Reliability and validity of pain, fatigue, depression, and social roles participation PRO measures in populations of individuals aging with a physical disability.

Aims: To evaluate the reliability and validity of patient reported outcome measures across the domains of pain, depression, fatigue, and social roles participation in four different rehabilitation study populations. Methods: Individuals with muscular dystrophy (MD) (N=340), multiple sclerosis (MS) (N=584), spinal cord injury (SCI) (N=492), and post-polio syndrome (PPS) (N=446) participated in a self-report survey related to aging with a disability. Measures examined included the PHQ-9 depression scale and PROMIS short forms for fatigue, pain interference, depression, and social roles participation. Analyses for each scale and within each population included examinations of floor and ceiling effects, percentage of variance accounted for by the first factor and ratio of first and second eigenvalues using exploratory factor analyses (EFA), and calculations of Chronbach’s alpha. Results: EFA analyses supported the unidimensionality of scores from all scales in all four populations. Across scales and populations the lowest eigenvalue ratio was 8.2 (PHQ-9) and the highest was 52 (PROMIS pain). The variance accounted for by the first factor ranged from 49% (PHQ-9) to 86% (PROMIS pain). The PROMIS pain and depression scales demonstrated floor effects, especially in individuals with MS and MD. On the pain scale, 20.5% and 23.8% of MD and MS individuals respectively were at the floor while 10.9% and 15.7% were at the floor on the depression scale. The PHQ-9 also demonstrated moderate floor effects with up to 9.5% of individuals at the bottom of the scale. With the exception of the social roles scale none of the scales had more than 2% of individuals at the ceiling of the scale. Ceiling effects on the social roles scale ranged from 1.2% (PPS) to 7.2% (MD). Chronbach’s alpha values ranged from 0.80 (fatigue in PPS) to 0.96 (pain in MS). Conclusions: EFA results support the unidimensionality of the response data, and thus the construct validity of the studied measures. Although there were some floor effects, these appear to be due to lack of symptoms in portions of the sample (e.g., persons reporting no pain), rather than inadequacy of the scale. The results of the study support the use of the measures in MS, MD, PPS, and SCI.