These letters represent a wide range of views about the usefulness of advance directives. Dr. Simcic's skepticism is similar to mine. Dr. Carter gives advance directives modest support, using them for now but hoping for better approaches in the future. Drs. Brasic and Hammes believe strongly that the current system of advance directives prevents overly aggressive treatment of dying patients, saves resources, and promotes comfort.

The view of Drs. Brasic and Hammes, whose end-of-life work I admire, differs most from my own. I believe their enthusiasm for advance directives rests on 2 faulty assumptions. The first is that future illness is sufficiently predictable to permit useful advance instructions about care. Two studies indicate otherwise (1, 2). One study found that, despite receiving detailed patient-specific prognoses, physicians typically wrote do-not-resuscitate orders no sooner than 2 days before patients died, suggesting that "hopelessness" becomes apparent only late in a patient's course (1). The other study involved frequent functional assessments of elderly patients before they died (2). Functional decline followed 5 patterns, each accounting for about 20% of deaths: treatable but eventually terminal disease, such as cancer (slow decline over months, then a rapid decline to death); organ system failure, such as heart failure (waxing and waning function with frequent rescues until an eventual, unpredictable death); chronic frailty, such as dementia (low and slowly diminishing function with an unpredictable death); sudden death; and "unclassifiable" (2).

Today's patients, who often are suffering from multiple diseases simultaneously, might follow any of several possible patterns. The attendant unpredictability severely limits the usefulness of advance directives in planning care.

The second faulty assumption is that people behave logically and consistently. As my own studies suggest, few patients who believe that advance directives can promote their care wishes actually sign them (3). Furthermore, many signers do not grasp the implications of their advance directives (which may, therefore, inaccurately reflect their wishes in the event of a crisis), many proxies do not understand patients' wishes (4) or act on them faithfully, and many physicians do not incorporate those wishes into treatment.

I used to believe strongly in the predictive powers of medicine and in the rationality of patients, so I advocated vigorously for advance directives. Yet repeated disappointments have tempered my enthusiasm (5). I now believe that advance directives have only a limited role in prompting and structuring advance care planning and that their effect must not be exaggerated. I also believe there must be a better way to prepare patients, families, and health professionals for the arduous, unpredictable trials of end-of-life care. Health professionals must take the lead in finding that better way.