Factor structure and measurement invariance of the Brief Addiction Monitor

Abstract

Substance use disorders (SUDs) are of great concern for health care providers working with military veteran populations. Systematic evaluation of progress and outcomes within the Veterans Affairs (VA) is a critical component of care provided for the veteran population. The Brief Addiction Monitor (BAM) is a 17-item instrument used within VA to assess substance use and related constructs among veterans participating in SUD care. Initial evaluations, using a version containing continuous items, suggested that the items form three factors reflecting substance use, risk factors, and protective factors. Subsequent work, using the BAM version containing Likert-style items collected from a single VA Medical Center sample, did not support the proposed 3-factor solution. The current study used a nationwide sample of 4955 veterans to evaluate the factor structure of the BAM and its usefulness over time. Exploratory factor analyses conducted did not provide evidence of the originally proposed BAM factor structure but instead supported a 4-factor model (reflecting alcohol use, stress, risk, and stability) formed from 13 of the items. Further analyses conducted within a structural equation modeling framework showed that the four-factor model exhibited invariance across occasions of measurement, although internal consistency was found to be low for most subscales. Results provide caution against using BAM subscale scores to track treatment outcomes over time.

Keywords: Addiction; Brief Addiction Monitor; Factor analysis; Measurement; Structural equation modeling; Veteran.