SUPPLEMENTARY DATA

Supplementary Figure 1.

Patients with suspected MI at participating sites during trial period 1

Randomized in DETO2X-AMI trial (N=6629)

Enrollment

Allocation Allocated to oxygen treatment (N=3311) Allocated to ambient air (N=3318)

All except AMI All except AMI excluded excluded (N=826) (N=793)

Enrolled in DETO2X-DM trial Enrolled in DETO2X-DM trial (N=2485 of which 445 with DM, 2040 without) (N=2525 of which 489 with DM, 2036 without)

Follow-Up

Lost to follow-up (N=0)

Analysis Analyz ed in primary ITT analysis 2 Analyzed in primary ITT analysis 2 (N=2485) (N=2525)

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA

Eligible patients with suspected myocardial infarction who presented to the ambulance service,

emergency departments, coronary care units, or catheterization laboratories of participating

were evaluated for inclusion. Shown are the numbers of patients who were enrolled in

the main trial (DETO2X-AMI), randomly assigned to a study group, and followed up in the

present subgroup analysis of patients with confirmed acute myocardial infarction (AMI) divided

by known diabetes status at enrollment (DETO2X-DM).

DETO2X•AMI denotes Determination of the Role of Oxygen in Suspected Acute Myocardial

Infarction trial, DM diabetes mellitus, ITT intention to treat.

1: The trial design and limited resources didn’t allow a detailed screening log on all patients with

suspected AMI evaluated for enrollment as the majority of the patients with suspected AMI only

at presentation but who receive a different final diagnosis, never get registered in

SWEDEHEART.

2: As the present analysis consists of a prespecified subgroup analysis from the DETO2X-AMI

trial, only an ITT analysis was performed. Of all the patients analyzed (N=5010), 288 (5.7%)

developed hypoxemia and received supplemental outside the study protocol (2.3% in the oxygen

group, 9.1% in the ambient-air group).

Furthermore, 217 (4.3%) did not complete the allocated treatment period, 164 (3.2%) patients

allocated to oxygen, and 52 (1.1%) allocated to ambient air.

Reasons for not fulfilling randomized therapy were as follows (oxygen/ambient air): patients’

unwillingness to continue therapy, 4.4%/0.4%; change of clinical circumstances, 0.8%/0.4%;

mistakes (technical, communication), 1.0%/1.0%.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA CONSORT CHECKLIST

Table. CONSORT 2010 Checklist of Information to Include When Reporting a Randomized Triala Reported Item on Section and Topic No. Checklist Item Page No. Title and abstract 1a Identification as a randomized trial in the title 1b Structured summary of trial design, methods, results, and conclusions (for specific guidance see CONSORT for abstracts) Introduction Background 2a Scientific background and explanation of rationale and objectives 2b Specific objectives or hypotheses Methods Trial design 3a Description of trial design (such as parallel, factorial) including allocation ratio 3b Important changes to methods after trial commencement (such as eligibility criteria), with reasons Participants 4a Eligibility criteria for participants 4b Settings and locations where the data were collected Interventions 5 The interventions for each group with sufficient details to allow replication, including how and when they were actually administered Outcomes 6a Completely defined prespecified primary and secondary outcome measures, including how and when they were assessed 6b Any changes to trial outcomes after the trial commenced, with reasons Sample size 7a How sample size was determined 7b When applicable, explanation of any interim analyses and stopping guidelines Randomization Sequence 8a Method used to generate the random allocation sequence generation 8b Type of randomization; details of any restriction (such as blocking and block size) Allocation concealment 9Mechanismusedtoimplementtherandomallocationsequence(suchassequentiallynumbered mechanism containers), describing any steps taken to conceal the sequence until interventions were assigned Implementation 10 Who generated the random allocation sequence, who enrolled participants, and who assigned participants to interventions Blinding 11a If done, who was blinded after assignment to interventions (for example, participants, care providers, those assessing outcomes) and how 11b If relevant, description of the similarity of interventions Statistical 12a Statistical methods used to compare groups for primary and secondary outcomes methods 12b Methods for additional analyses, such as subgroup analyses and adjusted analyses Results Participant flow 13a For each group, the numbers of participants who were randomly assigned, received intended treatment, (a diagram is strongly and were analyzed for the primary outcome recommended) 13b For each group, losses and exclusions after randomization, together with reasons Recruitment 14a Dates defining the periods of recruitment and follow-up 14b Why the trial ended or was stopped Baseline data 15 A table showing baseline demographic and clinical characteristics for each group Numbers analyzed 16 For each group, number of participants (denominator) included in each analysis and whether the analysis was by original assigned groups Outcomes 17a For each primary and secondary outcome, results for each group, and the estimated effect size and its and estimation precision (such as 95% confidence interval) 17b For binary outcomes, presentation of both absolute and relative effect sizes is recommended Ancillary analyses 18 Results of any other analyses performed, including subgroup analyses and adjusted analyses, distinguishing prespecified from exploratory Harms 19 All important harms or unintended effects in each group (for specific guidance see CONSORT for harms) Comment Limitations 20 Trial limitations, addressing sources of potential bias, imprecision, and, if relevant, multiplicity of analyses Generalizability 21 Generalizability (external validity, applicability) of the trial findings Interpretation 22 Interpretation consistent with results, balancing benefits and harms, and considering other relevant evidence Other information Registration 23 Registration number and name of trial registry Protocol 24 Where the full trial protocol can be accessed, if available Funding 25 Sources of funding and other support (such as supply of drugs), role of funders aWe strongly recommend reading this statement in conjunction with the CONSORT 2010 Explanation and Elaboration for important clarifications on all the items. If relevant, we also recommend reading CONSORT extensions for cluster randomized trials, noninferiority and equivalence trials, nonpharmacological treatments, herbal interventions, and pragmatic trials. Additional extensions are forthcoming: for those and for up-to-date references relevant to this checklist, see http://www.consort-statement.org.

©2010 American Medical Association. All rights reserved. (Reprinted) JAMA, July 7, 2010—Vol 304, No. 1 E1

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA

SUPPLEMENTAL MATERIAL

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA This supplement to the DETO2X trial contains the following items:

1. Original protocol

2. Original statistical analysis plan

3. Participating centers and investigators

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 1 SUPPLEMENTARY DATA 1. Original protocol

This protocol was the original and final version of the DETO2X trial. No major changes occurred during the trial.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 2 SUPPLEMENTARY DATA

DETermination of the role of OXygen in suspected Acute Myocardial Infarction (DETO2X-AMI) trial

version March 2013, 26/11-2013

Principal investigators

Leif Svensson, MD, PhD

Karolinska Institutet, Department of Medicine, Solna, , and Department of Cardiology, Södersjukhuset, 11863 Stockholm, Sweden

Email: [email protected] Phone: +46732703844

Johan Herlitz, MD, PhD

Department of Health Sciences, University of Borås, Sweden and Department of Cardiology,

Sahlgrenska University , 41345 Gothenburg, Sweden

Email: [email protected] Phone: +46734612002

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 3 SUPPLEMENTARY DATA Steering committee

Robin Hofmann, MD

Nils Witt, MD, PhD

Mats Frick, MD, PhD

Karolinska Institutet, Department of Clinical Science and Education, Division of Cardiology, Södersjukhuset, Stockholm

Tomas Jernberg, MD, PhD

Karolinska Institutet, Department of Medicine and Karolinska University Hospital, Department of Cardiology, Stockholm

Mattias Ekström, MD, PhD

Rickard Linder, MD, PhD

Karolinska Institute, Department of Clinical Science and Department of Cardiology, Danderyd University Hospital, Stockholm

Lennart Nilsson, MD, PhD

Joakim Alfredsson, MD, PhD

Department of Cardiology and Department of Medical and Health Sciences, Linköping University, Linköping

David Erlinge, MD, PhD

David Zughaft, BSc

Ulf Ekelund, MD, PhD

Department of Cardiology and Emergency Medicine, Lund University, Lund

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 4 SUPPLEMENTARY DATA Annica Ravn-Fischer, MD, PhD

Elmir Omerovic, MD, PhD

Department of Cardiology, Sahlgrenska University Hospital, Gothenburg

Thomas Kellerth, MD

Department of Cardiology and Faculty of Health, Örebro University, Örebro

Stefan James, MD, PhD

Bertil Lindahl, MD, PhD

Ollie Östlund, PhD

Department of Medical Sciences, Cardiology and Uppsala Clinical Research center, Uppsala University, Uppsala

Executive committee

Leif Svensson (chair)

Johan Herlitz

Robin Hofmann

Nils Witt

David Erlinge

Tomas Jernberg

Bertil Lindahl

Stefan James

Adjunct:

Eva Jacobsson, senior project manager, Uppsala Clinical Research center, Uppsala

Ollie Östlund, senior statistician, Uppsala Clinical Research center, Uppsala

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 5 SUPPLEMENTARY DATA

Table of contents

Table of contents 6

List of abbreviations 8

1. Introduction 10

1.1 Aim 12

1.2 Hypothesis 12

2. Study design, population and methods 12

2.1 Design and population 12

2.2 Methods 14

3. Flow chart 16

4. Efficacy outcomes 16

5. Safety assessment 17

5.1 Withdrawal criteria 18

6. Statistics 18

7. Ethical considerations and risk-benefit analysis 19

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 6 SUPPLEMENTARY DATA

8. Good clinical practice 19

9. Approval of the study protocol 20

10. Local study organisation 20

11. Time schedule 20

12. Source data 20

13. Maintenance of records 20

14. Monitoring 21

15. Insurance 21

16. Reporting and publication 21

17. Authorship of publications 22

18. References 23

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 7 SUPPLEMENTARY DATA List of abbreviations

ACS: Acute coronary syndrome

AMI: Acute myocardial infarction

CAD: Coronary artery disease

CRF: Case report form

DETO2X-AMI: DETermination of the role of OXygen in suspected Acute Myocardial Infarction

ED:

EMS: Emergency Medical Service

IC: Informed consent

IR: Ischemia-reperfusion

ITT: Intention-to-treat

MACE: major adverse cardiac events

NSTEMI: Non-ST-elevation myocardial infarction

PCI: Percutaneous coronary intervention

PP: Per-protocol

PROBE: Prospective Randomized Open Blinded End-point

RCT: Randomized controlled trial

ROS: Reactive oxygen species

RRCT: Randomized, Registry based Clinical Trial

SDV: Source data verification

STEMI: ST-elevation myocardial infarction

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 8 SUPPLEMENTARY DATA SWEDEHEART: Swedish Web-system for Enhancement and Development of Evidence-based care in Heart Disease Evaluated According to Recommended Therapies

UCR: Uppsala Clinical Research Center

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 9 SUPPLEMENTARY DATA 1 Introduction

Supplemental oxygen treatment has routinely been used in patients with suspected acute myocardial infarction (AMI) for more than a century1. Also, the pre-hospital and hospital use of oxygen treatment during acute coronary syndromes (ACS) is variably recommended in ACS-guidelines, despite very limited data supporting a beneficial clinical effect2,3. This practice is based on the belief that the administration of supplemental oxygen leads to increased oxygen delivery to the ischemic myocardium and thereby reduces infarct size and subsequent risk of complications such as heart failure and malignant arrhythmias.

A few experimental animal studies and some small clinical trials in humans support the role of supplemental oxygen in non-hypoxic ACS patients. In two studies on anesthetized dogs, 40-100% oxygen after coronary artery occlusion was reported to reduce myocardial infarct size and improved left ventricular ejection fraction as compared to air4,5. In a human study, 17 patients with anterior AMI received 100% oxygen and reduction in ST-segment elevation was seen on precordial ECG- mapping6. Encouraged by the assumption that a higher arterial oxygen tension reduces infarct size, other administration techniques have also been tested. In a multi-center trial, 112 patients with STEMI were randomized to receive hyperbaric or normobaric supplemental oxygen during thrombolysis for anterior STEMI7. No significant benefit was found using hyperbaric oxygen. In another multi-center trial, 269 STEMI patients were randomly allocated to either intracoronary hyperoxemic reperfusion after PCI of the target vessel or normooxemic reperfusion8. At 30 days there were no significant differences between the two groups in regard to infarct size, ST-segment resolution or wall motion score index.

In contrast to these studies with positive or neutral results, there are several smaller trials suggesting potentially harmful effects from supplemental oxygen. Indeed, oxygen treatment has been found to cause a decrease in cardiac output, coronary blood flow, left ventricular perfusion, coronary and systemic oxygen delivery, and an increase in coronary vascular resistance9-15.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 10 SUPPLEMENTARY DATA Coronary vasoconstriction during hyperoxemia is caused by direct effects on the regulation of vessel tension as well as an increased generation of reactive oxygen species (ROS) and subsequent decrease in the bioavailability of nitric oxide14,16-18. Also, the production of ROS might lead to enhanced leukocyte chemotaxis and inflammation, thereby potentially aggravating myocardial damage19.

A Cochrane report in 2010 reviewed the evidence available from randomized controlled trials (RCTs) on supplemental oxygen in acute myocardial infarction20. Only three studies met the inclusion criteria for the meta-analysis: Rawles and Kenmure21 from 1976, Wilson and Channer22 from 1995 and Ukholkina et al.23 from 2005, adding up to a total of 387 patients. No significant difference was found between oxygen therapy and air concerning mortality or analgesic effect although there a non-significant trend towards more deaths in the oxygen group was found.

In summary, the existing data are conflicting and fail to clarify the role of supplemental oxygen in patients with AMI. Available studies are outdated, of poor quality or largely under-powered to show clinically relevant effects. The need to clarify this important issue in a large randomized trial is urgent.

Since 2012 two RCTs have been started that investigate oxygen therapy in STEMI patients: The Australian AVOID study24 is aiming to include 490 patients and to look at infarct size by cardiac biomarkers as primary endpoint, whereas the Swedish SOCCER trial assesses myocardial salvage index by cardiac magnetic resonance imaging (CMR) in 100 patients25.

The Swedish DETO2X-AMI study has been designed, with appropriate power, to evaluate whether supplemental oxygen therapy influences mortality in patients with suspected AMI.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 11

SUPPLEMENTARY DATA 1.1 Aims

 To evaluate the effect of supplemental oxygen on one-year total mortality in patients with suspected AMI  To assess the effect of supplemental oxygen on major adverse cardiac events in patients with suspected AMI  To study if the effect of supplemental oxygen differs between pre-specified sub groups  To study the effect of supplemental oxygen on health economy in patients with suspected AMI

1.2 Hypothesis

Our hypothesis is that the use of supplemental oxygen in non-hypoxic patients with suspected AMI lowers mortality as compared to room air.

2 Study design, population and methods

2.1 Study design and population

The DETO2X-AMI trial is a prospective, multicenter, controlled, Randomized, Registry based Clinical Trial (RRCT) based on the SWEDEHEART (Swedish Web- system for Enhancement and Development of Evidence-based care in Heart disease Evaluated According to Recommended Therapies) registry.

The RRCT concept was first introduced in the Thrombus Aspiration in ST-Elevation myocardial infarction in Scandinavia (TASTE) trial26. The basis for this concept is the fact that in Sweden almost all patients with acute coronary syndrome are registered in the SWEDEHEART registry which is comprised of RIKS-HIA (national registry of acute coronary care), SCAAR (national registry of angiography and angioplasty), Swedish heart registry and SEPHIA (national registry of secondary prevention)27.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 12 SUPPLEMENTARY DATA The registry is web-based with all data registered on-line directly by the caregiver and transferred in an encrypted format to a central server. During registration the whole process of care is kept together in one record even if the patient is transferred between different units and hospitals. The technical platform, OpenQreg, is published as open source software that can receive data via the Internet or from other databases and electronic patient journals. The platform is in direct contact with the Swedish National Population Registry for immediate access to personal data and statistics on deaths. For patients admitted to hospital because of symptoms suggestive of ACS, information is collected prospectively for 106 variables and includes: patient demographics, admission logistics, risk factors, past medical history, medical treatment prior to admission, electrocardiographic changes, biochemical markers, other clinical features and investigations, medical treatment in hospital, interventions, hospital outcome, discharge diagnoses and discharge medications.

SWEDEHEART provides manuals, education and technical advice, including a telephone help desk for all users of the registry. The system has error checking routines for range and consistency with immediate feedback to the caregiver. Definitions are easily available when data are entered. To ensure the correctness of the data entered, monitors visit the hospitals on a regular basis. The agreement between key variables in the registry and medical records has consistently been 95- 96%.

In the DETO2X-AMI trial, randomization will be performed by means of a separate and easily accessible online randomization module in which inclusion criteria need to be affirmed and exclusion criteria denied before a patient is allocated to one of the treatments. Randomized patients will be automatically registered in the SWEDEHEART registry, and all relevant data will be obtained directly from the registry. No other documentation is needed. Mortality data outside hospital will be obtained by merging with the Swedish population registry, which includes information of the vital status of all Swedish citizens. Other clinical endpoints are collected by merging with the National Patient Registry, which includes diagnoses on all patients

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 13 SUPPLEMENTARY DATA hospitalized in Sweden. Thanks to the unique 10-digit personal identification number of all Swedish citizens, complete follow up is ensured.

The authors are solely responsible for the design and conduct of this study – all study analyses, the drafting and editing of the paper and its final contents.

2.2 Methods

2.2.1 Inclusion/exclusion criteria Patients ≥ 30 years of age with normal oxygen saturation (≥90% on pulse oximeter) presenting to the Emergency Medical Service (EMS) system or the emergency department (ED) with classical symptoms suggestive of ACS within the last 6 hours and significant ECG changes (characterized by ST-segment elevation ≥ 0.2 mV in leads V1-V4, or ≥ 0.1 mV in other leads, ST-segment depression ≥0.1 mV in any lead, negative ischemic T-wave in leads V2-V6, pathological Q-wave in at least 2 adjacent leads, or left bundle branch block) or elevated cardiac biomarkers in ED (above local decision-limit for AMI) are evaluated for inclusion.

Patients will be excluded if they do not consent, have continuous oxygen therapy or have suffered cardiac arrest prior to inclusion. If supplemental oxygen therapy is started before evaluation for inclusion for less than 20 minutes, a new evaluation can take place after discontinuation of oxygen delivery and ten minutes of wash-out. If deemed clinically necessary patients randomized to air may receive oxygen, but cross-over between randomized groups is discouraged. Unintentional crossover will be minimized with the use of stickers and patient bracelets that indicate the group to which a patient has been randomized.

2.2.2 Informed consent and randomization If a patient is deemed eligible, oral informed consent is obtained by EMS staff or ED personnel prior to 1:1 randomization to oxygen or room air. The patient will receive written study information directly after being admitted to a ward and will be asked to confirm informed consent by signature. Patients who withdraw their study consent will receive whatever alternative appears to be clinically relevant and will be excluded

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 14 SUPPLEMENTARY DATA from any study specific activities. However, they will remain in the national registries according to routine and be available for evaluation of the primary endpoint because mortality is publicly available. Patients who decline participation in the national registry will be withdrawn from the registry database but remain in the study data base. The DETO2X-AMI trial is using an open-label trial with blinded endpoint (PROBE)28. Blinding to reduce bias has been discussed but was deemed impossible for several reasons. Due to the lack of pressurized air in Swedish ambulances and that fact that most companies only use closed Hudson masks which require a minimum flow of 6L/min air or oxygen to reduce the risk of carbon dioxide retention, pre-hospital blinding would put patients at risk. With 1-year total mortality as the primary outcome, the reduction of the placebo effect by blinding should be of less importance.

The concept of a trial design using national registries as the only basis for follow-up is potentially limited by the lack of formal central adjudication of clinical events. Therefore, we have chosen one-year all-cause mortality obtained from the Swedish population registry as the primary end point of the trial. The implementation of a Data and Safety Monitoring Board has been extensively discussed but due to expected rapid inclusion rate and relatively long follow-up to event time, it was deemed impractical.

2.2.3 Sources of funding The DETO2X-AMI study group has received unrestricted study grants from The Swedish Heart-Lung Foundation, Swedish Research Council and the Swedish Foundation for Strategic Research for the execution of the study. The SWEDEHEART registry has granted funds for administration, data management work, and statistical analysis which are carried out at Uppsala Clinical Research Center at Uppsala University Hospital, Sweden.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 15

SUPPLEMENTARY DATA 3 Flow chart

4 Efficacy outcomes

4.1 Primary efficacy outcome

 One-year all-cause mortality in all patients with suspected AMI.

4.2 Secondary efficacy outcomes

 MACE: all-cause mortality or rehospitalization with heart failure or readmission with myocardial infarction  rehospitalization with heart failure  rehospitalization with AMI  health economy

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 16

SUPPLEMENTARY DATA Primary and secondary outcomes will be assessed at 30 days and one year of follow up.

4.3 Prespecified subgroup analyses

Predefined subgroups to be analyzed include gender (male/female), age ( median), AMI/Non-AMI, Type-I AMI (STEMI/NSTEMI), smokers (yes/no), Hb ( LNV), previous MI (yes/no), previous PCI (yes/no), oxygen fractions (95-100%/90- 94%), patients with chronic obstructive pulmonary disease (yes/no), chronic kidney disease (eGFR 60ml/min), and diabetes mellitus (yes/no).

4.4 Sub studies

4.4.1 DETO2X-Biomarkers

This is a multi-center sub-study to the DETO2X-AMI trial assessing if oxygen treatment enhances oxidative stress, systemic inflammation, and markers of apoptosis and MMPs in ACS patients, thereby potentially increasing myocardial damage and cell death, and potentially the prognosis (see separate trial protocol or clinicaltrials.gov NCT02290080 for details).

4.4.2 DETO2X-OXYPAIN 2

This is a multi-center sub-study to the DETO2X-AMI trial at centers with catheter laboratories evaluating a possible analgesic effect of oxygen in using visual-analog scale (VAS).

5 Safety assessments

Any unexpected adverse reaction to study treatment or study related procedures will be recorded in the CRF.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 17 SUPPLEMENTARY DATA 5.1 Withdrawal criteria

The study subject has the right to withdraw consent and discontinue the study at any time. In case of withdrawal the reason should be recorded in the CRF.

6. Statistics

The main focus of this trial is to gain clinically relevant information on the effect of supplemental oxygen in the setting of suspected acute myocardial infarction. The primary outcome is time from randomization to all-cause death within one year, in all randomized patients. All outcomes will be analyzed using the intention to treat (ITT) principle where patients randomized to a certain group will be followed irrespective of the actual treatment, offering unbiased assessments of treatment efficacy29. As a supplement, a per-protocol (PP) analysis and analyses based on patients with diagnosed AMI only will also be performed. Survival probabilities will be displayed and calculated using Kaplan-Meier methodology. Hazard ratios between treatment groups will be calculated using a Cox proportional hazard model, adjusted for age (as a linear covariate on the log-hazard scale) and gender. Patients without events will be censored at one year after randomization, or time of withdrawal of informed consent for secondary outcome events. A supplementary analysis will be performed using the full follow-up time available for each patient at time of final analysis. The same model will be applied to time from randomization to first MACE, and the individual MACE sub-categories (re-infarction and hospitalization due to heart failure). The same model will be also used for analyses in the PP and the diagnosed AMI populations. Subgroup analyses will be performed by introducing a treatment- subgroup interaction term in the Cox model, excluding any patients not possible to classify. Supplemental analyses of the number of patients with death and with MACE, within 30 days and within 1 year of randomization, will be performed using logistic regression adjusted for age and gender. Estimates of treatment differences will be presented with two-tailed 95% confidence intervals and associated p-values. A two- tailed P-value of <.05 is considered statistically significant. Secondary outcomes will be analyzed without adjustment for multiplicity.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 18 SUPPLEMENTARY DATA To calculate sample size two different sources of data were used – data from Herlitz et al. 30,31 supplying prospective data and analysis from RIKS-HIA from 2005-2010. The 1-year total mortality in patients treated with room air was estimated to be 14.4 %. A clinically relevant effect of supplemental oxygen was defined as a relative risk reduction of 20%. To be able to reject the null hypothesis at a significance level of 0.05 with a power of 0.90, 2,900 patients per group are needed. Loss to follow-up is considered negligible due to virtually complete coverage in the national registries. To control for cross-over and failure to complete the protocol, sample size is increased to 3,300 patients per group, resulting in a total of 6,600 patients. The calculation is based on an analysis of proportions at one year, and the power for the pre-specified adjusted Cox analysis is expected to be as good or better.

7 Ethical considerations and risk-benefit analysis

Apart from study treatment allocation, participation in the DETO2X-AMI study involves no other study specific procedures. The delivery of oxygen (6L/min for 6-12 hours) by Oxymask® is used in clinical routine and is considered to be safe and well- accepted by patients. For those randomized to no oxygen, treatment protocol assures that signs and symptoms of hypoxia will lead to immediate initiation of oxygen therapy (cross-over).

The potential benefit of the study is to provide new knowledge on the effect of supplemental oxygen on cardiovascular morbidity and mortality in patients with suspected AMI. If oxygen treatment is found to render no benefit or even to harm in an ACS setting, it should be excluded from the treatment algorithm.

Overall, we believe the risk of participation in the present study is very low and is markedly outweighed by potential benefits in terms of new knowledge.

8 Good Clinical Practice

The study will be conducted in accordance with the Declaration of Helsinki and Good Clinical Practice Guidelines in the latest version.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 19

SUPPLEMENTARY DATA 9 Approval of the Study Protocol

Approval of the protocol and amendments has been obtained from the Regional Ethical Review Board of Gothenburg (DNR 287-12) as well as from the Swedish Medical Products Agency (EudraCT 2013-002882-20).

10 Local study organizations

As both STEMI and NSTEMI patients are eligible for inclusion, hospitals without primary PCI facilities can also take part, thus offering access to a broad population. It is expected that a majority of the 71 hospitals in the country with acute cardiac care facilities enrolling in the SWEDEHEART registry will participate.

11 Time schedule

Recruitment is intended to start in spring 2013 and expected to be completed during 2015.

12 Source data

Source Data is defined as original documents, data and records. For this study the CRF is considered as Source Data. The study period per patient ends at 12 hours post start treatment. All follow-up data are collected via registries and will not be monitored. CRF for study participants will be stored in the archives of each enrolling centre. Study data is registered in the ordinary SWEDEHEART and a study data base is constructed and kept at UCR. The patients’ identities will always be kept confidential. The investigators are responsible for ensuring the accuracy, completeness, legibility and timeliness of the data recorded in the CRFs.

13 Maintenance of records

The printed CRFs will be stored at the primary investigators facility after completion of the trial for the maximum period of time permitted by the hospital but not less than 10 years after completion or discontinuation of the trial.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 20

SUPPLEMENTARY DATA 14 Monitoring

The monitoring activities are described in detail in a “monitoring plan” (see appendix). In summary: The monitoring is mainly centralized and performed by UCR in cooperation with RH and LS. Study Subjects will be monitored throughout the study period, i.e. study day (12 hours). Source Data Verification (SDV) will be done on 5% of the CRFs for enrolled subjects at the five sites enrolling most patients. The clinics will be asked to fax the requested CRFs to the Monitor. The monitor will check the data base against the CRF. Central monitoring activities will be used to monitor the informed consent (IC) procedure. In order to monitor the IC-process, the site will log all completed informed consents on an ongoing basis and send an updated log to UCR at the end of each month. The UCR Monitor will check the log against the database for a random subset for each site (10%). If inconsistencies are noted between the log and database, the site will be contacted by the monitor both for clarification and to agree on any action required. In addition to the study specific monitoring activities, the ordinary SWEDEHEART monitoring system is put into place by independent professionals to ensure the quality of SWEDEHEART registered data. During the study period, monitors will have regular contact with the participating departments to ensure that the trial is conducted in compliance with the protocol and applicable regulatory requirements. The monitors will also provide information and support to the investigator(s).

Investigators and other responsible personnel must be available during audits and inspections and should devote sufficient time to these processes.

15 Insurance

The Swedish “Patientförsäkring” (patient insurance) is valid and covers all study- associated complications in regard to participating hospitals and patients.

16 Reporting and publication

The results of this study will be presented at national and international meetings and published in peer-review journals.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 21

SUPPLEMENTARY DATA

17 Authorship of publications

Authors of publications must meet the International Committee of Medical Journal Editors guidelines for authorship and must satisfy the three criteria that follow:

a) Authors must make substantial contributions to the conception and design of the trial, acquisition of data, or analysis of data and interpretation of results.

b) Authors must draft the publication or, during draft review, provide contributions (data analysis, interpretation, or other important intellectual content) leading to significant revision of the manuscript with agreement by the other authors.

c) Authors must provide written approval of the final draft version of the publication prior to submission.

All contributors who do not meet these three criteria for authorship should be listed in an acknowledgments section within the publication, if allowed by the journal.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 22 SUPPLEMENTARY DATA 18 References

1. Kones R. Oxygen Therapy for Acute Myocardial Infarction-Then and Now. A Century of Uncertainty. American Journal of Medicine 2011;124:1000-5. 2. Hamm CW, Bassand JP, Agewall S, et al. ESC Guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation. European Heart Journal 2011;32:2999-3054. 3. Steg PG, James SK, Atar D, et al. ESC Guidelines for the management of acute myocardial infarction in patients presenting with ST-segment elevation. European Heart Journal 2012;33:2569-619. 4. Kelly RF, Hursey TL, Parrillo JE, Schaer GL. EFFECT OF 100-PERCENT OXYGEN ADMINISTRATION ON INFARCT SIZE AND LEFT-VENTRICULAR FUNCTION IN A CANINE MODEL OF MYOCARDIAL-INFARCTION AND REPERFUSION. American Heart Journal 1995;130:957-65. 5. Maroko PR, Radvany P, Braunwald E, Hale SL. Reduction of infarct size by oxygen inhalation following acute coronary occlusion. Circulation 1975;52:360-8. 6. Madias JE, Madias NE, Hood WB, Jr. Precordial ST-segment mapping. 2. Effects of oxygen inhalation on ischemic injury in patients with acute myocardial infarction. Circulation 1976;53:411-7. 7. Stavitsky Y, Shandling AH, Ellestad MH, et al. Hyperbaric oxygen and thrombolysis in myocardial infarction: The 'HOT MI' randomized multicenter study. Cardiology 1998;90:131-6. 8. O'Neill WW, Martin JL, Dixon SR, et al. Acute myocardial infarction with hyperoxemic therapy (AMIHOT) - A prospective, randomized trial of intracoronary hyperoxemic reperfusion after percutaneous coronary intervention. Journal of the American College of Cardiology 2007;50:397-405. 9. Bodetoft S, Carlsson M, Arheden H, Ekelund U. Effects of oxygen inhalation on cardiac output, coronary blood flow and oxygen delivery in healthy individuals, assessed with MRI. European Journal of Emergency Medicine 2011;18:25-30. 10. Bourassa MG, Campeau L, Bois MA, Rico O. EFFECTS OF INHALATION OF 100 PERCENT OXYGEN ON MYOCARDIAL LACTATE METABOLISM IN CORONARY HEART DISEASE. American Journal of Cardiology 1969;24:172-&. 11. Ganz W, Donoso R, Marcus H, Swan HJ. Coronary hemodynamics and myocardial oxygen metabolism during oxygen breathing in patients with and without coronary artery disease. Circulation 1972;45:763-8. 12. Kenmure ACF, Murdoch WR, Beattie AD, Marshall JC, Cameron AJ. CIRCULATORY AND METABOLIC EFFECTS OF OXYGEN IN MYOCARDIAL INFARCTION. British Medical Journal 1968;4:360-&. 13. McNulty PH, Maduakor E, King N, et al. Administration of oxygen to patients undergoing cardiac catheterization acutely reduces coronary blood flow. Circulation 2003;108:529-. 14. Moradkhan R, Sinoway LI. Revisiting the Role of Oxygen Therapy in Cardiac Patients. Journal of the American College of Cardiology 2010;56:1013-6. 15. Sukumalchantra Y, Levy S, Danzig R, Rubins S, Alpern H, Swan HJ. Correcting arterial hypoxemia by oxygen therapy in patients with acute myocardial infarction. Effect on ventilation and hemodynamics. The American journal of cardiology 1969;24:838-52.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 23

SUPPLEMENTARY DATA 16. Mak S, Azevedo ER, Liu PP, Newton GE. Effect of hyperoxia on left ventricular function and filling pressures in patients with and without congestive heart failure. Chest 2001;120:467-73. 17. McNulty PH, Robertson BJ, Tulli MA, et al. Effect of hyperoxia and vitamin C on coronary blood flow in patients with ischemic heart disease. Journal of Applied Physiology 2007;102:2040-5. 18. Zweier JL, Talukder MAH. The role of oxidants and free radicals in reperfusion injury. Cardiovascular Research 2006;70:181-90. 19. Kalogeris T, Bao Y, Korthuis RJ. Mitochondrial reactive oxygen species: a double edged sword in ischemia/reperfusion vs preconditioning. Redox Biol 2014;2:702-14. 20. Cabello JB, Burls A, Emparanza JI, Bayliss S, Quinn T. Oxygen therapy for acute myocardial infarction. Cochrane Database of Systematic Reviews 2010. 21. Rawles JM, Kenmure AC. Controlled trial of oxygen in uncomplicated myocardial infarction. Br Med J 1976;1:1121-3. 22. Wilson AT, Channer KS. Hypoxaemia and supplemental oxygen therapy in the first 24 hours after myocardial infarction: the role of pulse oximetry. Journal of the Royal College of Physicians of London 1997;31:657-61. 23. Ukholkina GB, Kostyanov IY, Kuchkina NV, Grendo EP, Gofman YB. Effect of oxygenotherapy used in combination with reperfusion in patients with acute myocardial infarction. Kardiologiya 2005;45:59-. 24. Stub D, Smith K, Bernard S, et al. A randomized controlled trial of oxygen therapy in acute myocardial infarction Air Verses Oxygen In myocarDial infarction study (AVOID Study). American Heart Journal 2012;163:339-U277. 25. Khoshnood A, Carlsson M, Akbarzadeh M, et al. The Effects of Oxygen Therapy on Myocardial Salvage in ST Elevation Myocardial Infarction Treated with Acute Percutaneous Coronary Intervention: The Supplemental Oxygen in Catheterized Coronary Emergency Reperfusion (SOCCER) Study. Cardiology 2015;132:16-21. 26. Frobert O, Lagerqvist B, Gudnason T, et al. Thrombus Aspiration in ST-Elevation myocardial infarction in Scandinavia (TASTE trial). A multicenter, prospective, randomized, controlled clinical registry trial based on the Swedish angiography and angioplasty registry (SCAAR) platform. Study design and rationale. American Heart Journal 2010;160:1042-8. 27. Jernberg T, Attebring MF, Hambraeus K, et al. The Swedish Web-system for enhancement and development of evidence-based care in heart disease evaluated according to recommended therapies (SWEDEHEART). Heart 2010;96:1617-21. 28. Hansson L, Hedner T, Dahlof B. Prospective randomized open blinded end-point (PROBE) study. A novel design for intervention trials. Prospective Randomized Open Blinded End-Point. Blood pressure 1992;1:113-9. 29. Montori VM, Guyatt GH. Intention-to-treat principle. Canadian Medical Association Journal 2001;165:1339-41. 30. Thang ND, Karlson BW, Bergman B, et al. Characteristics of and outcome for patients with chest pain in relation to transport by the emergency medical services in a 20- year perspective. The American journal of emergency medicine 2012;30:1788-95. 31. Thang ND, Karlson BW, Bergman B, et al. Patients admitted to hospital with chest pain - Changes in a 20-year perspective. Int J Cardiol 2013;166:141-6.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 24

SUPPLEMENTARY DATA 2. Statistical analysis plan

This present statistical analysis plan is the original and final version which was established before data base lock and unblinding of the trial data.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 25

SUPPLEMENTARY DATA

Statistical analysis plan

DETO2X-AMI Ollie Östlund, PhD, biostatistician

SAP version: 1.0

Based on protocol version and date: Protocol version 2013-03, date 2013-11-26

Biostatistician: Ollie Östlund

UCR Project Leader: Eva Jacobsson

Client representative: Leif Svensson

Review send list: Robin Hofmann Tomas Jernberg Bertil Lindahl Stefan James Leif Svensson

Uppsala Clinical Research Center 2016 Author: Ollie Östlund, [email protected] www.ucr.uu.se ©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA Innehållsförteckning Introduction ...... 30

Objectives and outcome variables ...... 30

Objectives ...... 30

Primary Objective ...... 30

Secondary Objective(s) ...... 30

Outcome variables ...... 30

Primary outcome variable ...... 30

Secondary outcome variables ...... 30

Study design ...... 32

Design ...... 32

Population ...... 32

Interventions ...... 32

Follow-up ...... 32

Blinding ...... 32

Definition of Analysis Populations ...... 32

ITT (Intention-to-treat population) ...... 32

AMI (Acute myocardial infarction population) ...... 32

PP (Per-protocol population) ...... 33

STEMI-PCI (ST-elevation MI treated with percutaneous coronary intervention) ...... 33

Non-DETO2X AMI (Eligible patients with AMI discharge diagnosis, not randomised) ...... 33

Non-DETO2X STEMI-PCI (Eligible patients with ST-elevation MI treated with PCI) ...... 33

Description of statistical analysis ...... 34

Study conduct and Subject/Patient disposition ...... 34

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 27

SUPPLEMENTARY DATA Baseline Characteristics and Treatment Group Comparability ...... 34

Treatment Administration/Compliance ...... 34

Efficacy analyses ...... 34

General analytical considerations ...... 34

General considerations for descriptive statistics ...... 34

All-cause death (Död), primary end point ...... 35

Secondary efficacy analyses ...... 36

NYHA heart failure classification and EQ-5D ...... 36

Ventricular function in hospital ...... 37

Other end points ...... 37

Handling of Missing Data or data outside range ...... 38

Safety analyses ...... 39

Other planned analyses ...... 39

Analysis data base definitions ...... 39

Data sources and terminology ...... 39

Analysis populations ...... 41

Study conduct and Subject/Patient disposition ...... 42

Treatment Administration/Compliance ...... 43

Baseline Characteristics and Treatment Group Comparability ...... 45

Hospitalisation ...... 56

Discharge medication ...... 58

STEMI-PCI population additional background and procedural variables from SCAAR ...... 59

Index PCI ...... 59

1.1.1. STEMI-PCI additional background and clinical presentation variables ...... 59

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 28

SUPPLEMENTARY DATA PCI procedural data ...... 62

Index PCI procedural variables summarised from lesion level data ...... 71

1.1.1. Complications to index PCI ...... 78

Efficacy outcomes ...... 82

Follow-up time per patient ...... 82

Outcome events in all populations ...... 82

Outcome events in the PCI-STEMI population only ...... 85

SEPHIA outcome variables ...... 88

Clean file documentation data included in the analysis data base ...... 90

Unintentional randomisations not included in ITT ...... 90

Corrections to DETO2X module data ...... 90

Re-matching of DETO2X modules and/or SCAAR PCI records to RIKS-HIA admission periods ...... 91

Corrections to DISCHARGE data ...... 91

Mortality, age and sex for patients not in SWEDEHEART ...... 91

Determination of sample size ...... 91

Interim Analysis Plan ...... 93

Changes in the Planned Analysis ...... 93

Description of Output ...... 94

Statistical software ...... 95

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 29

SUPPLEMENTARY DATA Introduction The purpose of this SAP is to describe the planned statistical analyses and data presentation for the primary follow-up period for the DETO2X-AMI trial. The SAP describes all pre-defined analyses based on the primary follow-up cut-off date, December 30, 2016, one year after inclusion of the last patient. Long-time follow-up is planned but not described in this SAP. Note that outcome variables derived from the registries outside SWEDEHEART will be available at a later time point than SWEDEHEART variables, and analyses will be performed when data becomes available. Health economic analyses are not described in this SAP. The output specification is given in a separate document. Objectives and outcome variables

Objectives

Primary Objective  To compare survival up to one year between treatment with oxygen according to guidelines and no oxygen treatment, in patients with suspected AMI.

Secondary Objective(s)  To compare major adverse cardiac events (MACE), defined as all-cause death or rehospitalisation with MI or heart failure, up to one year, between treatment with oxygen according to guidelines and no oxygen treatment, in patients with suspected AMI.  To compare survival and adverse cardiac events up to one year in subgroups of patients, including diagnosis of AMI, type of AMI (STEMI and NSTEMI), diagnosis of COPD, renal failure, and diabetes mellitus, oxygen saturation at inclusion (90-94% and 95-99%), and gender.  Perform health economic analyses (not covered in this SAP).

Outcome variables

Primary outcome variable  Time to all-cause death as reported in the Swedish population registry, censored at 365 days, in the ITT population.

Secondary outcome variables Pre-defined in the trial protocol  Time to first of all-cause death, rehospitalisation with MI defined as a post-discharge admission period in the SWEDEHEART registry with discharge diagnosis including ICD-10 code I21 or I22 with subcategories, or rehospitalisation with heart failure defined as a post-discharge registration in the Swedish inpatient registry with main diagnosis containing ICD-10 code I50 with subcategories, or I11.0, I13.0 or I13.2, censored at 365 days, in the ITT population.  Primary outcome in the AMI population (most important secondary variable).

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 30 SUPPLEMENTARY DATA  Time to first of all-cause death, rehospitalisation with MI or rehospitalisation with heart failure in the AMI population. Secondary outcome variables not defined in the trial protocol  Time to first of all-cause death or rehospitalisation with heart failure, censored at 365 days, in the ITT population (considered the most important secondary variable, after death in the ITT and AMI population), and in the AMI population.  The primary outcome in the STEMI-PCI population.  Time to first rehospitalisation with MI, censored for death or at 365 days, in the ITT, AMI and STEMI-PCI population.  Time to first of all-cause death or rehospitalisation with MI, censored at 365 days, in the ITT, AMI and STEMI-PCI population.  Time to first rehospitalisation with heart failure, censored for death or at 365 days, in the ITT and AMI population.  Time to cardiovascular death, defined as a registered death in the Swedish cause of death registry with underlying cause ICD-10 code in the "I" range, or underlying cause missing, censored at 365 days, in the ITT and AMI population.  NYHA heart failure classification at 6-10 weeks and at 12-14 months as reported in the SEPHIA registry, in AMI population patients with data.  EQ-5D domain scores and mean score at 6-10 weeks and at 12-14 months as reported in the SEPHIA registry, in AMI population patients with data.  EQ-5D VAS score at 6-10 weeks and at 12-14 months as reported in the SEPHIA registry, in AMI population patients with data.

These outcomes were specified after the trial had started, with access to blinded outcome data from the SWEDEHEART registries, but before any treatment comparisons were available. Due to unexpected low mortality in the trial patients, the trial has low power to show effect on the primary endpoint of death, and secondary outcomes may be better powered to estimate the effect of oxygen treatment. To aid interpretation of the multiple secondary endpoints, in particular for a non-significant primary outcome, the investigators decided in March 2014 that the composite of death and rehospitalisation with heart failure is to be considered the most important secondary endpoint, after the primary endpoint of death in the ITT and AMI populations. Compared to the protocol-defined secondary outcome, this composite does not include rehospitalisation with MI, because MI not believed to be very responsive to the possible physiologic effect of oxygen treatment. See also the sample size calculations in Section 0. STEMI-PCI population only (for definitions, see Section 0)  Time to first of all-cause death, rehospitalisation with MI or cardiogenic shock, or stent thrombosis, censored at 365 days  Time to first of all-cause death, rehospitalisation with MI or cardiogenic shock, stent thrombosis, or target vessel revascularisation, censored at 365 days  Time to target vessel revascularisation, censored at death or 365 days  Time to target segment revascularisation, censored at death or 365 days  Time to restenosis, censored at death or 365 days  Time to stent thrombosis, censored at death or 365 days  Time to rehospitalisation with cardiogenic shock, censored at death or 365 days

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 31

SUPPLEMENTARY DATA Study design

Design Parallel group randomised trial, with 1:1 randomisation performed online in SWEDEHEART.

Population Normoxic (≥90% oxygenation) patients aged 30 or above, presenting in ambulance or at ER with suspected acute myocardial infarction, as indicated by symptoms within the last 6 hours plus ECG changes or increased troponin.

Interventions  Oxygen: Oxygen for 6-12 hours  Room air: No oxygen given

Follow-up Mortality, new hospitalisations with MI or heart failure as defined by diagnose codes, new vascularisations, and heart failure classification and EQ-5D at 1 and 6 months, obtained from SWEDEHEART, the national inpatient registry, and the population registry. Primary cut-off for follow-up: one year after inclusion of the last patient. Primary observation time for analyses of patient data: one year after inclusion of the patient.

Blinding The treatment was administered in an unblinded fashion. The randomisation list is kept at the UCR quality registry section with access restricted to authorised support personnel, and will not be made available for analysis until after clean file and data base lock, to keep the steering committee blinded to any emerging treatment differences. Blinded summaries of background characteristics and mortality in the pooled treatment arms were available for central monitoring throughout the trial.

Definition of Analysis Populations

ITT (Intention-to-treat population) All intentionally randomised patients possible to identify for follow-up. Randomisation numbers assigned unintentionally, such as by clicking the wrong box or randomising for the wrong patient record in SWEDEHEART, will be recorded in the clean file documentation and removed from the analysis data base. Attempts will be made for randomisation numbers with incorrect patient identifiers to identify if there is an intentionally randomised patient linked to the number. Randomisation numbers for which it is not possible to clearly identify a patient for follow-up of mortality will be recorded and removed from the analysis data base, even if a patient might have been randomised. No patients will be removed from the ITT population due to violated inclusion or exclusion criteria or deviations from the treatment algorithm.

AMI (Acute myocardial infarction population) All ITT patients with a SWEDEHEART discharge diagnosis including ICD-10 codes I21 (acute myocardial infarction) or I22 (reinfarction), including all subcategories.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 32

SUPPLEMENTARY DATA PP (Per-protocol population) All ITT patients, excluding patients with reported treatment not completed, due to other reason than hypoxia. Patients excluded from the PP will be recorded in the clean file documentation and tabulated in the report.

STEMI-PCI (ST-elevation MI treated with percutaneous coronary intervention) All patients with discharge diagnosis STEMI/LBBB (DISCHARGE::INFARCTTYPE=1), that have a PCI record in SCAAR for the index admission period (matched by MCEID), with general success (PCI::SUCCESS) not missing and not equal to "Endast diagnostik" (numerical value 2).

Non-DETO2X AMI (Eligible patients with AMI discharge diagnosis, not randomised) All patients aged ≥30 not included in the ITT, recorded in SWEDEHEART as having been admitted to a DETO2X-reqruiting CCU between the first date a patient was included at that centre and December 30, 2015 (for Sundsvall site only patients up to date of last included patient, 2014-06-06), with discharge diagnosis including ICD-10 code I21 or I22 including all subcategories, that have a Swedish personal number. If a patient was admitted at recruiting centres several times during the period, only the first admittance is counted. The definition of DETO2X-recruiting CCU is based on the first 7 characters of DETO2X::CENTREID and admission page CENTREID. For example, patients are considered Non-DETO2X AMI patients at the DETO2X recruiting site "SE10011DTX" if they have admission page CENTREID "SE10011HIA". Note that only patients admitted at a randomising CCU in this sense are included in Non-DETO2X AMI, although DETO2X patients were regularly admitted to another CCU than the one performing the randomisation. The reason to exclude patients without Swedish personal number is that they cannot be followed for mortality in Swedish registries. The Non-DETO2X AMI population constitutes an observational registry study parallel to the DETO2X trial, to compare DETO2X and non-DETO2X patients in terms of background characteristics and outcome. In general, the Non-DETO2X AMI patients are expected to represent a more vulnerable population than the DETO2X AMI patients, including hypoxic and unconscious patients.

Non-DETO2X STEMI-PCI (Eligible patients with ST-elevation MI treated with PCI) Non-DETO2X AMI patients satisfying the same criteria as the STEMI-PCI population.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 33 SUPPLEMENTARY DATA Description of statistical analysis

Study conduct and Subject/Patient disposition Inclusion will be described as Kaplan-Meier plots of patients randomised per calendar day, first and last inclusion date, descriptive statistics of time from randomisation to death or censoring, and number of patients by randomising centre, for the ITT, AMI, and STEMI-PCI populations.

Baseline Characteristics and Treatment Group Comparability Baseline characteristics will be described by randomised treatment and in total for the ITT, AMI, PP, STEMI-PCI and Non-DETO2X AMI population. Categorical data will be described as total number and percentage, with missing data as a separate category. Numerical data will be described using number of patients with data, and median, quartiles, arithmetic mean and standard deviation for patients with data. "Statistical testing" comparing the randomised treatment groups will be performed using chi- square tests for categorical variables and Wilcoxon's test for numerical data, using observed cases. The result will be presented as p-values and used only if needed to satisfy a non-CONSORT journal house style that requires Table 1 p-values. Since the groups are randomised, all perceived differences will be due to chance.

Treatment Administration/Compliance Completed randomised treatment or reason for not completing randomised treatment will be presented as total number and percentage, by randomised treatment, for the ITT, AMI and STEMI-PCI population. Any loss to follow-up for mortality or SWEDEHEART events will be presented as patient listings. For statistical analyses of outcomes, the total number of patients with data for the analysis will be tabulated in the descriptive statistics for the outcome in question. The report will contain sufficient information to construct a CONSORT diagram. Duration of treatment and final saturation will be presented descriptively by randomised group using median and quartiles, for observed cases, in the ITT, AMI, STEMI-PCI and PP population.

Efficacy analyses

General analytical considerations All efficacy analyses will compare randomised treatment, and the results will be presented as treatment contrasts with 95% confidence interval and two-sided p-value. No formal adjustment for multiplicity above designating a primary outcome and analysis model will be used. While the ITT population is of primary interest in this pragmatic real-world study, estimands relating to the AMI population are clearly clinically relevant. The purpose of the STEMI-PCI population is to assess cardiovascular end points in this subset, but these patients also have registry variables that are not recorded for all ITT or AMI patients.

General considerations for descriptive statistics Tables will present outcome data by randomised treatment group and in total for the ITT, PP, AMI and STEMI-PCI populations, and in total for the Non-DETO2X-AMI population. Categorical data will be presented as the number and percentage of patients in each category. Tables for events will present the number and percentage of patients with event, during entire follow-up, 365 days, and 30 days. For this purpose, any patient lost to follow-up before that time

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 34

SUPPLEMENTARY DATA without an event will be counted as no event. The number of patient-years and events/patient year will also be tabulated for the entire follow-up and with cut-offs 365 and 30 days. Events will be presented graphically by randomised treatment as Kaplan-Meier curves up to 1 year and for the entire follow-up time, for the ITT, PP, AMI and STEMI-PCI populations, and in additional plots showing the entire DETO2X-AMI group against the entire Non-DETO2X- AMI group.

All-cause death (Död), primary end point Primary analysis The hazard ratio between the oxygen group and the room air group will be estimated, in the ITT population, using a Cox proportional hazard model with factors randomised treatment and sex, and age at randomisation in years as an untransformed covariate. Death after day 365 (with day of randomisation as day 0) will be censored on day 365. A two-sided p-value <0.05 will be regarded as statistically significant. Sensitivity analyses Primary analysis repeated with censoring of events after day 30. Primary analysis repeated without censoring at one year, using the full follow-up time per patient. Logistic regression for the outcome death before or on day 365, with the same covariates as the primary analysis. Logistic regression for the outcome death before or on day 30, with the same covariates as the primary analysis. Primary analysis repeated without adjustment for sex and age. AMI population All analyses will be repeated in the AMI population. The primary analysis result in the AMI population is considered a key secondary outcome, and the other analyses being sensitivity analyses. In addition, differential effect of oxygen in the AMI population on the hazard ratio scale will be investigated using a Cox regression analysis in the ITT population, censored at 365 days, with factors randomised treatment, AMI or Non-AMI, sex, age, and randomised treatment – AMI subgroup interaction, and presented as hazard ratios with confidence intervals for each group, the ratio of the AMI to the non-AMI hazard ratio with confidence interval, and the interaction p- value. The AMI population hazard ratio from this analysis will be considered secondary and the estimate from the primary analysis model primary, if there is a difference. Per-protocol population All analyses will be repeated in the PP population, as sensitivity analyses. STEMI-PCI population All analyses will be repeated in the STEMI-PCI population. Subgroup analyses The subgroup of primary interest is the DETO2X-AMI population, described above. The subgroup analyses of mortality below will also be performed. All subgroup analyses will be performed in a Cox regression model, censored at 365 days, with factors randomised treatment, age, gender, subgroup indicator, and treatment – subgroup interaction, and presented as hazard ratios for each subgroup with confidence intervals, ratio of hazard ratios with confidence interval, and interaction p-value. All analyses will be based on observed cases only. For each subgroup, the

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 35

SUPPLEMENTARY DATA total number of patients, and the number and percentage of patients with event within 365 days will be tabulated. STEMI/non-STEMI patients in the AMI population (excluding, if there are any, AMI population patients with non-infarction classification according to INFARCTIONTYPE). Male patients/female patients in the ITT population. Previous hospitalisation with diagnosis of COPD or not, in the ITT population. This analysis will be performed when data on historical diagnoses has been obtained from Socialstyrelsen. Oxygen saturation <95%/≥95% in the ITT population. Renal failure (Njursvikt) or not in the ITT population. Diabetes mellitus or not in the ITT population. Age <68/≥68 (ITT population median) in the ITT population. Checking of assumptions Homogeneity of adjusted hazard ratios over time will mainly be assessed by comparison of the pre-defined sensitivity analyses. A notably varying hazard ratio will manifest as different estimated hazard ratios at 30 days and at 1 year, and will in that case be interpreted using the short and long time estimates of hazard and odds ratios. The fit of the pre-defined functional form for age adjustment is not vital to the interpretation of the hazard ratios and will not be checked. The p- values from the pre-defined primary analysis model will be used for hypothesis testing regardless of model fit.

Secondary efficacy analyses

NYHA heart failure classification and EQ-5D Quality of life and heart failure classification (shortness of breath) will be presented for observed cases in the AMI population, using values from both physical visits and telephone visits. Measurements at 6-10 week follow-up and at 12-14 month follow-up will be handled as separate variables. Visits outside these visit windows will not be excluded; all data registered as visit 1 will be included as "6-10 weeks" and all data registered as visit 2 will be included as "12-14 months". To assess the amount of missing data, in the population of AMI patients <75 years of age (the age group for mandatory SEPHIA follow-up), the number and percentage of patients with some data for the 6-10 week visit, and for the 12-14 month visit, will be tabulated by treatment. Time of visit will be tabulated as the number of days from inclusion to visit, as mean, quartiles, and extreme values for all AMI patients with visits. NYHA classification For NYHA classification, the number and percentage of patients in each class will be tabulated by treatment group, with the denominator being patients with data. NYHA class will be analysed using proportional odds logistic regression with treatment group as a factor, and presented as the common odds ratio with 95% confidence interval and p-value. A supportive analysis will be provided using cumulative logistic regression without assumption of proportional odds, and presented as estimated odds ratios for class I-IV vs 0, II-IV vs 0-I, and III-IV vs 0-II with 95% marginal confidence intervals, and the overall p-value for treatment difference. For these analyses, NYHA class IV and class III will be collapsed to one category, since blind review of the data while preparing the SAP indicated only three patients with NYHA class IV at 6-10 weeks and 1 patient at 12-14 months, in data extracted in April 2016.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 36

SUPPLEMENTARY DATA EQ-5D Number and percentages of patients in the three different response categories will be presented by treatment for each domain in tables and graphs, with the denominator being patient with data. The primary analysis model will be unadjusted proportional odds logistic regression, estimating a common odds ratio with 95% confidence interval and p-value. For this purpose, category 2 and 3 will be collapsed for the Mobility and Hygiene domains, since blind review of data extracted in April 2016 showed that only a few patients reported category 3 Mobility or Hygiene. Additional descriptive statistics will be given by treatment as mean domain score and standard deviation, and a raw summary score across domains will be calculated as the mean of non- missing domain scores and presented using mean and standard deviation. Supportive statistical analyses will be given using both cumulative logistic regression without assumption of proportional odds, presented as cut-point specific odds ratios with marginal 95% confidence intervals and the global p-value, and using a linear model with treatment group as a factor, and presented as the difference in mean score with 95% confidence interval and p-value, the latter also for the raw summary score. EQ-5D VAS scale will be summarised by treatment group as mean and standard deviation and analysed using a linear model with treatment group as a factor, and presented as the difference in mean with 95% confidence interval and p-value. The full empirical cumulative distribution will be presented graphically by treatment.

Ventricular function in hospital Ventricular function in hospital, last value in admission period, will be presented descriptively as the number and percentage of patients in each category for the ITT and AMI population. Ventricular function will be analysed using observed cases by cumulative logistic regression without assumption of proportional odds with treatment group as a factor, and presented as estimated odds ratios for all cut-points with 95% marginal confidence intervals, and the overall p- value for treatment difference.

Other end points For presentation of end points, also see general considerations above. ITT, PP and AMI population

Cardiovascular death (CV-död) This outcome will be presented using the same primary and sensitivity analyses as for the primary outcome. Composite of all-cause death or rehospitalisation with heart failure (Död eller återinläggning med hjärtsvikt) This composite will be presented using the same primary and sensitivity analyses as for the primary outcome. This is considered the main secondary end point in the ITT population since March 2014. Composite of all-cause death, rehospitalisation with MI, or rehospitalisation with heart failure (Död, återinläggning med hjärtinfarkt eller återinläggning med hjärtsvikt) This composite will be presented using the same primary and sensitivity analyses as for the primary outcome. This is the pre-defined secondary end point based on the protocol.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 37

SUPPLEMENTARY DATA Composite of all-cause death or rehospitalisation with MI (Död eller återinläggning med hjärtinfarkt) This composite will be presented using the same primary and sensitivity analyses as for the primary outcome. The death or reinfarction outcome is mainly of interest for the AMI population, but will be presented for ITT as well. Rehospitalisation with myocardial infarction (Återinläggning med hjärtinfarkt) This outcome will be presented descriptively and using the primary analysis model. Rehospitalisation with heart failure (Återinläggning med hjärtsvikt) This outcome will be presented descriptively and using the primary analysis model.

STEMI-PCI population Composite of all-cause death, rehospitalisation with MI or cardiogenic shock, or stent thrombosis; STEMI-PCI MACE 1 (Död, återinläggning med hjärtinfarkt eller kardiogen chock, eller stenttrombos) This outcome will be presented using the same primary and sensitivity analyses as for the primary outcome. Composite of all-cause death, rehospitalisation with MI or cardiogenic shock, or stent thrombosis or target vessel revascularisation; STEMI-PCI MACE 2 (Död, återinläggning med hjärtinfarkt eller kardiogen chock, eller stenttrombos eller target vessel revascularisation) This outcome will be presented using the same primary and sensitivity analyses as for the primary outcome. Composite of all-cause death or rehospitalisation with MI (Död eller återinläggning med hjärtinfarkt) This composite will be presented using the same primary and sensitivity analyses as for the primary outcome. Rehospitalisation with myocardial infarction (Återinläggning med hjärtinfarkt) This outcome will be presented descriptively and using the primary analysis model. Target vessel revascularisation This outcome will be presented descriptively and using the primary analysis model. Target segment revacularisation This outcome will be presented descriptively and using the primary analysis model. Restenosis (Restenos) This outcome will be presented descriptively and using the primary analysis model. Stent thrombosis (Stenttrombos) This outcome will be presented descriptively and using the primary analysis model. Rehospitalisation with cardiogenic shock (Återinläggning med kardiogen shock) This outcome will be presented descriptively and using the primary analysis model.

Handling of Missing Data or data outside range Any loss to follow-up will be handled as censoring in analyses of time to event, and as "no event after drop-out" in analyses of patients with event. No or very little loss to follow-up is expected since the patients are followed through registries and no patients have withdrawn consent to follow-up at time of SAP completion. SEPHIA data will be analysed using observed cases.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 38

SUPPLEMENTARY DATA Background and treatment administration data will be presented with missing data as a separate category, or with the number of patients with data for numerical data. Time from symptom onset to randomisation, time from symptom onset to PCI, time from ECG to randomisation and time from ECG to PCI will be set to missing if <0 or >48 hours. See also Sections 0. Obviously incorrect durations of treatment or of hospital stay will be corrected or removed as part of data cleaning and recorded in the clean file documentation. A few background variables are collected both in RIKS-HIA and SCAAR. Since the DETO2X trial is based on RIKS-HIA, the RIKS-HIA value will be used by default. If the RIKS-HIA value is missing or recorded as unknown, for STEMI-PCI population patients (DETO2X and Non-DETO2X), the SCAAR value for the index PCI in the admission period will be used, if not missing.

Safety analyses SWEDEHEART-reported events, procedures and complications during index admission period will be presented for the ITT, AMI and STEMI-PCI populations as number and percent of patients. RIKS-HIA data (see Section 0will be presented for all populations, and SCAAR data relating to the index PCI (see Section 0) for the STEMI-PCI population.

Other planned analyses Other exploratory analyses, long-time follow up and health economic analyses may be performed but are not covered by this SAP.

Analysis data base definitions

Data sources and terminology Data will be collected from SWEDEHEART registries RIKS-HIA (CCU care), SCAAR (angiography and PCI), and SEPHIA (secondary prevention follow-up visits for patients up to 75 years old), from the DETO2X-module in SWEDEHEART, and from the national in- and outpatient care registries and cause of death registry. Date of death is obtained from the Swedish population registry through SWEDEHEART to which the information is routinely added every month. Each DETO2X-module is coupled with a RIKS-HIA patient admission module, where the randomisation was initiated. A basic structure in RIKS-HIA is a admission period, which consists of one admission module, one discharge module (discharge from last SWEDEHEART-reporting unit), and any number of medical care modules, and possibly biomarker modules, and is identified by the unique admission period identifier MCEID. Angiography and PCI events in SCAAR and post-hospitalisation follow-up in SEPHIA are also coupled to a RIKS-HIA admission period by MCEID. All individual admission, discharge, medical care, angiography/PCI procedure and SEPHIA modules are given a unique SID (subevent ID). SCAAR data is also recorded on sub-procedure level (typically one record per treated lesion, with location identified on coronary vessel segment level) repeated within the PCI SID, identified by SEGID. "Index hospitalisation" will be defined as the index RIKS-HIA admission period. Index follow-up visits (recorded in SEPHIA) and angiographies/PCI:s during index hospitalisation (recorded in SCAAR) will be obtained as those linked to the RIKS-HIA index admission period by MCEID.

Source Prefix in tables Comment

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 39

SUPPLEMENTARY DATA ADB (analysis data base) None Trial-specific DETO2X- DETO2X:: module SWEDEHEART common PATIENT:: patient information module RIKS-HIA admission INITIALCARE:: module RIKS-HIA discharge module DISCHARGE:: RIKS-HIA discharge DIAGNOSE:: diagnoses module RIKS-HIA care module MEDICALCARE:: (repeated within admission period, one for each involved CCU) SCAAR PCI lab admission REGANGIO:: module SCAAR segment module SEGMENT:: (may be repeated even within segment, unique row key SEGID, matched to REGANGIO by SID) SCAAR restenosis data RESTENOS:: (matched to PCI where restenosis was registered in REGANGIO by SID, and to original segment procedure in SEGMENT by SEGID) SEPHIA FOLLOWUP:: External sources Clean file documentation Data resulting from queries to site and corrections of obvious data entry errors, for example matching of randomisations without linked admission period with the correct admission period. End point follow-up external to the registries may be performed for some patients, in particular manual checks of mortality and death dates for randomised patients that are not present in SWEDEHEART, and

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 40

SUPPLEMENTARY DATA matching of future events in SWEDEHEART if the patient becomes included at a later date. National inpatient registry PAR_SV:: For the first version of the statistical report and the main National outpatient registry PAR_OV:: result publication on mortality, registry data obtained in spring 2017, complete up to 2015-12-31, will be used for background data. For the final report and possibly for subsequent publications, registry data obtained in autumn 2017, complete up to 2016-12-31, will be used for background and outcome data. Hence differences in reported background data may appear between the first statistical report and the final statistical report. National cause of death DORS:: registry

Analysis populations Label Format Name Sources Derivation ted values

ITT Ja/nej ITT DETO2X::RANDNO DETO2X::RANDNO not missing and not excluded Clean file randomization number. documentation.

PP Ja/nej PP TREATMENTCOMPLETE ITT=Ja and D_X TREATMENTCOMPLETED _X is Ja or Nej: Hypoxi

AMI Ja/nej DETO2X_AMI ITT, DIAGNOSE::CODE ITT=Ja and diagnosis fields of index hospitalisation contains string 'I21' or 'I22'

STEM Ja/nej STEMIPOP DISCHARGE::INFARCTT ITT=Ja and I-PCI YPE, PCI::SUCCESS DISCHARGE::INFARCT TYPE=1 and

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 41 SUPPLEMENTARY DATA PCI::SUCCESS not in (.,2) for a PCI event matched on MCEID.

Note: The PCI with lowest SID in the admission period will be considered the index PCI, for which background variables, procedures and complications are reported. Any other PCI in the admission period will be reported only as outcome data.

Non- Ja/nej NON_DETO2X_AMI ITT, Not ITT=Ja, and DETO DETO2X::ARRIVALDATE (ADMISSION_DATE 2X , between ARRIVALDATE AMI INITIALCARE::ADMISSIO of first ITT patient at N_DATE, same CENTREID_NEW CENTREID_NEW, and December 30, 2015, PATIENT::D_DOB, or June 6, 2014 iff DIAGNOSE::CODE, CENTREID_NEW='SE6 PATIENT::DEAD 2010' (Sundsvall)), and diagnosis fields of index hospitalisation contains string 'I21' or 'I22', and FLOOR((ADMISSION_ DATE- D_DOB)/365.25)>=30, and DEAD non-missing (it is missing iff the patient lacks personal number).

Non- Ja/nej NON_DETO2X_STE DISCHARGE::INFARCTT Non-DETO2X AMI DETO MIPOP YPE, patients with STEMI and 2X PCI, as derived for the PCI::SUCCESS STEM STEMI-PCI population. I-PCI NON_DETO2X_AMI

Study conduct and Subject/Patient disposition Label Formatted Name Sources Derivation values

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 42

SUPPLEMENTARY DATA Inklusionsd STARTDAT ITT: For ITT patients, atum E DETO2X::ARRIVALD DETO2X::ARRIV ATE, ALDATE (randomisation Non-DETO2X: date). For Non- INITALCARE::ADMI DETO2X AMI, SSION_DATE ADMISSION_DA (subevent date for TE INITIALCARE) Clean file documentation for any corrections (excel sheet) Inklusionsti STARTTIME ITT: d [not DETO2X::ARRIVALT tabulated] IME, Non-DETO2X: INITALCARE::ADMI SSION_TIME (subevent time for INITIALCARE) Clean file documentation for any corrections (excel sheet) Center Hospital CENTREID_ ITT: Substr(CENTREI names, format NEW DETO2X::CENTREI D,1,7)||"D" derived from D, Non-DETO2X: format of INITIALCARE::CENT DETO2X::CE REID NTREID Utskrivning DISCHARG DISCHARGE::DISCH Missing values sdatum E_DATE ARGE_DATE filled in. (not tabulated) _F Clean file documentation

Treatment Administration/Compliance Label Formatte Name Sources Derivation d values Fullföljde Ja/ TREATMENTCO Clean file documentation The exported s MPLETED_X (derived from DETO2X variables behandlin Nej: will be filled in with gen? Hypoxi/ DETO2X:: information by TREATMENTCOMPLE Robin Hofmann, to Nej: TED, Patiente an excel file

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 43

SUPPLEMENTARY DATA n DETO2X::NOTCOMPL enclosed with the avböjde ETEDREASON clean file fortsatt DETO2X::OXYGENGI documentation, deltagan VEN, which is then de/ DETO2X::NOTCOMPL imported to the ETEDCOMMENT, analysis data set. Nej: Filled in data Tidig DETO2X::COMMENT) (TREATMENTC utskrivni OMPLETED_F ng/ and Nej: NOTCOMPLETE Annat/ DREASON_F) are joined to one Nej: variable. Change of If clinical NOTCOMPLETE circumst DREASON is ances/ filled in with 7 (=completed), Nej: TREATMENTCO Mistakes MPLETED_F and (technica _X is set to Yes. l, commun ication)

Varför Text NOTCOMPLETE DETO2X:: fullföljdes DCOMMENT NOTCOMPLETEDCO inte MMENT behandlin gen? Kommen Text COMMENT DETO2X::COMMENT tar Tid från Num TREATMENT_TI DETO2X::NOTCOMPL Missing and randomis ME ETEDDATE notably incorrect ering till values filled in as behandlin DETO2X::NOTCOMPL specified in the gsavslut ETEDTIME clean file (timmar) DETO2X::TREATEND documentation. _DATE Time from DETO2X::TREATEND STARTDATE, _TIME STARTTIME. STARTDATE Time to TREATEND_DA STARTTIME TE_F (including filled-in and

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 44

SUPPLEMENTARY DATA Clean file documentation corrected values) if non-missing, otherwise to NOTCOMPLETE D_DATE, and to TREATEND_TIM E_F (including filled-in and corrected values) if non- missing, otherwise time to NOTCOMPLETE DTIME. Missing unless both date and time non- missing. Syremättn Num SATURATION_F DETO2X::SATURATIO IF ad vid N SATURATION_F behandlin =0 THEN gsavslut Clean file documentation SATURATION_F is set to missing. Missing data is not filled in systematically, but data was added for some patients where it was available. Inskrivni Ej STATUS_IC SUBEVENT::STATUS Derived by QReg, ngssida påbörjad included in RIKS- /Ej DETOX_IC_DC HIA klar/Kla data set. finns r Utskrivni Ej STATUS_DC SUBEVENT::STATUS ngssida påbörjad RIKS- /Ej HIA klar/Kla finns r

Baseline Characteristics and Treatment Group Comparability

Label Formatted Name Sources Derivation values

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 45

SUPPLEMENTARY DATA Kön Man/Kvi D_GENDER PATIENT::D_GEND Derived by the nna ER registry from personal number. For patients Clean file without a documentation SWEDEHEART Ålder vid Num D_AGE PATIENT::D_AGE record, filled in from inklusion excel sheet. (år) Clean file documentation Ålder Ja/nej AGE_68 D_AGE D_AGE>=68 >=68 år Längd (cm) Num HEIGHT_X INITIALCARE::HEI INITIALCARE will GHT be used primarily, missing and unknown REGANGIO::HEIGH values will be filled in T from the STEMI-PCI (DETO2X and Non- DETO2X) population index PCI if possible. If one value is unknown and the other missing, it will be set to unknown. Vikt (kg) Num WEIGHT_X INITIALCARE::WEI INITIALCARE will GHT be used primarily, missing and unknown REGANGIO::WEIG values will be filled in HT from the STEMI-PCI population (DETO2X and Non- DETO2X) index PCI if possible. If one value is unknown and the other missing, it will be set to unknown. BMI Num BMI INITIALCARE::WEI BMI=WEIGHT/(H (kg/m2) GHT EIGHT/100)**2 INITIALCARE::HEI GHT Rökning From SMOKING_ST INITIALCARE::SMO INITIALCARE will INITIALC ATUS_X KING_STATUS be used primarily, ARE:: missing and unknown REGANGIO::SMOKI values will be filled in NG_STATUS from the STEMI-PCI

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 46

SUPPLEMENTARY DATA SMOKIN population G_STATU (DETO2X and Non- S DETO2X) index PCI if possible. If one value is unknown and the other missing, it will be set to unknown. Diabetes Ja/nej DIABETES_X INITIALCARE::DIAB INITIALCARE will (SWEDEH X ETES be used primarily, EART+PA missing and unknown R) REGANGIO::DIABE values will be filled in TES from the STEMI-PCI PAR_SV::HDIA population (DETO2X and Non- PAR_SV::DIAGNOS DETO2X) index PCI PAR_SV::INDATUM if possible. If one value is unknown and PAR_OV::HDIA the other missing, it PAR_OV::DIAGNOS will be set to unknown. PAR_OV::INDATUM

If combined variable No or unknown, set to Yes if there is an ICD-code E10-E14, including all subcategories, as main or secondary diagnosis, in- or outpatient care, (STARTDATE- 5YR)UTD ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV) Set to No otherwise. Diabetes- Ja/nej DIAB_MED_I INITIALCARE::DIAB INITIALCARE will behandling NSULIN_X _MED_INSULIN be used primarily, insulin missing and unknown REGANGIO::DIABE values will be filled in TESINSULIN from the STEMI-PCI population (DETO2X and Non- DETO2X) index PCI

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 47

SUPPLEMENTARY DATA if possible. If one value is unknown and the other missing, it will be set to unknown. Previous to that, DIABETESINSULI N is set to 0 if REGANGIO::DIAB ETES=0. Hypertoni Ja/nej HYPERTENSI INITIALCARE::HYP If INITIALCARE (SWEDEH ON_XX ERTENSION variable No or EART+PA unknown, filled in R) REGANGIO::HYPER with REGANGIO TON variable from PAR_SV::HDIA STEMI-PCI population PAR_SV::DIAGNOS (DETO2X and Non- PAR_SV::INDATUM DETO2X) index PCI PAR_SV::UTDATUM if non-missing (both variables denote PAR_OV::HDIA current or previous PAR_OV::DIAGNOS hypertension with pharmaceutical PAR_OV::INDATUM treatment). If still unknown or missing, set to Yes if there is an ICD-code I10-I15, including all subcategories, as main or secondary diagnosis, in- or outpatient care, (STARTDATE- 5YR)UTD ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV). Set to No otherwise.

Tidigare Ja/nej PREVIOUS_M INITIALCARE::PRE If INITIALCARE hjärtinfarkt I_XX VIOUS_M variable No or (SWEDEH unknown, for PCI- EART+PA REGANGIO :TIDIN STEMI patients R) F (DETO2X and Non-

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 48

SUPPLEMENTARY DATA PAR_SV::HDIA DETO2X) use the REGANGIO PAR_SV::DIAGNOS variable if non- PAR_SV::INDATUM missing. If still No or unknown, set to Yes PAR_SV::UTDATUM if there is an ICD- PAR_OV::HDIA code I21-I22, including all PAR_OV::DIAGNOS subcategories, as main PAR_OV::INDATUM or secondary diagnosis, in- or outpatient care, (STARTDATE- 5YR)UTD ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV)Set to No otherwise. Tidigare Ja/nej previous_pci_X INITIALCARE::previ INITIALCARE will PCI ous_pci be used primarily, missing and unknown REGANGIO:TIDPCI values will be filled in from the STEMI-PCI population (DETO2X and Non- DETO2X) index PCI if possible. If one value is unknown and the other missing, it will be set to unknown. Tidigare Nej/CAB prior_cardiac_su INITIALCARE::prior_ INITIALCARE will hjärtkirurgi G/Annan/ rgery_x cardiac_surgery be used primarily, (avser ej Okänd missing and unknown pacemaker) REGANGIO::TIDCA values will be filled in BG with CABG from the STEMI-PCI population (DETO2X and Non- DETO2X) index PCI, if TIDCABG=1 the variable will be set to CABG.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 49

SUPPLEMENTARY DATA Känd From history_of_chf_ INITIALCARE::histor If INITIALCARE nedsatt INITIALC XX y_of_chf variable No or vänsterkam ARE PAR_SV::HDIA unknown, set to “Ja, marfunktio variable men okänd grad” if n PAR_SV::DIAGNOS any of the strings (SWEDEH PAR_SV::INDATUM 'I50', 'I110', 'I130' or EART+PA 'I132' is present in the R) PAR_SV::UTDATUM main or secondary PAR_OV::HDIA diagnosis, in- or outpatient care, PAR_OV::DIAGNOS ((STARTDATE- PAR_OV::INDATUM 5YR)UTD ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV). Set to No otherwise. Tidigare Ja/nej previous_stroke INITIALCARE::previ If INITIALCARE stroke (ej _xX ous_stroke variable No or TIA) PAR_SV::HDIA unknown, set to Yes (SWEDEH if there is an ICD- EART+PA PAR_SV::DIAGNOS code I60-I64, R) PAR_SV::INDATUM including all PAR_SV::UTDATUM subcategories, as main or secondary PAR_OV::HDIA diagnosis, in- or PAR_OV::DIAGNOS outpatient care, (STARTDATE- PAR_OV::INDATUM 5YR)UTD ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV). Set to No otherwise. Intagnings From PRESENTING INITIALCARE::PRES orsak INITIALC _SYMPTOMS ENTING_SYMPTO ARE MS variable Ambulans From AMBULANCE INITIALCARE::AMB INITIALC ULANCE ARE variable

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 50

SUPPLEMENTARY DATA Systoliskt Num SYSTOLIC_BL INITIALCARE::SYST blodtryck OOD_PRESSU OLIC_BLOOD_PRES RE SURE Hjärtfrekve Num HEART_RATE INITIALCARE::HEA ns RT_RATE Lungrassel Ja/nej PULMONARY INITIALCARE::PUL _RALES MONARY_RALES Cardiogen Ja/nej CARDIAC_SH INITIALCARE::CAR chock vid OCK DIAC_SHOCK ankomst Syremättna Num SYREMATTN DETO2X::SYREMAT d inklusion AD TNAD Syremättna Ja/nej SYREMATTN SYREMATTNAD SYREMATTNAD> d =>95% AD_HIGH =95 vid inklusion HLR före Ja/nej CPR_BEFORE INITIALCARE::CPR_ sjukhus _HOSPITAL BEFORE_HOSPITA L Prehospital Ja/nej THROMB_BE INITIALCARE::THR trombolys FORE_HOSPI OMB_BEFORE_HOS TAL PITAL

EKG STT From ECG_STT_CH INITIALCARE::ECG INITIALC ANGES _STT_CHANGES ARE variable EKG rytm From ECG_RHYTH INITIALCARE::ECG INITIALC M _RHYTHM ARE variable EKG QRS From ECG_QRS_AN INITIALCARE::ECG INITIALC NOTATION _QRS_ANNOTATIO ARE N variable Vänstergre From LEFT_BLOCK INITIALCARE::LEFT nblock INITIALC _BLOCK ARE variable

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 51 SUPPLEMENTARY DATA ASA Ja/nej ASPIRIN_REG INITIALCARE::ASPI RIN_REG Övriga From OTHER_ANTI INITIALCARE::OTH trombocyt INITIALC PLATELET_R ER_ANTIPLATELET hämmare ARE EG _REG variable Betablocke Ja/nej BETA_BLOCK INITIALCARE::BET rare ERS_REG A_BLOCKERS_REG Statiner Ja/nej STATINS_RE INITIALCARE::STAT G INS_REG Övriga From OTHER_LIPI INITIALCARE::OTH lipidsänkar INITIALC D_LOW_AGE ER_LIPID_LOW_AG e ARE NTS_REG ENTS_REG variable ACE- Ja/nej ACE_INHIBIT INITIALCARE::ACE hämmare ORS_REG _INHIBITORS_REG A2- Ja/nej ANGIOTENSI INITIALCARE::ANG blockerare N_II_BLOCK_ IOTENSIN_II_BLOC REG K_REG Ca- Ja/nej CALCIUM_AN INITIALCARE::CAL hämmare TAGONIST_R CIUM_ANTAGONIS EG T_REG Diuretika Ja/nej DIURETICS_R INITIALCARE::DIUR EG ETICS_REG Antikoagul From ORAL_ANTIC INITIALCARE::ORA antia INITIALC OAGULANTS L_ANTICOAGULAN ARE _REG TS_REG variable ACE- Ja/nej ACE_A2_REG INITIALCARE::ACE Yes if at least one yes, hämmare _X _INHIBITORS_REG, otherwise no if both eller A2- INITIALCARE::ANG no, otherwise missing blockerare, IOTENSIN_II_BLOC if at least one missing, härledd K_REG otherwise unknown Lipidsänka Ja/nej HYPERLIP_TR INITIALCARE::STAT Yes if at least one nde medel EAT_REG_XX INS_REG yes/non-no (0< (statin eller OTHER_LIPID_LO annat) INITIALCARE:: W_AGENTS_REG< OTHER_LIPID_LO 9), otherwise no if W_AGENTS_REG both no, otherwise REGANGIO::HYPER missing if at least one LIP missing, otherwise unknown.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 52

SUPPLEMENTARY DATA For STEMI-PCI patients, missing and unknown is filled in with REGANGIO::HYPE RLIP for index PCI, if non-missing. Kreatinin Num S_CREATININ INITIALCARELAB::S First non-missing (µmol/L) _XX _CREATININ, measurement in MEDICALCARECRE admission period, AT::S_CREATININ, ordered by SID CREATDATE MEDICALCARECRE CREATETIME ID AT:CREATDATE (Comment: Subevent MEDICALCARECRE ID, SID, is lower for AT:CREATETIME the first registered initial care or medical REGANGIO::SKREA care page, which TININ usually contains the first measurement; date and time are CRP Num CRP_X INITIALCARELAB:: ordered with missing CRP, before non-missing MEDICALCARECRP: since it is more :CRP common to record time for repeated MEDICALCARECRP: measurements than CRPDATE for the first MEDICALCARECRE measurement, and the AT:CRPTIME ID order usually is higher for later Hemoglobi Num HEMOGLOBI INITIALCARELAB:: measurements.) n (g/L) N_X HEMOGLOBIN, MEDICALCAREHB:: For Creatinine, if HEMOGLOBIN unknown or missing for STEMI-PCI MEDICALCAREHB: patients (DETO2X HBDATE or Non-DETO2X), MEDICALCAREHB: filled in with HBTIME REGANGIO variable. Hb

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 53

SUPPLEMENTARY DATA Women: HEMOGLOBIN_X <120

eGFR Num EGFR S_CREATININE_XX CKD-EPI CKD-EPI formula for (mL/min/ White or 1.73m2) Other:

if D_GENDER ='Kvinna' then

EGFR=141 * min(S_CREAT ININE_XX /61.9,1)**- 0.329 * max(S_CREAT ININE_XX /61.9,1)**- 1.209 * 0.993**D_AG E * 1.018; else if D_GENDER ='Man' then EGFR=141 * min(S_CREAT ININE_XX /79.6,1)**- 0.411 * max(S_CREAT ININE_XX /79.6,1)**- 1.209 * 0.993**D_AG E;

Njursvikt Ja/nej RENALFAILU EGFR EGFR<60 (eGFR<60 RE_X mL/min/1 .73m2) Infarkttyp As in INFARCTTYP DISCHARGE::INFAR registry E CTTYPE Subklassifi As in INFARCTCLA DISCHARGE::INFAR cering av registry SSIFICATION CTCLASSIFICATIO hjärtinfarkt N

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 54

SUPPLEMENTARY DATA Historisk Ja/nej COPD_HIST_S PAR_SV::HDIA J44, including all KOL- S subcategories, as main diagnos PAR_SV::DIAGNOS or secondary PAR_SV::INDATUM diagnosis, in- or outpatient care, PAR_SV::UTDATUM (STARTDATE- PAR_OV::HDIA 5YR)UTD PAR_OV::INDATUM ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV) Tidigare Ja/nej DIALYSIS_HIS PAR_SV::HDIA Z49, including all dialys T_SS subcategories, or PAR_SV::DIAGNOS Z99.2, as main or PAR_SV::INDATUM secondary diagnosis, in- or outpatient care, PAR_SV::UTDATUM (STARTDATE- PAR_OV::HDIA 5YR)UTD PAR_OV::INDATUM ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV) Tidigare Ja/nej CANCER_HIS PAR_SV::HDIA C14-C20 including cancer T_SS subcategories as main PAR_SV::DIAGNOS or secondary PAR_SV::INDATUM diagnosis, in- or outpatient care, PAR_SV::UTDATUM (STARTDATE- PAR_OV::HDIA 5YR)UTD PAR_OV::INDATUM ATUM (PAR_SV) or STARTDATE>IND ATUM (PAR_OV) Tidigare Ja/nej PVD_HIST_SS PAR_SV::HDIA I70-I73 including perifer subcategories as main kärlsjukdo PAR_SV::DIAGNOS or secondary m PAR_SV::INDATUM diagnosis, in- or PAR_SV::UTDATUM outpatient care, (STARTDATE- PAR_OV::HDIA 5YR)UTD PAR_OV::INDATUM ATUM (PAR_SV) or

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 55

SUPPLEMENTARY DATA STARTDATE>IND ATUM (PAR_OV)

Hospitalisation Label Formatted Name Sources Derivation values CABG 0=Nej, CABG_X MEDICALCARE::CA Summarised over 1=Ja, akut BG MCEID: Akut if any CABG, Akut, else Under 2=Ja, under vårdtillfället if any vårdtillfället, Under vårdtillfället, 3=Planerad else Planerad efter efter utskrivning if any utskrivning Planerad efter utskrivning, No if at least one No, missing if all records missing or unknown (9). CPAP Ja/nej CPAP_X MEDICALCARE::CP Yes if any medical AP care page in the admission period has Yes, else Unknown if at least one page with Unknown, else No if at least one page with No. Missing if no non-missing record. PCI under Ja/nej PCI_MC_X PCI::SUCCESS Yes if any vårdkedjan REGANGIO::REGT ANGIO/PCI YP registration in SCAAR under index MCEID, with REGTYP 1 (PCI) or 2 (PCIADHOC), and SUCCESS not in (.,2), i.e., general success not set to diagnostic only. No otherwise. Angiografi Ja/nej ANGIO_MC_ REGANGIO::FYND FYND not missing under X for a record matched vårdkedjan on MCEID. Vänsterka From LEFT_VENT MEDICALCARE:: Last non-missing (.) mmarfunkt MEDICAL RICULAR_FU LEFT_VENTRICUL and non-unknown (9) ion CARE:: NCTION_X AR_FUNCTION value in admission period, sorted by

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 56 SUPPLEMENTARY DATA LEFT_VE EVENT_DATE NTRICULA EVENT_TIME R_FUNCTI ORDNR ON Vänsterka From EF_OBTAIN MEDICALCARE::EF Value from same care mmarfunkt MEDICAL ED_BY_X _OBTAINED_BY page as ion mätt CARE:: Vänsterkammarfunkti med EF_OBTAI on NED_BY Reinfarkt Ja/nej MYOCARDIA MEDICALCARE::MY Yes if any medical under L_REINFARC OCARDIAL_REINF care page in the vårdtillfälle T_HOSPITAL ARCT_HOSPITAL admission period has t _X Yes, else Unknown if Cardiogen Ja/nej CARDIOGEN MEDICALCARE::CA at least one page with chock IC_SHOCK_X RDIOGENIC_SHOC Unknown, else No if K at least one page with AV-block Ja/nej AV_BLOCKA MEDICALCARE::AV No. Missing if no GE_X _BLOCKAGE non-missing record. Nytt Ja/nej NEW_FIBRIL MEDICALCARE::NE förmaksfli LATION_FLU W_FIBRILLATION_ mmer TTER_X FLUTTER Blödning Dödlig, BLEEDING_ MEDICALCARE::BL Summarised over all under vård Cerebral, TIMI_HOSPI EEDING_TIMI_HO medical care pages, Krävande TAL_X SPITAL using the priority op/transfusi Mortal, otherwise on, Okänt, cerebral, otherwise Nej requiring transfusion, otherwise Unknown, otherwise No. Missing if no non- missing record. HLR Ja/nej CARDIAC_AR MEDICALCARE:: RESUSCITATED_C under vård REST_inhospit RESUSCITATED_CA ARDIAC_ARREST al_X RDIAC_ARREST is grouped into Yes (1=VF/VT or 8=Other) or No, and summarised over all medical care pages as for other MEDICALCARE Yes/no variables. Mekanisk Fri MECHANICA MEDICALCARE::ME Summarised over all komplikati väggruptur/ L_COMPLICA CHANICAL_COMPL medical care pages, on under VSD/MI TION_X ICATION using the priority 1) vård (Akut fri väggruptur 2) VSD allvarlig MI) 3) akut allvarlig MI, Okänd, Nej. Avliden Ja/nej DEATH DISCHARGE::DEAT H

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 57

SUPPLEMENTARY DATA Antikoagul Ja/nej ANTICOAGU MEDICALCARE::AN Yes if any medical antia under LANTS_INH TICOAGULANTS_I care page in the vård OSP_X NHOSP admission period has iv Ja/nej BETA_BLOC MEDICALCARE::BE Yes, else Unknown if Betablocke KERS_TREA TA_BLOCKERS_TR at least one page with rare T_X EAT Unknown, else No if iv Ja/nej DIURETICS_ MEDICALCARE::DI at least one page with Diuretika TREAT_X URETICS_TREAT No. Missing if no iv Inotropa Ja/nej INOTROPES_ MEDICALCARE::IN non-missing record. X OTROPES iv Ja/nej NITRATES_T MEDICALCARE::NI Nitroglycer REAT_X TRATES_TREAT in

Tid på Num TIME_IN_HO STARTDATE DISCHARGE::DISC sjukhus SPITAL DISCHARGE::DISC HARGE_DATE_F – (dygn) HARGE_DATE STARTDATE (Subevent date for (discharge in same discharge) calendar day counted Clean file as zero days). Missing documentation and detected incorrect dates (identified as negative time in hospital) filled in.

Discharge medication Label Format Name Sources Derivatio ted n values ASA Ja/nej ASPIRIN_DISCHARGE DISCHARGE::ASP IRIN_DISCHARG E Övriga As in OTHER_ANTIPLATEL DISCHARGE::OT trombocythämmare registry ET_DISCHARGE HER_ANTIPLATE LET_DISCHARGE Betablockerare Ja/nej BETA_BLOCKERS_DIS DISCHARGE::BET CHARGE A_BLOCKERS_DI SCHARGE Statiner Ja/nej STATINS_DISCHARGE DISCHARGE::STA TINS_DISCHARG E Övriga lipidsänkare As in OTHER_LIPID_LOW_ DISCHARGE::OT registry AGENTS_DISCHARGE HER_LIPID_LOW _AGENTS_DISCH ARGE

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 58

SUPPLEMENTARY DATA ACE-hämmare eller Ja/nej ACE_A2_DISCHARGE DISCHARGE::ACE Yes if at A2-blockerare vid _X _INHIBITORS_DI least one utskrivning, härledd SCHARGE yes, no if DISCHARGE::AN both no, GIOTENSIN_II_B unknown LOCK_DISCHAR if both GE unknown , missing otherwise ACE-hämmare Ja/nej ACE_INHIBITORS_DIS DISCHARGE::ACE CHARGE _INHIBITORS_DI SCHARGE A2- blockerare Ja/nej ANGIOTENSIN_II_BL DISCHARGE::AN OCK_DISCHARGE GIOTENSIN_II_B LOCK_DISCHAR GE Ca-hämmare Ja/nej CALCIUM_ANTAGON DISCHARGE::CAL IST_DISCHARGE CIUM_ANTAGON IST_DISCHARGE Diuretika Ja/nej DIURETICS_DISCHAR DISCHARGE::DIU GE RETICS_DISCHAR GE Antikoagulantia As in ORAL_ANTICOAGUL DISCHARGE::OR registry ANTS_DISCHARGE AL_ANTICOAGU LANTS_DISCHAR GE

STEMI-PCI population additional background and procedural variables from SCAAR

Index PCI The index PCI is selected as the REGANGIO subevent with index admission period MCEID, REGTYP in (2,3) and not (SUCCESS in (.,2)) (i.e. PCI or PCI ad hoc registration, non-missing General success not set to Diagnostic only), with the lowest INTERDAT (date of intervention), and if several such events on the same INTERDAT, the lowest SID (subevent ID).

1.1.1. STEMI-PCI additional background and clinical presentation variables Label Formatt Name Sources Derivation ed values Indikation As in INDIKATIO REGANGIO::INDIK registry N ATION

Fynd As in FYND REGANGIO::FYND registry

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 59

SUPPLEMENTARY DATA Jourtid As in JOURTID REGANGIO::JOURT registry ID

Killip klass I/II/III/I KILLIPKLA REGANGIO::KILLIP V/Okän SS KLASS t

Stresskardiomyopa Ja/Nej/ STRESSKA REGANGIO::STRES ti Misstän RDIOMYOP SKARDIOMYOPATI kt ATI

Tid från Num SYMPTOM_ REGANGIO:: Set to missing if symtomdebut till TO_RAND SYMPTOM_ONSET <0 or >48h. randomisering _DATE Primarily derived (min) REGANGIO:: from SYMPTOM_ONSET REGANGIO. If _TIME REGANGIO data is incomplete or INITIALCARE:: outside window SYMPTOM_ONSET and _DATE INITIALCARE INITIALCARE:: data is complete SYMPTOM_ONSET and inside _TIME window, STARTDATE INITIALCARE data will be used. STARTTIME

Tid från Num SYMPTOM_ REGANGIO:: Set to missing if symtomdebut till TO_PCI SYMPTOM_ONSET <0 or >48h. punktion (min) _DATE Primarily derived REGANGIO:: from SYMPTOM_ONSET REGANGIO. If _TIME REGANGIO data is incomplete or

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 60

SUPPLEMENTARY DATA INITIALCARE:: outside window SYMPTOM_ONSET and _DATE INITIALCARE data is complete INITIALCARE:: and inside SYMPTOM_ONSET window, _TIME INITIALCARE REGANGIO::STICK data will be used. DAT

REGANGIO::STICK TID

Tid från Num ECG_TO_R REGANGIO:: Set to missing if reperfusionsgrunda AND PREHOSPITAL_EC <0 or >48h. nde EKG till G_DATE Primarily derived randomisering REGANGIO:: from (min) PREHOSPITAL_EC REGANGIO. If G_TIME REGANGIO data is incomplete or INITIALCARE:: outside window PREHOSPITAL_EC and G_DATE INITIALCARE INITIALCARE:: data is complete PREHOSPITAL_EC and inside G_TIME window,

STARTDATE INITIALCARE data will be used. STARTTIME

Tid från Num ECG_TO_PC REGANGIO:: Set to missing if reperfusionsgrunda I PREHOSPITAL_EC <0 or >48h. nde EKG till G_DATE Primarily derived punktion (min) from REGANGIO. If

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 61

SUPPLEMENTARY DATA REGANGIO:: REGANGIO data PREHOSPITAL_EC is incomplete or G_TIME outside window and INITIALCARE:: INITIALCARE PREHOSPITAL_EC data is complete G_DATE and inside INITIALCARE:: window, PREHOSPITAL_EC INITIALCARE G_TIME data will be used.

REGANGIO::STICK DAT

REGANGIO::STICK TID

PCI procedural data Label Formatt Name Sources Derivation ed values Komplett Ja/nej KOMPREV REGANGIO::KOMP revaskularisering REV

General success Ja/nej SUCCESS REGANGIO::SUCCE SS

Punktionställe A PUNKT REGANGIO::PUNKT femoral is/A brachial is/A axillaris /A radialis

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 62

SUPPLEMENTARY DATA höger/ A radialis vänster/ A radialis + A femoral is/A femoral is konvert erad från radialis/ A radialis konvert erad från A femoral is

Punktionställe, Radial/f PUNKT_X REGANGIO::PUNKT Radial=Brachialis sammanslaget emoral/ , Axillaris, Övriga radialis höger, radialis vänster. Femoral all other non-other alternative (femoral, radial+femoral, femoral converted from radial, radial

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 63

SUPPLEMENTARY DATA converted from femoral).

Artärförslutning As in VASCCLOS REGANGIO::VASCC registry UREDEV LOSUREDEV

Genomlysningstid Num GENOMLYS REGANGIO::GENO (mmm:ss) NINGSTID MLYSNINGSTID

Kontrastmedelsmä Num KONTRAST REGANGIO::KONT ngd (ml) MEDELMA RASTMEDELMANG NGD D

Stråldos (µGym2) Num STRALDOS REGANGIO::STRAL DOS

Adjuvant therapy

Någon adjuvant Ja/nej ADJUVANT REGANGIO::ADJUV terapi ANT

Aortaballongpump Ja/nej ADJPUMP REGANGIO::ADJPU Set to 0 (No) if MP ADJUVANT=0 (No) Annan Ja/nej ADJKAM REGANGIO::ADJKA vänsterkammarassi M st

Pacemaker Ja/nej ADJPACE REGANGIO::ADJPA CE

Distal protection Ja/nej ADJDISTAL ADJDISTAL device

Trombectomy Ja/nej ADJTROMB ADJTROMB

LUCAS Ja/nej ADJLUK REGANGIO::ADJLU K

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 64

SUPPLEMENTARY DATA Annat Ja/nej ADJANN REGANGIO::ADJAN N

Antithrombotic medication before PCI (<24h) and at PCI

Antitrombotisk Ja/nej ANTIFORE REGANGIO::ANTIF medicinering före ORE PCI (före PCI)

Antitrombotisk Ja/nej ANTIUND REGANGIO::ANTIU medicinering ND under/direkt i anslutning till PCI (under PCI)

ASA (före PCI) Ja/nej ASAFORE REGANGIO::ASAFO Set to No if RE ANTIFORE=No

ANTIFORE

ASA (under PCI) Ja/nej ASAUND REGANGIO::ASAU Set to No if ND ANTIUND=No

ANTIUND

ASA nyinsatt före Ja/nej ASAFORE_ ASAFORE Set to No if ASA PCI STAGE before admission ASPIRIN_REG

ASA nyinsatt Ja/nej ASAUND_S ASAUND Set to No if ASA under PCI TAGE before admission ASAFORE or before PCI ASPIRIN_REG

Clopidogrel/ticlopi Ja/nej CLOFORE REGANGIO::CLOFO Set to No if din (Plavix/Ticlid) RE ANTIFORE=No (före PCI) ANTIFORE

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 65

SUPPLEMENTARY DATA Clopidogrel/ticlopi Ja/nej CLOUND REGANGIO::CLOU Set to No if din (Plavix/Ticlid) ND ANTIUND=No (under PCI) ANTIUND

Clopidogrel/ticlopi Ja/nej CLOFORE_ CLOFORE No if clopidogrel din (Plavix/Ticlid) STAGE before admission INITIALCARE::OTH nyinsatt före PCI ER_ANTIPLATELET _REG

Clopidogrel/ticlopi Ja/nej CLOUND_S CLOUND CLOFORE No if clopidogrel din (Plavix/Ticlid) TAGE before admission INITIALCARE::OTH nyinsatt under PCI or PCI ER_ANTIPLATELET _REG

Heparin (före PCI) Ja/nej HEPFORE REGANGIO::HEPFO Set to No if RE ANTIFORE=No

ANTIFORE

Heparin (under Ja/nej HEPUND REGANGIO::HEPUN Set to No if PCI) D ANTIUND=No

ANTIUND

Dalteparin Ja/nej DALFORE REGANGIO::DALFO Set to No if (Fragmin) (före RE ANTIFORE=No PCI) ANTIFORE

Dalteparin Ja/nej DALUND REGANGIO::DALU Set to No if (Fragmin) (under ND ANTIUND=No PCI) ANTIUND

Enoxaparin Ja/nej ENOFORE REGANGIO::ENOFO Set to No if (Klexane) (före RE ANTIFORE=No PCI) ANTIFORE

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 66 SUPPLEMENTARY DATA Enoxaparin Ja/nej ENOUND REGANGIO::ENOU Set to No if (Klexane) (under ND ANTIUND=No PCI) ANTIUND

Annat Ja/nej ANNFORE REGANGIO::ANNF Set to No if lågmolekylärt ORE ANTIFORE=No heparin (före PCI) ANTIFORE

Annat Ja/nej ANNUND REGANGIO::ANNU Set to No if lågmolekylärt ND ANTIUND=No heparin (under ANTIUND PCI)

Fondaparinux Ja/nej FONFORE REGANGIO::FONFO Set to No if (Arixtra) (före PCI) RE ANTIFORE=No

ANTIFORE

Fondaparinux Ja/nej FONUND REGANGIO::FONU Set to No if (Arixtra) (under ND ANTIUND=No PCI) ANTIUND

Abciximab Ja/nej ABCFORE REGANGIO::ABCFO Set to No if (Reopro) (före RE ANTIFORE=No PCI) ANTIFORE

Abciximab Ja/nej ABCUND REGANGIO::ABCU Set to No if (Reopro) (under ND ANTIUND=No PCI) ANTIUND

Eptifibatid Ja/nej EPTFORE REGANGIO::EPTFO Set to No if (Integrilin) (före RE ANTIFORE=No PCI) ANTIFORE

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 67

SUPPLEMENTARY DATA Eptifibatid Ja/nej EPTUND REGANGIO::EPTUN Set to No if (Integrilin) (under D ANTIUND=No PCI) ANTIUND

Tirofiban Ja/nej TIRFORE REGANGIO::TIRFO Set to No if (Aggrastat) (före RE ANTIFORE=No PCI) ANTIFORE

Tirofiban Ja/nej TIRUND REGANGIO::TIRUN Set to No if (Aggrastat) (under D ANTIUND=No PCI) ANTIUND

Warfarin (Waran) Ja/nej WARFORE REGANGIO::WARF Set to No if (före PCI) ORE ANTIFORE=No

ANTIFORE

Warfarin (Waran) Ja/nej WARUND REGANGIO::WARU Set to No if (under PCI) ND ANTIUND=No

ANTIUND

Dabigatran Ja/nej DABFORE REGANGIO::DABF Set to No if (Pradaxa) (före ORE ANTIFORE=No PCI) ANTIFORE

Dabigatran Ja/nej DABUND REGANGIO::DABU Set to No if (Pradaxa) (under ND ANTIUND=No PCI) ANTIUND

Ticagrelor Ja/nej TICFORE REGANGIO::TICFO Set to No if (Brilique) (före RE ANTIFORE=No PCI) ANTIFORE

Ticagrelor Ja/nej TICUND REGANGIO::TICUN Set to No if (Brilique) (under D ANTIUND=No PCI)

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 68

SUPPLEMENTARY DATA ANTIUND

Ticagrelor Ja/nej TICFORE_S TICFORE No if brilique INITIALCARE::OTH (Brilique) nyinsatt TAGE before admission ER_ANTIPLATELET före PCI _REG

Ticagrelor Ja/nej TICUND_ST TICFORE No if brilique (Brilique) nyinsatt AGE before admission TICUND under PCI or PCI INITIALCARE::OTH ER_ANTIPLATELET _REG

Övriga (före PCI) Ja/nej OVRFORE REGANGIO::OVRF Set to No if ORE ANTIFORE=No

ANTIFORE

Övriga (under PCI) Ja/nej OVRUND REGANGIO::OVRU Set to No if ND ANTIUND=No

ANTIUND

Trombolys (före Ja/nej TROFORE REGANGIO::TROFO Set to No if PCI) RE ANTIFORE=No

ANTIFORE

Trombolys (under Ja/nej TROUND REGANGIO::TROU Set to No if PCI) ND ANTIUND=No

ANTIUND

Bivalirudin Ja/nej BIVFORE REGANGIO::BIVFO Set to No if (Angiox) (före RE ANTIFORE=No PCI) ANTIFORE

Bivalirudin Ja/nej BIVUND REGANGIO::BIVUN Set to No if (Angiox) (under D ANTIUND=No PCI) ANTIUND

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 69

SUPPLEMENTARY DATA Prasugrel (Efient) Ja/nej PRAFORE REGANGIO::PRAFO Set to No if (före PCI) RE ANTIFORE=No

ANTIFORE

Prasugrel (Efient) Ja/nej PRAUND REGANGIO::PRAU Set to No if (efter PCI) ND ANTIUND=No

ANTIUND

Prasugrel (Efient) Ja/nej PRAFORE_S PRAFORE No if prasugrel nyinsatt före PCI TAGE INITIALCARE::OTH before admission ER_ANTIPLATELET _REG

Prasugrel (Efient) Ja/nej PRAUND_S PRAUND No if prasugrel nyinsatt efter PCI TAGE before admission PRAFORE or PCI INITIALCARE::OTH ER_ANTIPLATELET _REG

Rivaroxaban Ja/nej RIVFORE REGANGIO::RIVFO Set to No if (Xarelto) (före RE ANTIFORE=No PCI) ANTIFORE

Rivaroxaban Ja/nej RIVUND REGANGIO::RIVUN Set to No if (Xarelto) (under D ANTIUND=No PCI) ANTIUND

Apixaban (Eliquis) Ja/nej APIXABAN REGANGIO::APIXA Set to No if (före PCI) FORE BANFORE ANTIFORE=No.

ANTIFORE NB! Added to SCAAR 2013-11- 28, missing for earlier patients.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 70

SUPPLEMENTARY DATA Index PCI procedural variables summarised from lesion level data

Index segment data are selected using SID.

Label Formatt Name Sources Derivation ed values

Procedur med Ja/nej EFTERDILAT SEGMENT::EFTER Yes if Yes for any efterdilatation ATION DILATATION lesion level subprocedure, including diagnostic, No otherwise

Maxdiameter stent Num DIAM SEGMENT::DIAM Maximum over i procedur (mm) all stents in the PCI

Total längd stent i Num STENTLANG SEGMENT::STENT Sum over all procedur (mm) D LANGD stents in the PCI.

Kombination String, TREATED_V SEGMENT::SEGM Vessel derived on behandlade kärl i ESSELS2_X ENT SEGMENT level [RCA/] procedur (TREATED_V as: [LM/][ SEGMENT::GRAF ESSEL_X) LAD/][ T if segment in LCx/][ (1,2,3,4,18,19) Graft] then TREATED_VES SEL_X=1; [RCA]

else if segment=5 then TREATED_VES SEL_X=2; [LM]

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 71

SUPPLEMENTARY DATA else if segment in (6,7,8,9,10,20) then TREATED_VES SEL_X=3; [LAD]

else if segment in (11,12,13,14,15,1 6,17) then TREATED_VES SEL_X=4; [LCx]

if graft=1 then TREATED_VES SEL_X=5; [Graft]

if graft=2 then TREATED_VES SEL_X=5; [Graft]

Summarised over all lesion level subprocedures in the PCI, including diagnostic.

NB. TREATED_VES SELS2_X is a more readable recoding of

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 72

SUPPLEMENTARY DATA TREATED_VES SEL_X.

Antal behandlade 0-5 NUMBER_TR TREATED_VESSE Number of kärl i procedur EATED_VES L_X vessels according SELS_X to TREATED_VES SEL_X (TREATED_VES SELS2_X)

RCA only Ja/nej RCA2 TREATED_VESSE Yes/No variables L_X based on LAD only LAD2 TREATED_VES LCX only LCX2 SEL_X

LM only LM2 (TREATED_VES SELS2_X). Graft only GRAFT2 Exclusive per patient, No for all patients with subprocedures in several vessels.

RCA (and possibly Ja/nej RCA3 TREATED_VESSE Yes/No variables other) L_X based on TREATED_VES LAD (and possibly LAD3 SEL_X other) (TREATED_VES LCX (and possibly LCX3 SELS2_X). Non- other) exclusive per

LM (and possibly LM3 patient, each other) patient may have yes for several Graft (and possibly GRAFT3 vessels. other)

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 73

SUPPLEMENTARY DATA RCA, vs. left Ja/nej RIGHTLEFT RCA2 if rca3=0 and (lad3= or vessel (excluding 1 RCA3 lcx3=1 or pat. with both) lm3=1) then LAD3 rightleft=0;

LCX3 else if rca2=1 then LM3 rightleft=1;

Behandlat kärl String TREATED_V TREATED_VESSE Set to missing if (missing om flera RCA/L ESSEL_ONE LS2_X NUMBER_TRE kärl) M/LAD ONLY ATED_VESSEL NUMBER_TREAT /LCx/G S_X>1. ED_VESSELS_X raft

Procedur med Ja/nej BALLONGVI SEGMENT::PROCT Any lesion level ballongvidgning DGNING YP subprocedure in the PCI of this Procedur med DIREKTSTEN type. No such direktstent T procedure is Procedur med BALLONG_S coded as no. ballong+stent TENT

Procedur med LASERLEDA laserledare RE

Procedur med ATHERECTO atherectomi MI

Procedur med ROTABLATO rotablator R

Procedur med CUTTING_B cutting ballon ALLON

Procedur med BRACHYTER brachyterapi API

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 74 SUPPLEMENTARY DATA Procedur med LEDARFORS ledarförsök OK

Procedur med LAKEMEDEL läkemedelsballong SBALLONG

Procedur med LAKEMEDEL läkemedelsballong SBALLONG_ +stent STENT

Procedur med ANNAN_TER annan terapi API

Procedur med DIAGNOSTIK diagnostik

Proximal lesion Ja/nej PROXIMAL_ SEGMENT::SEGM Derived on X ENT SEGMENT level as

if segment=. then proximal_x=.;

else if segment in (1,2,5,6,7,11) then proximal_x=1;

else proximal_x=0;

Summarised to PATIENT level over all lesion level subprocedures, including diagnostic; yes if any subprocedure

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 75

SUPPLEMENTARY DATA in proximal segment, no otherwise.

Stenostyp DeNovo Ja/nej DENOVO SEGMENT::STENO Any lesion level något segment STYP subprocedure in the PCI, where Stenostyp Övriga RESTENOTH stenosis is of this restenoser något ER type. None such segment coded as no. Stenostyp In-stent RESTENINST restenosis något segment

Alla ACC/AHA String STENOSSUM SEGMENT::STENO Summarised over stenosklasser i [A/][B1 (STENOSKLA SKLASS all lesion level procedur /][B2/][ SS) subprocedures in C/][B1 the PCI, including bif/][B2 diagnostic. bif/][Cb STENOSSUM is if][Othe the formatted r] version of STENOSKLASS.

A only Ja/nej SA_2 STENOSKLASS Yes/No variables based on B1 only SB1_2 STENOSKLASS B2 only SB2_2 (STENOSSUM).

C only SC_2 Exclusive per patient, No for all B1 Bif only SB1B_2 patients with B2 Bif only SB2B_2 stenoses of several classes. C Bif only SCB_2

Other only SO_2

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 76 SUPPLEMENTARY DATA A (and possibly Ja/nej SA_3 STENOSKLASS Yes/No variables other) based on STENOSKLASS B1A (and possibly SB1_3 (STENOSSUM). other) Non-exclusive per B2A (and possibly SB2_3 patient, each other) patient may have

C (and possibly SC_3 yes for several other) categories.

B1 Bif (and SB1B_3 possibly other)

B2 Bif (and SB2B_3 possibly other)

C Bif (and possibly SCB_3 other)

Other (and possibly SO_3 other)

Antal ACC/AHA Num STENOSNUM STENOSKLASS Number of stenosklasser i stenosis classes procedur according to STENOSKLASS (STENOSSUM).

Bifurkation (något Ja/nej BIFURK STENOSKLASS Yes if any segment) stenosis of class B1bif, B2bif or Cbif, No otherwise

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 77

SUPPLEMENTARY DATA 1.1.1. Complications to index PCI Label Formatt Name Sources Derivation ed values Complications to PCI in PCI lab

Komplikation Lab Ja/nej LABKOMP REGANGIO::LABK OMP

Allergisk reaktion Ja/nej LABALLERG REGANGIO::LABA Set to 0 (No) if lätt/måttlig Lab ILATT LLERGILATT LABCOMP=0 (No) Allergisk reaktion Ja/nej LABALLERG REGANGIO::LABA allvarlig Lab IALLV LLERGIALLV

Behandlingskrävan Ja/nej LABBEHARY REGANGIO::LABB de arytmi Lab TMI EHARYTMI

Hemodynamisk Ja/nej LABHEMO REGANGIO::LABH komplikation Lab EMO

Neurologisk Ja/nej LABNEURO REGANGIO::LABN komplikation Lab EURO

Vaskulär (endast Ja/nej LABVASK REGANGIO::LABV icke coronara kärl) ASK Lab

Tappat stent Lab Ja/nej LABTAPPAT REGANGIO::LABT APPAT

Bestående Ja/nej LABBESTSID REGANGIO::LABB sidogrensocklusion O ESTSIDO Lab

Perforation Lab Ja/nej LABPERF REGANGIO::LABP ERF

Tamponad Lab Ja/nej LABTAMP REGANGIO::LABT AMP

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 78

SUPPLEMENTARY DATA Akut CABG från Ja/nej LABAKUTCA REGANGIO::LABA lab Lab BG KUTCABG

Annan allvarlig Ja/nej LABANNAN REGANGIO::LABA komplikation Lab ALLV NNANALLV

Avliden Lab Ja/nej LABDODSFA REGANGIO::LABD LL ODSFALL

Procedurrelaterad Ja/nej LABPROCED REGANGIO::LABP Set to 0 (No) if död Lab URDOD ROCEDURDOD LABCOMP=0 (No)

Or LABPROCEDU RDOD=0

Complications to PCI in ward

Komplikation Ja/nej AVDKOMP ANGIOPCICOMP:: (Avd) AVDKOMP

Allergisk Ja/nej AVDALLERG ANGIOPCICOMP:: Set to 0 if senkomplikation ISK AVDALLERGISK AVDKOMP=0 (Avd)

Någon form av Ja/nej AVDBLODNI ANGIOPCICOMP:: blödning (Avd) NG AVDBLODNING

Vaskulär (endast Ja/nej AVDANNAN ANGIOPCICOMP:: icke coronara kärl) VASK AVDANNANVASK (Avd)

Neurologisk Ja/nej AVDNEURO ANGIOPCICOMP:: komplikation AVDNEURO (Avd)

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 79

SUPPLEMENTARY DATA Nytillkommen Ja/nej AVDNJURIN ANGIOPCICOMP:: njurinsufficiens SUFF AVDNJURINSUFF Avd

Tamponad (Avd) Ja/nej AVDTAMP ANGIOPCICOMP:: AVDTAMP

Komplikation - Re- Ja/nej AVDREPCI ANGIOPCICOMP:: PCI (behandlat AVDREPCI segment) (Avd)

CABG (p g a Ja/nej AVDCABG ANGIOPCICOMP:: komplikation, ej AVDCABG från lab) (Avd)

Hjärtinfarkt (Avd) Ja/nej AVDHJARTI ANGIOPCICOMP:: NFARKT AVDHJARTINFAR KT

Annan allvarlig Ja/nej AVDANNAN ANGIOPCICOMP:: komplikation ALLV AVDANNANALLV (Avd)

Avliden (Avd) Ja/nej AVDDODSFA ANGIOPCICOMP:: LL AVDDODSFALL

Blödning Major Ja/nej AVDBLODM ANGIOPCICOMP:: Set to 0 if (Avd) AJOR AVDBLODMAJOR AVDBLODNIN G=0 AVDBLODNING

Blödning Minor Ja/nej AVDBLODMI ANGIOPCICOMP:: (Avd) NOR AVDBLODMINOR

AVDBLODNING

Blödning mindre Ja/nej AVDBLODMI ANGIOPCICOMP:: (Avd) NI AVDBLODMINI

AVDBLODNING

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 80

SUPPLEMENTARY DATA Behandlingskrävan Ja/nej AVDBEHPSE ANGIOPCICOMP:: de pseudoaneurysm UDO AVDBEHPSEUDO (Avd) AVDBLODNING

Hematom > 5 cm Ja/nej AVDHEMAT ANGIOPCICOMP:: diameter (Avd) OM AVDHEMATOM

AVDBLODNING

Hb fall> 20 g/L Ja/nej AVDHBFALL ANGIOPCICOMP:: (Avd) AVDHBFALL

AVDBLODNING

Förlängd Ja/nej AVDFORLAN ANGIOPCICOMP:: kompressionstid > GDKOMPTID AVDFORLANGDK 6 tim (Avd) OMPTID

AVDBLODNING

Förlängd vårdtid > Ja/nej AVDVARDTI ANGIOPCICOMP:: 1 dygn (Avd) D AVDVARDTID

AVDBLODNING

Gjordes Ja/nej AVDULTRAL ANGIOPCICOMP:: ultraljudsundersök JUD AVDULTRALJUD ning / CT (Avd) AVDBLODNING

Blodtransfusion Ja/nej AVDBLODTR ANGIOPCICOMP:: (Avd) ANSFUSION AVDBLODTRANS FUSION

AVDBLODNING

Kirurgisk åtgärd Ja/nej AVDKIRURG ANGIOPCICOMP:: (Avd) ISKATGARD AVDKIRURGISKA TGARD

AVDBLODNING

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 81

SUPPLEMENTARY DATA Annan behandling Ja/nej AVDANNAN ANGIOPCICOMP:: utöver kompression BEHUTOVER AVDANNANBEHU (Avd) KOMP TOVERKOMP

AVDBLODNING

Förtida utsättning / Ja/nej AVDFORTID ANGIOPCICOMP:: uppehåll av AUTSATTNI AVDFORTIDAUTS antitrombotisk beh. NG ATTNING (Avd) AVDBLODNING

Procedurrelaterad Ja/nej AVDPROCED ANGIOPCICOMP:: Set to 0 if död (Avd) URDOD AVDPROCEDURD AVDDODSFAL OD L=0

AVDDODSFALL

Efficacy outcomes

Follow-up time per patient The final PAR_SV and DORS data will contain events up to 2016-12-31, while the final SWEDEHEART data, obtained mid-february 2017, may contain events from early 2017, but this data may be incomplete due to lag in reporting. For consistency, all outcome variables will be considered censored on 2016-12-30, one year after the last patient entered the trial, when reporting events during the entire follow-up time. Note that all primary analyses are based on censoring on day 365, and this affects only sensitivity analyses using "all follow-up time per patient".

Outcome events in all populations Label For Name Sources Derivation matt ed value s Tid till Num SURVTIME PATIENT::BEFDO BEFDODDTM-STARTDATE. död eller DDTM censureri If not DEAD, censored on 2016- ng Clean file 12-30. (If decided before clean file (dygn) documentation that patients are censored for death before this date, those censoring dates will documented and hardcoded).

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 82

SUPPLEMENTARY DATA If PATIENT::DEAD is missing (concerns patients without SWEDEHEART record or without Swedish personal number), death date is followed up outside the registry and filled in. Död Ja/n DEAD PATIENT::DEAD If PATIENT::DEAD is missing under ej (concerns patients without uppföljni Clean file SWEDEHEART record or without ngstiden documentation Swedish personal number), death is followed up outside the registry and filled in. Återinläg Ja/n MI_EVENT INITIALCARE::A Difference from STARTDATE gning ej DMISSION_DATE (day 0) to first med , ADMISSION_DATE for a post- hjärtinfa index admission period rkt (INITIALCARE::ADMISSION_D DISCHARGE_DA ATE>DISCHARGE_DATE_F) Tid till MI_TIME TE_F with discharge diagnosis fields återinläg containing string 'I21' or 'I22'. gning DIAGNOSE::COD Censored at SURVTIME (i.e. either med E time of death or censoring date). hjärtinfa rkt eller censureri ng (dygn) Återinläg Ja/N HF_EVENT PAR_SV::HDIA Difference from STARTDATE gning ej (day 0) to INDATUM; med PAR_SV::INDATU DISCHARGE_DATE_F hjärtsvik M

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 83

SUPPLEMENTARY DATA hjärtinfa MI_TIME DEAD_MI_TIME=min(SURVTI rkt ME,MI_TIME) Tid till Num DEAD_MI_ död eller TIME återinläg gning med hjärtinfa rkt (dygn) Död Ja/n DEAD_HF_ DEAD DEAD_HF_EVENT= eller ej EVENT max(DEAD,MI_EVENT) återinläg SURVTIME DEAD_HF_TIME=min(SURVTI gning HF_EVENT med ME,HF_TIME) hjärtsvik HF_TIME t Tid till Num DEAD_HF_ död eller TIME återinläg gning med hjärtsvik t (dygn) Död Ja/n DEAD_MI_ DEAD_HF_EVEN DEAD_MI_HF_EVENT=max(D eller ej HF_EVENT T EAD_HF_EVENT,MI_EVENT) återinläg gning DEAD_HF_TIME DEAD_MI_HF_TIME=min(DEA D_HF_TIME,MI_TIME) med MI_EVENT hjärtinfa rkt eller MI_TIME hjärtsvik t Tid till Num DEAD_MI_ död eller HF_TIME återinläg gning med hjärtinfa rkt eller hjärtsvik t (dygn) CV-död Ja/n CVDEATH_ DORS:ULORSAK CVDEATH_EVENT=(DEAD=1 ej EVENT and (SURVTIME<=2016-12-30)

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 84

SUPPLEMENTARY DATA Tid till Num CVDEATH_ DEAD and (index(ULORSAK,'I')>0 OR CV-död TIME missing(ULORSAK))) (dygn) SURVTIME CVDEATH_TIME=SURVTIME

Outcome events in the PCI-STEMI population only PCI segment data is considered post-index PCI if interdat>interdat_t or (interdat=interdat_t and sid>sid_t), where SID_T and INTERDAT_T are the REGANGIO SID and INTERDAT for the index PCI.

Label For Name Sources Derivation matt ed value s Target Ja/n TSR_EVEN REGANGIO ::INT Intervention with PROCTYP^=9, section ej T ERDAT in segment number (or graft) with revascul SEGMENT ::PROC intervention with PROCTYP^=9 at arisation TYP index PCI. Time counted from Tid till Num TSR_TIME SEGMENT ::SEG STARTDATE to INTERDAT for target MENT the first new intervention. Censored section SEGMENT ::GRAF at death or censoring date for death. revascul T arisation STARTDATE (dygn) Target Ja/n TVR_EVEN REGANGIO ::INT Intervention with PROCTYP^=9, vessel ej T ERDAT in a vessel (or graft) with revascul SEGMENT ::PROC intervention with PROCTYP^=9 at arisation TYP index PCI. Vessel is defined as for Tid till Num TVR_TIME SEGMENT ::SEG TREATED_VESSELS2_X above. target MENT Time counted from STARTDATE vessel SEGMENT ::GRAF to INTERDAT for the first new revascul T intervention. Censored at death or arisation STARTDATE censoring date for death. (dygn) Resteno Ja/n RESTENOSI RESTENOS::SEGI Restenosis date and supplemental s ej S_EVENT D stent thrombosis flag for restenosis Tid till Num RESTENOSI RESTENOS::SID set as below are added to the restenos S_TIME REGANGIO::INT SEGMENT level data set by (dygn) ERDAT matching on SEGID: Restenosis Stenttro Ja/n ST_EVENT REGANGIO::INT date as INTERDAT of intervention mbos ej ERDAT when restenosis was detected Tid till Num ST_TIME SEGMENT::SEGI obtained by matching on SID for stenttro D new intervention; Stent thrombosis mbos SEGMENT::SID flag (also) is set for the restenosis if (dygn) RESTENOS::OCK RESTENOS::OCKLUSION in LUSION (2,4) [2=Yes, with acute STARTDATE presentation, 4=No, but suspected thrombus].

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 85

SUPPLEMENTARY DATA Index PCI segments are selected and STARTDATE added by merging with supplemented index PCI ANGIOPCI data, and data is summarised for the index PCI by selecting the first non- missing restenosis date and first stent thrombosis flagged restenosis date for all segments treated at index PCI. No date is considered no event. Time is counted from STARTDATE. Censored at time of death or 2016-12-30. Återinlä Ja/n SHOCK_EV INITIALCARE::A Difference from STARTDATE ggning ej ENT DMISSION_DATE (day 0) to first med DISCHARGE_DA ADMISSION_DATE for a post- kardioge TE_F index admission period n chock STARTDATE (INITIALCARE::ADMISSION_D Tid till Num SHOCK_TI INITIALCARE::CA ATE>DISCHARGE_DATE_F) återinläg ME RDIAC_SHOCK with CARDIAC_CHOCK at gning REGANGIO::KILL admission or an angio/PCI record med IPKLASS in the admission period with Killip kardioge class III-IV. Censored at n chock SURVTIME (i.e. either time of death or censoring date). Död, Ja/n STEMI_MA Any of the events and time to first återinläg ej CE1_EVEN event. Censored at 2016-12-30. gning T med hjärtinfa rkt eller kardioge n chock, eller stenttro mbos (STEMI -PCI MACE 1) Tid till Num STEMI_MA död, CE1_TIME återinläg gning med hjärtinfa rkt eller kardioge n chock,

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 86 SUPPLEMENTARY DATA eller stenttro mbos (STEMI -PCI MACE 1) Död, Ja/n STEMI_MA Any of the events and time to first återinläg ej CE2_EVEN event. Censored at 2016-12-30. gning T med hjärtinfa rkt eller kardioge n chock, stenttro mbos eller TVR (STEMI -PCI MACE 2) Tid till Num STEMI_MA död, CE2_TIME återinläg gning med hjärtinfa rkt eller kardioge n chock, stenttro mbos eller TVR (STEMI -PCI MACE 2)

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 87

SUPPLEMENTARY DATA

SEPHIA outcome variables Label Form Name Sources Derivation atted values Andnöd 1=I, SHORTNESS_O FOLLOWUP::ORDNR FOLLOWUP::ORD 6-10 F_BREATH1 NR=1 for 6-10 weeks veckor 2=II, FOLLOWUP:: 3=III, SHORTNESS_OF_BRE 4=IV, ATH 5=0 FOLLOWUP::MOBILIT (NB!) Y Rörlighet 1,2,3 MOBILITY1 FOLLOWUP::HYGIEN 6-10 E veckor FOLLOWUP::ACTIVITI Hygien 6- 1,2,3 HYGIENE1 ES 10 veckor FOLLOWUP::PAIN_TR Aktiviteter 1,2,3 ACTIVITIES1 OUBLES 6-10 FOLLOWUP::DISCOMF veckor ORT_SADNESS Smärtor/b 1,2,3 PAIN_TROUBL esvär 6-10 ES1 veckor Oro/nedst 1,2,3 DISCOMFORT_ ämdhet 6- SADNESS1 10 veckor EQ-5D Num EQ5D_MEAN1 MOBILITY1 Mean of non-missing medelpoä domain scores ng 6-10 HYGIENE1 veckor ACTIVITIES1 PAIN_TROUBLES1 DISCOMFORT_SADNE SS1 Nuvarand 0-100 PRESENT_HEA FOLLOWUP::PRESENT e LTH_STAT1 _HEALTH_STAT hälsotillstå nd 6-10 veckor

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 88 SUPPLEMENTARY DATA Uppföljni Num FOLLOWUP_D FOLLOWUP::ORDNR FOLLOWUP::FOLL ngsdag AY1 OWUP_DATE- från FOLLOWUP::FOLLOW STARTDATE, inklusion, UP_DATE ORDNR=1 6-10 STARTDATE veckor SEPHIA- Ja/nej SEPHIA_FLAG1 SHORTNESS_OF_BRE At least one non- data 6-10 ATH1 missing veckor MOBILITY1 HYGIENE1 ACTIVITIES1 PAIN_TROUBLES1 DISCOMFORT_SADNE SS1 PRESENT_HEALTH_S TAT1 Andnöd 1=I, SHORTNESS_O FOLLOWUP::ORDNR FOLLOWUP::ORD 12-14 F_BREATH2 NR=2 for 12-14 månader 2=II, FOLLOWUP:: months 3=III, SHORTNESS_OF_BRE 4=IV, ATH 5=0 FOLLOWUP::MOBILIT (NB!) Y Rörlighet 1,2,3 MOBILITY2 FOLLOWUP::HYGIEN 12-14 E månader FOLLOWUP::ACTIVITI Hygien 1,2,3 HYGIENE2 ES 12-14 FOLLOWUP::PAIN_TR månader OUBLES Aktiviteter 1,2,3 ACTIVITIES2 FOLLOWUP::DISCOMF 12-14 ORT_SADNESS månader

Smärtor/b 1,2,3 PAIN_TROUBL esvär 12- ES2 14 månader Oro/nedst 1,2,3 DISCOMFORT_ ämdhet SADNESS2 12-14 månader

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 89

SUPPLEMENTARY DATA EQ-5D Num EQ5D_MEAN2 MOBILITY2 Mean of non-missing medelpoä domain scores ng 12-14 HYGIENE2 månader ACTIVITIES2 PAIN_TROUBLES2 DISCOMFORT_SADNE SS2 Nuvarand 0-100 PRESENT_HEA FOLLOWUP::ORDNR ORDNR=2 e LTH_STAT2 hälsotillstå FOLLOWUP::PRESENT nd 12-14 _HEALTH_STAT månader Uppföljni Num FOLLOWUP_D FOLLOWUP::ORDNR FOLLOWUP::FOLL ngsdag AY2 OWUP_DATE- från FOLLOWUP::FOLLOW STARTDATE, inklusion, UP_DATE ORDNR=2 12-14 STARTDATE månader SEPHIA- Ja/nej SEPHIA_FLAG2 SHORTNESS_OF_BRE At least one non- data 12-14 ATH2 missing månader MOBILITY2 HYGIENE2 ACTIVITIES2 PAIN_TROUBLES2 DISCOMFORT_SADNE SS2 PRESENT_HEALTH_S TAT2

Clean file documentation data included in the analysis data base

Unintentional randomisations not included in ITT Read to analysis data base from file “Excluded randomisation numbers CLEAN FILE.xlsx” and removed. The removed DETO2X::, INITIALCARE:: and DISCHARGE:: data is saved in data set Excluded_randnumbers for reference.

Corrections to DETO2X module data Read to analysis data base from file “All changes to DETO2X module data CLEAN FILE.xlsx”. Empty fields in the excel sheet are not used for filling in data.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 90

SUPPLEMENTARY DATA Corrected data fields are randomisation date and time (ARRIVALDATE, ARRIVALTIME), treatment completion and reason for noncompletion (TREATMENTCOMPLETED, NOTCOMPLETEDREASON), treatment end date and time, or noncompletion date and time (TREATEND_DATE, TREATEND_TIME); note that complemented and corrected times are given as treatment completion times regardless or whether treatment was completed, since the end time variables are merged anyhow, and treatment end saturation (SATURATION). The correction variables from the file are in the data base with names appended _F0. Variables after correction have names appended _F, except for randomisation date and time, which are corrected in the original variables, and the registry data variables saved as ARRIVALDATE_ORIGINAL and ARRIVALTIME_ORIGINAL.

Re-matching of DETO2X modules and/or SCAAR PCI records to RIKS-HIA admission periods Read from file “All matchings CLEAN FILE.xlsx”. Re-matching consist of changing the admission period key, MCEID, and possibly the patient key, PID, when the DETO2X module was originally associated with the wrong or a non-existing personal number. Original DETO2X module keys are saved as PID_ORIGINAL and MCEID_ORIGINAL. For rare cases with duplicated index admission periods, PCI registrations belonging to one admission period may be re-matched with the other, more complete, admission period in the same way. The registry SCAAR MCEID is saved as MCEID_PCI_ORIGINAL.

Corrections to DISCHARGE data Read to analysis data base from file “All complementing DISCHARGE data CLEAN FILE.xlsx”. Empty fields in the excel sheet are not used for filling in data. Fields are filled in after re-matching of admission periods. Corrected data fields are hospital discharge date (DISCHARGE_DATE), and the first two discharge diagnosis codes (may be extended to more codes if necessary) (DIAG1, DIAG2). The correction variables from the excel sheet are in the data base with names appended _NEW. The corrected discharge date is DISCHARGE_DATE_F, while diagnosis codes are corrected in the original registry variables, with the registry values saved as DIAG1_ORIGINAL and DIAG2_ORIGINAL.

Mortality, age and sex for patients not in SWEDEHEART Randomised patients without a SWEDEHEART record or without a Swedish personal number will be followed up for mortality outside the registry in January 2017. The variables age (PATIENT::D_AGE) and sex (PATIENT::D_GENDER) are derived from the personal number, and will be filled in for these patients. The variables DEAD_F, BEFDODDTM_F, D_GENDER_F and D_DOB_F will be read to the analysis data base from file “Complementing data for patients not in SWEDEHEART CLEAN FILE.xlsx” and used to fill in all missing values in the original variables.

Determination of sample size Assuming a 1-year mortality of 14% in the Room air group, for a least interesting contrast of 20% relative risk reduction at 1 year, 5900 patients will give 90% power to detect a difference using an unadjusted chi-square test. To leave room for treatment non-compliance, 6600 patients were planned to be randomised. The pre-planned analysis will use time to event, and adjust for age and sex, which was assumed to further increase power.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 91

SUPPLEMENTARY DATA Blinded interim checks of the assumptions in the sample size calculation were performed in March 2014 based on accumulated data up to 2014-03-03, and in April 2015 based on accumulated data up to 2014-12-12. The first analysis predicted a 1-year mortality of about 5% (both trial arms combined), based on observed 30-day mortality in the trial and the historical relationship between 30-day and 1-year mortality in SWEDEHEART registry data, giving an expected power of about 50%. However, it was not considered feasible to increase the sample size of the trial, since 80% power would require doubling the sample size. It was decided in March 2014 to include the composite of death and heart failure as an outcome variable in the ITT population since registry data indicated that this composite would have approximately twice the incidence rate of mortality alone. It was briefly discussed to change primary variable, but finally decided to keep mortality as primary, with the understanding that the power is low so that the new main secondary composite variable might be more relevant for interpretation of the results. The second blinded power check in 2015 took place after a number of patients had reached 1 year of follow-up, and confirmed a 1-year mortality of about 5%. Furthermore, a simulation based on observed mortality in age groups showed that the pre-defined age adjustment unfortunately did not seem to increase power considerably.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 92

SUPPLEMENTARY DATA Interim Analysis Plan No interim treatment comparisons were performed in the trial. Trial data blinded as to treatment arm were available to the study team for monitoring of conduct, quality and power assumptions throughout the trial. See also Section 0. Changes in the Planned Analysis There are no changes in the planned analyses, but secondary outcome variables were discussed and clarified during the trial, leading to additions and clarifications. Main additions are as below:  Cardiovascular death, all components of the pre-defined composite, and the composites of death and MI and death and heart failure are introduced as secondary outcomes. The definition of cardiovascular death follows the TASTE trial. Due to the unexpected low mortality in the trial population the power for mortality is low and the composite of mortality and heart failure is considered the most important secondary variable for the ITT population.  New myocardial infarctions are now considered of main interest for the DETO2X- AMI subpopulation only. However, since the composite of death, new hospitalisation with MI, or new hospitalisation with heart failure was a pre-defined secondary variable for the ITT population in the protocol, it will also be presented for the ITT population.  NYHA heart failure and EQ-5D, assessed at standard follow-up visits, are added as secondary variables.  The Per-Protocol population and sensitivity analyses are defined in this SAP.  The STEMI-PCI population is defined in this SAP, as well as the outcome events specific for this population; TSR, TVR, restenosis, stent thrombosis, and rehospitalisation with cardiogenic chock. The outcomes follow the TASTE trial, except for cardiogenic chock, which was a post-hoc addition in TASTE which is defined somewhat differently in this trial.  Exact definitions for MI and heart failure are provided. The definition of MI and cardiovascular death follows the TASTE trial. The TASTE trial used only ICD-10 code I50 to identify heart failure, adding the categories I11.0, I13.0 and I13.2 was decided by e-mail 2016-05-02. The trial protocol pre-specified that MI should be obtained from SWEDEHEART as in the TASTE trial, but did not specify the data source for heart failure. The data source for heart failure is specified in this SAP as the Swedish inpatient registry, which was the original intention and follows the TASTE trial. The rationale not to include inpatient registry data in the MI definition was that MI reporting in SWEDEHEART is known to be very complete and allows to distinguish the index MI hospitalization from a new hospitalization due to reinfarction within the first 30 days which is not possible in the inpatient registry. For hospitalisation with heart failure, the inpatient registry is the only complete data base available.  Data quality decisions on which checks and queries were considered necessary were added during SAP development. See Section 0 for information on which corrections are added to DETO2X and registry data exported from SWEDEHEART.  The original protocol did not contain any information on variables in the trial data base except outcomes. All other variables are specified in this SAP.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 93 SUPPLEMENTARY DATA

Description of Output Tables and figures are presented in separate output specification documents. All tables will be in Swedish since SWEDEHEART is in Swedish. All figures will use American English unless otherwise specified to be suitable for use in article submissions. All figures will be provided in vector based PDF format. All Kaplan-Meier curves will be provided both with y-axis from 0-100% and with y-axis adapted to the data. The file name convention is "Kaplan_Meier [ENDPOINT] [Max number of days in plot] [Max y-axis value] [POPULATION].PDF", for example "Kaplan_Meier DEAD_MI_EVENT 365 100 ITT.PDF".

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 94 SUPPLEMENTARY DATA Statistical software Statistical analyses will primarily be performed using SAS v. 9.4 or later. R version 3.2.2 or later may be used for additional output.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 95 SUPPLEMENTARY DATA 3. DETO2X-AMI

PARTICIPATING CENTERS AND INVESTIGATORS

Danderyd University Hospital, Stockholm, Sweden. Rickard Linder, Mattias Ekström.

Enköping Hospital, Sweden. Lena Forsman.

Gothenburg University Hospital (Sahlgrenska), Sweden. Johan Herlitz, Annica Ravn-

Fischer, Elmir Omerovic, Oskar Angerås.

Gothenburg University Hospital (Östra), Sweden. Björn Hornestam.

Gävle Hospital, Sweden. Robert Kastberg, Espen Haugen.

Hallands Hospital, Halmstad, Sweden. Markus Lingman.

Hallands Hospital, Varberg, Sweden. Markus Lingman.

Härnosand Hospital, Sundsvall, Sweden. Anna Millbourn.

Kalmar Regional Hospital, Sweden. Anders Engström, Jörg Carlsson.

Karlstad Hospital, Sweden. Urban Haaga.

Karolinska University Hospital, Huddinge, Sweden. Tomas Jernberg.

Karolinska University Hospital, Solna, Sweden. John Pernow, Linda Mellbin, Dinos

Verouhis.

Kiruna Hospital, Sweden. Fredrik Kjellberg.

Kristianstad Hospital, Sweden. Raluca Jumatate.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA Köping Hospital, Sweden. Lennart Malmqvist, Gull-Britt Eriksson.

Lindesberg Hospital, Sweden. Thomas Kellerth.

Linköping University Hospital, Sweden. Joakim Alfredsson, Lennart Nilsson, Eva

Swahn, Dimitrios Venetsanos.

Ljungby Hospital, Sweden. Carina Nilsson.

Norrlands University Hospital, Umeå, Sweden. Krister Lindmark.

Norrtälje Hospital, Sweden. Tommy Pettersson, Melvin Pourbazargan.

Nyköping Hospital, Sweden. Martin Serrander.

Ryhov Hospital, Jönköping, Sweden. Jörg Lauermann, Jan-Erik Karlsson, Neshro

Barmano.

Sahlgrenska Universitetssjukhus Mölndal, Sweden. Martin Risenfors.

Skaraborgs Hospital, Lidköping, Sweden. Magnus Peterson.

Skaraborgs Hospital, Skövde, Sweden. Ylwa Wallström.

Skåne University Hospital, Lund, Sweden. David Erlinge, David Sparv, Ulf Ekelund.

Skåne University Hospital, Malmö, Sweden. David Erlinge, David Sparv.

St: Göran Hospital, Stockholm, Sweden. Marianne Erlandsson.

Södersjukhuset, Stockholm, Sweden. Robin Hofmann, Nils Witt, Mats Frick, Leif

Svensson, Ellinor Berglund.

Trelleborg Hospital, Sweden. Troels Yndigegn.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1 SUPPLEMENTARY DATA Uppsala University Hospital, Sweden. Stefan James, Bertil Lindahl, Gabriel Arefalk,

Bo Lagerqvist.

Vrinnevi Hospital, Norrköping, Sweden. Christofer Digerfeldt.

Växjö Hospital, Sweden. Olle Bergström.

Örebro University Hospital, Sweden. Thomas Kellerth

Örnsköldsvik Hospital. Björn Byström.

©2019 American Diabetes Association. Published online at http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc19-0590/-/DC1