Background: Several methods for testing and adjusting for funnel plot asymmetry can provide useful resultswhen applied to a large set of fairly homogeneous trials [1]. As Cochrane reviews typically include few or heterogeneous studies, methods to adjust for small study effects are not often used. Networks of trials comparing several interventions for the same condition are increasingly considered in evidence synthesis and they can offer, under certain assumptions, an interesting opportunity to explore small study effects particularly for those comparisons with few studies. Objectives: To explore extensions of established regression approaches for testing and adjusting for small study effects in the context of multiple-treatments meta-analysis. Methods: We present how different methods for testing and evaluating small study effects adapt to multiple-treatments meta-analysis context. We use multiple-treatments meta-regression extending the variations of Egger’s meta-regression as described in a recent review paper [1]. The fit of the models and changes in inconsistency are monitored and several assumptions regarding the exchangeability of coefficients across comparisons of interventions in the same network or between networks in the same field are evaluated. We apply the methods to various networks of interventions and we observe the impact of adjustment in the relative ranking of the competing treatments. Results: Funnel plot asymmetry was present in some pairwise comparisons and adjustment had some effect in the relative effectiveness of pairs of treatments. However, when the entire network was considered, adjustment for small study effects was not associated with material changes in the fit, heterogeneity or consistency.
Reference
1. Moreno SG, Sutton AJ, Ades AE et al. Assessment of regressionbased methods to adjust for publication bias through a comprehensive simulation study. BMC.Med Res Methodol. 2009; 9: 2.