Selecting the best scale for measuring treatment effect in a network meta-analysis: a case study in childhood nocturnal enuresis

摘要: Dichotomous outcomes in pairwise meta-analysis are typically summarised using the odds ratio (OR), relative risk (RR) or difference (RD). The hazard (HR) may also be used where events occur over time. Choice of scale is often determined by ease interpretation mathematical properties a measure, as there frequently insufficient power to compare goodness-of-fit across different scales. Network (NMA) combines evidence network treatment comparisons. NMA allows combination greater numbers trials, so potential use statistics determine an appropriate on which effects treatments additive. In this paper, we explore choice childhood nocturnal enuresis for outcome 'failure achieve 14 days consecutive dry nights'. We OR, RR both harmful (RRH) and beneficial (RRB) outcomes, RD HR. Using Bayesian framework, posterior mean residual deviance information criterion evaluate model fit selection between summary effect measures. I(2) examine within-comparison heterogeneity analyses. results suggest that HR RRB provide best fit. conclude should based physiological epidemiological understanding disease process, together with empirical assessment adequacy. given consideration underlying time-to-event process. demonstrate how can transformed alternative aid interpretability. Copyright © 2012 John Wiley & Sons, Ltd.