Objectives: To study the diagnostic accuracy of clinical and laboratory features in the diagnosis of central nervous system (CNS) infection and bacterial meningitis.
Methods: We included consecutive adult episodes with suspected CNS infection who underwent cerebrospinal fluid (CSF) examination. The reference standard was the diagnosis classified into five categories: 1) CNS infection; 2) CNS inflammation without infection; 3) other neurological disorder; 4) non-neurological infection; and 5) other systemic disorder.
Results: Between 2012 and 2015, 363 episodes of suspected CNS infection were included. CSF examination showed leucocyte count >5/mm3 in 47% of episodes. Overall, 89 of 363 episodes were categorized as CNS infection (25%; most commonly viral meningitis [7%], bacterial meningitis [7%], and viral encephalitis [4%]), 36 (10%) episodes as CNS inflammatory disorder, 111 (31%) as systemic infection, in 119 (33%) as other neurological disorder, and 8 (2%) as other systemic disorders. Diagnostic accuracy of individual clinical characteristics and blood tests for the diagnosis of CNS infection or bacterial meningitis was low. CSF leucocytosis differentiated best between bacterial meningitis and other diagnoses (area under the curve [AUC] 0.95) or any neurological infection versus other diagnoses (AUC 0.93).
Conclusions: Clinical characteristics fail to differentiate between neurological infections and other diagnoses, and CSF analysis is the main contributor to the final diagnosis.
Keywords: Cerebrospinal fluid; Diagnostic test assessment; Encephalitis; Meningitis.
Copyright © 2016 The British Infection Association. Published by Elsevier Ltd. All rights reserved.