Applying generalizability theory methods to assess continuity and change on the Adolescent Quality of Life-Mental Health Scale (AQOL-MHS).

Abstract

The Adolescent Quality of Life-Mental Health Scale (AQOL-MHS) is designed to measure quality of life in clinical samples of Latino adolescents aged 12-18years. Initial findings support its reliability, validity and conceptual model for use of its three domains (emotional regulation, self-concept and social context). Our current study tests the usefulness of the AQOL-MHS for tracking changes in HRQOL during the course of service use. Three waves of data were collected from 59 participants who were recruited from mental health clinics, where waves were spaced 4months apart. Participants were receiving services at baseline assessment and were tracked for follow-up appointments regardless of treatment status. We analyze conventional reliability statistics for individual differences (e.g., Cronbach's alpha and test-retest correlations), and to estimate the reliability of change, we carried out a variance decomposition analysis. The psychometric analyses from Chavez et al. (Qual Life Res 23(4):1327-1335, 2014) were replicated with comparable results. A generalizability theory (GT) analysis revealed that the AQOL-MHS domains can measure change reliably. The reliability estimates varied from .65 to .78. Although there was reliable change at the individual level, on the average the AQOL-MHS means improved only slightly over time. The reliability of change for all three scales in the AQOL-MHS was acceptable, but consistently lower than the Cronbach's alpha values for each wave. Future work will use this approach to adapt our current measure to provide a better reliability of within-person change and thus broaden its applicability for prospective use.