RESUMEN
OBJECTIVES: To explore the accuracy of using a point-of-care (POC) glucometer for cerebrospinal fluid (CSF) glucose screening. METHODS: A cross-sectional study was conducted. A glucose analysis of CSF samples collected from infants <90 days with suspected meningitis was paired between tests by using a POC glucometer (POC-CSF glucose) and a laboratory glucose analysis (laboratory-CSF glucose). Accuracy and limits of agreement were compared, as well as the glucometer performance to detect a laboratory-CSF glucose level <45 and 60 mg/dL. RESULTS: Seventy-three CSF samples were analyzed. Subjects' mean gestational age was 32.2 (SD 4.0) weeks, the mean weight was 1947.7 (SD 814.5) g, and the median age was 8 (interquartile range: 2 to 19.5) days. POC-CSF glucose levels ranged from 26 to 126 mg/dL. The mean (±1.96 SD) difference between POC-CSF and laboratory-CSF glucose levels was -1.6 (interquartile range: -12.6 to 9.4) mg/dL. A POC-CSF glucose level <45 mg/dL has a sensitivity and negative predictive value (NPV) to detect a laboratory-CSF glucose level <45 mg/dL of 82% and 94%, respectively. For a laboratory-CSF glucose level <60 mg/dL, a POC glucose level <60 mg/dL provides a sensitivity and NPV of 96% and 90%, respectively, whereas sensitivity and NPV reach 100% at a POC glucose level <70 mg/dL. CONCLUSIONS: A POC glucometer for CSF glucose can detect a potential abnormal glucose level with an appropriate cutoff level. This may facilitate rapid decisions for empirical antibiotics in suspected meningitis, pending laboratory results in limited-resource settings, but requires robust validation in future studies before implementation.