Performance of comorbidity measures to predict stroke and death in a community-dwelling, hypertensive Medicaid population

Abstract

Background and purpose: The Charlson and Elixhauser comorbidities are widely used to control for differences in comorbidity in epidemiological studies but have not been validated for outpatient studies of hypertension. This study sought that validation using death and stroke outcomes.

Methods: Using Cox models in a retrospective cohort study of 49,479 hypertensive patients, Modified Charlson Index was compared with 6 alternative approaches to assessing comorbidity: individual Charlson comorbidities, Elixhauser comorbidities, prior major cardiovascular disease event, traditional risk factors for cerebrovascular accident, healthcare utilization, and antihypertensive medication utilization. Comorbidity measures were calculated at baseline and for a period before occurrence of the study outcome of interest or study conclusion.

Results: The Charlson comorbidities had the smallest Akaike information criterion value for both the stroke and death outcomes when baseline data were used. The Elixhauser comorbidities had the smallest Akaike information criterion value for both the stroke and death outcomes when follow-up data were used. Modified Charlson Index also predicted stroke and death, but alternative models were more robust.

Conclusions: This study indicates that both the Charlson and Elixhauser comorbidities are valid prediction tools that could enable clinicians and health systems to better assess risk for stroke and death in patients with hypertension. However, the Charlson comorbidities perform better when comorbidities are assessed using baseline data, whereas the Elixhauser comorbidities perform better for short follow-up periods when comorbidities are assessed proximal to events of interest.