C-reactive protein concentration as a predictor of in-hospital mortality after ICU discharge : a nested case-control study

PURPOSE: To assess the ability of potential clinical predictors and inflammatory markers to predict in-hospital mortality after patient discharge from the intensive care unit. SETTING AND PARTICIPANTS: 1272 patients who survived their index admission to a 22-bed multidisciplinary ICU of a university hospital in 2004. DESIGN: Nested case-control study with two concurrent control patients for each case of post-ICU discharge in-hospital mortality. RESULTS: There were 29 unexpected in-hospital deaths after ICU discharge (2.3%). C-reactive protein (CRP) concentrations within 24 hours of ICU discharge were available for 14 of these 29 patients and 22 concurrent control patients. CRP concentration at ICU discharge was associated with subsequent mortality (mean CRP concentrations: cases, 204 mg/L v controls, 63 mg/L; P=0.001). CRP concentration remained significantly associated with post-ICU mortality after adjustment with other potential predictors of mortality (odds ratio [OR] of death for a 10 mg/L increase in CRP concentration, 1.27; 95% CI, 1.09-1.49; P=0.005) and with propensity score (OR, 1.19; 95% CI, 1.05-1.33; P=0.004). The area under the receiver operating characteristic curve for CRP concentrations to predict in-hospital mortality was 0.87 (95% CI, 0.73- 0.99; P=0.001). The destination and timing of ICU discharge, SOFA (Sequential Organ Failure Assessment) score, white cell count and fibrinogen concentration at ICU discharge were not significantly associated with in-hospital mortality after ICU discharge. CONCLUSIONS: A high CRP concentration at ICU discharge is an independent predictor of subsequent in-hospital mortality. Prospective cohort studies in ICUs with different casemix, discharge criteria and post-ICU mortality rates are needed to validate and generalise our findings.