Prof Freedom Gumedze (UCT)
Network meta-analysis with random inconsistency effects and outliers
Abstract: Network meta-analysis (NMA) expands the scope of a standard meta-analysis to simultaneously handle multiple treatment comparisons. The studies in a network meta-analysis may be heterogeneous and the
network may be inconsistent. Random effects may be used to describe any inconsistency in the network. In addition, some trials may appear to deviate markedly from the others and thus be inappropriate to be synthesized in the NMA. In addition, the inclusion of these trials in evidence synthesis may lead to bias in estimation. Therefore the presence of
such outliers could substantially alter the conclusions in a network meta-analysis. This paper proposes a methodology for identifying and, if desired, down weighting studies that do not appear representative of the population they are thought to represent. An outlier is taken as a study result with an inflated random effect variance. We used the likelihood ratio test statistic as an objective measure for determining whether observations have inflated variance and are therefore considered outliers. The NMA model is formulated using a contrast-based approach. The proposed methodology is then applied to a network meta-analytic dataset from the literature.