Aim: To compare diagnostic accuracy of procalcitonin for early diagnosis of serious bacterial infection (SBI) in children presenting with fever and no focus of infection.
Methods: Prospective, observational study involving 72 children (1-36 mo) presenting to the paediatric units of two university hospitals. All children had blood cultures, urine cultures, white blood cell counts (WBC), chest X-ray, C-reactive protein (CRP) and procalcitonin (PCT) done at presentation.
Results: Eight (11.1%) children had SBI (1 pneumonia, 2 meningitis, 4 septicaemia/occult bacteraemia, 2 pyelonephritis), 19 (26.4%) had possible bacterial infection (received antibiotic treatment, but no organism grown) and 45 (62.5%) had viral or possible viral infection (virus isolated and/or uneventful recovery without antibiotics). PCT (>2 ng/l), CRP (>50 mg/l) and McCarthy's score (<9) had sensitivities and specificities of 50%/85.9%, 75%/68.7% and 87.5%/67.2%, respectively. Negative and positive likelihood ratios for CRP (>50 mg/l), PCT (>2 ng/l), white blood cells (>15 x 10(5)/l) and McCarthy's score (<9) were 0.36/2.4, 0.58/3.5, 0.94/1.1 and 0.19/2.7, respectively. A combination of PCT, CRP and WBC generated a positive likelihood ratio of 10.6, changing the post-test probability to 54%.
Conclusion: For early diagnosis of SBI in children presenting with fever and no focus of infection, the diagnostic utility of procalcitonin is similar to the traditional markers infection and clinical scoring. While a low procalcitonin level cannot be used to exclude SBI in this population, a combination of PCT, CRP and WBC may be more useful in predicting SBI.