Study objectives: To evaluate the usefulness of procalcitonin serum levels as a predictor of etiology and prognosis in adult patients with community-acquired pneumonia (CAP) when they are stratified according to severity.

Design: One-year, population-based, prospective study.

Setting: University teaching hospital.

Patients: All adult patients who received a diagnosis of CAP throughout the study period.

Interventions and measurements: An extensive noninvasive microbiological workup was performed. In patients who gave informed consent, a blood sample was collected at the time the diagnosis of CAP was established to measure biological markers. Procalcitonin levels were measured by a commercially available monoclonal immunoluminometric assay (limit of detection, 0.1 microg/L). Patients were classified according to microbial diagnosis, Patients Outcome Research Team pneumonia severity index (PSI), and outcome measures, and procalcitonin levels were compared among groups.

Results: Of 240 patients who received a diagnosis of CAP during the study period, procalcitonin concentrations were measured in 185 patients (77.1%). Levels were higher in patients with high-severity risk classes (PSI classes III-V) [p = 0.01] and in those with complications (p = 0.03) or death (p < 0.0001). Among patients classified into PSI low-severity risk classes (classes I-II), levels tended to be higher in those with bacterial etiology (p = 0.08); in this group, a serum procalcitonin level > or = 0.15 microg/L was more frequently found in patients with bacterial pneumonia than in those with nonbacterial pneumonia (p = 0.03). In patients with higher-severity risk classes, no significant differences were observed in procalcitonin levels among etiologic groups, but higher concentrations were associated with development of complications (p = 0.01) and death (p < 0.0001).

Conclusions: Procalcitonin contribution to the evaluation of CAP varies according to severity. While procalcitonin may have a role to predict the microbial etiology in patients with a low PSI score, in patients classified within high PSI risk classes, it is a prognostic marker rather than a predictor of etiology.