Computerized adaptive testing for assessing substance use and substance use disorder: A systematic review

Computerized Adaptive Tests (CATs) are adaptive assessments whereby an individual's initial item responses determine the subsequent items they receive. CATs can measure a construct with relative precision in less time than fixed-length assessments. Although several CATs have been developed to assess substance use and substance use disorder (SUD), they have not been systematically evaluated.

We conducted a pre-registered systematic review (PROSPERO ID = CRD42023422464) of CAT development and validation studies that assessed substance use or SUD to determine the size and scope of this literature as well as describe its strengths and weaknesses. We searched seven electronic databases and manually searched reference lists. Articles were included if they described a CAT for substance use or SUD developed among individuals with any substance use or SUD, resulting in 15 eligible records representing 11 CATs. Information on study characteristics, CAT methodological choices (e.g., start rule, stop rule, scoring algorithm), psychometric outcomes (e.g., construct validity), and study quality were extracted in duplicate.

The 11 CATs assessed a range of constructs (e.g., craving, substance use) across several substances (e.g., alcohol, cannabis) with various assessment goals (e.g., screening, diagnosis). There was notable heterogeneity with respect to CAT methodological choices and the psychometric properties of the CATs were minimally evaluated in the original studies.

In general, CAT development and validation research would benefit from greater attention to the measurement of substance use and SUD in the initial item battery, inclusion of more diverse samples, more rigorous tests of validity, and greater attention to implementation considerations.