Prognostication in urgent intensive care unit referrals: a cohort study

Area of Science:

• Critical Care Medicine
• Clinical Prognostication
• Healthcare Outcomes Research

Background:

• Accurate prognostication is vital for clinical decision-making in acutely ill patients.
• Variability in physician predictions can lead to suboptimal care.
• Predicting hospital mortality in deteriorating patients requires careful assessment.

Purpose of the Study:

• To evaluate the accuracy of physician predictions of hospital mortality.
• To assess prognostication in patients referred for urgent intensive care unit (ICU) admission.
• To identify factors influencing prediction accuracy.

Main Methods:

• Prospective cohort study of patients referred for urgent ICU admission.
• Physicians' prognosis assessments (survival with/without disabilities, no survival) recorded.
• Prognosis dichotomized into good or poor for mortality prediction.

Main Results:

• Analysis included 2103 patients with complete data.
• Physician predictions showed poor accuracy (AUC 0.61) for hospital mortality.
• Low sensitivity (31%) in detecting poor prognosis was a key limitation.

Conclusions:

• Physician prognostication is linked to hospital mortality but lacks overall accuracy.
• The primary deficit lies in the low sensitivity for identifying patients at high risk.
• Improving prediction accuracy is crucial for appropriate critical care resource allocation.