C-reactive protein in viral and bacterial respiratory infection in children

C-reactive protein (CRP) was studied in 209 children treated in hospital due to middle or lower respiratory tract infection with serologically demonstrated viral or bacterial aetiology. Of the 110 patients with serological evidence of bacterial infection, either alone or in association with viral infection, 52% had CRP > 20 mg/l, 35% > 40 mg/l and 15% > 80 mg/l. Of the 99 patients with serological evidence of viral infection alone, 35% had CRP > 20 mg/l, but only 12% > 40 mg/l and 5% > 80 mg/l. Nearly all, 88%, of the 25 patients with CRP > 40 mg/l in association with viral infection had either an infectious focus, specific microbial or non-specific laboratory evidence suggestive of bacterial infection. By calculating diagnostic parameters at 3 cut-off levels of CRP, the level 40 mg/l seemed more useful than 20 mg/l or 80 mg/l for differentiation between viral and bacterial infections. By using a CRP value of 40 mg/l as a screening limit sensitivity was 0.55, specificity 0.88, positive predictive value 0.76, negative predictive value 0.55, and likelihood ratios of a positive and negative test result 2.9 and 0.74, respectively. It is concluded that low CRP values do not rule out bacterial aetiology of respiratory infection in children. On the other hand viral infection without bacterial involvement is very improbable if CRP is > 40 mg/l. Our results suggest that high CRP values rule out viral infection as a sole aetiology of infection; bacterial infection and antibiotic treatment should be considered in these cases.