Predicting the Truly Sick Child with Fever
Source: Pulliam PN, et al. C-reactive protein in febrile children 1 to 36 months of age with clinically undetectable serious bacterial infection. Pediatrics 2001;108:1275-1279.
Febrile illness is a troublesome problem for children seen in the emergency department (ED). Many have an identifiable source of fever, including some who are thought to have viral syndromes. For the 20% in whom no etiology is apparent, much diagnostic uncertainty remains. Furthermore, febrile children younger than age 3 have serious infections or bacteremia documented more often than do older children. What is the best option for the ED physician? Admit for observation? Culture and treat? Culture and wait? Controversy abounds, and clinical guidelines are not applied uniformly.
In an attempt to reduce the difficulty of identifying young children with occult but serious bacterial infections (SBI), Pulliam and colleagues studied 77 children ages 1-36 months in an academic ED in Delaware. Each had a temperature exceeding 39°C but an undetectable source for fever after clinical evaluation. Each patient had a complete blood count, urinalysis, blood culture, and C-reactive protein (CRP) measured. Chest films (CXR) were done at the discretion of the attending physician.
The average age was 10 months, with a mean temperature of 39.5° C, and an average fever duration of 24 hours. Of the 77 children, 14 (18%) were diagnosed with SBI. Etiologies included urinary tract infection (UTI) in six (all were Escherichia coli), bacteremia in five (all were Streptococcus pneumoniae), and pneumonia by CXR in four (one of whom also had pneumococcal bacteremia). Comparison of laboratory data revealed an average white blood cell count (WBC) of 22,300/mm3 for the SBI group of 14 children, vs. 12,500/mm3 for the non-SBI children (p = 0.003). Absolute neutrophil counts (ANC) were 13,900/mm3 vs. 7300/mm3 for SBI and non-SBI cases, respectively (p < 0.0001). Band counts were 5.7% vs. 3.6% (p = 0.11). CRP concentration in SBI patients averaged 9.7 mg/dL, vs. 1.0 mg/dL for non-SBI children (p = 0.002).
Diagnostic performance of WBC, ANC, and CRP was analyzed by receiver operating characteristic curve to select cutoff values that maximized sensitivity and specificity. A cutoff for WBC of 15,000 had sensitivity of 64% and specificity of 67%, while an ANC cutoff of 10,200 had 71% sensitivity and 76% specificity. For CRP, a cutoff of 7.0 mg/dL was correlated with 79% sensitivity and 91% specificity, with positive and negative predictive values of 65% and 95%, respectively. A CRP value less than 5 mg/dL excluded SBI with 98% negative predictive value. The authors conclude that CRP levels are more definitive than other common tests in predicting SBI in young febrile children.
Commentary by Michael Felz, MD
The sensitivity, specificity, and predictive values of CRP exceeded those for total WBC and ANC in this cohort of 77 children younger than age 3 with high fever. While the use of CRP in the diagnosis of pediatric infections is not new, this study was performed in 2000 during an era when pneumococcal bacteremia in young children remains well-disguised, as was well documented in five of 14 cases. A cutoff CRP value of 7 mg/dL correlated strongly with pneumococcal SBI. Correlation was equally strong with UTI and radiologic pneumonia, also common occurrences in the pediatric ED. Low levels of CRP (< 5 mg/dL) predicted the absence of SBI with strong statistical significance.
I find this data quite persuasive. While no available test approaches 100% reliability, this study convincingly advances CRP as a definitive marker of SBI, outperforming more traditional tests such as WBC, ANC, and band counts—three tests upon which many of us rely heavily for diagnosis. When confronted by a febrile child with no readily apparent source of infection, we must "ride the horns" of several dilemmas in deciding how aggressively to pursue invasive evaluation (i.e., blood culture, bladder catheterization, lumbar puncture) and whether to treat, admit, discharge, or observe. In my opinion, CRP warrants addition to the list of fast-track diagnostic tools in the pediatric ED. It is rapid, easily obtained (by fingerstick), inexpensive, and reliable in documenting (value > 7 mg/dL) or excluding (value < 5 mg/dL) SBI in fever scenarios in the pediatric ED. Predictive values of CRP conceivably could render speculative empiric antibiotic therapy nearly obsolete, especially if WBC and ANC point in the same direction as CRP in predicting/excluding SBI. I welcome this new data on an "old" test.
Dr. Felz, Associate Professor, Department of Family Medicine, Medical College of Georgia, Augusta, is on the Editorial Board of Emergency Medicine Alert.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content