Childhood sexual abuse, marital discord, and marital disruption: Propensity score analyses

BACKGROUND: Childhood sexual abuse (CSA) is associated with greater relationship discord and a higher likelihood of marital disruption; however, these associations may be confounded by co-occurring childhood adversities and sociodemographic factors. Propensity score methods, which seek to improve comparability between exposed and unexposed groups on observed characteristics, provide a complementary approach to traditional covariate adjustment for addressing measured confounding. OBJECTIVE: To examine the association between CSA and marital discord and disruption using traditional covariate and propensity score methods. PARTICIPANTS AND SETTING: Participants were from the National Comorbidity Survey Replication, a nationally representative probability sample of United States adults. METHODS: Associations between CSA and marital discord and disruption were examined using four approaches: bivariate (unadjusted), traditional covariate-adjusted, propensity score-adjusted, and propensity score matching (PSM) analyses. Covariates included demographic and childhood characteristics. Post-matching models additionally adjusted for covariates with residual absolute standardized differences ≥0.10. RESULTS: CSA was associated with greater marital discord in the bivariate, traditional covariate-adjusted, propensity score-adjusted, and PSM analyses. After additional post-matching covariate adjustment, the estimated association remained positive and similar in magnitude. CSA was associated with higher odds of marital disruption across all models, including the post-matching covariate-adjusted model. CONCLUSIONS: For both marital outcomes, estimates were consistent in direction and similar in magnitude across traditional covariate-adjusted and propensity score approaches, which address measured confounding differently. Future research should investigate underlying mechanisms using longitudinal and dyadic designs.