Precision and comparability of adverse event rates of newer antidepressants

Due to the scarcity of clinical trials (CTs) directly comparing newer antidepressants, clinicians and reviewers often compare drugs on the basis of active drug and placebo-subtracted adverse event rates (AERs) published in references such as the Physician's Desk Reference in the United States, the Compendium of Pharmaceutical Specialities in Canada, and product monographs. This study examined the suitability of the above data and methods for comparing AERs of nine newer antidepressants: bupropion, citalopram, fluoxetine, fluvoxamine, moclobemide, nefazodone, paroxetine, sertraline, and venlafaxine. First, the authors examined data precision and comparability across drugs by analyzing placebo-arm AERs and their 95% confidence limits for each drug and by testing the homogeneity of placebo-arm AERs across drugs. The rationale was that placebo AER heterogeneity is likely a reflection of heterogeneity of nonpharmacologic factors among CTs. For all 16 adverse events examined, placebo AERs were found to be significantly heterogenous, suggesting significant nonpharmacologic heterogeneity across drugs. Second, using placebo AER heterogeneity as a measure of nonpharmacologic heterogeneity, the authors estimated the degree to which active drug and placebo-subtracted AERs were related to nonpharmacologic heterogeneity by estimating the degree of association between active drug and placebo AERs and between placebo-subtracted and placebo AERs. Also, 95% confidence limits for placebo-subtracted AERs were calculated. It was found that most active drug AERs and some placebo-subtracted AERs were significantly correlated with placebo AERs, suggesting a relationship with nonpharmacologic heterogeneity. Placebo-subtracted AERs were less strongly related than were active drug AERs. Finally, the authors discuss factors that may be associated with the observed heterogeneity and offer suggestions that may improve the methodology and reporting of CTs.