NPC Archive Item: Observational study of hypoglycaemic drugs has limitations

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo

15th January 2010

Compared with metformin monotherapy, the fully adjusted analysis in this UK cohort study found no increased risk of heart failure or myocardial infarction with pioglitazone▼ or rosiglitazone, and a lower risk of all-cause mortality with pioglitazone, It also found an increased risk of all-cause mortality with all sulfonylureas and an increased risk of heart failure with some. However, questions over the analysis limit the confidence that can be placed in these findings, many of which are not supported by or are in contrast to the results of randomised controlled trials.

Level of evidence:
Level 2 (limited quality patient-oriented evidence) according to the SORT criteria.

Health professionals should continue to follow NICE guidance on the management of type 2 diabetes and use metformin as the mainstay of glucose lowering treatment for most people. Other drugs, such as sulfonylureas and glitazones, may be considered to control blood glucose further. However, in accordance with NICE guidance, individual targets for HbA1c should be agreed with each patient. These could be greater than the 6.5% (48mmol/mol) target set for people with type 2 diabetes in general, and highly intensive management to levels of less than 6.5% (48mmol/mol) should be avoided. In accordance with principles of good prescribing, agreed targets and treatment should take into account patient preferences, the balance of likely benefits and harms, and the burden of extra medication.

Health professionals considering prescribing a glitazone should continue to follow MHRA advice published in October 2007 and December 2007. Neither pioglitazone nor rosiglitazone should be used in people with heart failure or a history of heart failure. Rosiglitazone should be used in patients with previous or current ischaemic heart disease only after a careful evaluation of the individual patient’s risk.

What is the background to this?
As we have concluded previously, it remains uncertain whether intensive glucose control (the addition of hypoglycaemic drugs to reduce HbA1c to levels significantly below that of standard treatments) offers any significant benefit in addition to that achievable by successful implementation of other interventions to reduce cardiovascular risk (including smoking cessation, exercise, losing weight, controlling blood pressure, taking statins, etc). Moreover, the MHRA has advised that pioglitazone and rosiglitazone can cause fluid retention, which may exacerbate or precipitate heart failure. There is also evidence that rosiglitazone might be associated with a small increased risk of cardiac ischaemia, particularly in combination with insulin. Questions have also been raised about the cardiovascular safety of sulfonylurea drugs with conflicting data available, much of it observational.

This cohort study looked at these questions using the very large UK general practice research database (GPRD).  The authors compared the rates of first occurrence of myocardial infarction (MI), congestive heart failure, and all-cause mortality among people prescribed glitazones, sulfonylureas, metformin and other hypoglycaemic drugs. The data for this study came from more than 90,000 people aged 35 to 90 years with diabetes, between 1st January 1990 and 31st December 2005.

What does this study claim?
The authors completed a number of analyses, adjusted just for sex and duration of diabetes (model 1); or additionally for previous complications from diabetes, previous peripheral arterial disease, previous cardiovascular disease and co-prescribed drugs (model 2); or with these adjustments and also adjusted for body mass index, cholesterol concentration, systolic blood pressure, HbA1c, creatinine concentration, albumin concentration and smoking status (model 3). Importantly, periods when patients were receiving insulin were not included in the analyses.

Sulfonylurea drugs
Sulfonylureas were grouped as first generation (e.g. chlorpropamide and tolbutamide) or second generation (e.g. glibenclamide and gliclazide). In the fully adjusted model (model 3) for the whole period 1990 to 2005 and compared to metformin monotherapy, first and second generation sulfonylureas did not show a statistically significant different risk of first episode of MI. First generation sulfonylureas also did not show a statistically significantly different risk of congestive heart failure compared to metformin monotherapy, but second generation sulfonylureas showed an increased risk (hazard ratio [HR] 1.18, 95% confidence interval [CI]1.04 to 1.34, P=0.011). Both groups of sulfonylureas were associated with a statistically significant increased risk of all-cause mortality (HR 1.37, 95%CI 1.11 to 1.71, P=0.0003 and 1.24 95%CI 1.14 to 1.35, P<0.001 respectively). In the model 2 analysis, a statistically significant increased risk of MI and heart failure, as well as all-cause mortality, was seen with first and second generation sulfonylureas.

In the fully adjusted model (model 3) for the whole period 1990 to 2005, and compared to metformin monotherapy, there was no observed statistically significant difference in risk of first of episode of MI associated with rosiglitazone alone, rosiglitazone in combination, or pioglitazone analysed as a composite of monotherapy and combination use. This was also the case with respect to the first episode of congestive heart failure. Rosiglitazone alone, or in combination, was not associated with a statistically significantly different risk of all-cause mortality. However, pioglitazone (all use) was associated with a lower risk of all-cause mortality (HR 0.69, 95%CI 0.49 to 0.98, P=0.024).

Glitazones were introduced only in 2000.  Analysis of the data for the period from 2000 to 2005 only is not published in the paper but as an online data supplement. Only the model 2 analysis is published (i.e. with adjustment for a number of variables, but importantly not smoking status). This found similar results for the glitazones to those above except that a statistically significant increased risk of congestive heart failure compared to metformin monotherapy was seen with rosiglitazone in combination therapy (HR 1.33, 95%CI 1.10 to 1.60, P=0.003).

So what?
These data are interesting, but should not change practice. There are a number of limitations of the analysis, which are discussed in the rapid responses. Observational data are prone to confounding. Unlike a randomised controlled trial (RCT), in ‘real life’ (on which observational studies are based), treatments are chosen, changed, or actively not chosen in the light of individual patients’ risk factors, preferences and tolerability or response to other drugs. Thus observed differences in outcomes may well be due to differences among the patients, not only the different treatments. Observational studies attempt to adjust for these differences by statistical modelling, but this requires certain assumptions and of course can take into account only those factors that have been recognised.

In this study, the more differences (covariates) that were included in the modelling, the fewer patients with a full data set who were available, and therefore the fewer the number of events. This limits the power of the analysis to detect any differences that might really exist, at conventional levels of statistical significance. The number of events in the population available for model 3 analysis was small. This led to wide confidence intervals for many outcomes and so although point estimates suggested a difference in risk these did not reach conventional levels of statistical significance. When fewer covariates were included more data were available for analysis but important possible confounding factors – notably smoking status – were excluded. Smoking history is likely to be one of the most important risk factors for cardiovascular events in people with diabetes, so the omission of this from the model 2 analysis limits the confidence one can place in the conclusions from that analysis.

Results of observational studies like this one should be compared with those of large, well conducted RCTs. Two such RCTs of pioglitazone (PROACTIVE) and rosiglitazone (RECORD) both found an increased risk of heart failure with the glitazone compared to controls. In addition, a meta-analysis of seven RCTs of pioglitazone or rosiglitazone found these drugs increased the risk of congestive heart failure compared with controls across a wide range of cardiac risk, with no heterogeneity between studies (relative risk 1.72, 95%CI 1.21 to 2.42, P=0.002).

With regard to myocardial ischaemia, the December 2007 issue of Drug Safety Update stated that a meta-analysis of data from clinical trials showed that the overall incidence of cardiac ischaemic events was higher for rosiglitazone than for comparators (1.99% vs.1.51%, HR 1.31, 95%CI 1.01 to1.70). However, there was no increase in overall mortality. The risk of cardiac ischaemic events seems to be particularly marked when rosiglitazone is combined with insulin (which was excluded from the observational study discussed above), or in patients with a previous or current ischaemic heart disease. Analysis of pioglitazone trials did not suggest an increased risk of cardiac ischaemia. A meta-analysis of 19 RCTs of pioglitazone found a statistically significantly lower risk of the combined outcome of death, MI or stroke (HR 0.82, 95% CI 0.72 to 0.94, P=0.005) for pioglitazone compared with controls, but not for the individual outcome components. However, comparative data do not provide good evidence of a difference between rosiglitazone and pioglitazone for risk of cardiac ischaemia events.

This observational study does not provide any grounds for moving away from metformin as the first choice drug for most patients, as recommended by NICE, or for routinely using glitazones in preference to sulfonylureas.

Study details
Tzoulaki I et al. Risk of cardiovascular disease and all cause mortality among patients with type 2 diabetes prescribed oral antidiabetes drugs: retrospective cohort study using UK general practice research database. BMJ 2009;339:b4731

Design: retrospective cohort study

Patients: 91,521 patients aged 35-90 years with an episode of care between 1 January 1990 and 31 December 2005 and a diagnostic (Read) code associated with a clinical or referral event for diabetes, extracted from the GPRD

Intervention, comparison and outcomes: Cox regression analysis comparing rate of first occurrence of incident MI, congestive heart failure, and all-cause mortality associated with use of rosiglitazone monotherapy, rosiglitazone combination therapy (with other antidiabetes drugs), pioglitazone monotherapy and pioglitazone combination therapy (analysed together), monotherapy with first generation sulphonylureas (acetohexamide, chlorpropamide, tolbutamide, or tolazamide), monotherapy with second generation sulphonylureas (glipizide, gliquidone, glimepiride, glibenclamide, or gliclazide), other oral antidiabetes drugs (for example, acarbose, nateglinide, repaglinide), and combination therapies excluding glitazones and insulin; compared with metformin monotherapy.

Results: this study generated a large quantity of data; key results are summarised above.

Sponsorship: “No funding was obtained for this study. This study is based on data from the full feature general practice research database funded through the Medical Research Council’s license agreement with the UK Medicines and Healthcare Products Regulatory Agency. The interpretation and conclusions contained in this study are those of the authors alone.”

More information can be found drug therapies and treatment approaches and priorities in type 2 diabetes can on the type 2 diabetes section of NPC.

Please comment on this blog in the NPC discussion rooms, or using our feedback form.

Make sure you are signed up to NPC Email updates — the free email alerting system that keeps you up to date with the NPC news and outputs relevant to you