Objective: To compare the accuracy of procalcitonin and C-reactive protein as diagnostic markers of bacterial infection in critically ill children at the onset of systemic inflammatory response syndrome (SIRS).
Design: Prospective cohort study.
Setting: Tertiary care, university-affiliated pediatric intensive care unit (PICU).
Patients: Consecutive patients with SIRS.
Interventions: From June to December 2002, all PICU patients were screened daily to include cases of SIRS. At inclusion (onset of SIRS), procalcitonin and C-reactive protein levels as well as an array of cultures were obtained. Diagnosis of bacterial infection was made a posteriori by an adjudicating process (consensus of experts unaware of the results of procalcitonin and C-reactive protein). Baseline and daily data on severity of illness, organ dysfunction, and outcome were collected.
Measurements and main results: Sixty-four patients were included in the study and were a posteriori divided into the following groups: bacterial SIRS (n = 25) and nonbacterial SIRS (n = 39). Procalcitonin levels were significantly higher in patients with bacterial infection compared with patients without bacterial infection (p = .01). The area under the receiver operating characteristic curve for procalcitonin was greater than that for C-reactive protein (0.71 vs. 0.65, respectively). A positive procalcitonin level (>or=2.5 ng/mL), when added to bedside clinical judgment, increased the likelihood of bacterial infection from 39% to 92%, while a negative C-reactive protein level (<40 mg/L) decreased the probability of bacterial infection from 39% to 2%.
Conclusions: Procalcitonin is better than C-reactive protein for differentiating bacterial from nonbacterial SIRS in critically ill children, although the accuracy of both tests is moderate. Diagnostic accuracy could be enhanced by combining these tests with bedside clinical judgment.