Safety assessment of an anti-obesity drug (sibutramine): a retrospective cohort study.

BACKGROUND: Obesity is a serious and rapidly growing health problem worldwide. Few therapies are available beyond diet, exercise and bariatric surgery. A previously approved medication, sibutramine, has been withdrawn from the market due to concerns over the potential of increased risk of cardiovascular (CV) events, based on a phase IV clinical trial that included only individuals at high risk for CV events. OBJECTIVE: The aim of the study was to compare sibutramine users and matched non-users on rates of CV events, both overall and stratified by whether the patient qualified for on-label sibutramine use, using data from real-life clinical practice. METHODS: A retrospective cohort was constructed from electronic medical record data from physician office practices (mostly primary care) in the UK and Germany, using the LifeLink™ database from IMS Health Incorporated. For patients with at least one physician visit in which sibutramine was prescribed between 1 April 1999 and 31 October 2008, the date of their first such prescription was their index date. Users and non-users were matched 1 : 1 on index date (within 30 days), sex, age group (six categories), Charlson Comorbidity Index and evidence of obesity (high body mass index [BMI] or, if BMI was missing, diagnosis of obesity or very high weight relative to height). The resultant total samples analysed were 6186 in Germany and 7264 in the UK. User and non-user cohorts in the samples were compared according to the ratio of their crude incidence rates of acute myocardial infarction (AMI), stroke and either AMI or stroke per 1000 patient-years of follow-up. Cox regression analysis was used to compare the risk of CV events as a hazard ratio (HR) with 95% confidence intervals (CIs) between sibutramine user and non-user cohorts, controlling for label status and/or history of prior CV disease at baseline. RESULTS: The risk of AMI, stroke and either AMI or stroke was not higher among sibutramine users than comparable non-users of sibutramine in both Germany and the UK [Germany: HR 0.47 (95% CI 0.17, 1.26), 0.43 (0.23, 0.81) and 0.44 (0.26, 0.75), respectively; UK: HR 0.44 (0.15, 1.31), 0.63 (0.25, 1.60) and 0.54 (0.27, 1.10), respectively]. Regardless of whether or not the model controlled for prior CV disease (CVD), the direction and statistical significance of the differences did not change. In the sensitivity analyses including only those without a history of CVD in the 365 days prior to the index date there was no increased risk of CV events in either Germany or the UK. CONCLUSION: This study offers a framework for the safety assessment of anti-obesity drugs using an observational epidemiological study design. Large electronic health databases were used to construct retrospective cohorts to examine the risk in a population using one specific anti-obesity drug. Use of sibutramine in general practice settings was not found to increase the risk of acute CV events.