• Journal Article

External validity of the cardiovascular health study: a comparison with the Medicare population

Citation

DiMartino, L. D., Hammill, B. G., Curtis, L. H., Gottdiener, J. S., Manolio, T. A., Powe, N. R., & Schulman, K. A. (2009). External validity of the cardiovascular health study: a comparison with the Medicare population. Medical Care, 47(8), 916-923. DOI: 10.1097/MLR.0b013e318197b104

Abstract

BACKGROUND: The Cardiovascular Health Study (CHS), a population-based prospective cohort study, has been used to identify major risk factors associated with cardiovascular disease and stroke in the elderly.

OBJECTIVE: To assess the external validity of the CHS.

RESEARCH DESIGN: Comparison of the CHS cohort to a national cohort of Medicare beneficiaries and to Medicare beneficiaries residing in the CHS geographic regions.

SUBJECTS: CHS participants and a 5% sample of Medicare beneficiaries.

MEASURES: Demographic and administrative characteristics, comorbid conditions, resource use, and mortality.

RESULTS: Compared with both Medicare cohorts, the CHS cohort was older and included more men and African American participants. CHS participants were more likely to be enrolled in Medicare managed care than beneficiaries in the national Medicare cohort. Compared with the Medicare cohorts, mortality in the CHS was more than 40% lower at 1 year, approximately 25% lower at 5 years, and approximately 15% lower at 10 years. There were minimal differences in comorbid conditions and health care resource use.

CONCLUSION: The CHS cohort is comparable with the Medicare population, particularly with regard to comorbid conditions and resource use, but had lower mortality. The difference in mortality may reflect the CHS recruitment strategy or volunteer bias. These findings suggest it may not be appropriate to project absolute rates of disease and outcomes based on CHS data to the entire Medicare population. However, there is no reason to expect that the relative risks associated with physiologic processes identified by CHS data would differ for nonparticipants.