Session: Methods for Mediation and Moderation Data Synthesis (Society for Prevention Research 22nd Annual Meeting)

The purpose of this symposium is to describe recent methodological developments in how to combine mediation and moderation results from different research studies. The symposium describes methods to synthesize data from research trials to gain information about how treatments affect outcomes through mediating variables and how treatment effects differ across subgroups.  The topics cover methods to generate comparable measures known as harmonization across studies, network methods to combine results across studies, limits of data synthesis because of different information from available studies, data synthesis as a missing data problem, methods to address the redundancy of studies added to a body of research literature and likelihood approaches to combining information across trials. In the poster entitled, Theoretical and Empirical Limits in Synthesis: Findings from Integrating Individual Data across Similar and Diverse Randomized Trials” the major challenges to synthesizing trials that may differ by population characteristics, trial level characteristics and intervention characteristics are outlined. Strategies for dealing with these issues are illustrated by the Familias Unidas prevention studies of drug abuse, HIV risk behaviors, and depression in Hispanic adolescents. The poster entitled, “Statistical Power for Moderation in Network Meta-analysis” describes how modern network analysis can be used to combine information from research trials in a way to maximize power to detect intervention effects. Simulated data and data from classroom based research trials illustrate these issues. In the poster, “Using IRT Models with Individual Participant Data to Evaluate Construct Equivalence and Harmonization”, approaches to combining information on different variables across different trials are outlined based on item response theory.  The poster “Adjusting for Study Redundancy in Second-Order Meta-Analysis” describes issues with combining the results from new trials with an existing database of research trials.  The poster entitled, “Multiple Imputation for Harmonizing Non-Commensurate Measures across Multiple Prevention Trials” describes the problem of missing measures across trials and describes how multiple imputation addresses this problem.  The poster, “An R Program for Synthesizing Mediation Analyses across Multiple Trials” describes a likelihood based method and computer program to combine mediation results across trials.  As prevention science matures, more research trials will be conducted. Rather than treat each study alone, more information can be extracted by synthesizing the results across trials.  This poster symposium describes the major issues in combining data across trials, suggests several solutions, and applies these solutions in the analysis of real data.