Post-operative hypoalbuminaemia and procalcitonin elevation for prediction of outcome in cardiopulmonary bypass surgery.

Abstract

Because few studies have addressed postoperative hypoalbuminaemia in relation to hospital mortality, we evaluated this association and the prognostic value of increased procalcitonin (PCT) after cardiopulmonary bypass (CPB) surgery. In 454 consecutive patients undergoing CPB, minimal serum albumin, colloid osmotic pressure (COP) and maximal PCT were retrospectively obtained from the 2nd to 10th postoperative day. Receiver operating characteristic (ROC) and multiple regression analyses determined independent predictive strength for 28-day mortality from preoperative albumin, Euroscore, postoperative minimal albumin and COP, and maximal PCT. Cut-off points for the four strongest predictors were calculated by the area under the curve (AUC) in the ROC for the 28-day mortality. Maximal PCT showed the largest AUC (0.85; 95% CI 0.79-0.90) and the highest relative risk (RR 12.17; 95%CI 5.26-28.16; P < 0.001), compared with postoperative albumin (AUC 0.72; 95% CI 0.62-0.81; RR 5.35; 95%CI 2.99-9.56; P < 0.001) and EuroSCORE (AUC 0.73; 95%CI 0.63-0.83; RR 4.48; 95%CI: 1.78-11.28; P < 0.01). By logistic regression, postoperative albumin was the strongest predictor of mortality (odds ratio 0.86; 95% CI 0.84-0.89). Cut-off values for predicting 28-day mortality were found for postoperative albumin and PCT at 17.8 g l(-1) and 2.5 ng l(-1), respectively. A slight but significant inverse correlation between PCT and albumin was found. Patients with albumin less than the cut-off showed significantly higher median values for PCT levels (2.5 vs. 1.0 g l-1), a higher 28-day mortality rate (20.8% vs. 4.5%), and a longer ICU stay (6 vs. 3 days) in comparison with patients with minimal albumin greater than 18 g l(-1). Post-operative serum albumin <18 g l(-1) and PCT >2.5 ng l(-1) are predictive for a higher 28-day mortality rate in cardiosurgical patients. Both peak PCT and minimal albumin were better outcome predictors than the Euroscore, which better represents the preoperative condition of the patient.