Is ACEs Screening for Adolescent Mental Health Accurate and Fair?

• NCJ Number: 307185
• Journal: Prevention Science Volume: 23 Dated: 2022 Pages: 1216-1229
• Author(s): Joseph R. Cohen; Jae Wan Choi
• Date Published: 2022
• Length: 14 pages
• Grant Number(s): 2018-R2-CX-0022
• Agencies: NIJ-Sponsored
• Publication Type: Research (Applied/Empirical)
• Award: A Novel, Trauma-Informed Screening Approach for Teen Dating Violence Perpetration in Racially Diverse Adolescents: A Multi-Site Study

Annotation

This study examines the accuracy and fairness of an adverse childhood experiences (ACEs) measure as an index of environmental risk for adolescent mental health diagnoses.

Abstract

This study examining whether an adverse childhood experiences (ACEs) measure was accurate and fair as an index of environmental risk for adolescent mental health diagnoses found that ACEs screening can potentially produce biased results when used to inform mental health policy and prevention. Reasons for why results from ACEs screening may vary across adolescent subpopulations and the importance of testing statistical fairness for preventive mental health screening are discussed. Increasingly, adversity-focused assessment tools are being introduced into preventive mental health screening protocols. However, few studies have explicitly examined whether use of these instruments serves as equitable, clinically useful measures of mental health risk in adolescents. Secondary data analyses were conducted on the National Comorbidity Survey-Adolescent Supplement. Adolescents (N = 10,148; AgeMean = 15.20; 51.3% male; 65.6% White, 15.1% Black, and 14.4% Hispanic) answered ten questions concerning childhood adversities and completed diagnostic interviews for PTSD, depression, and externalizing disorders. In the overall sample, ACEs showed some clinical utility (e.g., area under the curve (AUCs) ≥ 0.64), diagnostic likelihood ratios (DLRs) > 4.0) and acceptable calibration (i.e., expected/observed indices’ confidence intervals included 1) across diagnoses. Within subpopulations, however, predictive validity varied. The AUCs were lower for multiple diagnoses in Black male and Hispanic female adolescents and DLRs suggested greater clinical utility for indexing mental health in White, female adolescents. Finally, models were not well-calibrated between adolescent subpopulations, suggesting recommended ACEs screening can potentially produce biased results when used to inform mental health policy and prevention. Reasons for why results from ACEs screening may vary across adolescent subpopulations and the importance of testing statistical fairness for preventive mental health screening are discussed.