Objectives are unreliable on life support withdrawal

Study offers no clear guidance on predicting costs

Objective methods to determine whether ICU patients whose outlook is futile should be kept alive and worth the cost in hospital life support efforts may be misleading and should not be used by nurses or physicians as a guide, according to researchers at the University of Vermont School of Medicine in Burlington.

In fact, existing objective measures for evaluating the cost-effectiveness of withdrawing life support are unreliable. Physicians may never have an objective means of making decisions to withdraw life-support from patients whose outcomes are deemed terminal during the hospitalization.1

Furthermore, even when accepted survival scoring systems such as the widely accepted APACHE III (Acute Physiology and Chronic Health Evaluation III) are used, the cost savings are likely to be quite small. Part of the reason is that decisions to withdraw life support involve a relatively small group of patients.

Another factor is that the results of systems such as APACHE III when used in calculating survival in terms of cost savings don’t deviate much from results obtained when physicians use individual clinical judgments, according to researchers.

With the number of ICUs commonly grappling with whether to withdraw life support once physicians decide that further medical care is futile, clinicians have tried to rely on objective measures to justify their decisions.

"If it is reasonable to withdraw support from patients who are extremely unlikely to benefit from ICU care, objective means of identifying patients receiving medically futile care should be useful," according to Laurent G. Glance, MD, a University of Vermont anesthesiologist and study co-author.

In a wide-ranging retrospective study of more than 4,000 non-cardiac patients, a prognostic scoring system to predict the cost-effectiveness of withdrawing life support from relevant patients did not prove significantly valuable as an objective measure.

The study involved a nine-year review of patients at a surgical ICU who had a probability of death of greater than 90% within 48 hours of admission. The study used a mortality risk estimate taken from APACHE III scores.

Investigators constructed a model to compare the cost-effectiveness of two clinical strategies. One involved patients who were discharged, died, or had life support withdrawn based on subjective criteria. The second involved patients who were discharged, died, or had life-support withdrawn based on subjective criteria but also were predicted to have a greater than 90% risk of mortality within 48 hours using a predictive scoring system.

"The use of scoring systems to assist in the decision to discontinue critical care is extremely controversial," Glance and his colleagues wrote, "Although prognostic scoring systems would be expected to have advantages over clinical judgment, the explanatory power of APACHE III . . . is only slightly better than physician judgment."


1. Glance LG, Osler T, Shinozaki T. Intensive care unit prognostic scoring systems to predict death: A cost-effectiveness analysis. Crit Care Med 1998; 26:1,842-1,849.