The variance shared across forms of childhood trauma is strongly associated with liability for psychiatric and substance use disorders

Abstract

Introduction: Forms of childhood trauma tend to co-occur and are associated with increased risk for psychiatric and substance use disorders. Commonly used binary measures of trauma exposure have substantial limitations. Methods: We performed multigroup confirmatory factor analysis (CFA), separately by sex, using data from the Childhood Trauma (CT) Study's sample of twins and siblings (N = 2594) to derive three first-order factors (childhood physical abuse, childhood sexual abuse, and parental partner abuse) and, as hypothesized, one higher order, childhood trauma factor (CTF) representing a measure of their common variance. Results: CFA produced a good-fitting model in the CT Study; we replicated the model in the Comorbidity and Trauma (CAT) Study's sample (N = 1981) of opioid-dependent cases and controls. In both samples, first-order factors are moderately correlated (indicating they measure largely unique, but related constructs) and their loadings on the CTF suggest it provides a reasonable measure of their common variance. We examined the association of CTF score with risk for psychiatric and substance use disorders in these samples and the OZ-ALC GWAS sample (N = 1538) in which CT Study factor loadings were applied. We found that CTF scores are strongly associated with liability for psychiatric and substance use disorders in all three samples; estimates of risk are extremely consistent across samples. Conclusions: The CTF is a continuous, robust measure that captures the common variance across forms of childhood trauma and provides a means to estimate shared liability while avoiding multicollinearity. Confirmatory factor analysis was used to derive a higher order, childhood trauma factor representing a measure of the common variance across three forms of trauma: childhood physical abuse, childhood sexual abuse, and parental partner abuse. We replicated the model in a second sample. We then examined the association of childhood trauma score with risk for psychiatric and substance use disorders in these samples and a third sample in which the primary sample's factor loadings were applied finding factor scores to be strongly and consistently associated with liability for psychiatric and substance use disorders in all three samples.