Senior Scientist Optum Epidemiology Boston, United States
Background: In the absence of a validated algorithm, it has been challenging to identify atopic dermatitis (AD) in claims data due to its heterogenous clinical presentation and lack of International Classification of Diseases, 10th Revision (ICD-10) codes that specify disease severity.
Objectives: To develop and validate a claims-based algorithm for moderate-to-severe (mod-sev) AD using medical records.
Methods: Patients in the Optum Research Database with a diagnosis code for AD (ICD-10: L20*) between March 2017 and November 2019 were identified. Within this source population, several candidate claims-based algorithms were applied to identify patients with mod-sev AD. The algorithms included proxy measures for mod-sev AD, such as number of AD diagnoses, number and types of AD therapies, and treating provider specialty. Among patients who met ≥ 1 candidate algorithms, a subset was randomly selected for medical record review. Additionally, medical records from a random sample of patients from the source population were reviewed to assess sensitivity. Overall, 278 records were sought and 200 were adjudicated, including 100 from patients with potential mod-sev AD and 100 from the random sample of patients with an AD code. Case status was determined separately for presence of AD and presence of mod-sev disease by a dermatologist with AD expertise. The positive predictive value (PPV) of each algorithm was calculated and compared to a pre-specified threshold of ≥ 70%.
Results: For the initial algorithms, which required ≥ 1 AD diagnosis, PPVs ranged from 38-100%. For algorithms with PPVs ≥ 70%, only 6 patients met the criteria. The algorithms were then modified to require ≥ 2 AD diagnoses; the PPVs of the revised algorithms ranged from 57-100%. The final algorithm selected for mod-sev AD had a PPV of 76% (95% confidence interval [CI]: 53% - 90%) and a sensitivity of 35% (95% CI: 21% - 52%). This final algorithm required either: 1) ≥ 2 claims for AD from any physician and either ≥ 1 dispensing of dupilumab or ≥ 2 dispensings of high potency topical corticosteroids or systemic immunosuppressants; or 2) ≥ 2 claims for AD from a dermatologist or allergist and ≥ 3 dispensings of medium potency topical corticosteroids, topical tacrolimus, phototherapy, or oral/parenteral corticosteroids.
Conclusions: The first validated claims-based algorithm for identifying patients with mod-sev AD was successfully developed. This rigorous study utilized expert dermatologist review of 200 medical records. It also required a high PPV threshold, which was met by revising initial candidate algorithms. This algorithm can be used in future studies including those seeking to evaluate the safety and effectiveness of AD therapies in the post-marketing setting.
