Validation of PROMIS® depression item bank computer adaptive testing among patients with heart failure

Background & objective Depressive symptoms are frequent among patients with heart failure (HF). The Patient-Reported Outcomes Measurement Information System Depression Computer Adaptive Test (PROMIS-D CAT) uses item response theory (IRT) to measure depressive symptoms, allowing for tailored, precise, and efficient assessment. We aim to assess the validity and reliability of PROMIS-D CAT to screen depressive symptoms among patients hospitalized with HF. Methods Cross-sectional convenience sample of adults hospitalized with HF, who completed the PROMIS-D CAT and Patient Health Questionnaire-9 (PHQ-9). PHQ-9 score ≥ 15 defined moderate-severe depressive symptoms. Participants completed the EuroQol 5-Dimension 5-Level (EQ-5D-5L), Kansas City Cardiomyopathy Questionnaire (KCCQ), Edmonton Symptom Assessment System-revised (ESAS-r). Convergent validity was measured by Spearman's correlation between PROMIS-D CAT and PHQ-9, ESAS-r depression item, and EQ-5D-5L depression/anxiety item. Known-group analysis compared PROMIS-D CAT and PHQ-9 scores between groups expected to have different levels of depression. Reliability of PROMIS-D CAT was calculated on individual and group level from standard error of measurement, according to IRT. Area under receiver-operating characteristics (AUROC) curve assessed discrimination. Results 337 participants completed PROMIS-D CAT & legacy measures. Mean (standard deviation [SD]) age was 67 (16) years, 219 (65%) were male, 118 (35%) were female. Moderate or severe depressive symptoms were reported by 70 (21%) of participants. PROMIS-D CAT T-scores were moderately correlated with PHQ-9 scores (rho = 0.66) and other legacy measures. Participants completed fewer items with PROMIS-D CAT (median [IQR] 4[2]) vs 9 items for PHQ-9. PROMIS-D CAT showed high reliability (mean [SD] 0.91 [0.07]) across the T-score spectrum. Known-group analysis further supported construct validity. PROMIS-D CAT demonstrated excellent discrimination for identifying patients with moderate-severe depressive symptoms compared to a PHQ-9 cutoff ≥ 15, with an AUROC of 0.86. Conclusions PROMIS-D CAT is a valid and reliable measure for depressive symptoms, with excellent discrimination, and reduced question burden among patients hospitalized with HF. Impact statement This article provides evidence of construct validity and reliability of the PROMIS-D CAT to screen for potentially clinically significant depressive symptoms among patients hospitalized with HF.