Purpose: The purpose of this study was to prospectively analyze the predictive role of classic predictors for suspected infection (temperature, WBC and derivatives) with two biomarkers, procalcitonin and lactate, on the incidence of culture-proven infection in the surgical intensive care unit (SICU).
Materials and methods: One hundred forty-six consecutive patients admitted for suspected infection had admission and 12-h procalcitonin values, admission and every 6-h lactate values for 24 h, and admission temperature, leukocyte count, lymphocyte count and percentage measured and analyzed in this study.
Results: Peak (highest measured value ≤24-h of admission) procalcitonin values were not predictive for culture-proven infection. However, a culture-negative subset was identified when peak procalcitonin values were < 2.9 ng/mL and when peak lactate values were < 1.3 mmol/L with a probability of 98.3% (P < .001). No other admission predictor was statistically associated with culture-proven infection. Following boosted-tree partitioning, a C-index of 0.85 was calculated with a misclassification rate of 23.3%.
Conclusions: The ability to utilize procalcitonin values in the diagnosis of culture-proven infection was not realized in this study. However, the association of admission peak procalcitonin values with admission peak lactate values identified a group of patients who were culture-negative for suspected infection. No other admission predictor was associated with culture-proven infection.
Keywords: Biomarkers/procalcitonin/lactate; Infection/surgical; Intensive care/surgical patients; Length of stay/surgical intensive care unit/hospital; Misclassification rates; Partitioning/boosted-trees.
Copyright © 2018 Elsevier Inc. All rights reserved.