A composite score combining procalcitonin, C-reactive protein and temperature has a high positive predictive value for the diagnosis of intensive care-acquired infections

Laurent Robriquet, Caroline Séjourné, Eric Kipnis, Michele D’herbomez, François Fourrier

BACKGROUND: Nosocomial infection diagnosis in the intensive care unit (ICU) remains a challenge. We compared routine measurements of procalcitonin (PCT), C-reactive protein (CRP), white blood cell count (WBC) and temperature in the detection of ICU-acquired infections. METHOD: Prospective observational cohort study in a University hospital Medicosurgical ICU. All patients admitted to the ICU >= 5 days (n = 141) were included into two groups, either infected (documented infection, n = 25) or non-infected (discharged from the ICU without diagnosis of infection, n = 88). RESULTS: PCT, CRP, WBC and temperature progression from day -4 (D-4) to day 0 (D0) (day of infection diagnosis or ICU discharge) was analysed. Differences (Delta) were calculated as D0 levels minus the lowest preceding value. D0 PCT and CRP were significantly increased in infected compared to non-infected patients (median, 1st and 3rd quartiles): 3.6 ng/mL (0.92-25) for PCT, 173 mg/L (126--188) for CRP versus 0.02 ng/mL (0.1-0.9) and 57 mg/mL (31--105) respectively (p < 0.0001). In multivariate analysis, D0 temperature > 38.6[degree sign]C, PCT > 1.86 ng/mL, and CRP > 88 mg/L, performed well (AUCs of 0.88, 0.84, and 0.88 respectively). The sensitivity/specificity profiles of each marker (76%/94% for temperature, 68%/91% for PCT, and 92%/70% for CRP) led to a composite score (0.068 x D0 PCT + 0.005 x D0 CRP + 0.7 x temperature) more highly specific than each component (AUC of 0.90 and sensitivity/specificity of 80%/97%). CONCLUSION: Combining CRP, PCT and temperature is an approach which may increase of nosocomial infection detection in the ICU.