Psychometric Properties of the Behavioral Risk Factor Surveillance System (BRFSS) Health Related Quality of Life (HRQOL) Items (B-HRQOL) in Arthritis
T. Mielenz, E. Jackson, L.F. Callahan, R. DeVellis, S. Currey University of North Carolina, Chapel Hill, NC
Purpose: Measuring HRQOL is important in arthritis and SF-36v2 is the current gold standard. It is not clear to what extent the B-HRQOL corresponds to HRQOL for people with arthritis. This study's purpose is to assess the psychometric properties of the nine items B-HRQOL (four item Healthy Days Core Module and five item Healthy Days Symptom Module) in an arthritis sample using SF-36v2 as a comparison.
Methods: A cross-sectional study on data from two databases: 1) cohort from 16 family practices; 2) cohort from university subspecialty clinics and private rheumatology practices. In fall 2002, a self-report survey including the B-HRQOL and SF-36v2 was sent to 4,183 individuals reporting OA, RA, and fibromyalgia with 2181 (52%) responses. The first item of both the B-HRQOL and SF-36v2 is general health (GEN) with five responses (poor to excellent). All eight other B-HRQOL items ask healthy days in the past month. Exploratory principal components analyses with rotation when necessary were conducted to determine the underlying factor structure of B-HRQOL. Distributional properties of B-HRQOL and physical component summary (PCS) and mental component summary (MCS) of SF-36v2 were compared. The multi-trait and multimethod matrix (MTMM) was used to analyze the shared variation among the B-HRQOL and the PCS and MCS. The relative contribution of the B-HRQOL in predicting the PCS and MCS was determined by regression. Analyses were performed removing GEN from the B-HRQOL.
Results: All nine B-HRQOL items loaded primarily onto one factor (explaining 57% of the item variance) representing a reasonable solution for capturing overall HRQOL. After rotation a two factor interpretation for the nine items was clear, with four items capturing physical health (physical, activity, pain, and energy days) and three items capturing mental health (mental, depression, and stress days). GEN loaded with physical health and sleep did not load. All of the loadings for these two factors were greater than 0.70. Both the single factor and two factor versions were analyzed by MTMM. The eight item B-HRQOL correlated equally with PCS (r=-.65, p<0.0001) and MCS (r=-.67, p<0.0001). The B-HRQOL physical health factor correlated more with PCS (r=-.78, p<0.0001) than MCS (r=-.50, p<0.0001). The B-HRQOL mental health factor correlated more strongly with MCS (r=-.71, p<0.0001) than PCS (r=-.35, p<0.0001). The relative contribution of the B-HRQOL in predicting PCS was 73% (r-square .73) when GEN was included in the B-HRQOL score and 65% (r-square .65) when GEN was removed. The relative contribution of the B-HRQOL in predicting MCS was 56% (r-square .56) when GEN was included and removed.
Conclusions: The B-HRQOL appears to have strong psychometric properties in individuals with arthritis when compared to the SF-36v2. A one factor solution for the eight items is reasonable for overall quality of life and the two factors, representing physical and mental health, are reasonable for examining those aspects.
Disclosures: T. Mielenz, None; E. Jackson, None; L.F. Callahan, None; R. DeVellis, None; S.Currey, None.
Get email updates
To receive email updates about this page, enter your email address:
- Centers for Disease Control and Prevention
Quality of Life
1600 Clifton Rd
Atlanta, GA 30333
TTY: (888) 232-6348
- Contact CDC-INFO