which clinical parameters predict a positive csf diagnosis of meningitis in a population with high hiv prevalence?

Background. The HIV epidemic has changed the aetiology of meningitis in sub-Saharan Africa, and frontline clinicians are faced with a variety of meningitic presentations. Doctors working in resource-limited settings have the challenge of appropriately selecting patients for lumbar puncture (LP), a potentially risky procedure that requires laboratory analysis.  Methods. In a rural South African hospital, the practice of performing LPs was audited against local guidelines. Data were collected retrospectively between February and June 2013. Symptoms and signs of meningitis, HIV status, investigations performed prior to LP and cerebrospinal fluid (CSF) results were recorded. With the aim of determining statistically significant clinical predictors of meningitis, parameters were explored using univariate and multivariate logistic regression analyses. Results. A total of 107 patients were included, of whom 43% had an abnormal CSF result. The majority (76%) of patients were HIV-positive (CD4+ cell count <200 cells/µl in 46%). Cryptococcal meningitis (CCM) was the most prevalent microbiological diagnosis, confirmed in 10 out of 12 patients. Of the non-microbiological diagnoses, lymphocytic predominance was the most common abnormality, present in 17 out of 33 patients. Confusion (p=0.011) was the most statistically significant predictor of an abnormal CSF result. Headache (p=0.355), fever (p=0.660) and photophobia (p=0.634) were not statistically predictive. Conclusion. The high incidence of CCM correlates with previous data from sub-Saharan Africa. In populations with high HIV prevalence, the classic meningitic symptoms of headache, fever and photophobia, while common presenting symptoms, are significantly less predictive of a meningitis diagnosis than confusion.