OBJECTIVES: Breast cancer (BC) continues to be a public health burden and has higher incidence (BCI) in certain disease populations. Currently, BCI and Atopic Dermatitis (AD) studies produce mixed findings. Administrative data may assist in detangling this relationship. However, use of validated algorithms (VAs) for administrative data targeting the general population may not be reliable in certain disease indications. Therefore, we explored BCI VA reliability in both the general population and an AD population.

METHODS: Published VAs from 1992 to 2018 were compared based on sensitivity, specificity, positive and negative predictive value, and study population. Two VAs were chosen; one included both procedural and diagnostic codes, while the other only included diagnostic. IQVIA Pharmetrics and Optum Clinformatics databases from 2014 to 2019 were used for VA implementation. BCI rates (IR) per 100,000 person-years were calculated. This was replicated in an AD only cohort defined ≥2 AD diagnoses at least 30 days apart (ICD10: L20.X).

RESULTS: For general population estimates in Pharmetrics 4,408 and 380 incident cases were identified using the procedural and diagnostic VA, respectively. The diagnostic VA IR was 184.67 cases (95%CI: 179.07,190.44), while the procedural VA IR was 17.25 cases (95%CI: 15.61,19.08). In the AD population 312 and 17 incident cases were identified using the diagnostic VA and procedural VA, respectively. The IR for the AD population varied from 7.90 (4.96,12.96) to 145.35 (130.11, 162.41) BCI cases. Similar IRs were calculated from Clinformatics.

CONCLUSIONS: The two VAs provided significantly different IRs, in both data sources, which diverged from the Surveillance, Epidemiology, and End Results (SEER) general population estimates and expected AD-specific population estimates. This suggests limitations with the use of claims data and the algorithms themselves. Researchers should carefully consider their data source and indicated population when adopting previously validated algorithms in their studies.