Pneumonia is one of the leading causes of death in the intensive care unit (ICU). Many biomarkers for predicted prognosis have been suggested; among these, procalcitonin (PCT) is known to increase in cases of bacterial infection. However, there have been many debates regarding whether PCT is an appropriate prognostic marker for pneumonia. Therefore, we investigated whether PCT can serve as a biomarker for pneumonia, and compared it with CURB-65, which is a known tool for predicting the prognosis of pneumonia.

Levels of PCT and CURB-65 scores were compared between 30-day non-survival (n = 30) and survival (n = 101) patients. Relationships between PCT and CURB-65 were determined by using linear regression analysis, as well as by using receiver operating characteristic (ROC) curve analysis and calculation of the area under the curve (AUC). High and low PCT groups were compared.

High PCT and high CURB-65 score were positively associated with 30-day mortality. For the prediction of 30-day mortality, initial PCT and CURB-65 exhibited AUCs of 0.63 and 0.66; these were not significantly different (P = 0.132). We found that the high PCT group had a higher rate of initial treatment failure (91%, P = 0.004).

Initial PCT can be a prognostic biomarker for mortality in severe pneumonia, similar to the CURB-65 score. Initial high PCT was positively associated with initial treatment failure.