Analysis of these elderly informants' narratives revealed several clinically important findings that contradict current clinical practice. First, in decision making during serious illness, patients focus on outcomes
, not treatments. To our informants, specific interventions such as CPR or mechanical ventilation were largely irrelevant beyond the ability of these treatments to restore patients' valued life activities. Patients' outcomes orientation contrasts sharply with physicians' own focus on patients' preferences for specific life-sustaining treatments, as they seek to ascertain patients' “code status.” Such disparate orientations may explain the difficulties providers experience in end-of-life discussions with their patients. Our findings also suggest that highly specific advance directive documents, in which patients are asked to choose from a checklist of many possible interventions in a variety of illness scenarios,17
may not match patients' own models of end-of-life planning. Our informants' perspectives were more closely aligned with an approach that identifies patients' most valued life activities or “states worse than death.”18,19
These outcomes-oriented formats, designed to clarify health states in which any life-prolonging medical treatments would be unacceptable, appeared to provide a better match to informants' end-of-life objectives.
The role of advanced age in end-of-life decisions has been the subject of considerable debate. Age per se has been inconsistently associated with survival after CPR20–23
or other interventions used in serious illness,24–29
yet studies have consistently shown that elderly patients receive less aggressive treatment for many illnesses, including breast cancer,30,31
coronary artery disease,32
and acute critical illness.33
The discrepancy between treatment outcomes and the provision of such treatments has been attributed to age bias on the part of providers31,33
; alternatively, it may reflect a disparity between the preferences of elderly and younger patients for aggressive life-sustaining care.34
The informants in this study, who largely represent the “oldest old,” felt that advanced age was an important contributor to their own treatment preferences. The meaning they assigned to old age reflected concerns over their current health state and future decline, the deaths of significant loved ones, and the personal significance of having lived a “natural life span.” While these concerns may not be important to all elderly individuals facing treatment decisions, they led most of our informants to be less willing to risk substantial suffering or functional loss for the opportunity to live longer.
Some experts have argued that assigning decision-making authority to a proxy is an even more important element of advance planning than is documenting one's advance wishes.35
We found that authority is granted both to family members, based on their concern for patients' interests, and physicians, for their technical expertise. Our findings are consistent with other studies reporting patients' strong desire to have family members as proxies in end-of-life decisions,36,37
even granting them substantial leeway to override patients' own advance treatment wishes.38
They also match the findings from several studies examining authority preferences in cancer treatment decisions, where up to 60% percent of cancer patients preferred that their doctor make the final treatment decisions.39–41
The model for dual authority identified in this study, however, contrasts both with historic models for end-of-life decision making that embraced a paternalistic perspective and contemporary legal standards that do not grant any formal authority for physicians. Our data suggest that patients' goals may be achieved best by a model of collaborative surrogate decision making by families and physicians. Families can represent patients' interests by defining acceptable outcomes and delineating goals of care based on their understanding of patients' values. Physicians can be responsible for providing accurate prognostic information, for outlining a range of achievable goals of care, and for making specific recommendations to use or withhold curative treatments based on mutually defined goals.
Our inquiry also found that expressions of care—patients' concerns about burdening loved ones on the one hand, and caregiver obligations to ill family members on the other—are potent values in settings of poor prognosis. High36,37
has reported extensively on families' filial expectations, and other authors have found that concerns about family burden are a principal reason that patients reject life-sustaining treatments.42
In another recent qualitative study, Singer et al.43
reported that acts of caring—relieving burden and strengthening relationships with loved ones—are critical to patients' perceptions of quality end-of-life care.
The importance of family caring behaviors in settings of poor prognosis has important implications with regard to physician-family interactions in the care of mentally incapacitated patients. In situations where patients' survival is unlikely, physicians can support family authority by focusing decision making on caring behaviors, the family's primary domain. Support can be expressed for example, by seeking the family's perceptions regarding the role that nutrition and hydration might play as an expression of caring. Physicians can also assuage family members' feelings of guilt or abandonment by making explicit the types of comfort-oriented interventions that will be added when decisions are made to discontinue curative treatments.
The decision to use a qualitative approach for this study was made with an understanding of the trade-off between gaining an in-depth understanding of the research questions and the study's generalizability. Despite our efforts to assure the trustworthiness of our findings, the small size, advanced age, and ethnic and racial homogeneity of our study sample are limitations to its generalizability to other patient populations. In particular, the high rate of advance directive completion at both sites (through programs to encourage their use) may reflect that informants are not representative of the general population in whom advance care planning occurs much less frequently. In addition, because informants' personal experiences with end-of-life decisions were retrospectively elicited, it is unclear whether their narratives represent clinical decision making as it actually occurred. Prospective studies conducted in actual decision-making settings may help to confirm our findings. Finally, this study focused on older individuals' perspectives as family members or as patients faced with serious illness. Physician perspectives, which are an integral part of any model for end-of-life decision making, were absent and should be pursued in future work.