(C-196) Varying period prevalence of idiopathic inflammatory myopathies by enrollment factors and case definition: an exploration in a U.S. administrative claims database

Background: Idiopathic inflammatory myopathies (IIMs) are a collection of rare inflammatory muscle disorders with limited data on current prevalence. Administrative healthcare claims data are increasingly used to describe disease occurrence. While they offer large, presumably representative samples, which is advantageous for the study of rare diseases like IIM, their use relies on scientist specification to produce meaningful and interpretable results.

Objectives: To demonstrate the impact of design specifications on estimated prevalence of IIMs in a U.S. administrative claims database.

Methods: Optum’s de-identified Clinformatics® Data Mart Database was used to identify adults (18+) between 1Jan2019-31Dec2023. IIM cases were defined in 3 ways using ICD-10-CM diagnosis codes and procedure codes related to deep muscle biopsy used for confirmatory diagnosis: Definition A. Full diagnosis list (91 codes); Definition B. Full diagnosis list plus biopsy evidence (to increase diagnostic certainty) (CPT 20205) for “unspecified” or “other” myositis in unspecified, other, or multiple sites (M60.9, M60.89, M60.88, M60.80); Definition C. Restricted diagnosis list for dermatomyositis, polymyositis, and interstitial myositis (49 codes). Codes were selected based on prior literature and medical input. Definitions required ≥1 inpatient or ≥2 outpatient claims with diagnosis code in any position, separated by 30-364 days. Continuous enrollment requirements varied from 1+ day to 365+ days. Age- and sex-adjusted period prevalence was estimated for each year; 2023 was selected for comparison.

Results: 2023 period prevalence of IIM varied from 135.7 per 100,000 persons (95% CI 130.4-141.1) (Definition A, 365+ day enrollment) to 22.6 per 100,000 persons (95% CI 20.8-24.4) (Definition C, 1+ day enrollment). Reducing enrollment requirement to 1+ days decreased the estimate for Definition A to 110.2 per 100,000 persons (95% CI 106.3-114.1). Requiring presence of a biopsy code in Definition B substantially reduced the prevalence estimate from Definition A (135.7 to 43.5 per 100,000 (95% CI 40.5-46.5) with 365+ day enrollment).

Conclusions: Estimated prevalence of IIM was highly dependent on code list specification and selection of continuous enrollment periods. When designing and interpreting studies of disease occurrence in claims-based data sources, scientists should consider clinical insight on medical practice and the impact of imposing enrollment restrictions to increase confidence on the interpretability of the results from these data sources.