Drs. Webster, Grossman, and GuntherMaher each help to explain the results of our study by placing them within the social context of U.S. medicine. They suggest that constraints related to the societal concepts of "the physician," "the patient," and "the health care system" itself may lead physicians away from frank disclosure of prognoses to patients with cancer at the end of life. Regarding the physician, Dr. Webster suggests that until physicians stop viewing death as preventable and thus a professional failure, their efforts to cultivate good deaths for their patients will be scant (1). Regarding the patient, Dr. Grossman suggests that patients too view death as preventable and a failure and develop cognitive and emotional filters that help them to deny poor prognoses from physicians and to yearn for further futile anticancer therapy (2). Finally, Dr. GuntherMaher suggests that the "harried" and "fractionated" health care system favors nondisclosure and optimistic disclosure over frank disclosure because these styles simply take less time. In sum, nonfrank disclosure of prognoses at the end of life is easier, quicker, and cheaper for physicians, patients, and the health care system.

In the current context of U.S. medicine, undue optimism regarding prognosis at the end of life may be a Pareto optimal solution. This suggests that with respect to physicians' prognostic behavior at the end of life, change will occur only as the societal constraints related to U.S. medicine are relaxed. We suspect that as Americans—patients, physicians, and health care administrations—come to understand that death is inevitable and that palliation is both kinder and perhaps less expensive than futile anticancer therapy, the prognoses physicians give their dying patients will more often be frank and efforts to cultivate a good death will be more common.