Objectives: Previous validation of the HIV Treatment Satisfaction Questionnaire status version (HIVTSQs) found that nine of the 10 items performed well, but the demands item needed revision. This study investigated the psychometric properties of the revised HIVTSQs and new change version (HIVTSQc).
Methods: English-speaking Americans completed the HIVTSQs at baseline and Week 48 of a clinical trial of HIV treatments, and the HIVTSQc at Week 48. Demographic and viral load information was collected. Psychometric validation used item frequency distributions, Confirmatory Factor Analysis (CFA), item-total correlations, Cronbach's alpha, Spearman's rank correlation, Kruskal-Wallis and Mann-Whitney tests.
Results: At baseline, 126 of the 152 patients completed the HIVTSQs fully (100 of 106 at Week 48). The negatively skewed distribution of the revised demands item resembled that of the other nine, with comparable missing data. CFA (baseline and Week 48) supported the general satisfaction/clinical subscale (alpha 0.83; 0.85), lifestyle/ease subscale including demands (alpha 0.82; 0.85), and 10-item treatment satisfaction scale (alpha 0.89; 0.91). Subscale and scale scores differed significantly between ethnic groups. Viral load was not significantly related to subscale or scale scores. At Week 48, 97 of 106 patients completed the HIVTSQc fully. All items had negatively skewed distributions. CFA supported two subscales (general satisfaction/clinical change alpha 0.85; lifestyle/ease change alpha 0.88) and 10-item treatment satisfaction change scale (alpha 0.92). Viral load change between baseline and Week 48 correlated significantly with patients' perceptions of change in HIV control (control(c) item), but not with scale or subscale scores.
Conclusion: The excellent psychometric properties of the HIVTSQs and HIVTSQc support their use in clinical trials.