2026 American Academy of Neurology Autoimmune Abstract Website

Objective:

To identify clinical factors associated with intensive care unit (ICU) admission in autoimmune encephalitis (AE).

Background:

ICU admission is frequent in AE, but predictors of ICU requirement in low- and middle-income settings remain poorly characterized.

Design/Methods:

We conducted a retrospective analysis of the multicenter REAL-LABIC registry, including patients with probable or definite AE from six Latin American countries (2017–2022). The primary outcome was ICU admission. Age-stratified analyses were performed. Associations between clinical variables and ICU admission were evaluated using logistic regression models. An exploratory clinical severity score (0–5 points) was constructed based on bedside features at presentation, assigning one point each for dysautonomia, movement disorders, seizures, altered level of consciousness, and heart rate >90/min.

Results:

Among 154 patients, 86 (55.8%) required ICU admission, more frequently in children than adults (65.5% vs 43.3%, p=0.006). In unadjusted analyses, multiple features were associated with ICU admission, particularly in adults, including dysautonomia, movement disorders, tachycardia, abnormal EEG and brain CT. In adjusted models, fewer independent associations were identified. In adults, shorter time from symptom onset to diagnosis (OR 0.97, 95% CI 0.96–0.98), abnormal brain CT (OR 7.14, 95% CI 2.47–20.60), and a higher clinical severity score (≥3 points) (OR 16.90, 95% CI 1.83–155.86) were associated with ICU admission. In children, only female sex remained associated (OR 2.85, 95% CI 1.20–6.79). The severity score showed good discrimination for ICU admission in adults (AUC 0.81, 95% CI 0.70–0.91) but not in children (AUC 0.49, 95% CI 0.37–0.62).

Conclusions:

ICU admission in AE is associated with global clinical severity rather than isolated features, with distinct age-related patterns. A simple bedside severity score may help identify adults at higher risk of ICU admission, although external validation is required. Pediatric predictors were limited, suggesting differences in disease presentation and risk stratification.