1 - (IOP 6) What Is the Effect of Child Maltreatment on Psychopathology? The Impact of Contamination and Solutions for Correcting It to Improve the Accuracy of Causal Estimates

Significance
Child maltreatment is a highly-prevalent, world-wide public health concern affecting the onset and course of multiple forms of psychopathology. However, variation in the significance and magnitude of maltreatment effects across independent research has resulted in a growing number of replication and reproducibility failures. These failures suggest that maltreatment effects may not be as important or severe as we think, affecting how we plan future research and whether we consider maltreatment in the etiology of a presenting complaint.

Innovation
The presence of contamination can help explain these replication and reproducibility failures. Contamination occurs when those assigned to a comparison group are exposed to maltreatment. This minimizes the significance and magnitude of maltreatment effects because the severity of psychopathology in a control group will approximate the severity of psychopathology in a maltreatment group simply because some members of the control group were exposed to maltreatment. Unfortunately, there are no established solutions for how to correct contamination in maltreatment research. This presentation examines the impact of contamination and ways to correct it to improve the accuracy of maltreatment effects.

Approach
We review the prevalence of contamination in seminal cohorts examining maltreatment effects on psychopathology, where it ranges from 15%-65%. We then present results from studies (N=4023) examining the impact of contamination on the significance and magnitude of maltreatment effects on psychopathology, where it increased replication failures by up to 50% and where effect magnitudes were reduced by 15%-130%. Finally, we present novel results from propensity score models that estimated maltreatment effects on psychopathology in two independent, nationally-representative, racially and ethnically diverse, prospective cohorts. Cohort 1: The National Survey of Child and Adolescent Well-being-II (N=5872) is a multi-wave, national probability sample of children 0-17.5 years of age involved in the U.S. child welfare system. When contamination was ignored, maltreatment had no effect on internalizing behaviors (b = -0.68, p = .14) and led to statistically significant reductions in externalizing behaviors (b = -.98, p = .04), failing to replicate prior research. When contamination was corrected, maltreatment led to significantly more severe internalizing (b = 10.30, p < .001) and externalizing (b = 8.46, p = .001) behaviors that were large in magnitude (SMD’s = .85-1.03). Cohort 2: The Longitudinal Studies of Child Abuse and Neglect (N=1354) is a multi-wave, multi-site cohort of maltreatment exposure from birth to age eighteen. Maltreatment prior to age four predicted more severe transdiagnostic psychopathology symptoms in adolescence before and after correcting contamination (all p’s < .05). However, effect magnitudes increased 135% in models that corrected contamination (SMD=.35) relative to models that ignored it (SMD=.14). These results suggest contamination is a feature, not a bug, of conducting research on child maltreatment. Developing and using solutions for correcting it can improve the accuracy of maltreatment effects on psychopathology and beyond.