Objectives: The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT) represents one of the largest and most comprehensive efforts to describe patient preferences in seriously ill patients, and to evaluate how effectively patient preferences are communicated. Our objective was to review findings from SUPPORT describing the communication of seriously ill patients' preferences for end-of-life care.
Methods: We identified published reports from SUPPORT describing patient preferences and the communication of those preferences. We abstracted findings that addressed each of the following questions: What patient characteristics predict patient preferences for end of life care? How well do physicians, nurses, and surrogates understand their patients' preferences, and what variables are correlated with this understanding? Does increasing the documentation of existing advance directives result in care more consistent with patients' preferences?
Results: Patients who are older, have cancer, are women, believe their prognoses are poor, and are more dependent in ADL function are less likely to want CPR. However, there is considerable variability and geographic variation in these preferences. Physician, nurse, and surrogate understanding of their patient's preferences is only moderately better than chance. Most patients do not discuss their preferences with their physicians, and only about half of patients who do not wish to receive CPR receive DNR orders. Factors other than the patients' preferences and prognoses, including the patient's age, the physician's specialty, and the geographic site of care were strong determinants of whether DNR orders were written. In SUPPORT patients, there was no evidence that increasing the rates of documentation of advance directives results in care that is more consistent with patients' preferences.
Conclusions: SUPPORT documents that physicians and surrogates are often unaware of seriously ill patients' preferences. The care provided to patients is often not consistent with their preferences and is often associated with factors other than preferences or prognoses. Improving these deficiencies in end-of-life care may require systematic change rather than simple interventions.