In short Read article

New authorisation policy for diabetes drugs

The rosiglitazone file


The oral blood glucose-lowering drug rosiglitazone has had a turbulent history. After its authorisation, it was found that the risk of myocardial infarction was larger with rosiglitazone than with other oral blood glucose lowering drugs. This led to changes in the authorisation policy for these drugs. Since 2020, the American FDA has demanded that pre-authorisation studies of blood glucose lowering drugs should concentrate more on a representative and sufficiently large sample of potential users, making it possible to determine both the effectiveness and the adverse effects. The European regulatory authority has also adjusted its authorisation policy on blood glucose lowering drugs in 2012. It is currently working on an amendment which says that the effectiveness of blood glucose lowering drugs must be proven in randomised double-blind placebo-controlled trials and randomised double-blind non-inferiority trials, in which the new drug is compared with an existing alternative. Many questions remain, however, concerning, for instance, the methodology of the trials. Nor is it clear to what extent the regulatory authorities will be extending this policy to other drug categories.

What is Ge-Bu’s opinion?
  • Although rosiglitazone would probably not have been authorised under the new FDA requirements, it remains questionable if the new policy will sufficiently prevent problems in the future.
  • In particular, the absence of requirements regarding hard endpoints, the unclear description of the methodology required for research into adverse effects and the absence of post-authorisation requirements raises questions about the effectiveness of the measures taken.
  • There should be a requirement for all new drugs that their effectiveness must, where possible, be proven by randomised controlled trials and controlled observational studies to detect adverse effects.
  • Superiority regarding both effectiveness and adverse effects, compared to existing drugs, should where possible be shown on hard endpoints.

The American regulatory authority FDA and the European Medicines Agency (EMA) adjusted their policy on the authorisation of new blood glucose lowering drugs in 2020 and 2012, respectively. One of the reasons to adjust this policy was the problems surrounding the diabetes drug rosiglitazone. After authorisation by both the FDA and the EMA, research showed that rosiglitazone might possibly not reduce the number of cardiovascular complications, but actually increase them.1 The new policy involves stricter requirements on the research into the effectiveness and adverse effects of these drugs. 

Despite these recent adjustments, the question remains whether this has solved all problems. In addition, it is unclear whether this policy change concerns only the blood glucose lowering drugs, and whether similar problems could still arise with other new drugs. 

An article published in the New England Journal of Medicine in 2020 discusses the nature of the change in the FDA policy, using the authorisation of rosiglitazone as an example.2 The present Ge-Bu article summarises the main points and investigates the EMA’s policy on pre-authorisation research.

The firm which developed rosiglitazone established two objectives for the new blood glucose lowering drug. The first was that rosiglitazone should have a greater blood glucose lowering effectiveness than the oral blood glucose lowering drugs available to patients with type 2 diabetes mellitus at the time (in the late 1990s). The second was that the new drug should reduce the increased risk of arterial diseases and events associated with type 2 diabetes, and specifically the risk of myocardial infarction.

Neither of these objectives was met. No clinically relevant difference regarding the surrogate endpoint of HbA1c was found between rosiglitazone and other oral blood glucose lowering drugs. Later, research found that the risk of myocardial infarction while using rosiglitazone was possibly higher than that associated with other blood glucose lowering drugs.

Both the FDA and EMA authorised rosiglitazone as a monotherapy for type 2 diabetes, in 1999 and 2000, respectively. No added value compared to the other drugs on the market at the time had been established, but that was apparently no precondition for authorisation.2

The original policy 

Before 2008, the FDA  assessed the safety of new blood glucose lowering drugs on the basis of studies comprising a few thousand user years, studies which often excluded patients with cardiovascular diseases. A 2007 meta-analysis proved a turning point. This meta-analysis included 42 small randomised placebo-controlled and industry-sponsored studies, and concluded that rosiglitazone did not reduce, but actually increased the risk of myocardial infarction, even after a short period of use.1 The FDA then issued a warning. The EMA warned of a possibly elevated risk of myocardial infarction in patients with existing coronary artery disease. Neither of the two authorities saw a reason to withdraw the authorisation of rosiglitazone.

Reason for change

Meta-analysis of cardiovascular problems

The above-mentioned systematic review and meta-analysis of 42 randomised studies attempted to answer the question whether rosiglitazone prevents cardiovascular diseases in patients with type 2 diabetes. The authors of the meta-analysis included studies with a follow-up of more than 24 weeks. Ten studies compared rosiglitazone with placebo, 25 with other oral blood glucose lowering drugs, 5 with insulin, and 2 with ‘usual care’. The quality of the studies and the risk of bias were not assessed in the meta-analysis. The primary outcome measures were the number of myocardial infarctions and the mortality from cardiovascular causes. The meta-analysis found an odds ratio for the number of myocardial infarctions of 1.43 (95% confidence interval (CI) 1.03 to 1.98) for rosiglitazone compared to the combined control groups. The odds ratio for the number of deaths from cardiovascular causes was found to be 1.64 (0.98 to 2.74).1

Problems identified in post-marketing research

An editorial accompanying the published meta-analysis commented that the FDA’s reporting system is unsuitable for identifying rare adverse effects of rosiglitazone, like myocardial infarctions.3 The FDA requests post-marketing studies (phase 4 studies), but it is the manufacturer who decides on the study design. As a result, drugs are often compared with an inferior treatment, with a comparable drug but using an inferior dosage or with an uncommon drug, or results are influenced by including or excluding specific patients, or bias arises due to the methodology used. 

In addition, it was found that in the 1998–2003 period, only 25% of the requested phase-4 studies had been completed, and that by 2006, the results of 899 phase-4 studies had still not been made available. The authors of the editorial therefore asked for large-scale randomised studies specifically focusing on the potential users. Well-designed post-marketing research is even more important when a drug is authorised on the basis of surrogate endpoints. The actual effectiveness on hard endpoints, for instance in the case of blood glucose lowering drugs, will then only become apparent after the drug has been on the market for a longer period of time. In addition, a drug may have adverse effects that only become manifest after prolonged use.3

Results of the RECORD trial

The RECORD trial, set up by the manufacturer, was published in 2009.4 This study, which was initiated after the FDA authorisation in 1999, and focused on the relationship between rosiglitazone and myocardial infarction, found no difference in risk between rosiglitazone and metformin/sulfonylurea derivatives.

This long-lasting randomised open-label non-inferiority trial investigated the cardiovascular adverse effects of rosiglitazone.4 In particular, it assessed the number of myocardial infarctions. Patients with type 2 diabetes who used metformin or sulfonylurea derivatives as monotherapy were included and randomised to a rosiglitazone group and a metformin-sulfonyl-urea group. Rosiglitazone was combined with metformin or a sulfonylurea derivative. The patients had a mean baseline HbA1c value of 63 mmol/mol (7.9%). The outcome measure was the number of patients who were hospitalised for cardiovascular disease or who died. After an average follow-up of 5.5 years, 64 myocardial infarctions had occurred in the rosiglitazone group (n=2220) against 56 in the metformin-sulfonylurea group (n=2227), with a hazard ratio of 1.14 (95% CI 0.80 to 1.63). In the rosiglitazone group, 154 of the 2220 patents died of a myocardial infarction or a stroke, compared to 165 of the 2227 patients in the metformin-sulfonylurea group (HR 0.93 [0.74 to 1.15]). The manufacturer’s initial objective, that rosiglitazone should reduce the risk of cardiovascular disease compared to existing drugs, was thus not achieved. 

Measures taken by the regulatory authorities after the RECORD trial

The FDA has not yet made a decision on the basis of the RECORD trial, but has commissioned a re-evaluation and reanalysis of the results by an American agency.2 The EMA has decided to suspend the authorisation of rosiglitazone because of the previously established increased risk of myocardial infarction and because of methodological problems in the RECORD trial.

Interim adjustments to the FDA’s authorisation policy (2008 to 2020)

In 2008, partly as a result of the rosiglitazone issue (see the timeline in the Background Information), the FDA tightened its requirements for studies into potential new blood glucose lowering drugs. The FDA demanded randomised studies with a larger number of participants, and with greater diversity in terms of comorbidity, age and sex. In addition, the number of user years would have to be larger, for instance by having a longer follow-up period. These studies would have to include a percentage of patients with cardiovascular problems and kidney diseases that corresponds with the percentage encountered in practice. Furthermore, sufficient numbers of hard endpoints, such as myocardial infarctions, would have to have occurred in the study sample to allow an increase in relative risk to be calculated. The upper limit of the 95% confidence interval of the relative risk of an important adverse effect had to be lower than 1.3, in order to be able to exclude a risk increase of at least 30% compared to the control treatment. As the FDA authors reported in 2020, none of the new blood glucose lowering drugs had shown an increased risk of cardiovascular incidents since the introduction of this new requirement. A number of studies even found a reduced risk.2

FDA imposes restrictions on supplying rosiglitazone

In 2011, the FDA warned of the cardiovascular risks of rosiglitazone, and imposed restrictions which meant that it could only be supplied by specified pharmacies. In the same year, a review summarised 15 observational studies: rosiglitazone used by 360,688 patients with type 2 diabetes was compared with pioglitazone (218,815 patients), using various cardiovascular endpoints as outcomes, including myocardial infarction. This review found an odds ratio for the endpoint of myocardial infarction of 1.16 (95% CI 1.07 to 1.24) to the disadvantage of rosiglitazone, a significant difference.5 

Re-evaluation of the RECORD trial

The results of the re-evaluation of the RECORD trial commissioned by the FDA were published in 2013.6,7 The FDA’s request was prompted by the methodological limitations of the study, including the low number of myocardial infarctions that had occurred during the study period.8 The results were checked for events that had not been included, and the original data were reassessed. Using these new criteria, 68 versus 60 myocardial infarctions were found (HR 1,13 [95% CI 0.80 to 1.59]). Applying the FDA definitions to the outcome measures resulted in a hazard ratio of 1.14 (0.80 to 1.63). Although the re-evaluation thus found higher hazard ratios, the differences remained non-significant.7

The FDA’s reaction to the re-evaluation

Based on the results of the re-evaluation, the FDA decided to withdraw its previous restriction which meant that rosiglitazone could only be supplied by specified pharmacies. From then on, rosiglitazone could be supplied by any pharmacy without restrictions. The European authorisation of rosiglitazone remained suspended.

Systematic review based on individual patient-level data

In 2020, a systematic review was published, based on individual patient-level data (IPD) from randomised studies (including the RECORD trial). This review found a statistically significant difference in the number of myocardial infarctions and cases of heart failure, but not in the number of deaths, compared to a control group, to the disadvantage of rosiglitazone (see Background Information).9

New policy in force in the USA from 2020 

In light of the revised results of the RECORD trial, the FDA discussed new draft guidelines with representatives of various professions in October 2018. The development of new blood glucose lowering drugs would have to meet the newly established requirements, in order to ensure their safety in the broadest possible sense.10 In addition to the importance of cardiovascular safety, numerous other issues were raised, including comorbidity and diabetic complications. This necessarily implied more comprehensive checks on the safety profile of new blood glucose lowering drugs, and greater emphasis was put on sufficient amounts of data from long-term users.11

New requirements for blood glucose lowering drugs

It was decided that the results of phase 3 studies would have to be based on at least 4000 patient years. In addition, 1500 patients would have to have used the drug for at least 1 year, and 500 for at least 2 years.

Furthermore, the study groups would have to be sufficiently representative in terms of cardiovascular history, chronic kidney diseases and age, for the purpose of evaluating the safety for future users. It was further decided that at least 500 patients with advanced chronic kidney disease, at least 600 patients with cardiovascular disorders like myocardial infarction, stroke or peripheral vascular disease, and at least 600 patients older than 65 years would have to be included. In all, at least 1200 different persons belonging to one of the above categories would have to be included in the study. It is unclear what exactly is the rationale behind the numbers required by the FDA.

In some cases, where concerns about a particular rare unexpected adverse effect have arisen during the pre-authorisation phase, the FDA stated that this adverse effect also had to be sufficiently common in the sample to be assessed to allow a reliable  calculation of a potential risk. In addition, the FDA can also impose post-authorisation measures on the manufacturer.

In the FDA’s view, these measures, which it considers to be ‘holistic’, allow the safety of blood glucose lowering drugs to be more fully ensured without forming too high a barrier to the development of these drugs.2,10

In September 2010, the EMA suspended the authorisation of de rosiglitazone in view of the increased cardiovascular risk.12 Since then, the drug has no longer been on the market.

New requirements for diabetes drugs

In 2012, the EMA published a guideline focussing specifically on the authorisation of new drugs to treat diabetes.13 This guideline contains the requirements that would have to be met by research into the effectiveness and adverse effects of blood glucose lowering drugs.

Effectiveness of blood glucose lowering drugs

As of 2012, the effectiveness of drugs for diabetes has to be tested in randomised studies with the following characteristics:

  • Division of the sample into parallel groups (each of the study groups receiving only one of the treatments to be investigated);
  • if possible a double-blind study design;
  • a controlled study design with a placebo and with an alternative drug;
  • superiority of the new drug over placebo to be established in at least one monotherapy study, for instance in a phase 2 study using HbA1c as its primary endpoint; 
  • superiority of the new drug over a placebo to be established in a study in which both groups are also given usual care;
  • non-inferiority of the new drug to be shown compared to an active and well-known alternative drug;
  • the duration of one of the studies to be at least 12 months, the others having a duration of at least 6 months;
  • the studies preferably to take account of special patient groups in the various European states, particularly young and older people.

The EMA is currently working on a new version of the guideline, by including the requirement that all three of the above-mentioned superiority and non-inferiority studies must be submitted for authorisation.14 The use of HBA1c as a primary endpoint is still required, as are cardiovascular risks as a secondary endpoint, despite the fact that independent research found that lowering the HbA1c in patients with cardiovascular problems did not reduce the number of cardiovascular events and even increased the mortality risk.15,16

Adverse effects 

For the purpose of improved detection of adverse effects, the EMA requests that, starting from phase 2 studies, those designing the studies should determine how known and serious adverse effects can be detected and addressed. This particularly concerns the size and duration of the study. It remains unclear whether these have to be controlled studies. If uncommon side effects are detected in database studies (especially relating to cardiovascular issues), new study proposals for this have to be submitted, which also mention the duration of the study. This duration should be at least 18 to 24 months.13

Evidence of clinically relevant effectiveness

The precise implications of these rules for practice are unclear. Will the stricter requirements regarding control groups, patients to be included and duration of studies be sufficient to prove the clinically relevant effectiveness of a drug? As regards rosiglitazone, the new regulations would have meant that the manufacturer’s primary objectives would have had to be proven beforehand. In the original procedure, the objective of lowering HbA1c further than existing blood glucose drugs was not achieved, which would probably have meant that the drug would not have been authorised. Proving clinically relevant effectiveness on cardiovascular outcomes before authorisation would probably not have been feasible. This raises the question whether the EMA and the FDA should also have imposed stricter and more compulsory post-authorisation requirements to allow the lack of clinically relevant effectiveness to be detected sooner.

The EMA has commented on superiority studies comparing with placebo, but not with a commonly used alternative drug. Both the FDA and the EMA persist in the use of HBA1c as the primary endpoint, whereas it is especially in the debate about cardiovascular risks in patients with type 2 diabetes that it is unclear whether this is a suitable endpoint.16,15 Ge-Bu has previously underlined the value of superiority studies.17

Identifying adverse effects

Neither the FDA nor the EMA explicitly indicates how adverse effects are to be investigated. The word ‘controlled’ is not mentioned anywhere, but is essential. After the authorisation, well-controlled and independent observational studies should be conducted to identify adverse effects that are less common or only occur after prolonged use.18 These studies must obviously be of sufficient size and duration and be based on a user population with representative characteristics. As was previously concluded in the Ge-Bu article ‘Level of evidence: randomised or observational studies?’, controlled observational studies but also randomised studies have their methodological problems. Despite these problems, however, both are needed to obtain a clear picture of the effectiveness and adverse effects of a drug.19

Thiazolidinedione derivatives

Rosiglitazone belongs to the thiazolidinedione derivatives. This group of drugs came on the market in the early years of this century, with the claim that they would have a preventive effect on cardiovascular complications of diabetes. The group is also known as PPAR-γ agonists (Peroxisome Proliferator-Activated Receptor-γ agonists). The PPA receptors are located in the cell’s nucleus, and their activation results in the expression of proteins that play a role in the glucose and lipid metabolisms. More free fatty acids are taken up by the adipose tissue, lipogenesis is boosted and the transport of glucose increases. The insulin sensitivity in the liver and the adipose and muscle tissues increases. The drug is only effective in the presence of sufficient insulin.20

Timeline of events since rosiglitazone was authorised 


Event 21,9


The FDA approves rosiglitazone as a first-line monotherapy or combined with metformin; the EMA rejects rosiglitazone as a first-line monotherapy.


The EMA approves rosiglitazone as a second-line treatment.


The FDA rejects the rosiglitazone and insulin combination therapy, as research has found too little improvement on surrogate endpoints relative to the increased risk of cardiovascular adverse effects. The manufacturer starts a randomised trial into the cardiovascular outcomes of rosiglitazone, the RECORD trial.


The FDA approves the rosiglitazone and insulin combination therapy.


Ge-Bu concludes that rosiglitazone (and pioglitazone) appear mostly suitable for overweight patients, but that they are not the drugs of first choice for monotherapy. As regards their blood glucose lowering effect, they are comparable to sulfonylurea derivatives and to metformin. No long-term data are available regarding effectiveness and adverse effects.22


A meta-analysis based on 42 studies by GlaxoSmithKline (GSK) suggests a 43% increased risk of myocardial infarction.1

The EMA warns against its use for patients with ischaemic heart disease, and removes the contraindication for the rosiglitazone and insulin combination therapy. The FDA advises against this combination therapy.

A Cochrane review of 18 randomised studies in which 3888 patients with type 2 diabetes had been randomised for rosiglitazone for at least 24 weeks showed no effect on the endpoints of morbidity and mortality. No clinically relevant difference regarding the surrogate endpoint of HbA1c was found between rosiglitazone and other oral blood glucose lowering drugs.11


The first results of the RECORD trial show no difference in cardiovascular risk between rosiglitazone and a combination of a sulfonylurea derivative and metformin.


The EMA suspends the authorisation of rosiglitazone, as it is of the opinion that the advantages no longer outweigh the disadvantages. As of 2010, rosiglitazone is no longer on the market in Europe. This suspension has so far not been reversed.


The FDA warns about the cardiovascular risks of rosiglitazone and imposes restrictions which mean that the drug can only be supplied by specified pharmacies.

A review study summarises the findings of 15 observational studies which compared rosiglitazone with pioglitazone, one of the outcomes being myocardial infarction. This review found a significant difference to the disadvantage of rosiglitazone.5


The findings of the RECORD trial have been reanalysed at the FDA’s request, and as a result the restrictions imposed by the FDA in 2011 are withdrawn again.6,7


The authors of a study based on individual patient-level data (IPD) conclude that rosiglitazone is associated with an elevated cardiovascular risk, especially of heart failure. IPD data can be used to generate hypotheses but are not conclusive in finding a causal relationship.23,9

Rosiglitazone is still available in the USA.


Results of a meta-analysis based on individual patient-level data 

Three datasets were analysed for myocardial infarction using various calculating methods (the Peto method, which takes small numbers into account , the ‘fixed-effects’ model  and the ‘random-effects’ model ). As regards the number of myocardial infarctions, only the dataset with individual patient data, analysed with the Peto method, yielded a statistically significant difference, to the disadvantage of rosiglitazone (Table 1). As regards the outcome measure of heart failure, a statistically significant difference was found for all datasets and with all analysis methods, OR 1.80 (95% CI 1.46 to 2.22) using the Peto method, combining IPD with the RECORD trial) (Table 2). No difference was found in the number of deaths.9

Table 1. Myocardial infarctions in the rosiglitazone group versus the control group


Peto method
(OR [95% CI])

Fixed-effects model (OR [95% CI])

Random-effects model (OR [95% CI])


1.30 (1.02 – 1.67)*

1.25 (0.99 – 1.60)

1.17 (0.92 – 1.51)


1.17 (0.99 – 1.38)

1.15 (0.98 – 1.36)

1.11 (0.94 – 1.32)

IPD + other**

1.19 (0.96 – 1.48)

1.13 (0.92 – 1.39)

1.09 (0.88 – 1.35)

IPD + RECORD + other

1.13 (0.97 – 1.32)

1.10 (0.95 – 1.28)

1.08 (0.92 – 1.26)

IPD: dataset with all individual patient data; RECORD: results of the RECORD trial; other: results of other studies; OR: Odds ratio; CI: confidence interval
* statistically significant difference; ** no individual patient data available

Table 2. Heart failure in the rosiglitazone group versus the control group


Peto method
(OR [95% CI])

Fixed-effects model (OR [95% CI])

Random-effects model (OR [95% CI])


1.66 (1.24 – 2.22)*

1.60 (1.20 – 2.14)*

1.54 (1.14 – 2.09)*


1.80 (1.46 – 2.22)*

1.78 (1.44 – 2.20)*

1.75 (1.41 – 2.18)*

IPD: dataset with all individual patient data; RECORD: results of the RECORD trial; OR: Odds ratio; BI: confidence interval
* statistically significant difference; ** no individual patient data available

Study details

meta-analysis by Nissen et al 20071
Design: systematic literature review and meta-analysis 
Primary endpoint(s): number of participants with a myocardial infarction, number of deaths from cardiovascular causes
Inclusion criteria: randomised control groups, study duration >24 weeks use of drugs, availability of data on number of myocardial infarctions or deaths from cardiovascular causes
Number of articles analysed: 42 met the inclusion criteria (no further specification)
Number of patients: 15,565 in the rosiglitazone group, 12,282 in the control group
Funding: not reported
Conflicts of interest: reported by one of the two authors
RECORD trial and re-evaluation, 2009 and 20134,6,7
Study name: Rosiglitazone evaluated for cardiovascular outcomes in oral
agent combination therapy for type 2 diabetes (RECORD).
Design: international multicentre randomised open-label non-inferiority study. 
Inclusion criteria: patients with type 2 diabetes (40–75 years) using a sulfonylurea derivative or metformin as monotherapy, with insufficiently controlled diabetes (HbA1c 53–75 mmol/mol [7.0–9.0%]). BMI>25 kg/m2
Main exclusion criteria: hospitalisation for serious cardiovascular events in previous 3 months, heart failure.
Intervention: rosiglitazone in combination with a sulfonylurea derivative or metformin, versus a sulfonylurea derivative combined with metformin (1:1)
Primary endpoint(s) and duration: time to first hospitalisation due to a cardiovascular event or death from cardiovascular causes. Study duration was 7 years and 9 months
Intended number of patients and power: based on a median study duration of 6 years, 99% power, 11% events a year in the active control group and 2% annual loss-to-follow-up, 4000 patients had to be included to prove non-inferiority
Randomisation: not reported
Blinding: treatment was not blinded
Population analysed: intention-to-treat
Number of patients included: 4447 patients, approx. 50% male, average age 58 years, baseline HbA1c 63 mmol/mol (7.9%)
Trial registration: NCT00379769
Funding: by GSK, the manufacturer of rosiglitazone
Conflicts of interest: 9 of the 9 authors
Study based on individual patient-level data 20209
Design: meta-analysis of individual patient-level data from various randomised trials
Primary endpoint(s): composite outcome measure was myocardial infarction, heart failure, death from cardiovascular causes and death from non-cardiovascular causes
Inclusion criteria: randomised trials comparing rosiglitazone with an active control or placebo among adults. Study duration >24 weeks. Composite outcome measure including myocardial infarction, heart failure, death from cardiovascular causes and death from non-cardiovascular causes 
Number of articles analysed: 33 studies using IPD, 103 studies for the outcome measure of myocardial infarction and 103 studies for the outcome measure of death from heart failure
Number of patients: 21,156 patients in the set with individual patient-level data, 23,683 patients in studies with the outcome measure of myocardial infarction, 22,772 patients in studies with the outcome measure of death from cardiovascular causes
Funding: Laura and John Arnold Foundation (Collaboration for Research Integrity and Transparency at Yale)
Conflicts of interest: 3 of the 10 authors

  1. Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med. 2007 Jun 14;356(24):2457-71. doi: 10.1056/NEJMoa072761. Erratum in: N Engl J Med. 2007 Jul 5;357(1):100.
  2. Chong WH, Yanoff LB, Andraca-Carrera E, Thanh Hai M. Assessing the Safety of Glucose-Lowering Drugs - A New Focus for the FDA. N Engl J Med. 2020 Sep 24;383(13):1199-1202. doi: 10.1056/NEJMp2004889.
  3. Psaty BM, Furberg CD. Rosiglitazone and cardiovascular risk. N Engl J Med. 2007 Jun 14;356(24):2522-4. doi: 10.1056/NEJMe078099.
  4. Home PD, Pocock SJ, Beck-Nielsen H, Curtis PS, Gomis R, Hanefeld M, et al, RECORD Study Team. Rosiglitazone evaluated for cardiovascular outcomes in oral agent combination therapy for type 2 diabetes (RECORD): a multicentre, randomised, open-label trial. Lancet. 2009 Jun 20;373(9681):2125-35. doi: 10.1016/S0140-6736(09)60953-3.
  5. Loke YK, Kwok CS, Singh S. Comparative cardiovascular effects of thiazolidinediones: systematic review and meta-analysis of observational studies. BMJ. 2011 Mar 17;342:d1309. doi: 10.1136/bmj.d1309.
  6. Lopes RD, Dickerson S, Hafley G, Burns S, Tourt-Uhlig S, White J, et al. Methodology of a reevaluation of cardiovascular outcomes in the RECORD trial: study design and conduct. Am Heart J. 2013 Aug;166(2):208-216.e28. doi: 10.1016/j.ahj.2013.05.005.
  7. Mahaffey KW, Hafley G, Dickerson S, Burns S, Tourt-Uhlig S, White J, et al. Results of a reevaluation of cardiovascular outcomes in the RECORD trial. Am Heart J. 2013 Aug;166(2):240-249.e1. doi: 10.1016/j.ahj.2013.05.004.
  8. Bijl D. Veiligheid van de thiazolidinedionderivaten pioglitazon en rosiglitazon. Gebu. 2007;41(10):105-112.
  9. Wallach JD, Wang K, Zhang AD, Cheng D, Grossetta Nardini HK, Lin H, et al. Updating insights into rosiglitazone and cardiovascular risk through shared data: individual patient and summary level meta-analyses. BMJ. 2020 Feb 5;368:l7078. doi: 10.1136/bmj.l7078. Erratum in: BMJ. 2021 Jun 2;373:n1302.
  10. Food and Drug Administration. Type 2 diabetes mellitus: evaluating the safety of new drugs for improving glycemic control; guidance for industry [Draft]. U.S. Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research (CDER). March 2020. Available from:
  11. Richter B, Bandeira-Echtler E, Bergerhoff K, Clar C, Ebrahim SH. Rosiglitazone for type 2 diabetes mellitus. Cochrane Database Syst Rev. 2007 Jul 18;2007(3):CD006063. doi: 10.1002/14651858.CD006063.pub2.
  12. European Medicines Agency. European Medicines Agency recommends suspension of Avandia, Avandamet and Avaglim. 23-09-2010. Available from:
  13. European Medicines Agency. Committee for Medicinal Products for Human Use. Guideline on clinical investigation of medicinal products in the treatment or prevention of diabetes mellitus. 14 May 2012. Available from: Accessed 2 June 2021.
  14. European Medicines Agency. Committee for Medicinal Products for Human Use. Guideline on clinical investigation of medicinal products in the treatment or prevention of diabetes mellitus [Draft]. 29 January 2018. Available from: Accessed 2 June 2021.
  15. ACCORD Study Group, Gerstein HC, Miller ME, Genuth S, Ismail-Beigi F, Buse JB, Goff DC Jr, et al. Long-term effects of intensive glucose lowering on cardiovascular outcomes. N Engl J Med. 2011 Mar 3;364(9):818-28. doi: 10.1056/NEJMoa1006524.
  16. Hiatt WR, Kaul S, Smith RJ. The cardiovascular safety of diabetes drugs--insights from the rosiglitazone experience. N Engl J Med. 2013 Oct 3;369(14):1285-7. doi: 10.1056/NEJMp1309610.
  17. Dekkers OM. Het non-inferioriteitsonderzoek. Gebu. 2015;49(3):27-34.
  18. 18. Lundh A, Lexchin J, Mintzes B, Schroll JB, Bero L. Industry sponsorship and research outcome. Cochrane Database Syst Rev. 2017 Feb 16;2(2):MR000033. doi: 10.1002/14651858.MR000033.pub3.
  19. Dekkers OM. Bewijskracht: gerandomiseerd of observationeel onderzoek? Gebu. 2020;54(1):1-8.
  20. Informatorium Medicamentorum. Den Haag: KNMP, 2021. Available from: Accessed 31 March 2021.
  21. Pouwels KB, van Grootheest K. The rosiglitazone decision process at FDA and EMA. What should we learn? Int J Risk Saf Med. 2012;24(2):73-80. doi: 10.3233/JRS-2012-0559.
  22. Tack CJ, Goudswaard AN, Smits P. Thiazolidinedionderivaten bij diabetes mellitus type 2. Gebu. 2005;39(10):107-115.
  23. Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data: rationale, conduct, and reporting. BMJ. 2010 Feb 5;340:c221. doi: 10.1136/bmj.c221.


  • Frans M. Helmerhorst, prof. dr
  • Anton J.F.A. Kerst, dr
  • Marielle A.E. Nieuwhof, pharmacist