OBJECTIVES: This study aimed to develop and validate algorithms to identify adult patients with non-relapsing secondary progressive multiple sclerosis (nrSPMS) in electronic health records (EHRs) or healthcare claims databases in the US.

METHODS: Eight algorithms were developed based on cognitive interviews with neurologists. Medical record data (including International Classification of Diseases-10, Current Procedural Terminology, and Healthcare Common Procedure Coding System codes) from 195 MS patients were collected. Algorithm performance (sensitivity, specificity, positive predictive value [PPV] and negative predictive value [NPV]) were tested in the full medical record data and code-only data subset. Face validity was further tested in the IQVIA Pharmetrics Plus^® claims database (2016-2020) among MS patients from 2016-2018 who had ≥1 inpatient claim or ≥2 outpatient claims for primary diagnosis of MS ≥30 days apart and were continuously enrolled for 2 years since one of their MS diagnoses.

RESULTS: Out of the 8 clinically recommended algorithms and hundreds of variations, the 2 best-performing algorithms used evidence of multi-system involvement (i.e., mobility dysfunction, brain/brain stem dysfunction, other) or ≤1 DMT use, and lack of relapse during study period to identify the nrSPMS state. In full medical record/code-only data, both algorithms resulted in 93%/92% sensitivity, 74%/84% specificity, 86%/90% PPV, and 87%/86% NPV. In the IQVIA database, 33,244 patients with MS were identified. The algorithms identified 19,661 and 19,783 patients with nrSPMS, respectively. Demographic, clinical, and utilization characteristics of these patients were reported.

CONCLUSIONS: The proposed algorithms showed high performance when tested in medical record data. Additionally, the algorithms identified a cohort of patients in claims data that appeared consistent with clinically identified patients with nrSPMS. These algorithms can be applied in other US EHR or claims-based datasets to facilitate further research and better describe the nrSPMS population.