This research letter describes the findings of a rapid assessment of the safety of varenicline tartrate, which was requested by the US FDA due to concerns that it may increase the risk of certain cardiovascular events. The assessment was carried out by the Mini-Sentinel program, which is part of the FDA’s Sentinel Initiative (a system under development for the monitoring of medical product safety).
Healthcare databases (with over 350 million person-years of longitudinal observation time) were used to identify individuals aged ≥20 years who had a diagnosis code for tobacco use disorder, with no existing cardiovascular disorder, and who filed a first prescription for varenicline (n=89,519) or a comparator (buproprion; n=113,378) in the study period. The outcome of interest was a composite of three cardiovascular endpoints - acute myocardial infarction, intermediate coronary syndrome or unstable angina, and acute coronary occlusion without myocardial infarction (determined using primary diagnosis coding in inpatient or emergency settings).
The adjusted incidence rate ratios (IRR) for varenicline versus bupropion for the outcome of interest were as follows:
• 1.02 (95% CI 0.71-1.47) for the primary analysis
• 0.98 (0.43-2.23) when all initiators of varenicline therapy were compared with initiators of therapy with a bupropion product specifically approved for smoking cessation (Zyban®) and its generics
• 1.52 (1.21-1.91) when all initiators of varenicline therapy were compared with all initiators of bupropion therapy (i.e. no restriction to patients with a tobacco use disorder code or those who used a bupropion product approved for smoking cessation).
The latter would have included patients receiving bupropion for other indications (e.g. depression) and therefore was expected to have been more confounded.
The authors concluded that there was no consistent evidence of increased cardiovascular risk during the first treatment episode with varenicline compared with bupropion in those with no recent diagnosis of cardiovascular events. These findings are from patients treated in clinical practice and therefore complement those from randomised controlled trials. They note that this was a rapid analysis and is not an evaluation of causality, and therefore the results need to be considered in the context of the primary purpose of the research (to provide safety information quickly), and the limitations of using electronic health data.