A-126 Social Determinants Measures: Dichotomous or Dimensional?

Abstract

Abstract Objective: The study aimed to examine the use of dichotomous versus dimensional scores across two social determinant measures regarding their ability to predict cognitive and psychiatric outcomes in a diverse, urban, clinical sample. The goal was to better guide clinicians in the selection of appropriate research methodology pertaining to the use of cut-scores versus dimensional constructs. Method: The study included 215 adults referred for outpatient neuropsychological evaluation. Patients were classified into at-risk/low-risk groups based on predetermined Rapid Estimate of Adult Literacy in Medicine - Revised (REALM-R) and Adverse Childhood Experiences (ACE) cut-scores. Nine cognitive tests and two psychiatric self-report scales were administered. A series of hierarchical regression models were conducted to investigate the predictive power of dichotomous scores alone, and account for the incremental predictive power of dimensional scores. Results: For REALM-R, dichotomous scores alone accounted for significant variance, demonstrating robust predictive power across all cognitive tests (p < 0.001 to p < 0.05), but not for psychiatric outcomes. Dimensional scores provided significant incremental variance in five cognitive tests (p < 0.001to p < 0.05) during which the parameter estimate for the dichotomous scores was reduced to non-significance. Regarding ACE, dichotomous scores were positively associated with psychiatric symptoms (p < 0.001), but not with cognitive outcomes. The addition of dimensional scores did not predict significantly more variance. Conclusion: This study demonstrated the robust predictive power of dichotomous scores in REALM-R and ACE. Further, dimensional scores provided potent incremental predictive power for REALM-R and should be used to enhance clinical decision-making when applicable.