To the Editor: We believe that the analysis reported by Dr Bassler and colleagues1 incorporated an important scientific and logical error that led to invalid conclusions. The authors separated trials into 2 groups: one that stopped early because of apparent efficacy and another that did not stop early. The main conclusion was that trials that stopped early for efficacy yielded biased estimates for the treatment effect, and the authors presented analyses suggesting that the relative risk (RR) in studies that stopped early was 0.71 relative to those that did not stop early.
Scott M. Berry, Bradley P. Carlin, Jason Connor. Bias and Trials Stopped Early for Benefit. JAMA. 2010;304(2):156–159. doi:10.1001/jama.2010.930