OBJECTIVES: Insurance claim databases are a potentially valuable source for evidence generation in NMOSD, a rare autoimmune condition of the central nervous system with scarce real-world data. The use of one International Classification of Diseases (ICD-10) code for NMOSD (G36) has shown limited validity in detecting NMOSD patients from insurance claims. To date, there is no specific ICD code to identify NMOSD relapses. This systematic literature review aimed to identify algorithms to detect NMOSD patients and relapses using insurance claims data.

METHODS: We searched the MEDLINE, Embase and SciSearch databases using a combination of indexed and free terms to identify articles published between 2010–2020. Complementary searches of relevant congress abstracts were conducted. Given the paucity of NMOSD literature and similarities in disease presentation, the scope was extended to multiple sclerosis (MS) to gain insights from a more advanced field. Two independent reviewers screened the abstracts for relevance against pre-determined criteria, followed by an in-depth full-text review and data extraction.

RESULTS: Ten publications reported NMOSD algorithms, developed using various criteria including prescription claims and exclusion of MS diagnosis, as well as by combining inpatient/outpatient claims with ICD-9/ICD-10 NMO/NMOSD codes and core symptoms. Only one publication validated algorithms against chart reviews to identify NMOSD patients. Two publications described a similar, non-validated algorithm to identify NMOSD relapses. The literature was more abundant for identifying MS patients and relapses, with 13 and 8 publications reporting validated algorithms, respectively. Most were evaluated by external validation; two used indirect validation. Extending the timeframe of encounters, inclusion of prescription data, and combining inpatient and outpatient claims all increased sensitivity, whereas increasing the number of claims improved positive predictive value at the cost of sensitivity.

CONCLUSIONS: Due to the limited evidence, further research is needed to establish robust methods to detect both NMOSD patients and their relapses.