Evaluation of item matching strategies to harmonize assessment tools for psychopathology in children and adolescents

The Reproducible Brain Charts initiative aims to aggregate and harmonize phenotypic and neuroimage data to delineate novel mechanisms regarding the developmental basis of psychopathology in youth and yield reproducible growth charts of brain development. To reach this objective, the second step of our project is to test item-wise matching strategies of phenotypic harmonization between studies using bifactor models of psychopathology. We focused on this model because general and specific aspects of mental health problems can dissociated, so more specific relationships with the brain could be established. In the current study, we benchmarked six item matching strategies for harmonizing the Child Behavioral Checklist (CBCL) and the Sstrenghts and Difficulties Qquestionnaire (SDQ) within a bifactor model framework in two samples that were assessed with both instruments. It proceded in the following steps: 1) harmonization of items according to the six strategies, 2) estimated bifactor models with harmonized items for each sample separately, 3) estimated factor score correlation between assessment tools in each sample, 4) estimated factor reliability, 5) tested the assessment’s invariance according to each strategy and 6) calculated the root expected mean square difference (REMSD) to estimate the factor score difference of using a proxy measure instead of a target measure while integrating the two samples. We expect that the results of this study can encourage the use of the best streategy to date to increase reproducibility in the field while aggregating data from different contexts and instruments in the context of the bifactor model of psychopathology.