Value and role of intensive care unit outcome prediction models in end-of-life decision making

Crit Care Clin. 2004 Jul;20(3):345-62, vii-viii. doi: 10.1016/j.ccc.2004.03.002.


In the United States, intensive care unit (ICU) admission at the end of life is commonplace. What is the value and role of ICU mortality prediction models for informing the utility of ICU care?In this article, we review the history, statistical underpinnings,and current deployment of these models in clinical care. We conclude that the use of outcome prediction models to ration care that is unlikely to provide an expected benefit is hampered by imperfect performance, the lack of real-time availability, failure to consider functional outcomes beyond survival, and physician resistance to the use of probabilistic information when death is guaranteed by the decision it informs. Among these barriers, the most important technical deficiency is the lack of automated information systems to provide outcome predictions to decision makers, and the most important research and policy agenda is to understand and address our national ambivalence toward rationing care based on any criterion.

Publication types

  • Review

MeSH terms

  • Advance Care Planning / organization & administration
  • Attitude to Health
  • Conflict, Psychological
  • Critical Care / organization & administration*
  • Decision Support Techniques*
  • Discriminant Analysis
  • Forecasting
  • Health Care Rationing / organization & administration
  • Health Policy
  • Hospital Mortality
  • Humans
  • Outcome Assessment, Health Care / organization & administration*
  • Predictive Value of Tests
  • ROC Curve
  • Reproducibility of Results
  • Risk Assessment / organization & administration
  • Risk Factors
  • Survival Analysis
  • Terminal Care / organization & administration*
  • Total Quality Management / organization & administration
  • United States / epidemiology