Methods/Statistics
Multilevel network meta-regression for population-adjusted treatment comparisons: transporting treatment effects to target populations for decision-making David Phillippo* David Phillippo
Network meta-analysis (NMA) and indirect comparisons combine aggregate data from multiple randomised controlled trials to estimate relative treatment effects, assuming that any effect modifiers are balanced across populations. Population adjustment methods aim to relax this assumption, using individual patient data available from one or more studies to adjust for differences in effect modifiers between populations. These methods are closely related to those in the causal inference literature for transporting estimates from trials to target populations, and to recent developments in causally-interpretable meta-analysis.
We introduce multilevel network meta-regression (ML-NMR), a general method for synthesising individual and aggregate data in networks with any number of studies and treatments, extending the standard NMA framework. An individual-level regression model is defined, and aggregate study data are incorporated appropriately by integrating this model over the covariate distributions of the respective studies, which avoids aggregation bias. This marginalisation integral is evaluated using Quasi-Monte Carlo integration, providing a general approach that is applicable to models and likelihoods of any form. Crucially for decision making, estimates may be transported into any target population with a given covariate distribution.
We demonstrate this approach with a network of trials and treatments for moderate-to-severe plaque psoriasis, transporting estimates into a decision population with characteristics informed by a representative registry study. We discuss the links between population adjustment methods like ML-NMR and the literature on transportability and causally-interpretable meta-analysis, and suggest areas for future research and cross-pollination. A user-friendly R package multinma is available for performing ML-NMR analyses.