Systematic review of the diagnostic accuracy of C-reactive protein to detect bacterial infection in nonhospitalized infants and children with fever.

Abstract

To determine the accuracy of C-reactive protein (CRP) for diagnosing serious bacterial and bacterial infections in infants and children presenting with fever. Systematic review of diagnostic accuracy studies. We included studies comparing the diagnostic accuracy of CRP with microbiologic confirmation of (a) serious bacterial and (b) bacterial infection. For differentiating between serious bacterial infection and benign or nonbacterial infection (6 studies), the pooled estimate of sensitivity was 0.77 (95% CI, 0.68, 0.83); specificity, 0.79 (95% CI, 0.74, 0.83); positive likelihood ratio, 3.64 (95% CI, 2.99, 4.43); and negative likelihood ratio, 0.29 (95% CI, 0.22, 0.40). In multivariate analysis, CRP is an independent predictor of serious bacterial infection. 3 studies investigating the accuracy of CRP for diagnosing bacterial infection could not be pooled, but all showed a lower sensitivity compared with studies using serious bacterial infection as the reference diagnosis. CRP provides moderate and independent information for both ruling in and ruling out serious bacterial infection in children with fever at first presentation. Poor sensitivity means that CRP cannot be used to exclude all bacterial infection.