Objectives: To determine whether patient preferences are a barrier to hospice enrollment.
Design: Prospective cohort study.
Setting: Fifteen ambulatory primary care and specialty clinics and three general medicine inpatient units.
Participants: Two hundred three seriously ill patients with cancer (n=65, 32%), congestive heart failure (n=77, 38%), and chronic obstructive pulmonary disease (n=61, 30%) completed multiple interviews over a period of up to 24 months.
Measurements: Preferences for high- and low-burden life-sustaining treatment and site of death and concern about being kept alive by machines.
Results: Patients were more likely to enroll in hospice after interviews at which they said that they did not want low-burden treatment (3 patients enrolled/16 interviews at which patients did not want low-burden treatment vs 47 patients enrolled/841 interviews at which patients wanted low-burden treatment; relative risk (RR)=3.36, 95% confidence interval (CI)=1.17-9.66), as were interviews at which patients said they would not want high-burden treatment (5/28 vs 45/826; RR=3.28, 95% CI=1.14-7.62), although most patients whose preferences were consistent with hospice did not enroll before the next interview. In multivariable Cox regression models, patients with noncancer diagnoses who desired low-burden treatment (hazard ratio (HR)=0.46, 95% CI=0.33-0.68) were less likely to enroll in hospice, and those who were concerned that they would be kept alive by machines were more likely to enroll (HR=5.46, 95% CI=1.86-15.88), although in patients with cancer, neither preferences nor concerns about receiving excessive treatment were associated with hospice enrollment. Preference for site of death was not associated with hospice enrollment.
Conclusion: Overall, few patients had treatment preferences that would make them eligible for hospice, although even in patients whose preferences were consistent with hospice, few enrolled. Efforts to improve end-of-life care should offer alternatives to hospice that do not require patients to give up life-sustaining treatment, as well as interventions to improve communication about patients' preferences.