Objectives: To identify ICU-specific predictors of mortality.
Design: An inception cohort study.
Setting: Barnes Hospital, an academic tertiary care center.
Patients: Consecutive patients, requiring mechanical ventilation, admitted to the medical intensive care unit (ICU) (75 patients), surgical ICU (100 patients), and cardiothoracic ICU (102 patients).
Interventions: Prospective data collection and outcomes evaluation.
Measurements and main results: Stepwise logistic regression analysis identified the following variables to be independent predictors of mortality for the individual ICUs: medical ICU, an Organ System Failure Index (OSFi) greater than or equal to 3; surgical ICU, OSFI greater than or equal to 3; cardiothoracic ICU, OSFI greater than or equal to 3, requiring acute dialysis, and the occurrence of an iatrogenic event. The same analysis was repeated after removing the OSFI as a potential confounding variable. Independent predictors of mortality identified in this subsequent analysis were as follows: medical ICU, occurrence of renal failure; surgical ICU, supine head positioning, acute physiology score greater than or equal to 10, preadmission lifestyle score greater than or equal to 2; cardiothoracic ICU, requiring acute dialysis, occurrence of ventilator-associated pneumonia, and the occurrence of an iatrogenic event.
Conclusions: We identified the presence of ICU-specific predictors of mortality amongst the three ICUs examined. These data suggest that ICU-specific interventions could be developed to improve the quality of patient care and potentially to reduce patient mortality.