Methods. We identified all 169,036 VA patients over age 67 who died in fiscal year 2000 or 2001, and we obtained records of all their VA and Medicare inpatient and outpatient utilization. To identify patients with dementia, we adapted the disease classification method developed by the VA Dementia Registry Task Force. We measured aggressiveness of medical services using ICU admission and four surgical procedures identified by a panel of physicians for a previous study: ventilator, pulmonary artery monitor, cardiac catheterization, and dialysis. We compared the use of each aggressive treatment between acute hospital patients with dementia and those without dementia using both bivariate analyses and Probit regressions controlling for age, sex, race, principal diagnosis, and Charlson Comorbidity Score. Our primary analyses investigated care received in the final 30 days of life, but we also analyzed data for care in the final 90 days and final year of life.

Results. We identified 31.6% of decedents as having dementia, with higher prevalence among older patients, women, and blacks. In bivariate analyses stratified by age, we found that across age groups, patients with dementia were less likely to receive each aggressive treatment during a given medical/surgical hospital stay. Probit regressions found that patients with dementia were substantially less likely to receive each aggressive treatment, controlling for other factors. The Probit model predicted patients with dementia to be 20% less likely to be admitted to the ICU, 32% less likely to be placed on ventilator support, 26% less likely to receive cardiac catheterization, 53% less likely to receive pulmonary artery monitoring, and 14% less likely to receive dialysis, compared to patients without dementia. We found similar results when including care in the final 90 days and final year of life, and when limiting analyses to only VA or only Medicare.

Conclusion. Our study demonstrated a substantial disparity between the aggressiveness of in-hospital, end-of-life care for persons with dementia compared to those without dementia. These differences may be related to patient preferences, or they may reflect inappropriate care. Such striking differences warrant further research into whether the observed disparity represents appropriate care.