ABSTRACT
Introduction: Community-acquired pneumonia (CAP) presents high mortality rates and high healthcare costs worldwide. C-reactive protein (CRP) has been widely used as a biomarker for the management of CAP. We evaluated the performance of CRP threshold values and ΔCRP as predictors of CAP survival and length of hospital stay. Methods: A total of 173 adult patients with CAP were followed for up to 30 days. We measured serum CRP levels on days 1, 4, and 7 (D1, D4, and D7) of hospitalization, and their variations between different days were calculated (ΔCRP). A multivariate logistic regression model was created with CAP 30-day survival and length of hospital stay as dependent variables, and absolute CRP values and ΔCRP, age, sex, smoking habit (pack-years), pO2/FiO2 ratio on D1, WBC on D1, and CURB-65 score as independent variables. Results: A total of six patients with CAP died (30-day mortality 3.47%). No difference was found in CRP levels and ΔCRP between survivors and non-survivors. Using a cut-off level of 9 mg/dL, the AUC (95% CI) for the prediction of survival of CRP on D4 and D7 were 0.765 (0.538−0.992) and 0.784 (0.580−0.989), respectively. A correlation between CRP values on any day and length of hospital stay was found, with it being stronger for CRPD4 and CRPD7 (p < 0.0001 and p = 0.0024, respectively). A reduction of CRP > 50% from D1 to D4 was associated with 4.11 fewer days of hospitalization (p = 0.0308). Conclusions: CRP levels on D4 and D7, but not ΔCRP, could fairly predict CAP survival. A reduction of CRP > 50% by the fourth day of hospitalization could predict a shorter hospital stay.
