Gaetano Iapichino, Davide Corbella, Cosetta Minelli, Gary H. Mills, Antonio Artigas, David L. Edbooke, Angelo Pezzi, Jozef Kesecioglu, Nicolò Patroniti, Mario Baras, Charles L. Sprung

To identify factors influencing triage decisions and investigate whether admission to the intensive care unit (ICU) could reduce mortality compared with treatment on the ward. A multicentre cohort study in 11 university hospitals from seven countries, evaluating triage decisions and outcomes of patients referred for admission to ICU who were either accepted, or refused and treated on the ward. Confounding in the estimation of the effect of ICU admission on mortality was controlled by use of a propensity score approach, which adjusted for the probability of being admitted. Variability across centres was accounted for in both analyses of factors influencing ICU admission and effect of ICU admission on mortality. Eligible were 8,616 triages in 7,877 patients referred for ICU admission. Variables positively associated with probability of being admitted to ICU included: ventilators in ward; bed availability; Karnofsky score; absence of comorbidity; presence of haematological malignancy; emergency surgery and elective surgery (versus medical treatment); trauma, vascular involvement, liver involvement; acute physiologic score II; ICU treatment (versus ICU observation). Multiple triages during patient's hospital stay and age were negatively associated with ICU admission. The area under the receiver operating characteristic (ROC) curve of the model was 0.83 [95% confidence interval (CI): 0.81-0.84], with Hosmer-Lemeshow test P = 0.300. ICU admission was associated with a statistically significant reduction of both 28-day mortality [odds ratio (OR): 0.73; 95% CI: 0.62-0.87] and 90-day mortality (0.79; 0.66-0.93). The benefit of ICU admission increased substantially in patients with greater severity of illness. We suggest that intensivists take great care to avoid ICU admission of patients judged not severe enough for ICU or with low performance status, and they tend to admit surgical patients more readily than medical patients. Interestingly, they do not judge age per se as a reason for refusal of ICU admission. Admission to ICU was associated with a reduction of both 28- and 90-day mortality, particularly in patients with greater severity of illness at time of triage.