Development of anti-LGI1 autoimmune encephalitis outcome rating score using the Rasch analysis

Anti-LGI1 autoimmune encephalitis is an immunotherapy-responsive condition with growing interest in clinical trial development. However, the lack of reliable outcome measures is a major barrier. We aim to design a disease-specific outcome measure for anti-LGI1 autoimmune encephalitis.

The anti-LGI1 autoimmune encephalitis outcome rating (LOR) score was proposed through expert consensus and literature review. It was applied to a development cohort of 81 patients, followed by Rasch analysis to evaluate item-person targeting, reliability, and dimensionality. Adjustments were made to improve its psychometric performance. The finalized score was re-applied to the development cohort and validated in a separate cohort of 54 patients. Finally, the score was assessed in a longitudinal cohort of 53 patients for responsiveness.

The LOR score consists of 6 items: seizure type, seizure frequency, memory, orientation, behavioral disturbance, and gait, with a total score of 19. Compared to the CASE score, the LOR score can better capture patients across varying levels of severity, without major floor or ceiling effects. The scoring within each item proportionally reflected the item's difficulty. The score was relatively unidimensional. The reliability (0.63) and separation (1.34) indicated very small random errors. The inter-rater and intra-rater reliability (κ > 0.85) was excellent. In the longitudinal cohort, the changes in LOR score are significantly different among different clinical impression of change (IOC) categories (p < 0.01), indicating its ability to detect clinically meaningful changes.

The LOR score is a psychometrically reliable, sensitive, responsive, and relatively unidimensional outcome measure, which may be used in clinical practice and trial settings.