Abstract
Early and accurate diagnosis of central nervous system infections (CNSIs) is crucial for clinical treatment. Cerebrospinal fluid (CSF) analysis is considered the gold standard for diagnosis, and the sensitivity of its results directly affects treatment decisions. However, due to physiological differences in CSF composition between the ventricles and the lumbar space, the choice of sampling site may lead to diagnostic discrepancies, including false negatives. This study aims to compare the diagnostic sensitivity of CSF samples obtained via external ventricular drainage (EVD) and lumbar puncture (LP) in patients with CNSIs following craniotomy. METHODS: This study prospectively collected data from patients who underwent craniotomy and had EVD placement between January 2024 and December 2024. For patients suspected of CNSIs, CSF samples were simultaneously collected via LP and EVD, and the differences in cell counts and biochemical markers were compared. The Kappa index was used to assess diagnostic sensitivity and correlation, and statistical analysis was performed using McNemar's χ(2) test. RESULTS: A total of 41 patients were included, with 41 LP samples and 41 EVD samples collected. Among the 82 samples, 29 met the diagnostic criteria for CNSIs, with 21 (72.4%) from LP samples and 8 (27.6%) from EVD samples. Among the 21 LP-diagnosed infection cases, 14 EVD samples did not meet the infection criteria, while among the 8 EVD-diagnosed infection cases, only 1 LP sample did not meet the infection criteria. The Kappa correlation index between LP and EVD diagnostic results was 0.279, and McNemar's χ(2) test yielded p = 0.001. CONCLUSION: LP CSF demonstrates higher sensitivity than EVD CSF for early diagnosis of CNSIs in post-craniotomy patients with indwelling EVDs. In clinical practice, when EVD results are negative but there is high clinical suspicion of CNSIs, concurrent LP should be performed for further confirmation.