Fig. 2

Power and conditional power to detect a difference in response between triple therapy vs methotrexate combined with etanercept as a function of the sample size. The difference anticipated in a new study was of OR = 0.71 in American College of Rheumatology (ACR) 50 response favoring methothrexate/etanercept. Power for a single randomized trial considered in isolation (dotted line), conditional power for a fixed effect pairwise meta-analysis (dashed line, two studies), and network meta-analysis (solid line). Expected event rate in the triple therapy was assumed equal to the average observed in the network (49%). Calculations were performed using the conditional power method [12]