Replication and assessment of a claims-based algorithm to predict multiple sclerosis disease severity in the Medicare and commercially insured populations

Toliver, Joshua Caleb
Journal Title
Journal ISSN
Volume Title

Objectives: Patients with multiple sclerosis (MS) may have low health-related quality of life, increased mortality and morbidity, and increased healthcare costs. This study applied a claims-based algorithm to categorize disease severity and determine the risk for MS relapses and MS-related hospitalizations based on disease severity (low, moderate, and high) and insurance (commercial, Medicare age-eligible, and Medicare disability-eligible) groupings. Methods: Using the Optum Commercial® (2016-2018) and Humana Medicare (2013-2015) claims databases, a sample of commercial, Medicare age-eligible, and Medicare disability-eligible were created. A claims-based algorithm categorized MS disease severity using pre-index MS symptoms and healthcare utilization. Linear regression analysis assessed the relationship between pre-index disease severity scores and total costs excluding disease modifying therapies (DMTs) over a 12-month follow-up period. Flexible parametric models were used to assess the risk of follow-up MS relapses and MS-related hospitalizations. Results: A total of 17,988 MS patients were categorized into three insurance-based groups which differed significantly at baseline, with disability-eligible Medicare patients having the most symptoms and comorbidities. The proportion of variance accounted for in follow-up costs (excluding DMTs) by the pre-index severity scores was moderate to high for the three groups (53%-93%). The risk of MS-related relapses and MS-related hospitalizations increased as MS disease severity scores increased. Overall, 14% of patients had at least one MS relapse, with a range of 9% for commercial, low-severity patients to 43% for disability-eligible, high-severity patients. Overall, 5% of patients had at least one follow-up MS-related hospitalization, with a range of 2% for commercial, low-severity patients to 29% for age-eligible, high-severity patients. Conclusion: The pre-index claims-based disease severity algorithm performed well in predicting total follow-up healthcare costs. Patient severity groupings were related to the prevalence of follow-up MS relapses and MS-related hospitalizations. This claims-based algorithm may be a useful tool in determining MS disease severity in claims data.