Diagnostic accuracy of clinical and laboratory characteristics in suspected non-surgical nosocomial central nervous system infections

Abstract

BackgroundThe diagnosis of meningitis in nonsurgical hospitalised patients is often difficult and diagnostic accuracy of clinical, laboratory, and radiological characteristics is unknown. MethodsIn a prospective multi-centre cohort study in the Netherlands with adults suspected of central nervous system (CNS) infections, we included consecutive patients who underwent a lumbar puncture for the suspicion of a nonsurgical nosocomial CNS infection. All episodes were categorized into five final clinical diagnosis categories, as reference standard: CNS infection, CNS inflammatory disease, systemic infection, other neurological disease, or non-systemic, non-neurological disease. Diagnostic accuracy of clinical characteristics, the index test, was assessed by determining sensitivity, specificity, and positive and negative predictive values (PPV; NPV). ResultsBetween 2012-2022, 114 of 1275 (9%) patients included in the cohort had suspected nonsurgical nosocomial CNS infection. Sixteen (14%) patients had a confirmed diagnosis of nonsurgical nosocomial CNS infection, including 4 (25%) with bacterial meningitis, 9 (56%) with viral CNS infections, 2 (13%) fungal meningitis and 1 (6%) parasitic meningitis. Diagnostic accuracy of individual clinical characteristics was generally low. Elevated CSF leukocyte count had the highest sensitivity (81%; 95% CI 54-96) and NPV (96%; 95% CI 90-99). When combining the presence of abnormalities in neurological or CSF examination, sensitivity for diagnosing a CNS infection was 100% (95% CI 79-100) and NPV 100% (95% CI 78-100). CSF examination changed clinical management in 47% of patients. ConclusionOf patients suspected of nonsurgical nosocomial CNS infections, 14% were diagnosed with such infections. Diagnostic accuracy for individual clinical characteristics was low, with elevated CSF leukocyte count having the highest sensitivity and NPV.