Reliability and construct validity of PROMIS® measures for patients with heart failure who undergo heart transplant

Abstract

Purpose: To evaluate the reliability and construct validity of measures from the Patient-Reported Outcomes Measurement Information System® (PROMIS®) for patients with heart failure before and after heart transplantation. Methods: We assessed reliability of the PROMIS short forms using Cronbach’s alpha and the average marginal reliability. To assess the construct validity of PROMIS computerized adaptive tests and short-form measures, we calculated Pearson product moment correlations between PROMIS measures of physical function, fatigue, depression, and social function and existing PRO measures of similar domains (i.e., convergent validity) as well as different domains (i.e., discriminate validity) in patients with heart failure awaiting heart transplant. We evaluated the responsiveness of these measures to change after heart transplant using effect sizes. Results: Forty-eight patients were included in the analyses. Across the many domains examined, correlations between conceptually similar domains were larger than correlations between different domains of health, demonstrating construct validity. Health status improved substantially after heart transplant (standardized effect sizes, 0.63–1.24), demonstrating the responsiveness of the PROMIS measures. Scores from the computerized adaptive tests and the short forms were similar. Conclusions: This study provides evidence for the reliability and construct validity (including responsiveness to change) of four PROMIS domains in patients with heart failure before and after heart transplant. PROMIS measures are a reasonable choice in this context and will facilitate comparisons across studies and health conditions.