COMBINING EVENT RATES FROM CLINICAL-TRIALS - COMPARISON OF BAYESIAN AND CLASSICAL METHODS

OBJECTIVE: To compare an empirical Bayesian, a fully Bayesian, and a c lassical fixed-effect (Peto) method for pooling event rates from separ ate epidemiologic studies or clinical trials. DESIGN: Four data sets u sed in meta-analyses by previous authors were evaluated. The first dat a set concerned death rates observed in clinical trials of beta-blocke rs, the second to lung cancer and smoking in 14 case-control studies, the third to drowsiness induced by the antihistamine compound chlorphe niramine, and the fourth to the use of intravenous magnesium in patien ts with suspected myocardial infarction. Randomly chosen data points w ere made more extreme to test the methods further. MAIN OUTCOME MEASUR ES: Pooled estimates of effect expressed as odds ratios and their asso ciated 95% confidence intervals. RESULTS: All three methods gave compa rable results with respect to the 95% confidence interval, although th e Bayesian methods gave generally wider interval estimates. However, t he point estimates for the individual studies were substantially diffe rent, particularly for small studies. CONCLUSIONS: For the data sets c onsidered, Bayesian methods, which are computer intensive but intuitiv ely appealing, provided results that were consistent with the classic fixed-effect Peto method. Introduction of the more extreme data points did not alter this conclusion.