Different Oral Antidiabetics Associated with Different Adverse Outcomes?

This observational cohort study used the UK's General Practice Research Database (GPRD) to investigate the risk for incident myocardial infarction (MI), congestive heart failure (CHF), and all-cause mortality associated with prescription oral antidiabetic agents. The GPRD comprises clinical and prescribing data from patient-based clinical records of about 5 million people.

The investigators obtained data on patients aged 35-90 years with an episode of care for diabetes between January 1, 1990, and December 31, 2005. An interval of oral drug treatment was used as the unit of observation; this was defined as the period from onset of a drug treatment to onset of the next drug treatment, until the patient was censored, or until the event of interest occurred. For example, 2 intervals would be contributed to a patient prescribed monotherapy with a sulfonylurea (SU) at entry to the cohort and then prescribed a combination of an SU and metformin and who remained on that combination until occurrence of MI.

Primary events were first occurrence of incident MI, CHF, and all-cause mortality. A series of Cox regression analyses adjusted for the results of sex and diabetes duration; then for the addition of complications and nondiabetes drugs; and then for the addition of body mass index, lipids, blood pressure, hemoglobin A1c, creatinine, albumin, and smoking status.

After exclusion of periods when patients received insulin and events during these periods, there were 2,843,007 intervals of oral antidiabetes treatments among 91,521 patients (mean age, 65 years). The median follow-up period was 24 days per interval; the mean follow-up per patient was 7.1 years. During the study period, there were 3,588 first events of MI; 6,900 first events of CHF; and 18,548 deaths.

Compared with metformin, SU monotherapy was initially associated with higher risk of first MI, but the relationship became nonsignificant after full multivariable adjustment. However, second-generation SUs were associated with an elevated risk of heart failure (hazard ratio [HR], 1.18; 95% confidence interval [CI], 1.04-1.34]) and with all-cause mortality (HR, 1.24; 95% CI, 1.14-1.35). Neither rosiglitazone nor pioglitazone was associated with a statistically significant increased risk of heart failure (HRs were 1.21 and 1.17, respectively). Pioglitazone was associated with reduced all-cause mortality compared with metformin (HR, 0.69; 95% CI, 0.49-0.98).

Head-to-head comparisons of the relative risks and benefits of the increasing number of antidiabetes drugs will almost certainly never be randomized, so we must turn to observational studies for answers. Observational studies are important in their own right because clinical trial results can be difficult to apply to real-world populations. However, results of observational studies can always be questioned because the reasons for which patients were presumably prescribed a drug may be associated with the study outcomes. This phenomenon is known as "confounding by indication." As in the current study, researchers do their best to eliminate the potential bias with statistical techniques, but one can never be certain that the effort was successful.

Although it was reassuring that there was no association between the drugs studied and MI, the increased risk of heart failure associated with SUs cannot be ignored, especially because previous studies have reported similar results. The higher mortality risk associated with SUs reported in the current study further suggests that SU therapy should perhaps be used sparingly, as suggested by recent guidelines.

Kent DM, Hayward RA. Limitations of applying summary results of clinical trials to individual patients: the need for risk stratification. JAMA. 2007;298:1209-1212.