University of Groningen Statin Treatment in Type 2 Diabetes

University of Groningen

Statin treatment in type 2 diabetes patients de Vries, Folgerdiena Maria

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record

Publication date: 2016

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): de Vries, F. M. (2016). Statin treatment in type 2 diabetes patients. Rijksuniversiteit Groningen.

Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.

Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 04-10-2021 Statin treatment in type 2 diabetes patients

Folgerdiena Maria de Vries The research presented in this thesis was supported by the Graduate School for Health Research (SHARE), University of Groningen.

Printing of this thesis was partially supported by the University of Groningen and the Graduate School for Health Research (SHARE).

Colophon: Author: Folgerdiena Maria de Vries Cover Design: Folgerdiena Maria de Vries Design and lay-out: Mirakels Ontwerp Printing: Gildeprint - The Netherlands ISBN: 978-90-367-8618-8 (printed version) ISBN: 978-90-367-8617-1 (digital version)

Copyright © 2016, Folgerdiena Maria de Vries. All rights reserved. No part of this thesis may be reproduced, stored in a retrievall system, or transmitted in any form or by any means without prior permission in writing from the author. Statin treatment in type 2 diabetes patients

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen op gezag van de rector magnificus prof. dr. E. Sterken en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op

vrijdag 26 februari 2016 om 16.15 uur

door

Folgerdiena Maria de Vries

geboren op 26 februari 1989 te Delfzijl Promotores Prof. dr. E. Hak Prof. dr. P. Denig Prof. dr. M.J. Postma

Beoordelingscommissie Prof. dr. E. Buskens Prof. dr. G.E.H.M. Rutten Prof. dr. O.H. Klungel Paranimfen Kirsten de Vries Birgit Kamminga-Bräuer TABLE OF CONTENTS

General introduction 8

Part I: Effects and cost-effectiveness of statin treatment

Chapter 1 Primary prevention of cardiovascular and cerebrovascular 20 events with statins in diabetic patients: a meta-analysis.

Chapter 2 38 for the secondary prevention of cardiovascular and Efficacy of standard and intensive statin treatment cerebrovascular events in diabetes patients: a meta- analysis.

Chapter 3 Adherence to standard-dose or low-dose statin treatment 58 and LDL-cholesterol response in type 2 diabetes patients.

Chapter 4 Cost-effectiveness of statins for primary prevention in 78 patients newly diagnosed with type 2 diabetes in the Netherlands.

Part II: Patient & physician behaviour regarding statin treatment

Chapter 5 Prescribing patterns, adherence and LDL-cholesterol 102 response of type 2 diabetes patients initiating statin on low-dose versus standard-dose treatment: a descriptive study.

Chapter 6 Does a cardiovascular event change adherence to statin 122 treatment in patients with type 2 diabetes? A matched cohort design.

Chapter 7 Preferences of diabetes patients for lipid-lowering 138 treatment: A discrete choice experiment.

6 Table of Contents

General discussion 158

Nederlandse samenvatting 176

Appendices 184

Curriculum Vitae 192

Dankwoord 196

SHARE publications 202

General introduction General introduction

The prevalence of type 2 diabetes is increasing in the Netherlands as in many countries worldwide [1]. In 2011, more than 830,000 patients were diagnosed with diabetes in the Netherlands, 90% of these patients with type 2 diabetes [2]. An increasing number of patients with obesity or physical inactivity, early diagnosis of type 2 diabetes, and aging of the population are among the main reasons for this increase in prevalence rate [2]. healthcare costs. The total Dutch healthcare costs related to diabetes were estimated at The high prevalence of diabetes reflects a substantial disease burden with considerable 1.7 billion euro’s for the year 2011 [3]. An important part of diabetes-related healthcare costs is caused by costs due to micro- and macro-vascular complications of diabetes [4-6].

Cardiovascular complications of diabetes Patients with type 2 diabetes are more likely to develop cardio- and cerebrovascular disease than individuals without diabetes. It has been estimated that patients with diabetes without a prior myocardial infarction have a risk for myocardial infarction as high as patients without diabetes, but with a history of myocardial infarction [7]. In addition, mortality rates following a myocardial infarction are higher among patients with diabetes [8]. Around 60% of patients with diabetes die from any type of cardiovascular disease or stroke [9]. Therefore, the control of cardiovascular risk factors, including dyslipidaemia, is essential.

Control of dyslipidaemia with statin treatment HMG coenzyme-A reductase inhibitors (statins) are the most important drug class for controlling dyslipidaemia. HMG-CoA reductase catalyses the conversion of HMG- CoA to mevalonate, which is the rate-limiting step in the cholesterol synthesis. Competitive inhibition of this enzyme by statins decreases hepatocyte cholesterol synthesis and stimulates expression of LDL-cholesterol receptors [10]. This results in a reduction in the intracellular cholesterol concentration which causes an increased extraction of LDL-cholesterol from the blood and therefore decreases circulating LDL- cholesterol concentrations [11]. Statins are also linked to various pleiotropic effects that are independent of their effects on lipids and associated with a variety of improved outcomes [12].

users in the Netherlands [13]. The extent to which the statin lowers LDL-cholesterol In 2014, five different statins were available on the Dutch market with over 1,900,000 is dependent on the potency and dosing (potency: rosuvastatin> atorvastatin> [14]. In general, doubling the statin dose has been associated with an additional 7% reduction in LDL-cholesterol from baseline [15]. Meta- simvastatin> pravastatin> fluvastatin) analyses have reported a reduced risk for cardiovascular events with statin treatment in both primary and secondary prevention populations [16,17]. Besides the reductions in

10 General introduction

cholesterol levels and cardiovascular events, adverse events have been associated with statin treatment. Muscle toxicity and effects on liver enzymes are well acknowledged, the risk is, however, low [18]. Rhabdomyolysis is a severe, but rare, form of muscle toxicity involving muscle breakdown which can cause renal failure and can be fatal [18]. Although the risk for adverse events is low, for patients with a low cardiovascular risk they could possibly outweigh the benefits of treatment. Statin treatment recommendations The need for statin treatment for secondary prevention of cardiovascular events is widely recognized. Since 2006, Dutch guidelines recommend statin treatment for both primary and secondary prevention for almost all diabetes patients [19,20]. Standard-dose statin treatment (simvastatin 40 mg) is recommended for all diabetes patients with a history of cardiovascular disease [19,20]. The use of statin treatment for primary prevention is, however, more controversial. The Cochrane Collaboration, for example, emphasized in 2011 that caution should be taken in prescribing statins to diabetes patients without risk-elevating medical conditions [17].

Treatment recommendations in the Dutch guidelines are based on individual estimated cardiovascular risk scores. Statin treatment is recommended for all patients with a acknowledges that for patients with low estimated risks or with low LDL-cholesterol 10-year cardiovascular risk of ≥10% and an LDL-cholesterol level >2.5 mmol/l. This levels statin treatment might not be cost-effective nor improve patient quality of life [17]. When following the current guideline an increasing number of diabetes patients might not be eligible for statin treatment as type 2 diabetes is nowadays diagnosed earlier and at a younger age, with lower cardiovascular risk at diabetes diagnosis as a result.

Cost-effectiveness of statin treatment The recommendation in the Dutch guidelines to start statin treatment with simvastatin 40 mg is based upon cost-effectiveness analyses [19]. In 2003, simvastatin was the simvastatin price. This price reduction had a positive effect on cost-effectiveness of first generic statin that entered the Dutch market resulting in a large reduction in the simvastatin treatment and allowed broader prescribing of simvastatin.

Cost-effectiveness calculations are usually based on efficacy data from clinical trials. Also, in clinical trials patients are treated under well controlled conditions, which is not Clinical trial populations often do not reflect the population treated in clinical practice. the case in clinical practice [21]. This probably affects cost-effectiveness, and may lead to overestimations.

11 General introduction

Treatment non-response in clinical practice Observational studies have shown that in clinical practice cholesterol targets are not reached by at least a third of patients [22,23]. This lack of treatment response could be due to being prescribed low-dose treatment [22,24] [25,26,27] [28,29]. , lack of treatment intensification and/or non-adherence to treatment Being prescribed a dose lower than recommended is a problem in clinical practice. Patients treated with a low dose are more likely not to reach their LDL-cholesterol target [22]. In 2006, 20 mg was the most often prescribed dose of simvastatin [22]. Although the treatment dose has been increasing over the years, patients were on average treated with a dose lower than recommended in the Dutch guideline in 2006 [19,22,24].

Furthermore, it has been shown that at least a third of the patients that are not reaching [25,26,27]. This could be related to non-adherence, as suspicion of non-adherence to treatment could their LDL-cholesterol target do not receive any treatment modification be reason for the physician not to intensify treatment. However, the ability of physicians to recognize non-adherence is poor [30]. There is some evidence of possible redundant [31], but also lack of treatment [26]. More insight in prescribing and adherence treatment intensification in non-adherent patients patterns is needed to improve treatment response. intensification in adherent patients

Adherence to treatment Non-adherence to statin treatment is an important problem, which is associated with less decrease in LDL-cholesterol [28,29] and with a higher risk for cardiovascular events and all-cause mortality [32]. The World Health Organization (WHO) has adopted the following The extent to which a person’s behaviour – taking medication, following a diet, and/or executing lifestyle changes, corresponds with definition of adherence to long-term treatment: agreed recommendations from a health care provider [33]. Adherence can be measured using prescription data as the Proportion of Days Covered (PDC) which expresses the proportion of days for which a patient has received medication in the study period [34]. Patients are often categorized as adherent or non-adherent for which an arbitrary cut- off point of 80% PDC is commonly used [35]. It has been reported that only 50% of the [36,37]. patients remains adherent, that is PDC >80%, during the first years of treatment treatment [38] Also discontinuation rates of statin treatment may be high during the first years of preferences. Non-adherence and discontinuation appear to be more common in patients . Adherence to and discontinuation of treatment is influenced by patient’s that are female, have no cardiovascular history, have experienced adverse effects and have a lower socioeconomic status or education [39,40,41]. Better insight in patient’s preferences for lipid-lowering drugs according to patient characteristics could be useful

12 General introduction

for motivating patients to continue using statin treatment and thereby improve health and economic benefits. Research aims and outline of the thesis Type 2 diabetes patients have a higher risk for developing cardiovascular and cerebrovascular disease therefore statins are recommended for almost all diabetes patients. This thesis will focus on statin treatment in type 2 diabetes patients. Part I is about the effects of statin treatment on LDL-cholesterol and cardiovascular and cerebrovascular outcomes and about cost-effectiveness estimates, using clinical trial and observational data from routine medical practice. It provides updates on meta- analyses and precise effect estimates for statin treatment in diabetes patients. For clinical decision making and cost-effectiveness analysis, it is important to have such precise estimates. Moreover, the impact of real-world adherence and dosing of statin treatment on the outcomes are assessed. These factors are expected to affect the cost-effectiveness of statin treatment in daily practice. Part II elaborates further on both patient-related factors (e.g. adherence to treatment, patient’s preferences) and physician-related factors

(e.g. treatment dosing, and modifications) that affect outcomes of statin treatment.

Part I. Effects and cost-effectiveness of statin treatment

• determine precise effect estimates of statins for primary and secondary prevention The aims of the first part are to: of cardiovascular events in diabetes patients; • assess the effect of dosing and adherence on cholesterol outcomes in diabetes patients; • determine the cost-effectiveness of statins for primary prevention of cardiovascular and cerebrovascular events in newly diagnosed type 2 diabetes patients taking adherence into account.

In Chapter 1 a meta-analysis is presented in which we aimed to determine the effect of standard-dose statin treatment on the risk of cardiovascular and cerebrovascular events in a diabetes population without cardiovascular history. In Chapter 2 meta-analyses are presented in which the effects of standard-dose and intensive-dose treatment are determined in a secondary prevention population. The effect of dose and adherence to statins on LDL-cholesterol outcomes is determined in a retrospective cohort study in Chapter 3.

13 General introduction

In Chapter 4 a cost-effectiveness analysis is presented for primary prevention of cardiovascular and cerebrovascular events in newly diagnosed type 2 diabetes patients. For this analysis the baseline risk of a population with type 2 diabetes is assessed using observational patient data, also the effect of real-world adherence rates on cost- effectiveness is assessed.

Part II. Patient & physician behaviour regarding statin treatment

The aims of the second part are to: • describe adherence and prescribing patterns of statin treatment in diabetes patients; • determine patient’s preferences for lipid-lowering drugs of diabetes patients.

Chapter 5 describes and evaluates statin prescribing and adherence patterns in relation to LDL-cholesterol response in a diabetes cohort study with two year follow-up after statin initiation. In Chapter 6 the effect of the occurrence of a cardiovascular event while being on statin treatment on adherence rates is determined in diabetes patients, using a matched cohort design. In the study presented in Chapter 7 we determined the willingness of patients to continue using lipid-lowering treatment, and the importance they attach to specific treatment characteristics. We explored whether these aspects are influenced by patient characteristics such as age, gender, clinical history and education. discussion. Finally, the main findings of the studies are summarized and discussed in the general

14 General introduction

REFERENCES

1. Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res Clin Pract 2010; 87(1): 4-14. 2. -

Available at: http://www.nationaalkompas.nl/gezondheid-en-ziekte/ziekten-en-aandoe ningen/endocriene-voedings-en-stofwisselingsziekten-en-immuniteitsstoornissen/ 3. diabetes-mellitus/omvang/ Assessed on 8 May 2015. - Available at: http://www.nationaalkompas.nl/gezondheid-en-ziekte/ziekten-en-aandoe ningen/endocriene-voedings-en-stofwisselingsziekten-en-immuniteitsstoornissen/ diabetes-mellitus/diabetes-mellitus-effectiviteit-kosten-en-gebruik-van-preventie- 4. en-zorg/Clarke P, GrayAssessed A, Legood on 10 R, June Briggs 2015. A, Holman R. The impact of diabetes-related complications on healthcare costs: results from the United Kingdom Prospective Diabetes Study (UKPDS Study No. 65). Diabet Med 2003; 20(6): 442-50. 5. Ringborg A, Yin DD, Martinell M, Stålhammar J. The impact of acute myocardial infarction and stroke on health care costs in patients with type 2 diabetes in Sweden. Eur J Cardiovasc Prev Rehabil 2009; 16(5): 576-82. 6. Redekop WK, Koopmanschap MA, Rutten GE, Wolffenbuttel BH, Stolk RP, Niessen LW. Resource consumption and costs in Dutch patients with type 2 diabetes mellitus: results from 29 general practices. Diab Med 2002; 19(3): 246-53. 7. Haffner SM, Lehto S, Rönnemaa T, Pyörälä K, Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Eng J Med 1998; 339(4): 229-34. 8.

Miettineninfarction. H, The Lehto FINMONICA S, Salomaa V,myocardial et al. Impact infarction of diabetes register on mortality study after group. the firstDiabetes myocardial Care 1998;21(1): 69-75. 9. Van der Meulen A. Sterfte aan diabetes. Bevolkingstrends. 2005b; 53(1): 64-68. Available

from: http://www.cbs.nl/NR/rdonlyres/CF0D2FF7-1E85-4F72-A83C-CD59EE2CA9E2 10. /0/2005k1b15p064art.pdfVaughan CJ, Gotto AM. Update Assessedon statins: 23 2003. July Circulation 2015. 2004; 110(7): 886-892. 11. Schachter M. Chemical, pharmacokinetic and pharmacodynamics properties of statins: an update. Fundam Clin Pharmacol 2005; 19(1): 117-125. 12. Mihos CG, Salas MJ, Santana O. The pleiotropic effects of the hydroxyl-methyl-glutaryl-CoA reductase inhibitors in cardiovascular disease. A comprehensive review. Cardiol Rev 2010; 18 (6) :298-304. 13.

The Drug Information System of National Health Care Institute. Available from: https:// www.gipdatabank.nl/databank.asp?tabel=01-basis&geg=gebr&item=C10AA.July 2015 Accessed 23

15 General introduction

14. Smith MEB, Lee NJ, Haney E, Carson S. Drug class review: Hmg-coa reductase inhibitors

(statins)Portland (OR); and fixed-dose 2009 Drug combination Class Reviews. products containing a statin: Final report update 5. 15. Roberts WC. The rule of 5 and the rule of 7 in lipid-lowering by statin drugs. Am J Cardiol 1997; 80(1): 106-107. 16.

adherencetimeliness. toPLoS guidelines One 2011; in 6(9):diabetes e24278. care according to different definitions of adequacy and 24. Geleedst-De Vooght M, Maitland-van der Zee AH, Schalekamp T, Mantel-Teeuwisse A, Jansen P. Statin prescribing in the elderly in the Netherlands: A pharmacy database time trend study. Drugs Aging 2010; 27(7): 589-596. 25.

Rodondihypertension, N, Peng dyslipidemia, T, Karter AJ, and et diabetes al. Therapy mellitus. modifications Ann Intern in Med response 2006; to144(7): poorly 475-484. controlled 26. Schmittdiel JA, Uratsu CS, Karter AJ, et al. Why don’t diabetes patients achieve recommended J Gen Intern Med

risk2008; factor 23(5): targets? 588-594. Poor adherence versus lack of treatment intensification. 27. -

Viranication forSS, elevatedWoodard cholesterol LD, Chitwood levels SS, in et patients al. Frequency with cardiovascular and correlates disease. of treatment Am Heart intensifi J 2011; 162(4): 725-732.e1. 28. Parris ES, Mohn LA, Lawrence DB, Long LB. Adherence to statin therapy and LDL cholesterol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care 2005; 28(3): 595-599. 29. Watanabe JH, Bounthavong M, Chen T. Revisiting the medication possession ratio threshold for adherence in lipid management. Curr Med Res Opin 2013; 29(3): 175-180.

16 General introduction

30. Osterberg L, Blaschke T. Adherence to medication. N Eng J Med 2005; 353(5): 487-97. 31. Pittman DG, Fenton C, Chen W, Haffner S, Pendergrass M. Relation of statin nonadherence and

32. treatmentDe Vera MA, intensification. Bhole V, Burns Am LC, J Lacaille Cardiol D.2012; Impact 110(10): of statin 1459-1463. adherence on cardiovascular disease and mortality outcomes: a systematic review. Br J Clin Pharmacol 2014; 78(4) :684-698. 33.

AvailableAssessed on at:10 June http://www.who.int/chp/knowledge/publications/adherence_report/en/ 2015. 34. Vink NM, Klungel OH, Stolk RP, Denig P. Comparison of various measures for assessing

medication18(2): 159-165. refill adherence using prescription data. Pharmacoepidemiol Drug Saf 2009; 35. Simpson RJ, Mendys P. The effects of adherence and persistence on clinical outcomes in patients treated with statins: a systematic review. J Clin Lipidol 2010; 4(6): 462-471. 36. Donnelly LA, Doney AS, Morris AD, Palmer CN, Donnan PT. Long-term adherence to statin treatment in diabetes. Diabet Med 2008; 25(7): 850-855. 37. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. JAMA 2002; 288(4): 455-461. 38. Zhang H, Plutzky J, Skentzos S, et al. Discontinuation of statins in routine care settings. Ann Intern Med 2013; 158(7): 526-534. 39. Mann DM, Woodward M, Muntner P, Falzon L, Kronish I. Predictors of nonadherence to statins: a systematic review and meta-analysis. Ann Pharmacother 2010; 44(9): 1410-1421. 40. Wei MY, Ito MK, Cohen JD, Brinton EA, Jacobson TA. Predictors of statin adherence, switching, and discontinuation in the USAGE survey: understanding the use of statins in America and gaps in patient education. J Clin Lipidol 2013; 7(5): 472-483. 41. McGinnes B, Olson KL, Magid D, et al. Factors related to adherence to statin therapy. Ann Pharmacother 2007; 41(11): 1805-1811.

PART I

Effects and cost-effectiveness of statin treatment

CHAPTER 1

Primary prevention of major cardiovascular and cerebrovascular events with statins in diabetic patients: a meta-analysis

Folgerdiena M. de Vries1 Petra Denig2 Koen B. Pouwels1 Maarten J. Postma1 Eelko Hak1

Drugs 2012;72(18):2365-2373.

1. University Groningen, Department of Pharmacy, Unit of PharmacoEpidemiology & PharmacoEconomics, Groningen, the Netherlands 2. Department of Clinical Pharmacy and Pharmacology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands CHAPTER 1

ABSTRACT

Background: Patients with diabetes are at increased risk of developing cardiovascular disease. Controlling lipid levels has a preventive effect on the occurrence of major cardiovascular and cerebrovascular events. Individual trials

therapy in the primary prevention of such events in diabetes. have shown varying data on the efficacy of treatment with lipid-lowering statin

Objective:

The objective of this study was to assess the efficacy of statins in the primary prevention of the first-time occurrence of a major cardiovascular or cerebrovascular event in diabetes patients. Secondary end points were fatal/non- fatal stroke, fatal/non-fatal myocardial infarction and all-cause mortality. Methods: A systematic search for trial reports was conducted in PubMed, EMBASE, The Cochrane library and clinicaltrials.gov for the years 1966-2011. Reference lists of reviews and meta-analyses of related subjects were searched. High-quality, randomized, double-blinded clinical trials comparing a statin with placebo for the primary prevention of major cardiovascular and cerebrovascular events in diabetes patients were selected. Only large studies with a minimum of at least 500 diabetic participants followed-up for at least 2 years were included. End points were major cardiovascular and cerebrovascular events.

Trial and patient characteristics were extracted by three researchers. The quality of the included studies was tested with the Jadad score. The combined effect on

Publication bias was examined with a funnel plot. primary as well as secondary end points was measured with a fixed-effect model.

Results: Four trials were included, with a total of 10,187 participants. Treatment with statins in the primary prevention of major cardiovascular and cerebrovascular

events in diabetes patients resulted in a significant relative risk (RR) reduction in the first-time occurrence of major cardiovascular or cerebrovascular events (RR 0.75; 95% confidence interval [CI]: 0.67-0.85), fatal/non-fatal stroke (RR 0.69; 95%CI: 0.51-0.92) and fatal/non-fatal myocardial infarction (RR 0.70; 95%CI: 0.54-0.90) and a non-significant RR reduction in all-cause mortality (RR 0.84; the individual effect estimates and no publication bias. 95%CI 0.65-1.09). Among the studies there was non-significant heterogeneity in

22 Statins for primary prevention in diabetes

Conclusion: Treatment with statins in primary prevention among diabetes patients

has a significant beneficial effect on event rates of the first-time occurrence of a 1 major cardiovascular or cerebrovascular event, fatal/non-fatal stroke and fatal/ cause mortality. non-fatal myocardial infarction. There was a non-significant RR reduction in all-

INTRODUCTION

Cardiovascular diseases are the number one cause of death in Western countries [1]. The American Heart Association considers diabetes as one of the six major controllable risk factors for cardiovascular disease [2]. Patients with diabetes are two to four times more likely to develop cardiovascular disease or stroke than non-diabetic individuals, and mortality rates after a myocardial infarction are also higher among diabetes patients [2,3]. At least 65 per cent of patients with diabetes dies from any type of cardiovascular disease or stroke [2]. Hence, the control of cardiovascular risk factors in people with diabetes is essential and needs to be targeted at controlling hyperglycaemia and hypertension as well as controlling dyslipidaemia.

Hydroxymethylglutaryl coenzyme-A reductase inhibitors (statins) are important in and mortality in patients with established cardiovascular disease [4,5]. There is, however, regulating lipids. Statins have shown high efficacy in reducing cardiovascular morbidity controversy over their value in the primary prevention of cardiovascular disease. For that reason, the Cochrane Collaboration emphasizes that caution should be taken in prescribing statins to diabetes patients without risk-elevating medical conditions [6].

Diabetes patients are an important subgroup in the population of primary prevention because of their higher risk of cardiovascular disease [2]. In a 2008 meta-analysis the authors concluded that reduction of major vascular events with statin treatment was similar in diabetes patients with and without vascular disease [7]. This analysis did not include the more recent ASPEN trial [8] (see Table 1 for full study names) which showed no benefit of statins in primary prevention diabetes patients. Since then, two patients without cardiovascular disease. Brugts et al. [9] compared the effects of statins other meta-analyses have been conducted assessing the efficacy of statins in diabetes in diabetes and non-diabetes patients but did not present a calculation of risk reduction including studies with only diabetes patients, such as the ASPEN trial [8]. Chen et al. [10] did include this trial but also included patients from the PROSPER trial [11] some of whom

23 CHAPTER 1

were secondary prevention patients. To retrieve accurate and precise estimates we aimed to assess the effect of statins in the primary prevention of major cardiovascular and cerebrovascular events in diabetes patients. Because some end points may be vulnerable to performance and detection bias, [12] and we aimed to derive high-quality estimates, only double-blinded, placebo-controlled trials were eligible for inclusion in our meta-analysis.

METHODS

Search strategy We searched EMBASE, PubMed, The Cochrane library and clinicaltrials.gov for English- language randomized clinical trials that compared treatment with a statin versus placebo in diabetes patients in the primary prevention of major cardiovascular and cerebrovascular events (1966 to November 2011). To identify these trials, the medical subject heading (MeSH) terms “HMG-CoA reductase inhibitors” and “diabetes mellitus” were used and we added the terms “statin”, “diabetes” and “cardiovascular disease”. We then searched for randomized clinical trial reports. We also examined the reference lists of reviews and meta-analyses on subjects related to the study. Two researchers performed the search (FV and KP).

Study selection Studies were included if they were double-blinded randomized clinical trials that compared a statin with placebo in diabetes patients for the primary prevention of major cardiovascular and cerebrovascular events. Separate data on primary prevention diabetes patients needed to be available from the report. Studies were required to have at least 500 diabetic participants. The mean follow-up had to be at least 2 years, because we chose to examine long-term relevant clinical endpoints. The quality of the selected studies was scored by the Jadad score [13]. Scores could range from 0 to 5; higher scores

Three researchers (FV, KP, EH) individually extracted the trial and patient characteristics, indicate better quality. The included studies were all of good quality with scores ≥ 4. the outcome results, and conducted the quality scoring.

Endpoints The primary outcome in our meta-analysis was major cardiovascular and cerebrovascular infarction, non-fatal stroke, unstable angina or revascularization. Secondary endpoints events, which are the first-time occurrence of cardiovascular death, non-fatal myocardial assessed were non-fatal and fatal stroke, non-fatal and fatal myocardial infarction and all-cause mortality. Included events for the endpoints are reported in Table 2. Effects

24 Statins for primary prevention in diabetes

overall occurrence of the selected events. In the primary endpoint, events needed to in the studies were measured either as the first-time occurrence of an event or the be measured as the first-time occurrence of an event, which caused small unavoidable 1 differences in endpoint definitions. Data analysis and statistical methods the absolute risk reduction and the number needed to treat (NNT) with 95%CI were For each endpoint, the relative risk (RR) with 95% confidence interval (95%CI), calculated [14] and the random-effect model in the computer program RevMan [15,16]. After calculating . The results of the separate studies were pooled with the fixed-effect model model to use in this case. both models, we compared them and concluded that the fixed-effect model was the best

The decision was made on the results of the Tau2 test in the random-effect model and the number of trials included in the analysis [17]. The Tau2 test gives the underlying estimate of the between-studies variance. In general, the Tau2 test does not systematically increase with the number or size of studies included in a meta-analysis. Therefore, making a decision based on the Tau2 test is better than making a decision based on the I2 test, which systematically increases with the size of the studies [18]. The quantity I2 measures the percentage of total variation across studies that is due to heterogeneity rather than chance [19]. There is one caveat in making a decision based on the Tau2: when the number of studies involved is very small, the estimation of the Tau2 will have poor precision. Because of this, although the outcome of the Tau2 in all-cause mortality prefers the were included [17]. The test statistic Q was calculated to test for heterogeneity [15]. Testing mixed-effect model, the fixed-effect model is also used in this case since only two studies for publication bias was done by visually examining funnel plots.

25 CHAPTER 1

TABLE 1. Full trial name of study acronyms Study acronyms Trial name ASCOT-LLA, 2005 [20] Anglo-Scandinavian Cardiac Outcomes Trial–Lipid-Lowering Arm ASPEN, 2006 [8] Atorvastatin Study for Prevention of coronary heart disease Endpoints in Non-insulin-dependent diabetes mellitus PROSPER, 2002 [11] PROspective Study of Pravastatin in the Elderly at Risk CARDS, 2004 [21] Collaborative AtoRvastatin Diabetes Study HPS, 2005 [22] Heart Protection Study of cholesterol-lowering with simvastatin in people with diabetes 4D, 2005 [25] Die Deutsche Diabetes Dialyse Studie MEGA, 2006 [24] Management of Elevated cholesterol in the primary prevention Group of Adult Japanese ALLHAT-LLT, 2002 [23] Antihypertensive and Lipid-Lowering treatment to Prevent Heart Attack Trial

RESULTS

Randomized clinical trials Results from four randomized clinical trials were included in the meta-analysis (Figure 1): the ASCOT-LLA [20], the ASPEN [8], the CARDS [21] and the HPS [22] (see Table 1 for full study names). The study and patient characteristics are represented in Table 3 and Table 4. In total, 10,187 participants were included in this meta-analysis, 5,100 of whom were treated with statins and 5,087 of whom were allocated to the placebo group. Those in the treatment group were treated with 40 mg simvastatin or 10 mg atorvastatin. The mean follow-up was 3.8 years (range 2.4 to 4.8). The mean age was 62 years, and 70% of participants was male. On average the low-density lipoprotein (LDL)-cholesterol was lowered by 1.0 mmol/L (range -0.9 to -1.2).

26 Statins for primary prevention in diabetes

1 Non-fatal MI, including Non-fatal CHD; fatal/ MI, fatal silent stroke non-fatal MI; fatal/ Fatal/non-fatal stroke non-fatal non-CV CV mortality; mortality MI, MI, non-fatal Fatal MI; stroke including silent any from type; death of any cause Other reversible reversible CV mortality, non-fatal MI, unstable/ non-fatal CV mortality, angina, life-threatening stable heart non-fatal arrhythmias, peripheral stroke, non-fatal failure, thrombosis, disease, retinal arterial TIA, revascularization, deficits ischaemic neurological MI, non-fatal non-fatal CV death, arrest, cardiac resuscitated stroke/TIA, angina requiring or unstable worsening CABG recanalization, hospitalization, MI, fatal/non-fatal Fatal/non-fatal coronary TIA, other acute stroke, cardiac resuscitated heart death, angina, coronary unstable arrest, revascularization coronary any from MI, death Non-fatal type, coronary/ of any disease, strokes other revascularization Included events in endpoints Included events and Major cardiovascular events cerebrovascular Pts aged 40–79 y, with 40–79 y, Pts aged and T2DM two hypertension cholesterol total other risk factors, TG if <6.5 mmol/L, fasting even <4.5 mmol/L levels with T2DM, 40–75 y, Pts aged ≤4.1 mmol/L LDL-C levels, with T2DM and 40–75 y, Pts aged factor, one additional risk least at of ≤4.14 mmol/L and LDL-C levels of ≤6.78 mmol/L TGs with diabetes, 40–80 y, Pts aged of ≥3.5 cholesterol total non-fasting mmol/L Inclusion criteria (participants) Inclusion criteria Previous MI, currently treated treated MI, currently Previous event angina, cerebrovascular 3 mo within previous MI, interventional Previous procedures vascular MI, angina, coronary cerebrovascular surgery, peripheral severe accident, disease vascular other vascular MI, other CHD, disease Exclusion criteria (primary criteria Exclusion prevention) 2005 [20] 2004 [21] 2006 [8] 2005 [22] TABLE and inclusion criteria 2. Exclusion ASPEN, CARDS, HPS, pts: infarction; myocardial MI: cholesterol; lipoprotein LDL-C: low-density cardiovascular; heart disease; CV: coronary CHD: grafting; bypass artery CABG: coronary TG: triglycerides. mellitus; T2D: type 2 diabetes attack; ischemic TIA: transient patients; Studies ASCOT-LLA, ASCOT-LLA,

27 CHAPTER 1

Potentially relevant reports screened for retrieval (N=515)

Reports excluded on basis of title or abstract (N=507)

Extra reports screened for retrieval from reviews and meta-analyses (N=3)

Reports retrieved for more detailed evaluation (N=11)

Reports excluded on basis of exclusion criteria (N=7): • not double-blinded (n=2) • surrogate endpoint (n=1) • no separate results (n=1) • small number participants (n=2) • specific population (n=1)

Included randomised controlled trials (N=4)

FIGURE 1. Flow diagram of study selection

The exclusion criteria for primary prevention in the studies were not exactly alike (Table 2). All studies excluded patients with prior myocardial infarction but only two also excluded those with cerebrovascular disease. There were also some differences in the events included in the reported major cardiovascular and cerebrovascular endpoint (Table 2). HPS [22] did not include unstable angina or a transient ischaemic attack (TIA), whereas the ASCOT-LLA study [20] included several minor events.

Primary endpoint When looking at the separate results of the included studies for the primary endpoint (major cardiovascular and cerebrovascular events) three out of fours studies showed (Figure 2). In this meta-analysis, treatment with a statin a significant RR reduction major cardiovascular and cerebrovascular events in diabetes patients (RR 0.75, 95%CI: was associated with a significant 25% risk reduction in the first-time occurrence of 0.67-0.85) (Figure 2). In the statin treatment group, 8.5% of subjects had a major

(Table 5) cardiovascular or cerebrovascular event (434/5,100), whereas 11.3% of those in the time occurrence of one major cardiovascular or cerebrovascular event, 35 people needed control group experienced such an event (576/5,087) . To prevent the first-

28 Statins for primary prevention in diabetes

to be treated for 3.8 years (NNT 35, 95%CI: 25-58) (Table 5). There was no publication bias and non-significant heterogeneity in the individual effect estimates. TABLE 3. Study characteristics Study Diabetes Statin Compliance No. of participants Mean follow- 1 pts (%) (statin/control) up (y) ASCOT-LLA, 2005 [20] 100 ATO 10 mg NA 2,532 3.3 (1,258/1,274)

ASPEN, 2006 [8] 100 ATO 10 mg Excluded when 1,905 2.4 less than 80% (959/946) compliant

CARDS, 2004 [21] 100 ATO 10 mg Excluded when 2,838 4 less than 80% (1,428/1,410) compliant

HPS, 2003 [22] 100 SIM 40 mg Compliant 2,912 4.8 when taking at (1,455/1,457) least 80% ATO: atorvastatin; NA: not applicable; pts: patients; SIM: simvastatin.

Secondary endpoints Participant and event numbers and outcome results for the secondary outcomes are listed in Table 5 of 31% (RR 0.69, 95%CI: 0.51-0.92). Two out of three of the studies included in the . We found a significant RR reduction in fatal and non-fatal stroke meta-analysis did not find a significant RR reduction. For fatal and non-fatal myocardial infarction there was a significant 30% RR reduction (RR 0.70, 95%CI 0.54-0.90). The reduction of 16% (RR 0.84, 95%CI 0.65-1.09) (Table 2 and Table 5). NNT was 86 (95%CI 50-290). For all-cause mortality, we found a non-significant risk

29 CHAPTER 1

TABLE 4. Patient characteristics a

Characteristic ASCOT-LLA[20] ASPEN[8] CARDS[21] Men (%) 77 62 68

Mean age (y) 64 60 62

Smoking (%) 20.4 12 23

Hypertension (%) 100 52 84

Mean BMI (kg/m2) 30 29 29

Mean SBP (mmHg) 165 133 144

Mean DBP (mmHg) 92 77 83

Total cholesterol (mmol/L) 5.3 5.0 5.4

LDL-cholesterol (mmol/L) 3.3 (−0.9) 2.9 (−0.9) 3.0 (−1.2)

HDL-cholesterol (mmol/L) 1.2 1.2 1.4

TG (mmol/L) 1.9 1.6 1.7

a Of the 5,963 patients in the HPS trial, [22] this meta-analysis only examined data in a subset of 2,912 patients who had no history of any arterial disease. Baseline characteristics of this subset of patients have not been published separately.

BMI: body mass index; DBP: diastolic blood pressure; HDL: high-density lipoprotein; LDL: low-density lipoprotein; SBP: systolic blood pressure; TG: triglycerides.

TABLE 5. The combined results for the primary and secondary endpoints Major cardio- and Fatal/non-fatal Fatal/non-fatal All-cause cerebrovascular stroke MI mortality events No. of participants 5,100/5,087 3,645/3,630 3,645/3,630 2,387/2,356 (statin/control) No. of events (statin/ 434/576 75/109 99/141 105/123 control) RR (95% CI) 0.75 (0.67–0.85) 0.69 (0.51–0.92) 0.70 (0.54–0.90) 0.84 (0.65–1.09) Q 5.59 2.14 2.65 1.85 I2 46% 7% 25% 46% Tau 2 0.01 0 0.02 0.03 NNT (95% CI) 35 (25–58) 101 (59–362) 86 (50–290) 130 (50–∞) NNT: number needed to treat; RR: relative risk.

30 Statins for primary prevention in diabetes 2

1.5 1 Favours control Favours 1 Risk Ratio 0.7 M-H, Fixed (95% CI) 0.5 Favours experimentalFavours Risk Ratio 0.97 (0.75-1.26) 0.78 (0.62-0.98) 0.78 0.69 (0.56-0.85) 0.69 0.65 (0.49-0.84) 0.65 0.75 (0.67, 0.85) 0.75 (0.67, M-H, Fixed (95% CI) 100% 17.8% 26.0% 22.2% 34.0% Weight 946 1274 1410 1457 5087 Total Placebo 576 151 102 127 196 Events 959 1428 1455 5100 1258 Total Experimental 83 116 135 434 100 Events 2005

Results for the primary prevention of major cardiovascular and cerebrovascular events with statins in diabetic patients in diabetic statins with events and cerebrovascular cardiovascular of major the primary prevention for Results 2004 2006

(95% CI) 2003 FIGURE 2. 2. FIGURE ASPEN CARDS ASCOTT-LLA Total Total events Total Heterogeneity: Chi² l² = 5.59, df = 0.13); = 46% = 3 (P forTest overall effect:(P < 0.00001) Z = 4.72 HPS Study or Subgroup

31 CHAPTER 1

DISCUSSION

This meta-analysis shows that treatment with statins is associated with a 25% risk events in diabetes patients without established vascular disease. In order to prevent reduction in the first-time occurrence of major cardiovascular and cerebrovascular one major cardiovascular or cerebrovascular event, 35 people should be treated for

3.8 years. Significant risk reductions of about 30% in fatal or non-fatal stroke and fatal reduction of 16% in all-cause mortality. or non-fatal myocardial infarction were also found. There was a non-significant risk because the ASPEN study [8] As mentioned earlier, there is conflicting evidence from the randomized clinical trials and cerebrovascular events. This study was not included in the previous meta-analysis did not show significant risk reductions in cardiovascular by Kearney et al. [7] estimates. The ASPEN study [8] population was younger, less hypertensive and included but inclusion in our meta-analysis resulted in precise beneficial fewer smokers and men than the CARDS [21] and ASCOT-LLA [20] studies. The follow-up period of 2.4 years was the shortest of the included studies. These might be reasons for the finding of non-significant results. double-blinded studies. This caused the ALLHAT-LLT [23] and MEGA [24] studies to be Our study differed from previous meta-analyses in that our inclusion criteria specified excluded. The ALLHAT-LLT study [23] mortality and coronary heart disease. The non-blinded study design was given as a did not find significant risk reductions of all-cause possible reason. When looking at differences in lipid lowering between double-blinded and non-blinded studies, it is evident that the double-blinded studies [8,20-22] show much larger differences between statin and control groups than the non-blinded studies [23,24]. This could be due to higher use of statins or use of non-pharmacologic cholesterol interventions in the control group. In the MEGA trial, [24] the control group was a diet group which could also cause a decrease in the difference in lipid lowering between the two groups.

The 4D trial [25] study population. In the 4D study, which compared atorvastatin with placebo on was also excluded from the meta-analysis because of the specific cardiovascular outcomes in 1,255 type 2 diabetes patients on maintenance hemodialysis the investigators found that after a median follow-up of 4 years atorvastatin (20 mg/ high number of cardiovascular events and an overall 24% cardiovascular mortality. day) decreased the RR by 8% (95% confidence interval, 0.77-1.10; P=0.37) despite a Additional pathogenic pathways in cardiovascular disease could be the reason for not finding significant reductions in the incidence of cardiovascular events in this trial. 32 Statins for primary prevention in diabetes

The initiation of the statin treatment might come too late to translate into consistent improvement of the incidence of cardiovascular events in patients with end-stage renal disease [25].

Inclusion of secondary prevention patients is a problem in primary prevention studies. 1 It is likely that all studies are dealing with inclusion of some secondary prevention patients. In the ASCOT-LLA study, [20] 12% of diabetes patients had some pre-existing cardiovascular disease, which was not in their exclusion criteria for primary prevention. reduction of 25%, which is greater than the 23% risk reduction they originally found for When these patients were excluded, the investigators still found a significant risk their primary endpoint (total cardiovascular events and procedures). When excluding the ASCOT-LLA study [20] from our primary endpoint results we found a 26% (RR 0.74, or cerebrovascular events. 95%CI: 0.65-0.85) risk reduction in the first-time occurrence of a major cardiovascular

Another problem can be early termination of a trial. The CARDS study [21] terminated 2 was reached. We believe this should not be a reason for excluding the CARDS study [21] years earlier than expected because the pre-specified early stopping rule for efficiency from the meta-analysis, especially as included patients were followed-up for a median duration of 3.9 years (interquartile range 3.0-4.7 years). When we did exclude the CARDS study [21] from the meta-analysis we found a 22% (RR 0.78, 95%CI: 0.69-0.89) event which is not substantially different from the overall results. risk reduction in the first time occurrence of a major cardiovascular or cerebrovascular

There are some limitations to this meta-analysis. First, the diagnostic criteria for diabetes for diabetes, the ASCOT-LLA [20] study based the diagnosis of diabetes on a self-reported differed among the studies. Where most used the World Health Organisation definition history and receiving any treatment; this included dietary manoeuvres. In their group of diabetes patients only around 50% were receiving oral hypoglycaemics and around 7% were treated with insulin. This might make their diabetes patients less far advanced. A second limitation is that the criteria that were set in the separate studies for primary prevention were not all the same which made a difference in the baseline risk of the populations. Third, the events included in the primary endpoint differed. Also, not all studies used the same statin; three used atorvastatin 10 mg and one used simvastatin 40 mg. Further, not all patients who were prescribed statins were still taking them at the end of follow-up; furthermore, patients in the placebo group were also receiving statins, which will lead to lower risk reductions. Finally, the use of subgroups from the ASCOT-LLA study [20] and HPS trial [22] could have caused some bias, but we do not think to balance treatment groups in terms of major prognostic factors. this had a major influence, because minimization randomisation techniques were used 33 CHAPTER 1

Primary prevention by statins could reach a large group of diabetes patients, but it needs to be acknowledged that the absolute reduction in events is rather low because of the low baseline risk. However, the NNT found in randomized clinical trials is not high and have changed in recommending statins for a larger group of people [26,27]. In 2006, the economic analyses are needed to establish its efficiency. In the last decade, guidelines Dutch guidelines [28,29] changed, recommending statin treatment in almost all patients with diabetes. However, exceptions were made for patients with low estimated risks or with low LDL-cholesterol levels, thereby acknowledging that, for such diabetes patients, willingness of patients to take statin treatment also play an important role in decision statin treatment may not have an optimal benefit-risk balance. Non-adherence and making. The overall better health of people without vascular complications makes them less likely to be adherent to drug therapy for primary prevention [30]. Observational data showed that the proportion of diabetes patients that were adherent 1 year after the start of statin treatment was less than 50%. Long-term adherence was also poor in diabetes patients, especially in those patients with few other cardiovascular risk factors [31]. Effectiveness data from observational studies could be useful for economic evaluations. Olafsdottir et al. [32] examined the effect of statins on mortality in older type 2 diabetes patients without cardiovascular heart disease. The outcomes in that study supported the effects found in clinical trials [32].

CONCLUSION

cardiovascular and cerebrovascular events in diabetes patients and may reduce all- Treatment with statins can have a beneficial effect in the primary prevention of major cause mortality. New insights from observational studies that also take effects of non- adherence into account could be of great importance for economic evaluations and decision making.

34 Statins for primary prevention in diabetes

REFERENCES

World Health organisation. Global Health Observatory 2008. http://www.who.int/gho/ 2. mortality_burden_disease/causes_death_2008/en/index.html 1 American Heart Association. Cardiovascular disease & Diabetes statistics update 2010. http:// www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetesMatters/Cardiovascular- 3. Disease-Diabetes_UCM_313865_Article.jsp Miettineninfarction. H, The Lehto FINMONICA S, Salomaa myocardial V, et al. Impact infarction of diabetes register on mortality study group. after Diabetes the first myocardialCare 1998; 21(1): 69-75. 4.

Baigentprospective C, Keech meta-analysis A, Kearney of PM,data et from al. Efficacy90,056 participantsand safety of in cholesterol-lowering 14 randomised trials treatment: of statins. Lancet 2005; 366(9493): 1267-1278. 5. Wei L, Ebrahim S, Bartlett C, Davey PD, Sullivan FM, MacDonald TM. Statin use in the secondary prevention of coronary heart disease in primary care: cohort study and comparison of inclusion and outcome with patients in randomised trials. BMJ 2005; 330(7495): 821. 6. Taylor F, Ward K, Moore TH, et al. Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev 2011; (1): CD004816. 7.

Kearney686 people PM, with Blackwell diabetes L, in Collins 14 randomised R, et al. Efficacytrials of ofstatins: cholesterol-lowering a meta-analysis. therapy Lancet 2008; in 18 371(9607): 117-125. 8.

Knoppprevention RH, of d’Emden cardiovascular M, Smilde end JG,points Pocock in subjects SJ. Efficacy with andtype safety2 diabetes: of atorvastatin The Atorvastatin in the study for prevention of coronary heart disease endpoints in non-insulin-dependent diabetes mellitus (ASPEN). Diabetes Care 2006; 29 (7): 1478-1485. 9. -

Brugtsvascular JJ, disease Yetgin T, but Hoeks with SE, cardiovascular et al. The benefits risk factors: of statins meta-analysis in people without of randomised established controlled cardio trials. BMJ 2009; 338: b2376. 10. Chen YH, Feng B, Chen ZW. Statins for primary prevention of cardiovascular and cerebrovascular events in diabetic patients without established cardiovascular diseases: a meta-analysis. Exp Clin Endocrinol Diabetes 2011; 120(2): 116-120. 11. Shepherd J, Blauw GJ, Murphy MB, et al. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet 2002; 360(9346): 1623-1630. 12. Jüni P, Altman DG, Egger M. Systematic reviews in health care: Assessing the quality of controlled clinical trials. BMJ 2001; 323 (7303): 42-46. 13. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17(1): 1-12. 14. Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment. BMJ 1995; 310(6977): 452-454.

35 CHAPTER 1

15. Scholten RJ, Kostense PJ, Assendelften WJ, Bouter LM. De praktijk van systematische reviews. IV. Het combineren van de resultaten van afzonderlijke onderzoeken. [In Dutch] Ned Tijdschr Geneeskd 1999; 143(15): 786-791. 16. Review Manager (RevMan) [Computer program]. Version 5.1. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011. 17.

Borensteineffect versus M, random Hedges effects LV, Higgins models. JPT, John et al. Wiley (2009). & Sons Introduction, Ltd. (59-86) to meta-analysis. Part 3 fixed- 18. Rücker G, Schwarzer G, Carpenter JR, Schumacher M. Undue reliance on I(2) in assessing heterogeneity may mislead. BMC Med Res Methodol 2008; 8: 79. 19. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327(7414): 557-560. 20. Sever PS, Poulter NR, Dahlöf B, et al. Reduction in cardiovascular events with atorvastatin in 2,532 patients with type 2 diabetes: Anglo-Scandinavian Cardiac Outcomes Trial –lipid- lowering arm (ASCOT-LLA). Diabetes Care 2005; 28(5): 1151-1157. 21. Colhoun HM, Betteridge DJ, Durrington PN, et al. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo-controlled trial. Lancet 2004; 364(9435): 685-696. 22.

Theoutcomes ALLHAT in moderatelyofficers and hypercholesterolemic, coordinators for the ALLHAThypertensive collaborative patients researchrandomized group. to pravaMajor- statin vs usual care. The antihypertensive and lipid-lowering treatment to prevent heart attack trial (ALLHAT-LLT). JAMA 2002; 288(23): 2998-3007. 24. Nakamura H, Arakawa K, Itakura H, et al. Primary prevention of cardiovascular disease with pravastatin in Japan (MEGA Study): a prospective randomised controlled trial. Lancet 2006; 368(9542): 1155-1163. 25. Wanner C, Krane V, März W, et al. German diabetes and dialysis study investigators. Atorva- statin in patients with type 2 diabetest mellitus undergoing hemodialysis. N Eng J Med 2005; 353(3): 238-248. 26. Grundy S, Cleeman JI, Merz NB, et al., for the Coordinating Committee of the National Choles- terol Education Program. Implications of recent clinical trials of the National Cholesterol Education Program Adult Treatment Panel III Guidelines. Circulation 2004; 100(2): 227-239. 27. -

Reinerpidaemias. Z, Catapano The task AL, force De forBacker the management G, et al. ESC/EAS of dyslipidaemias guidelines for of the the management European Society of dysli of Cardiology (ESC) and the European Atherosclerosis Society (EAS). Eur Heart J 2011; 32(14): 1769-1818. 28. Rutten GEHM, De Grauw WJC, Nijpels G. NHG-Standaard Diabetes mellitus type 2. Huisarts Wet. 2006; 49(3): 137-152.

36 Statins for primary prevention in diabetes

29. Wiersma TJ, Goudswaard AN. NHG-standaard cardiovascular risk management. [In Dutch]

30. Houten:Bates TR, Bohn Connaughton Stafleu van VM, Loghum; Watts GF. 2006. Non-adherence to statin therapy: a major challenge for preventive cardiology. Expert Opin Pharmacother 2009; 10(18): 2973-2985. 31. Donnelly LA, Doney AS, Morris AD, Palmer CN, Donnan PT. Long-term adherence to statin 1 treatment in diabetes. Diab Med 2008; 25(7): 850-855. 32. Olafsdottir E, Aspelund T, Sigurdsson G, et al. Effects of statin medication on mortality risk associated with type 2 diabetes in older patients: the population-based AGES-Reykjavik Study. BMJ Open 2011; 1(1): e000132.

CHAPTER 2

Efficacy of standard and intensive statin treatment for the secondary prevention of cardiovascular and cerebrovascular events in diabetes patients: a meta -analysis

Folgerdiena M. de Vries1 Johan Kolthof1 Maarten J. Postma1 Petra Denig2 Eelko Hak1

PLOS ONE 2014;9(11):e111247.

ABSTRACT

Aims: secondary prevention of major cardiovascular and cerebrovascular events in To estimate the efficacy of standard and intensive statin treatment in the diabetes patients.

Methods: A systematic search was conducted in Medline over the years 1990 to September 2013. Randomized, double-blind, clinical trials comparing a standard- dose statin with placebo or a standard-dose statin with an intensive-dose statin for the secondary prevention of cardiovascular and cerebrovascular events in diabetes patients were selected. Trial and patient characteristics were extracted independently by two researchers. The combined effect on the composite primary

examined with a funnel plot. endpoint was measured with a fixed-effect model. Potential publication bias was

Results: Five trials were included in the analysis comparing standard-dose statins with placebo with a total of 4,351 participants. Four trials were included for comparing standard-dose with intensive-dose statins, including 4,805 participants.

relative risk (RR) reduction of 15% in the occurrence of any major cardiovascular Compared with placebo, standard-dose statin treatment resulted in a significant or cerebrovascular event (RR 0.85, 95% CI 0.79-0.91). Compared with standard- dose statin treatment, intensive-dose statin treatment resulted in an additional 9% relative risk reduction (RR 0.91, 95% CI 0.84-0.98).

Conclusion: Treatment with standard-dose statins to prevent cardiovascular or cerebrovascular events in diabetes patients with manifest cardiovascular disease results in an estimated 15% relative risk reduction and intensive-dose statin treatment adds 9%. If proven cost-effective, more intensive statin treatment should be recommended for diabetes patients at high cardiovascular risk.

40 Standard and intensive dose statins for secondary prevention in diabetes

INTRODUCTION

Cardio- and cerebrovascular diseases are ranked among the major causes of mortality worldwide [1]. Patients with diabetes have a two- to four-fold higher risk of cardiovascular events than age-matched non-diabetes patients [2]. Especially those diabetes patients with a history of cardio- or cerebrovascular disease are at increased risk for recurrent events [3]. Hence, cardiovascular risk management is an essential part of the management of diabetes [4]. 2

The need for statin treatment for secondary prevention of cardiovascular events is widely leading to the conclusion that statin therapy reduces the occurrence of major vascular recognized. Significant benefits of statin treatment were reported in two meta-analyses, events in diabetes patients with and without vascular disease [5, 6]. These analyses did not statin in diabetes patients [7]. Recently Chang et al. [8] performed a meta-analysis and include the ASPEN study, which did not show significant results with a standard-dose in diabetes patients. When focusing on high quality trials for secondary prevention in came to the conclusion that there is still uncertainty regarding the benefits of statins diabetic patients only, being the ASPEN trial and the 4D trial [9] of statin treatment were seen [8]. In these previous meta-analyses several studies with , no significant benefits heterogeneity was seen. Moreover, most trials included in the previous meta-analyses selective subgroups, such as patients on hemodialysis, were included and a significant were limited to interventions with standard-dose statins. Baigent et al. reported, however, that more intensive treatment is associated with even further reductions in the risk for major vascular events in secondary prevention patients [10]. As LDL cholesterol targets have been set at lower levels, targets are not met in at least a third of these patients and more intensive treatment may be needed [11,12]. Randomized clinical trials comparing standard with intensive statin treatment reported partly significant major cardiovascular or cerebrovascular complications within the diabetes subgroup and partly insignificant effects of intensive treatment on reducing the occurrence of [13,14,15,16]. For clinical decision making and cost-effectiveness analysis it is important to have precise effect estimates of standard and intensive statin treatment for secondary prevention diabetes patients.

We aimed to perform two meta-analyses: (1) to assess the effect of standard-dose statins over placebo and (2) to assess the additional effect of intensive-dose statins over standard-dose statins, both for the secondary prevention of major cardiovascular and cerebrovascular events in a non-restrictive diabetes population. Only high-quality and double-blinded studies were eligible for inclusion.

41 CHAPTER 2

METHODS

Search strategy We searched the Medline and Embase databases and ClinicalTrials.gov for randomized clinical trials that either compared statin treatment to placebo or standard to intensive statin treatment for the secondary prevention of cardiovascular and cerebrovascular events (1990 to September 2013) reported in the English-language. Trials were identified with the use of the medical subject heading (MeSH) terms ‘statins’, ‘HMG-CoA reductase inhibitor’, ‘randomized controlled trial’, ‘secondary prevention’, ‘intensive’, Reference lists of reviews and meta-analysis related to the study were examined. The ‘moderate’, ‘diabetes’, ‘coronary heart disease’, ‘myocardial infarction’ and ‘stroke’. search was independently performed by two researchers (FV and JK).

Study selection Inclusion criteria for both meta-analyses were: randomized clinical trials (1); including a non-restrictive secondary prevention diabetes population with previous coronary heart disease, cerebrovascular disease, myocardial infarction or unstable angina pectoris (2); and reporting major cardiovascular and cerebrovascular events as endpoint (3). For a placebo arm, for the second meta-analysis studies were included which compared a the first meta-analysis studies were required to include a standard-dose statin arm and standard-dose with an intensive-dose statin arm. Standard-dose refers to commonly

[17,18,19]. Higher daily doses prescribed daily doses of atorvastatin <=20 mg, simvastatin <=60 mg, rosuvastatin <=10 were categorized as intensive-dose statin treatment. The quality of selected studies mg and any dosing of pravastatin, lovastatin and fluvastatin was scored with the Jadad score [20]. The Jadad score evaluates on a scale from 0-5 the appropriateness of the randomization technique (1), the method for double-blinding (2) and the description of withdrawals and dropouts (3). Two researchers (JK, FV) individually extracted the trial and patient characteristics and the outcome results.

Endpoints The primary endpoint for both meta-analyses was a composite of major cardiovascular and infarction (MI), fatal and non-fatal stroke, revascularization and hospitalization for cerebrovascular events, including the first occurrence of fatal and non-fatal myocardial unstable angina. The meta-analysis comparing standard-dose statins with placebo also assessed the secondary endpoints: fatal and non-fatal MI (a); fatal and non-fatal stroke (b); and all-cause mortality (c).

42 Standard and intensive dose statins for secondary prevention in diabetes

Data analysis and statistical methods

(CI) and the number needed to treat (NNT) with 95% CI were calculated [21]. The results For each endpoint, the relative risk (RR) with corresponding 95% confidence interval effect model in the computer program RevMan from the Cochrane Collaboration [22,23]. of the separate studies were pooled with the fixed-effect model and the random- sampling error whereas the random-effect model assumes that the separate studies are The fixed-effect model assumes that differences in effects between studies are due to measuring different effects. Model selection was based on heterogeneity testing which 2 was assessed by calculating the Q statistic, that tests the homogeneity hypothesis, and the I2 index [24], that calculates the percentage of variability in the effect estimates that is effect model was used for all endpoints. Evidence for potential publication bias was due to heterogeneity rather than chance. Based on the heterogeneity testing the fixed- examined by visually studying funnel plots.

Potentially relevant reports screened for retrieval (N=1,194)

Reports excluded on basis of title or abstract (N=1,173)

Reports retrieved for more detailed evaluation (N=21)

Reports excluded on basis of exclusion criteria (N=12): • not double-blinded (n=3) • no secundary prevention population (n=3) • selective population (n=4) • small number participants (n=2)

Included randomised controlled trials (N=9)

FIGURE 1. Flow diagram of study selection

RESULTS

Description of included randomized clinical trials Data from nine randomized clinical trials were included in the current meta-analyses (Figure 1): the 4S [25], the ASPEN [7], the CARE [26], the HPS [27] and the LIPID [28] trial for the comparison of a standard-dose statin with placebo and the A to Z [13], the PROVE-

43 CHAPTER 2

IT TIMI [14], the SEARCH [15] and the TNT [16] trials for the comparison of standard-dose statin treatment with intensive-dose statin treatment (see Table 1 for full trial names). The study and patient characteristics of the included studies can be found in Table 2 and Table 3

. In total, 4,351 (2,153 standard-dose statin/ 2,198 placebo) participants were included in the standard-dose statin/placebo analysis and 4,805 (2,409 intensive- dose statin/ 2,396 standard-dose statin) participants were included for the comparison placebo analysis participants were treated with pravastatin 40 mg, atorvastatin 10 intensive-dose statin/standard-dose statin treatment. Within the standard-dose statin/ mg or simvastatin in the dosage of 20 or 40 mg daily. Within the analysis comparing standard-dose statin treatment with intensive-dose statin treatment, patients in the standard-dose statin group were treated with simvastatin 20 mg, pravastatin 40 mg or atorvastatin 10 mg and patients in the intensive-dose statin group were treated with simvastatin 80 mg or atorvastatin 80 mg.

The weighted mean follow-up was 5.3 years (ranging from 4.0 in the ASPEN trial [7] and 6.0 in the LIPID trial [28] (ranging from 2.0 years in the A to Z trial [13] and the PROVE-IT TIMI trial [14] to 6.7 years ) in the standard-dose statin/placebo analysis and 4.6 years in the SEARCH trial [15] Participants in both analyses did not differ according to age and gender. ) in the standard-dose statin/intensive-dose statin analysis.

TABLE 1. Full trial name of study acronyms Study acronyms Trial name 4D[9] Die Deutsche Diabetes Dialyse Studie 4S [25] Scandinavian Simvastatin Survival Study ASPEN [7] Atorvastatin Study of Prevention of coronary heart disease Endpoints in Non-insulin-dependent diabetes mellitus A to Z [13] Phase Z of the A to Z trial CARE [26] Cholesterol And Recurrent Events GISSI [30] Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico Prevenzione trial HPS [27] Heart Protection Study of cholesterol-lowering with simvastatin in people with diabetes LIPID [28] Long-Term Intervention with Pravastatin in Ischemic Disease LIPS [31] The Lescol Intervention Prevention Study Post CABG[29] Post Coronary Artery Bypass Graft Trial PROVE IT TIMI [14] The Pravastatin or Atorvastatin Evaluation and Infection Therapy Thrombolysis in Myocardial Infarction trial SEARCH [15] Study of the Effectiveness of Additional Reductions in Cholesterol and Homocysteine TNT [16] Treating to New Targets

44 Standard and intensive dose statins for secondary prevention in diabetes

TABLE 2. Patient and trial characteristics of the included studies Trial Intervention Patients Drugs DM Outcome used for Jadad [20]# primary endpoint 4S,1997 [25] SDS/ Plac MI or AP Sim 20mg T1/T2 CHD death, MI, 5 revascularization, stroke, PVE ASPEN,2006 [7] SDS/ Plac MI or IP Ato 10mg T2 CHD death, 4* MI, stroke, 2 revascularization, UAP CARE,1998 [26] SDS/ Plac MI Pra 40mg T1/T2 CHD death, MI, 5 revascularization HPS,2003 [27] SDS/ Plac CVD Sim 40mg T1/T2 CHD death, 5 MI, stroke, revascularization LIPID,2003 [28] SDS/ Plac MI or Pra 40mg T1/T2 CHD death, 5 UAP MI, stroke, revascularization, UAP A to Z,2004 [13] SDS/ IDS ACS Sim T1/T2 CHD death, MI, 5 20mg/ stroke, AP Sim 80mg PROVE-IT SDS/ IDS ACS Pra 40mg/ T1/T2 CHD death, 4* TIMI,2006 [14] Ato 80mg MI, stroke , AP, revascularization SEARCH,2010 [15] SDS/ IDS MI Sim T1/T2 CHD death, 5 20mg/ MI, stroke, Sim 80mg revascularization TNT,2006 [16] SDS/ IDS CHD Ato 10mg/ T1/T2 CHD death, 4* Ato 80mg MI, stroke, revascularization, AP, CHF # score ranging 1-5; * method of randomization not described ACS: acute coronary syndrome; AP: angina pectoris; Ato: atorvastatin; CHD: coronary heart disease; CVD: cardiovascular disease; CHF: congestive heart failure; DM: diabetes mellitus; IDS: intensive-dose statin; IP: interventional procedures; MI: myocardial infarction; Pra: pravastatin; PVE: peripheral vascular event; SDS: standard-dose statin; Sim: simvastatin; T1: Type 1 diabetes; T2: Type 2 diabetes; UAP: unstable angina pectoris.

45 CHAPTER 2

TABLE 3. Patient characteristics of the included studies Trial No. DM patients Age Men Baseline TC Baseline Follow-up (exp/ con) (yr.) (%) (mmol/l) LDLC (yr.) (mmol/l) 4S,1997 [25] 202 (105/97) 60 72 6.7 4.8 5.4 ASPEN,2006 [7] 505 (252/253) 63 82 4.9 2.9 4.0 CARE,1998 [26] 586 (282/304) 61 80 5.3 3.6 5.0 HPS,2003 [27] 1981 (972/1009) NA NA NA NA 5.0 LIPID,2003 [28] 1077 (542/535) 64 81 5.6 3.7 6.0 A to Z,2004 [13] 1059 (529/530) NA NA NA NA 2 PROVE-IT TIMI,2006 [14] 978 (499/479) 60 72 4.6 2.6 2 SEARCH,2010 [15] 1267 (633/634) NA NA NA NA 6.7 TNT,2006 [16] 1501 (748/753) 63 72 4.5 2.5 5 Con: controls; DM: diabetes mellitus; Exp: experimental; LDLC: low-density lipoprotein cholesterol; NA: not available TC: total cholesterol.

trials included in both meta-analyses (Table 2). Some studies included patients with The definition for secondary prevention patients was slightly different for the individual previous unstable angina or cerebrovascular disease, while others did not. Furthermore, some slight differences in inclusion criteria were made in age, cholesterol and triglyceride included in the composite endpoint of major cardiovascular and cerebrovascular events. levels. Small differences were also present in the definition of endpoints that were

Relative risk of standard-dose statin versus placebo Within a hypothetical high risk population with a 10-year risk for cardiovascular disease of 50%, 27 patients need to be treated for 5 years with a standard-dose statin to prevent one major cardiovascular or cerebrovascular event. Standard-dose statin treatment was or cerebrovascular events (RR 0.85, 95% CI 0.79-0.91) (Figure 2, Figure 3). In the statin associated with a significant relative risk (RR) reduction of 15% for major cardiovascular treatment group, 35.2% of subjects had a major cardiovascular or cerebrovascular event

(763/2,153), whereas 41.4% of those in the placebo group experienced such an event in the individual effect estimates. (914/2,198). There was no observed publication bias and non-significant heterogeneity

In Table 4, participant and event numbers and outcome results for secondary endpoints are presented. For fatal and non-fatal stroke, there was a significant 33% relative risk for fatal and non-fatal MI (RR 0.73, 95% CI 0.53-1.00) and 22% for all-cause mortality reduction (RR 0.67, 95% CI 0.49-0.90). Non-significant relative risk reductions of 27% (RR 0.78, 95% CI 0.53-1.14) were found.

46 Standard and intensive dose statins for secondary prevention in diabetes 2 1.5 Favours (control) Favours 1

Risk Ratio 2 M-H, Fixed (95% CI) 0.7 Favours (experimental) Favours 0.5 Risk Ratio 0.70 (0.53,0.70 0.91) 0.78 (0.62,0.78 0.99) 0.86 0.97) (0.76, 0.89 1.00) (0.79, 0.85 (0.64, 1.12) 0.85 (0.79, 0.91) M-H, Fixed (95% CI) 7.0% 8.6% 41.2% 31.3% 11.9% Weight 100.0% 97 535 253 304 2198 1009 Total Control 61 78 914 381 112 282 Events 972 252 105 542 282 2153 Total Experimental 81 66 46 245 763 325 Events Results of the primary endpoint of major cardiovascular and cerebrovascular events comparing standard-dose statins with placebo statins standard-dose comparing events and cerebrovascular of major cardiovascular of the primary endpoint Results (95% CI) FIGURE 2. 2. FIGURE ASPEN CARE Study or Subgroup 4S HPS LIPID Total Heterogeneity: Chi² df = 0.53); l² = 3.14, = 4 (P = 0% forTest overall effect: Z = 4.33(P < 0.0001) Total events Total

47 CHAPTER 2

Relative risk of intensive-dose statin versus standard-dose statin To prevent one major cardiovascular or cerebrovascular event, 17 patients need to be treated with an intensive-dose statin for 5 years in a hypothetical high risk population with a 10-year risk for cardiovascular disease of 50%. Compared to treatment with standard-dose statins, treatment with intensive-dose statins resulted in an additional 9% relative risk reduction (RR 0.91, 95% CI 0.84-0.98) (Figure 4, Figure 5, Table 4). In the intensive-dose statin treatment group, 31.7% of subjects had a major cardiovascular or cerebrovascular event (764/2,409), whereas 34.9% of those in the standard-dose statin treatment group experienced such an event (837/2,396).

FIGURE 3. Funnel plot of the meta-analysis comparing placebo with standard-dose statin treatment

48 Standard and intensive dose statins for secondary prevention in diabetes 2 1.5 Favours (control) Favours 1 2 Risk Ratio M-H, Fixed (95% CI) 0.7 Favours (experimental) Favours 0.5 Risk Ratio 0.91 (0.75, 1.10) 0.91 (0.75, 0.90 1.15) (0.71, 0.90 (0.80, 1.02) 0.92 (0.80, 1.06) 0.91 (0.84, 0.98) M-H, Fixed (95% CI) 13.1% 39.5% 29.2% 18.3% Weight 100.0% 479 753 530 634 2396 Total Control 110 245 332 837 150 Events 748 529 633 499 2409 Total Experimental 99 142 764 225 298 Events 2006 Results of the primary endpoint of major cardiovascular and cerebrovascular events comparing standard-dose statins with intensive- statins standard-dose comparing events and cerebrovascular of major cardiovascular of the primary endpoint Results 2010 2004 (95% CI) 2006 Z to FIGURE 4. 4. FIGURE dose statins PROVE-IT-TIMI PROVE-IT-TIMI A A SEARCH Total events Total Total Study or Subgroup Heterogeneity: Chi² = 0.04, df = 1.00); = 3 (P l² = 0% forTest overall effect: Z = 2.37(P = 0.02) TNT TNT

49 CHAPTER 2

FIGURE 5. Funnel plot of the meta-analysis comparing standard-dose with intensive- dose statin treatment

TABLE 4. Overall results of the primary and secondary endpoints in the meta-analyses comparing standard-dose statins with placebo and intensive-dose statins with standard- dose statins No. of patients No. of events RR (95% CI) I2 Q NNT (statin/control) (statin/control) (95% CI) Standard-dose vs. placebo: MCCE 2153/2198 763/914 0.85 (0.79-0.91) 0% 3.14 16 (11-30) F/NF stroke 1181/1189 65/98 0.67 (0.49-0.90) 0% 2.0 37 (21-142) F/NF MI 534/557 56/81 0.73 (0.53-1.0) 0% 0.22 25 (13-711) All-cause mortality 357/350 41/51 0.78 (0.53-1.14) 41% 1.70 32 (12-∞) Intensive-dose vs. standard-dose: MCCE 2409/2396 764/837 0.91(0.84-0.98) 0% 0.04 31 (17-180) F: fatal; MCCE: Major cardiovascular and cerebrovascular events; MI: myocardial infarction; NF: non- fatal.

50 Standard and intensive dose statins for secondary prevention in diabetes

DISCUSSION

The results of this study show that treatment with standard-dose statins for the secondary prevention of major cardiovascular or cerebrovascular events in diabetes with an intensive-dose statin instead of a standard-dose statin will reduce the relative patients is associated with a 15% significant relative risk reduction. Treating patients risk for such events with an additional 9%. Within a hypothetical high risk population with a 10-year risk for cardiovascular events of 50% this will reduce the number needed 2 to treat for 5 years to prevent one major cardiovascular and cerebrovascular event from 27 to 17.

Secondary endpoints in the analysis of a standard-dose statin compared to placebo also achieved a significant relative risk reduction of 33% for fatal and non-fatal stroke and all-cause mortality. The number of participants for these analyses were small, however, non-significant relative risk reductions of 27% for fatal and non-fatal MI and 22% for standard-dose statin treatment and intensive-dose statin treatment for secondary leading to wide confidence intervals. Due to lack of data we were not able to compare endpoints in a meta-analysis. The TNT [16] and PROVE-IT TIMI [14] trials, however, did risk reduction of 33% (RR 0.67, 95% CI 0.43-1.04) for stroke. The PROVE-IT TIMI trial report results on some of these endpoints. In the TNT trial there was a non-significant on intensive-dose statin treatment vs 2.2% on standard-dose statin treatment. Also, reported a non-significant difference in stroke event rate of 2.6% for diabetes patients were observed [14,16]. regarding event rates for MI and all-cause mortality, only non-significant differences

Previous meta-analyses comparing standard-dose statin treatment with placebo for secondary prevention in diabetes patients reported risk reductions of 20% [5] and 21% [6] [8], whereas our estimate was 15%. These differences can be explained by differences in study inclusion criteria. The estimates of as well as a non-significant reduction of 11% 20% and 21% included non-blinded studies, such as the Post-CABG [29] in the analysis of Costa et al. [6] and the GISSI [30] in the analysis of Kearney et al. [5]. Furthermore, Chang et al. included studies in restrictive subpopulations, such as patients with heart failure or on hemodialysis [8]. Analyses of Chang et al. showed that including these studies resulted in higher risk reductions than when including only double-blinded studies in diabetes patients [8]. All previous analyses included the LIPS [31] which was conducted in patients after a successful percutaneous coronary intervention. We did not include non-blinded studies nor studies in restrictive subpopulations.

51 CHAPTER 2

On the other hand, we did include the ASPEN trial that was not included by Kearney et al. [5] and Costa et al. [6] [7] but . The non-significant results from the ASPEN trial caused conflicting contributed to our risk reduction estimate of 15%. evidence from the randomized clinical trials regarding the efficacy of statins

In addition to the previous meta-analyses comparing standard-dose statins with placebo, we conducted a meta-analysis comparing standard-dose with intensive-dose statins including all diabetes patients from relevant studies. Through using similar inclusion criteria for both comparisons, we were able to determine the estimate of the overall effect of statins in a general secondary prevention population with diabetes and the effect of intensive-dose over standard-dose statins.

secondary prevention in the separate studies, which could have resulted in differences There are some limitations to our study. There were small differences in the definition for in the baseline risk of the populations. Also, the diagnostic criteria for diabetes differed diabetes, CARE [26] interviewed the patients and asked whether they had diabetes. Most among the studies. Where most used the World Health Organization definition for included type 1 and type 2 diabetes patients. Furthermore, the events included in the primary endpoint were not exactly similar. The differences especially concern the inclusion of unstable angina, congestive heart failure or peripheral vascular events in addition to CHD death, MI and stroke in some studies. Including additional events in a composite endpoint may lead to larger or smaller risk reductions depending on the effect of statins on such events. The larger risk reductions seen for our secondary endpoints MI and stroke may suggest that we have underestimated the overall risk reduction for the major events. Furthermore, not all patients who were prescribed statins were still taking them at the end of follow-up, and some patients in the placebo group may also receive statins during follow-up, which will lead to lower risk reductions. In the analysis comparing standard-dose with intensive-dose statin treatment, patients could also have changed treatment intensity over time. For the ASPEN [7], HPS [27], TNT [16], PROVE-IT TIMI [14], A to Z [13] and SEARCH [15] trials, we used subgroup results which can cause bias. However, these trials used minimization randomization techniques we do not expect that this will have a high impact on the meta-analysis results.

Cardiovascular risk management is an important part of diabetes treatment. The Dutch and European guidelines [32,33], for example, recommend statin treatment in almost all diabetes patients, and especially in those patients with a history of cardiovascular disease. LDL-cholesterol targets are lowered for these high risk patients, therefore

[11,12,34] reaching targets is more difficult and intensive-dose statin treatment may be needed . Our analysis shows that intensive-dose statin treatment has a significant effect 52 Standard and intensive dose statins for secondary prevention in diabetes

in reducing major cardiovascular and cerebrovascular events compared to standard- dose statin treatment for the secondary prevention in diabetes patients. We should individual endpoints. While the 2012 ESC guidelines just recommend statin treatment acknowledge, however, that these reductions are not confirmed yet by analyzing [33] secondary prevention patients [35]. In contrast, the current Dutch guidelines recommend , the recent ACC/AHA guideline recommends intensive-dose statin treatment for all to start with a standard-dose statin, such as simvastatin 40 mg. This choice is largely driven by economic considerations [32]. Now that the patent for atorvastatin has expired, 2 however, the cost-effectiveness needs to be reassessed.

Besides the reductions in cardiovascular events, there are adverse events associated with statins, among which muscle toxicity and effects on liver enzymes are well acknowledged [36]. With higher doses there is an increased risk for statin-induced adverse events, especially regarding their effect on liver enzymes [36,37,38]. The risk of myopathy remains low at high doses, with an estimated incidence of 0.5% for simvastatin 80 mg and 1.5% for atorvastatin 80 mg [37,38]. The incidence of rhabdomyolysis was found to be too low

[36,37,38]. to detect significant differences between standard and intensive dose statin treatment form a more selective population than patients included in observational studies. A few The efficacy of statins as shown here is based on clinical trials in which patients usually observational studies have been conducted in secondary prevention patients. These repeat hospitalization for acute coronary syndrome and for mortality were reported in studies show similar results as found in the clinical trials. Significant risk reductions for observational studies comparing standard-dose with intensive-dose statins [39,40].

Translation of trial evidence into guideline recommendations and of guideline recommendations into practice is subject to interpretation of the evidence. Meta- analyses can support and strengthen this process. Some guidelines now recommend intensive-dose statin treatment for all secondary prevention patients, whereas others recommend to start statin treatment on a standard-dose and to switch to an intensive- dose if LDL cholesterol targets are not reached. Our meta-analyses show that standard- dose statin treatment is associated with a 15% relative risk reduction of cardiovascular and cerebrovascular events and that the use of intensive-dose statins results in a reduction of 9% compared to standard-dose statin treatment in a secondary prevention diabetes population. These estimates are useful for further cost-effectiveness analyses. If proven cost-effective, a more differentiated advise can be given, where more intensive treatment is recommended for diabetes patients at high cardiovascular risk.

53 CHAPTER 2

REFERENCES

World Health organisation. Global Health Observatory 2012. http://www.who.int/gho/ 2. mortality_burden_disease/en/American Heart Association. Cardiovascular Accessed 8 September disease 2014and diabetes [online]. Available from -

http://www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetesMatters/Cardiovas 3. cular-Disease-Diabetes_UCM_313865_Article.jspNational Cholesterol Education Program (NCEP) Accessed Expert Panel 8 September on Detection, 2014 Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Circu-

Treatmentlation 2002; of 106(25): High Blood 3143-3421. Cholesterol in Adults (Adult Treatment Panel III) final report. 4. Buse JB, Ginsberg HN, Bakris GL, et al. Primary prevention of cardiovascular diseases in people

withAmerican diabetes Diabetes mellitus: Association. a scientific Diabetes statement Care 2007;from the30(1): American 162-172. Heart Association and the 5.

Costadiabetic J, Borgesand non-diabetic M, David patients: C, Vaz Carneiro meta-analysis A. Efficacy of randomised of lipid lowering controlled drug trials. treatment BMJ 2006; for 332(7550): 1115-1124. 7.

Knoppprevention RH, of d’Emden cardiovascular M, Smilde end JG,points Pocock in subjects SJ. Efficacy with andtype safety 2 diabetes: of atorvastatin the Atorvastatin in the Study for Prevention of Coronary Heart Disease Endpoints in non-insulin-dependent diabetes mellitus (ASPEN). Diabetes Care 2006; 29(7): 1478-1485. 8.

Changprevention YH, Hsiehof cardiovascular MC, Wang CY,disease Lin KC,in diabetic Lee YJ. patients Reassessing - A thesystematic benefits review of statins and inmeta- the analysis. Rev Diabet Stud 2013; 10(2-3): 157-170. 9. Warner C, Krane V, Marz W, et al. Dialysis Study I. Atorvastatin in patients with type 2 diabetes mellitus undergoing hemodialysis. N Engl J Med 2005; 353(3): 238-248. 10.

Baigentcholesterol: C, Blackwell a meta-analysis L, Emberson of data J, etfrom al. Efficacy170 000 and participants safety of morein 26 intensiverandomised lowering trials. Lancetof LDL 2010; 376(9753): 1670-1681. 11. Heintjes EM, Penning-van Beest FJA, Plat AW, et al. Cholesterol level goal attainment with statins: clinical management guideline recommendations versus management in actual clinical practice. Pharmacotherapy 2012; 32(7): 631-641. 12. Sidorenkov G, Haaijer-Ruskamp FM, de Zeeuw D, Denig P. A longitudinal study examining

adherencetimeliness. toPLoS guidelines One 2011; in 6(9):diabetes e24278. care according to different definitions of adequacy and

54 Standard and intensive dose statins for secondary prevention in diabetes

13. De Lemos JA, Blazing MA, Wiviott SD, et al. Early intensive vs a delayed conservative simvastatin strategy in patients with acute coronary syndromes: phase Z of the A to Z trial. JAMA 2004; 292(11): 1307-1316. 14. Ahmed S, Cannon CP, Murphy SA, Braunwald E. Acute coronary syndromes and diabetes: is Eur Heart J 2006;

intensive27(19): 2323-2329. lipid lowering beneficial? Results of the PROVE IT-TIMI 22 trial. 15. Armitage J, Bowman L, Wallendszus K, et al. Intensive lowering of LDL cholesterol with 80 mg versus 20 mg simvastatin daily in 12,064 survivors of myocardial infarction: a double-blind 2 randomised trial. Lancet 2010; 376(9753): 1658-1669. 16. Shepherd J, Barter P, Carmena R, et al. Effect of lowering LDL cholesterol substantially below currently recommended levels in patients with coronary heart disease and diabetes. The treating to new targets (TNT) study. Diabetes Care 2006; 29(6): 1220-1226. 17. Law MR, Wald NJ, Rudnicka AR. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: systematic review and meta-analysis. BMJ 2003; 326(7404): 1423. 18. Smith MEB, Lee NJ, Haney E, Carson S (2009) Drug class review: Hmg-coa reductase inhibitors (statins) and fixed-dose combination products containing a statin: Final report update 5. Portland (OR). 19. Weng TC, Yang YH, Lin SJ, Tai SH. A systematic review and meta-analysis on the therapeutic equivalence of statins. J Clin Pharm Ther 2010; 35(2): 139-151. 20. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996; 17(1): 1-12. 21. Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment. BMJ 1995; 310(6977): 452-454. 22. Review Manager (RevMan) [Computer program]. Version 5.1. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011. 23. Scholten RJ, Kostense PJ, Assendelften WJ, Bouter LM. De praktijk van systematische reviews. IV. Het combineren van de resultaten van afzonderlijke onderzoeken. [In Dutch] Ned Tijdschr Geneeskd 1999; 143(15): 786-791 . 24. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327(7414): 557-560. 25.

Pyŏrälälowering K, with Pedersen simvastatin TR, Kjekshus improves J, Faergemanprognosis O,of Olssondiabetic AG, patients Thorgeirsson with coronary G. Cholesterol heart disease. A subgroup analysis of the Scandinavian Simvastatin Survival Study (4S). Diabetes Care 1997; 20(4): 614-620. 26. Goldberg RB, Mellies MJ, Sacks FM, et al. Cardiovascular events and their reduction with pravastatin in diabetic and glucose-intolerant myocardial infarction survivors with average cholesterol levels: subgroup analyses in the cholesterol and recurrent events (CARE) trial. The Care Investigators. Circulation 1998; 98(23): 2513-2519.

55 CHAPTER 2

27.

Collinslowering R, Armitagewith simvastatin J, Parish inS, Sleigh5963 P,people Peto R. with MRC/BHF diabetes: Heart a randomised Protection Study placebo-controlled of cholesterol- trial. Lancet 2003; 361(9374): 2005-2016. 28. Keech A, Colquhoun D, Best J, et al. Secondary prevention of cardiovascular events with long-term pravastatin in patients with diabetes or impaired fasting glucose: results from the LIPID trial. Diabetes Care 2003; 26(10): 2713-2721. 29. Post Coronary Artery Bypass Graft Trial Investigators. The effect of aggressive lowering of low-density lipoprotein cholesterol levels and low-dose anticoagulation on obstructive changes in saphenous-vein coronary-artery bypass grafts. N Engl J Med 1997; 336(3): 153-162. 30. GISSI Prevenzione Investigators (Gruppo Italiano per lo Studio della Sopravvivenza nell’In- farto Miocardico). Results of the low-dose (20 mg) pravastatin GISSI Prevenzione trial in 4271 patients with recent myocardial infarction: do stopped trials contribute to overall knowledge? Ital Heart J 2000; 1(12): 810-820. 31. Serruys PW, de Feyter R, Macaya C, et al. Fluvastatin for prevention of cardiac events following JAMA

succesfull2002; 287(24): first 3215-3222. percutaneous coronary intervention: a randomized controlled trial. 32. Cardiovasculair risicomanagement (Tweede herziening) [In Dutch] Huisarts Wet 2012; 55(1): 14-28. 33. Perk J, De Backer G, Gohlke H, et al. European Guidelines on cardiovascular disease prevention in clinical practice (version 2012). The Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). Eur Heart J 2012; 33(13): 1635-1701. 34. Stark Casagrande S, Fradkin JE, Saydah SH, Rust KF, Cowie CC. The prevalence of meeting A1C, blood pressure, and LDL goals among people with diabetes, 1988-2010. Diabetes Care 2013; 36(8): 227-229. 35.

Stonecholesterol NJ, Robinson to reduce J, atheroscleroticLichtenstein AH, cardiovascular et al. ACC/AHA risk guideline in adults: on a thereport treatment of the American of blood Circu-

Collegelation 2014; of Cardiology/American 129(25 Suppl 2): S1-S45. Heart Association Task Force on Practice Guidelines. 36. Armitage J. The safety of statins in clinical practice. Lancet 2007; 370(9601) :1781-1790. 37. Silva M, Matthews ML, Jarvis C, et al. Meta-analysis of drug-induced adverse events associated with intensive-dose statin therapy. Clin Ther 2007; 29(2): 253-260. 38. Newman C, Tsai J, Szarek M, Luo D, Gibson E. Comparative safety of atorvastatin 80 mg versus 10 mg derived from analysis of 49 completed trials in 14,236 patients. Am J Cardiol 2006; 97(1): 61-67. 39. Ko DT, Wijeysundera HC, Jackevicius CA, Yousef A, Wang J, Tu JV. Diabetes mellitus and cardiovascular events in older patients with myocardial infarction prescribed intensive-dose and moderate-dose statins. Circ Cardiovasc Qual Outcomes 2013; 6(3): 315-322.

56 Standard and intensive dose statins for secondary prevention in diabetes

40. Fintel D, Joyce A, Mackell J, Graff J, Kuntze E, Ollendorf DA. Reduced mortality rates after intensive statin therapy in managed-care patients. Value Health 2007; 10(2): 161-169.

CHAPTER 3

Adherence to standard-dose or low-dose statin treatment and LDL-cholesterol response in type 2 diabetes patients

Folgerdiena M. de Vries1 Jaco Voorham1 Eelko Hak2 Petra Denig1

Current Medical Research and Opinion 2015

1. Department of Clinical Pharmacy and Pharmacology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands 2. University Groningen, Department of Pharmacy, Unit of PharmacoEpidemiology & PharmacoEconomics, Groningen, the Netherlands CHAPTER 3

ABSTRACT

Objective: To determine the association between adherence, treatment dose and LDL-cholesterol response in patients with type 2 diabetes initiating statin treatment.

Research design and methods: This cohort study was performed using data for 2007-2012 from the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database. The association between adherence to a standard-dose statin and LDL-cholesterol response was assessed using linear regression, adjusting for covariates. The effect of low-dose versus standard-dose was assessed in a

Days Covered (PDC), were estimated between statin initiation and LDL-outcome propensity-score matched cohort. Adherence rates, defined as the Proportion of measurement.

Main outcome measure: LDL-cholesterol level at follow-up.

Results: The effect of adherence on LDL-cholesterol response, measured in 2,160 patients, was dependent on the baseline LDL-cholesterol level. For patients with

reduction in LDL-cholesterol was predicted. In the matched sample of 1,144 a baseline LDL-cholesterol of 3.7 mmol/l and an adherence rate of 80%, a 40% patients, the treatment dose showed a difference in impact on the outcome for adherence rates higher than 50%. It was estimated that a patient with a baseline

dose and 63% on standard-dose treatment to reach the LDL-cholesterol target of LDL-cholesterol of 3.7 mmol/l will need an adherence rate of at least 76% on low-

2.5 mmol/l. Limitations: Adherence was measured as the PDC, which is known to overestimate actual adherence. Also, we were not able to adjust for lifestyle factors.

Conclusions: We determined the concurrent effect of treatment adherence and treatment dose on LDL-cholesterol outcomes. Given the adherence levels seen in clinical practice, diabetes patients initiating statin treatment are at high risk of not reaching the recommended cholesterol target, especially when they start on a low- dose statin.

60 Statin treatment and LDL-cholesterol response

INTRODUCTION

Patients with type 2 diabetes are at high risk for developing cardiovascular disease (CVD), and the use of statins is associated with a reduction in the risk of cardiovascular events [1,2]. Clinical guidelines recommend statin treatment for almost all diabetes patients unless they have a very low cardiovascular risk [3,4,5]. Dutch guidelines recommend to start with a standard-dose statin for both primary and secondary prevention, and aim cost-effective strategy [3]. It is expected that not all patients will achieve the lipid targets for an LDL-cholesterol level of <2.5 mmol/l (<97 mg/dl), which is considered the most on standard-dose treatment, depending on their LDL-cholesterol level at initiation, 3 in which case they should switch to a high-dose statin. In case of low LDL-cholesterol levels, one may start with a low-dose statin [3].

Observational studies show that lipid targets are not met in at least a third of the patients [6,7] [6,8] or non-adherence to treatment [9-12]. In the last decade, the prescribed daily dose of statins . Insufficient treatment response could be due to prescribing low-dose treatment has increased but on average patients receive less than the recommended standard dose [8,13]. Adherence to statins has also improved but remains sub-optimal with up to 30% of [14]. It is not clear, however, what the concurrent impact of statin dose and non-adherence is on treatment response patients being non-adherent in the first year after initiation in clinical practice.

Several studies assessed the association between non-adherence and LDL-cholesterol response in statin initiators but most used arbitrary cut-off points for adherence [9-12] clinically relevant decrease in treatment response. Moreover, few studies differentiated . This makes it difficult to determine which level of non-adherence may lead to a between the choice or dose of statins [9,11]. One study observed LDL-cholesterol outcomes levels increasing from 20% to 100%, and also observed differences between statins, not decreasing from 3.6 mmol/l (140 mg/dl) to 2.4 mmol/l (95 mg/dl) with adherence taking account of dose [9]. Another study showed that high risk patients on low-dose statins were less likely to achieve the LDL-cholesterol target across several adherence categories [11]. For clinical practice it is important to gain better insight in the impact of the level of non-adherence in relation to the dose of statin on treatment response.

The aim of this study is to determine the association between statin adherence and LDL-cholesterol response in type 2 diabetes patients initiating on the recommended standard-dose statin treatment, and to assess the effect of low-dose versus standard- dose on this association.

61 CHAPTER 3

METHODS

Study design A retrospective cohort study was performed in patients with type 2 diabetes initiating lipid-lowering treatment between January 2007 and December 2012. First, the association between adherence to statin treatment and LDL-cholesterol response was determined in patients on standard-dose statin treatment. Second, the effect of low- dose versus standard-dose statin treatment on this association was determined in a propensity-score matched cohort.

Setting This study was performed in the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database. The GIANTT database contains anonymized longitudinal information retrieved from electronic medical records of general practitioners and is maintained by the University Medical Center Groningen [15]. These records include medical history, prescription data, routine laboratory test results and physical examinations of type 2 diabetes patients from the northern part of the Netherlands that are managed in primary care. Medical history consists of date of diabetes diagnosis and comorbidity [16] or text descriptions that are coded manually. data, which is based on the International Classification of Primary Care (ICPC)

Patient selection Patients managed in general practice for type 2 diabetes initiating statin treatment were included (Anatomical Therapeutic Chemical (ATC) code C10AA) [17]. Since the documented date of diagnosis is not always exact in our database, we allowed for a grace period of up to 180 days before the documented diagnosis in which the statin lowering medication (ATC code C10) in the preceding 360 days. Patients needed to have might be initiated. Statin initiation was defined as having no prescription for any lipid

270 days. Patients with temporary absence from the database, for example due to being sufficient medical history to be classified as initiators, and a follow-up period of at least included as initiators. institutionalized, as identified by long-term gaps in all prescribed medication, were not

Patients were included in the standard-dose treatment group when they were prescribed simvastatin 40 mg, atorvastatin 20 mg, or rosuvastatin 5 or 10 mg from statin initiation till the LDL-cholesterol measurement at follow-up. These doses are recommended in average by 37% to 43% [3,18,19]. Patients were included in the low-dose group when they the first steps of the Dutch guideline and expected to reduce the LDL-cholesterol on were prescribed pravastatin 40 mg, simvastatin 20 mg or atorvastatin 10 mg from statin

62 Statin treatment and LDL-cholesterol response

initiation till the LDL-cholesterol measurement at follow-up. Such low-dose treatment is expected to reduce the LDL-cholesterol on average by 32% to 38% [3,18,19].

For assessing the LDL-cholesterol response, which was calculated using the Friedewald equation, patients without an LDL-cholesterol measurement at baseline or follow-up, calculations [20], or who used co-medication that interacts with statin treatment were or with triglyceride levels >4.5 mmol/l which may result in unreliable LDL-cholesterol excluded. We included as interacting medication [21]: (1) medication causing an increase of the statin concentration (amiodaron, amlodipine, diltiazem, verapamil, ciclosporin, imatinib, ticagrelor, danazol, and colchicine), and (2) medication causing a decrease 3 of the statin concentration (bosentan, carbamazepin, phenobarbital, phenytoin, and primidon).

Adherence measurement Adherence was estimated over a period from statin initiation till the LDL-cholesterol measurement at follow-up. It was calculated as the Proportion of Days Covered (PDC), which expresses the proportion of days for which a patient has received medication in the study period [22]. Patients receiving treatment in daily packages or with single prescription durations longer than 270 days or with missing prescription data were excluded, since the PDC cannot be reliably calculated in such cases.

LDL-cholesterol response days and not more than 540 days after statin initiation. For patients with multiple The primary outcome measure was the LDL-cholesterol level (mmol/l) at least 270 LDL-cholesterol measurements in this period, the one nearest to 360 days after statin initiation was selected. As secondary outcome, percentages of patients achieving the

LDL-cholesterol target of 2.5 mmol/l were estimated. Confounders The association between statin adherence and LDL-cholesterol response may be confounded by patient characteristics. In addition to baseline LDL-cholesterol level, the following baseline characteristics possibly related to cardiometabolic treatment were considered: (1) age, gender, and diabetes duration; (2) HDL-cholesterol, total- cholesterol, triglycerides, systolic blood pressure (SBP), diastolic blood pressure (DBP), fasting glucose, HbA1c, creatinine, potassium, hemoglobin, albuminuria, body mass index (BMI) and smoking status; (3) comorbidity, including lipid disorders (ICPC code

T93), high blood pressure (K85/K86/K87), stroke/cerebrovascular accident (K89), transient ischemic attack (K90), atherosclerosis (K91), myocardial infarction/ischemic heart disease (K75/K76), angina pectoris (K74), heart failure (K77), bypass/angioplasty 63 CHAPTER 3

(including: coronary artery bypass grafting (CABG), percutaneous transluminal coronary angioplasty (PTCA), peripheral bypass and percutaneous transluminal angioplasty neuropathy (N94.2), diabetic foot, retinopathy (F83) and blindness. (PTA)) and microvascular complications including: proteinuria (U98/U90), diabetic

Statistical analyses Descriptive statistics are presented for patients on low-dose and standard-dose treatment who persisted on the same dose up to the LDL-cholesterol measurement at follow-up, and they were compared on baseline characteristics and adherence rates using chi-square tests, independent sample t-tests and Wilcoxon rank-sum tests.

To determine the association between adherence and LDL-cholesterol response in model adjusted for baseline LDL-cholesterol and possible confounders. Confounders patients on standard-dose statin treatment, we first used a multivariate linear regression were selected via forward selection and included with a p-value of 0.05. Missing data on SBP, DBP, HbA1c, fasting glucose, HDL-cholesterol, triglycerides, total-cholesterol, creatinine, potassium, hemoglobin, albuminuria, and BMI were imputed using multiple imputation. Patients who had no data on SBP and DBP and HbA1c and BMI were excluded from this analysis, since imputation was considered unreliable for these patients.

Next, to determine the effect of low-dose versus standard-dose statin treatment on the association between adherence and LDL-cholesterol response a propensity-score matched cohort was formed [23,24]. Since patients initiating low-dose treatment may differ from patients initiating standard-dose treatment, confounding by indication can be expected. Propensity-score matching is a method that is recommended to balance treatment groups and reduce indication bias [25]. Within the matched sample, the association between adherence, baseline LDL-cholesterol and treatment dose (low-dose versus standard-dose) and the interaction terms adherence/baseline LDL-cholesterol model adjusted for baseline LDL-cholesterol. and adherence/treatment dose were determined using a multivariate linear regression

The propensity-scores were calculated using logistic regression based on all available demographic and clinical characteristics and comorbidity data (see above, section Confounders). Patients on low-dose and standard-dose treatment were matched 1:1 on the propensity score using a nearest neighbor matching algorithm with a caliper width equal to 0.01 [23,24]. The standardized differences of characteristics between the low-dose and standard-dose group were calculated, and variables with a standardized [24]. We adjusted variable selection difference of ≤0.10 were assumed to be balanced

64 Statin treatment and LDL-cholesterol response

balanced in the matched cohort. of the propensity-score generating model until all baseline variables were sufficiently homoscedasticity of residuals and collinearity of the model. This resulted in log Robustness of the regression models was verified by examining normality and transforming the outcome variable and centering the variables adherence and LDL- cholesterol at baseline. The analyses were performed with Stata version 13 (Stata Corp., College Station, TX). Sensitivity analysis 3 To examine whether the period of follow-up may affect the results, we conducted two additional analyses with more restricted follow-up periods of 270 to 450 days (around after statin initiation. 1 year plus/minus 90 days) and 315 to 405 days (around 1 year plus/minus 45 days)

RESULTS

In total 6,046 of the type 2 diabetes patients initiating statin treatment between 2007 and 2012 received a standard dose statin, and 1,525 patients received a low dose statin (Figure 1). More females (chi-square test; p-value <0.001), older patients, with a longer diabetes duration and lower glucose levels (t-tests; during the first year of treatment p-value <0.001) were on low-dose treatment compared to standard-dose treatment (Table 1) for patients initiating low-dose and standard-dose treatment respectively (t-test; p-value . The average baseline LDL-cholesterol levels were 3.7 mmol/l and 3.8 mmol/l 0.029). More patients were above the LDL-cholesterol target level at baseline in the low- dose as compared to the standard-dose treatment group (66% vs. 62%; chi-square test: p-value 0.007). In the standard-dose group more patients had missing LDL-cholesterol measurement at baseline (29% vs. 32%; chi-square test: p-value 0.016). There were small differences in history of angina pectoris and myocardial infarction or ischemic heart disease between the groups (Table 1).

65 CHAPTER 3

Exclusion criteria Base population 11,994

Treated by a specialist 235

Started >180 days before DM diagnosis 906

No history 73

Follow-up <270 days 764

Not remaining on same dose category till follow-up 2,445

7,571

Standard-dose Low-dose

6,046 1,525

Patients using daily packages 134 32

Triglycerides >4,5mmol/l 343 56

Missing LDL cholesterol measurements 2,740 683

Comedication use influencing statin concentration 381 87

Prescription duration >270 days 25 5

Missing data on SBP, DBP, Hba1c and BMI 21 3

Adherence calculation impossible 242 80

Population with adherence and LDL measurements

2,160 579

BMI: body mass index; DBP: diastolic blood pressure; DM: Diabetes Mellitus; LDL-cholesterol: low- density lipoprotein cholesterol; SBP: systolic blood pressure. FIGURE 1. Selection of standard- and low-dose patients

66 Statin treatment and LDL-cholesterol response

TABLE 1. Characteristics of baseline populations on low-dose and standard-dose statin treatment. Variable Low-dose (n=1,525) Standard-dose (n=6,046) p-value

Age, mean (SD) 64.6 (11.6) 62.3 (12.1) <0.001

Male sex 46.5% 52.4% <0.001

SBP, mmHg (SD) 143.7 (19.3) 144.5 (20.1) 0.171

DBP, mmHg (SD) 82.0 (10.5) 82.7 (10.6) 0.036

Diabetes duration (IQR) 2 (0-6) 1 (0-5) <0.001 Fasting glucose, mmol/l 7.9 (2.4) 8.2 (2.9) <0.001 3 HbA1c (%) 7.0 (1.2) 7.3 (1.6) <0.001

LDL cholesterol, mmol/l 3.68 (0.8) 3.75 (1.0) 0.029

HDL cholesterol, mmol/l 1.2 (0.3) 1.2 (0.3) <0.001

Triglycerides, mmol/l 2.1 (1.8) 2.3 (1.7) <0.001

Total cholesterol, mmol/l 5.6 (1.0) 5.7 (1.2) <0.001

Creatinine, μmol/l 82.6 (22.5) 82.6 (21.6) 0.993

Potassium, mmol/l 4.3 (0.42) 4.2 (0.42) 0.053

Hemoglobin, mmol/l 8.9 (0.9) 9.0 (0.9) 0.022

Body mass index 30.6 (5.6) 30.6 (5.8) 0.997

Smoking 22.9% 28.5% 0.006

Comorbidity:

Hypertension 37.4% 36.2% 0.396

Angina pectoris 4.7% 3.4% 0.020

Myocardial infarction/IHD 5.1% 4.1% 0.071

Heart failure 1.7% 2.3% 0.189

Bypass/angioplasty 2.6% 2.6% 0.925

Lipid disorder 7.8% 7.5% 0.682

Stroke/CVA/TIA/ atherosclerosis 4.6% 4.5% 0.856

Microvascular complications 4.1% 3.6% 0.301

CVA: cerebrovascular accident; DBP: diastolic blood pressure; HDL: high-density lipoprotein; IHD: ischemic heart disease; IQR: interquartile range; LDL: low-density lipoprotein; SBP: systolic blood pressure; SD: standard deviation; TIA: transient ischemic attack.

67 CHAPTER 3

Less patients in the low-dose treatment group reached the target in comparison to the standard-dose group (Table 2). The median adherence was 85% in the low-dose group (interquartile range (IQR): 47%-99%) and 89% in the standard-dose group (IQR: 55%- 99%) (Wilcoxon rank-sum test; p-value 0.014) (Figure 2).

TABLE 2. LDL-cholesterol outcomes at follow-up LDL-cholesterol at follow-up: Low-dose Standard-dose p-value LDL-cholesterol missing, n (%) 494 (32.4) 1,951 (32.3) 0.926 Mean LDL-cholesterol level, mmol/l (SD) 2.63 (0.87) 2.44 (0.95) <0.001 LDL-cholesterol at target, n (%) 566 (37.1) 2,627 (43.5) <0.001 LDL-cholesterol above target, n (%) 465 (30.5) 1,468 (24.3) <0.001 SD: standard deviation.

FIGURE 2. Boxplots showing the Proportion of Days Covered (PDC) till the follow-up LDL- cholesterol measurement for patients on standard-dose and low-dose treatment

68 Statin treatment and LDL-cholesterol response

Adherence and LDL-cholesterol response A total of 2,160 patients were included in the standard-dose treatment group for analyzing the association between adherence and LDL-cholesterol response (Figure 1). Most patients were excluded because of missing LDL-cholesterol measurements. Excluded patients had a slightly longer duration of diabetes (Appendix I; Supplemental Table S1). The multivariate linear model included adherence, LDL-cholesterol at baseline, age, lipid disorder, fasting glucose, hypertension and diabetic neuropathy (Table below Figure 3) at follow-up, with every 10% increase in adherence resulting in an expected decrease . Adherence was significantly associated with LDL-cholesterol 3 in LDL-cholesterol of 0.068 mmol/l (adjusted coefficient -0.0068, p-value <0.001). For was associated with a 48% reduction in LDL-cholesterol; an adherence rate of 80% was patients with a baseline LDL-cholesterol value of 3.7 mmol/l, an adherence rate of 100% associated with a reduction of 40%; an adherence rate of 60% was associated with a reduction of 31% in LDL-cholesterol. In this range, a 10% decrease in adherence was associated with 4% less reduction in LDL-cholesterol.

Sensitivity analysis Restricting the follow-up period to 270-450 days resulted in the inclusion of 1,595 patients; with a follow-up period of 315-405 days only 941 patients could be included. Using these restrictions the observed associations between adherence and LDL- cholesterol response were similar (Appendix I; Supplemental Figure S1 and S2).

Low-dose versus standard-dose treatment A propensity-score matched sample of 1,144 patients on low-dose and on standard- dose was created, for which the baseline characteristics were balanced (Appendix I; Supplemental Table S2). The multivariate linear model included adherence, LDL- cholesterol at baseline, low-dose treatment, and the interaction terms adherence/LDL- cholesterol at baseline and adherence/low-dose treatment. The average LDL-cholesterol (Table below level of these patients at baseline was 3.7 mmol/l. Being on low-dose treatment was Figure 4). Again adherence was found to be associated with LDL-cholesterol at follow- significantly associated with less LDL-cholesterol response at follow-up up (coefficient -0.0067, p-value <0.001). This association was however dependent on showing a stronger effect of adherence on LDL-cholesterol at follow-up in patients LDL-cholesterol at baseline (coefficient -0.0009 for the interaction term, p-value 0.013), with higher baseline LDL-cholesterol levels. Also the treatment dose affected the the interaction term, p-value 0.010), showing a stronger effect of adherence on LDL- relation between adherence and LDL-cholesterol at follow-up (coefficient -0.0016 for cholesterol for patients on low-dose treatment.

69 CHAPTER 3

Variable Coeffcient (ß) Standard error p-value Adherence (PDC) (%) -0.0068 0.00025 <0.001 LDL-cholesterol at baseline (mmol/l) 0.1594 0.00660 <0.001 Age (years) -0.0026 0.00055 <0.001

Lipid disorder (yes/no) 0.0808 0.02464 0.001

Fasting glucose (mmol/l) -0.0107 0.00247 <0.001

Hypertension (yes/no) -0.0266 0.01323 0.044

Diabetic neuropathy (yes/no) -0.2636 0.12583 0.036

Constant 1.001 0.05218 <0.001 R² = 0.384, p-value = <0.001 LDL-c, LDL-cholesterol; PDC, proportion of days covered

FIGURE 3. Predicted LDL-c response in patients (n=2,160) on standard-dose treatment for different adherence rates using a multivariate linear model

70 Statin treatment and LDL-cholesterol response

Variable Coeffcient (ß) Standard error p-value Low-dose treatment 0.0923 0.0159 <0.001 LDL-cholesterol at baseline 0.1668 0.0090 <0.001

Adherence (PDC) -0.0067 0.0045 <0.001

Adherence (PDC) x LDL-cholesterol at -0.0009 0.0004 0.013 baseline Adherence (PDC) x low-dose 0.0016 0.0006 0.010 treatment Constant 0.8154 0.0112 <0.001 R² = 0.385, p = <0.001 CI, Confidence interval; LDL-c, LDL-cholesterol; PDC, proportion of days covered

FIGURE 4. Predicted LDL-cholesterol response for low-dose and standard-dose statin treatment in a propensity-score matched sample (n=1,144). Associations are presented for a LDL-cholesterol baseline level of 3.7 mmol/l.

71 CHAPTER 3

Following from the model presented in Figure 4, an adherence rate of 80% in patients associated with a 34% reduction in LDL-cholesterol compared to 40% reduction for with a baseline LDL-cholesterol of 3.7 mmol/l treated with low-dose statin was patients treated with standard-dose statin treatment. For adherence rates higher than least 10% more adherent than patients on standard-dose treatment to reach the same 50% and a baseline LDL-cholesterol of 3.7 mmol/l, patients on low-dose need to be at LDL-cholesterol level (Figure 4). For example, to reach the LDL-cholesterol target of 2.5 have an adherence rate of at least 76% on low-dose treatment and of 63% on standard- mmol/l (97 mg/dl), a patient with a baseline LDL-cholesterol of 3.7 mmol/l needs to dose treatment. In our study population, 38% in the low-dose and 28% in the standard- dose treatment group did not reach these adherence rates. Patients with a lower baseline least 37% on low-dose and 33% on standard-dose treatment to reach this target. With a LDL-cholesterol level of, for example, 2.7 mmol/l need to have an adherence rate of at on standard-dose, whereas patients on low-dose treatment are not expected to reach higher baseline LDL-cholesterol level of 4.7 mmol/l an adherence rate of 88% is needed their target even with an adherence rate of 100%.

DISCUSSION

We determined the concurrent effect of treatment dose and adherence on LDL-cholesterol outcomes in diabetes patients starting on statins, where previous research focused on adherence disregarding the statin dose (or vice versa). In patients with a baseline LDL- standard-dose treatment is associated with 4% less reduction in LDL-cholesterol. Such cholesterol level of 3.7 mmol/l we estimated that a 10% reduction in adherence rate on patients need to take more than 60% of their medication to reach the target level of 2.5 patients on standard-dose treatment to reach the same LDL-cholesterol level. mmol/l. Patients on low-dose treatment need to be at least 10% more adherent than with low-dose statin treatment [6,8]. Nevertheless, a substantial number of patients In line with previous findings, we observed that reaching lipid targets was more difficult started on low-dose statin treatment. Remarkably, we found that these patients had a starting on standard-dose treatment. As observed before, we found that older and high average baseline level of 3.7 mmol/l, which was almost similar to that of patients female patients are more often treated with a low-dose [26,6]. Fear for adverse events in such patients may be the reason to initiate on a low dose but in patients with high LDL- cholesterol levels up-titration to a maximally tolerated dose would be the next step [4,27]. The patients in our study, however, remained on low-dose statin treatment till the end of follow-up.

72 Statin treatment and LDL-cholesterol response

The predicted LDL-cholesterol reduction of 40% in patients who were 80% adherent to standard-dose treatment was similar to reductions reported in the reviews based on clinical trials [18,19]. It has been acknowledged that also in trials patients may not be adherent [28]. Regarding the association between adherence and outcomes, one small sample size study found that 10% less medication taken would correlate with 3.1% less reduction in LDL-cholesterol in patients who had an average baseline LDL-cholesterol of [12]. Most patients in that study were on low-dose simvastatin or atorvastatin treatment. Our study provides new insights in the effect of being non-adherent on 3.9 mmol/l standard-dose as compared to low-dose treatment. For patients on standard-dose, a 10% decrease in adherence is associated with 4% less LDL-cholesterol reduction. The 3 difference in LDL-cholesterol reductions seen between low-dose and standard-dose treatment becomes smaller, when the adherence rate is lower. Our results suggest that for patients with less than 50% adherence there is little difference in expected outcomes regardless of the statin dose. Since most patients have higher adherence rates, physicians should preferably use standard-dose statin as initial treatment, as low-dose treatment is more sensitive to treatment non-response due to non-adherence [27].

Reviews and meta-analyses on the therapeutic equivalence of statins have resulted in

Dutch guidelines, which was based on the review of Law et al [19] and provides estimates slightly varying classifications on statin doses. We followed the classification used in the evidence about simvastatin compared to atorvastatin, there is some debate about the with confidence intervals. Given these partly overlapping intervals and also conflicting equivalence of statins but estimates based on results of more than 50 head-to-head trials conducted by the Canadian Drug Effectiveness Review Project resulted in the same [18]. classification as we used in our study A strength of our study is that we estimated the association between statin adherence and LDL-cholesterol using adherence as a continuous variable. Often a cut-off point of 80% adherence is used to determine the effect of adherence on outcomes [29]. Our study achieved by increasing either the adherence or the dose, especially for patients with showed that even in patients with adherence rates above 80% relevant benefits can be high baseline LDL-cholesterol levels. Although patients with missing LDL-cholesterol levels for determining an association between adherence and reduction in LDL- measurements could not be included, we had sufficient variation in LDL-cholesterol cholesterol. Also, selection by matching did not affect the observed association in the patients on standard-dose treatment, nor did restricting the time period for the outcome measurements. We did not differentiate between primary and secondary prevention patients, since the effect of statins is expected to be similar in both groups [30].

73 CHAPTER 3

A limitation of our study is the use of PDC as measure for adherence, which is known to overestimate actual adherence [31]. This limitation is likely to result in observing a weaker association between adherence and LDL-cholesterol response. Also adherence adherence to treatment may decrease over time. However, we conducted sensitivity was measured over a flexible time period, which could have biased this relation as analyses showing similar associations between adherence and LDL-cholesterol response after restricting the maximal difference follow-up period to 180 or 90 days. Finally, we were not able to adjust for lifestyle factors that might be related to adherence and LDL- cholesterol response.

CONCLUSION

Diabetes patients initiating statin treatment are at high risk of not reaching the recommended cholesterol target, especially when they start on a low-dose statin. Patients with high baseline LDL-cholesterol levels starting on low-dose statin are not expected to reach their target even with an adherence rate of 100%. To prevent the lack of effective statin treatment more attention should be given to improve current management strategies in clinical practice.

74 Statin treatment and LDL-cholesterol response

REFERENCES

1. de Vries FM, Denig P, Pouwels KB, Postma MJ, Hak E. Primary prevention of major cardiovascular and cerebrovascular events with statins in diabetic patients: a meta-analysis. Drugs 2012; 72(18): 2364-2373. 2.

detreatment Vries FM, for Kolthof the secondary J, Postma prevention MJ, Denig P,of Hak cardiovascular E. Efficacy ofand standard cerebrovascular and intensive events statin in

3. diabetesNHG-standaard patients: Cardiovasculair a meta-analysis. risicomanagement PLoS One 2014; 9(11): (Eerste DOI:0.1371/journal.pone.0111247. herziening) Huisarts Wet 2012; 55(1): 14-28. 3 4. Rutten GEHM, De Grauw WJC, Nijpels G, et al. NHG-Standaard Diabetes mellitus type 2 (derde herziening). Huisarts Wet 2013; 56: 512-525. 5. American Diabetes Association. Standards for medical care in diabetes – 2015. Diabetes Care 2015;38(Suppl. 1):S1–S90 6. Heintjes EM, Penning-van Beest FJA, Plat AW, et al. Cholesterol level goal attainment with statins: clinical management guideline recommendations versus management in actual clinical practice. Pharmacotherapy 2012; 32(7): 631-641. 7. Sidorenkov G, Haaijer-Ruskamp FM, de Zeeuw D, Denig P. A longitudinal study examining

adherencetimeliness. toPLoS guidelines One 2011; in 6(9)diabetes :e24278. care according to different definitions of adequacy and 8. Geleedst-De Vooght M, Maitland-van der Zee AH, Schalekamp T, Mantel-Teeuwisse A, Jansen P. Statin prescribing in the elderly in the Netherlands: A pharmacy database time trend study. Drugs Aging 2010; 27(7): 589-596. 9. Parris ES, Mohn LA, Lawrence DB, Long LB. Adherence to statin therapy and LDL cholesterol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care 2005; 28(3): 595-599. 10. Watanabe JH, Bounthavong M, Chen T. Revisiting the medication possession ratio threshold for adherence in lipid management. Curr Med Res Opin 2013; 29(3): 175-180. 11. Shalev V, Goldshtein I, Halpern Y, Chodick G. Association between persistence with statin therapy and reduction in low-density lipoprotein cholesterol level: analysis of real-life data from community settings. Pharmacotherapy 2014; 34(1):1-8. 12. Cheng CW, Woo KS, Chan JC, Tomlinson R, You JH. Association between adherence to statin therapy and lipid control in Hong Kong Chinese patients at high risk of coronary heart disease. Br J Clin Pharmacol 2004; 58(5): 528-535. 13.

14. AvailableStuurman-Bieze at: https://www.gipdatabank.nl/databank.asp. AG, Hiddink EG, van Boven JF, Vegter S. Proactive Accessed pharmaceutical8 April 2015. care interventions improve patients’ adherence to lipid-lowering medication. Ann Pharmacother 2013; 47(11): 1448-1456.

75 CHAPTER 3

15. Voorham J, Denig P. Computerized extraction of information on the quality of diabetes from text in electronic medical patient records of general practitioners. J Am Med Inform Assoc 2007; 14(3): 349-354. 16. Lamberts H, Wood M (eds). International Classification of Primary Care (ICPC). Oxford: Oxford University Press; 1987 17.

18. ATC/DDDSmith MEB, index. Lee NJ, Available Haney E, at: Carson http://www.whocc.no/atc_ddd_index/ S. Drug class review: Hmg-coa reductase Accessed inhibitors 8 April (statins) 2015 and fixed-dose combination products containing a statin: Final report update 5. Portland (OR); 2009 19. Law MR, Wald NJ, Rudnicka AR. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: systematic review and meta-analysis. BMJ 2003; 326(7404): 1423. 20. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clinical Chemistry 1972; 18(6): 499-502. 21. Interacting medication. Available at:

. Accessedhttps://kennisbank.knmp.nl/article/Informatorium_ 10 November 2014 22. Medicamentorum-_-G633.htmlVoorham J, Haaijer-Ruskamp FM, Wolffenbuttel BHR, Stolk RP, Denig P. Medication adherence Clin Ther 2011; 33(1):

121-134.affects treatment modifications in patients with type 2 diabetes. 23. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011; 46(3): 399-424. 24. Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Statist Med 2009; 28(25): 3083-3107. 25. Stürmer T, Wyss R, Glynn RJ, Brookhart MA. Propensity scores for confounder adjustment when assessing the effects of medical interventions using nonexperimental study designs. J Intern Med 2014; 275(6): 570-580. 26. Penning-van Beest FJA, Termorshuizen F, Goettsch WG, Klungel OH, Kastelein JJ, Herings RM. Adherence to evidence-based statin guidelines reduces the risk of hospitalizations for acute myocardial infarction by 40%: a cohort study. Eur Heart J 2007;28(2):154-159. 27. Stroes ES, Thompson PD, Corsini A, et al; European Atherosclerosis Society Consensus Panel. Statin-associated muscle symptoms: impact on statin therapy-European Atherosclerosis Society Consensus Panel Statement on Assessment, Aetiology and Management. Eur Heart J 2015; 36(17): 1012-1022. 28. Dodd S, White IR, Williamson P. Nonadherence to treatment protocol in published randomised controlled trials: a review. Trials 2012; 13: 84. 29. De Vera MA, Bhole V, Burns LC, Lacaille D. Impact of statin adherence on cardiovascular disease and mortality outcomes: a systematic review. Br J Clin Pharmacol 2014; 78(4): 684-698.

76 Statin treatment and LDL-cholesterol response

30.

Kearneypeople with PM, Blackwelldiabetes inL, Collins14 randomised R, et al. Efficacy trials of of statins:cholesterol-lowering a meta-analysis. therapy Lancet in 18,6862008; 371(9607): 117-125. 31. Vink NM, Klungel OH, Stolk RP, Denig P. Comparison of various measures for assessing Pharmacoepidemiol Drug Saf 2009;

medication18(2): 159-165. refill adherence using prescription data.

CHAPTER 4

Cost-effectiveness of statins for primary prevention in patients newly diagnosed with type 2 diabetes in The Netherlands

Folgerdiena M. de Vries1,2 Petra Denig2 Sipke T. Visser1 Eelko Hak1 Maarten J. Postma1

Value in Health 2014;17:223-230.

ABSTRACT

Objectives: Statins are lipid-lowering drugs that reduce the risk of cardiovascular events in patients with diabetes. The aim of this study is to determine whether statin treatment for primary prevention in newly diagnosed type 2 diabetes is cost- effective, taking non-adherence, baseline risk, and age into account.

Methods: A cost-effectiveness analysis was performed using a Markov model with a time horizon of 10 years. The baseline 10-year cardiovascular risk was estimated in a Dutch population of primary prevention patients with newly diagnosed diabetes from the Groningen Initiative to Analyse Type 2 Diabetes Treatment database (GIANTT), using the UKPDS risk engine. Statin adherence was measured as pill days covered (PDC) in the IADB.nl pharmacy research database. Cost-effectiveness was measured in costs per quality-adjusted life-year (QALY) from the healthcare- payers’ perspective.

Results: or an average patient of 60 years, the base case, statin treatment was highly cost-effective at €2,245 per QALY. Favorable cost-effectiveness was robust in sensitivity analysis. Differences in age and 10-year cardiovascular risk showed large differences in cost-effectiveness from almost €100,000 per QALY to almost being cost saving. Treating all patients under 45 at diabetes diagnosis was not cost- effective (weighted cost-effectiveness of almost €60,000 per QALY).

Conclusions: Despite the non-adherence levels observed in actual practice, statin treatment is cost-effective for primary prevention in patients newly diagnosed with type 2 diabetes. Due to large differences in cost-effectiveness according to different

patients with relatively higher cardiovascular risk and higher ages. risk and age groups, the efficiency of the treatment could be increased by targeting

80 Cost-effectiveness of statins for primary prevention in diabetes

INTRODUCTION

The prevalence of diabetes is growing in the Netherlands, as in many countries worldwide, due to aging of the population and the occurrence of diabetes at younger ages[1]. Patients with type 2 diabetes have an increased risk of cardiovascular complications[2]. These diabetes [3,4,5]. Statins are lipid-lowering drugs that have shown to reduce the risk of complications reflect a high burden for health and cause substantial costs related to cardiovascular and cerebrovascular events in diabetes patients [6,7,8]. After 2006, the Dutch guidelines for diabetes and cardiovascular risk management changed and recommended statin treatment for the prevention of cardiovascular events for practically all type 2 diabetes patients, unless they have a low cardiovascular risk [9,10]. Although risk tables changed in recent guidelines, it was found using different risk calculation methods that few type 2 diabetes patients are at such low risk [11]. Also in other countries, statin 4 treatment is recommended for a large proportion of the diabetes population [12,13].

Previous pharmacoeconomic analyses on the primary prevention of cardiovascular and cerebrovascular events estimated that statin treatment is likely to be cost-effective in type 2 diabetes patients [14,15,16]. These pharmacoeconomic results, however, are limited since they made use of data from the same clinical trial, which included a restricted patient population with high adherence levels. Adherence to statins is a well-known problem in clinical practice [17,18], and therefore important to take into account when assessing cost-effectiveness. Furthermore, attention should be paid to the risk level where treatment becomes cost-effective. Greving et al. [19] recently demonstrated for a hypothetical primary prevention population that statins were not cost-effective in persons at low risk in the Netherlands. Whether this may also be true for type 2 diabetes patients at relatively low risk is not known. Earlier detection of type 2 diabetes and younger age of these patients cause that patients with lower risk for cardiovascular [20]. For better guidance on treatment decisions, it is important to gain insight in the cost-effectiveness and cerebrovascular events are being identified and treated with statins of statin treatment in type 2 diabetes patients using the cardiovascular risk estimates and adherence levels based on patient data from actual practice.

The objective of this study is to determine whether statin treatment for primary prevention, started at the time of diagnosis of type 2 diabetes, is cost-effective from a healthcare payers’ perspective in the Netherlands, taking non-adherence, baseline risk and age into account.

81 CHAPTER 4

METHODS

Markov Model The cost-effectiveness of statin treatment for primary prevention, started at the time of type 2 diabetes diagnosis, was determined with a Markov model (Figure 1). A population with statin treatment and a population without the use of any lipid-regulating treatment were simulated with the model. The design of the model was the same for both populations, only probabilities of transitions to events changed by the use of statin treatment. We assumed that all patients got optimal secondary prevention. Differences in costs and effects were evaluated.

Because we aimed to determine the cost-effectiveness of primary prevention, all having experienced no previous cardiovascular or cerebrovascular event. Ten different patients started in the ‘Otherwise healthy’ state, that is, patients with type 2 diabetes health states were included in the Markov model and patients could shift to another health state after a 1-year cycle, for which half cycle corrections were performed [21]. The adverse events rhabdomyolysis and myopathy were taken into account. Patients that run. The correction for patients discontinuing treatment due to adverse events was done had experienced an adverse event returned to the ‘Otherwise healthy’ state in the next by applying real world adherence rates, which include patients who have stopped statin treatment after an adverse event. In the base case a time horizon of 10 years was used, as we are focusing on primary prevention.

Myopathie Otherwise Healthy Rhabdomyolysis

Fatal CHD Non-fatal CHD Non-fatal stroke Fatal stroke

Post CHD Post stroke

Death

CHD: coronary heart disease

FIGURE 1. Markov model

82 Cost-effectiveness of statins for primary prevention in diabetes

Event probabilities Probabilities of shifting to an event state were taken from the results of the UKPDS risk engine with which the annual rates for an event over a 10-year period can be calculated (see below) [22,23,24]. When patients experienced a non-fatal event in the model, they moved to the ‘Post event’ state in the next cycle. All patients in the cohort experienced from the Dutch lifetime tables, taking the ageing of the Dutch population explicitly into age-specific probabilities of dying from other causes. These probabilities were taken account [25]

. As we evaluate primary prevention, we used a simplified adjustment on the the probabilities of death, taken from the Dutch lifetime tables, were increased two-fold death rate after the occurrence of an event. That is, for patients in the ‘Post event’ states based on previous studies [26,27].

The UKPDS risk engine was used to estimate the patients’ 10-year risks for coronary 4 heart disease (CHD) and stroke on the basis of patient characteristics, including age, lipid ratio and other risk factors [22,23,24]. These patient characteristics were obtained from the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database. The GIANTT database contains anonymised longitudinal information retrieved from electronic medical records of general practitioners and is maintained by the University Medical Center Groningen [28]. Information on age, lipid ratio and smoking status were taken at the time of type 2 diabetes diagnosis, in line with Dutch guidelines that advise statin treatment for type 2 diabetes patients immediately after diagnosis for practically all patients [9,10]. Since it is to be expected that after diagnosis HbA1c and blood pressure values will decrease due to initiation of glucose- and blood pressure lowering treatment, we averaged the values for HbA1c and systolic blood pressure at one and two years after the diagnosis for estimating the 10-year risks. In this way, we aim to estimate the risk factors levels achieved with such treatment and assume that patients remain adequately treated for these other risk factors during follow-up. This allows us to assess the risk reduction of statin treatment in addition to the risk reduction caused by concurrent treatment for the other risk factors. Clinical data retrieved from the GIANTT database for a cohort of patients diagnosed with type 2 diabetes in 2007, 2008 or 2009 were inserted into the UKPDS risk engine. Patients with a history of CHD or stroke (International [29] or who had a prescription for any lipid-regulating treatment (Anatomical Therapeutic chemical (ATC) C10) [30] during the Classification of Primary Care codes K75, K76, K90) year before diagnosis were excluded.

In total, 4,683 newly diagnosed type 2 diabetes patients without a history of cardiovascular and cerebrovascular events were extracted from the GIANTT database and included in the UKPDS risk engine. The average patient was 61.3 years of age and there were slightly more women than men (51.4%). On average patients had a HbA1c

83 CHAPTER 4

high-density lipoprotein cholesterol), 18% of the patients were smoking and 1.6% had of 6.7%, systolic blood pressure of 140 mm Hg and a lipid ratio of 5 (total cholesterol/ atrial fibrillation according to their medical records. The average 10-year risks of the population aged 25-65 years at type 2 diabetes diagnosis, was 16% for CHD, 8% for fatal CHD, 4% for stroke and 0.7% for fatal stroke. on the risks. Patients aged 56-65 at diagnosis of type 2 diabetes had average risk of 21% Stratifying the 10-year cardiovascular risks by age showed that age has a large influence for CHD, 11% for fatal CHD, 6% for stroke and 1% for fatal stroke, patients aged 46-55 years had risks of 15%, 6%, 3% and 0.4%, respectively. Patients aged 36-45 years were estimated to have average risks of 9% for CHD, 3% for fatal CHD, 1% for stroke and 0.1% for fatal stroke.

Statin treatment and adherence meta-analysis of randomized clinical trials on the primary prevention of cardiovascular Statin efficacy was modelled by implementing the relative risk reduction found in a recent and cerebrovascular events in diabetes patients with statins [8]. Overall relative risk reductions of 31% for non-fatal CHD and fatal CHD and 30% for non-fatal stroke and fatal stroke were applied. The risks for adverse events myopathy and rhabdomyolysis were taken from a systematic review [31].

Adherence to statins was modelled using real-world data from the IADB.nl database, which is a pharmacy research database in the Netherlands with a sample of the Dutch population of 500,000 persons. Patients with type 2 diabetes and no history of proxies (Appendix II) cardiovascular and cerebrovascular events were identified by the use of medication Covered (PDC) [32]. Adherence rates were measured for patients starting statin treatment . Statin adherence rates were defined as the Proportion of Days from 2008 up to and including 2010 (ATC C10AA, C10BA, C10BX). Patients were considered as a starter of statin treatment when they did not use any lipid-lowering medication (ATC C10) in the year before the start of the statin treatment. Adherence rates were measured up to a maximum of three years, and in the base case we assumed that statin adherence stayed constant after year three, as was observed in a previous study [17].

[33] Non-adherence was modelled by reducing the statin efficacy. Statin treatment with ≥80% PDC was assumed to be associated with full efficacy . Patients with ≤20% of intermediary levels of PDC were approximated with linear interpolation in the base PDC were assumed to be associated with no efficacy of the statin treatment. Efficacy

84 Cost-effectiveness of statins for primary prevention in diabetes

case, as studies indicate that there is a linear association between adherence levels and reductions in LDL-cholesterol and in hospitalizations [34,35,36].

In 1,660 patients who started statin treatment the mean PDC was 81% during their three, respectively. In total, 1,049 patients were followed for at least two years and 479 first year of statin use. Adherence rates decreased to 77% and 75% in year two and for three years. As we allowed new users to be included during the study period, some patients could only be followed for one year. The decrease in the number of patients in the follow-up years was mainly due to later inclusion into the study rather than early dropout.

Health effects and costs We determined the difference in quality-adjusted life-years (QALY) between no lipid- 4 regulating treatment and treatment with a statin. QALYs were calculated by multiplying state. For patients experiencing a CHD or stroke disutility scores were 0.06 and 0.22, the time in a specific health state with the quality of life associated with the health respectively [37]. These utility scores were constant over time. A small disutility of 0.001 for taking a statin pill every day was included [38,39].

All cost estimates were updated to 2012 Euro’s with the Dutch consumer price indices on statin pill costs (€7.3 for 40 mg generic simvastatin) [40], two GP visits (€59.5) [41], (http://statline.cbs.nl). Annual costs for statin treatment in the base case were based annually lipid spectrum testing (€20.3) [42], and four pharmacists’ prescription fees (€23). Event costs were based on literature (Table 1).

85 CHAPTER 4

TABLE 1. Incidence, effectiveness, disutilities and cost data for the base case Parameter Base case Distribution Data source Reference Annual risks CHD yearly risk, 1.7, 1.8, 1.9, 2.0, 2.0, 2.1, Beta Real world data from GIANTT database - % (10-year risk) 2.2, 2.3, 2.4, 2.5 (21%) inserted in UKPDS risk engine Fatal CHD yearly risk, 0.7, 0.8, 0.9, 1.0, 1.1, 1.2, Beta Real world data from GIANTT database - % (10-year risk) 1.3, 1.4, 1.5, 1.6 (11%) inserted in UKPDS risk engine Stroke yearly risk, 0.3, 0.4, 0.4, 0.5, 0.6, 0.6, Beta Real world data from GIANTT database - % (10-year risk) 0.7, 0.8, 0.9, 1.0 (6%) inserted in UKPDS risk engine Fatal stroke yearly risk, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, Beta Real world data from GIANTT database - % (10-year risk) 0.1, 0,1, 0.2, 0.2 (1%) inserted in UKPDS risk engine Myopathy (per 11 Beta Systematic review 31 100,000 person years) Rhabdomyolysis (per 3 Beta Systematic review 31 100,000 person years) Overall mortality rate 12% - Actual rates 25 (10-year risk) Statin treatment effectiveness (relative risk) MI 0.69 Lognormal Meta-analysis 8 Stroke 0.70 Lognormal Meta-analysis 8 Mean adherence rate First year 81% Beta Real world data from the IADB.nl - prescription database Second year 77% Beta Real world data from the IADB.nl - prescription database Third until tenth year 75% Beta Real world data from the IADB.nl - prescription database Disutility MI 0.06 Triangular Meta-analysis 37 Stroke 0.22 Triangular Meta-analysis 37 Act of taking statin -0.001 Triangular Assumption 38,39 Annual cost data (€) Statin treatment Simvastatin 40 mg € 7.30 - Official tariff 40 Doctor visits € 59.5 - Official tariff 41 Laboratory fee € 20.3 - Official tariff 42 Pharmacists fee € 24.3 - Official tariff 43 Myopathy € 59.5 Gamma Assumption - Rhabdomyolysis € 11,126 Gamma Cost study 44 MI First year € 5,012 Gamma Cost study 45 Subsequent years € 1,885 Gamma Cost study 5 Stroke First year € 13,480 Gamma Cost study 46 Subsequent years € 1,885 Gamma Cost study 5 Death € 2,902 Gamma Assumption 19 CHD: coronary heart disease

86 Cost-effectiveness of statins for primary prevention in diabetes

Economic analysis The model was run over a time horizon of 10 years. Costs were discounted at 4% and health effects at 1.5%, following the Dutch guidelines [41]. In the base case a 60 year old patient was assessed, based on the average age of 61.3 years at type 2 diabetes diagnosis in the GIANTT population. Mean 10-year cardiovascular risks of patients aged 56-65 at of the parameters on the outcome of the cost-effectiveness was assessed. Additionally, type 2 diabetes diagnosis were used. With the use of a tornado diagram the influence the effects of age and baseline risk for CHD and stroke were evaluated. To assess the impact of individual parameters that may vary between countries or that were based on assumptions, one-way sensitivity analyses were performed. This included sensitivity analyses for statin costs, adverse event rates, discounting rates, disutility of taking a pill every day, time horizon, adherence rates, and relation between non-adherence Table 2. We also included 4 a one-way sensitivity analysis to account for the possible overestimation by using the and efficacy. Parameter variations tested are presented in UKPDS risk engine, assuming a maximum reduction in the estimates of 50% [47]. Finally, a probabilistic sensitivity analysis with 5,000 iterations was performed to deal with parameter uncertainty [48]. The parameter distributions used are listed in Table 1.

87 CHAPTER 4

TABLE 2. One-way sensitivity analyses (base case values between brackets)

Assumption Cost-effectiveness Base case €2,245

Statin costs (€7.3) €18.5 (actual costs IADB.nl population) €2,644 €50 €3,766 €100 €5,547

Myopathy (11 on 100.000 years) 100 €2,247

Discounting rate (effects 1.5%;costs 4%) Effects 3%; costs 3% €3,041 Effects 4%; costs 4% €3,144

Disutility taking a pill (0.999) No disutility €2,155

Time horizon (10 years) 5 years €12,424

Adherence rate (81%, 77%, 75% respectively first, second and subsequent years) Full adherence (80-100% PDC) in all years €1,534 50% adherence in all years €3,312 Adherence decreasing 2% each year after year 3 €2,266 Adherence decreasing 5% each year after year 5 €2,297

Relation statin efficacy and non-adherence (≥80%=1;20-80%=linear;≤20%=0) ≥70%=1;30-70%=linear;≤30%=0 €2,217 ≥80%=1;0-80%=linear €2,175 ≥60%=1;0-50%=linear €2,108

UKPDS risk overestimation Lowering risk with 50% €6,500 PDC: Proportion of Days Covered.

88 Cost-effectiveness of statins for primary prevention in diabetes

TABLE 3. Outcomes of the cost-effectiveness analysis stratified by age and risk for coronary heart disease and stroke Age 10-year risk for Proportion QALY’s 1,000 Costs per Cost-effectiveness CHD; fatal-CHD; per age patients 1,000 stroke; fatal stroke group patients <45 1. 3;1;1;0 27.5 -8 €814,012 Inferior 2. 6;1;1;0 28.5 -3 €789,256 Inferior 3. 9;3;1;0 21.8 11 €738,926 €66,537 4. 13;4;2;0 9.5 43 €689,576 €16,085 5. 16;6;3;1 12.7 77 €634,993 €8,223 Weighted cost-effectiveness for treating all diabetes patients under 45 in the Netherlands = €57,244 45-55 1. 6;2;1;0 21.1 9 €758,216 €88,440 2. 10;4;2;0 28.7 49 €674,895 €13,828 4 3. 15;6;3;0.5 20.8 75 €623,327 €8,345 4. 20;7;4;0.5 16.7 96 €559,641 €5,810 5. 26;10;5;1 12.6 145 €486,131 €3,353 Weighted cost-effectiveness for treating all diabetes patients between 45 and 55 in the Netherlands = €8,295 55-65 1. 6;3;3;1 6.8 43 €638,881 €14,824 2. 10;4;3;1 30.4 61 €618,594 €10,119 3. 21;11;6;1 34.7 193 €432,890 €2,245 4. 27;13;7;1 15.9 197 €382,633 €1,939 5. 35;24;9;2 12.3 453 €276,148 €609 Weighted cost-effectiveness for treating all diabetes patients between 55 and 65 in the Netherlands = €2,480

> €80,000 €50,000 - €80,000 €20,000 - €50,000 < €20,000

89 CHAPTER 4

RESULTS

Economic analysis In the base case, cost-effectiveness is €2,245 per QALY (Table 2). Sensitivity analyses were performed to investigate the differences in risks for CHD and stroke according to age. In Table 3 and 10-year risks for CHD and stroke. The risks chosen for each age group were based on the outcomes of the cost-effectiveness analysis are shown stratified by age the actual risk ranges within the GIANTT population. Treating all type 2 patients under the age of 45 for primary prevention of cardiovascular and cerebrovascular events was not cost-effective, with costs of €57,243 per QALY. Treating all patients aged 45-55 at diagnosis was cost-effective with €8,295 per QALY. However, for 21,1% of the patients cost-effectiveness would be high and not cost-effective (Table 3).

The tornado diagram (Figure 2) for CHD and stroke and the adherence rates have a large effect on the outcome of the shows that the efficacy of the statins, the baseline risk cost-effectiveness. In the one-way sensitivity analyses (Table 2), statin costs also have a major effect on cost-effectiveness. In the Netherlands, statin costs for 40 mg generic simvastatin were only €7.3 annually. The actual annual statin costs per patient in the IADB.nl population were €18.5, which results in a negative effect on cost-effectiveness (€2,644 per QALY). Changing the time horizon to 5 years will lead to a substantial increase in costs (€12,424 per QALY). Increasing adherence to statin treatment will improve cost-effectiveness. Full adherence reduces the costs to €1,534 for the base case. However, changing the assumptions on the adherence rate after year three does not have a great impact on cost-effectiveness (Table 2).

The acceptability curves of the probabilistic sensitivity analyses in Figure 3 show that the probability that statin treatment is cost-effective for the average 60 year old patient is 100%, using a threshold of €20,000 per QALY. Considering the average 40 year old patient, the analysis shows that with this threshold of €20,000 per QALY, the probability that it is cost-effective is less than 15% (Figure 3).

90 Cost-effectiveness of statins for primary prevention in diabetes

FIGURE 2. Tornado diagram; all parameters are increased and decreased with 25%

60 year old 50 year old 40 year old FIGURE 3. Cost-effectiveness acceptability curves from the probabilistic sensitivity analysis

91 CHAPTER 4

CONCLUSION

We estimated that statin treatment for primary prevention starting at the time of diagnosis of type 2 diabetes is cost-effective at €2,245 per QALY in the base case. There is great diversity in the risks for CHD and stroke in the type 2 diabetes population. Differences in age and 10-year risks for CHD and stroke result in large differences in cost-effectiveness from losing QALYs to being almost cost neutral. With the adherence rates seen in practice, it can be concluded that treating all type 2 diabetes patients under the age of 45 at diagnosis with statins for primary prevention is not cost-effective. For patients between 45 and 55 years at diagnosis, statin treatment is cost-effective except when the 10-year risk for CHD is as low as 6%. For the other patients, statin treatment is expected to be cost-effective.

60-year old patient is in line with other pharmacoeconomic analyses [14,15,16] Our finding that statin treatment for primary prevention is cost-effective for the average . The finding from Greving et al. [19] Cost-effectiveness, however, appears slightly better in our diabetes that statin treatment is not cost-effective in patients at low risk is in line with the findings population. For a 65-year old patient with a 10-year risk of 15%, Greving et al estimated the costs at €12,652 per QALY whereas we estimated the costs at €10,119 per QALY for a 60-year old patient with a 10-year risk of 13%.

Currently, there are over 800,000 diabetes patients in the Netherlands [49]. Within the GIANTT population, 14% of the patients has a cardiovascular history, implying that 688,000 patients might be eligible for primary prevention with statin treatment. With the current price levels, such statin treatment would cost the Dutch healthcare system around €75.7 million each year. The broad range for cost-effectiveness between by focussing on patients with higher cardiovascular risk and higher ages. Within the the individual patients suggests that the efficiency of treatment could be increased GIANTT primary prevention population, 19% of the patients had an age at diagnosis under 50 years in whom starting statin treatment at the time of diagnosis may not be cost effective.

Cost-effectiveness was assessed taking various parameters into account, such as adherence and adverse events. Adherence to statin treatment obviously plays an important role in cost-effectiveness analysis. In our study population, adherence rates for statin treatment were higher than reported in other studies [17,50]. Nevertheless, further improving adherence would lead to improvements in cost-effectiveness. Furthermore, in the base case analysis the statin costs of 40 mg generic simvastatin (€7.3) were applied; use of the average annual costs of statin treatment from the IADB.nl population (€18.5)

92 Cost-effectiveness of statins for primary prevention in diabetes

[9,10] recommend 40 mg simvastatin or 40 mg pravastatin (€16.10) when statin treatment is started. has a minor negative influence on cost-effectiveness. Dutch guidelines

Nevertheless, some patients need a switch to more expensive statins to reach lipid Treating patients with 40 mg simvastatin has beneficial effects on cost-effectiveness. goals [51]. In the base case analysis, we chose a time horizon of 10 years. Shortening the time horizon to 5 years clearly affects the cost-effectiveness, indicating that primary the treatment for a longer period. prevention with statins is especially beneficial in diabetes patients who can and will take

Strengths We analysed the baseline risk of a population of type 2 diabetes patients using observational patient data that were inserted into the UKPDS risk engine. Since the GIANTT cohort includes an unrestricted population of type 2 diabetes patients in 4 practice as compared to the patients included in trials. Adherence rates were also primary care, these data better reflect the population treated with statins in actual measured in the actual population. Both populations were taken from the northern part of the Netherlands. Using the tornado diagram, the parameters that have a large the baseline risk for CHD and stroke and adherence rates- were analysed in detail. From influence on the outcome of the cost-effectiveness analysis -such as the efficacy of statins, this, it becomes clear that the utilities and the drug costs did not have a large impact on age. As a result of the broad ranges in the 10-year risks for CHD and stroke between the the results. Another strength of the study is that we stratified the patients according to different age groups, broad variations in cost-effectiveness could be shown.

Limitations First, the UKPDS risk engine was used for determining cardiovascular risks for type 2 diabetes patients. There is evidence that the UKPDS risk engine overestimates the risks for CHD and stroke [47,52]. As shown in the sensitivity analysis, the costs per QALY could more than double when risks are overestimated by a factor of 2. However, it seems that at the risk levels seen in the primary prevention diabetes population the overestimation is probably not that high [47]. Second, adherence rates of the patients were measured in a pharmacy research databases in which the indication of the patient was not known. Medication proxies had to be used for identifying diabetes primary prevention patients. The proxy for diabetes is known to give a high sensitivity [53]. The proxy for primary prevention showed a high sensitivity of 85%. However, around 15% of the included patients can be expected to be secondary prevention patients. Third, the adherence time afterwards. In addition, for the intermediate adherence levels we assumed a linear rates were only measured for the first three years and assumed to stay constant over relation with the efficacy. Although there is some evidence that these assumptions 93 CHAPTER 4

are acceptable, [17,34,35,36] we tested them in the sensitivity analyses which showed that various changes in the assumptions did not have much impact on the results. We used real patient data for the risk estimation and adherence rates, but treatment efficacy, MI and post-stroke utilities, we made use of estimates from meta-analyses, which applied disutilities and cost data were based on literature. For statin efficacy as well as for post- minimum criteria for including appropriate data. For costs, we tried to retrieve the most recent data available but had to rely on multiple sources published in the past decade. Only direct health care costs were included. Including indirect costs would obviously be favourable for cost-effectiveness.

This study is an illustration for the Netherlands. In the Netherlands, statin costs are very low, and such medication is reimbursed for all patients. Other countries may have higher statin costs or higher non-adherence rates due to patient co-payments, and cost- effectiveness might be less favourable. The populations used in this analysis were from the northern part of the Netherlands, which is known to have a lower rate of immigrants and a lower socio-economic-status in comparison to other parts of the Netherlands [54,55]. As one could speculate that this might affect the estimated risks for cardiovascular events and the adherence rates in opposite directions, it is not clear to what extent this would result in relevant changes in our findings. We used a time horizon of 10 years with cycle lengths of one-year. Due to data availability is was not possible to shorten the cycle length but by performing half cycle corrections the effect of this longer cycle length was minimized. As we aimed to assess the effect of primary prevention of cardiovascular and cerebrovascular events, we did not perform a life-time analysis but did shorten the time horizon in the sensitivity analysis to 5 years.

In general, primary prevention with statin treatment at the time of diagnosis is not cost- effective for all type 2 diabetes patients. In the recent Dutch guidelines for cardiovascular management 40 years is mentioned as age limit for diabetes patients below which statin treatment would seldom be indicated [9]. Our study indicates that this age limit may be set higher, since on average starting statin treatment in patients below 45 years is not expected to be cost-effective. On the other hand, our results indicate that statin treatment, even with the levels of non-adherence and variations in choice of statin seen in actual practice, is highly cost-effective for patients of more than 55 years at diagnosis with a time horizon of 10 years. Although it is acknowledged that there is no single be considered, these simple age-based recommendations could be helpful for guiding treatment appropriate for all patients and the individual risk profile of a patient should practice.

94 Cost-effectiveness of statins for primary prevention in diabetes

REFERENCES

1. The national institute for public health and the environment. Diabetes in the Netherlands

2007.9 April Available 2013 from: http://www.rivm.nl/bibliotheek/rapporten/260322001.pdf Accessed 2.

American Heart Association. Cardiovascular disease & Diabetes statistics update 2010. Available from: http://www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetes 3. Matters/Cardiovascular-Disease-Diabetes_UCM_313865_Article.jspClarke P, Gray A, Legood R, Briggs A, Holman R. The impact of diabetes-related Accessed 9 complications April 2013 on healthcare costs: results from the United Kingdom Prospective Diabetes Study (UKPDS Study No. 65). Diabet Med 2003; 20(6): 442-450. 4. Ringborg A, Yin DD, Martinell M, Stålhammar J, Lindgren P. The impact of acute myocardial infarction and stroke on health care costs in patients with type 2 diabetes in Sweden. Eur J 4 Cardiovasc Prev Rehabil 2009; 16(5): 576-582. 5. Redekop WK, Koopmanschap MA, Rutten GE, Wolffenbuttel BH, Stolk RP, Niessen LW. Resource consumption and costs in Dutch patients with Type 2 diabetes mellitus. Results from 29 general practices. Diabet Med 2002; 19(3): 246-253. 6.

Baigentprospective C, Keech meta-analysis A, Kearney of PM,data et from al. Efficacy90,056 participantsand safety of in cholesterol 14 randomised lowering trials treatment: of statins. Lancet 2005 ; 366(9493): 1267-1278. 7. Wei L, Ebrahim S, Bartlett C, Davey PD, Sullivan FM, MacDonald TM. Statin use in the secondary prevention of coronary heart disease in primary care: cohort study and comparison of inclusion and outcome with patients in randomised trials. BMJ 2005; 330(7495): 821. 8. de Vries FM, Denig P, Pouwels KB, Postma MJ, Hak E. Primary prevention of major cardiovascular and cerebrovascular events with statins in diabetic patients: a meta-analysis. Drugs 2012; 72(18): 2364-2373. 9. NHG-standaard Cardiovasculair risicomanagement (Eerste herziening) Huisarts Wet 2012; 55(1): 14-28. 10. Wiersma TJ, Goudswaard AN. NHG-standaard cardiovascular risk management. [In Dutch]

11. Houten:De Grauw Bohn W, Bakx Stafleu C, Vanvan Loghum;Gerwen W. 2006. Iedere patiënt met diabetes mellitus type 2 een statine, tenzij… Verwarring na twee standaarden? Huisarts en Wetenschap, jaargang 2007, nummer 4: 144-147. 12. American Diabetes Association. Standards of medical care in diabetes 2012. Diabetes Care 2012; 35(Suppl. 1): S11-S63. 13. National Collaborating Centre for Chronic Conditions. NICE clinical guideline 2006. In Type 2 Diabetes: National Clinical Guideline for Management in Primary and Secondary Care (Update). London, Royal College of Physicians, 2008.

95 CHAPTER 4

14. Lafuma A, Colin X, Solesse A. Cost-effectiveness of atorvastatin in the prevention of cardiovascular events in diabetic patients: a French adaptation of CARDS. Arch Cardiovasc Dis 2008; 101(5): 327-332. 15. Raikou M, McGuire A, Colhoun HM, et al. Cost-effectiveness of primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes: results from the Collaborative Atorva- statin Diabetes Study (CARDS). Diabetologia 2007; 50(4): 733-740. 16. Annemans L, Marbaix S, Webb K, Van Gaal L, Scheen A. Cost effectiveness of atorvastatin in patients with type 2 diabetes mellitus. A pharmacoeconomic analysis of the collaborative atorvastatin diabetes study in the belgian population. Clin Drug Investig 2010; 30(2): 133-142. 17. Donnelly LA, Doney AS, Morris AD, Palmer CN, Donnan PT. Long-term adherence to statin treatment in diabetes. Diabet Med 2008; 25(7): 850-855. 18. Penning-van Beest FJ, Termorshuizen F, Goettsch WG, Klungel OH, Kastelein JJ, Herings RM. Adherence to evidence-based statin guidelines reduces the risk of hospitalizations for acute myocardial infarction by 40%: a cohort study. Eur Heart J 2007; 28(2): 154-159. 19. Greving JP, Visseren FL, de Wit GA, Algra A. Statin treatment for primary prevention of vascular disease: whom to treat? Cost-effectiveness analysis. BMJ 2011; 342: d1672. 20. Baan CA, Schoemaker CG. [Diabetes tot 2025 Preventie en zorg in samenhang] Rapport

21. 260322004/2009,Sonnenberg FA, Beck RIVM, JR. MarkovBilthoven, Models 2009. in Medical Decision Making: A Practical Guide. Med Decis Making 1993; 13(4): 322-328. 22. Stevens RJ, Kothari V, Adler AI, Stratton IM. The UKPDS risk engine: a model for the risk of coronary heart disease in Type II diabetes (UKPDS 56). Clin Sci (London) 2001; 101(6): 671-679. 23. Kothari V, Stevens RJ, Adler AI, et al. UKPDS 60: Risk of stroke in type 2 diabetes estimated by the UK Prospective Diabetes Study risk engine. Stroke 2002; 339(7): 1776-1781. 24. Stevens RJ, Coleman RL, Adler AI, Stratton IM, Matthews DR, Holman RR. Risk factors for myocardial infarction case fatality and stroke case fatality in type 2 diabetes: UKPDS 66. Diabetes Care 2004; 27(1): 201-207. 25. Statistics Netherlands. Lifetables: Dutch population by age and sex. Statistics Netherlands, 2010. 26. Haffner SM, Lehto S, Rönnemaa T, Pyörälä K, Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med 1998; 339(4): 229-234. 27. Brønnum-Hansen H, Davidsen M, Thorvaldsen P. Long-term survival and causes of death after stroke. Stroke 2001; 32(9): 2131-2136. 28. Voorham J, Denig P. Computerized extraction of information on the quality of diabetes from text in electronic medical patient records of general practitioners. J Am Med Inform Assoc 2007; 14(3): 349-354.

96 Cost-effectiveness of statins for primary prevention in diabetes

29. Lamberts H, Wood M (eds). International Classification of Primary Care (ICPC). Oxford: Oxford University Press; 1987. 30.

ATC/DDD2013 index. Available from: http://www.whocc.no/atc_ddd_index/ Accessed 13 March 31. Law M, Rudnicka AR. Statin safety: a systematic review. Am J Cardiol 2006; 97(8A): 52C-60C. 32. Peterson AM, Nau DP, Cramer JA, Benner J, Gwadry-Sridhar F, Nichol M. A checklist for medication compliance and persistence studies using retrospective databases. Value Health 2007; 10(1): 3-12. 33. Simpson RJ, Mendys P. The effects of adherence and persistence on clinical outcomes in patients treated with statins: a systematic review. J Clin Lipidol 2010; 4(6): 462-471. 34. Sokol MC, McGuigan KA, Verbrugge RR, et al. Impact of medication adherence on hospitalization risk and healthcare cost. Med care 2005;43:521-530. 35. Parris ES, Lawrence DB, Mohn LA, Long LB. Adherence to statin therapy and LDL choles- 4 terol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care 2005; 28(3): 595-599. 36. Watanabe JH, Bounthavong M, Chen T. Revisiting the medication possession ratio threshold for adherence in lipid management. Curr Med Res Opin 2013; 29(3): 175-180. 37. Lung TW, Hayes AJ, Hayen A, Farmer A, Clarke PM. A meta-analysis of health state valua- tions for people with diabetes: explaining the variation across methods and implications for economic evaluation. Qual Life Res 2011; 20(10):1669-1678. 38.

Nagliedecision IG, analysis. Detsky Med AS. Decis Treatment Making of 1992; chronic 12(4): nonvalvular 239-249. atrial fibrillation in the elderly: a 39. Pignone M, Earnshaw S, Tice JA, Pletcher MJ. Aspirin, statins, or both drugs for the primary prevention of coronary heart disease events in men: a cost-utility analysis. Ann Intern Med 2006; 144(5): 326-336. 40. Dutch Health Care Insurance Board. Drug costs. Available from: www.medicijnkosten.nl Accessed 7 April 2013 41. Hakkaart-van Roijen L, Tan SS, Bouwmans CAM. Guideline for costing research. Methods and standardized prices for economic evaluations in health care. Health Care Insurance Board (The Netherlands), 2010. 42. Dutch Healthcare Authority. Rates for various professionals and various laboratory tests. 2010. Available from: www.nza.nl. Accessed 8 March 2013 43.

Rateskosten%20voor%20farmaceutische%20zorg.asp for pharmacy fee. Available from: http://www.fk.cvz.nl/voorna/i/inl%20de%20 Accessed 9 April 2013 44. Pletcher MJ, Lazar L, Bibbins-Dommingo K, et al. Comparing impact and cost-effectiveness of primary prevention strategies for lipid-lowering. Ann Intern Med 2009; 150(4): 243-254. 45. Soekhlal RR, Burgers LT, Redekop WK, Tan SS. Treatment costs of acute myocardial infarction in the Netherlands. Neth Heart J 2013; 21(5): 230-235.

97 CHAPTER 4

46. Kroes ME. Nationaal Programma Grieppreventie: het succes van de griepprik. Rapport 162. Amstelveen: College voor zorgverzekeringen, 2003. 47. van der Heijden AA, Ortegon MM, Niessen LW, Nijpels F, Dekker JM. Prediction of coronary heart disease risk in a general, pre-diabetic, and diabetic population during 10 years of follow-up: accuracy of the Framingham, SCORE, and UKPDS risk functions: The Hoorn Study. Diabetes Care 2009; 32(11): 2094-2098. 48. Briggs A, Sculpher M, Claxton K. Decision Modelling for Health Economic Evaluation. New York: Oxford University Press Inc, 2006. 49.

National Institute for Public Health and the Environment. Available from: http://www. rivm.nl/Documenten_en_publicaties/Algemeen_Actueel/Nieuwsberichten/2013/Meer_ 50. dan_800_000_mensen_met_diabetes_in_Nederland_toename_forsJackevicius CA, Mamdani M, Tu JV. Adherence with statin therapy Accessedin elderly 31patients October with 2013 and without acute coronary syndromes. JAMA 2002; 288(4): 462-467. 51. Harley CR, Gandhi S, Blasetto J, Heien H, Sasane R, Nelson SP. Low-density lipoprotein cholesterol (LDL-C) levels and LDL-C goal attainment among elderly patients treated with rosuvastatin compared with other statins in routine clinical practice. AM J Geriatr Pharmacother 2007; 5(3): 185-194. 52. Van Dieren S, Peelen LM, Nöthlings U, et al. External validation of the UK Prospective Diabetes Study (UKPDS) risk engine in patients with type 2 diabetes. Diabetologia 2011; 54(2): 264-270. 53. Voorham J, Haaijer-Ruskamp FM, van der Meer K, et al. Kwaliteit van de behandeling van type 2-diabetes: resultaten van het GIANTT-project 2004-2007. Ned Tijdschr Geneeskd 2010; 154: A775. 54.

Nationale atlas volksgezondheid. Socioeconomic status 2010 Available from: http://www. zorgatlas.nl/beinvloedende-factoren/sociale-omgeving/ses/sociaaleconomische-status/Accessed 10 April 2013 55.

Nationale atlas volksgezondheid. Demographic ethnicity. Available from: http://www. zorgatlas.nl/beinvloedende-factoren/demografie/etniciteit/ Accessed 10 April 2013

98 Cost-effectiveness of statins for primary prevention in diabetes

PART II

Patient & physician behaviour regarding statin treatment

CHAPTER 5

Prescribing patterns, adherence and LDL- cholesterol response of type 2 diabetes patients initiating statin on low-dose versus standard-dose treatment: a descriptive study

Folgerdiena M. de Vries1 Jaco Voorham1 Eelko Hak2 Petra Denig1

Submitted

ABSTRACT

Aims: and discontinuation, adherence levels, and response to treatment in patients with The aim of this study was to describe and compare treatment modifications type 2 diabetes initiating on low-dose versus standard-dose statin treatment.

Methods: A 2-year follow-up cohort study was performed using data from the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database in patients with type 2 diabetes initiating statin treatment between January 2007 and December 2012. First we determined whether there were differences in treatment

on a low-dose versus standard-dose. Second, we looked at differences in adherence modifications and discontinuation after statin initiation between patients starting and LDL-cholesterol response after 2 years follow-up between these groups.

Results: Around 22% of patients initiated statin treatment on a dose lower than recommended. More than half of them remained on a low dose during a 2-year follow-up period, whereas less than 15% received a dose increase. Of the patients initiating on standard-dose, also more than half remained on the same treatment during this period, whereas 8% received a dose decrease without subsequent increase. Over 25% of patients starting on low-dose or standard-dose treatment

discontinued treatment, often within the first 180 days after initiation or after levels and were less likely to have adequate LDL-cholesterol response compared to a first treatment change. Patients on low-dose treatment had lower adherence patients on standard-dose after 2 years follow-up.

Conclusions: Current patterns of statin treatment in patients with type 2 diabetes are suboptimal, with discontinuation, inadequate adherence levels and lack of

response after 2 years of follow-up. Patients starting on low-dose had more treatment intensification seen in those who had inadequate LDL-cholesterol

to those starting on standard-dose treatment, which calls for a closer look at the treatment modifications, discontinuation and adherence problems as compared rationale of starting patients on low-dose statin treatment.

104 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

INTRODUCTION

Patients with type 2 diabetes have an increased risk of developing cardiovascular disease, therefore statin treatment is recommended for almost all type 2 diabetes patients. Statins are associated with a reduction in risk of cardiovascular disease [1,2]. Dutch guidelines recommend to start with simvastatin 40 mg for both primary and

[3]. However, statins are not optimally used in clinical practice and lipid targets are not secondary prevention, and aim for an LDL-cholesterol level of ≤2.5 mmol/l (≤97 mg/dl) reached in about a third of the patients [4,5].

This lack of treatment response could be due to being prescribed low-dose treatment [4,6,7] [8,9,10] [11,12]. In the last decade, the prescribed daily dose of statins has increased but on average patients , lack of treatment intensification and/or non-adherence to treatment receive less than the recommended dose [6,13] response. Also, it has been shown that at least a third of the patients with lipid levels , which could lead to insufficient treatment [8,9,10]. Suspicion of non-adherence to 5 treatment could be reason for the physician not to modify treatment, however the ability above target do not receive treatment modifications of physicians in recognizing non-adherence is poor [14]. There is evidence of redundant [15], but also lack of treatment [9]. Also, the relation between adherence and LDL- treatment intensification in non-adherent patients cholesterol response is related to the treatment dose [7]. The initial treatment dose could intensification in adherent patients have an impact on subsequent adherence and treatment modification patterns. Better insight in such patterns of statin treatment modifications and adherence is needed for treatment in daily practice. the development and refinement of interventions aimed at improving outcomes of statin discontinuation, adherence levels, and cholesterol response in patients with type 2 The aim of this study was to describe and compare treatment modifications and diabetes initiating on low-dose versus standard-dose statin treatment.

105 CHAPTER 5

METHODS

Study design A 2-year follow-up cohort study was performed in patients with type 2 diabetes initiating statin treatment between January 2007 and December 2012. First we determined statin initiation between patients starting on a low-dose versus standard-dose. Second, whether there were differences in treatment modifications and discontinuation after

LDL-cholesterol response after 2 years follow-up between patients starting on a low- we looked at differences in treatment modifications, discontinuation, adherence and dose versus standard-dose.

Setting This study was performed using data from the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database. The GIANTT database contains anonymized longitudinal information retrieved from electronic medical records of general practitioners and is maintained by the University Medical Center Groningen [16]. These records include medical history, prescription data, routine laboratory test results and physical examinations of type 2 diabetes patients from the northern part of the Netherlands. Medical history consists of date of diabetes diagnosis and comorbidity [17] or text descriptions that are coded manually. data, which is based on the International Classification of Primary Care (ICPC)

Patient selection Patients managed in general practice for type 2 diabetes initiating lipid-lowering treatment exclusively on a statin were included (Anatomical Therapeutic Chemical (ATC) code C10AA) [18]. Since the documented date of type 2 diabetes diagnosis is not always precise, we allowed for a grace period of 180 days before the documented diagnosis date any lipid lowering medication (ATC code C10) in the preceding 360 days. Patients needed for a statin initiation to be included. Initiation was defined as having no prescription for at least 720 days. Patients with temporary absence from the database, for example due to have sufficient medical history to be classified as initiators, and a follow-up period of were not included as initiators. Patients receiving treatment in daily packages or with a to being institutionalized, as identified by long-term gaps in all prescribed medication, single prescription duration longer than 270 days or with missing prescription attributes were excluded, since adherence cannot be reliably calculated in such cases.

Treatment changes Statins are expected to be prescribed sequentially, but prescription information in the medical records do not necessarily reflect actual drug taking. Therefore, before 106 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

treatment changes were assessed the prescriptions were pre-processed to correct for, artefacts caused by entry errors. This approach is described in more detail elsewhere [19]. amongst others, stockpiling, erroneous prescription durations due to modifications and

The first and second change in treatment in up to two years after statin treatment being a dose adjustment, treatment switch or addition, and treatment discontinuation. initiation were determined. We differentiated between a treatment modification,

Modifications in which the patient stayed on the same type of statin but changed dosing lipid-lowering drug, and had a treatment stop <90 days after or before initiation of the were classified as a ‘dose increase’ or ‘dose decrease’. Patients that started a different new lipid-lowering drug, were defined as switchers. To classify switchers (increase/ categories [20,21]; low-dose, standard-dose and high-dose (Table 1). Patients with a decrease/similar) all lipid-lowering drugs and dosings were classified into three dose decrease’ was a switch to a lipid-lowering drug in a lower dose category, switching to ‘switch increase’ switched to a lipid-lowering drug in a higher dose category, a ‘switch 5 patients had a different lipid-lowering drug started without the initial lipid-lowering a lipid-lowering drug in a similar dose category was defined as ‘switch similar’. When drug being stopped before or in the 90 days after initiation of the new lipid-lowering was used, which is the usual length of one prescription in the Netherlands. If patients drug, the modification was defined as an ‘addition’. An overlap period of at least 90 days had a period of 180 days in which they had no medication available this was defined as ‘discontinuation’. Treatment turbulence during follow-up was expressed as the number of treatment modifications divided by number of prescriptions x 100%.

TABLE 1. Dose-classification of different types and dosings of lipid-lowering treatment

Lipid-lowering drug: Low-dose Standard-dose Intensive-dose

Fluvastatin All doses - -

Pravastatin ≤ 40 mg >40 mg -

Simvastatin ≤ 20 mg >20 mg - ≤60 mg >60 mg

Atorvastatin ≤ 10 mg >10 mg - ≤30 mg >30 mg

Rosuvastatin - ≤ 10 mg >10 mg

Lipid-lowering drugs other than statins were classified in the low-dose treatment group.

107 CHAPTER 5

Adherence measurement Adherence was estimated over the two years of follow-up. It was calculated as Proportion of Days Covered (PDC), which expresses the proportion of days for which a patient has received medication in the study period [22] adherent, patients with a PDC < 80% as non-adherent [23]. . Patients with a PDC ≥80% were classified as

LDL-cholesterol response All LDL-cholesterol levels are based on the Friedewald equation [24]. At statin initiation was the most recent measurement in the 180 days before or at statin initiation. LDL- (baseline) we determined whether LDL-cholesterol was at target (≤2.5 mmol/l). This cholesterol response was determined after two years of follow-up, that is, using the measurement closest to 720 days after statin initiation within a period 540 and 900 days after initiation. Adequate treatment response was defined as either achieving the LDL-cholesterol, which is the expected decrease for standard-dose treatment [20,21]. target LDL-cholesterol level of 2.5 mmol/l or a decrease of at least 40% from baseline

Data analyses Descriptive statistics are presented for patients initiating on a low-dose and standard- dose. Groups were compared on baseline characteristics using chi-square tests, independent sample t-tests or Wilcoxon rank-sum tests. First and second treatment changes were described for patients initiating on low-dose or standard-dose treatment. Differences between patients initiating on low-dose and standard-dose in (1) proportion turbulence during two years of follow-up, and (3) LDL-cholesterol response after of first treatment modifications and discontinuation, (2) treatment adherence and two years follow-up were tested using chi-square tests, Wilcoxon rank-sum tests, or independent sample t-tests. Kaplan-Meier curves were used to present the time till the first treatment modification, separate for treatment intensification, reduction, switch low-dose versus a standard-dose was determined using log-rank tests. similar, and discontinuation. Significance of differences between patients initiating on a

RESULTS

In total 7,772 type 2 diabetes patients who initiated statin treatment between 2007 and 2012 were included in the analyses. Of these 86% started on simvastatin treatment; 69% started on simvastatin 40 mg which is the recommended dose in Dutch guidelines[3] (Figure 1). In total 1,776 patients initiated on a low dose compared to 5,842 that initiated on a standard dose. Patients that initiated on low-dose treatment were for example older (t-test; p-value <0.001), more often female (chi-square test; p-value

108 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

<0.001), had a longer diabetes duration (chi-square test; p-value <0.001), and better glucose regulation (t-test; p-value <0.001) compared to patients initiating on standard- dose treatment. There was no difference in LDL-cholesterol at baseline (t-tests; p-value 0.703), nor in the proportion of patients with a baseline LDL-cholesterol level at target (chi-square test; p-value 0.131) (Table 2).

Exclusion criteria Base population 10,349

Follow-up <720 days 2,324

Patients using daily packages 174

Prescription duration >270 days 79

Initiation dose Included patients 7,772 5 Pravastatin Simvastatin Atorvastatin Rosuvastatin Other drugs 178 (2.3%) 6,706 (86.3%) 523 (6.7%) 353 (4.5%) 12 (0.2%)

80 mg 80 mg 80 mg 4 15 0 (0%) (0.1%) (0.2%)

40 mg 40 mg 40 mg 40 mg 109 5,333 103 7 (1.4%) (68.6%) (1.3%) (0.1%)

20 mg 20 mg 20 mg 20 mg 50 1,194 187 24 (0.6%) (15.4%) (2.4%) (0.3%)

10 mg 10 mg 10 mg 10 mg 19 169 216 173 (0.2%) (2.2%) (2.8%) (2.2%)

5 mg 147 (1.9%)

Others Others Others Others 0 6 2 2 (0.0%) (0.1%) (0.0%) (0.0%)

Initiation dose: dark grey = intensive-dose; grey = standard-dose; light grey = low-dose

FIGURE 1. Flow chart of the patient selection and initiation dose of the patients

109 CHAPTER 5

TABLE 2. Characteristics of the total baseline population and patients classified by treatment initiation dose Variable Total population Standard-dose Low-dose p-value (n=7,772) (n=5,842) (n=1,776) Age, mean (SD) 62.1 (11.7) 61.5 (11.7) 63.9 (11.5) <0.001 Male sex 50.2% 51.7% 45.3% <0.001 SBP, mmHg (SD) 144.5 (19.8) 144.5 (19.9) 144.6 (19.1) 0.969 N 6,337 4,734 1,490 % missing 18.5 19.0 16.1 0.006 DBP, mmHg (SD) 82.8 (10.6) 83.1 (10.6) 82.2 (10.6) 0.008 N 6,336 4,734 1,489 % missing 18.5 19.0 16.2 0.007 Diabetes duration: ≤2 years (n) 65.7% (5,008) 68.8% (4,020) 55.6% (988) <0.001 2-10 years (n) 26.8% (2,039) 24.5% (1,434) 34.1% (605) >10 years (n) 7.5% (571) 6.6% (388) 10.3% (183) Fasting glucose, mmol/l (SD) 8.3 (2.9) 8.4 (3.1) 8.0 (2.5) <0.001 N 5,601 4,258 1,261 % missing 27.9 27.1 29.0 0.120 HbA1c % (SD) 7.3 (1.6) 7.4 (1.7) 7.1 (1.3) <0.001 N 6,288 4,728 1,459 % missing 19.1 19.1 17.8 0.249 LDL cholesterol, mmol/l (SD) 3.8 (0.9) 3.8 (1.0) 3.8 (0.9) 0.703 N 5,202 3,904 1,236 % missing 33.1 33.2 30.4 0.029 > target (n) 91.5% (4,760) 91.4% (3,570) 92.8% (1,147) 0.131 <= target (n) 8.5% (442) 8.6% (334) 7.2% (89) HDL cholesterol, mmol/l (SD) 1.19 (0.33) 1.18 (0.33) 1.22 (0.33) 0.002 N 5,298 3,995 1,237 % missing 31.8 31.6 30.3 0.313 Triglycerides, mmol/l (IQR) 1.9 (1.4-2.7) 2.0 (1.4-2.8) 1.9 (1.3-2.6) <0.001 N 5,332 4,013 1,251 % missing 31.4 31.3 29.6 0.171 Total cholesterol, mmol/l (SD) 5.9 (1.2) 5.9 (1.2) 5.9 (1.1) 0.183 N 5,186 3,921 1,199 % missing 33.3 32.9 32.5 0.757 Body mass index (SD) 30.6 (5.7) 30.6 (5.7) 30.6 (5.5) 0.805 N 3,786 2,877 849 % missing 51.3 50.8 52.2 0.287 Comorbidity: Macrovascular complications (n) 11.7% (911) 10.9% (634) 12.3% (219) 0.084 Microvascular complications (n) 3.6% (279) 3.3% (195) 4.3% (76) 0.061 DBP: diastolic blood pressure; IQR: inter quartile range; SBP: systolic blood pressure; SD: standard deviation.

110 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

Treatment changes and standard-dose treatment are shown in Figure 2. More than half of the patients The first two treatment changes during follow-up for patients initiating on low-dose that initiated on a low-dose either did not have any treatment change during follow- up (n=865, 48.7%) or did not receive a dose increase (n=110, switch similar without dose increase; n=11, dose decrease without dose increase). Almost 15% received a dose increase, including around 9.0% without further change (n=160) and 2.3% with a subsequent switch or second dose increase (n=40). Furthermore, more than 22% discontinued treatment either without a restart (n=330, 18.6%) or after a switch to a similar low-dose statin (n=36, 2.0%) or a dose increase (n=35, 2.0%). Patients restarting treatment after discontinuation (n=176, 9.9%) mostly restarted with a similar low-dose statin (n=139). Of the patients that initiated on a standard dose, the majority did not have a treatment change (n=3,236, 55.4%) or switched to a similar dose statin without subsequent change 5 (n=186, 3.2%). More than 8% received a dose decrease without subsequent dose increase (n=499, 8.5%). Less than 3% received a dose increase, either as first change without subsequent decrease or discontinuation (n=129, 2.2%) or after a switch to a similar statin (n=30, 0.5%). Furthermore, almost 19% discontinued treatment either without a restart (n=903, 15.5%) or after a dose decrease (n=127, 2.2%) or switch similar or dose increase (n=74, 1.3%). Patients restarting treatment after discontinuation (n=606, 10.4%) mostly received a similar standard-dose statin (n=469) or a low-dose statin (n=118). The percentage of patients that did not have any treatment change during follow-up was lower in the patients initiating on low-dose compared to standard-dose (48.7% versus 55.4%, chi-square test p-value <0.001). Patients initiating on low-dose were more likely to discontinue treatment (28.5% versus 25.8%, chi-square test p-value 0.026), and to or a switch to a similar dose category (7.3% versus 5.6%, chi-square test p-value 0.009) receive a treatment intensification (14.6% versus 2.9%, chi-square test p-value <0.001) patients on standard-dose (10.2% versus 0.8%, chi-square test p-value <0.001). Also, as first treatment change. As could be expected, a reduction was more common in the time to first treatment reduction was shorter for patients starting on standard- initiating on low-dose (Figure 3a/b) dose treatment, whereas the time to treatment intensification was shorter for patients (Figure 3c/d). In both groups, most . The time to the first switch or to discontinuation were significantly shorter in the low-dose group switches and discontinuations occurred within the first 180 days after treatment initiating on low-dose during follow-up, which was mainly caused by the difference in initiation. Furthermore, there was significantly more treatment turbulence for patients

111 CHAPTER 5

the amount of patients with no treatment change (67.3% for low-dose versus 70.8% for standard-dose) (chi-square test p-value 0.004).

Start First change Second change

No change* 55.4% (3,236)

Decrease* 10.2% (598) No change 53.7% (321) Decrease 1.2% (7) Dose decrease 3.5% (205) Increase 16.6% (99) Switch similar 7.4% (44) Switch decrease 6.7% (393) Discontinuation 21.2% (127)

Increase* 2.9% (170) No change 62.9% (107) Decrease 14.1% (24) Dose increase 1.0% (56) Increase 5.3% (9) Treatment start (5,842) Switch increase 1.3% (77) Switch similar 7.6% (13) with standard-dose statin Discontinuation 10.0% (17) Addition 0.6% (37)

No change 56.5% (186) Decrease 11.6% (38) Switch similar* 5.6% (329) Increase 9.1% (30) Switch similar 5.5% (18) Discontinuation 17.3% (57)

No restart 59,8% (903) Restart with: Discontinuation* 25.8% (1,509) Decrease 7.8% (118) Increase 1.3% (19) Switch similar 31.1% (469) Same statin 0% (0)

No change 48.7% (865)

Decrease 0.8% (15) No change 46.7% (7) Decrease 0.0% (0) Dose decrease 0.8% (15) Increase 26.7% (4) Switch similar 13.3% (2) Switch decrease not possible Discontinuation 13.3% (2)

Increase 14.6% (260) No change 61.5% (160) Decrease 9.6% (25) Treatment start (1,776) Dose increase 9.4% (167) Increase 10.0% (26) with low-dose statin Switch increase 4.7% (84) Switch similar 5.4% (14) Discontinuation 13.5% (35) Addition 0.5% (9) No change 45.4% (59) Decrease 0.8% (1) Switch similar 7.3% (130) Increase 15.4% (20) Switch similar 10.8% (14) Discontinuation 27.7% (36)

No restart 65.2% (330) Restart with: Discontinuation 28.5% (506) Decrease 0% (0) Increase 7.3% (37) Switch similar 27.5% (139) Same statin 0% (0)

FIGURE 2. First and second treatment changes for patients initiating on low-dose and standard-dose statin treatment

112 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

low-dose standard-dose FIGURE 3. Days till first treatment modification or discontinuation according tothe treatment initiation dose. Part a. shows results when the first modification is an intensification; Part b. when the first modification is a reduction; Part c. when the first modification is a switch to a similar dose; Part d. shows the time till discontinuation.

Adherence and LDL-cholesterol achievement The adherence rate during follow-up was slightly lower in the low-dose group (PDC: median 83%; IQR 46-96) compared to the standard-dose group (PDC: median 86%; IQR 52-97) (Wilcoxon-test; p-value <0.001) (Figure 4). More than 80% of adherent patients without a change in treatment had an adequate LDL-cholesterol response, whereas this was just over 70% for adherent patients with a treatment change (Table 3). For patients without treatment change, regardless of the adherence level, the LDL-cholesterol group (Table 3). For adherent patients with a treatment change, the LDL-cholesterol response was significantly better for the standard-dose group in comparison to low-dose was similar for both groups. The poorest response was seen for non-adherent patients with at least one treatment change during follow-up, with only 40% of patients initiating on low-dose and 52% initiating on standard-dose treatment having an adequate LDL- cholesterol response (Table 3) change, almost half had an adequate LDL-cholesterol response after 2 years (44.0% for . Of the patients with a discontinuation as first treatment low-dose and 49.8% for standard-dose).

113 CHAPTER 5

Overall, of the 502 patients initiating on low-dose treatment without an adequate LDL-cholesterol response after 2 years, 209 (42%) had a discontinuation and 128 (26%) showed poor adherence rates, whereas 108 (21%) had remained on a low dose while showing an adequate adherence rate and 57 (11%) had a change in treatment with adequate adherence (Table 3). In comparison, of the 1179 patients initiating on standard-dose treatment without an adequate LDL-cholesterol response, 543 (46%) had a discontinuation, 284 (24%) showed poor adherence rates, 217 (18%) had remained on the same dose and 135 (11%) had a change in treatment with adequate adherence.

TABLE 3. LDL-cholesterol level and response 2 years after treatment initiation for adherent and non-adherent patients starting on low-dose or standard-dose treatment, and categorized by receiving no treatment change, treatment change or discontinuation Adherent Non-adherent Low-dose Standard-dose Low-dose Standard-dose No change: LDL-c missing (n) 23.5% (167) 21.8% (579) 28.1% (43) 24.2% (139) Mean LDL-c [SD] (n) 2.3 [0.7] (545) 2.1 [0.7] (2,082) 2.7 [0.9] (110) 2.4 [0.9] (436) LDL-c response + (n) 80.2%* (437) 89.6% (1,865) 63.6%* (70) 77.3% (337) LDL-c response – (n) 19.8% (108) 10.4% (217) 36.4% (40) 22.7% (99) Change: LDL-c missing (n) 13.1% (30) 17.7% (110) 16.5% (29) 19.3% (92) Mean LDL-c [SD] (n) 2.6 [0.9] (199) 2.4 [0.9] (511) 3.3 [1.1] (147) 3.1 [1.1] (384) LDL-c response + (n) 71.4% (142) 73.6% (376) 40.1%* (59) 51.8% (199) LDL-c response – (n) 28.6% (57) 26.4% (135) 59.9% (88) 48.2% (185) Discontinuation: LDL-c missing (n) - - 26.3% (133) 28.4% (428) Mean LDL-c [SD] (n) - - 3.2 [1.0] (373) 3.1 [1.0] (1,081) LDL-c response + (n) - - 44.0% (164) 49.8% (538) LDL-c response – (n) - - 56.0% (209) 50.2% (543) *significant difference between low-dose and standard-dose treatment. LDL-c: Low-density lipoprotein cholesterol; SD: standard deviation

114 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

FIGURE 4. Overall adherence calculated as proportion of days covered over the first two years of treatment, for patients initiating on low-dose and standard-dose 5 DISCUSSION

Around 22% of type 2 diabetes patients initiated statin treatment on a dose lower than recommended. More than half of them remained on a low dose during a 2-year follow- up period, whereas less than 15% received a dose increase. Of the patients initiating on standard-dose, more than half remained on the same treatment during this period, whereas 8% received a dose decrease without subsequent increase. More than 25% of patients starting on low-dose or on standard-dose treatment discontinued treatment, other hand, 35-40% of them restarted treatment within 720 days, often with a similar often within the first 180 days after initiation or after a first treatment change. On the dose statin. Patients that initiated on low-dose were older and more often female than patients initiating on standard-dose treatment. There was no difference in baseline LDL-cholesterol level. Patients that initiated on low-dose showed a higher treatment turbulence and received treatment changes earlier than patients initiating on standard- dose treatment. Patients on low-dose treatment had lower adherence levels and were less likely to have adequate LDL-cholesterol response compared to patients on standard- dose after 2 years follow-up. Discontinuation or poor adherence accounted for 68-70% of inadequate response, whereas no treatment change in adherent patients accounted for 18-21% of inadequate LDL-cholesterol response after 2 years follow-up.

Previous studies showed that more than half of the patients started statin treatment on a low dose [25,26]. In our population of diabetes patients this proportion was considerably lower but still substantial with 22%. Having only a moderately increased LDL-cholesterol 115 CHAPTER 5

level could be a reason to start on a low-dose. However, we observed no difference in baseline LDL-cholesterol between patients initiating on low-dose and standard-dose statin treatment. With an average baseline LDL-cholesterol level of 3.8 mmol/l, it can be on low-dose treatment [20,21]. After initiation on a low-dose upward titration should be expected that many of the patients will not reach the LDL-cholesterol target of 2.5 mmol/l the next step but this occurred in less than 15% of such patients. Others also reported [26,27]. The strategy to start with low-dose statin treatment and titrate treatment till the LDL-cholesterol target that treatment intensification was uncommon in clinical practice has been reached might be suboptimal. This strategy can delay effective treatment with 18 months or more [28]. Whereas a previous study found that treatment changes [26], we found that initiating on a low dose was associated with higher treatment turbulence, including and discontinuation were not influenced by the initial statin potency more and earlier treatment changes. Such increased treatment turbulence needed for dose titration could contribute to poorer patient adherence [29,30]. This is in accordance with the higher discontinuation and lower adherence rates we observed for patients starting on a low dose. Future studies using longitudinal modelling are needed to get better insight in the relationship between dosing, treatment turbulence and adherence, as well as interpersonal variation in treatment response.

Fear for adverse events could be another reason to initiate on low-dose statin. Muscle toxicity and effects on liver enzymes are well-acknowledged adverse events associated with statin treatment [31,32,33]. With higher doses there is an increased risk for statin- induced adverse events [31,32,33]. Patients initiating on a low dose in our study were older and more often female. Especially in patients vulnerable for adverse events, such as the elderly, physicians might be inclined to start on a low dose statin. However, patients that was generally similar to that of younger adults [31,34,35]. Experiencing adverse events up to the age of 80 years have been included in clinical trials, showing a safety profile during treatment can be a reason to become non-adherent or discontinue treatment [36]. We did not have information about adverse events but observed a lower adherence rate and higher discontinuation rate in patients that initiated on low-dose as compared to standard-dose treatment. This suggests that starting on a low dose did not prevent patients becoming non-adherent.

Treatment dosing and adherence affect the likelihood of reaching lipid targets [4,6,7]. This was also seen in our study. In addition, we observed that LDL-cholesterol response was worse for patients with treatment changes than for patients without treatment changes, regardless of the initial dose or adherence level. Treatment changes are more likely in patients that have not yet reached the LDL-cholesterol target level [26], but can also be made to reduce adverse events. Our study shows that such changes, especially in

116 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

patients who were non-adherent, did not lead to adequate LDL-cholesterol response in around half of the patients after 2-year follow-up.

Initiation on low-dose statin is a problem that is common in clinical practice [25,26] but little was known about the subsequent treatment modifications and outcomes. Our cholesterol response during 2 years of follow-up between patients that initiated on study is one of the first providing insight in differences in treatment patterns and LDL- low-dose and standard-dose statin treatment. We had detailed information about statin dosing and subsequent changes in treatment, enabling to describe various trajectories such as discontinuations after treatment changes and restart of treatment after discontinuation.

Our descriptive approach is a limitation, since we cannot draw conclusions about possible cholesterol target was reached or when a reduction in LDL-cholesterol of 40% was causal relationships. We defined LDL-cholesterol response as adequate when the LDL- achieved, thereby allowing for patients with very high baseline LDL-cholesterol levels 5 to have a response considered to be adequate. Around 23% of the patients, however, did not have an LDL-cholesterol measurement with no difference for patients starting on low-dose or standard-dose. For adherence measurement, we used the PDC which is known to overestimate actual adherence [37]. Finally, we did not have information about actual adverse event rates in our study population.

In conclusion, our study illustrates that current patterns of statin treatment in patients with type 2 diabetes are suboptimal, with discontinuations followed by inadequate adequate LDL-cholesterol response. Adherence levels are lower and discontinuation adherence levels and lack of treatment intensification seen in those who do not have rates are higher in patients starting on low-dose as compared to standard-dose treatment, and when restarting patients often do so on similar low-dose statins. These on low-dose statins. findings call for a closer look at the rationale of initiating patients with type 2 diabetes

117 CHAPTER 5

REFERENCES

detreatment Vries FM, for Kolthof the secondary J, Postma prevention MJ, Denig P,of Hak cardiovascular E. Efficacy ofand standard cerebrovascular and intensive events statin in diabetes patients: a meta-analysis. PLoS One 2014; 9(5): e111247. 3. NHG-standaard Cardiovasculair risicomanagement (Eerste herziening) Huisarts Wet 2012; 55(1): 14-28. 4. Heintjes EM, Penning-van Beest FJ, Plat AW, et al. Cholesterol level goal attainment with statins: clinical management guideline recommendations versus management in actual clinical practice. Pharmacotherapy 2012; 32(7): 631-641. 5. Sidorenkov G, Haaijer-Ruskamp FM, de Zeeuw D, Denig P. A longitudinal study examining

adherencetimeliness. toPLoS guidelines One 2011; in 6(9)diabetes :e24278. care according to different definitions of adequacy and 6. Geleedst-De Vooght M, Maitland-van der Zee AH, Schalekamp T, Mantel-Teeuwisse A, Jansen P. Statin prescribing in the elderly in the Netherlands: A pharmacy database time trend study. Drugs Aging 2010; 27(7): 589-596. 7. De Vries FM, Voorham J, Hak E, Denig P. Adherence to standard-dose or low-dose statin treatment and LDL-cholesterol response in type 2 diabetes patients. Curr Med Res Opin in press. 8.

Rodondihypertension, N, Peng dyslipidemia, T, Karter AJ, and et diabetes al. Therapy mellitus. modifications Ann Intern in Med response 2006; to144(7): poorly 475-484. controlled 9. Schmittdiel JA, Uratsu CS, Karter AJ, et al. Why don’t diabetes patients achieve recommended Journal of General

riskInternal factor Medicine targets? 2008; Poor 23(5):adherence 588-594. versus lack of treatment intensification. 10. -

Viranication forSS, elevatedWoodard cholesterol LD, Chitwood levels SS, in et patients al. Frequency with cardiovascular and correlates disease. of treatment Am Heart intensifi J 2011; 162(4): 725-732.e1. 11. Parris ES, Lawrence DB, Mohn LA, Long LB. Adherence to statin therapy and LDL cholesterol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care 2005; 28(3): 595-599. 12. Watanabe JH, Bounthavong M, Chen T. Revisiting the medication possession ratio threshold for adherence in lipid management. Curr Med Res Opin 2013; 29(3):175-180. 13.

14. AvailableOsterberg at: L, https://www.gipdatabank.nl/databank.aspBlaschke T. Adherence to medication. N Eng AccessedJ Med 2005; 8 April 353(5): 2015 487-497. 15. Pittman DG, Fenton C, Chen W, Haffner S, Pendergrass M. Relation of statin nonadherence and Am J Cardiol 2012; 110(10): 1459-1463.

treatment intensification. 118 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

16. Voorham J, Denig P. Computerized extraction of information on the quality of diabetes from text in electronic medical patient records of general practitioners. J Am Med Inform Assoc 2007; 14(3): 349-354. 17. Lamberts H, Wood M (eds). International Classification of Primary Care (ICPC). Oxford: Oxford University Press; 1987 18.

19. ATC/DDDVoorham J, index. Haaijer-Ruskamp Available at: FM,http://www.whocc.no/atc_ddd_index/ Wolffenbuttel BHR, Stolk RP, Denig P. Accessed Medication 8 April adherence 2015 Clin Ther 2011; 33(1):

affects121-134. treatment modifications in patients with type 2 diabetes. 20. Smith MEB, Lee NJ, Haney E, Carson S. Drug class review: Hmg-coa reductase inhibitors (statins) and fixed-dose combination products containing a statin: Final report update 5. Portland (OR); 2009 21. Law MR, Wald NJ, Rudnicka AR. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: systematic review and meta-analysis. BMJ 2003; 326(7404): 1423. 22. Peterson AM, Nau DP, Cramer JA, Benner J, Gwadry-Sridhar F, Nichol M. A checklist for 5 medication compliance and persistence studies using retrospective databases. Value Health. 2007; 10(1): 3-12. 23. Simpson RJ, Mendys P. The effects of adherence and persistence on clinical outcomes in patients treated with statins: A systematic review. J Clin Lipidol 2010; 4(6): 462-471. 24. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972; 18(6): 499-502. 25.

Willeylipoprotein JV, Bullanocholesterol MF, goal Shoetan attainment NN, Gandhi rates associated SK. Therapy with modifications the initiation of and generic low-density simvastatin. Curr Med Res Opin 2010; 26(1): 121-128. 26. Simpson RJ, Tunceli K, Ramey DR, et al. Treatment pattern changes in high-risk patients newly initiated on statin monotherapy in a managed care setting. J Clin Lipidol 2013; 7(5): 399-407. 27. Fox FM, Gandhi SK, Ohsfeldt RL, Blasetto JW, Davidson MH. Titration patterns with rosuvastatin as compared with other statins in clinical practice: A retrospective observational cohort study using an electronic medical record database. Clin Ther 2007; 29(11): 2385-2394. 28. Caspard H, Chan AK, Walker AM. Determinants of the differences in LDL-cholesterol after initiation of statin treatment. Ann Pharmacother 2006; 40(1): 21-26. 29. Düsing R. Adverse events, compliance and changes in therapy. Curr Hypertens Rep 2001; 3(6): 488-492. 30. Schuster H. Improving lipid management – to titrate, combine or switch. Int J Clin Pract 2004; 58(7): 689-694. 31. Armitage J. The safety of statins in clinical practice. Lancet 2007; 370(9601) :1781-1790.

119 CHAPTER 5

32. Silva M, Matthews ML, Jarvis C, et al. Meta-analysis of drug-induced adverse events associated with intensive-dose statin therapy. Clin Ther 2007; 29(2): 253-260. 33. Newman C, Tsai J, Szarek M, Luo D, Gibson E. Comparative safety of atorvastatin 80 mg versus 10 mg derived from analysis of 49 completed trials in 14,236 patients. Am J Cardiol 2006; 97(1): 61-67. 34.

Baigentprospective C, Keech meta-analysis A, Kearney of PM,data et from al. Efficacy90 056 participantsand safety of in cholesterol-lowering 14 randomised trials treatment: of statins. Lancet 2005; 366(9493): 1267-1278. 35.

Collinslowering R, Armitagewith simvastatin J, Parish in S, Sleight20,536 P,high-risk Peto R. MRC/BHF individuals: Heart a randomised Protection Studyplacebo-controlled of cholesterol trial. Lancet 2002; 360(9326):7-22. 36. Wei MY, Ito MK, Cohen JD, Brinton EA, Jacobson TA. Predictors of statin adherence, switching, and discontinuation in the USAGE survey: Understanding the use of statins in America and gaps in patient education. J Clin Lipidol 2013; 7(5): 472-483. 37. Vink NM, Klungel OH, Stolk RP, Denig P. Comparison of various measures for assessing Pharmacoepidemiol Drug Saf 2009;

medication18(2): 159-165. refill adherence using prescription data.

120 Statin prescribing patterns, adherence and LDL-cholesterol response in diabetes

121

CHAPTER 6

Does a cardiovascular event change adherence to statin treatment in patients with type 2 diabetes? A matched cohort design

Folgerdiena M. de Vries1 Petra Denig2 Stefan Vegter1 H. Jens Bos1 Maarten J. Postma1 Eelko Hak1

Current Medical Research & Opinion 2015;31(4):595-602.

ABSTRACT

Objective: To be effective, adherence to statin treatment is essential. We assessed

2 diabetes patients. the effect of an apparent first cardiovascular event on statin adherence rates in type

Methods: A matched cohort study was conducted among type 2 diabetes patients initiating statin treatment for primary prevention in the Groningen University IADB.nl pharmacy database. Patients who had a drug-treated cardiovascular event (index date) after statin initiation were matched to a reference patient without such an event with similar gender, age at statin initiation, initiation date, follow-up period and adherence level before the event. Adherence rates were measured as

Proportion of Days Covered (PDC), and shifts in adherence levels (non-adherent/ partial-adherent/full-adherent) and rates around the event were evaluated. Results: We could match 375 of the 855 eligible index patients to a reference patient. Index patients had on average a PDC of 81% after the index date; reference patients had a PDC of 71% (P<0.001) while both had a PDC of 79% before the index date. Index patients were 4.5 times more likely than reference patients to shift from non-adherent to full-adherent [95%CI 1.1-18.8] and 1.8 times more likely to shift from partial-adherent to full-adherent [95%CI 1.2-2.6]. In the index group, 26%

20% of patients became less adherent. of patients became more adherent after the first cardiovascular event. In contrast,

Limitations: Medication proxies were used, which could have caused

matched to a reference patient due to small ranges in matching criteria. misclassification. Furthermore, a substantial group of index patients could not be

Conclusions: The occurrence of a drug-treated cardiovascular event appeared to avert the declining statin adherence rate observed in diabetes patients without

such an event. On the other hand, one in five patients became less adherent after the event, indicating that there are still important benefits to achieve.

124 Cardiovascular events and adherence to statins in diabetes

INTRODUCTION

Cardiovascular disease is one of the most common causes of death [1]. Diabetes patients are at increased risk for developing cardiovascular disease [2]. The use of statins is associated with a reduction in the risk of cardiovascular events in diabetes patients with and without previous cardiovascular events [3,4]. To be effective, however, patients need to adhere to the advised dose regimens. The high adherence rates as accomplished in a clinical trial setting may in practice be much lower which reduces the health and economic impact of such preventive medication [5]. Having a cardiovascular history or a high risk for cardiovascular complications is associated with higher adherence rates than having no cardiovascular history or a low risk for cardiovascular complications [6-9].

Despite this difference between statin adherence rates in primary and secondary prevention, not much is known about adherence patterns of patients who experience common since the population for primary prevention has been extended in most a first cardiovascular event while using statins. This occurrence is becoming more Western countries, but treatment cannot prevent cardiovascular events in all patients [10,11]. Poluzzi et al. [12] reported in a sub-analysis of a cohort study that 53% of the patients on statin treatment used more statins after they experienced a cardiovascular 6 that more patients increased their use might be expected as secondary prevention is event, whereas 26% used the same and 21% used less than before the event. The finding associated with higher adherence rates. However, other investigators reported that the occurrence of a cardiovascular event during statin treatment was associated with lower adherence rates [8,9]. The perception that the drug had been ineffective was mentioned as a possible explanation. These three studies did not distinguish between a first or subsequent cardiovascular event, which might explain the disparate findings. prevention patients. None of the studies elucidate the effect of a first event on the adherence rate in primary

We designed a matched cohort study to assess the causal effects of the occurrence of patients. This design was chosen to reduce the potential for confounding in cohort a first drug-treated cardiovascular event on statin adherence rates in type 2 diabetes studies.

125 CHAPTER 6

METHODS

Study design An index-reference matched cohort study was conducted among type 2 diabetes patients initiating statin treatment for primary prevention (Figure 1). Based on their medication, after initiating statin treatment, whereas the reference group consisted of patients who patients in the index group were identified as having experienced a cardiovascular event did not experience such an event after initiating statin treatment. Matching was done on potential confounders for the association between cardiovascular event and adherence rates (see Study population).

Type 2 diabetes patients initiating statin treatment for a primary objective in both index and

reference groups. Index group patients experience a cardiovascular event at T1, time between 0T

and T1 should be at least one year, and follow up after 1T at least 180 days. Patients in the reference group are followed as long as their matched index patient, reference patients will not experience a cardiovascular event during follow up. FIGURE. 1 Study design.

Setting This cohort study was performed with the University of Groningen IADB.nl pharmacy prescription database which contains prescription data from 1994 until 2011 and covers an estimated population of 500,000 persons in the Netherlands. Age, gender, and prescription rates among the database population have found to be representative for the Netherlands as a whole, and the database is widely researched [13-15]. Each prescription record contains information on the date of dispensing, the quantity dispensed, the dose regimen, the number of days the prescription is valid, the prescribing physician, and the Anatomical Therapeutic Chemical code (ATC code) [16]. Each patient has a unique pharmacy commitment in the Netherlands, the medication records of each patient are anonymous identifier; date of birth and gender are known. Due to the high patient- virtually complete, except for over the counter (OTC) drugs and medication dispensed

126 Cardiovascular events and adherence to statins in diabetes

during hospitalization [17]. Data between January 2001 and December 2011 were used for the analyses.

According to the Code of Conduct for Health Research (Dutch FMWV Code approved by Dutch Data Protection Authority in 2004), no ethics committee approval is needed for research using anonymous medical records that do not lead to questions from the point of view of privacy.

Study population Individuals initiating statin treatment (ATC codes C10AA, C10BA, C10BX) between 2001 and 2011 were selected from the IADB.nl database. Patients were only included in the study when they were present in the database for at least one year before initiation of statin treatment. Type 2 diabetes patients were selected by the use of two prescriptions of non-insulin blood-glucose lowering drugs (A10B) [18] within the time-interval of 180 days before and 180 days after statin initiation. Patients on monotherapy of insulin were not included, as they are mainly type 1 diabetes patients. This implies that around 6% of type 2 diabetes patients that are exclusively treated with insulin were excluded from the study [19] of cardiovascular and cerebrovascular medication proxies (Appendix II . Identification of primary prevention diabetes patients was done by the use 6 of primary and secondary prevention patients with type 2 diabetes in a prescription : Identification database) [18]. Diabetes patients who had received at least two prescriptions for vitamin K antagonists (B01AA) or thrombocyte aggregation inhibitors (B01AC) within the interval of 180 days before and 180 days after statin initiation were excluded, as they were considered to be secondary prevention patients. The occurrence of a first cardiovascular event (T1) at the index date was defined by start of medication with a high specificity and sensitivity for indicating such events, i.e. a treatment start for Appendix II). Reference patients did not start thrombocyte aggregation inhibitors but thrombocyte aggregation inhibitors (≥2 prescriptions for B01AC within one year, see were allowed to use other cardiovascular medication. For calculating adherence rates, complete medication follow-up after the index date (T1) needed to last for at least 180 days. To enable a reliable calculation of adherence rate before the event, patients that shifted to secondary prevention within one year after statin initiation were excluded.

All index patients who started thrombocyte aggregation treatment were preliminary gender, date of statin initiation (between -180 and +180 days), and follow-up period. matched with potential reference patients on age at statin initiation (+/-1 year of age), By requiring a medication follow-up period for reference patients at least as long as their matched index patient we prevented so-called immortal time bias [20]. Next, the adherence level (see further) was measured until the occurrence of the cardiovascular

127 CHAPTER 6

event for index patients and in case of the reference patients for the same follow-up period as the matched index patient (see also Figure 1). Subsequently, all index patients were matched on all four criteria to one reference patient for the analysis.

Adherence duration of the prescription was estimated using the number of tablets dispensed and All statin prescriptions (ATC codes C10AA, C10BA, C10BX) were identified. The expected the prescribed number of tablets used per day. Patients were allowed to switch to another statin or another dosage. Adherence rates were measured as the Proportion of Days Covered (PDC) which is calculated as the number of days the patient has drugs

100 [21] available divided by the number of days within the specific time period multiplied by partial-adherent (level 2: 20%-80% PDC), or full-adherent (level 3: >80% PDC) [5]. . Patients were classified into three groups: non-adherent (level 1: <20% PDC),

Adherence rates were calculated for two periods: (1) from initiation of statin treatment until the occurrence of a cardiovascular event for index patients (T0-T1), and in case of the reference patients for the same follow-up period as their matched index patient; (2) from the time of the occurrence of a cardiovascular event until the end of follow-up for index patients which is the last date the patient was active within the IADB.nl database. Again we assessed the adherence rates for reference patients over the same follow-up period as their matched index patient.

Data analysis Descriptive statistics were used to compare matched with unmatched index patients, and index with reference patients regarding baseline characteristics and duration of follow-up. Furthermore, index and reference patients were compared regarding their use of RAS inhibitors (C09), beta-blockers (C07A), calcium antagonists (C08) and nitrates (C01DA) within the interval of 90 days before and 90 days after the start of until 180 days after T1 using Chi-square tests. statin treatment (T0) and from the occurrence of the first cardiovascular event (T1) index patients were compared using Wilcoxon signed rank tests. The mean adherence Adherence rates before and after the first cardiovascular event (T1) for reference and compared with that of the reference patients for the same follow-up period as their rate of the index patients after the occurrence of a first cardiovascular event (T1) was matched index patient by the use of Mann-Whitney U Tests. In addition, relative risks

(RR) with 95% confidence interval (95% CI) of shifting to another adherence level after the event/index date (T1) were compared between the index and the reference group.

128 Cardiovascular events and adherence to statins in diabetes

The amount of patients that shifted to a higher or lower adherence rate or stayed within difference of more than 10% PDC before and after the index date. The statistical the same adherence range were calculated. A shift was defined as having an absolute of Chi-square tests. significance of shifts were compared between index and reference patients by the use

Sensitivity analysis The adherence rate before the event could have been underestimated when patients are hospitalized for a period of several weeks. In such cases, our proxy for a drug-treated have occurred earlier in time leading to unreliable adherence rate calculations before event would lead to a misclassification regarding the date, where the actual event would the event. Therefore, we performed a sensitivity analysis excluding patients that may have been hospitalized for a long period. Index patients and their matched reference patients that had no prescription in the period of 90 to 30 days before the event date were excluded, as an indicator for patients not active in the database due to a long period of hospitalization.

RESULTS 6

Study Population reference patients. Patient characteristics of the matched and unmatched index patients We identified 855 index patients of whom 375 could be matched to one of the 5,829 and the reference patients are listed in Table 1. There were no substantial differences between the matched index patients and the unmatched index patients concerning age and gender distribution at baseline. Matched index patients were on average 62 years of age when initiating statin treatment and 49% was male. Within the reference group the use of calcium antagonists (Chi-square test, p-value 0.01) and nitrates (Chi-square test, p-value 0.01) around initiation of statin treatment was slightly lower compared to the matched index group. As expected, cardiovascular co-medication use was significantly apparent cardiovascular event. higher in the index group than in the reference group after the occurrence of the first

Adherence rates The mean PDC before the index date T1 was 79% for both the index group and the reference group (Mann-Whitney U Test, p-value 0.92, see Table 1). After the occurrence of an apparent event (T1), the mean PDC for the index group changed to 81% (Wilcoxon signed ranks test, p-value 0.032) a decrease to 71% (Wilcoxon signed ranks test, p-value <0.001) was reported in the reference group, see Table 2

. There was a significant 129 CHAPTER 6

difference between reference and index patients after T1 (Mann-Whitney U Test, p-value <0.001).

Adherence shifts Shifts in adherence levels before and after the index date T1 with the relative risks and 95% CI are shown in Table 3. Index patients were more likely to become full-adherent after the occurrence of a cardiovascular event (T1) compared to their reference patients. The relative risk to shift from being non-adherent (level 1) to full-adherent (level 3) was 4.5 (95% CI 1.1-18.8), and from partial-adherent (level 2) to full-adherent (level 3) was 1.8 (95% CI 1.2-2.6) for the index group. Reference patients were more likely to become non-adherent (level 1) after the index date (T1). The relative risk of remaining non- adherent (level 1) was 0.4 (95% CI 0.2-0.9), and the shift from partial-adherent (level 2) to non-adherent (level 1) was 0.5 (95% CI 0.3-0.8) for the index group.

When analyzing the absolute PDC adherence rates, 53% (n=200) of the index patients PDC difference, before and after the index date. Index patients shifted more often to a and 54% (n=203) of the reference patients had a similar adherence rate, i.e. 10% or less higher PDC adherence rate (26%, n=97), whereas reference patients shifted more often to a lower adherence rate (30%, n=111) (chi-square tests; p-value <0.05). Since the partial-adherence level has a wide range in PDC (20-80%), we explored whether the distributions within this partial-adherence group differed between the difference (Mann Whitney U test, p-value 0.89). index and reference group before the index date (T1), which showed no significant

Sensitivity analysis Of the index patients, 34 did not have a prescription in the period 90 to 30 days before the event date. After exclusion of these patients and their matched reference patients, the mean PDC was 80% before the event and 81% after the event for the index patients. For reference patients, the PDC was 79% before the event and 72% after the event. Index patients were more likely to shift from non-adherent to full-adherent (RR 1.4; 95% CI 1.0-2.0) and to shift from partial-adherent to full-adherent (RR 1.4; 95% CI 1.1-1.8). Reference patients were more likely to stay non-adherent (RR 0.5; 95% CI 0.3-1.0) and to shift from partial-adherent to non-adherent (RR 0.8; 95% CI 0.6-1.0). 20% of the patients decreased their use after the event.

130 Cardiovascular events and adherence to statins in diabetes

TABLE 1. Baseline characteristics of the index and reference patients Total index Unmatched Matched Reference group (n=855) index (n=480) index (n=375) (n=375)

Male (%) 51% 53% 49% 49% Age (mean, SD), years 63 (11.0) 63.6 (11.2) 62 (10.7) 62 (10.7) Follow-up, years 6.7 (2.1) 7.3 (1.7) 5.8 (2.1) 5.8 (2.1) Time until T1, years 3.4 (1.9) 3.6 (2.0) 3.0 (1.8) Time after T1, years 3.3 (2.1) 3.7 (2.2) 2.8 (1.8) Year of initiation, n 2001-2005 629 413 216 225 2006-2010 226 67 159 150

Adherence level before index date 1: Non-adherent (≤20%), % 11% 14% 7% 7% (n, mean percentage of days (93, 7.1%) (68, 7.8%) (25, 5.2%) (25, 8.3%) covered) 2: Partial-adherent (20-80%), 25% 25% 25% 25% % (n, mean percentage of days (213, 53%) (120, 54%) (93, 52%) (93,52%) covered) 3: Full-adherent (≥80%), % 64% 61% 69% 69% 6 (n, mean percentage of days (549, 96%) (292, 96%) (257, 96%) (257, 95%) covered) Mean percentage of days 75.5% 72.8% 79.0% 79.0% covered, (IQR) (60.3%-99.1%) (51.8%-98.9%) (68.2%-99.3%) (67.5%-99.2%) Co-medication at initiation:# RAS inhibitors (C09) 45% (386) 45% (217) 45% (169) 42% (156) Beta blockers (C07A) 23% (200) 23% (108) 25% (92) 21% (79) Calcium antagonists (C08) 15% (125) 13% (62) 17% (63) 10% (39)* Nitrates (C01DA) 2% (21) 2% (8) 3% (13) 1% (3)* Co-medication at T1:≠ RAS inhibitors (C09) 68% (581) 67% (322) 69% (259) 57% (212)* Beta blockers (C07A) 57% (484) 56% (267) 58% (217) 26% (98)* Calcium antagonists (C08) 27% (231) 27% (128) 27% (103) 15% (56)* Nitrates (C01DA) 18% (158) 19% (91) 18% (67) 1% (2)* IQR = inter quartile range #at least one prescription between 90 days before and 90 days after statin initiation

≠ at least one prescription between 1T and 180 days after 1T *p-value<0.05; comparing matched index patients with reference patients

T1: the occurrence of the first cardiovascular event of the index patients

131 CHAPTER 6

TABLE 2. Adherence level after the index date for index and reference patients Adherence level after Total index group Unmatched index Matched index Reference event/index date: (n=855) (n=480) (n=375) (n=375) 1: Non-adherent (≤20%) %, 11% 12% 10% 18% (n, mean PDC) (96, 3.7%) (58, 2.7%) (38, 5%) (66, 2.6%)

2: Partial-adherent (20-80%) %, 17% 16% 18% 19% (n, mean PDC) (144, 57%) (77, 58%) (67, 56%) (72, 53%)

3: Full-adherent (≥80%) %, 72% 72% 72% 63% (n, mean PDC) (615, 97%) (345, 97%) (270, 98%) (237, 96%)

Mean PDC, (IQR) 80% 79% 81% 71% (75.2%-100%) (75.3%-100%) (72.0%-100%) (49%-100%) IQR: inter quartile range; PDC: proportion of days covered.

TABLE 3. The relative risk (RR) and 95% confidence interval (CI) to shift from one adherence level to another Level before T1 Level after T1 Index (n) Reference (n) RR CI (95%)

Non-adherent Non-adherent 7 16 0.44 0.22-0.88

Non-adherent Partial-adherent 9 7 1.29 0.57-2.90

Non-adherent Full-adherent 9 2 4.50 1.08-18.77

Partial-adherent Non-adherent 21 39 0.54 0.34-0.84

Partial-adherent Partial-adherent 24 27 0.89 0.56-1.42

Partial-adherent Full-adherent 48 27 1.78 1.22-2.58

Full-adherent Non-adherent 10 11 0.91 0.44-2.27

Full-adherent Partial-adherent 34 38 0.89 0.58-1.37

Full-adherent Full-adherent 213 208 1.01 0.94-1.09

T1: the occurrence of the first cardiovascular event of the index patients

132 Cardiovascular events and adherence to statins in diabetes

DISCUSSION

This matched cohort study showed that mean statin treatment adherence rate of type 2 diabetes patients before and after a drug-treated cardiovascular event had a small increase in adherence, while among reference patients statin treatment adherence rates declined. Index patients had an average adherence rate of 81% after the occurrence of an apparent cardiovascular event (T1), whereas reference patients had a mean adherence rate of 71% when considering the same follow-up period, while both had mean adherence rates of 79% before the index date (T1). Index patients were more likely to shift from being non-adherent or partial-adherent to full-adherent, whereas reference patients were more likely to stay non-adherent or to shift from being partial-adherent to non-adherent. Still, 20% of the index patients became less adherent after the occurrence of a first event. Our results show that a drug-treated cardiovascular event averts the decline in adherence patients that increased than decreased in their adherence to treatment after an event rates for statins seen in patients without such an event. The finding that there were more was in line with previous results from Poluzzi et al. [12]. This previous study did not focus on diabetic patients and reported that most patients increased statin adherence after 6 the occurrence of a cardiovascular event, whereas in our study most patients did not change their adherence. This difference could be the result of including also very small changes in the adherence calculation before and after the event as a decrease or increase in adherence in the previous study, whereas our criteria of change were based on an absolute change in PDC of more than 10%. Other investigators [8,9] did not adjust for adherence rates before the event which could lead to confounding by indication. Also, and included both patients in primary and secondary prevention. Thus, it may be that previous studies did not distinguish between first or subsequent cardiovascular events study- but may decrease after subsequent cardiovascular events. adherence to statin treatment does not decrease after a first event -as shown in our

The percentage of patients becoming less adherent after an event in our study was around 20%, similar as found in a previous study [12]. This amount of patients is lower than compared to patients who did not experience an event but is still a matter of concern. after an event is worrying. This suggests that patients might discontinue treatment after Especially our finding that 10% of patients had an adherence rate of less than 20% PDC a cardiovascular event, which has shown to increase mortality rates [22]. To address this problem better in clinical practice, the underlying reasons for this phenomenon need to be studied. Furthermore, it is important to look whether similar patterns of non- adherence are seen for other medications. What becomes clear from our study, is that we

133 CHAPTER 6

should tailor the approach to managing individual patients after a cardiovascular event. We cannot assume that all patients will be motivated to keep using their statin treatment after such an event, importantly these patients are at increased risk [23].

A strength of our study is that we analysed the effect of the transition from primary to secondary prevention on statin treatment adherence. This transition is important because of the difference between adherence rates in primary and secondary prevention patients. We adjusted for possible bias and confounding by indication, and corrected for immortal time bias by requiring a follow-up period for reference patients to be as long as their matched index patient [20]. Furthermore, adherence rates in our study were not affected by increased medication costs for patients after a cardiovascular event, since drug costs are fully reimbursed in the Netherlands, as is not the case in many other healthcare systems where patients are confronted with co-payments [8].

The fact that we had to use medication proxies to identify cardiovascular events is a clear limitation to our study. Nevertheless, the use of medication proxies to identify patients [7,24]. with specific diseases or events is common when reliable diagnostic coding is lacking We tested the medication proxies for sensitivity and specificity, however, misclassification of the event on the adherence rates and biased the effects towards the null. Further, could have occurred. Such misclassification would lead to underestimating the impact exclusively treated with insulin were thereby excluded, which might have resulted in type 2 diabetes was identified by the use of oral anti-diabetic medication. Patients excluding some of the more severe cases. Patients using oral anti-diabetic medication for other indications could not be excluded. As the use of medication proxies could have linking event data with diagnostic codes to pharmacy dispensing records which would caused some misclassification, we suggest for future research to conduct the analysis probably result in stronger associations. Furthermore, we were not able to include data on medication dispensed during hospitalization which could have underestimated our adherence rate calculations in the index group. Excluding patients that appeared not to be active in our database, possibly due to a long period of hospitalization, indeed of shifting to higher adherence levels. It did, however, not lead to relevant changes in decreased the point estimate as well as the confidence intervals for the relative risks possible concern is that, a substantial group of index patients could not be matched adherence rates nor in the direction or significance of our findings. Another matter of to a reference patient due to the small ranges in our matching criteria. Determining these ranges is balancing between the inclusion of more patients on the one hand and introducing more bias on the other hand. The use of strict matching criteria is preferred to prevent the introduction of bias due to poor matching. Matched and unmatched index patients were relatively similar in terms of age, gender, co-medication use and adherence

134 Cardiovascular events and adherence to statins in diabetes

levels after the index date. Finally, we did not differentiate between non-adherence and non-persistence, which were both included in the calculation of the adherence level. For the purpose of our study there is no need to distinguish between these two types of non-adherence, since we aim at comparing statin drug-taking behaviour over the entire follow-up period.

CONCLUSION

In type 2 diabetes patients the occurrence of a drug-treated cardiovascular event during statin treatment is likely to avert the decline in adherence observed in patients without such an event. Since one in five of the patients with an event became less adherent, pharmaceutical management of individual patients. there are still important health and economic benefits to achieve by tailoring the

135 CHAPTER 6

REFERENCES

World Health organisation. Global Health Observatory 2012. Available at: http://www.who. 2. int/gho/mortality_burden_disease/en/American Heart Association. Cardiovascular Accessed disease 8 September and diabetes 2014 [online]. Available at: -

http://www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetesMatters/Cardiovas 3. cular-Disease-Diabetes_UCM_313865_Article.jspde Vries FM, Denig P, Pouwels KB, Postma MJ, Hak Accessed E. Primary 8 September prevention 2014 of major cardiovascular and cerebrovascular events with statins in diabetic patients: A meta-analysis. Drugs 2012; 72(18): 2365-2373. 4.

detreatment Vries FM, for Kolthof the secondary J, Postma prevention MJ, Denig P,of Hak cardiovascular E. Efficacy ofand standard cerebrovascular and intensive events statin in diabetes patients: a meta-analysis. PLoS One 2014; 9(11): e111247. 5. Simpson RJ, Mendys P. The effects of adherence and persistence on clinical outcomes in patients treated with statins: A systematic review. J Clin Lipidol 2010; 4(6): 462-471. 6. Chodick G, Shalev V, Gerber Y, et al. Long-term persistence with statin treatment in a not-for-

profitisrael. Clin health Ther maintenance 2008; 30(11): organization: 2167-2179. A population-based retrospective cohort study in 7.

Latryin a real-life P, Molimard setting M, isDedieu better B, when Couffinhal associated T, Bégaud cardiovascular B, Martin-Latry risk factorsK. Adherence increase: with a statinscohort study. BMC Cardiovasc Disord 2011; 11: 46. 8. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. JAMA 2002; 288(4) :455-461. 9. Donnelly LA, Doney AS, Morris AD, Palmer CN, Donnan PT. Long-term adherence to statin treatment in diabetes. Diabet Med 2008; 25(7): 850-855. 10.

EuropeanDe Backer Association G, et al; ESC for Committee Cardiovascular for Practice Prevention Guidelines & Rehabilitation, (CPG) 2008-2010 Reiner andZ, Catapano 2010-2012 AL,

Committees.management ESC/EAS of dyslipidaemias guidelines of for the the European management society of ofdyslipidaemias: cardiology (ESC) The and task the force European for the atherosclerosis society (EAS). Eur Heart J 2011; 32(14): 1769-1818. 11. Practice guideline cardiovascular risk management from the Dutch General Practitioners Society (Tweede herziening) [In Dutch, NHG-standaard Cardiovasculair risicomanagement] Huisarts Wet 2012; 55(1): 14-28. 12. Poluzzi E, Piccinni C, Carta P, et al. Cardiovascular events in statin recipients: Impact of adherence to treatment in a 3-year record linkage study. Eur J Clin Pharmacol. 2011; 67(4): 407-414. 13. Monster TB, Janssen WM, de Jong PE, de Jong-van den Berg LT. PREVEND Study Group Prevention of REnal and Vascular ENT Stage Disease. Pharmacy data in epidemiological studies: An easy to obtain and reliable tool. Pharmacoepidemiol Drug Saf 2002; 11(5): 379-384.

136 Cardiovascular events and adherence to statins in diabetes

14. Vegter S, de Jong-van den Berg LT. Misdiagnosis and mistreatment of a common side-effect- -angiotensin-converting enzyme inhibitor-induced cough. Br J Clin Pharmacol 2010; 69(2): 200-203. 15. Vegter S, Nguyen NH, Visser ST, de Jong-van den Berg LT, Postma MJ, Boersma C. Compliance, persistence, and switching patterns for ACE inhibitors and ARBs. Am J Manag Care 2011; 17(9): 609-616. 16.

17. ATC codes. Available: http://www.whocc.no/atc_ddd_index/. Accessed 8 September 2014. - Leufkenslands. In: AnonymousHG, Urquhart (eds) J (2006, Pharmacoepidemiology. 2007) Automated Pharmacy Wiley, New Record York, Linkagepp 311–322. in The Nether 18. de Vries FM, Denig P, Visser ST, Hak E, Postma MJ. Cost-effectiveness of statins for primary prevention in patients newly diagnosed with type 2 diabetes in the Netherlands. Value Health 2014; 17(2): 223-230. 19. Voorham J, Haaijer-Ruskamp FM, van der Meer K, et al. Quality of the treatment of type 2 diabetes: Results from the GIANTT project 2004-2007. Ned Tijdschr Geneeskd 2010; 154: A775. 20. Levesque LE, Hanley JA, Kezouh A, Suissa S. Problem of immortal time bias in cohort studies: Example using statins for preventing progression of diabetes. BMJ 2010; 340: b5087. 21. Peterson AM, Nau DP, Cramer JA, Benner J, Gwadry-Sridhar F, Nichol M. A checklist for medication compliance and persistence studies using retrospective databases. Value Health 6 2007; 10(1): 3-12. 22. Daskalopoulou SS, Delaney JA, Filion KB, Brophy JM, Mayo NE, Suissa S. Discontinuation of statin therapy following an acute myocardial infarction: a population-based study. Eur Heart J 2008; 29(17): 2083-2091. 23. Haffner SM, Lehto S, Rönnemaa T, Pyörälä K, Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med 1998; 339(4) :229-234. 24. Abraha I, Montedori A, Stracci F, Rossi M, Romagnoli C. Statin Compliance in the Umbrian population. Eur J Clin Pharmacol 2003; 59(8-9): 659-661. 25. Voorham J, Denig P. Computerized extraction of information on the quality of diabetes care from free text in electronic patient records of general practitioners. J Am Med Inform Assoc 2007; 14(3): 349-354.

137

CHAPTER 7

Preferences of diabetes patients for lipid-lowering treatment: a discrete choice experiment

Folgerdiena M. de Vries1,2 Sieta T. de Vries1 Thijs Dekker3 Eelko Hak2 Petra Denig1

Submitted

ABSTRACT

Aims: To assess the willingness of patients with type 2 diabetes to continue

using lipid-lowering treatment, the importance they attach to specific treatment characteristics, and the influence of patient characteristics on this. Methods: Patients being prescribed at least an oral glucose-lowering and lipid- lowering drug completed a questionnaire with discrete choice experiment. This experiment contained choice sets with two hypothetical lipid-lowering treatments and the option to stop lipid-lowering treatment. These treatment options differed in their characteristics (i.e. effects and intake moments). Multinomial logit models were used to assess the importance attached to the characteristics in relation to patient characteristics.

Results: In total, 175 patients were included in the study (mean age 66 years; 59% male). In general, patients were willing to continue using lipid-lowering treatment when their cholesterol level was too high. All treatment characteristics included

in the experiment had a significant impact on the patients’ treatment choices. with lower educational levels and a history of cardiovascular disease were more Cholesterol reduction and risk of adverse events were most influential. Patients willing to continue treatment. Age, education, cardiovascular history and having

experienced adverse effects all influenced the importance patients attached to specific treatment characteristics when choosing between medication alternatives. Conclusions: Patients were generally willing to continue using lipid-lowering

treatment characteristics depended on patient characteristics. These insights can treatment but the willingness and importance patients attached to specific be useful for motivating patients to continue lipid-lowering treatment and making treatment decisions more in accordance with patient preferences when current

treatment is considered insufficient.

140 Preferences for lipid-lowering treatment in diabetes

INTRODUCTION

Lipid-lowering drugs are associated with a reduction in risk of cardiovascular events [1,2]. Lipid-lowering treatment is recommended for almost all patients with type 2 diabetes [3,4], as such patients have an increased risk of developing cardiovascular disease [5]. However, lipid-lowering drugs are not optimally used in clinical practice. Discontinuation rates of [6,7]. lipid-lowering treatment are high during the first years of treatment Discontinuation of treatment is related to patient’s preferences. Discontinuation appears to be more common in patients that are female, have no cardiovascular history, have experienced adverse effects, and have a lower socioeconomic status or education [8-11].

[12]. A recent Additionally expected benefits and adverse effects of a drug, especially when interfering study assessing patients’ trade-off preferences for statin treatment showed that about with daily activities, influence the willingness to take preventive medication drawbacks [13]. These trade-off preferences for preventive medication might be related 40% of the patients valued efficacy equal to or lower than adverse effects and other to age, as older patients may value quality of life over life extension [14,15]. Better insight in patients’ preferences according to patient characteristics could be useful for motivating patients to continue using lipid-lowering treatment and to make treatment decisions in line with those preferences [16].

Patients’ preferences for lipid-lowering treatment can be assessed using a discrete 7 choice experiment. In this experiment a hypothetical situation is presented to the patient and drawbacks [13,17]. The aim of this study was to assess (1) the willingness of patients with treatment choices described by their characteristics regarding expected benefits with type 2 diabetes to continue using lipid-lowering treatment, (2) the importance patient characteristics such as age, gender, clinical history and education. they attach to specific treatment characteristics, and (3) whether both are influenced by

141 CHAPTER 7

METHODS

Study design, sample and data collection We conducted a questionnaire study incorporating a discrete choice experiment to assess patients’ preferences regarding lipid-lowering treatment. Patients were recruited via pharmacies in the northern part of the Netherlands. Patients were eligible if they were 18 blood glucose-lowering treatment) and had received statin treatment in the six months years or older, had type 2 diabetes (identified as at least one prescription for non-insulin acid and derivatives or other lipid modifying agents were excluded. Eligible patients before selection. Patients with a prescription for fibrates, bile acid sequestrants, nicotic were informed about the study via an information letter including an informed consent form. Patients that returned the informed consent form received the questionnaire by mail. Patients were called when they did not return the questionnaire within eight weeks or when there was data missing in the general part of the questionnaire to reduce the amount of non-responders and missing items. Questionnaires with an incomplete discrete choice experiment were returned for further completion.

Gender and age of all the contacted patients were provided by the pharmacy to compare responders with non-responders. Of consenting patients, a medication list of the last six months was collected from their pharmacy. Medication use during this period was [18]. classified using ATC codes The Medical Ethics Committee of the University Medical Center Groningen (METc UMCG) in the Netherlands determined that ethical approval was not needed for this study (reference number M14.150721). In accordance to the Dutch law, all patients consented to participation.

Discrete choice experiment The general part of the questionnaire consisted of questions about age, gender, clinical history and education. The second part of the questionnaire consisted of a discrete choice experiment. Discrete choice experiments are used to analyse trade-off preferences regarding treatments, based on the choices the patient makes between hypothetical treatment options. Information about treatment characteristics, including effects, and drug administration aspects, was given. Varying levels of these treatment efficacy regarding cholesterol level and cardiovascular risk reduction, risk of adverse characteristics were shown to assess their influence on the patient’s treatment choices. relevance for discontinuation of treatment [6,7] and patient preferences for preventive These characteristics were selected based on previous findings indicating their treatment [13]. Levels for cholesterol reduction and cardiovascular risk reduction were

142 Preferences for lipid-lowering treatment in diabetes

based on clinical trial data of lipid-lowering treatment [1,2,19]. The levels for the risk of adverse effects were based on reported risks for muscle pain [20,21]. Levels for the intake moment were based on available options [22]. The treatment characteristics and levels

2 nurse practitioners, 3 general practitioners, 2 specialists, and 3 pharmacists. Table 1 were discussed and finalized in interviews with ten healthcare professionals, including contains the final characteristics and levels of the discrete choice experiment. For the experiment, patients needed to imagine they used a lipid-lowering drug (1 tablet per day) without experiencing adverse effects and that their LDL-cholesterol level was too high (3 mmol/l). Patients were asked to choose between continuing treatment with ‘Medication A’, continuing treatment with ‘Medication B’ or ‘I consider stopping’. Within due to a heart attack and risk of limitations due to stroke increased, whereas the risk of the ‘I consider stopping’ option, levels of cholesterol, risk of death, risk of limitations [23] was used to generate the Medication A and Medication B options, and select a limited adverse effects became zero just as the number of tablets per day. An efficient design number of choice sets that maximize the precision of the model parameters. Ngene

[23]. These 30 choice sets were divided in three blocks of 10 to reduce the burden for (version 1.1.1) was used to develop a D-efficient design with 30 different choice sets assign patients to one of the blocks containing 10 choice sets, leading to 10 choices per the respondents. Stratified sampling based on age, gender and pharmacy was used to patient. To identify irrational responders, one dominant choice set was added in which all characteristics of Medication A favoured the ones of Medication B. 7

143 CHAPTER 7

FIGURE 1. Example of a choice set presented in the questionnaire

144 Preferences for lipid-lowering treatment in diabetes

The questionnaire was tested in a pilot study with 10 patients, after which some changes in wording and a separate instruction form were made. An example of a choice set is shown in Figure 1.

Statistical analyses Descriptive statistics were used to compare responders with non-responders, and rational with irrational responders. The treatment choices made by the patients (medication A, medication B or consider stopping treatment) were analysed using a multinomial logit model, which estimates the probability of a particular choice made by a patient given a set of treatment and patient characteristics. First, a model was estimated which included the main effects of the six treatment characteristics to assess the willingness to continue using lipid-lowering treatment (compared to stopping such treatment) and the relative importance of the included treatment characteristics (Table 1). A random utility model was used assuming that patients choose their preferred option by maximizing utility:

U continue treatment cholesterol early death limitations heart attack limitations stroke +

= β0adverse effects +one β1tablet every other+ day β2 half tablet + every β3 day + β4 β5 + β6 + β7 + ε

U = utility that a patient assigns to a treatment

ε = unexplainable component 7 β0 = estimate of willingness to continue versus to stop treatment β1- β7 = relative importance of each of the treatment characteristics Second, the impact of the patients’ characteristics on the willingness to continue using lipid-lowering drugs and on the relative importance of the treatment characteristics was assessed by including interaction terms of patient characteristics with all six treatment characteristics in the model. The impact of the following patient characteristics was tested (1) age, (2) gender, (3) cardiovascular history, (4) education and (5) history of adverse effects. Main effects and interaction terms were selected via backward selection using the log-likelihood as the guiding measure of goodness-of-fit. The relative range, which is the difference between the smallest part worth utility and the largest importance of treatment characteristics was determined by calculating the coefficient part worth utility within the levels of the treatment characteristics, and dividing it by the [24]. The analyses were performed with Stata version 13 (Stata Corp., College Station, TX). sum of the coefficient ranges of all treatment characteristics and ranges

145 CHAPTER 7

TABLE 1. Treatment characteristics and levels used in the discrete choice experiment Coefficient Abbreviation in regression formula Coding Cholesterol level β1 Cholesterol Increase from 3.0 mmol/l to 4.0 mmol/l* 4 Staying at 3.0 mmol/l# 3 Decrease from 3.0 mmol/l to 2.0 mmol/l 2 Risk of early death β2 early death 9 out of 100 patients (9%)* 9 7 out of 100 patients (7%)# 7 5 out of 100 patients (5%) 5 Limitations due to heart attack β3 limitations heart attack 5 out of 100 patients (5%)* 5 4 out of 100 patients (4%)# 4 3 out of 100 patients (3%) 3 Limitations due to stroke β4 limitations stroke 5 out of 100 patients (5%)* 5 4 out of 100 patients (4%)# 4 3 out of 100 patients (3%) 3 Risk of adverse events β5 adverse events No adverse events* 0 5 out of 100 patients (5%) 5 10 out of 100 patients (10%)# 10 Intake moment No tablets* - Every day one tablet (reference)# - Every other day one tablet β6 one tablet every other day - Every day half a tablet β7 half tablet every day - * Levels included in the ‘I consider to stop’ option (not used for medication A or medication B); # Levels used for the non-preferable treatment in the dominant choice set.

RESULTS

Of the 904 patients that were contacted by the pharmacies, 251 expressed their willingness to participate (Figure 2)

. There was no difference in gender (p = 0.189) completed the questionnaire. Eight patients had to be excluded because they did not and age (p = 0.283) between non-responders and responders. Of the responders, 203 meet the inclusion criteria, and 20 were identified as irrational responders. Rational and irrational responders did not differ in age (p = 0.457) or gender (p = 0.542). 146 Preferences for lipid-lowering treatment in diabetes

Contacted Information letters sent (n=904)

- Did not meet inclusion criteria (n=3) - Other reasons (n=2) - Non-responders (n=648, 72%)

Questionnaire sent to patients (n=251, 28%) Responders mean age 66.6 (SD 10.1) 57% males

- No/incomplete questionnaire returned (n=48)

Questionnaires with complete discrete choice experiment Completers (n=203, 23%) mean age 66.1 (SD 10.0) 59% males

- No medication list (n=3) - No lipid- or glucose-lowering drugs (n=5) - Irrational choice dominant choice set (n=20) 7 Questionnaires used for analyses (n=175, 19%) Included mean age 66.2 (SD 9,7) 59% males

FIGURE 2: Patient inclusion flowchart

147 CHAPTER 7

TABLE 2. Patient characteristics (n=175) Characteristics General: Age, (SD) 66.2 (9.7) Female (%) 72 (41.1) Education (%) Low educated a. 80 (45.7) Moderate educated b. 72 (41.1) High educated c. 20 (11.4) Other 3 (1.7) BMI (IQR) 29.2 (25.7-32.0) Smoking (%) Current smokers 21 (12.0) Past smokers 99 (56.6) Non smokers 55 (31.4) Patients hospitalized for cardiovascular disease (%) 64 (37%) Cardiovascular event 55 (86%) Cerebrovascular event 7 (11%) Cardiovascular and cerebrovascular event 2 (3%) Prescribed medication: Number of lipid-lowering drugs 1 lipid-lowering drug 175 Classes of cholesterol lowering drugs (ATC code) Simvastatin (C10AA01) 82 (46.9) Pravastatin (C10AA04) 9 (5.1) Atorvastatin (C10AA05) 54 (30.9) Rosuvastatin (C10AA07) 30 (17.1) Number of glucose-lowering drugs (%) 1 glucose-lowering drug 86 (49.1) 2 glucose-lowering drug 78 (44.6) >2 glucose-lowering drug 11 (6.3) Number of blood pressure-lowering drugs (%) No blood pressure-lowering drug 34 (19.4) 1 blood pressure-lowering drug 39 (22.3) 2 blood pressure-lowering drugs 51 (29.2) >2 blood pressure-lowering drugs 51 (29.2) >>

148 Preferences for lipid-lowering treatment in diabetes

Number of cholesterol-lowering drugs used according to the patient (%) None 6 (3.4) One 160 (91.4) Two 3 (1.7) More than two 1 (0.6) I do not know 5 (2.8) Did you ever experience side effects from cholesterol-lowering drugs? (%) No 108 (61.7) Yes 46 (26.3) I do not know 21 (12.0) SD: Standard deviation; IQR: Interquartile range a No education; Elementary school; Junior secondary vocational education b Junior general secondary education; Senior secondary vocational education c Senior general secondary education; Higher professional education; University education

In total, 175 patients were included in the study with a mean age of 66 years (SD 9.7) and 59% of them were male. Almost half of the patients was lower educated (Table 2). There were 64 patients (37%) that reported that they had been hospitalized for a cardiovascular or cerebrovascular event. Adverse effects from lipid-lowering drugs were experienced by 26% of the patients (Table 2). According to the medication lists, all included patients used one lipid-lowering drug and almost half of the patients used simvastatin (Table 2). 7

Impact of treatment characteristics on preferences Only 7 patients considered for one or more choice sets to stop lipid-lowering treatment. that patients prefer to continue using treatment when their cholesterol level is too This was confirmed by the positive constant term in the logit model, which indicates high (Table 3) associated with the patients’ treatment preferences (Table 3). Looking at the relative . All treatment characteristics included in the model were significantly treatment on the cholesterol level, followed by the effect on the risk of adverse events, importance, patients’ preferences are most strongly influenced by the effect of the early death, limitations due to a heart attack and limitations due to a stroke.

149 CHAPTER 7

TABLE 3. Relative importance of treatment characteristics Variables Coefficient CI 95% p-value Relative importance* (ranking) Constant (continue treatment) 2.06 1.492 - 2.627 <0.001 Cholesterol -0.641 -0.750 - -0.532 <0.001 23.65 (1) Early death -0.290 -0.346 - -0.235 <0.001 21.40 (3) Limitations heart attack -0.291 -0.400 - -0.182 <0.001 10.74 (4) Limitations stroke -0.227 -0.337 - -0.117 <0.001 8.37 (5) Adverse effects -0.119 -0.141 - -0.097 <0.001 21.95 (2) Half a tablet every day -0.440 -0.592 - -0.287 <0.001 8.12 (6) One tablet every other day -0.313 -0.469 - -0.157 <0.001 5.77 (7) Number of observations 1,750 (175 patients * 10 choice sets) Log Likelihood -1087.89 Log Likelihood ratio 650.23 *The relative importance of treatment characteristics was determined by calculating the coefficient range, which is the difference between the smallest part worth utility and the largest part worth utility within the levels of the treatment characteristics, and dividing it by the sum of the coefficient ranges of all treatment characteristics and ranges. [24] CI: confidence interval.

Impact of patient characteristics on preferences Patients with a history of cardiovascular disease were more willing to continue using lipid- lowering treatment. Whereas, patients with a higher education level were less willing to continue lipid-lowering treatment (Table 4). With increasing age, the importance of the risk reduction of early death and limitations due to a heart attack decreased, whereas the importance of a low risk of adverse effects increased. Additionally with increasing age, patients were less willing to take a tablet every other day as compared to daily intake. Having a cardiovascular history was associated with attaching more importance to a low risk of limitations due to a heart attack. For higher educated patients a lower cholesterol level and a lower risk of early death was more important. Patients that had experienced adverse effects from lipid-lowering treatment before, attached more continue treatment nor the importance of treatment characteristics. importance to a lower risk of adverse effects. Gender did not influence the willingness to

150 Preferences for lipid-lowering treatment in diabetes

TABLE 4. Relative importance of treatment characteristics when including interaction terms with patient characteristics Variables Coefficient CI 95% p-value Constant (continue) 3.424 2.326 – 4.522 <0.001 Constant (continue) x Cardiovascular history* 2.176 0.133 – 4.219 0.037 Constant (continue) x Education† -2.477 -3.682 - -1.272 <0.001 Cholesterol -0.501 -0.656 - -0.347 <0.001 Cholesterol x Education† -0.314 -0.535 - -0.094 0.005 Early death -0.864 -1.226 - -0.502 <0.001 Early death x Age# 0.010 0.005 – 0.015 <0.001 Early death x Education† -0.165 -0.276 - -0.055 0.003 Limitations heart attack -0.979 -1.695 - -0.264 0.007 Limitations heart attack x Age# 0.012 0.001 – 0.022 0.032 Limitations heart attack x Cardiovascular history* -0.236 -0.467 - -0.006 0.044 Limitations stroke -0.245 -0.357 - -0.133 <0.001 Adverse effects x Age# -0.001 -0.002 – -0.001 <0.001 Adverse effects x Experienced adverse effects‡ -0.090 -0.138 – -0.042 <0.001 Half a tablet every day -0.457 -0.611 - -0.302 <0.001 Tablet every other day x Age# -0.005 -0.007 – -0.003 <0.001 Number of observations 1,750 (175 patients * 10 choice sets) Log Likelihood -1045.7 Log Likelihood ratio 583.72 CI: confidence interval. 7 #age: (continouos variable) *cardiovascular history: (Have you ever been hospitalized for a cardiovascular or cerebrovascular event? yes (coded 1) or no (coded 0)) † education (What is your highest completed education? Lower educated: No education; Elementary school; Junior secondary vocational education (coded 0); Higher educated: Junior general secondary education; Senior secondary vocational education; Senior general secondary education; Higher professional education; University education (coded 1)) ‡ history of adverse effects (Did you ever experience side effects from cholesterol-lowering drugs? Yes (coded 1); No (coded 0); I do not know (coded 0)).

DISCUSSION

In general, current users of lipid-lowering treatment were willing to continue using the willingness to continue treatment, but patients with lower educational levels and treatment when their cholesterol level was too high. Age and gender did not influence with a history of cardiovascular disease were more willing to do so. Age, education, cardiovascular history and having experienced adverse effects all influenced the 151 CHAPTER 7

between medication alternatives. importance patients attached to specific treatment characteristics when choosing discontinuation of statin treatment [8,9]. Our study suggests that elderly patients do not Previous studies found conflicting results regarding the relationship between age and have a decreased willingness to continue treatment but that they may differ in how they age, patients may attach less importance to life-extension and put more value on quality weigh beneficial and adverse effects when choosing between treatments. With increasing of life [14,15] of early death and more importance to a low risk of adverse effects supports this. It has . The finding that older patients in our study gave less importance to a low risk been suggested that females may discontinue treatment more often [8], but this might in part be due to experiencing more adverse effects [25]. When adjusting for adverse effects, statin discontinuation [9] a recent study found that female gender became an insignificant factor for predicting . Our study did not find a gender effect on treatment decisions. It is acknowledged that adverse effects and concerns about developing adverse effects are reasons for discontinuing lipid-lowering treatment [9,10,26]. Most patients in our study, however, were willing to continue with some kind of medication treatment that offered discontinue treatment which would result in experiencing no adverse effects but also benefits at the costs of potential adverse effects. Only seven patients considered to

[8,10,26]. In addition, discontinuers had less knowledge and were less less beneficial effects. Previous studies found that discontinuers also expect limited [9,10]. Therefore, it is possible benefits of statins satisfied with information they received about the treatment more patients are inclined to choose medication treatment over discontinuation of that with sufficient information about benefits and drawbacks, as provided in our study, treatment. When being confronted with actual bothersome adverse effects, however, the willingness to continue treatment is likely to decrease [9,26,27]. A recent study assessing patients’ preferences when choosing between pairs of cholesterol-lowering treatments effects and other drawbacks when such drawbacks were presented as certainties [13]. In showed that around 40% of the patients valued efficacy equal to or lower than adverse contrast, having experienced adverse effects in the past did not affect the willingness to continue treatment in our study. One explanation could be that we studied a cohort of current users, who apparently have been able to cope with these adverse effects. These patients did attach more importance to a low risk of adverse effects when choosing between treatments, perhaps because they are more aware of the effects of adverse effects on the quality of life.

152 Preferences for lipid-lowering treatment in diabetes

A higher willingness to continue treatment was seen in patients with a history of cardiovascular disease, which corresponds to the lower discontinuation rates seen in patients with cardiovascular diseases or diabetes [8]. Patients with a cardiovascular history gave more importance to a low risk of limitations due to a heart attack but not to limitation due to a stroke or a low risk of early death. As most patients with a cardiovascular history in our study experienced a cardiovascular event (i.e. 89% compared to 14% with a cerebrovascular event) they might have been more aware of the consequences of a heart attack.

We found that higher educated patients were more likely to discontinue treatment. This appears to be in contrast to a Danish study, where a lower education was associated with more discontinuation [11] [9]. One

and a US study that did not find an impact of education in countries where the treatment is not fully reimbursed. When corrected for socio- reason for such conflicting results is confounding by socio-economic-status, especially economic-status, the association between low education and discontinuation became [11]. In the Netherlands, lipid-lowering treatment is reimbursed. We could speculate that higher educated patients more often have a healthy insignificant in the Danish study lifestyle and a lower cardiovascular risk, and therefore perceive medication treatment as less essential. When making a choice between treatments, higher educated patients gave more importance to a low cholesterol level and a low risk of early death suggesting that 7 they focus more on the efficacy of treatment. A possible limitation to our study is the response rate and possible related selection bias. However, there were no differences in gender and age between responders and non-responders. Also, almost half of the participants were lower educated which corresponds with the national type 2 diabetes population [28]. We included current users of lipid-lowering treatment. Therefore, patients that had permanently discontinued such treatment because of negative beliefs or experiences were not included in our study. Another limitation to the study is that patients needed to consider a hypothetical the dominant option we tried to exclude responders who may not have understood the situation which might be a difficult task for some. By excluding patients that did not prefer at the same time in a discrete choice experiment is a strength of our study. Furthermore, task. The fact that patients needed to consider benefits and adverse effects of treatment the option to stop lipid-lowering treatment was included, therefore the choices resemble real-world choice options [29].

153 CHAPTER 7

CONCLUSIONS

We conclude that education level and cardiovascular history are relevant for the willingness of diabetes patients to continue lipid-lowering treatment, and that age, education, cardiovascular history and having experienced adverse effects all influenced between medication alternatives. These insights can be useful for motivating patients the importance these patients attach to specific treatment characteristics when choosing to continue using lipid-lowering treatment and making treatment decisions more in accordance with patient preferences when current treatment is considered insufficient.

154 Preferences for lipid-lowering treatment in diabetes

REFERENCES

detreatment Vries FM, for Kolthof the secondary J, Postma prevention MJ, Denig P,of Hak cardiovascular E. Efficacy ofand standard cerebrovascular and intensive events statin in

3. diabetesNHG-standaard patients: Cardiovasculair a meta-analysis. risicomanagement PLoS One 2014; DOI: (Eerste 10.1371/journal.pone.0111247. herziening) Huisarts Wet 2012; 55: 14-28. 4. Rutten GEHM, De Grauw WJC, Nijpels G, et al. NHG-standaard Diabetes mellitus type 2. Huisarts Wet 2006; 49: 137-52. 5. American heart association. cardiovascular disease and diabetes: Statistics update 2010. -

Available: URL:http://www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetes 6. Matters/Cardiovascular-Disease-Diabetes_UCM_313865_Article.jsp.Zhang H, Plutzky J, Skentzos S, et al. Discontinuation of staitns in routine Accessed care 15 settings. March 2013. Ann Intern Med 2013; 158(7): 526-534. 7. Simpson RJ, Tunceli K, Ramey DR, et al. Treatment pattern changes in high-risk patients newly initiated on statin monotherapy in a managed care setting. J Clin Lipidol 2013; 7(5): 399-407. 8. Mann DM, Woodward M, Muntner P, Falzon L, Kronish I. Predictors of nonadherence to statins: a systematic review and meta-analysis. Ann Pharmacother 2010; 44(9): 1410-1421. 9. Wei MY, Ito MK, Cohen JD, Brinton EA, Jacobson TA. Predictors of statin adherence, switching, 7 and discontinuation in the USAGE survey: understanding the use of statins in America and gaps in patient education. J Clin Lipidol 2013; 7(5): 472-483. 10. McGinnes B, Olson KL, Magid D, et al. Factors related to adherence to statin therapy. Ann Pharmacother 2007; 41(11): 1805-1811. 11. Rasmussen JN, Gislason GH, Rasmussen S, et al. Use of statins and beta-blockers after acute myocardial infarction according to income and education. J Epidemiol Community Health 2007; 61(12): 1091-1097. 12.

Friedpersons’ TR, willingness Tinetti ME, toTowle take V,medication O’Leary JR, for Iannone primary L. cardiovascularEffects of benefits prevention. and harms Arch on Intern older Med 2011; 171(10): 923-928. 13.

WoutersAssessing H, patients’ Van Dijk L,trade-off Van Geffen preferences EC, et al. Dowith the conjoint benefits analysis.of statins Intoutweigh J Cardiol their 2014; drawbacks? 176(3): 1220-1222. 14. Fried TR, Van Ness PH, Byers AL, Towle VR, O’Leary JR, Dubin JA. Changes in preferences for life-sustaining treatment among older persons with advanced illness. J Gen Intern Med 2007; 22(4): 495-501.

155 CHAPTER 7

15. Sherbourne CD, Keeler E, Unutzer J, Lenert L, Wells KB. Relationship between age and patients’ current health state preferences. Gerontologist 1999; 39(3): 271-278. 16. Porzsolt F, Clouth J, Deutschmann M, Hippler HJ. Preferences of diabetes patients and physicians: a feasibility study to identify the key indicators for appraisal of health care values. Health Qual Life Outcomes 2010; 8: 125. 17. Joy SM, Little E, Marathur NM, Purnell TS, Bridges JF. Patient preferences for the treatment of type 2 diabetes: a scoping review. Pharmacoeconomics 2013; 31(10): 877-892. 18.

ATC/DDD2013. index. Available from: http://www.whocc.no/atc_ddd_index/. Accessed 13 March 19. Smith MEB, Lee NJ, Haney E, Carson S. Drug class review: Hmg-coa reductase inhibitors (statins) and fixed-dose combination products containing a statin: Final report update 5. Portland (OR); 2009. 20.

Bruckerthigh-dosage E, Hayem statin therapy G, Dejager in hyperlipidemic S, Yau C, Bégaud patients B. Mild the to moderatePRIMO study. muscular Cardiovasc symptoms Drugs Therwith 2005; 19(6): 403-414. 21. Silva M, Matthews ML, Jarvis C, et al. meta-analysis of drug-induced adverse events associated with intensive-dose statin therapy. Clin Ther 2007; 29(2): 253-260. 22. Keating AJ, Campbell KB, Guyton JR. Intermittent nondaily dosing strategies in patients with previous statin-induced myopathy. Ann Pharmacother 2013; 47(3): 398-404. 23. Reed Johnson F, Lancsar E, Marshall D, et al. Constructing experimental designs for discrete- choice experiments: report of the ISPOR Conjoint Analysis Experimental Design Good Research Practices Task Force. Value Health 2013; 16(1): 3-13. 24. Laba TL, Brien J, Jan S. Understanding rational non-adherence to medications. A discrete choice experiment in a community sample in Australia. BMC Fam Pract 2012; 13: 61. 25. Hsue PY, Bittner VA, Betteridge J, et al. Impact of female sex on lipid lowering, clinical outcomes, and adverse effects in atorvastatin trials. Am J Cardiol 2015; 115(4): 447-453. 26. Fung V, Sinclair F, Wang H, Dailey D, Hsu J, Shaber R. Patients’ perspectives on nonadherence to statin therapy: a focus-group study. Perm J 2010; 14(1): 4-10. 27. Mann DM, Allegrante JP, Natarajan S, Halm EA, Charlson M. Predictors of adherence to statins for primary prevention. Cardiovasc Drugs Ther 2007; 21(4): 311-316. 28.

https://www.volksgezondheidenzorg.info/onderwerp/diabetes-mellitus/cijfers-context/bevolkingsgroepen#node-ongediagnosticeerde-diabetes-naar-opleidingsniveau Accessed 8 April 2015. 29. Veldwijk J, Lambooij MS, de Bekker-Grob EW, Smit HA, de Wit GA. The effect of including an opt-out option in discrete choice experiments. PLoS One 2014; 9(11): e111805.

156 Preferences for lipid-lowering treatment in diabetes

157

General discussion General discussion

This thesis focused on statin treatment in type 2 diabetes patients of which different aspects were studied. In Part I the effect of statins on LDL-cholesterol and cardiovascular outcomes was determined. Also, cost-effectiveness of statin treatment was assessed in which we took account of information from clinical practice. Part II focussed on patient methodological considerations, and future perspectives will be discussed for each part and physician behaviour regarding statin treatment. The main findings, implications, in this chapter. At the end of this chapter an overall conclusion is presented.

Part I. Effects and cost-effectiveness of statin treatment

For secondary prevention of cardiovascular and cerebrovascular events, the need for statin treatment is widely recognized, however, for primary prevention the use of statins is more controversial. The absolute risk reduction of statin treatment for primary prevention could be low because of a low baseline cardiovascular risk. Some have reported that statins are associated, with a fifth reduction in the incidence of cardiovascular and [1]. Others, however, reported that there is only limited cerebrovascular events per mmol/l reduction in LDL-cholesterol, largely irrespective of evidence that statins are cost-effective and improve patients quality of life for primary specific patient characteristics prevention [2]. There is no precise effect estimate of statin treatment in diabetes patients [1,2]. Such evidence is important for clinical decision making and cost-effectiveness optimal. In clinical trials patients are often treated under controlled conditions, which analysis, especially for primary prevention where the benefit-risk balance might not be is different from clinical practice. In clinical practice statin treatment response and cost- [3-6]. The precise relation between adherence to treatment and treatment response is unclear and likely effectiveness will be influenced by for example adherence to treatment to be affected by the treatment dose.

• determine precise effect estimates of statins for primary and secondary prevention Therefore the aims of the first part of this thesis were to: of cardiovascular events in diabetes patients; • assess the effect of dosing and adherence on cholesterol outcomes in diabetes patients; • determine the cost-effectiveness of statins for primary prevention of cardiovascular and cerebrovascular events in newly diagnosed type 2 diabetes patients, taking adherence into account.

160 General discussion

Main findings In Chapter 1 cerebrovascular events in diabetes patients was determined in a meta-analysis. Based the effect of statins on the first time occurrence of cardiovascular and on four trials including 10,187 participants, treatment with standard-dose statins for the primary prevention of a major cardiovascular or cerebrovascular event resulted in a significant relative risk reduction of 25%. In order to prevent one major cardiovascular risk reductions of 31% for fatal or non-fatal stroke and 30% for fatal or non-fatal or cerebrovascular event 35 patients should be treated for 3.8 years. Significant relative were found. myocardial infarction, and a non-significant risk reduction of 16% for all-cause mortality prevention of major cardiovascular and cerebrovascular events in diabetes patients was The efficacy of standard-dose and intensive-dose statin treatment for the secondary assessed in Chapter 2

. Using data from five trials including 4,351 participants, standard- occurrence of any major cardiovascular or cerebrovascular event compared to placebo. dose statin treatment resulted in a significant relative risk reduction of 15% in the

For fatal or non-fatal stroke, there was a significant 33% relative risk reduction. Non- and 22% for all-cause mortality were found. In an analysis using data from four trials significant relative risk reductions of 27% for fatal or non-fatal myocardial infarction including 4,805 participants, standard-dose statin treatment compared to intensive- dose statin treatment resulted in an additional 9% relative risk reduction in any major cardiovascular or cerebrovascular event.

In Chapter 3 we performed a cohort study using observational data from the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database. The association between adherence, treatment dose and LDL-cholesterol response in type 2 diabetes patients initiating statin treatment was assessed. For patients with a baseline LDL- reduction in LDL-cholesterol. It was estimated that a 20% reduction in adherence rate cholesterol level of 3.7 mmol/l, an adherence rate of 100% was associated with a 48% on standard-dose treatment was associated with 8-9% less reduction in LDL-cholesterol adherence rate of at least 63% on standard-dose treatment and of 76% on low-dose in such patients. Patients with a baseline LDL-cholesterol of 3.7 mmol/l will need an treatment to reach the LDL-cholesterol target of 2.5 mmol/l. In Chapter 4 the cost-effectiveness of statins for primary prevention of cardiovascular and cerebrovascular events in newly diagnosed type 2 diabetes patients was assessed. The effect of real-world adherence rates was assessed. Taking expected adherence rates into account, starting statin treatment at diagnosis of type 2 diabetes was cost- effective at €2,245 per QALY in the base case (a 60-year old patient with mean 10-year

161 General discussion

cardiovascular risks as estimated for patients aged 56-65 at type 2 diabetes diagnosis). Differences in age and 10-year risks for CHD and stroke resulted in large differences in cost-effectiveness from losing QALY’s to being almost cost neutral. With the adherence rates seen in clinical practice, it can be concluded that treating all type 2 diabetes patients under the age of 45 at diagnosis with statins for primary prevention is not cost- effective. For patients between 45 and 55 years at diagnosis, statin treatment is cost- effective except when the 10-year risk for CHD is 6% or lower. For all other diabetes patients, statin treatment is expected to be cost-effective at a threshold level of 20,000 euro per QALY or even cost-saving. Notably, also higher thresholds have been suggested and are being applied [7].

Implications Nowadays, there is a broad indication for statin treatment [8,9]. The need for statin treatment for the secondary prevention of cardiovascular events is widely recognized, especially for diabetes patients [8,9,10]. Recommendations on the dose needed, however, are less clear. American guidelines recommend intensive-dose statin treatment for secondary prevention [10] whereas Dutch guidelines recommend to start on standard- dose treatment irrespective of patient characteristics [8]. In Chapter 2 we found a for such treatment. There was an additional risk reduction in any major cardiovascular significant risk reduction of 15% for major cardiovascular and cerebrovascular events or cerebrovascular event of 9% for intensive-dose treatment compared to standard- dose treatment. Although the risk for adverse events is increasing with higher dosing, [11,12]. This implies that intensive-dose treatment can be used in high-risk patients, as this is also likely to this is not likely to outweigh the health benefits in high risk patients be cost-effective [13].

Furthermore, there is discussion on the use of statin treatment for primary prevention [2]. The 25% risk reduction with acceptable numbers needed to treat as observed in our meta-analysis (Chapter 1), imply that standard-dose statin treatment can be used for primary prevention in type 2 diabetes patients. However, the number needed to treat can be high because of a low baseline cardiovascular risk. Standard-dose treatment is currently recommended in Dutch guidelines for patients with a 10-year [8]. We determined that for patients >45 years at diabetes diagnosis standard-dose statin treatment for primary prevention of cardiovascular and cardiovascular risk of ≥10% cerebrovascular events is cost-effective unless the 10-year cardiovascular risk is lower than 6% (Chapter 4).

Diabetes patients initiating statin treatment are at high risk of not reaching the recommended LDL-cholesterol target, especially when they are on low-dose treatment.

162 General discussion

Although for patients with a low or moderate cardiovascular risk intensive-dose treatment might not be preferred due to an increased risk for adverse events, low-dose treatment should be avoided [11,12]. Especially in patients with a high baseline LDL-cholesterol level as they are not expected to reach the LDL-cholesterol target even with an adherence level of 100% on low-dose. Both the decision whether to start statin treatment and the choice for the dose needed, should be based on cost-effectiveness of treatment and should be assessed whether the low absolute reduction in cardiovascular events is likely include a benefit-risk evaluation. Especially for patients with low cardiovascular risk it to outweigh the drawbacks of statin treatment.

The results from Chapter 1 Chapter 2 are based on clinical trials, which are often performed in patients who are treated under controlled conditions [14]. The difference in & caused by non-adherence to treatment [15,16]. In line with others, [3-6] we found that non- efficacy in trials and effectiveness of statin treatment in clinical practice could partly be adherence to treatment has a large impact on LDL-cholesterol outcomes, it however also considerably impacts cost-effectiveness of treatment (Chapter 4). In addition, Chapter 3 showed how the relation between adherence and LDL-cholesterol response was affected by statin dose. Patients on low-dose treatment need to be at least 10% more adherent than patients on standard-dose treatment to reach the same LDL-cholesterol level. Therefore, low-dose treatment should especially be avoided in patients that are vulnerable for non-adherence to treatment.

The studies in this thesis showed that statin treatment resulted in substantial health of statin treatment in clinical practice will be affected by adherence and dosing. This benefits in diabetes patients in clinical trials. However, health and economic outcomes means that suboptimal treatment outcomes can be due to physician-related factors treatment, especially in low risk patients for which treatment might be considered (e.g. treatment dosing) and/or patient-related factors (e.g. adherence). Suboptimal less essential, can lead to very high numbers needed to treat and high costs per QALY in clinical practice. More insight in physician-related and patient-related factors can be helpful for improving health and economic outcomes of statin treatment in clinical practice.

Methodological considerations Type 2 diabetes patients are an important group in cardiovascular risk management due to their increased risk for cardiovascular disease [17,18]. In clinical trials heterogeneity of responsiveness to treatment and the vulnerability to adverse effects [19]. Therefore it treatment effects can be a problem which reflects patient diversity in risk of disease, is important to take account of the population characteristics to retrieve precise effect

163 General discussion

estimates for the target population. We tried to determine the effect of statins in a non-restrictive diabetes population in our meta-analyses. We therefore excluded trials in restrictive subpopulations, for example the Gissi-HF trial [20], which was performed in heart failure patients. Including such trials will affect the relative risk reductions found [19].

The relation between adherence, treatment dose and LDL-cholesterol was assessed in a type 2 diabetes population managed in general practice (Chapter 3). Although we do not expect that the relation between adherence and LDL-cholesterol response will differ in a non-diabetes population, caution should be taken when using these results in non- diabetes populations [19].

In contrast to the cardiovascular and cerebrovascular events used as outcome in the meta-analyses (Chapters 1 & 2), we determined the effect of statins on LDL-cholesterol in the cohort study (Chapter 3). For determining the effect of medical treatment, clinical endpoints like cardiovascular and cerebrovascular events instead of surrogate endpoints as cholesterol levels are commonly preferred [21]. However, statin induced LDL-cholesterol reduction is clearly related to a reduction in cardiovascular events [22]. Determining effects of treatment on clinical endpoints in a heterogeneous population requires a large cohort with a long follow-up. Also, the risk of cardiovascular and cerebrovascular events is not only affected by statin treatment but also by other factors, like the control and treatment of blood pressure and blood glucose [23,24] a non-randomised study. Therefore, we assessed the effect of adherence and treatment , which are difficult to control for in dosing on LDL-cholesterol level instead of cardiovascular events.

Future perspectives standard-dose statin treatment in a type 2 diabetes population. For patients with a The meta-analyses in this thesis show that overall benefits can be expected from

[13]. Future Dutch guidelines on cardiovascular cardiovascular history intensive-dose statin treatment has a significant additional risk management could be more explicit about the limited value of low-dose statin benefit. This is likely to be cost-effective treatment and the added value of intensive-dose statin treatment.

In clinical practice, there can be heterogeneity of treatment effects which will also result in differences in cost-effectiveness. It was, for example, shown that differences in age and 10-year cardiovascular risk resulted in large differences in cost-effectiveness from losing QALY’s to being almost cost neutral (Chapter 4). The concept of personalized or precision medicine (i.e. treatment targeted to the needs of individual patients on the basis of genetics, biomarkers, phenotypic, or psychosocial characteristics [26]) has as goal

164 General discussion

to improve outcomes of individual patients and minimizing adverse events in patients that are less likely to respond to treatment. For example, although current evidence suggests that common genetic variation is not an important determinant for statin LDL- cholesterol response, it seems that SLCO1B1 can be used to predict myopathy [27]. This can be useful in clinical practice as fear of adverse events can be a reason for physicians not to treat patients with statins [28] or not to intensify statin treatment as the risk for adverse events increases with higher doses [11,12,29]. Precision medicine could play a role in improving outcomes of statin treatment by minimizing adverse events and it could also be helpful for more precise targeting at populations in which treatment is cost-effective, but this all requires additional research. The implementation of precision medicine in clinical practice will be challenging especially for primary care providers as they are early signs of disease and navigate referral paths that will have many more branches as the first line in the health care delivery system with the task to prevent disease, identify a result of precision medicine [26].

Furthermore, we did not consider adverse events in our studies. Within the GIANTT population, the number of reported adverse events was low (<2%) and therefore probably underreported. However, it has also been reported that only a small proportion of symptoms reported as statin induced are genuinely caused by statins [30]. Physicians symptoms that are spontaneous or attributable to the nocebo effect (i.e. patients might find it difficult to differentiate between adverse events caused by statins and experiencing symptoms due to negative expectations of treatment) [30]. Uncertainty about the cause of the symptom could be reason that adverse events are not well documented in clinical practice [31]. Active involvement of patients in the process of adverse event monitoring should be further explored [32].

Part II. Patient & physician behaviour regarding statin treatment

of cardiovascular and cerebrovascular events in diabetes patients in Part I of this Benefits of statin treatment were shown for primary and secondary prevention adherence can lead to very high numbers needed to treat, resulting in high costs per thesis. However, it also showed that suboptimal treatment by low-dosing and/or non- QALY. Treatment can be suboptimal due to low-dose treatment [33,34], lack of treatment [35,36,37] [3,4]. Better insights in prescribing and adherence patterns, and patient preferences are needed to improve intensification and/or non-adherence to treatment treatment outcomes in clinical practice.

165 General discussion

Therefore the aims of the second part of this thesis were to: • describe adherence and prescribing patterns of statin treatment in diabetes patients; • determine patient’s preferences for lipid-lowering drugs of diabetes patients.

Main findings In Chapter 5 adherence and LDL-cholesterol outcomes between type 2 diabetes patients initiating on we described differences in treatment modifications, discontinuation, low-dose versus standard-dose statin treatment. This study was performed with data from the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database. Around 22% of the patients started on a dose lower than recommended, more than half of them remained on a low dose during a 2-year follow-up period, whereas less than 15% received a dose increase. Of the patients initiating on standard-dose, also more than half remained on the same treatment during this period, whereas 8% received a dose decrease without subsequent increase. More than 25% of the patients starting on low-dose or on standard-dose treatment discontinued treatment. Patients that initiated treatment at a low dose were older and more often female. There was no difference in baseline LDL-cholesterol. Patients that initiated on low-dose received more and earlier treatment changes than patients initiating on standard-dose treatment. Patients on low- dose treatment had lower adherence levels and were less likely to have adequate LDL- cholesterol response compared to patients on standard-dose after 2 years of follow-up.

treatment was assessed in Chapter 6. As statin treatment is increasingly used for primary The effect of the occurrence of a first cardiovascular event on adherence to statin prevention but treatment cannot prevent all events, an increasing amount of patients know that adherence is higher in secondary prevention populations compared to primary is on statin treatment when experiencing their first cardiovascular event. Although we prevention populations [39-42], the effect of an event while being on statin treatment on adherence was unclear. This matched cohort study was performed with data from the Groningen University IADB.nl pharmacy database. Patients with a cardiovascular event during follow-up were matched to a patient without a cardiovascular event (reference) during follow-up. Adherence was compared before and after the event (for the reference patients for the same follow-up period as their match). Patients with a cardiovascular event had on average a Proportion of Days Covered (PDC) of 81% after the event (index which had a PDC of 71%, while both had a PDC of 79% before the index date. Among the date) which was significantly higher than the patients without a cardiovascular event patients with a cardiovascular event, 26% of the patients became more adherent after after the event. the first cardiovascular event. However, 20% of these patients became less adherent

166 General discussion

Patients’ preferences regarding lipid-lowering treatment were assessed in a questionnaire study with a discrete choice experiment (Chapter 7). We determined (1) the willingness of patients with type 2 diabetes to continue using lipid-lowering drugs,

(2) the importance they attach to specific treatment characteristics, and (3) whether education. Patients needed to imagine that their cholesterol level was too high, and both are influenced by patient characteristics such as age, gender, clinical history and had to choose between two hypothetical treatment options or the option to stop lipid- lowering treatment. Most patients were willing to continue using lipid-lowering drugs. Patients with a history of cardiovascular disease were more willing to continue using lipid-lowering drugs than patients without a history of cardiovascular disease. Patients with a higher education level were less willing to continue using lipid-lowering drugs. Treatment characteristics including cholesterol reduction, risk of early death, risk of limitations due to heart attack, risk of limitations due to stroke, risk of adverse events reduction and risk of adverse events were most important for the choice of treatment. and intake moment, had a significant impact on patients’ treatment choices. Cholesterol Age, education, cardiovascular history and previously experienced adverse effects increasing age, for example, risk reduction of early death and limitations due to a heart influenced the importance patients attach to specific treatment characteristics. With attack became less important for a patient, whereas avoiding adverse effects became more important.

Implications A large proportion of type 2 diabetes patients is not reaching their LDL-cholesterol target [33,38], which was also demonstrated in our study population (Chapter 5). was uncommon [39,40]. Given the adherence rates seen in practice, this is problematic Moreover, 20% of patients started on low-dose treatment and treatment intensification since patients on a low-dose are more likely not to reach the LDL-cholesterol target due to non-adherence (Chapter 3).

There are many factors contributing to non-adherence [41,42], therefore improving adherence in clinical practice is complex and interventions with some degree of success a cardiovascular history was found to be associated with better adherence [43-46]. This are often multifaceted interventions. However, some factors are non-modifiable. Having Chapter 6, where we observed that the occurrence of a cardiovascular event may avert the declining statin adherence rate seen in patients without such an was confirmed in event. In addition, we observed in Chapter 7 that patients with a cardiovascular history patients became less adherent after a cardiovascular event (Chapter 6), indicating that were more willing to continue using lipid-lowering treatment. However, one in five there are still important benefits to achieve. 167 General discussion

In clinical practice high discontinuation rates of statin treatment have been reported [39,47]. Some of these patients, however, may reinitiate treatment [48]. Taking such restarts into account, we showed in Chapter 5 of this thesis that more than a quarter of diabetes Chapter 7 we observed that patients using statins were generally willing to continue such patients still discontinue treatment in the first two years after initiation. In treatment. Surprisingly, having experienced adverse events in the past did not affect the willingness to continue using lipid-lowering treatment, whereas this is often reported as an important reason to discontinue treatment [47,49]. Since our study was performed in a cohort of current users, this indicates that these patients were able to accept or cope with the drawbacks of treatment. Previous research found that discontinuers expect [50-52] information they received about the treatment [51,52]. We observed that higher educated limited benefits . In addition they had less knowledge and were less satisfied with patients were less willing to continue using lipid-lowering treatment (Chapter 7), it and drawbacks of treatment, are less convinced about the usefulness of statin treatment might be that higher educated patients, based on the same information about benefits than lower educated patients.

Part I, can be limited by low-dosing, non- adherence and discontinuation of treatment. The studies in Part II illustrate that to The health benefits of statins, as illustrated in improve the outcomes of statin treatment in practice, a joint approach towards patients of treatment may be needed, taking educational level, age and previous experiences and physicians is needed. Providing tailored information about benefits and drawbacks with cardiovascular events and adverse events into account. In addition, physician- related reasons for low-dosing of treatment should also be addressed in interventions to improve statin use.

Methodological considerations discontinuation rates, adherence and LDL-cholesterol outcomes for patients initiating To provide insight in current practice, we described treatment modification patterns, on low-dose and standard-dose statin treatment in Chapter 5. We were able to use a large unrestricted cohort of diabetes patients managed in the province of Groningen in The Netherlands. We gave a clear view of statin prescribing patterns in this diabetes population, these patterns might not be generalizable to other populations. Also, we did not look into specific subpopulations for example patients with a cardiovascular history. Possible selection bias could have been introduced in the study presented in Chapter 6, where we matched index patients to a reference. This was not possible for a substantial amount of patients because of the strict matching criteria used. The use of strict matching is preferred to prevent the introduction of bias due to poor matching. This

168 General discussion

could have limited the generalizability of the results, however, we found that there were no substantial differences in the age and gender distribution between matched and unmatched index patients.

The low response rate and possible related selection bias is a limitation in Chapter 7. In total, 251 patients (28%) were willing to participate, and 175 patients (19%) could be included in the analysis. There were no differences in gender and age between responders and non-responders. Also, almost half of the participants were lower educated, which corresponds with the national type 2 diabetes population [53]. Nonetheless, participants to surveys commonly have a higher socioeconomic status, better health and healthier lifestyle which could limit the generalizability of the results [54, 55].

Future perspectives Patients starting on low-dose statin treatment appeared to have a similar need for the recommended standard-dose treatment as patients starting on standard-dose treatment. Fear for adverse events was mentioned as reason for physicians not to start statin treatment [28] and could also be reason to start on low-dose instead of standard- dose treatment. As mentioned before precision medicine might be useful for identifying patients at high risk for adverse events. However, more research is needed to determine reasons for physicians to prescribe low-dose statin treatment, as this currently leads to limited health benefits of statin treatment. In addition to low-dosing of treatment, discontinuation of and non-adherence to statin events is often reported as the most important reason for patients to discontinue treatment are also limiting health benefits from statin treatment. Experiencing adverse treatment [49,51,52,56,57]. However, it has also been reported that only a small proportion of symptoms reported as statin induced are genuinely due to statins [30]. This is in line with due to an adverse event [47]. Therefore more effort should be placed on restarting statin findings that most patients can successfully restart statin treatment after discontinuation treatment and motivating patients to continue using statin treatment.

interventions to improve adherence are often not very effective [58]. To be effective For statin users health benefits can be increased by improving adherence. Current taking account of patients’ preferences is essential. Interventions with some degree of success are multifaceted combinations which include patient education, patient- physician communication enhancement, and increased patient monitoring and follow- up. In this precision medicine can also play a role as technology can assist in monitoring cholesterol levels, potential adverse events and adherence to medication [26]. Designing feedback systems and tailored incentives for the individual patient might help changing the behaviour of both the patient [58] and the physician [59, 60]. 169 General discussion

Overall conclusion

Part I with diabetes with and without cardiovascular history. The impact at a population of this thesis showed that statins result in substantial health benefits in patients level can, however, be limited by suboptimal treatment resulting from factors like low- dosing and non-adherence to treatment. Part II of this thesis showed that low-dosing, non-adherence and discontinuation of statin treatment are common in clinical practice. on (1) improving statin prescribing by determining physician-related reasons for Therefore, interventions to improve health benefits of statin treatment should target low-dosing of treatment, (2) improving adherence to treatment for which providing educational level, age and previous experiences with cardiovascular events and adverse tailored information about benefits and drawbacks of treatment may be needed, taking events into account, (3) more tailored prescribing of intensive dose statin treatment especially for those patients at high cardiovascular risk and (4) precision medicine which could help identifying patients at risk for adverse events.

this is currently limited in clinical practice due to low dosing and non-adherence. To be This thesis shows that health and economic benefits of statin treatment can be high, but able to improve prescribing and adherence rates, interventions should not only target patients, but also the physician.

170 General discussion

REFERENCES

Baigentprospective C, Keech meta-analysis A, Kearney of PM,data et from al. Efficacy90,056 participantsand safety of in cholesterol-lowering 14 randomised trials treatment: of statins. Lancet 2005; 366(9493): 1267-1278. 2. Taylor F, Ward K, Moore TH, et al. Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev 2011; (1): CD004816. 3. Parris ES, Mohn LA, Lawrence DB, Long LB. Adherence to statin therapy and LDL cholesterol goal attainment by patients with diabetes and dyslipidemia. Diabetes Care 2005; 28(3): 595-599. 4. Watanabe JH, Bounthavong M, Chen T. Revisiting the medication possession ratio threshold for adherence in lipid management. Curr Med Res Opin 2013; 29(3): 175-180. 5. Shalev V, Goldshtein I, Halpern Y, Chodick G. Association between persistence with statin therapy and reduction in low-density lipoprotein cholesterol level: analysis of real-life data from community settings. Pharmacotherapy 2014; 34(1): 1-8. 6. Cheng CW, Woo KS, Chan JC, Tomlinson B, You JH. Association between adherence to statin therapy and lipid control in Hong Kong Chinese patients at high risk of coronary heart disease. Br J Clin Pharmacol 2004; 58(5): 528-535. 7. Rozenbaum MH, Sanders EAM, van Hoek AJ, et al. Cost effectiveness of pneumococcal vacci- nation among Dutch infants: an economic analysis of the seven valent pneumococcal conju- gated vaccine and forecast for the 10 valent and 13 valent vaccines. BMJ 2010; 340: c2509. 8. NHG-standaard Cardiovasculair risicomanagement (Eerste herziening) Huisarts Wet 2012; 55(1): 14-28. 9. Rutten GEHM, De Grauw WJC, Nijpels G, et al. NHG-Standaard Diabetes mellitus type 2 (derde herziening). Huisarts Wet 2013; 56: 512-525. 10. American Diabetes Association. Standards for medical care in diabetes – 2015. Diabetes Care 2015; 38(Suppl. 1): S1–S90. 11. Silva M, Matthews ML, Jarvis C, et al. Meta-analysis of drug-induced adverse events associated with intensive-dose statin therapy. Clin Ther 2007; 29(2): 253-260. 12. Newman C, Tsai J, Szarek M, Luo D, Gibson E. Comparative safety of atorvastatin 80 mg versus 10 mg derived from analysis of 49 completed trials in 14,236 patients. Am J Cardiol 2006; 97(1): 61-67. 13. Taylor DC, Pandya A, Thompson D, et al. Cost-effectiveness of intensive atorvastatin therapy in secondary cardiovascular prevention in the United Kingdom, Spain, and Germany, based on the Treating to New Targets study. Eur J Health Econ 2009; 1D(3); 255-265. 14. Nallamothu BK, Hayward RA, Bates ER. Beyond the randomized clinical trial. The role of effectiveness studies in evaluating cardiovascular therapy. Circulation 2008; 118(12): 1294-1303. 15.

DavidsonAm J Manag MH. Care Differences 2006; 12(15 between Suppl): S405-11. clinical trial efficacy and real-world effectiveness.

171 General discussion

16. Frolkis JP, Pearce GL, Nambi V, Minor S, Sprecher DL. Statins do not meet expectations for lowering low-density lipoprotein cholesterol levels when used in clinical practice. Am J Med 2002; 113(8): 625-629. 17.

Available at: http://www.heart.org/HEARTORG/Conditions/Diabetes/WhyDiabetesMatters/ 18. Cardiovascular-Disease-Diabetes_UCM_313865_Article.jsp Assessed 8 May 2015. Miettineninfarction. H, The Lehto FINMONICA S, Salomaa myocardial V, et al. Impact infarction of diabetes register on mortality study group. after Diabetesthe first myocardialCare 1998 Jan; 21(1): 69-75. 19. Kravitz RL, Duan N, Braslow J. Evidence-based medicine, heterogeneity of treatment effects, and the trouble with averages. Milbank Q 2004; 82(4): 661-687. 20. Tavazzi L, Maggioni AP, Marchioli R, et al. Effect of rosuvastatin in patients with chronic heart failure (the GISSI-HF trial): a randomised, double-blind placebo-controlled trial. Lancet 2008; 372(9645): 1231-1239. 21. Psaty BM, Weiss NS, Furberg CD, et al. Surrogate end points, health outcomes, and the drug- approval process for the treatment of risk factors for cardiovascular disease. JAMA 1999; 282(8): 786-790. 22. Cholesterol Treatment Trialists’ (CTT) Collaboration, Baigent C, Blackwell L, Emberson J, et al.

Efficacy170,000 and participants safety of morein 26 intensiverandomised lowering trials. Lancetof LDL cholesterol:2010; 376(9753): a meta-analysis 1670-1681. of data from 23. Kothari V, Stevens RJ, Adler AI, et al. UKPDS 60: Risk of stroke in type 2 diabetes estimated by the UK Prospective Diabetes Study risk engine. Stroke 2002; 33(7): 1776-1781. 24. Stevens RJ, Kothari V, Adler AI, Stratton IM. The UKPDS risk engine: a model for the risk of coronary heart disease in Type II diabetes (UKPDS 56). Clin Sci (London) 2001; 101(6): 671-679. 25. Sokol MC, McGuigan KA, Verbrugge RR, Epstein RS. Impact of medication adherence on hospitalization risk and health care cost. Med Care 2005; 43(6): 521–530. 26. Jameson JL, Longo DL. Precision Medicine – Personalized, Problematic, and Promising. N Engl J Med 2015; 372(23): 2229-2234. 27. Hopewell JC, Reith C, Armitage J. Pharmacogenomics of statin therapy: any new insights in Curr Opin Lipidol 2014; 25(6):438-445.

28. efficacyAb E, Denig or safety? P, van Vliet T, Dekker JH. Reasons of general practitioners for not prescribing lipid- lowering medication to patients with diabetes: a qualitative study. BMC Family Practice 2009; 10: 24. 29. Armitage J. The safety of statins in clinical practice. Lancet 2007; 370(9601):1781-1790. 30. Finegold JA, Manisty CH, Goldacre B, Barron AJ, Francis DP. What proportion of symptomatic side effects in patients taking statins are genuinely caused by the drug? Systematic review of randomized placebo-controlled trials to aid individual patient choice. Eur J Prev Cardiol 2014; 21(4): 464-474.

172 General discussion

31. Skentzos S, Shubina M, Plutzky J, Turchin A. Structured vs. Unstructured: Factors affecting adverse drug reaction documentation in an EMR repository. AMIA Annu Symp Proc 2011; 2011: 1270-1279. 32. Budnitz DS, Layde PM. Outpatient drug safety: new steps in an old direction. Pharmacoepidemiol Drug Saf 2007;16(2): 160-165. 33. Heintjes EM, Penning-van Beest FJ, Plat AW, et al. Cholesterol level goal attainment with statins: clinical management guideline recommendations versus management in actual clinical practice. Pharmacotherapy 2012; 32(7): 631-641. 34. Geleedst-De Vooght M, Maitland-van der Zee AH, Schalekamp T, Mantel-Teeuwisse A, Jansen P. Statin prescribing in the elderly in the Netherlands: A pharmacy database time trend study. Drugs Aging 2010; 27(7): 589-596. 35.

Rodondihypertension, N, Peng dyslipidemia, T, Karter AJ, and et diabetes al. Therapy mellitus. modifications Ann Intern in Med response 2006; to144(7): poorly 475-484. controlled 36. Schmittdiel JA, Uratsu CS, Karter AJ, et al. Why don’t diabetes patients achieve recommended J Gen InternMed

risk2008; factor 23(5): targets? 588-594. Poor adherence versus lack of treatment intensification. 37. -

adherencetimeliness. toPLoS guidelines One 2011; in 6(9):diabetes e24278. care according to different definitions of adequacy and 39. Simpson RJ, Tunceli K, Ramey DR, et al. Treatment pattern changes in high-risk patients newly initiated on statin monotherapy in a managed care setting. J Clin Lipidol 2013; 7(5): 399-407. 40. Fox FM, Gandhi SK, Ohsfeldt RL, Blasetto JW, Davidson MH. Titration patterns with rosuvastatin as compared with other statins in clinical practice: A retrospective observational cohort study using an electronic medical record database. Clin Ther 2007; 29(11): 2385-2394. 41. Bates TR, Connaughton VM, Watts GF. Non-adherence to statin therapy: a major challenge for preventive cardiology. Expert Opin Pharmacother 2009; 10(18): 2973-2985. 42. Osterberg L, Blaschke T. Adherence to medication. N Eng J Med 2005; 353(5): 487-497. 43. Chodick G, Shalev V, Gerber Y, et al. Long-term persistence with statin treatment in a not-for-

profitisrael. Clin health Ther maintenance 2008; 30(1): organization: 2167-2179. A population-based retrospective cohort study in 44.

Latryin a real-life P, Molimard setting M, isDedieu better B, when Couffinhal associated T, Bégaud cardiovascular B, Martin-Latry risk factors K. Adherence increase: with A statinscohort study. BMC Cardiovasc Disord 2011; 11: 46. 45. Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein MC, Avorn J. Long-term persistence in use of statin therapy in elderly patients. JAMA 2002; 288(4): 455-461.

173 General discussion

46. Donnelly LA, Doney AS, Morris AD, Palmer CN, Donnan PT. Long-term adherence to statin treatment in diabetes. Diabet Med 2008; 25(7): 850-855. 47. Zhang H, Plutzky J, Skentzos S, et al. Discontinuation of statins in routine care settings: a cohort study. Ann Intern Med 2013; 158(7): 526-534. 48. Korhonen MJ, Helin-Salmivaara A, Huupponen R. Dynamics of long-term statin therapy. Eur J Clin Pharmacol 2011; 67(9): 925-931. 49. Wei MY, Ito MK, Cohen JD, Brinton EA, Jacobson TA. Predictors of statin adherence, switching, and discontinuation in the USAGE survey: understanding the use of statins in America and gaps in patient education. J Clin Lipidol 2013; 7(5): 472-483. 50. Mann DM, Woodward M, Muntner P, Falzon L, Kronish I. Predictors of nonadherence to statins: a systematic review and meta-analysis. Ann Pharmacother 2010; 44(9): 1410-1421. 51. McGinnes B, Olson KL, Magid D, et al. Factors related to adherence to statin therapy. Ann Pharmacother 2007; 41(11): 1805-1811. 52. Fung V, Sinclair F, Wang H, Dailey D, Hsu J, Shaber R. Patients’ perspectives on nonadherence to statin therapy: a focus-group study. Perm J 2010; 14(1): 4-10. 53.

https://www.volksgezondheidenzorg.info/onderwerp/diabetes-mellitus/cijfers-context/bevolkingsgroepen#node-ongediagnosticeerde-diabetes-naar-opleidingsniveau Accessed 8 April 2015. 54. Bergstrand R, Vedin A, Wilhelmsson C, Wilhelmsson L. Bias due to non-participation and heterogenous sub-groups in population surveys. J Chronic Dis 1983; 36(10): 725-728. 55. Van Loon AJ, Tijhuis M, Picavet HS, Surtees PG, Ormel J. Survey non-response in the Nether- lands: effects on prevalence estimates and associations. Ann Epidemiol 2003; 13(2): 105-110. 56. Maningat P, Gordon BR, Breslow JL. How do we improve patient compliance and adherence to long-term statin therapy? Curr Atheroscler Rep 2013; 15(1): 291. 57. Cohen JD, Brinton EA, Ito MK, Jacobsen TA. Understanding statin use in America and gaps in patient education (USAGE): An internet-based survey of 10,138 current and former statin users. J Clin Lipidol 2012; 6(3): 208-215. 58. Nieuwlaat R, Wilczynski N, Navarro T, et al. Interventions for enhancing medication adherence (Review). Cochrane Database Syst Rev 2014; 11: CD000011. 59. Asch DA, Muller RW, Volpp KG. Automated hovering in health care – Watching over the 5000 hours. N Engl J Med 2012; 367(1): 1-3. 60. Ostini R, Hegney D, Jackson C, et al. Systematic review of interventions to improve prescribing. Ann Pharmacother 2009; 43(3): 502-513.

174 General discussion

175

Nederlandse samenvatting Nederlandse samenvatting

Dit proefschrift richt zich op statinetherapie in type 2 diabetespatiënten; hiervan zijn verschillende aspecten onderzocht. In Deel I worden de effecten van statines op LDL- cholesterol en cardiovasculaire en cerebrovasculaire uitkomsten bepaald. Daarnaast is de kosteneffectiviteit van statinetherapie geschat waarbij gebruik is gemaakt van informatie uit de klinische praktijk. In Deel II ligt de focus op het gedrag van de patiënt en de arts ten aanzien van statinetherapie. Hieronder wordt per deel een korte introductie gegeven en worden de belangrijkste bevindingen uit de onderzoeken weergegeven.

Deel I. Effecten en kosteneffectiviteit van statinetherapie

Voor secundaire preventie van cardiovasculaire en cerebrovasculaire events is het belang van statinetherapie breed erkend; voor primaire preventie is het gebruik van statines meer controversieel. De absolute risicoreductie door statinetherapie kan klein zijn door een laag cardiovasculair risico. Het is gerapporteerd dat statinetherapie geassocieerd is met een vijfde reductie in de incidentie van cardiovasculaire en cerebrovasculaire patiëntkarakteristieken. Echter, andere onderzoeksgroepen hebben gerapporteerd dat events per mmol/l reductie in LDL-cholesterol, grotendeels ongeacht specifieke er slechts beperkt bewijs is dat statinetherapie bij primaire preventie kosteneffectief is en de kwaliteit van leven van de patiënt verbetert. Er is geen accurate effectschatting voor statinetherapie bij diabetespatiënten met en zonder doorgemaakt cardiovasculair event. Eerdere analyses in primaire preventie populaties includeerden ook secundaire preventie patiënten. Een accurate effectschatting is echter belangrijk bij het nemen van klinische beslissingen en voor kosteneffectiviteitsanalyses, voornamelijk bij primaire preventie waar de balans tussen voor- en nadelen van therapie mogelijk niet optimaal is. In klinische geneesmiddelonderzoeken worden patiënten vaak behandeld onder gecontroleerde condities, in tegenstelling tot de klinische praktijk. In de klinische praktijk wordt de respons en kosteneffectiviteit van statinetherapie beïnvloed door bijvoorbeeld therapietrouw. Het is onduidelijk in welke mate therapieontrouw leidt tot verminderde therapierespons; deze relatie wordt waarschijnlijk beïnvloed door de dosering.

De doelen van het eerste deel van dit proefschrift waren: • Het bepalen van de effect schattingen van statinetherapie voor de primaire en secundaire preventie van cardiovasculaire en cerebrovasculaire events bij diabe- tespatiënten; • Het effect van statine dosering en therapietrouw op cholesteroluitkomsten in diabetespatiënten schatten;

178 Nederlandse samenvatting

• De kosteneffectiviteit van statinetherapie voor de primaire preventie van cardiovasculaire en cerebrovasculaire events bij nieuw gediagnosticeerde type 2 diabe- tespatiënten bepalen, rekening houdend met therapietrouw.

Belangrijkste bevindingen In hoofdstuk 1 is het effect van statinetherapie op het optreden van een eerste cardiovasculair of cerebrovasculair event in diabetespatiënten bepaald aan de hand van een meta-analyse. Gebaseerd op resultaten van vier studies met gezamenlijk 10.187 deelnemers, resulteerde statinetherapie voor primaire preventie van een cardiovasculair of cerebrovasculair event te voorkomen moeten 35 patiënten 3,8 jaar behandeld worden. of cerebrovasculair event in een relatieve risico daling van 25%. Om één cardiovasculair

Significante relatieve risico reducties van 31% voor een fataal en niet-fataal beroerte en 16% voor algehele mortaliteit werden gevonden. 30% voor een fataal en niet-fataal hartaanval, en een niet-significante risico reductie van

De effectiviteit van standaarddosering en intensieve dosering statinetherapie voor de secundaire preventie van een cardiovasculair of cerebrovasculair event in diabetespatiënten is in hoofdstuk 2 bepaald. Gebaseerd op gegevens van vijf onderzoeken met gezamenlijk 4.351 deelnemers, resulteerde standaarddosering statinetherapie vergeleken met placebo in een significante relatieve risico reductie van 15% voor een cardiovasculair of cerebrovasculair event. Er was een significante relatieve risico reducties van 27% voor fataal en niet-fataal hartaanval en van 22% voor relatieve risico reductie van 33% voor fataal en niet-fataal beroerte. Niet-significante algehele mortaliteit zijn gevonden. Vergeleken met standaarddosering statinetherapie resulteerde een intensieve dosering statinetherapie in een extra 9% relatieve risico reductie voor een cardiovasculair of cerebrovasculair event. Dit was gebaseerd op data van vier onderzoeken met gezamenlijk 4.805 deelnemers.

In hoofdstuk 3 is een cohort studie uitgevoerd met observationele data van de Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) databank. De associatie tussen therapietrouw, dosering en LDL-cholesterol respons in type 2 diabetespatiënten die statinetherapie starten is bepaald. Voor patiënten met een baseline LDL-cholesterol een 48% reductie in LDL-cholesterol bij standaarddosering. Een 20% reductie in waarde van 3,7 mmol/l, werd een therapietrouw niveau van 100% geassocieerd met therapietrouw in deze patiënten werd geassocieerd met 8-9% minder daling in LDL- een therapietrouwniveau van tenminste 63% op een standaarddosering nodig om de cholesterol. Patiënten met een baseline LDL-cholesterol waarde van 3,7 mmol/l hebben van 76% nodig. streefwaarde van 2,5 mmol/l te halen, en bij lage dosering is een therapietrouwniveau

179 Nederlandse samenvatting

In hoofdstuk 4 is de kosteneffectiviteit van statinetherapie voor de primaire preventie van cardiovasculaire en cerebrovasculaire events in nieuw gediagnosticeerde type 2 diabetespatiënten bepaald. Het effect van therapietrouwniveaus zoals waargenomen in de praktijk is bepaald. Rekening houdend met verwachte therapietrouwniveaus (81%, 77%, 75% respectievelijk voor het eerste, tweede en opvolgende jaren) was het starten van statinetherapie op het moment van type 2 diabetes diagnose kosteneffectief met €2.245 per QALY in de base case (een 60-jarige patiënt met het gemiddelde 10-jaar cardiovasculair risico van patiënten in de leeftijd 56-65 jaar bij type 2 diabetes diagnose). Verschillen in leeftijd en 10-jaars risico voor cardiovasculaire en cerebrovasculaire events zorgden voor grote verschillen in kosteneffectiviteit, uiteenlopend van het verlies van QALY’s tot bijna kosten neutraal. Met de therapietrouwniveaus gezien in de klinische praktijk, kan geconcludeerd worden dat behandeling met statines voor alle type 2 diabetespatiënten onder de 45 jaar bij type 2 diabetes diagnose niet kosteneffectief is. Voor patiënten in de leeftijd van 45 tot 55 jaar bij type 2 diabetes diagnose is statinetherapie kosteneffectief tenzij het 10-jaars cardiovasculaire risico 6% of lager is. Voor alle andere patiënten wordt verwacht dat statinetherapie kosteneffectief is.

Deel II. Gedrag van patiënten en artsen ten aanzien van statinetherapie

Voordelen van statinetherapie zijn aangetoond voor primaire en secundaire preventie van cardiovasculaire en cerebrovasculaire events in diabetespatiënten in Deel I van dit proefschrift. Echter, er werd ook aangetoond dat een lage dosering en/of Dit betekent dat de behandeling enerzijds suboptimaal kan zijn door een lage dosering therapieontrouw voor inefficiëntie kan zorgen, wat resulteert in hoge kosten per QALY. of gebrek aan intensivering van de behandeling, en anderzijds door therapieontrouw. Betere inzichten in voorschrijf- en therapietrouwpatronen, en gerelateerde voorkeuren van patiënten ten aanzien van cholesterolverlagende middelen zijn nodig om uitkomsten van statinetherapie in de klinische praktijk te verbeteren.

De doelen van het tweede deel van dit proefschrift waren: • Het beschrijven van therapietrouw- en voorschrijfpatronen van statinetherapie in diabetespatiënten; • Het bepalen van patiënten preferenties voor cholesterolverlagende middelen in een diabetes populatie.

180 Nederlandse samenvatting

Belangrijkste bevindingen In hoofdstuk 5 zijn verschillen in therapieveranderingen, stoppen, therapietrouw en LDL-cholesterol uitkomsten beschreven tussen type 2 diabetespatiënten die starten op lage versus standaarddosering. Deze studie is uitgevoerd met data van de GIANTT databank. Ongeveer 22% van de patiënten startte op een lagere dosering dan aanbevolen, meer dan de helft van deze patiënten bleef op een lage dosering gedurende twee jaar follow-up, terwijl minder dan 15% een dosisverhoging kreeg. Van de patiënten die op een standaard dosis startten bleef ook meer dan de helft op dezelfde dosis gedurende twee jaar follow-up, terwijl 8% een dosisverlaging kreeg zonder opvolgende dosisverhoging. Meer dan 25% van de patiënten die zijn gestart op een lage of standaarddosering stopten met de behandeling. Patiënten die begonnen met een lage dosis waren ouder en vaker vrouw. Er was geen verschil in baseline LDL-cholesterol waarden. Patiënten die begonnen op een lage dosis kregen meer en eerder therapie veranderingen dan patiënten die op standaarddosering begonnen. Patiënten op een lage dosering hadden een lagere geschatte therapietrouw en een lagere kans op een adequate LDL-cholesterol respons vergeleken met patiënten op een standaarddosering na twee jaar follow-up.

Het effect van een eerste cardiovasculair event op statine therapietrouw is bepaald in hoofdstuk 6. Doordat statinetherapie steeds meer gebruikt wordt voor primaire preventie, maar behandeling niet alle cardiovasculaire events kan voorkomen, is er een steeds groter aantal patiënten dat statines gebruikt ten tijde van het eerste cardiovasculaire event. Alhoewel we weten dat therapietrouw hoger is in secundaire preventie populaties dan in primaire preventie populaties is het effect van het eerste cardiovasculair event op therapietrouw niet duidelijk. Deze gematchte cohort studie is uitgevoerd met data van de IADB.nl apotheekdatabank van de Rijksuniversiteit Groningen. Patiënten met een cardiovasculair event werden gematcht met een patiënt zonder cardiovasculair event (referentie). Therapietrouw werd vergeleken voor en na het cardiovasculair event (voor referentie patiënten gedurende dezelfde follow-up periode als hun match). Patiënten met een cardiovasculair event hadden gemiddeld een geschatte therapietrouw (gemeten in ‘proportion of days covered’ (PDC)) van van 71% hadden terwijl beide een PDC van 79% voor het event hadden. Onder de 81% na het event. Dit was significant hoger dan de referentie patiënten, die een PDC patiënten met cardiovasculair event had 26% een hogere therapietrouw waarde na het eerste cardiovasculair event. Aan de andere kant was 20% van de patiënten minder therapietrouw na het event.

Preferenties van patiënten ten aanzien van cholesterolverlagende middelen zijn bepaald in een vragenlijstonderzoek met een discrete choice experiment in hoofdstuk 7. Bepaald zijn (1) de bereidheid van patiënten met type 2 diabetes om door te gaan met

181 Nederlandse samenvatting

of uitkomsten voor hen, en (3) de invloed van patiënt karakteristieken zoals leeftijd, cholesterolverlagende middelen, (2) het belang van specifieke therapiekarakteristieken geslacht, klinische historie en educatie hierop. Patiënten moesten zich indenken dat hun cholesterol te hoog was en moesten kiezen tussen twee hypothetische behandelopties dan wel de optie om te stoppen met de cholesterolverlagende behandeling. De meeste patiënten waren bereid om door te gaan met de cholesterolverlagende middelen. Patiënten met een doorgemaakt cardiovasculair event waren meer bereid om door te gaan met de behandeling dan patiënten zonder doorgemaakt cardiovasculair event. Patiënten met hogere educatie waren minder bereid om door te gaan met de therapie. De behandeluitkomsten ten aanzien van cholesterolverlaging, risico op vroegtijdig overlijden, risico op beperkingen door een hartaanval, risico op beperkingen door een beroerte, risico op bijwerkingen en het inname moment hadden allemaal een risico op bijwerkingen waren het meest belangrijk. Leeftijd, educatie, cardiovasculaire significante invloed op de behandelkeuze van de patiënten. Cholesterolverlaging en therapiekarakteristieken. Met toenemende leeftijd bijvoorbeeld, werden het risico op historie en eerdere ervaringen met bijwerkingen beïnvloedden het belang van specifieke vroegtijdig overlijden en beperkingen door een hartaanval minder belangrijk voor de patiënt, terwijl het risico op bijwerkingen belangrijker werd.

Deel I van dit proefschrift heeft laten zien dat statines een aanzienlijk gezondheidsvoordeel hebben in diabetespatiënten met en zonder doorgemaakt cardiovasculair event. Het effect kan echter beperkt zijn door suboptimale therapie veroorzaakt door factoren als lage doseringen en therapieontrouw. Deel II van dit proefschrift heeft laten zien dat lage doseringen, therapieontrouw en stoppen van statinetherapie veelvoorkomend is in de klinische praktijk. Interventies om statinegebruik en uitkomsten te verbeteren kunnen zich richten op (1) het verbeteren van het voorschrijven van statines door arts- gerelateerde redenen voor het voorschrijven van lage doseringen te bepalen, (2) het voorschrijven van hoge dosering statinetherapie voor patiënten met een hoog risico op cardiovasculaire events, (3) het verbeteren van therapietrouw door het leveren van op maat gemaakte informatie over voor- en nadelen van therapie, rekening houdend met het educatie niveau, leeftijd en eerdere ervaringen met cardiovasculaire events en bijwerkingen, (4) ‘precision medicine’ die mogelijk kan helpen om bijvoorbeeld patiënten met een verhoogd risico op bijwerkingen te identificeren. Dit proefschrift geeft aan dat statinetherapie kan zorgen voor economische en gezondheid gerelateerde voordelen. In de praktijk zijn deze voordelen momenteel beperkt door het voorschrijven van lage doseringen en door therapieontrouw. Om de gezondheid en economische winst te verbeteren moeten interventies zich richten op de patiënten en de arts.

182 Nederlandse samenvatting

183

Appendices Appendices

APPENDIX I

SUPPLEMENTAL TABLE S1. Population on standard-dose excluded (missing measurements) versus included Variable Standard-dose excluded Standard-dose included (n=3,886) (n=2,160) Age, mean (SD) 62.5 (12.6) 61.4 (11.3) Male sex 55% 48% SBP, mmHg (SD) 145.1 (20.7) 143.5 (19.0) DBP, mmHg (SD) 82.8 (11.1) 82.5 (9.9) Diabetes duration, years (IQR) 1 (0-5) 0 (0-3) Fasting glucose, mmol/l 8.4 (3.2) 8.1 (2.7) HbA1c (%) 7.3 (1.6) 7.2 (1.6) LDL cholesterol, mmol/l 3.7 (1.0) 3.8 (0.9) > target (%) 1752 (90%)* 2000 (93%) <= target (%) 204 (10%) 160 (7%) HDL cholesterol, mmol/l 1.2 (0.3) 1.2 (0.3) Triglycerides, mmol/l 2.5 (2.2) 2.0 (0.8) Total cholesterol, mmol/l 5.7 (1.3) 5.8 (1.0) Creatinine, μmol/l 84.1 (23.4) 80.4 (18.5) Potassium, mmol/l 4.2 (0.4) 4.2 (0.4) Hemoglobin, mmol/l 8.9 (1.0) 9.1 (0.8) Body mass index 30.6 (5.8) 30.5 (5.6) Smoking 28.2% 28.9%

Comorbidity: Hypertension 38.1% 32.9% Angina pectoris 4.4% 1.6% Myocardial infarction/IHD 5.3% 1.9% Heart failure 2.9% 1.0% Bypass/angioplasty 3.3% 1.3% Lipid disorder 8.0% 6.6% Stroke/CVA/TIA/atherosclerosis 5.3% 3.0% Microvascular complications 3.8% 3.2% CVA: cerebrovascular accident; DBP: diastolic blood pressure; HDL: high-density lipoprotein; IHD: ischemic heart disease; LDL: low-density lipoprotein; SBP: systolic blood pressure; SD: standard deviation; TIA: transient ischemic attack. * Percentage calculated over population with LDL-cholesterol measurement.

186 Appendices

SUPPLEMENTAL TABLE S2. Balance between low-dose and standard-dose patients in the propensity-score matched cohort Variable low-dose mean low-dose SD standard-dose standard-dose SDD (n=572) mean (n=572) SD Age, mean 64.0 10.7 65.0 11.0 0.092 female sex 55.9 . 57.5 . 0.032 SBP, mmHg 143.4 19.3 144.7 19.6 0.067 DBP, mmHg 82.1 10.2 82.6 10.9 0.049 Diabetes duration 3.6 4.7 3.3 4.8 0.056 Fasting glucose, mmol/l 7.9 2.4 7.8 2.4 0.054 HbA1c (%) 7.1 1.4 7.0 1.3 0.012 LDL cholesterol, mmol/l 3.7 0.8 3.7 0.9 0.025 HDL cholesterol, mmol/l 1.3 0.3 1.3 0.3 0.072 Triglycerides, mmol/l 1.9 0.8 1.8 0.8 0.083 Total cholesterol, mmol/l 5.8 1.0 5.7 1.0 0.034 Albuminuria 17.7 . 17.6 . 0.002 Creatinine, μmol/l 80.3 18.5 79.2 18.1 0.057 Potassium, mmol/l 4.3 0.5 4.3 0.5 0.008 Hemoglobin, mmol/l 9.0 1.0 9.0 0.9 0.003 Body mass index 30.6 7.1 30.4 5.9 0.034 Smoking status 22.3 . 23.1 . 0.020 Comorbidity: Hypertension 34.6 . 32.7 . 0.041 Angina pectoris 3.7 . 2.8 . 0.049 Myocardial infarction/IHD 2.1 . 2.8 . 0.045 Heart failure 1.6 . 1.6 . 0.000 Bypass/angioplasty 1.2 . 0.5 . 0.075 Atherosclerosis 0.2 . 0.2 . 0.000 Lipid disorder 7.9 . 7.2 . 0.027 TIA 1.0 . 0.9 . 0.018 Stroke/CVA 1.0 . 1.2 . 0.016 PTA 0.0 . 0.0 . 0.000 PTCA 0.7 . 0.9 . 0.038 CABG 0.5 . 0.3 . 0.027 Peripheral bypass 0.2 . 0.0 . 0.059 Proteinuria 1.0 . 1.6 . 0.046 Nephropathy 0.0 . 0.0 . 0.000 Neuropathy 0.7 . 0.9 . 0.020 Diabetic foot 0.2 . 0.3 . 0.034 Retinopathy 1.7 . 1.4 . 0.028 Blindness 0.5 . 0.5 . 0.000

Pre-matching c-statistic: 0.648 Post-matching c-statistic: 0.570 CABG: coronary artery bypass grafting; CVA: cerebrovascular accident; DBP: diastolic blood pressure; HDL: high-density lipoprotein; IHD: ischemic heart disease; LDL: low-density lipoprotein; PTA: peripheral bypass and percutaneous transluminal angioplasty; PTCA: percutaneous transluminal coronary angioplasty; SBP: systolic blood pressure; SD: standard deviation; SDD: Standardized difference of the mean; TIA: transient ischemic attack.

187 Appendices

Variable Coefficient (β) Standard error p-value Adherence (PDC) (%) -0.0065 0.00029 <0.001 LDL-cholesterol at baseline (mmol/l) 0.1664 0.00767 <0.001 Age (years) -0.0025 0.00061 <0.001 Lipid disorder (yes/no) 0.0720 0.02808 0.010 Fasting glucose (mmol/l) -0.0107 0.00296 <0.001 Constant 0.9307 0.06003 <0.001 R2 = 0.383, p-value = <0.001 LDL-c, LDL-cholesterol; PDC, proportion of days covered

SUPPLEMENTAL FIGURE S1: Predicted LDL-c response on standard-dose treatment for different adherence rates using multivariate linear regression. Patients needed to have an LDL-cholesterol measurement in the period (initiation date + 360) -/+ 90 days, (n=1,595).

188 Appendices

Variable Coefficient (β) Standard error p-value Adherence (PDC) (%) -0.0065 0.00037 <0.001 LDL-cholesterol at baseline (mmol/l) 0.1712 0.00972 <0.001 Age (years) -0.0021 0.00079 0.008 Fasting glucose (mmol/l) -0.0107 0.00352 0.002 PTCA (yes/no) 0.2608 0.12177 0.032 constant 0.8887 0.07385 <0.001 R2 = 0.396, p-value = <0.001 LDL-c, LDL-cholesterol; PDC , proportion of days covered; PTCA, percutaneous transluminal coronary angioplasty SUPPLEMENTAL FIGURE S2: Predicted LDL-c response on standard-dose treatment for different adherence rates using multivariate linear regression. Patients needed to have an LDL-cholesterol measurement in the period (initiation date + 360) -/+ 45 days, (n=941).

189 Appendices

APPENDIX II

Identification of primary and secondary prevention patients with type 2 diabetes in a prescription database

Background When using a pharmacy database, such as the IADB.nl database, medication proxies are is to test the accuracy of medication proxies for identifying primary and secondary needed to identify patients with a specific condition or event. The aim of this analysis prevention patients from a type 2 diabetes population by using the Groningen Initiative to Analyse Type 2 Diabetes Treatment (GIANTT) database as reference, which contains general practitioner data [1].

Methods The accuracy of the medication proxies for identifying primary and secondary prevention patients was tested for patients from the cohort of 2008 who received two prescriptions for ATC A10B in this year, to meet the inclusion criteria for type 2 diabetes from the IADB. Multiple medication proxies were tested within the GIANTT database, using the cardiovascular events and disease history of the patients as documented in the general practitioner medical records as a gold standard (ICPC codes K75, K76, K90). Proxies were all based on having two prescriptions within the year 2008, to identify chronic use.

Sensitivity and specificity of the proxies were calculated. Results Medication proxies were tested on 6,400 diabetes patients from the GIANTT database.

Proxy 7 (thrombocyte aggregation inhibitors or vitamin K antagonists) showed the highest Sensitivity and specificity of the medication proxies are presented in table Appendix I. sensitivity, which is needed to exclude all patients likely to be secondary prevention patients. For identifying patients who experience a drug-treated cardiovascular event, much of the sensitivity. proxy 5 (thrombocyte aggregation inhibitors) showed a high specificity without loss of

Conclusion To identify patients that are expected to initiate statin treatment for primary prevention, patients receiving at least two prescriptions for a thrombocyte aggregation inhibitor and/or a vitamin K antagonist before or around statin initiation should be excluded. thrombocyte aggregation inhibitor is considered as the optimal medication proxy. For identification of the occurrence of a cardiovascular event, two prescriptions for a

190 Appendices

TABLE APPENDIX I. Sensitivity and specificity measurements of the medication proxies tested on general practitioner data from the GIANTT database Proxy (ATC) Sensitivity Specificity 1. Nitrates (C01DA) 0.17 0.96 2. Vitamin K antagonists (B01 AA) 0.19 0.91 3. Calcium antagonists (C08) 0.33 0.79 4. Beta-blockers (C07A) 0.68 0.62 5. Thrombocyte aggregation inhibitors (B01AC) 0.69 0.79 6. RAAS inhibitors (C09) 0.75 0.40 7. Thrombocyte aggregation inhibitors (B01AC) or 0.85 0.71 vitamin K antagonists (B01AA)a a Patients receiving at least two prescriptions from at least one of the medication groups

1. Voorham J, Denig P. Computerized extraction of information on the quality of diabetes from text in electronic medical patient records of general practitioners. J Am Med Inform Assoc 2007; 14(3): 349-354.

191

Curriculum Vitae 194 Curriculum Vitae

Dianna (Folgerdiena Maria) de Vries was born on February 26th 1989 in Delfzijl, The in 2007 she began her study Pharmacy at the University of Groningen. In 2010 she Netherlands. After finishing secondary school at the Ommelander College in Appingedam

Groningen. In 2011 she started working on her master thesis “The cost-effectiveness finished her bachelor Pharmacy and started the Pharmacy master at the University of of statins for the primary prevention of cardiovascular and cerebrovascular events in type 2 diabetes patients” at the Unit of Pharmacoepidemiology and Pharmacoeconomics in collaboration with the Department of Clinical Pharmacy and Pharmacology of the University Medical Center Groningen.

During her master she initiated her PhD project “Statin treatment in type 2 diabetes patients”, this project was an extension of her master thesis. This PhD project was also a collaboration of the Unit of Pharmacoepidemiology and Pharmacoeconomics of the University of Groningen and the Department of Clinical Pharmacy and Pharmacology

Pharmacy. of the University Medical Center Groningen. In 2014 Dianna finished her master in

As of November 2015 Dianna has been working as a post-doc researcher at the German Center for Neurodegenerative Diseases (DZNE) in Bonn on the Rhineland Study.

195

Dankwoord 198 Dankwoord

Allereerst wil ik mijn promotores, Prof. dr. E. Hak, Prof. dr. P. Denig, en Prof. dr. M.J. Postma, bedanken voor de mogelijkheid om mijn masterproject uit te breiden naar dit PhD project. Jullie variërende vakgebieden, gedachtes en meningen waren voor mij een uitdaging en hebben mij ontzettend veel geleerd.

Prof. dr. E. Hak, Beste Eelko, ik wil jou bedanken voor je inzet om dit nog inofficiële PharmD/PhD project mogelijk te maken. Ik heb onze samenwerking als erg prettig ideeën in te brengen. ervaren en vond het fijn dat jij mij altijd de mogelijkheid hebt gegeven om mijn eigen

Prof. dr. P. Denig, Beste Petra, in het begin van onze samenwerking, tijdens mijn masterproject, werd ik altijd een beetje nerveus van jouw e-mails. Nu ben ik blij met jouw aanwijzingen en kritiek destijds, het heeft mij erg veel geleerd en de kwaliteit van mijn werk aanzienlijk verhoogd. Jouw manier van werken en passie voor de wetenschap zijn een voorbeeld voor mij.

Prof. dr. M.J. Postma, Beste Maarten, bedankt voor de hulp die jij hebt geboden bij dit PhD project. Jouw relaxte kijk op zaken was erg prettig.

Dr. J. Voorham, Beste Jaco, ik wil jou bedanken voor jouw input bij de studies die ik met de GIANTT data heb uitgevoerd. Jouw kennis van de databank heeft mij enorm geholpen. Daarnaast wil ik je bedanken voor de tijd en het geduld die jij hebt gehad om mij te leren werken met Stata.

Dr. T. Dekker, Beste Thijs, bedankt voor jouw hulp bij de discrete choice studie, jouw input heeft het artikel naar een hoger niveau getild.

Bij deze wil ik overige co-auteurs bedanken voor hun input.

De beoordelingscommissie, bestaande uit Prof. dr. E. Buskens, Prof. dr. G.E.H.M. Rutten en Prof. dr. O.H. Klungel wil ik bedanken voor het beoordelen van dit proefschrift.

Mijn kamergenoten, Pieter, Bauke, Mishgina, Tomoko, Bianca, Koen, en Maarten, wil ik in het bijzonder bedanken voor de leuke tijd die ik gehad heb. Ik heb veel verschillende werkplekken gehad de afgelopen jaren en ben altijd met plezier naar mijn werk gegaan.

mijn vragen en voor advies, bedankt daarvoor. Beste Sieta, het was erg fijn om met jou samen te werken. Ik kon altijd bij jou terecht met

199 Dankwoord

Andere collega’s wil ik bedanken voor de leuke tijd die ik gehad heb op beide afdelingen.

ik denk dat jullie voor Kirsten en mij meer zijn dan een oom en tante. Wij hebben Beste oom Remmer en tante Birgit, officieel zijn jullie onze oudoom en oudtante maar het vroeger altijd erg naar ons zin gehad bij jullie. Misschien komt de Duitse invloed van toen nu goed van pas, ik weet in ieder geval wie de Kastelruther Spatzen zijn. J Ik wil jullie bedanken voor jullie interesse en steun. Tante Birgit, ik vind het erg leuk dat jij mijn paranimf wil zijn.

Helaas kan niet iedereen dit meemaken. Opa Hoppes en opa de Vries, ik weet dat jullie ontzettend trots zouden zijn. Op momenten als deze zegt mama vaak, de opa’s geven helaas nooit gekend, ik vind het mooi dat er Folgerdiena in dit proefschrift staat. elkaar een high-five daarboven, ik weet zeker dat dat nu zo is. Mama’s mama heb ik

Beste oma Hoppes en oma de Vries, ondanks dat het voor jullie denk ik lastig te begrijpen is wat promoveren inhoudt hebben jullie altijd interesse getoond, bedankt daarvoor.

Overige familie en vrienden wil ik bedanken voor hun belangstelling.

En dan de belangrijkste personen voor mij, de leden van de familie app 2.0, wat ben ik trots op jullie!

Lieve Kirsten, van kleins af aan zijn wij, hoe verschillend we ook zijn, altijd elkaars beste maatje geweest. Bij jouw HBO-V diploma uitreiking twee jaar geleden had ik tranen in mijn ogen, zo trots was ik toen op jou! De weg naar jouw diploma is onder andere door je gezondheid niet altijd gemakkelijk geweest. Kirs, ik kan me geen betere zus wensen, wij staan altijd voor elkaar klaar.

Lieve papa en mama, jullie hebben Kirsten en mij altijd het goede voorbeeld gegeven, en dat is denk ik het halve werk. Als ik terugkijk op de afgelopen jaren hebben jullie het soms behoorlijk te verduren gehad met ons. Jullie steunen ons in onze keuzes, en als deze achteraf niet de juiste blijken te zijn, helpen jullie de puinhoop op te ruimen. Bedankt dat jullie altijd voor ons klaar staan, ik hou van jullie!

200 Dankwoord

Lieve Adams, schat, de afgelopen vijf jaren waren soms heftig en zouden in een filmscript is over now, and I can see the sunshine’, dat nummer zou op ons kunnen slaan. Verhuizen niet misstaan. Gelukkig is dit een film met een goed einde. Zoals jij laatst zei, ‘The storm naar Groningen en alles opnieuw opbouwen is niet eenvoudig geweest, maar ik ben blij dat je deze stap voor mij hebt willen nemen. Ik ben trots op jou als persoon. Ik hoop dat de toekomst ons veel goeds zal brengen.

Groningen, oktober 2015

Dianna de Vries

201

SHARE publications SHARE publications

RESEARCH INSTITUTE SHARE

This thesis is published within the Research Institute SHARE (Science in Healthy Ageing and

Further information regarding the institute and its research can be obtained from our internetsite: healthcaRE) of the University Medical Center Groningen / University of Groningen. . http://www.share.umcg.nl/ More recent theses can be found in the list below. ((co-) supervisors are between brackets)

Otten E Introducing eHealth and other innovative options into clinical genetic patient care in view of

(prof I van Langen, prof AV Ranchor, dr E Birnie) increase efficiency and maintenance of quality of care; patients’ and providers’ perspectives

Pouwels K Self-controlled designs to control confounding (prof E Hak)

Voerman AE Living with prostate cancer; psychosocial problems, supportive care needs and social support groups (prof R Sanderman, prof M Hagedoorn, dr APh Visser)

Janse M The art of adjustment (prof AV Ranchor, prof MAG Sprangers, dr J Fleer)

Nigatu YT Obesity and depression; an intertwined public health challenge (prof U Bültmann, prof SA Reijneveld)

Aris-Meijer JL Stormy clouds in seventh heaven; a study on anxiety and depression around childbirth (prof CLH Bockting, prof RP Stolk, dr H Burger)

Kluitenberg B The NLstart2run study; running related injuries in novice runners (prof RL Diercks, dr H van der Worp, dr M van Middelkoop)

Benjaminse A Motor learning in ACL injury prevention (Prof E Otten, prof KAPM Lemmink, prof RL Diecks)

Potijk MR Moderate prematurity, socioeconomic status, and neurodevelopment in early childhood; a life course perspective (prof SA Reijneveld, prof AF Bos, dr AF de Winter)

204 SHARE publications

Beyer AR The human dimension in the assessment of medicines; perception, preferences, and decision making in the European regulatory environment (prof JL Hillege, prof PA de Graeff, prof B Fasolo)

Alferink M Psychological factors related to Buruli ulcer and tuberculosis in Sub-saharan Africa (prof AV Ranchor, prof TS van der Werf, dr Y Stienstra)

Bennik EC Every dark cloud has a colored lining; the relation between positive and negative affect and reactivity to positive and negative events (prof AJ Oldehinkel, prof J Ormel, dr E Nederhof, dr JACJ Bastiaansen)

Stallinga HA

Disability and Health (ICF) Human functioning in health care; application of the International Classification of Functioning, (prof RF Roodbol, prof PU Dijkstra, dr G Jansen)

Papageorgiou A How is depression valued? (prof AV Ranchor, prof E Buskens, dr KM Vermeulen, dr MJ Schroevers)

Spijkers W Parenting and child psychosocial problems; effectiveness of parenting support in preventive child healthcare (prof SA Reijneveld, dr DEMC Jansen)

Ark M van Patellar tendinopathy; physical therapy and injection treatments (prof RL Diercks, prof JL Cook)

Boven JFM van Enhancing adherence in patients with COPD: targets, interventions and cost-effectiveness (prof MJ Postma, prof T van der Molen)

Gerbers JG Computer assisted surgery in orthopedaedic oncology; indications, applications and surgical

(prof SK Bulstra, dr PC Jutte, dr M Stevens) workflow

Niet A van der Physical activity and cognition in children (prof C Visscher, dr E Hartman, dr J Smith)

For more 2015 and earlier theses visit our website

205