The Medicare Health Outcomes Survey (HOS) is intended to improve the health of beneficiaries by measuring and reporting health outcomes based on the physical (PCS) and mental (MCS) health component summary scores of the SF-36 items. The factor structure of the SF-36 has not been tested with the Medicare HOS population. The information is vital, however, to determine the appropriateness of the SF-36 as a measure of generic health for this population.

For this research we test the measurement invariance of the second-order factor models of the SF-36 with two populations enrolled in Medicare Advantage plans: persons with disabilities (PWD), younger than 65 years old, and Seniors, 65 years of age and older. Using confirmatory factor analysis, we estimate the factorial loadings of the SF-36 items in a study population of PWD (n=5,546) as well as with Seniors (n=80,196) to determine instrument performance for each group. We then took a random sample of the Seniors (n=5,500) along with all of PWD respondents to compare the structural invariance of the SF-36 between the groups.

The results suggest (a) strong factorial loadings (>0.80) for each of the eight latent constructs, and (b) Physical Health (PCS) and Mental Health (MCS) are highly correlated (r≥0.95) for both groups. We did find structural invariance between PWD and Seniors although the item factor loadings were different between the groups. These findings indicate that while the SF-36 may be appropriate to assess PWD and Seniors from the HOS at the subscale level, the component summary scores computed for PCS and MCS are inappropriate for this population.