Development and Validation of an Endometrial Cancer Algorithm in US Claims Data

Area of Science:

• Oncology
• Medical Informatics
• Epidemiology

Background:

• Accurate identification of endometrial cancer cases is crucial for epidemiological studies and post-authorization safety assessments.
• Existing coding systems require validation for precise case ascertainment in large datasets.

Purpose of the Study:

• To develop and validate algorithms for identifying endometrial cancer incidence using ICD-9-CM and ICD-10-CM codes.
• To support a post-authorization safety study on hormone therapies by ensuring reliable case identification.

Main Methods:

• Utilized national claims data from the HealthCare Integrated Research Database (HIRD).
• Developed screening algorithms based on diagnosis codes and adjudicated by expert review.
• Calculated positive predictive value (PPV) and conditional sensitivity for algorithm performance.

Main Results:

• An algorithm using two ICD-9-CM codes (182.0 or 182.8) achieved a PPV of 91.2% and sensitivity of 99.3%.
• An algorithm using two ICD-10-CM codes (C54.1, C54.8, or C54.9) achieved a PPV of 97.0% and sensitivity of 99.5%.
• Both algorithms demonstrated PPVs and sensitivities exceeding 75% across all test cohorts.

Conclusions:

• Two diagnosis codes for endometrial cancer accurately identify confirmed cases in both ICD-9-CM and ICD-10-CM systems.
• The developed algorithms exhibit high accuracy and minimal false positives for endometrial cancer.
• The validated algorithms are suitable for use in large-scale epidemiological and safety studies.