Value of C-reactive protein determination in the initial diagnostic evaluation of the febrile, neutropenic child with cancer.

We studied prospectively the value of administration C-reactive protein (CRP) in the diagnostic evaluation of the child with cancer hospitalized for fever and neutropenia. During a 7-month period 74 patients with malignant disease had 122 hospital admissions because of fever and neutropenia. All patients had a serum CRP obtained 8 to 24 hours after the onset of fever as part of their initial evaluation. There was a borderline correlation between serum CRP concentration and temperature at admission (P = 0.06). Patients with fever without an identifiable source had significantly lower CRP concentrations compared with those having focal or microbiologically documented infection (34.9 +/- 6 vs. 70.2 +/- 12 mg/liter; P = 0.0005). Twelve patients had positive blood cultures, 5 of which were coagulase-negative staphylococci considered to be central venous catheter-related infection or colonization. CRP concentrations were significantly lower in these 5 patients compared with the 7 patients with septicemia caused by other organisms (21 +/- 9 vs. 113 +/- 23 mg/liter; P = 0.01). In distinguishing between septicemic and nonsepticemic children, serum CRP was found to have excellent sensitivity and negative predictive value at concentration limits of 20, 50 and 100 mg/liter. However, both specificity and positive predictive value were low at these respective levels, thus limiting the overall utility of serum CRP in the initial empiric management of the febrile, neutropenic child with cancer.