Objective: To describe treatment preferences of nursing home residents, concordance with decisions by self-selected proxies and to establish the relationship of sociodemographic and functional measures to decisions.
Setting and subjects: 52 patient-proxy pairs at a Veterans Affairs nursing home.
Methods: Treatment preferences were elicited from residents and proxies regarding cardiopulmonary resuscitation, mechanical ventilation, and intensive care unit care. Hospitalization, intravenous antibiotics, intravenous fluid administration, and tube feeding were presented in three separate health scenarios. Concordance was determined for the entire interview and separately for each scenario. Treatment-seeking intensity and decision-making consistency were scored and used to explore associations with sociodemographic variables and function.
Results: Subjects were predominantly male (97%) and non-Hispanic white (74%); average age was 70 +/- 12 years, with 4 +/- 2.9 diagnoses. Residents accepted 70% of all treatments. The proportion of subjects accepting interventions declined parallel to health status in each scenario. Only 7/52 (13%) subjects made inconsistent decisions. Resident treatment acceptance was inversely associated with GDS scores but not associated with any other sociodemographic or functional measure. Concordance with proxies was no greater than chance. Proxies' decisions were not systematically biased against resident preferences or influenced by patient characteristics.
Conclusions: Veterans desired most treatments, but adjusted preferences according to health status and were not inconsistent. Depressive symptoms should be addressed prior to advance directive selection. The patient remains the best source of information, but proxies' decisions exhibit no bias and are not affected by patient status.