comes either to eliminate those cases or to rely on information given by a surrogate, usually a family member. The tools developed for SUPPORT reflect survey methodologies of the early 1980s, which had important limitations—including lowered patient expectations and subsequent high satisfaction with the quality of care. For example, Desbiens and colleagues (1996) reported that persons were satisfied with pain management despite reporting severe pain more than one-half the time.
Responding to the need to develop tools to measure quality of life and quality of care at the end of life, the Brown University Center for Gerontology and Health Care Research and the IOM have convened a series of multidisciplinary conferences (Teno et al., 1999). The result has been a series of recommendations for a “Toolkit of Instruments to Measure End of Life Care,” with the initial target of developing tools that measure the perspectives of the dying person and the family for the purposes of research and quality improvement.
Since medical decisions increasingly are based on quality of life and quality of life is a subjective concept, cancer patients must be allowed their desired role in decisionmaking. Medical records can document treatments received and whether physicians state that they discussed treatment decisions with patients and/or their families. Even though this can be useful information, a consumer perspective on communication, decisionmaking, coordination, and other domains is important when assessing the quality of care of the dying. Ultimately, it is not documentation of the event, but whether the information was provided in a way that the cancer patient could understand and use in making decisions that should be the ultimate judge of the quality of care.
Typically, “satisfaction measures” have been relied on for the consumer perspective on the quality of health care (Table 3-4). In these cases, consumers are asked to rate the quality of care using scales ranging from either “excellent” to “poor” or “very satisfied” to “very dissatisfied.” Typically, respondents must go through a cognitive process in which they first ask whether a particular event occurred, formulate their expectations regarding that event, and then rate that event using the provided response scale. Unfortunately, expectations are usually low, causing respondents to express high satisfaction with care that is less than optimal.
Newer methods have begun using either “patient-centered reports” (Cleary and McNeil, 1988) or “preference-based questions” (i.e., unmet needs) to capture consumer perspectives. These methodologies, unlike typical satisfaction questions that rely on ratings, provide information that can guide improvement of the quality of care. For example, knowing that 85 percent of patients believe a health care provider is “very good” does not supply that provider with information on how to improve. On the other hand, knowing that 20 percent of patients did not understand a provider’s