Background: The diagnosis of acute bacterial meningitis in children is difficult in sub-Saharan Africa, because the clinical features overlap with those of other common diseases, and laboratory facilities are inadequate in many areas. We have assessed the value of non-laboratory tests and incomplete laboratory data in diagnosing childhood acute bacterial meningitis in this setting.
Methods: We prospectively studied 905 children undergoing lumbar puncture at a rural district hospital in Kenya over 1 year. We related microbiological findings and cerebrospinal-fluid (CSF) laboratory measurements to tests that would typically be available at such a hospital.
Findings: Acute bacterial meningitis was proven in 45 children (5.0% [95% CI 3.7-6.6]) and probable in 26 (2.9% [1.9-4.2]). 21 of the 71 cases of proven or probable acute bacterial meningitis had neither neck stiffness nor turbid CSF. In eight of 45 children with proven disease the CSF leucocyte count was less than 10x10(6)/L or leucocyte counting was not possible because of blood-staining. The presence of either a leucocyte count of 50x10(6)/L or more or a CSF/blood glucose ratio of 0.10 or less detected all but two of the 45 children with proven acute bacterial meningitis; these two samples were grossly blood-stained.
Interpretation: The diagnosis of childhood acute bacterial meningitis is likely to be missed in a third of cases at district hospitals in sub-Saharan Africa without adequate and reliable laboratory resources. CSF culture facilities are expensive and difficult to maintain, and greater gains could be achieved with facilities for accurate leucocyte counting and glucose measurement.