Non steroidal anti-inflammatory drugs and cardiovascular risk: identifying evidence for channelling bias in a population based study

Abstract

ABSTRACT The non-steroidal anti-inflammatory drug (NSAID), diclofenac, has been associated with a high risk for cardiovascular events in observational studies. However, majority of studies identifying this association were conducted when diclofenac was the only NSAID that could be obtained as a combination product (i.e., formulated with misoprostol). As a result, channelling bias might have resulted if prescribers selected the combination of diclofenac/misoprostol (Diclo-Miso) in patients with poor health status frequently than other NSAID products. The main purpose of this study was to identify evidence for channelling bias in a cohort of patients with coronary heart disease (CHD) prescribed NSAIDs. Three independent, retrospective analyses were carried out using Saskatchewan’s health administrative databases. Patients were eligible if they were hospitalized with CHD event between January 1, 1994 and December 31, 2008. In the first analysis, a time series was conducted to examine trends in the use of NSAIDs following discharge from original hospitalization. In the second analysis, multivariate logistic regression models were constructed to identify characteristics of patients prescribed with Diclo-Miso in comparison to single-entity diclofenac. Finally, a nested case-control study was conducted to examine the risk for recurrent myocardial infarction (MI)/ Unstable Angina (UA) or death among patients prescribed with Diclo-Miso versus single-entity diclofenac. For each case, up to five controls were matched by age and sex. Between 1994 and 2008, NSAIDs were used by 20.1% (3,099/15,393) of patients in the year following discharge from their original MI/UA hospitalization. Use of these agents was relatively stable until 2004 when the COX-2 selective agent rofecoxib was withdrawn from the market. Following this date (i.e., September 30, 2004), the use of Diclo-Miso and single-entity diclofenac appeared to follow different trends. However, available patient and disease specific factors could not explain diverging utilization trends. Further, no differences were observed in the risk of experiencing recurrent MI/UA between patients receiving Diclo-Miso (OR 0.88, 95% CI 0.72-1.08, p=0.22) or single-entity diclofenac (OR 0.78, 95% CI 0.60-1.00, p=0.06) versus patients not exposed to NSAIDs. Based on the study’s result, channelling bias does not appear to be a major threat to the analysis of cardiovascular toxicity of diclofenac products.