Lab tests or clinical assessment for detecting serious bacterial infection?

To help establish the utility of surrogate markers of serious bacterial infection (SBI) in the post-pneumococcal vaccination era, investigators compared the diagnostic properties of these markers and clinical evaluation for detecting SBIs. Included in the study were 328 children aged 1 to 36 months with fever higher than 38^o C and no identified source of infection who visited a pediatric emergency department (ED) and who required a standard workup because of either appearing ill or the fever's duration.

Blood tests included a complete blood count, semiquantitative procalcitonin (PCT), C-reactive protein (CRP), total white blood cells (WBC), and the absolute neutrophil count (ANC). A blood culture and urine analysis and culture also were performed. Before test results were available, attending physicians performed the clinical evaluation part of the study by estimating the probability of an SBI on the basis of the child's history and physical examination.

A total of 54 (16%) children were diagnosed with SBI, of whom 48 (89%) had urinary tract infections (UTIs). Four children (7%) had pneumonia, 1 (2%) child had meningococcal meningitis, and 1 (2%) child had an occult Streptococcus pneumoniae bacteremia. Using the areas under the curves (AUCs) of the receiver operating characteristic (ROC) curves, investigators determined that all the surrogate markers of SBI were superior to clinical evaluation for making the diagnosis. Whereas the clinical evaluation had an AUC of only 0.59, the AUCs for the surrogate markers were not only higher but also similar: PCT, 0.82; CRP, 0.88; WBC, 0.81; and ANC, 0.80 (Manzano S, et al. Arch Dis Child. 2011;96[5]:440-446).

Receiver operating curves are generated by plotting the true-positive rate (sensitivity) against the false-positive rate (100% specificity). Plotting results for multiple tests on the same graph allows the accuracy of the tests to be compared; the test with the largest AUC is considered the most accurate.

In this study, the history and physical exam were less accurate than in other studies in detecting SBI in children younger than 3 years, a finding that hasn't changed with the decreasing rate of S pneumoniae disease in vaccinated communities. What has changed is the increased proportion of SBIs found that are UTIs (nearly 90%) versus bacteremia, meningitis, and pneumonia. The authors suggest that this trend increases the value of urinalysis and urine culture while decreasing the value of blood culture, except in the youngest children and in those who are incompletely immunized.

PCT, CRP, WBC, and ANC all had similar accuracy in predicting SBI. The researchers suggest best cutoff values for each test using the ROC curves, but they emphasize that degree of abnormality matters. So, a high PCT is concerning, but a very high PCT should be more concerning. The ideal would be a system that predicts probability of SBI based on all of the markers working in concert. -Michael Burke, MD