We aimed to validate and refine the encephalitis criteria proposed by the International Encephalitis Consortium in a cohort of adults initially suspected of a central nervous system (CNS) infection.

We included patients from two prospective cohort studies consisting of adults suspected of a CNS infection whom underwent a diagnostic lumbar puncture. We evaluated the test characteristics of the criteria for both possible and probable encephalitis. The reference standard was a final clinical diagnosis of encephalitis. Recalibration of the criteria was done by adjusting the weight of each criterion based on their respective odds.

In total 1446 episodes were evaluated, of whom 162 (11%) had a clinical diagnosis of encephalitis. Possible encephalitis had a sensitivity of 41% (95% CI 33–49) and a specificity of 88% (95% CI 86–90). Probable encephalitis had a sensitivity and specificity of respectively 27% (95% CI 20–34) and 95% (95% CI 94–96). Through odds-based weighting, we recalibrated the weight of each individual criterion, resulting in a model consisting of an altered mental status (weight of 2), seizures (weight of 3), elevated CSF leukocytes (weight of 5) and abnormalities on neuroimaging (weight of 9). We proposed a cut-off at 5 for possible encephalitis, (sensitivity 93% [95% CI 88–96]; specificity 51% [95% 49–54]), and at 8 for probable encephalitis (sensitivity 51% [95% CI 44–59]; specificity 91% [95% CI 89–92]).

We validated and refined the existing diagnostic criteria for encephalitis, leading to a substantially enhanced sensitivity. These updated criteria hold promise to facilitate the accurate identification of encephalitis.