Despite growing support for advance directives, there are few data validating their utility. We conducted this study to determine if the type of advance directive influences physicians' willingness to withhold specific therapies and if physicians are equally willing to withhold these therapies.
The 444 full-time faculty of the department of medicine of a university medical center were presented with patient scenarios and accompanying advance directives in three separate surveys that were mailed. They were asked if they would withhold each of 12 specific therapies based on their interpretation of the advance directive. Three types of advance directives were studied: general statement, therapy specific, and therapy specific combined with designation of a proxy and prior patient-physician discussion.
The mean proportion of all 12 therapies that were withheld varied by type of advance directive: general-statement advance directive, 55%; therapy-specific advance directive, 71%; and therapy-specific advance directive with proxy and prior patient-physician discussion, 83%. Respondents were more likely to withhold cardiopulmonary resuscitation when given a therapy-specific advance directive vs general-statement advance directive, 84% vs 73%, respectively. With a therapyspecific advance directive that was supported by a proxy and prior patient-physician discussion, 100% of physicians were willing to withhold cardiopulmonary resuscitation. With the therapy-specific advance directive that was supported by proxy and prior patient-physician discussion, physicians were not equally willing to withhold all therapies, ie, mechanically assisted ventilation, 98%; intravenous fluids, 82%; antibiotics, 80%; simple tests, 70%; and pain medications, 13%.
Detailed advance directives with a supportive proxy, coupled with physician-patient discussion, furnish the most reliable medical directives. Even with such directives, physicians are more likely to withhold life-saving therapies than simple tests, treatments, and pain medications.(Arch Intern Med. 1993;153:375-381)