The Limits of Prognostication - Hampered Outcome Prediction Accuracy Warrants Caution

Claudia Goettler and colleagues have a great article in the new Journal of Trauma. They review previous studies showing that while scoring systems like APACHE are highly accurate, they still have a high false positive rate.  That is, they predict that patients to die who actually survive.  Scoring systems work well to predict outcome in patient populations, but are less useful for driving individual patient care.  That leaves physicians with the task of judgment and prognostication.

In their study Goettler and colleagues found that the "educated guess" is better at outcome prediction for the individual patient than SOFA, APACHE, APACHE II, or TRISS scoring.  Still, they "recommend caution. . . .  Large outcome studies of many specialized patient populations are required to determine the patients who actually have a survival likelihood and are receiving futile care. . . .  Clearly, we all want to save as many patients as possible but we do not want to prolong the dying process in those who cannot be saved."

Interestingly, in a roundtable discussion appendix to the article, Dr. Goettler (from Greenville, North Carolina) notes that "in our region we are essentially in the buckle of the Bible belt and withdrawal of support in our patient population is extraordinarily uncommon even in patients who have been given a very, very poor prognosis."