Monday, March 13, 2006
Two articles from the latest Archives of Internal Medicine:
First is a systematic review of the accuracy of surrogate decision makers. The authors looked at 16 trials of patients and their surrogate decision makers; all trials looked at patients' and surrogates' responses to hypothetical situations. Agreement between them regarding medical decisions in a variety of situations was 68% overall, which was generally consistent across several studied scenarios (ranging ~58%-70% agreement). There was no improvement in this when comparing designated decision makers vs. 'default' ones (that is--who would be the designated decision maker by state law where applicable). The 2 'intervention' trials--all of which had some patients talk with proxies in an uncontrolled fashion prior to being surveyed--showed essentially no difference from the intervention (one showed in fact a tiny decrement in agreement with discussion). They didn't really look at this systematically but most trials found surrogates wanted more interventions than patients, although many showed no difference and a few showed the opposite.
So what to make of this? First off, none of this is new and the trials were quite heterogeneous, although it is interesting how this 2/3 agreement finding is consistent across the trials. I'm not sure whether I think 2/3 is great (that it's so high) or terrible (that it's so low). Clearly it's not going to be 100% but one I guess would hope for ~90. The authors use language in the conclusion stating that prior discussions don't improve surrogate decision making. While that's true in that it has not been shown, they are basing this off 2 articles in which the discussions were uncontrolled, suggesting that there's a lot more room for research on this matter. All of this, to me, gets to the problems of advanced directives and the attempt by some to use them as ways of specifically directing medical care. It is hard enough sitting at the beside or in a family conference discussing complicated medical decision making with patients/families--helping them understand the issues, the decisions to be made and their consequences etc.--that I can't imagine trying to express that in a written document without any knowledge of what decisions one would ever be faced with. Do most people really 'know' if they'd want to be intubated or not? Or if they'd agree to going to a long-term care facility (my state's form has a box you can check saying your surrogate can't admit you to a long-term care facility)? A few do, and things certainly get clearer the closer one is to death, but most don't, and advanced directives should be made when one is relatively well. AD's to me are best as statements of values (what is important to patients in the setting of a life threatening illness, what is of value in their life, and perhaps what they don't find valuable, such as a beating heart without a functioning brain) and permissions (explicitly giving permission for x, y, and z which is usually some variation of 'it's ok to stop burdensome or life-prolonging treatments in the setting of neurologic devastation, or a progressive life-threatening disease that's not likely to get better, etc.'). The permissions bit is important particularly to families feeling burdened with what seems to them as life & death decisions being plunked at their feet. More explicit than this seems inappropriate for most advanced directives, although there are exceptions. All of this is to say that it is unrealistic to expect that patients talking with families is going to be anything more than a discussion of values and permissions. Indeed it can't be more than that--it can't be "If X & Y happen to me and the doctor say P & Q I would want you to do Z." Medical decisions are messy, and part of what I'm wondering is if proxy decision making can't get cleaner than 2/3 accurate for the above reasons. I'd welcome a trial that proved me wrong, though!
More along the lines of hypothetical decision making.
The second article is about physician willingness to withhold or withdraw life-sustaining treatments. It is from a survey of 400 internists in the US (this represented a 41% response rate for the survey; most were in private practice; a mix of general and subspecialists were surveyed). This was one of those create several scenarios, vary the details deliberately, and see how people respond differently type of studies. The details that were mixed up were decision making capacity of the patient, whether they were terminally ill or quadriplegic, whether the scenario involved withholding or withdrawing a treatment, and the treatment in question (dialysis, ventilator, artificial nutrition/hydration, antibiotics); all scenarios involved explicit requests by patients or proxies to stop or withhold a treatment, and it was made clear the patients weren't depressed etc. The results were not surprising, but somewhat discouraging. The surveyed physicians were less willing to assent to the patients'/proxies' requests if it involved withdrawing vs. withholding a treatment, if it involved antibiotics or artificial nutrition/hydration vs. the vent or dialysis, and if the patients were alert vs. demented or terminally ill. Prepare to be assaulted by your internist if you're quadriplegic and need dialysis but don't want it! That's a little flip, and I've written for too long tonight already, but I'll close by emphasizing that this study was not of what the docs thought was best, or what they'd recommend, but what they'd do if a patient said "No."