Validating an algorithm for multiple myeloma based on administrative data using a SEER tumor registry and medical...

Area of Science:

• Health Informatics
• Oncology
• Epidemiology

Background:

• Administrative databases offer a valuable resource for studying large multiple myeloma patient cohorts in real-world settings.
• Accurate identification of multiple myeloma cases within these databases is essential for the validity of research findings.
• Developing robust algorithms is necessary to ensure reliable case ascertainment from administrative data.

Purpose of the Study:

• To develop and evaluate algorithms for identifying multiple myeloma cases using administrative data.
• To assess the accuracy and validity of these algorithms in real-world clinical settings.
• To enable large-scale studies of multiple myeloma patients through reliable data extraction.

Main Methods:

• Algorithms were developed and validated using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes (203.0x) at two study sites.
• Validation involved comparing algorithm results against tumor registry data and medical chart reviews.
• The selected algorithm required specific ICD-9-CM codes to appear before and after diagnostic procedure codes within a defined timeframe.

Main Results:

• At site 1, algorithm positive predictive values (PPVs) ranged from 0.54 to 0.88, with sensitivities from 0.30 to 0.88.
• A selected algorithm achieved a PPV of 0.81 and sensitivity of 0.73 at site 1.
• At site 2, the validated algorithm demonstrated a PPV of 0.86 (95% CI, 0.79-0.92).

Conclusions:

• Algorithms were successfully developed and validated for identifying multiple myeloma cases with adequate accuracy for claims database analyses.
• The optimal algorithm required at least two preceding ICD-9-CM codes (203.0x), followed by diagnostic procedure codes, and subsequent ICD-9-CM codes within a specific time window.
• These findings support the use of administrative data for robust multiple myeloma research.