9 66 43.3 53.7 (30.4, 56.2) (MDII) (Control) Ly (T1) 2013 (CSII) 32.4 19.4 79.9 46 43.9 46.3 (30.5, 57.3) (Intervention) Ly (T1) 2013 (CSII) 71.3 34.6 127.4 49 77.8 70.9 (57.9, 97.6) (Control)

Outcome: time spent in hypoglycemic range during night from 22.00-6.00 (minutes) < 3,0- 3,3 mmol/l or 60-70 mg/dl rtCGM vs SMBG (at follow-up)

Estimated Estimated Estimated Study (arm) Median LQ UQ n Mean SD 95% CI Ly (T1) 2013 (CSII) 11.5 1.9 25.4 46 12.9 18.0 (7.74, 18.1) (Intervention) Ly (T1) 2013 (CSII) 29.8 20.2 47.5 49 32.5 20.9 (26.7, 38.3) (Control) Outcome: time spent in hyperglycemic range >10 mmol/l or 180 mg/dl per day (minutes) rtCGM vs SMBG (at follow-up)

Estimated Estimated Estimated Study (arm) Median LQ UQ n Mean SD 95% CI Beck (T1) 2017 (MDII) 638 503 807 105 649.3 228.5 (606, 693) (Intervention) Beck (T1) 2017 (MDII) 740 625 854 53 739.7 174.5 (693, 787) (Control) Beck (T2) 2017 (MDII) 549 353 789 74 563.7 329.7 (489, 639) (Intervention) Beck (T2) 2017 (MDII) 571 422 883 72 625.3 348.7 (545, 706) (Control) Outcome: time spent in hyperglycemic range >13,9 mmol/l or 250 mg/dl per day (minutes) rtCGM vs SMBG (at follow-up)

Estimated Estimated Estimated Study (arm) Median LQ UQ n Mean SD 95% CI Beck (T1) 2017 (MDII) 223 128 351 105 234.0 167.6 (202, 266) (Intervention) Beck (T1) 2017 (MDII) 347 241 429 53 339.0 143.3 (300, 378) (Control)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 463 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Ruedy (T1+T2) 2017 89 37 208 61 111.3 129.8 (78.8, 144) (MDII) (Intervention) Ruedy (T1+T2) 2017 179 83 316 53 192.7 177.6 (145, 240) (MDII) (Control) Beck (T2) 2017 (MDII) 105 37 246 74 129.3 158.0 (93.3, 165) (Intervention) Beck (T2) 2017 (MDII) 118 48 288 72 151.3 181.6 (109, 193) (Control) Outcome: change in event rate of hypoglycaemia per 24h rtCGM vs SMBG

Estimated Estimated Estimated Study (arm) Median LQ UQ n Mean SD 95% CI Riddelsworth (T1) 2017 -0.1 - 0.1 103 -0.1 0.2 (-0.104, - (MDII) (Intervention) 0.2 0.0164) Riddelsworth (T1) 2017 0.0 - 0.1 53 0.0 0.2 (-0.0949, (MDII) (Control) 0.2 0.0282)

Forest plots including all values plotted into RevMan:

Change in HbA1c % rtCGM vs SMBG

Change in HbA1c % FGM vs SMBG

Time in normoglycaemia 70-180 mg/dl (or 3.9 -10.0 mmol/l) rtCGM vs SMBG

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 464 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time in normo FGM vs SMBG

Time in hyperglycaemia >180 rtCGM vs SMBG

Time in hyperglycaemia >250 (or 13.9 mmo/l) rtCGM vs SMBG

Time in hyper >240 (or 13.3 mmol/l) FGM vs SMBG

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 465 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Time in hypoglycaemia <70 rtCGM vs SMBG

Time in hypoglycaemia <50-60 (or 2.8-3.3 mmol/l) rtCGM vs SMBG

Time in hypo <70 FGM vs SMBG

Time in hypo <55 FGM vs SMBG

Patients with IHA & previous severe hypo episodes: Time in hypo and hypo/severe hypo events

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 466 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Patient Involvement

Diabetes Scotland has a formal membership offer – people don’t have to be a member to access services and resources; on average the Diabetes Scotland Helpline handles 12/15 calls and 20 - 25 emails per day. Formal membership - on April 2016 Diabetes Scotland figures were as follows: 9,000 individual membership (PLWD); 39 local group members and 4 family groups (membership list vary from 30 – 300 per group); 284 HCP members; 252 Primary care / AHP.

Impact of condition – Diabetes mellitus treated with insulin In Scotland there are 291,981 people living with diabetes of which 30,899 have Type 1 (T1) insulin dependent diabetes, and 257,728 have Type 2 (T2) this represents 5.4 per cent of the population in Scotland (Scottish Diabetes Survey 2016).

Diabetes is a serious, life-long health condition that occurs when the amount of glucose in the blood is too high because the body can’t synthesise it properly. If left untreated, high blood glucose levels can cause serious health complications, even death. There are two main types of diabetes: Type 1 and Type 2. They’re different conditions, with differing modes of onset.T1 has a rapid onset and is an auto immune response. T2 has a much more complex, multifactorial manifestation and usually has a slower, more insidious onset. Both conditions are serious and need to be treated and managed properly with a range of interventions. Those on injectable therapies such as insulin, are required to inject multiple times a day and recommended to test blood glucose levels between 4 – 10 times a day - on waking, before and after meals, before and after physical activity, before driving, before going to bed. If the person develops a mild illness such as cold or gastric upset, this not only impacts on blood glucose control quite markedly, they are also required to increase the frequencies of blood glucose testing. For those already struggling with needle phobia this can be an enormous struggle and cause further emotional and mental distress. One of the most challenging aspects of living with diabetes, primarily for those who are insulin dependent, is the prevention and management of hypoglycaemia (hypos), especially nocturnal (night time) hypos. If left untreated severe hypo can lead to seizures, coma, lasting neuro deficits, and even death. The fear of, and anxiety associated with, hypos is one of the most talked about topics covered in calls to Diabetes UK Helpline, in focus and support groups, and online forums. Hypos are distressing not only for the person living with the condition but also for parents, spouses and family members. They can be difficult and distressing to manage, the person may become aggressive, irritable, unco-operative, unsteady, confused etc.

An additional concern for people living with diabetes (PLWD) and parent/carers is the potential loss of hypo awareness - the ability to recognise early warning signs that their blood glucose levels are falling. This frequently results in individuals testing excessively, running their blood glucose level higher than advised targets – thus increasing the risk of micro and macro vascular damage and serious lasting complications such as blindness, stroke, kidney disease, neuropathy and amputation; and/or withdrawing from social events/ interaction because of the fear of hypoglycaemic episodes and the possible consequences (loss of control, hospital admission, injury through falling, discrimination and stigma). The impact and time required to recover from a severe hypo varies from individual to individual. On average it can take several hours before the person is able to resume normal daily tasks. Recovery from nocturnal hypo can be longer due to disturbed sleep and length of time to reach safe target levels. In some instances the person may not be able to attend school, college or university, thus missing vital education; or, if employed, attend work. All of which can impact on attainment, future prospects, security of tenure of employment and close relationships due to financial insecurity.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 467 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

PLWD often feedback that the need to constantly finger prick to test blood glucose is painful and inconvenient. They report that they frequently face discrimination in the work place for doing blood tests in public which results in them limiting the number of times they tests their levels. PLWD require regular review appointments on a monthly 3, 6, 9 or 12 monthly basis, depending on the stability and complexity of their diabetes. Whilst diabetes is recognised under the Disability Discrimination Act, employers and education establishments frequently do not make reasonable adjustments. Individuals have reported being subjected to competency or disciplinary proceedings as result of periods of absence due to attending frequent diabetes appointments or seeking treatment/support for diabetes related complications. A survey in 2017 by Diabetes Scotland revealed that 65% of people with diabetes found it difficult to manage their condition at work. The survey revealed that 32 per cent of people felt it was ‘not very easy’ or ‘not at all easy’ to take time away during work to self-manage their condition with blood testing, taking medication etc. Furthermore, 12 per cent of respondents said they had been refused time off to attend a diabetes healthcare appointment.

Making positive choices is particularly important for people with diabetes. Healthy eating and exercise are the foundation of good diabetes management. Understanding the nutritional profile of foods and drinks, including the calorie, fat, sugar and salt content levels, is vital for day to day and long term management of diabetes. This information is often limited or missing from menus in restaurants and other out of home food outlets. This lack of information often prohibits or stops people from going out for meals with family and friends and can lead to social isolation.

Living with diabetes is difficult. People face managing both the short and long term complications on a daily basis, which is frequently described as ‘relentless’, ‘life limiting’, ‘stressful’, ‘unpredictable’ and ‘exhausting’.

“Managing an invisible condition can be isolating, and other people do not always understand what it’s like to live with diabetes.” (Future of Diabetes: Diabetes UK 2017)

“Diabetes doesn’t just affect someone physically. The effect of varying blood sugar levels on mood – and the relentless need to manage the condition – affects your mental health.”

(Future or Diabetes: Diabetes UK 2017)

Diabetes is a complex condition. Optimum management, for instance administering insulin and testing blood glucose, and balancing insulin and food intake, requires skills in numeracy and dexterity. For young, pre-school and early years children the day to day management of the condition is largely reliant on the intervention of the main care giver/parents. Parents frequently report feelings of guilt, denial, anger and anxiety at diagnosis – why did this happen? Is it my fault? How will I/we cope?

Families find themselves faced with unexpected hospital stays, constant medication adjustments and lifestyle changes in order to cope with health complications from diabetes. Simple things such as going out, family events, and holidays require more planning which can lead to frustration, anger, resentment and more stress. Family dynamics can change due to one family member dominating the attention of others, causing feelings of jealously, abandonment etc. The strain of managing a child with diabetes, especially in the early months and years after diagnosis can cause breakdown of parental relationships. Similarly, a parent diagnosed with diabetes may require help from his or her children thus altering the established family roles.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 468 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Parents of young children frequently report interrupted sleeps for protracted periods (years) because they have to check their child’s blood glucose levels during the night to avoid life threatening hypos and, where necessary, take remedial actions such as waking the child for blood testing or treatment. For single parents this strain can be enormous, it can affect the parent’s ability to attend work, sustain employment etc.

“It is difficult to stress the horror we went through at the beginning of diagnosis when my daughter had to finger prick. She thoroughly hated it. It was the worst time. She would scream, cry and run away from us…”

(Mother of a child with Type 1).

People living with diabetes (PLWD) do not get a day off from the condition. Those dependent on insulin are required to undertake self-monitoring of blood glucose (SMBG) and self-manage 24 hours a day, 365 days a year. Not everyone with diabetes comes to terms with the fact that they are living with a long term condition, or are able to sustain the intense daily vigilance required to keep healthy. This need for constant vigilance can lead to ‘diabetes burnout’, anxiety, obsession and eating disorders such as anorexia and diabulimia.

Not taking the correct amount of insulin may occur for many reasons such as fear of hypos or under-estimating carbohydrate intake. However, when it is associated with weight control and happens over a prolonged period this is called diabulimia. Research which emerged last year from Toronto suggested that 60% of females with Type 1 will have experienced a clinically diag- nosable eating disorder by the age of 25. In the general population girls/women are 10 times more likely to experience eating disorders than men/boys.

Eating disorders in T1 diabetes contribute to poor diabetes control, rapid development of secondary complications such as retinopathy and neuropathy, and increased rates of severe hypo, hyperglycaemia leading to Diabetes Ketone Acidosis, and mortality compared with a person with T1 without eating disorders. Standard management strategies for eating disorders (avoid calorie counting, adopting more flexible eating patterns etc.) can prove problematic for those with dual diagnosis of T1 and anorexia/diabulimia. With the right psychological support, access to appropriate education and technology (insulin pumps, CGMs and Flash GM), diabulimia can be effectively managed and overcome (Diabetes Medicine 2018 Volume 35).

Diabetes (T1 and T2) can have an impact on the wellbeing of mothers during pregnancy and early motherhood. During pregnancy, blood glucose control can be very labile. Pregnant women can experience unpredictable swings in blood glucose levels. To ensure the health, and reduce the risk of diabetes related complications for both mother and baby, it is important that the mother maintains blood glucose levels within a safe target range. Providing devices such as SMBG meters, CGM and/or Flash CGM pre and post-partum has been shown to have significant impact on the health of both mothers and neonates (Diabetes Medicine 2018 Vol 35 Number 2) .

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 469 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Individuals with limited dexterity can encounter numerous challenges from inability to manipulate insulin vials, syringes or cartridges, to handling blood glucose strips, using glucose meters, and self-injections. Those with limited visual acuity can face similar challenges requiring meters with larger text display windows and/or audible voice facilities. Poor visual acuity can make it difficult for the person to administer anti-diabetes treatments, making them reliant on third party assistance. For individuals with language, literacy and numeracy disabilities, the need to understand carbohydrate counting and ratios to match intake of carbohydrates to insulin dosage often proves problematic and difficult to master.

With regard to all the above, providing information in more accessible mediums together with the use of appropriate, accessible technology, and active intervention and support from health care staff, can go a long way to help independence and self-management of the condition, improving quality of life for all PLWD.

Experiences with currently available health interventions: Self-monitoring blood glucose (SMBG) medical devices Self-monitoring of blood glucose (SMBG) is an integral component in the management of insulin dependent diabetes. As part of the day-to-day routine it can help with necessary lifestyle and treatment choices; it helps to monitor for signs of hypo or hyperglycaemia and prevent any long- term complications from developing. It is therefore essential that the individual is taught how to carry out a test properly from onset as poor technique may lead to incorrect results which could lead to inaccurate medication.

Despite being an essential component of effective diabetes management, PLWD have reported that access to essential meters and strips have been subjected to restrictions across some parts of UK. Limiting access to essential test strips on the grounds of fiscal policy is both dangerous to patient safety and a false economy. Achieving even a small percentage reduction in blood glucose levels can have a marked reduction in the risk of developing secondary complications and improve the quality of life for PLWD. Access to appropriate SMBG devices and sufficient test strips is essential to achieving this.

Re: restriction of testing strips:

“I had to reduce my testing. Hypo awareness has depleted again.

(Diabetes UK Survey 2017)

“I am constantly worried we’ll run out. I get cross with my daughter if she wastes a strip by dropping it or not producing enough blood”.

(Diabetes State of The Nation 2016: Diabetes UK)

Frequent testing can be painful, inconvenient and difficult to achieve due to the person’s daily work routine. For example it is not always practical/easy for a teacher to SMBG in front of pupils, leave in the middle of class to SMBG, or take appropriate action if their blood glucose levels are rising/falling. It is not always easy for people to wash their hands in order to test when they are on the move.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 470 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

There are devices available such as insulin pumps, Continuous Glucose Monitoring (CGM) and Flash Glucose Monitoring (Flash GM) to assist diabetes management.

Continuous Glucose Monitoring (CGMs) work by sensing glucose level in the fluid between tissues in the cells. CGM measure the levels every few minutes, sends results to receiver and provides regular feedback, alerting the person to changes in glucose levels. The device provides data on how blood glucose may vary in response to everyday tasks such as sleeping, after meals, physical activity or if unwell. CGM is potentially useful to anyone with diabetes especially those on multiple daily injections and insulin pump users. Some devices are integrated with the insulin pump with the capacity to suspend insulin delivery if the person’s glucose levels are falling. Results with CGM are not 100% accurate and regular blood glucose testing is still required.

“I’m woken by an alarm if my levels are low… My life is totally different, I can make plans… I never want to go back to worrying that I might go to bed and not wake up” (Female, age 24, living with diabetes)

“Technology has as made my life with T1 diabetes better. It gives the freedom and confidence to manage my diabetes and enjoy life”

(Student with Type 1 Diabetes)

Flash Glucose Monitoring (Flash GM) is a technology that has proven to be popular with PLWD. It comprises a small sensor worn on the skin and a scanner. Users access results by scanning the sensor. Flash GM shows if blood glucose level are rising or falling, allowing the user to take remedial action sooner to avoid hypos or hyperglycaemia. Flash GM can be used for short-term investigations into an individual’s glucose levels where an individual is having difficulty in managing their condition or for longer diabetes management. Those utilising the device have said it is “life changing”.

Unlike CGM, the Flash GM does not have a built alarm facility to alert the users to highs and lows or suspend insulin delivery. Flash GM would not be appropriate in those who have irreversibly lost their hypoglycaemia awareness, which means they cannot recognise when their glucose levels are low. CGM devices which provide alarms are more appropriate in this cohort of PLWD.

Users are still required to do ‘regular’ SGBM, especially if they are a driver (Flash GM is not recognised by Driving Vehicle Licensing Authority (DVLA) in the UK.

Access to SMBG devices, for those who are not using insulin, is limited. SGBM for defined period (immediately following diagnosis, prior to treatment changes etc.) can help the person to see, in real time, the impact differing foods, levels of physical activity and routine daily activities have on their individual blood glucose control. SMBG can provide reassurance, empower people to take control of their healthcare, and understand the relationship between their feelings and blood glucose readings. The benefits and need for testing blood glucose levels should be discussed, reviewed regularly, and any changes agreed by the person with diabetes and their GP or specialist team.

Flash GM has utility for both people with T1 and insulin dependent T2 diabetes.

While Flash GM (FreeStyle Libre®) has been approved for prescription on NHS it is not currently available in all areas of the UK. Many people have chosen to self-fund the technology but this is not an option for others, particularly those who live in areas of multiple deprivation or are on low

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 471 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin incomes. Research indicates that those living within socially deprived income groups are 2.5 times more likely to develop diabetes.

“We are lucky to be able to afford the Libre, it has made a huge difference to my daughter’s lifestyle and her diabetes care. It’s a travesty that those who can’t afford it don’t have access to such a life changing technology.” (Mother of a child with Type 1 diabetes)

In a recent survey (Diabetes UK 15 Health Care Essential Survey 2016) 28% of those who took part said they had encountered difficulties accessing medication or equipment needed to manage their diabetes. Particularly test strips, insulin pumps and continuous glucose monitoring.

Flash Glucose Monitoring (Flash GM), is a new technology that is potentially life-changing for many people living with diabetes. Those utilising the device have indicated that they can see a clear ‘direction of travel’ in terms of understanding how their diabetes affects their blood glucose levels. They have described the device as ‘discreet’ allowing them to scan blood glucose levels easily in public without drawing too much attention and/or negative responses. People indicate they feel more confident in managing their diabetes. People are relieved not to have to do as many painful finger prick tests. Carers/parents can monitor blood glucose levels during activities/sleep without disturbing/waking the child/PLWD unnecessarily

“Without a having a clear picture of what my glucose levels are doing, I’m essential poking around in the dark. Even with a finger prick test it’s a moment in time rather than a direction. Flash gives an interactive way of managing my diabetes. I can watch things unfolding and react accordingly. I am now trying to avoid events rather than avoiding recording the event. It takes away the stress and guesswork around testing and management of my condition”.

(Male who’s living with type 1 diabetes: Diabetes UK Big Conversation 2017).

“I was having problems with high and low glucose levels, with symptoms too, despite the blood glucose meter saying I was in target range. The Flash has helped resolve these issues.” (Female living with T2 diabetes)

“This little piece of tech has significantly changed how I manage my diabetes. I now test multiple times a day – typically more than 20 – yet it is far easier and more discreet than finger-prick testing. And I’m able to keep far closer to target levels… not only by reacting to the current blood glucose reading, but more importantly to the trend arrow.” (T1 diabetes using Flash GM since 2016) “My son is very active. He does judo, tennis, football, cricket, swimming and helps to walk the family dog too. At school he is physically and mentally active. No day is ever the same. The Flash device has allowed my son his freedom and independence to test. The arrows are the most important for us. At six years old my son can read his levels and immediately understand how to take care of himself. We have a better understanding of how to manage his diabetes with his active lifestyle, so he can join in and experience a ‘normal’ childhood.” (Mother of a child with Type 1 diabetes)

Adults with diabetes report that the most useful aspect of the Flash GM device (Freestyle Libre®) is the directional arrow that tells them whether their blood glucose is rising or falling. A key aspect of CGM is the alarm that warns of hypo or hyperglycaemia for people who have lost hypo awareness.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 472 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

In collaboration with diabetes professionals and The Juvenile Diabetes Federation (JRDF), Dia- betes UK published extensive guidance on who would benefit from Flash GM. They concluded that Flash GM devices should be made available to any adult or child with Type 1 diabetes and to people with other forms of diabetes when intensive insulin therapy becomes necessary because of severely reduced pancreatic function.

Additional information Those who use SMBG devices should make sure they have access to and follow the guidelines for usage, checking equipment for any signs of damage or breakage regularly. Choice of SMBG devices and monitoring equipment should be based on individual need, and ease of use. This involves discussing options, making agreed joint decisions based on individual need. Consultation should discussion use and frequency of testing, and agree targets for blood glucose levels. Those commencing SMBG, CGM or Flash GM should receive structured education to ensure they can maximise their use and benefit from such technology. Healthcare professionals working with people living with diabetes, require training on use of technology and will need to understand how to use and interpret glucose data in order to support PLWD. People who use a Flash GM monitor still require to test regularly and will need adequate supplies of test strips in order to finger prick test: If glucose levels are changing rapidly If scanned levels do not correspond with physical symptoms; If reader indicates low glucose To meet current essential Driving and Vehicle Licensing Authority requirements. Flash GM should not be considered as an alternative to Continuous Glucose Monitoring (CGM). People who meet the guidance and criteria for the use of CGM in people with Type 1 diabetes, as set out clearly in clinical guidance, should still be provided with CGM.

Key messages Hypoglycaemia (severe/moderate) is one of most significant and challenging aspects associated with diabetes (T1 and T2). Hypoglycaemia has an enormous impact on the person’s quality of life. There is related cost not only to NHS in terms of resources (ambulance call outs, hospital admission, increased length of stay associated with poor management when admitted for non-diabetes related reasons) but also to the individual in terms of stress and anxiety caused, working days lost, missed education and effect on family relationships.

Current medical devices for self-monitoring blood glucose (SMBG) provide just a snap shot of blood glucose at a single point time, results obtained are dependent on the skills/technique and circumstances of the users. Devices can produce aberrant readings, caused by defects in calibration or test strip accuracy (finite shelf life), and lack of optimum conditions for testing (e.g. handwashing facilities). There is also the potential for the misinterpretation of readings due to inade- quacies in size and configuration of display windows.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 473 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Flash technology is new but the experience of users indicates that when integrated in normal/usual care the clinical and quality of life benefits are significant. These benefits include: Re- ducing the need for painful and inconvenient finger prick glucose monitoring; Helping people to better manage their diabetes and engage people with self-management of their condition; Reduc- ing stress and anxiety for people with diabetes and their families; Reducing hypoglycaemia and increasing time in optimum range; Improving HbA1c to reduce the risk of developing costly complications.

Further comparative study comparing Flash GM and traditional CGM is required to help us understand the advantages of Flash GM and its role in diabetes management, both T1 and T2. Technology associated with the management of diabetes is developing fast and constantly changing. It is essential that the Scottish Government, Health Improvement Scotland and Scottish Health Technology Group are vigilant and horizon scan this area to ensure that diabetes services are future proofed, fit for purpose, and enable those living with diabetes to effectively self- manage, avoid unnecessary complications and live well with the condition. Scotland requires an E- Health Strategy that is robust, flexible and ensures access to devices, e-health learning pro- grammes (for HCP and PLWD) and digital platforms that integrate and capture patient information across community and acute based services (SCI-Diabetes) to provide seamless diabetes care for all.

International Diabetes Federation European Region Patient representatives were speaking on behalf of IDF Europe, but these are individual experiences and might not be representative for the entire European region.

Impact of condition – Diabetes mellitus treated with insulin Not all people living with diabetes are affected the same. However, a great number of individuals find it difficult and exhausting. Growing up with type 1 diabetes, there was not much awareness about the condition, and everyone kept it a secret. Some patients are ashamed of the diagnosis and always hide their treatment from their loved ones. This may have a negative effect in their daily living. For example, before I accepted the condition, I always had a problem with relationships, socialising, feeling like I fit in and more. There are also a number of other individuals who went through the same thing. Knowing that there are a great number of risks and complications make it more difficult to accept the condition and have control. Contrary to some people’s belief, diabetes is not just diet and exercise. It involves a lot more! This always frustrated me, when I ate healthy and exercised on a daily basis, yet my bloods where almost never perfect. Yes, diet and exercise are important factors, however, type 1 diabetes comes with a lot of anger, frustration, anxiety, stress and much more. These all taking a toll on one’s diabetes. This may result in the need for more insulin.

As one might know, weight gain is a common side effect for people who take insulin. This can have a greater stress on the person with diabetes, especially young girls. Gaining weight may be very stressful on the person, and may want to give up entirely causing more damage to the person’s health. women with Type 1 have close to two and a half times the chance of developing an eating disorder. One of these includes Diabulimia which is a life-threatening combination and the unhealthy practice of withholding insulin to manipulate or lose weight.

Diabetes treated with insulin affects many areas in a person’s life. Those areas include their social life, ability to engage in a specific activity or task at a certain moment due to high/low bloodsugar. Stigma definitely exists, but it’s impact is different by countries. Children with diabetes can experience isolation, depression, unwillingness to interact with peers because of the fear that they might

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 474 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin not be understood and their situation taken seriously. Diabetes requires a different, new lifestyle and family dinamic, where everyone needs to play an active role. A number of families who have a relative with type 1 diabetes share the same emotionally charged reaction to this chronic illness when they were informed about the diagnosis. It adversely affected them during the first few weeks/months after the diagnosis. They were traumatised by getting to know the diagnosis of diabetes, expressing strong disbelief and heartbreak. Some of the fears that they face are related to the stigma diabetes carries within society that sometimes drive the person to avoid informing others such as peers and school staff about it. There is also the constant fear experienced by family members of whether the person was managing his or her condition properly and the complications such as physical injury that could arise if it is not controlled. Managing their child’s diabetes is a great responsibility and constant care needs to be given. As parents, they want their children with type 1 diabetes to be independent and learn to self-manage however, their fear and anxiety on the condition never fades away. Most families want to give the best to their child, whether it is insulin, new methods of glucose reading or others, This may sometimes put the family in a financial burden since all diabetes medication are costly. This can result in other stresses which some families cannot handle.

There is a great burden for care-givers. First of all, they need to be available most of the time (if the patient is a child), they have to implement more daily tasks into their routine and that can be very draining, both physically and mentaly. They also have to be aware that at all times acute diabetes complications e.g. hypoglycaemia can be deadly at any moment.

Diabetes is never easy, whether you are a man, woman, child, elderly, people with disabilities or others. Having other conditions may make it more difficult for him/her to control their diabetes which may cause several issues. During my voluntary work, I have noticed that women have a harder time accepting the condition than men. Women have other factors such as pregnancy complications, menstruation (which causes a disruption of the glucose levels), social factors (such as not feeling comfortable in their own skin) and more. The lack of awareness in our societies may also cause issues in managing diabetes mellitus. Numerous time people pass comments such as, ‘you get diabetes from eating too much sugar’, ‘it is not a serious illness, not like cancer’, ‘but you do not look like you have diabetes’, ‘have a good diet and exercise, then you can stop taking insulin’, ‘are you sure you want to do your shot here?’, ‘should you really be eating that?’, ‘I would die if I had to give myself shots’ and much more. Such comments make the person with diabetes either feel angry or sad. On the rare occasions, they feel empowered to educate others.

No insulin-treated PWD has it easy. But some groups - such as mothers with diabetes that are active in the work force face issues of not taking enough care of themselves since they are taking care of their whole nuclear and often even, extended families. Children and adolescents have a tendency to hide from peers while injecting insulin or checking bloodsugar, since they don’t want to be labeled abnormal by others.

Experiences with currently available health interventions: Self-monitoring blood glucose (SMBG) medical devices

Currently, the continuous glucose monitoring system is gaining a number of users in Malta. This device has improved many lives of people with type 1 diabetes. This is because it gives you a clear picture of how your bloods are 24/7 without having to prick your finger multiple times a day. It does not take as much time as the normal meters therefore, people who are busy at school, work or in their daily lives, can check their blood whilst walking to their next class or meeting. It is discreet and painless. One can immediately know when a low or high is coming and can treat

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 475 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin accordingly. This device also improves the HbA1c since a number of people start checking more often and do their best to keep their bloods in the correct range.

However, since it monitors interstitial fluid glucose levels, it is not always accurate. Therefore, sometimes you still have to check with a normal blood glucose reader. But most of the time, it is accurate. Self-monitoring devices present a huge advantage in diabetes self-management. They provide you with more specific information at any given moment. That takes off a bit of the burden and fear of sudden changes in BG levels and the need to constantly prick your fingers. It can also be very useful for the care-giver who can be informed about the person’s status at all times even if they are not close by (most devices have this feature). In the end, they help you improve your self-management and directly prevent complications. The device is very easy to use and other persons with diabetes who use the device are all willing to help another set up the device and show them how it is used. The device is also easily covered by clothing, and if not it is a small sensor. Other people who do not know what it is usually think it is a device that helps you stop smoking. There where cases where others wanted to see what it is and took it of the person with diabetes. This caused the sensor to stop working, therefore the money spent to buy it was lost. Since there is no actual awareness about it, not a lot of people know what it is and usually do not ask. I have met some teenagers who do not wear it during the summer so that it is not seen by others. In my case for example, I still use it in Summer, however if there is a wedding/formal event coming up, I make sure that I do not have one for the date. Even though I know it is for my own good to use it, however, others may see it as an eye sore, especially when you are part of the wedding party. Most self-monitoring devices are quite easy to use, though the proccess of insertion for some can be somewhat complicated. The biggest barrier though are the costs of using such a system since in many countries they are not reimbursed by health insurance.

Further are listed: More control; Improved HbA1c; More readings; Less stress; More financial burden since it is an expensive device for families who do not afford; Not all devices have an alarm that advises you when blood sugar is going dangerously low/high which most parents seek for their children with diabetes; Some individuals are allergic to the components used forcing them to stop using the device; Fingers are no longer sore/numb from all the pricks done; Doc- tors/healthcare professionals have a clear picture of what insulin needs to be adjusted if needed; Carers are at ease since when in doubt (for example during the night) one can easily check without waking up the child; For example, if surgery is needed with general anaesthesia, the professionals can continuously check the blood glucose and have a clear picture whether the blood will be going low/high.

My personal experiences have been very positive. I have used the Dexcom G4® system and am currently using the Freestyle Libre® and they really make a huge difference in every day life. I can focus more on my daily activities and check my BG levels at any time in a matter of seconds. Though they can pose a problem e.g. when they are not accurate which can happen fairly often. The thing that I didn’t like about the Dexcom is the need to calibrate the device every 12 hours, but on the other hand it was usually more accurate than the non-calibrating Libre I am currently using. The biggest drawback of these systems is definitely the monthly cost since they are not covered by health insurance and that is the most important area that needs to be tackled in the future - making them more accessible.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 476 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

People with diabetes are only getting to know about these medical devices, therefore do not have the information/knowledge at hand. People may not like to have something stuck to them for a long period of time and may not know all the benefits there is to using this device. Also, many might believe that having this device limits them for example during exercise. The price is also a main factor. These are very expensive devices and with the cost of living continuously on the rise, many families may not afford such devices. Children and young adults with type 1 diabetes. Also expecting mothers, and people taking care of their family. Practically, everyone. This device can have a huge positive impact for the lives of people with diabetes. It gives a person, somewhat of a “normal” life. Investing in such devices, will decrease/prolong the number of complications. All patients could benefit from using the self-monitoring BG systems. But it could be most valuable to athletes and people with an active lifestyle. On the other hand for people with great BG fluctuations e.g. adolescents, pregnant women, “brittle” diabetes patients it would be precious to have the ability to know the BG levels at all times and act accordingly thus improving their management.

Additional information

The patient should be able the choose the treatment choice in collaboration with the endocrinologist. This especially goes for these systems e.g. do they want a pre-calibrated system or a one where you have to calibrate it yourself, is the ability to have alarms crucial or not etc. Patients should also go through an educational course at the start of using these devices. They must be aware that they make mistakes and to be taught what to do when a malfunctioning occurs.

Key messages

The current medical devices have been good for the past years, but with this fast-paced lifestyle of today, these medical devices are outdated. People treated with insulin need these new medical devices in order to have a better, more controlled life. There are thousands of people on these devices, and they are able to lead a “normal” life, without the fear of how their blood is, because in a simple second, they can check discreetly. They are great and should be accessible by all by making them reach different countries and affordable.

Focus groups - Croatia

Two Focus groups (with individual patients) were held in Croatia separately in March 2018 – one for adults (12 participants) and one for children with informal caregivers (7 children with 7 parents). All filled the EUnetHTA DOICU Form; none have any conflict of interest. Four to five-hours meetings were performed with predefined questions related to impact of condition; experience with currently available medical devices; experiences with, and expectations of, the medical devices being assessed; and additional information which patients believe would be helpful to the HTA researchers. In Summary: Adults: Some patients with long period of DM T1 or short period, on MDII or insulin pumps; intermittent use of rtCGM or FGM devices. Impact of condition; hypoglycaemia without symptoms the major problem for daily living and working; stigma, better quality of life with devices, problems with finger sticks, pain, sensation lost, financial aspects, problem of unexpected hyperglycaemia (some patients have daily significant glucose fluctuaction; importance of visible arrows and trend of glucose; variability of glucose valu- aes in different life cycles, much better control with CGM or FGM when big glucose oscillations are present, MD of utmost importance, SMBG with tests - not enough (only 4 per day), some

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 477 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin problems with device, visible, inconvenient in some situations, visible blood, not know the trend of glucose; problems on very low temperature outside, on the swimingpool (wronge value of glucose), indiscrete; not visible trends, stress from finger sticks, importance of education, importance of alarms, problems if sensors are lost, CGM or FGM for different periods of life, price should be lower, better regulation of glucose, better knowledge on food impact on glucose is provided; devices are visible, people must be educated and motivated, better sleep, beter regulation of insulin dose and precision; finger stics as short picture, CGM or FGM as film; normal life and quality, can be used more discreetly; must be affordable, life is much better controlled; malfunctioning coul occur; important for childer and youn people, with positive impact on caregivers; positivan impact and normal life, will prevent chronic complications; cost-benefit will be much better; experience with currently available medical devices; less stress Main messages: of utmost importance for patients with frequent hypohlycaemia or for people who are not aware of it; better disease regulation and less adverse events; more confidence in life and less fear on hypoglycaemia; time in range more important then HbA1c; important for better control of DM, prevention of acute and chronic complications; better quality of life; less stress with living with DM; more active life with more pfysical activities.

Children with parents: Some patients with long period of DM T1 or short period, on MDII or insulin pumps; intermittent use of rtCGM or FGM devices; some without CGM or FGM; Some expressed negative experience related to disease: without support of institutions, school, discriminations, could not go alone on school trips; parent must go with them; fears of hypoglycaemia; huge oscillations during the childhood; could not eat everything; education in school staff, children, alarms; problems with finger-sticks, FGM could be lost during the training or pool, no need to stick on nigh, just scan, better quality of life; life becomes easier; could not eat all kind of food; problems with SMBG, pain, disinfections of hands; problems with small amount on blood, several finger sticks needed, problems with waste, financial, organisational issues, medical devices could not be bought in Croatia, problems related to unavailability, non accessibility, finance, could be lost in water, need to be proper protection, trend of glucose is of utmost importance; hypo-hyper oscillations, no need for high number of finger-sticks, arrows and trend; education, communication, importance of insulin pumps; not proper accessibility; better QoL of children and parents; anxiety, stress; huge burden for all family; knowing when high or low is coming, proper reactions; not visible, easy to use; should be more accessible; the price, many families could not paid for such devices; positive impact and normal life; more freedom to children and parents, more peaceful, more autonomy, importance of education and motivation; life becomes easier and more quality, could see the trends and could react properly; Main messages: less finger sticks is needed; importance to see the trends to hypo or hyperglycaemia; better glucose control; no finger sticks during the nigh; important for better control of DM, prevention of acute and chronic complications; better quality of life of children and the whole family; better assesability of these medical devices is needed.

The most important messages: major emotional and social impact, changes to sleeping paterns; better quality of life; independence, more control and normal life.

In conclusion adults, children and parents with T1 DM reported very positive experience with rtCGM and FGM devices; different benefits were shown as well as the most important barriers – high cost and unavailability of rtCGM and FGM medical devices in Croatia.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 478 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Additional data: Questions for Focus groups

Impact of condition – Diabetes mellitus treated with insulin

How does diabetes mellitus treated with insulin, for which the health intervention is being assessed, affects you? What are the experiences of living with the diabetes mellitus treated with insulin?

Issues you could consider in your response: • Aspects of diabetes mellitus treated with insulin that are most difficult/distressing at different stages of the disease such as specific symptoms; difficulty in activities of daily living (dressing, eating, socialising, intimacy; loss of ability to work/go to school; social exclusion) • Activities that you find difficult or are unable to do such as engaging in sports and physical activities, housework, shopping • Emotional and psychological impacts such as fear, anxiety, uncertainty, stigma, embarrassment • Impacts of diabetes mellitus treated with insulin on family life • Financial implications such as cost of aids/interventions/support to control symptoms, manage disease, loss of income.

Your views on: How does diabetes mellitus treated with insulin affect the daily life of carers and what is the emotional impact for them? What is the burden on care-givers?

Issues you could consider in your response: • Challenges faced by family and friends who support a patient to manage diabetes mellitus/access the service such as disruption of usual daily routines • Pressures on carers/care-givers daily life such as emotional/psychological issues, relationship challenges, organisation of care, fatigue, stress, anxiety, depression, physical challenges, financial issues. Your views on: Are there groups of patients that have particular issues in managing their diabetes mellitus treated with insulin?

Issues you could consider in your response: • For example groups such as men, women, children, young adults, older people, those with disability, ethnic groups, those in deprived areas, other minority groups, those with a particular genotype • Issues they face (such as looking after family, managing diabetes mellitus alongside other conditions, access to treatment, social stigma).

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 479 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Experiences with currently available health interventions: Self-monitoring blood glucose (SMBG) medical devices

How well do currently available health interventions work, specifically any type of self-monitoring blood glucose medical devices? How does the use of any type of self-monitoring blood glucose medical devices affect activities of daily living? What is the effect of any type of self-monitoring blood glucose medical devices on quality of life?

Issues you could consider in your response: • Main self-monitoring blood glucose medical devices currently used by patients for diabetes mellitus treated with insulin • Extent to which currently available self-monitoring blood glucose medical devices control or reduce the most difficult/distressing aspects of diabetes mellitus treated with insulin (e.g. reduction in symptoms); Health-related quality of life and patient-reported functions like cognitive, physical and social (ability to dress, work, go to school, socialise; improve breathing, swallowing, walking) • The most important benefits of currently available self-monitoring blood glucose medical devices • The burden of currently available self-monitoring blood glucose medical devices on daily life (e.g. difficulty in using the interventions, challenges in recovering after treatment, need for rehabilitation, special clinic visits for treatments and examinations) • Side effects associated with currently available self-monitoring blood glucose medical devices that are distressing or difficult to tolerate • Financial implications to the patient and his/her family such as costs of purchasing the intervention, travelling costs, administration costs • Implications for carers/care-givers such as vulnerability to risk of infection, inflicting pain on loved one when using the intervention • Areas the current self-monitoring blood glucose medical devices do not address. • Is there an unmet need for patients with diabetes mellitus treated with insulin?

Your views on: Are there groups of patients that have particular issues using the currently available self-monitoring blood glucose medical devices? Are there groups of patients who currently don’t have good access to available self-monitoring blood glucose medical devices? Are there factors that could prevent a group or person from gaining access to available self-monitoring blood glucose medical devices?

Issues you could consider in your response: • Those who find it difficult to operate the device (e.g. children, older people, those with disability, those with breathing difficulties) • Those who need to use a device, which has poor social acceptability, in a public place • Those with a particular form of diabetes mellitus

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 480 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

For those with experience of these medical devices, what difference did it make to their lives? How does the use of these medical devices affect activities of daily living? What is the effect of these medical devices on quality of life? Was the use of these medical devices worthwhile?

Issues you could consider in your response: • Main reasons for use of these medical devices (advantages of the technology) • State any objectives set when starting the use of these medical devices and whether they were achieved • Reasons you do or don’t like these medical devices being assessed compared with self-monitoring blood glucose medical devices • Extent to which these medical devices being assessed controls or reduces the most difficult aspects of diabetes mellitus treated with insulin, or resolves it (e.g. reduction in symptoms; ability to dress, work, go to school, socialise; improve breathing, swallowing, walking) • Symptoms that have changed and impact on daily life and quality of life such as less pain, less fatigue, improved continence, less nausea, increased mobility, less time linked to assistive device (e.g. oxygen, dialysis, etc) • Limitations of these medical devices being assessed (disadvantages of the technology) • Unwanted outcomes (e.g. side effects) that are difficult to tolerate and those that patients are willing to tolerate • The burden of these medical devices on daily life (for example ease of use, challenges in recovering after treatment, need for recalibration, special clinic visits for treatments and examinations) • Financial implications to you and your family (e.g. costs of purchasing the intervention, travelling costs, administration costs) • Impact on use of healthcare services, such as fewer visits to clinic • Impact of intervention on carers/care-givers • Any aspects of the health intervention that you would like to change

For those without experience of these medical devices being assessed being assessed, what are the expectations of using it? How do patients perceive these medical devices under assessment?

Issues you could consider in your response:

• Whether the clinical studies have studied outcomes that are important to patients (e.g. symptoms that limit activities) • Do you think that outcomes assessed in this Rapid REA are relevant to you? EFF Domain

Primary outcomes Mortality Glycaemic control: change in HbA1c (glycosylated haemoglobin) and fasting plasma glucose Incidence of hypoglycaemia (i.e., severe, nocturnal, and overall hypoglycaemia)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 481 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Quality of life Patient satisfaction Hypoglycaemia fear

Secondary outcomes Incidence of hyperglycaemia Incidence of diabetic ketoacidosis (in type 1 diabetes) Incidence of hyperosmolar, hyperglycemic coma (in type 2 diabetes) Resource utilization (i.e., number of visits to emergency room, primary care, specialists; hospitalizations) Number of daily finger-sticks tests Number of sensor scans per day

SAF Domain Adverse events (AEs) (device related, i.e. Pain or discomfort related to glucose monitoring and not related to device) /Any AEs, Serious AE (SAE), most frequent AEs and SEAs, Death as SAE, withdrawals due AEs/

Primary outcomes for SAF domain will be frequency of any AEs and SAE.

Questions related to the Ethical and Social aspects

Ethical: Does comparing the new technology to the defined, existing comparators point to any differences that may be ethically relevant?

Social: Does the introduction of the new technology and its potential use/non-use instead of the defined, existing comparator(s) give rise to any new social issues?

• Minimum level of improvement of the most important symptoms that you would like to see • What would you most like to see from these medical devices being assessed (e.g. improved daily life, ability to work, improved mobility, greater symptom control, easier use, less intrusive application) • Main reasons why these medical devices being assessed may not be used • Perceived advantages and disadvantages • Financial implications to you and your family (e.g. costs of purchasing the intervention, travelling costs, administration costs) • Impact of these medical devices on carers/care-givers

Your views on: Which groups of patients might benefit most from these medical devices being assessed?

Issues you could consider in your response: • Groups that currently have very limited treatment options or who find current medical devices difficult to administer

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 482 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Additional information

What information may support you to make informed decisions about using these medical devices?

How are treatment choices explained to you?

What specific issues may need to be communicated to you to improve adherence?

Please include any additional information you believe would be helpful to the HTA reviewers (e.g. ethical or social issues, any potential equality issues, information needs about these medical devices).

8. Key messages

In up to five statements, please list the most important points for you.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 483 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

APPENDIX 2: REGULATORY AND REIMBURSEMENT STATUS

Table A11: Regulatory status

Country Institution Verbatim word- Specified contra- Date of Launched Approval issuing ing of the (antic- indications approval yes/no If number (if approval ipated) indica- (include no in- available) tion(s) expiry date clude for country date of of assess- launch

ment) UK FreeStyle BSI-British The FreeStyle Contraindications: EC certificate yes CE 597686 Libre® Libre Flash The FreeStyle Libre valid from 11- Abbott Glucose Monitor- Flash Glucose Dec-2017 to ing System is Monitoring System 05-Mar-2021 indicated for must be removed measuring inter- prior to Magnetic stitial fluid glu- Resonance Imaging cose levels in (MRI). people (age 4 Caution: and older) with • On rare occasions, diabetes mellitus, you may get inaccu- including preg- rate Sensor glucose nant women. The readings. If you indication for believe your glucose children (age 4 - readings are not 12) is limited to correct or are incon- those who are sistent with how you supervised by a feel, perform a caregiver who is blood glucose test at least 18 years on your finger to of age. The confirm your glu- caregiver is cose. If the problem responsible for continues, remove managing or the current Sensor assisting the and apply a new child to manage one. the FreeStyle • Performance of Libre Flash the System when Glucose Monitor- used with other ing System and implanted medical also for interpret- devices, such as ing or assisting pacemakers, has the child to inter- not been evaluated. pret FreeStyle • The Reader is for Libre readings. It use by a single is designed to person. It must not replace blood be used on more glucose testing in than one person the self- including other management of family members due diabetes with the to the risk of spread- exceptions listed ing infection. All below. Under the parts of the Reader following circum- are considered stances, use a biohazardous and blood glucose can potentially meter to check transmit infectious the current glu- diseases, even after cose readings performing the from the Free- cleaning procedure. Style Libre Flash • Some individuals Glucose Monitor- may be sensitive to ing System the adhesive that Sensor: • During keeps the Sensor times of rapidly attached to the skin. changing glucose If you notice signifi- levels, interstitial cant skin irritation glucose levels as around or under measured by the your Sensor, re- Sensor and move the Sensor reported as and stop using the current may not FreeStyle Libre accurately reflect system. Contact

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 484 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) blood glucose your health care levels. When professional before glucose levels continuing to use are falling rapidly, the FreeStyle Libre glucose readings system. from the Sensor may be higher than blood glucose levels. Conversely when glucose levels are rising rapidly, glucose readings from the Sensor may be lower than blood glucose levels. • In order to confirm hypoglycaemia or impending hypoglycaemia as reported by the Sensor. • If symptoms do not match the Free- Style Libre Flash Glucose Monitor- ing System reading. Do not ignore symptoms that may be due to low blood glucose or high blood glucose. US FreeStyle FDA The FreeStyle Sept 27, yes PMA Num- Libre® Libre Flash 2017 ber: Abbott Glucose Monitor- P160030 ing System is a continuous glucose monitoring (CGM) device indicated for the management of diabetes in persons age 18 and older. It is designed to replace blood glucose testing for diabetes treatment decisions. The System detects trends and tracks patterns aiding in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments. Interpretation of the System readings should be based on the glucose trends and several sequential read-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 485 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) ings over time. The System is intended for single patient use and requires a prescription Canada FreeStyle Health The FreeStyle June 27, Yes Licence Libre® Canada Libre Flash 2017 Number: Abbott Glucose Monitor- 99351 ing System is indicated for measuring interstitial fluid glucose levels in adults aged 18 years and older who have at least 2 years of experience in self- managing their diabetes. It is designed to replace blood glucose testing in the self- management of diabetes with the exceptions listed below. Treatment decisions should not be based on real-time Sensor glucose readings alone and instead should be based on the combination of the Sensor glucose reading, the Glucose Trend Arrow, and Glu- cose Graph. Under the following circumstances, use a blood glucose meter to check the current glucose readings from the Free- Style Libre Flash Glucose Monitor- ing System Sensor: • During times of rapidly changing glucose levels, interstitial glucose levels as measured by the Sensor and reported as current may not accurately reflect blood glucose levels. When glucose levels are falling rapidly, glucose readings from the Sensor may be higher

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 486 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) than blood glucose levels. Conversely when glucose levels are rising rapidly, glucose readings from the Sensor may be lower than blood glucose levels. • In order to confirm hypoglycaemia or impending hypoglycaemia as reported by the System’s Glu- cose Messages. • If symptoms do not match the FreeStyle Libre Flash Glucose Monitoring Sys- tem reading. Do not ignore symptoms that may be due to low blood glucose or high blood glucose. Australia FreeStyle TGA The FreeStyle Aug 24, 2017 Yes ARTG Libre® Libre Flash 269198 Abbott Glucose Monitor- ing System is indicated for measuring interstitial fluid glucose levels in people (age 4 and older) with diabetes mellitus. The indication for children (age 4 - 17) is limited to those who are supervised by a caregiver who is at least 18 years of age. The caregiver is responsible for managing or assisting the child to manage the FreeStyle Libre Flash Glucose Monitor- ing System and also for interpreting or assisting the child to interpret FreeStyle Libre readings. It is designed to replace blood glucose testing in the self- management of diabetes with the exceptions listed below. Under the

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 487 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) following circumstances, use a blood glucose meter to check the current glucose readings from the Free- Style Libre Flash Glucose Monitor- ing System Sensor: • During times of rapidly changing glucose levels, interstitial glucose levels as measured by the Sensor and reported as current may not accurately reflect blood glucose levels. When glucose levels are falling rapidly, glucose readings from the Sensor may be higher than blood glucose levels. Conversely when glucose levels are rising rapidly, glucose readings from the Sensor may be lower than blood glucose levels. • In order to confirm hypoglycaemia or impending hypoglycaemia as reported by the Sensor. • If symptoms do not match the Free- Style Libre Flash Glucose Monitor- ing System reading. Do not ignore symptoms that may be due to low blood glucose or high blood glucose. Dexcom The Dexcom G6® No June 12, ® Conformite Yes G6 Con- Continuous MRI/CT/Diathermy – 2018 tinuous European Glucose Monitor- MR Unsafe Glucose ing System ® Monitoring (Dexcom G6 Do not wear your System System or G6) is CGM (sensor, a glucose moni- transmitter, receiver, toring system or smart device) for indicated for magnetic resonance persons age 2 imaging (MRI), years and older. computed tomogra- The Dexcom G6 phy (CT) scan, or System is de- high-frequency signed to replace electrical heat (dia- finger-stick blood

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 488 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) glucose (BG) thermy) treatment. testing for treatment decisions. The G6 has not Interpretation of been tested in those the Dexcom G6 situations. The System results magnetic fields and should be based heat could damage on the glucose the components of trends and sev- the G6, which may eral sequential cause it to display readings over inaccurate G6 time. The Dex- sensor glucose ® com G6 Sys- readings (G6 read- tem also aids in ings) or may prevent the detection of alerts. Without G6 episodes of readings or hyperglycaemia alarm/alert notifica- and hypoglycae- tions, you might mia, facilitating miss a severe low or both acute and high glucose event. long-term therapy adjustments. The Dexcom G6 System is intended for use by patients at home and in healthcare facilities Dexcom The Dexcom G6® December 7th ® FDA (USA) Yes G6 Con- Continuous 2017 tinuous Glucose Monitor- Glucose ing System Monitoring (Dexcom G6 System System) is a real time, continuous glucose monitoring device indicated for the management of diabetes in persons age 2 years and older.

The Dexcom G6® System is intended to replace finger-stick blood glucose testing for diabetes treatment decisions. Interpreta- tion of the Dex- com G6 System results should be based on the glucose trends and several sequential readings over time. The Dexcom G6 System also aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 489 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) adjustments.

The Dexcom G6® System is also intended to autonomously communicate with digitally connected devices, including automated insulin dosing (AID) systems. The Dexcom G6® System can be used alone or in conjunction with these digitally connected medical devices for the purpose of managing diabetes. Dexcom Conformite The Dexcom G5 The following apply May 15, 2015 yes G5® Mo- European Mobile Continu- to both the G5 and bile Con- ous Glucose the G4, except that tinuous Monitoring Sys- the G4 data requires Glucose tem is a glucose confirmation by a Monitoring monitoring sys- BG meter reading to System tem indicated for make treatment the management decisions, whereas of diabetes in the G5 data can be persons age 2 used for treatment years and older. decisions without The Dexcom G5 BG meter reading Mobile CGM confirmation. System is de- The system must be signed to replace removed prior to finger-stick blood magnetic resonance glucose testing imaging, CT scan, for diabetes or diathermy treat- treatment deci- ment. Safety and sions. performance has Interpretation of not been studied in the Dexcom G5 these circumstanc- Mobile CGM es. System results Use of paracetamol- should be based containing medica- on the glucose tions when the trends and sev- sensor is inserted eral sequential may affect the readings over performance of the time. The Dex- device. com G5 Mobile If sensor readings CGM System do not fit with symp- also aids in the toms, patients detection of should measure episodes of their BG with a hyperglycaemia meter. and hypoglycae- Patients should mia, facilitating calibrate the G4 and both acute and G5 system at least long-term therapy every 12 hours. adjustments. Sensor glucose The Dexcom G5 readings may be Mobile CGM inaccurate if the System is in- sensor is not cali- tended for use by brated at least every patients at home 12 hours. The sys-

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 490 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) and in healthcare tem must be cali- facilities. brated initially 2 hours after the sensor has been inserted. Calibration must be based on a meter reading using Dexcom The Dexcom G5 blood obtained from August 19, yes ® G5 Mo- Mobile Continu- a finger stick. Pa- 2015 bile Con- ous Glucose tients must enter the tinuous Monitoring Sys- exact finger-stick Glucose tem is a glucose reading that their Monitoring monitoring sys- blood glucose meter System tem indicated for displays within 5 detecting trends minutes. Entering and tracking incorrect values or patterns in per- BG readings that sons (age 2 and occurred more than older) with diabe- 5 minutes ago will tes. The system affect device per- is intended for formance. single patient use The CGM System is and requires a not approved for prescription. The use in pregnant Dexcom G5 women or people on Mobile System is dialysis. indicated for use The safety and as an adjunctive effectiveness of the device to com- systems has not plement, not been evaluated for replace, infor- sensor insertion mation obtained sites other than the from standard skin of the abdomen home glucose or, in the case of monitoring devic- patients aged be- es. The Dexcom tween 2 and 17 G5 Mobile Sys- years, other than tem aids in the the abdomen or detection of upper buttocks. episodes of The system should hyperglycaemia not be used in and hypoglycae- critically ill patients. mia, facilitating It is not known how both acute and different conditions long-term therapy or medications adjustments, common to the which may mini- critically ill popula- mize these ex- tion may affect the cursions. Inter- performance of the pretation of the system. Dexcom G5 Patients should Mobile System establish a rotation results should be schedule for choos- based on the ing each new sen- trends and pat- sor location, and terns seen with should avoid sensor several sequen- locations that are tial readings over constrained by FDA (USA) time. clothing or accesso- Dexcom Conformite The Dexcom G4 ries, or are subject- October 25, yes G4® PLAT- European PLATINUM ed to vigorous 2012 INUM Continuous movement during Continuous Glucose Monitor- exercise. The sen- Glucose ing System is a sor should be Monitoring glucose- placed in different System monitoring device sites each time to indicated for allow the skin to detecting trends heal. and tracking

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 491 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) patterns in per- Patients should sons (age 2 and avoid injecting older) with diabe- insulin or placing an tes. The system insulin pump infu- is intended for sion set within 7.62 use by patients at cm (3 inches) of a home and in sensor in case the healthcare facili- insulin affects the ties. sensor glucose The Dexcom G4 readings. PLATINUM Blistering, redness, System is indi- mild swelling, mild cated for use as irritation, and possi- an adjunctive bly mild bleeding at device to com- the insertion site plement, not may occur with use replace, infor- of the device. mation obtained The transmitter has from standard a range of up to 6 home glucose meters without monitoring devic- obstruction, in air. es. The Dexcom G4 PLATINUM System aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments, which may minimize these excursions. Inter- pretation of the Dexcom G4 PLATINUM System results should be based on the trends and patterns seen with several sequential readings over time. Dexcom FDA (USA) The Dexcom G4 October 5, yes G4® PLAT- PLATINUM 2012 INUM Continuous Continuous Glucose Monitor- Glucose ing System is a Monitoring glucose monitor- System ing device indicated for detecting trends and tracking patterns in persons (age 18 and older) with diabetes. The system is intended for single patient use and requires a prescription.

The Dexcom G4 PLATINUM System is indicated for use as

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 492 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) an adjunctive device to complement, not replace, information obtained from standard home glucose monitoring devices.

The Dexcom G4 PLATINUM System aids in the detection of episodes of hyperglycaemia and hypoglycaemia, facilitating both acute and long-term therapy adjustments, which may minimize these excursions. Inter- pretation of the Dexcom G4 PLATINUM System results should be based on the trends and patterns seen with several sequential readings over time. USA Medtronic Stand-alone real- Please refer to Jan 27, Jun CE 8858 for Guardian™ time Continuous contraindications in 27 2017 all sensors Connect Glucose Monitor- IFU documentation and trans- ing (CGM) with provided and sum- mitters; Medtronic Guard- marised below: CE21024 for ian™ Connect EnliteTM Sensor: 2 different enables patients None known packages of to continuously GuardianTM Con- sensors; monitor their nect App: None glucose levels, known allowing them to CareLinkTM Con- track trends and nect: Refer to the patterns. It re- Guardian Connect quires a small app user guide for piece of equip- information on ment, a transmit- contraindications ter and a glucose GuardianTM Con- sensor that nect Transmitter: Do measures the not expose your patient’s intersti- transmitter to MRI tial glucose levels equipment, diather- every 5 minutes. my devices, or other This means devices that gener- patients will ate strong magnetic receive 288 fields. If your trans- readings every mitter is inadvertent- day, compared ly exposed to a with approxi- strong magnetic mately 6 to 10 field, discontinue readings when use and contact the they use a tradi- 24 Hour HelpLine or tional finger- your local repre- prick, blood sentative for further glucose monitor. assistance. The system lets One-press Serter:

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 493 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) patients see their Do not use the glucose values, serter on products as well as other than the Enlite whether their sensor. Medtronic levels are going cannot guarantee up or down, the safety or effica- warning the cy of this product if patients with used with other alarms. Predic- products. tive alerts can also be activated (alerts from 10 minutes to 1 hour ahead of hypos and hypers). In this way, the patients can be warned in advance to prevent the events. The MiniMed Guardi- anTM Connect is a, real time CGM system made up of various components that perform different functions as described below: 1. The EnliteTM Sensor, which is inserted under the skin using a small insertion device (serter) and taped in place for a single-use, six-days period. It measures glucose levels in the fluid surrounding the cells below the skin (interstitial fluid). The sensor sends glucose readings approximately every 5 minutes to the phone app via Bluetooth. It does not replace finger-stick tests for determining insulin requirements for meals and activities. It also requires a minimum of two finger-stick calibrations against the system’s glucose meter every day. 2. The Guardi- anTM Connect Transmitter, in conjunction with the glucose

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 494 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

ment) sensor, collects and wirelessly transmits interstitial glucose values to the App. The Guard- ianTM Connect system can store infinitive days of glucose sensor data. The transmitter comes with a charger described in the IFU. 3. The Guardi- anTM Connect App shown in the picture below displays sensor glucose data, trends and alerts directly to the SmartPhone. • 24/7 glucose monitoring and trends • Predictive alerts for high and low • Real-time remote monitoring and SMS alerts for care partners 4. CareLinkTM personal is a diabetes therapy management platform with a set of reports enabling patients to understand patterns and discuss therapy adjustment with the physician. Medtronic Conformite As submitted Yes Guardian™ European from July 16, Connect 2016

Sources: (65, 68, 70)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 495 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Table A12: Summary of (reimbursement) recommendations in European countries for the technology

Summary of Country and issuing Summary of reasons for (reimbursement) organisation e.g. G-BA, recommendations, rejections and recommendations and NICE restrictions restrictions http://www.codage.ext.cnamts.fr/cgi/ti ps/cgi- France – HAS/CEPS Full for T1 and T2 MDI fiche?p_code_tips=1102257&p_date Abbott FreeStyle Libre® _jo_arrete=%25&p_menu=FICHE&p _site=AMELI Austria - Full for T1 and T2 MDI Abbott FreeStyle Libre® Belgium - http://www.inami.fgov.be/SiteCollecti RIJKSINSTITUUT VOOR Full for T1 – copays for T2 onDocuments/overeenkomst_diabet Abbott FreeStyle Libre® ZIEKTE es_zelfregulatie.pdf Luxembourg Full for T1 and T2 MDI Abbott FreeStyle Libre®

Italia - Emilia Romagna ® Full for T1 and T2 MDI Abbott FreeStyle Libre Region Italia - Basilicata Region Full for T1 only Abbott FreeStyle Libre® Italia - Lazio Region Full for T1 and T2 MDI Abbott FreeStyle Libre® Italia - Tuscany Region Full for T1 and T2 MDI Abbott FreeStyle Libre® Italia - Lombardia Region Full for T1 and T2 MDI Abbott FreeStyle Libre® Italia - Umbria Region Full for T1 and T2 MDI Abbott FreeStyle Libre® Italia - Piemonte Region Full for T1 and T2 MDI Abbott FreeStyle Libre® Italia - Campania Region Full for T1 and T2 MDI Abbott FreeStyle Libre®

Italia - Friuli Venezia ® Full for T1 only Abbott FreeStyle Libre Giulia Region Italia - Trento Province Full for T1 and T2 MDI Abbott FreeStyle Libre® Italia - Veneto Region Full for T1 and T2 MDI Abbott FreeStyle Libre®

UAE – QATAR – Arabie ® Full – hospital dependent Abbott FreeStyle Libre Saoudite https://www.nhsbsa.nhs.uk/pharmacie Full for T1 and T2 MDI 73 ® UK – NHS7 - BSA s-gp-practices-and-appliance- Abbott FreeStyle Libre CCGs contractors/drug-tariff/drug-tariff-part-ix https://www.tk.de/centaurus/servlet/c ontentblob/48710/Datei/94145/TK- No national reimbursement Satzung-Stand-01-11-2016.pdf so far, this is basically up to ® Germany https://www.dak.de/dak/download/Sat Abbott FreeStyle Libre the individual payer and zung_der_DAK- contract. Gesundheit_Fassung_2__Nachtrag_ vom_24_10_2016-1854070.pdf http://www.dagensdiabetes.info/inde x.php/alla-senaste-nyheter/2706- landstingen-rekommenderas-av-nt- radet-libre-till-t2dm-m-flerdos-insulin- ® Sweden Full for T1 and T2 MDI Abbott FreeStyle Libre och-hba1c-over-70-eller- aterkommande-hypoglykemier- nationellt-pris-iphone-kan-lasa- sensor-med-gratisapp No national reimbursement, all diabetes supplies are in ® Finland Abbott FreeStyle Libre tender business in several county councils https://www.bag.admin.ch/bag/de/ho me/themen/versicherungen/krankenv Switzerland Full for T1 and T2 MDI ersicherung/krankenversicherung- Abbott FreeStyle Libre® leistungen-tarife/Mittel-und- Gegenstaendeliste.html

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 496 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Summary of Country and issuing Summary of reasons for (reimbursement) organisation e.g. G-BA, recommendations, rejections and recommendations and NICE restrictions restrictions

Full for T1 and T2 Insulin ® Denmark Abbott FreeStyle Libre User Patients Portugal Full for T1 Abbott FreeStyle Libre® Spain Full only for paediatrics Abbott FreeStyle Libre® Israël Full for T1 Abbott FreeStyle Libre® http://bipold.aotm.gov.pl/ind ex.php/zlecenia-mz- 2017/855-materialy- 2017/5248-zlecenie-175- 2017, for the treatment of type 1 diabetes in children ® Poland* Abbott FreeStyle Libre and adolescents from 4 to 17 years of age and in adults up to 26 years of age, particularly with difficult to control type 1 diabetes Stand-alone or in combination with CSII

T1D and T2D on insulin United States Employers Yes Dexcom CGM Commercial payors United States Gov’t Medicare Yes, for Dexcom only Dexcom CGM NICE diagnostic assessment; Clinical NICE: Limited with SAP; CCG fund- UK National Taxes Dexcom CGM Commissioning Groups on ing for T1 adults Individual Funding Request Sweden Social system funded from contributions Tenders; regional Yes majority Dexcom CGM by employees Norway Social system funded from contributions Tenders; regional Yes majority Dexcom CGM by employees France Social system funded from contributions HAS; national Yes; pricing TBD Dexcom CGM by employees Germany Social system funded from contributions GB-A; national Fully reimbursed from Sept 6, 2016 Dexcom CGM by employees Switzerland private National Yes Dexcom CGM insurance Full reimbursement in Piemonte, Friuli, Bolzano, Toscana, Italy National Taxes Provincial EmilioRomagna, Marche and Dexcom CGM Basalicata; limited reimbursement in Sardegna and Lazio Stand-alone

Yes, full reimbursement in 3 provinc- Dexcom CGM Spain National Taxes Provincial es Belgium Social system Dexcom CGM Full reimbursement under funded from contributions Sensor convention consideration by employees

Slovenia National Taxes National Yes – paediatrics Dexcom CGM

Netherlands Social system National Hospital DRG only Dexcom CGM

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 497 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Summary of Country and issuing Summary of reasons for (reimbursement) organisation e.g. G-BA, recommendations, rejections and recommendations and NICE restrictions restrictions & private insurance

partial; reimbursement includes Dexcom CGM Czech Republic National receivers and transmitters from Oct National; Patient capitation Taxes 2016; patient co-pay between 20- 40% of treatment costs Australia National Taxes National Paediatrics Dexcom CGM Slovakia National Taxes National Yes, paediatrics, patient co-pay Dexcom CGM Individual Funding Request Dexcom CGM South Africa private Reimbursed up to Savings Limit / funded from patients' insurance patient co-pay savings insurance plan Israel Regional Yes – paediatrics Dexcom CGM Austria / Insurance Medtronic Guardian™ National recommendation in Type 1 Competence Center Connect National coverage for type 1 full reimbursement and type 2 with copayment http://www.riziv.fgov.be/FR/DOCUM ENTRECHERCHE/Pages/default2.a spx?r=%22owstaxIdRITargetGroup% 22%3D%232ebaf0cf-7353-4273- Medtronic Guardian™ Belgium/ INAMI b1af- Connect 236262c84494%3A%22Patient%22 %20%22owstaxIdRITheme%22%3D %238ec480f0-fd0c-436a-98b8- 58cfcdd3f17c%3A%22Prestations%2 0de%20soins%20par%20%E2%80% A6%22#.WoGtc66nHcs National coverage for type 1, kids fully reimbursed adults 25% Medtronic Guardian™ Czech Republic/ MoH copayment Connect http://scoo.cz/wp- content/uploads/metodika_996.pdf Medtronic Guardian™ Denmark Regional funding/tender market Connect Medtronic Guardian™ Finland Funding/tender market Connect National coverage type 1 and type 2 on intensive insulin treatment https://www.g- ba.de/institution/presse/pressemitteil ungen/623/ Germany/GBA and https://www.iqwig.de/de/projekte- Medtronic Guardian™

IQWIG ergebnisse/projekte/nichtmedikamen Connect toese-verfahren/d12-01- kontinuierliche-interstitielle- glukosemessung-cgm-mit-real-time- messgeraten-bei-insulinpflichtigem- diabetes-mellitus.3258.html National coverage Ireland/General Medical Medtronic Guardian™ http://www.hse.ie/eng/services/list/1/ Services Connect schemes/lti/ Funding at the regional level for diabetes patients Medtronic Guardian™ Italy /MoH http://www.gazzettaufficiale.it/eli/gu/2 Connect 017/03/18/65/so/15/sg/pdf National coverage type 1 https://www.zorginstituutnederland.nl Medtronic Guardian™ Netherlands/ZIN /publicaties/standpunten/2010/11/01/ Connect continue-glucose-monitoring

Norway Funding/tender market Medtronic Guardian™

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 498 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

Summary of Country and issuing Summary of reasons for (reimbursement) organisation e.g. G-BA, recommendations, rejections and recommendations and NICE restrictions restrictions Connect Slovenia/ Health Medtronic Guardian™ Insurance institute of National coverage Connect Slovenia Medtronic Guardian™ Spain regional coverage Connect Medtronic Guardian™ Sweden Funding/tender market Connect National coverage in the basic insurance for type 1 https://www.bag.admin.ch/bag/de/ho Medtronic Guardian™ Switzerland/FOPH me/themen/mensch- Connect gesundheit/nichtuebertragbare- krankheiten/diabetes.html https://www.nice.org.uk/guidance/ng 18 Medtronic Guardian™ UK/NICE clinical guideline https://www.nice.org.uk/guidance/ng Connect 17 According to the Regulation of the Minister of Health of January 18, 2018 http://dziennikustaw.gov.pl/du/2018/2 81/1), the following elements of the rtCGM system are refunded in Po- land: sensor/electrode and transmitter (for patients up to 26 years old with type 1 diabetes, treated with a pump insulin, with an unconscious hypoglycaemia (no prodromal symptoms of hypoglycaemia; exclusion of alcoholic hypoglycaemia). In 2015 at the CGM system AOTMiT was assessed for the population of people with type 1 diabetes – there was a positive appraisal of the Transparency Council and the Poland* President of AOTMiT for children with type 1 diabetes treated with a pump insulin, with an unconscious hypoglycaemia (no prodromal symptoms of hypoglycaemia; exclusion of alcoholic hypoglycaemia http://bipold.aotm.gov.pl/index.php/zl ecenia-mz-2015/829-materialy- 2015/3864-zlecenie-017-2015). Then, in 2017 there was a positive appraisal of the Transparency Council and the President of AOTMiT, for patients up to 26 years old with type 1 diabetes, treated with a pump insulin, with frequent/repeated episodes of severe hypoglycaemia (http://bipold.aotm.gov.pl/index.php/z lecenia-mz-2017/5008-zlecenie-073- 2017). Sources: (65, 68, 70)

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 499 Continuous (real-time) and flash glucose monitoring as personal, standalone systems in patients with DM treated with insulin

APPENDIX 3: CHECKLIST FOR POTENTIAL ETHICAL, ORGANISATIONAL, PATIENT AND SOCIAL AND LEGAL ASPECTS

1 Ethical 1.1 Does the introduction of the new technology and its potential use/non-use Yes instead of the defined, existing comparator(s) give rise to any new ethical issues? Questions related to equity could be important to take into consideration. Could potential inequalities prevent access to the technology? And more specifically, are there factors that could prevent a group or person from gaining access to the technology? If so, is it possible to influence these factors or manage the utilisation of the technology in a way that gives equal access to those in equal need? Questions related to supportive actions and information could be important to focus on. Is there any specific information or any support patients (or decision-makers?) should seek to decide upon adopting the technology? Are there any particular challenges related to the use of the technology that the patient and/or caregivers need to be aware of? Is there clear and sufficient information available to understand the technology and possible risk related to unappropriated use? 1.2 Does comparing the new technology to the defined, existing comparators point to No any differences that may be ethically relevant? 2 Organisational 2.1 Does the introduction of the new technology and its potential use/non-use Yes instead of the defined, existing comparator(s) require organisational changes? Questions related to the involvement of patients and caregivers as well as proper education and training could be important to raise. What kind of involvement of patients/participants and/or caregivers is the most suited? And, what kinds of co-operation and communication of activities are needed? 2.2 Does comparing the new technology to the defined, existing comparator(s) point No to any differences that may be organisationally relevant? 3 Social 3.1 Does the introduction of the new technology and its potential use/non-use Yes instead of the defined, existing comparator(s) give rise to any new social issues? Questions related to patients' perspectives and perceptions as well as their expectations from using the technology could be important to discuss. This includes any positive or negative experiences that may arise as a consequence of using the technology (i.e., worries, satisfaction, stigmatisation, social status…). A new technology may allow patients to return to work, however since the technology possibly can be seen or alarm sound can be heard by co-workers, it may lead to undesired attention from the surroundings. 3.2 Does comparing the new technology to the defined, existing comparator(s) point No to any differences that may be socially relevant? 4 Legal 4.1 Does the introduction of the new technology and its potential use/non-use No instead of the defined, existing comparator(s) give rise to any legal issues? 4.2 Does comparing the new technology to the defined, existing comparator(s) point No to any differences that may be legally relevant?

For the purpose of transparency, a separate document with comments on the 2nd draft assessment from external experts and the MAH/manufacturer(s) (fact check), as well as responses from authors, is available on the EUnetHTA website.

Version 1.4, 27 July 2018 EUnetHTA Joint Action 3 WP4 500