PRM141 - Validity of QOL Impact Attributions to Specific Diseases: A Multitrait-Multimethod Comparison

Abstract

To test convergent-discriminant validity of quality of life (QOL) impact attributions to arthritis and respiratory conditions using the multitrait-multimethod (MTMM) approach. Chronically-ill adults (N=601) with osteoarthritis (OA) and respiratory (asthma, COPD) disease completed Internet-based surveys. Ages ranged from 18-93 (median=58), 66.5% female and 20.7% nonwhite. QOL impact was measured using 3 methods: QOL Disease Impact Scale (QDIS) and disease severity with specific attribution to each condition, specific symptoms (joint pain, shortness of breath), and generic physical (PCS) and mental (MCS) summary measures (SF-36). The resulting 10X10 correlation matrix (3 disease-specific, 2 generic) for each disease was evaluated to test convergent validity (correlations > 0.40 between different methods of measuring the same disease) and discriminant validity (significantly lower correlations between different diseases measured by the same method). The MTMM matrix was also analyzed using principal component analysis (PCA) to test for disease versus methods factors. The MTMM matrix yielded strong evidence that patients discriminated the impact of one condition from the other. In support of convergent validity, disease specific QOL impact, severity and symptom measures (joint pain for OA, SOB for respiratory) correlated substantially for asthma (0.516 - 0.765) and OA (0.465 - 0.774). In support of discriminant validity, different diseases measured by the same method correlated significantly lower (0.264 - 0.286). PCA yielded two disease-specific factors (OA and respiratory) with high loadings across methods for OA (0.748 - 0.871 and respiratory (0.687 - 0.890) and low secondary loadings (0.085 - 0.262). Generic PCS and MCS measures correlated lower than disease-specific measures, as hypothesized. 2-method (QOL impact, severity) MTMM validations replicated these results across 37 pairs of over-sampled comorbid conditions. Results strongly support the convergent-discriminant validity of survey measures of QOL impact based on attributions to OA and respiratory disease and suggest that disease-specific attributions may improve the validity of QOL impact attributions for other common pairs of comorbid conditions.