Impact of dupilumab approval on diagnostic coding patterns for atopic dermatitis

Summary: Abstract Body: The prevalence of atopic dermatitis (AD) has been increasing, the exact causes of which remain unknown. Estimates of AD prevalence derived from administrative claims or electronic health records data may be affected by changes in diagnostic coding patterns over time due to new drug approvals, among other reasons. To understand the potential impact of dupilumab approval for treatment of moderate-to-severe AD among adults on AD prevalence and ICD-based algorithms for identifying individuals with AD, we evaluated the International Classification of Diseases, Tenth Revision (ICD-10) coding patterns for AD and other dermatitides before and after drug approval. We performed a controlled interrupted time series analysis of Medicare inpatient and outpatient visits with at least one ICD-10 code for dermatitis (L20.0, L20.8x, L20.9, L23.x-L25.x, L30.0, L30.1, L30.5, L30.8, L30.9) between January 2016 and December 2019. The primary outcome was change in monthly proportions of visits associated with the codes L20.8x (other AD); L20.9 (AD, unspecified); L30.0 (nummular dermatitis); L30.8 (other specified dermatitis), and L30.9 (dermatitis, unspecified) before and after dupilumab approval in March 2017. L30.0 served as the reference code due to its low likelihood of being affected by dupilumab approval. Total number of dermatitis-associated visits ranged from 182,801 to 376,927 per month. The total effect (i.e., difference in expected and observed code frequency one year after approval) was positive for specific AD ICD codes L20.84 (intrinsic eczema; 3.26%; 95% confidence interval [2.44%, 4.08%]) and L20.9 (1.46% [1.01, 1.90]) while negative for non-specific dermatitis codes L30.8 (-2.98% [-3.79%, -2.16%]) and L30.9 (-3.26% [-4.72%, -1.80%]). Analyses assessing sensitivity of the results to study time period and modeling choices were consistent for L20.84 but not L20.9 and L30.8. Our study identified increased use of more specific codes for AD and decreased use of non-specific dermatitis codes post-dupilumab approval that may contribute to increasing AD prevalence estimates. Junko Takeshita¹, Jonathan Heintz¹, Nandita Mitra¹, Gary Hettinger¹ 1. University of Pennsylvania, Philadelphia, PA, United States. Clinical Research: Epidemiology and Observational Research