OBJECTIVES: To identify features (signs, symptoms, and co-morbidity) in the prodromal phase of Multiple sclerosis (MS) and utilize temporal association of each feature in predicting ‘at-risk’ individuals. The scope is to facilitate early intervention to prevent or delay the development of MS.

METHODS: MS cohort was created from Optum® de-identified Market Clarity Dataset. The time and age group considered was 1st of January 2016–31st December 2021 & 18-75 years respectively. Selection criteria for these patients was based on two claims (1IP & 2OP) at least 90 days apart using ICD-9-CM and ICD-10-CM code i.e., 340 & G35 respectively. Continuous eligibility and look-back period was of 3 years respectively. 9 sign and symptoms along with 16 co-morbidities were included in the study. Frequency, pattern, and time-to-event of each feature were also evaluated. Non-MS controls were matched to cases based on age and gender. Three models were built i.e., logistic regression (LR), random forest (RF), XGBoost (XG) and performance was evaluated based on area under the curve (AUC) and F1 score. Significance of association between prodromal features and MS incidence was calculated using odds ratio (OR) and P-value.

RESULTS: Total patient count in MS was 18,670 and median age at diagnosis was found to be 47 years. LR outperformed other models with AUC & F1 score of 0.74 & 0.76 respectively. Most common prodromal features in MS were Musculoskeletal (74%) and Ophthalmic (38%). Key differentiators were paralysis (37%) and facial neuralgia (45%) with OR of 5.6 and 2.6 respectively. Temporal analysis showcased 4.14% MS patients having psychiatric illness, Musculoskeletal & Ophthalmic symptoms in respective order.

CONCLUSIONS: Paralysis, facial-neuralgia and musculoskeletal symptoms were identified as prominent risk factors associated with MS in US population in the order of occurrence. These results improve our clinical knowledge of early MS, suggesting how underlying ailments in combination should be interpreted.