We agree with Dr. Leff that our study does not prove that improving the quality of advance directive discussions will be sufficient to increase their effectiveness. Other factors, ranging from problems with the portability of directives to insufficient institutional incentives for implementing directives, may well be contributing to their lack of effectiveness. However, when only 25% of discussions mention the use of artificial nutrition and hydration or only 55% mention uncertainty, it is not a mere leap of faith to suggest that many of these discussions will not aid physicians faced with such decisions in the future. Furthermore, immediately after these discussions, we presented physicians and patients with 20 identical hypothetical scenarios and in 90% of cases found poor agreement ( = –0.04 to 0.31) between physicians and patients about the patients' treatment preferences (1).

Physicians should be available for "sensitive counseling" when complex decisions arise. If our findings regarding advance directives are any indication, however, there is reason to question whether physicians are having such conversations. Not only did physicians ignore values, they rarely explored the reasons behind patients' stated preferences. Dr. Pascoe questions whether physician communication skills can be improved with teaching and states that, in any case, insufficient time exists to allow for adequate discussion. However, increasing evidence suggests that communication skills are no different from any other aspect of medical training and can be improved through proper education (2-4). We agree that good discussions certainly take time, and we support those who call for reimbursement of this activity (5).

Providing sufficient time, however, will not automatically result in better communication. In fact, high-quality, useful discussions can occur in a reasonable amount of time if the time is spent well. To be useful, these discussions need not cover every possible situation. Specifically, we encourage physicians to emphasize values, goals, and certain sentinel, important scenarios, such as disability states and uncertainty, and spend less time on the "dire" and "reversible" situations about which nearly all patients agree. Finally, in the ambulatory setting, physicians can have these discussions over several visits.