A Claims-Based Algorithm for Identifying Hidradenitis Suppurativa Severity.

Area of Science:

• Dermatology
• Health Informatics
• Epidemiology

Background:

• Administrative claims databases lack information on hidradenitis suppurativa (HS) severity.
• Accurate HS severity identification in claims data is crucial for treatment effect modification analysis.

Purpose of the Study:

• To develop and validate a claims-based algorithm for identifying mild, moderate, and severe HS.
• To improve the ability to analyze HS treatment effects based on disease severity.

Main Methods:

• Linked electronic health records (EHR) with Medicaid claims data (October 2016-December 2019).
• Utilized multinomial LASSO regression on a training sample to identify key claims-based variables for HS severity.
• Validated the algorithm using positive predictive values (PPVs) in a hold-out testing sample.

Main Results:

• The algorithm achieved an 89% PPV for HS with one ICD-10 diagnosis, improving to 100% with concurrent systemic medication use.
• PPVs for HS severity were 20% (mild/uncertain), 54% (moderate), and 67% (severe).
• Combining mild/moderate versus severe HS yielded a 71% PPV, indicating good performance in identifying severe cases.

Conclusions:

• The developed claims-based algorithm shows reasonable performance for identifying severe HS.
• The algorithm has limitations in distinguishing between mild and moderate HS.
• Combining severity into mild/moderate versus severe categories improved the algorithm's performance.