UMHS LOGOUniversity of Michigan
Department of Pediatrics

Evidence-Based Pediatrics Web Site

CRP Level Is Helpful in Determining if Febrile Children Without Source Have a Serious Bacterial Infection


  • In children less than 36 months with fever and no obvious source, how accurate is CRP as compared to total WBC, bands, and ANC in detecting serious bacterial infections (SBI) that need antibiotics?

Clinical Bottom Lines

  1. Quantitative CRP measurements can predict which febrile children are at risk for serious bacterial infections.
  2. A CRP of 7 (LR=8.3) has a better accuracy than using WBC of 15 or ANC of 10.2.

Summary of Key Evidence

  1. Prospective study done in Delaware at a pediatric ED from January to October.1
  2. All patients had WBC, band count, ANC, quantitative CRP, blood culture, and urine obtained.
  3. 77 children from age 1 to 36 months with T>39 and no obvious source were enrolled.
  4. 18% (N=14) of the patients had SBI. UTI was the most common (6 patients).
  5. CRP level of <5 had a likelihood ratio (LR) of 0.09 for SBI (NPV of 98%); thus, this could be useful in ruling out SBI.
  6. CRP level of >9 had a LR of 9.0 (PPV of 67%) for SBI.
  7. Using ROC analyses, a CRP of 7 has a LR of 8.3, a WBC of 15 has a LR of 1.9, and an ANC of 10.2 has a LR of 3.0.

Additional Comments

  • More studies needed to determine the optimal cutoff value for CRP. Also need to find how duration of fever affects the value and age of the patient.
  • The CRP seems to be more helpful if the fever has been present for >24 hours. Likely takes some time for the body to mount a measurable response (3 patients had SBI and CRP <7 and all had fever of <9 hours duration).
  • Limitations of the study: results may not be generalizable to all ED's that do not specialize in pediatrics and it was a small study overall. Well-appearing children that may have had up and down fevers may have been discharged from the ED without any laboratory workup. Also, the study was done at large referral center, so a lot of the otherwise well-appearing children may not have been sent in, only the children that looked most ill (possible sampling bias).
  • Can measure CRP from capillary sample and the test can be done quickly and it is inexpensive; however, it is not clear if this study used venous or capillary samples.
  • More recent studies are now examining the role of CRP combined with procalcitonin and other inflammatory makers to predict SBI.2


  1. Pulliam PN, Magdy W, Cronan KM. C-Reactive Protein in Febrile Children 1 to 36 months of Age With Clinically Undetectable Serious Bacterial Infection. Pediatrics 2001;108:1275-9.
  2. Galetto-Lacour A, Zamora S, Gervaix A. Bedside Procalcitonin and C-Reactive Protein Tests in Children with Fever Without Localizing Sings of Infection Seen in a Referral Center. Pediatrics 2003;112:1054-60.

CAT Author: Pat Gordon, MD

CAT Appraisers: Alex Kemper, MD

Date appraised: November 2, 2005

Last updated November 27, 2005
Department of Pediatrics and Communicable Diseases
© 1998-2002 University of Michigan Health System