Melanoma Tumor Depth Quality Audit: A Nonmatch Analysis

Area of Science:

• Oncology
• Cancer Surveillance Research
• Medical Informatics

Background:

• Concerns arose regarding melanoma tumor depth coding within the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program.
• The Surveillance Research Program (SRP) initiated efforts to address these coding inconsistencies.

Purpose of the Study:

• To develop and validate an algorithm for accurately identifying melanoma tumor depth measurements.
• To assess the accuracy of existing melanoma depth coding practices through a nonmatch analysis.

Main Methods:

• A natural language processing algorithm was developed to extract melanoma tumor depth values.
• A nonmatch analysis compared algorithm-identified values against a gold standard using 1,117 cancer cases (2010-2017).
• Statistical analyses were performed using SAS software (version 9.4) to evaluate discrepancies and their distribution.

Main Results:

• 76% of cases showed a match between originally reported and gold standard melanoma depth values.
• 16% of cases had incorrect AJCC-7 T staging based on original reporting, often misclassified as TX instead of T1.
• Overall, 24% of cases exhibited discrepancies in recorded melanoma depth, with decimal errors accounting for 3%.

Conclusions:

• A significant portion of melanoma cases (24%) had coding discrepancies in tumor depth.
• The developed algorithm can enhance cancer registry data quality by automating the identification of inconsistencies.
• Automated review and adjudication by registrars can optimize resource allocation and improve data accuracy.