Multiple Imputation for Harmonizing Non-Commensurate Measures Across Multiple Prevention Trials

 

Title:Multiple imputation for harmonizing non-commensurate measures across multiple prevention trials

There are many advantages to individual participant data meta-analysis for combining data from multiple studies. These advantages include greater power to detect effects, increased sample heterogeneity, and the ability to perform more sophisticated analyses. However, a fundamental challenge is that it is unlikely that all the studies to be combined use the same measure for the construct of interest. We propose that this situation can be viewed as a missing data problem and use multiple imputation to fill in missing measurements. We apply our method to 18 adolescent depression prevention trials (n=5292) consisting of 9 types of interventions, including both adolescent-only and family-based, where there is overlap in depressive symptom measures across studies, but no common measure used in all 18 trials. In addition, we present how to use related missing data procedures to combine covariate data across trials, such as parent education level, which is categorized differently across trials.  We describe the use of diagnostics to check whether imputed values are consistent with observed values. Our method can be implemented using standard imputation software and allows researchers to report the results of their analysis on an existing metric. To demonstrate the utility of our approach, we present the results of two analyses that can only be performed when missing depression measurements have been filled in for all participants. The first analysis compares treatment effects when the outcome is clinician-based versus a self-reported depression outcome. The second analysis looks at the moderating effect of baseline depression on intervention effects.