Development and Validation of Algorithms to Identify Individuals With Cutaneous Lupus From Healthcare Databases

Area of Science:

• Dermatology
• Rheumatology
• Epidemiology
• Health Informatics

Background:

• Accurate identification of cutaneous lupus erythematosus (CLE) patients from large datasets like electronic health records (EHRs) is currently lacking.
• This limitation hinders epidemiological studies of CLE.
• Validated methods are needed to extract CLE cases from healthcare databases.

Purpose of the Study:

• To develop and validate accurate algorithms for identifying individuals with cutaneous lupus erythematosus (CLE) using healthcare records.
• To improve the ability to study CLE epidemiology through data mining.
• To establish reliable case-finding methods for CLE in large patient populations.

Main Methods:

• Twelve case-finding algorithms were created using International Classification of Diseases (ICD)-10 codes, provider specialties, and medication data.
• Algorithms were validated on a cohort of 300 individuals from a multi-institutional healthcare network.
• Case definition standard involved dermatologist/rheumatologist documentation or biopsy findings, with performance measured by positive predictive values (PPVs), specificity, and sensitivity.

Main Results:

• Positive predictive values (PPVs) for the algorithms ranged from 58.0% to 92.9%.
• Algorithms relying solely on a single CLE diagnosis code from any provider showed poor PPVs.
• The most effective algorithm, achieving 89.0% PPV with maintained sensitivity, required at least one ICD-10 CLE diagnosis code from a dermatologist.

Conclusions:

• Employing specific cutaneous lupus erythematosus (CLE) diagnosis codes combined with dermatology as the coding provider specialty represents a valid approach.
• This method enables accurate identification of CLE patients within electronic health record systems.
• The findings support the use of these algorithms for epidemiological research and clinical data analysis.