aEstimates were adjusted for education, race/ethnicity, baseline existential well-being, baseline social support, advance care planning, end-of-life discussion, health insurance status, patient-physician relationship, degree of positive religious coping, preferences for aggressive care, religiousness, spiritual support from religious communities, Northern versus Southern recruitment site, and terminal illness awareness. Models repeated with cluster analysis by site with findings unchanged.
bSample reduced from 339 due to missing data; findings unchanged when analysis repeated with data imputed to mean values.