(171) Assessing the potential for sequential channeling bias for drugs with continuously expanded patient populations: an illustrative example using dupilumab

Background: The pharmacoepidemiologic literature has described that selective prescribing of newly approved medications, for example to patients with severe or poorly controlled disease, can lead to channeling bias in observational safety studies. However, the potential for channeling bias to occur sequentially when a therapy’s indication is expanded over time has not been described. Here, we use dupilumab as an illustrative example of a therapy for which channeling bias could exist in new user populations upon label expansion. Dupilumab was initially approved by the United States (US) Food and Drug Administration (FDA) in March 2017 to treat moderate-to-severe atopic dermatitis (AD) in adults. This indication has subsequently been expanded to include adolescents ≥12 years (March 2019), children 6–11 years (May 2020), and young children 6 months–5 years (June 2022).

Objectives: To examine characteristics of the evolving population of AD patients newly treated with dupilumab in the US at the time of each label expansion.

Methods: Optum’s de-identified Clinformatics® Data Mart Database was used to create an overall cohort of individuals whose first dispensing of dupilumab occurred between March 2017 and July 2023. All included individuals had ≥365 days prior continuous enrollment, during which they had at least one claim of an AD diagnosis (ICD-10 L20.x). Four temporal subgroups were created based on whether dupilumab initiation occurred within the first six months following initial FDA approval (March 2017) or any of the three subsequent label expansions for the AD indication (March 2019, May 2020, June 2022). Characteristics (demographics, atopic comorbidities, prior AD medications, and healthcare utilization) in the year before dupilumab initiation were described.

Results: A total of 14,078 new users of dupilumab with AD were identified. Those in the March 2017 subgroup (n=363), compared to the most recent label update (June 2022 [selected results], n=1,744), were older (mean 48 vs 43 years) and more likely to be male (56% vs 44%). In the March 2017 subgroup, compared to the June 2022 subgroup, a higher prevalence of comorbid Type 2 inflammatory diseases (e.g., asthma: 33% vs 22%; allergic rhinitis: 25% vs 19%), a larger proportion of prior conventional systemic immunosuppressant use (30% vs 3%), and more frequent dermatologist visits in the previous year (mean 6.2 vs 3.1 visits) were suggestive of a higher level of disease burden and AD severity.

Conclusions: Users of dupilumab in March 2017 had markers suggestive of severe AD, suggesting channeling of the new therapy as expected. As the label has been expanded and clinical experience with the product has grown, the AD population newly treated with dupilumab appears to have shifted towards a more moderate-to-severe phenotype.