LAUSR

Critical Care Medicine • Volume 46 • Number 1 (Supplement) www.ccmjournal.org Learning Objectives: Rationale: In the absence of appropriately derived clinical decision rules to guide the initial disposition of Emergency Department (ED) patients with pneumonia, severityofillness-scores (i.e. the CURB-65 score) have been potentially misapplied to decision making. Objectives: To assess the predictive performance of the CURB-65 score as it relates to the proximal and actionable end-point of critical care intervention (CCI) for patients admitted to the hospital with pneumonia. Methods: Methods: This was a secondary analysis of electronic health records from a single tertiary care center. Patients admitted to the hospital with a primary diagnosis of pneumonia were included. CURB-65 was calculated and the relationship to receipt of CCIs and in-hospital mortality determined. Results: Measurements and Main Results: 3,851 patients were admitted with pneumonia during the study period, and 1,147 (29.8%) patients were admitted to the ICU within 48-hours of ED triage. 636 (55.4%) of those received a critical care intervention. Of patients with a CURB-65 score of 01, 357 (18.7%) were admitted to the ICU within 48 hours, 181 (9.5%) received a CCI, and 38 (2.0%) died. Among patients with CURB-65 score 2, 790 (40.6%) were admitted to the ICU within 48 hours, 454 (23.4%) received a CCI and 168 (8.6%) died. The Area Under the ROC for CCI and in-hospital mortality were 0.68 and 0.71 respectively. The sensitivity of CURB-65 to predict CCI was 71.4% and was lower than that for mortality, 81.6%. Conclusions: Conclusions: Patients with low CURB-65 scores (0–1) were often admitted to the ICU and commonly received CCIs. This finding, combined with the relatively low sensitivity of CURB-65 for CCI, suggests caution in applying CURB-65 as a decision tool for disposition from the ED. Future decision rules for patients with pneumonia should be calibrated to proximal and actionable endpoints as compared to mortality.