Geographic and patient variation among Medicare beneficiaries in the use of follow-up testing and skin examinations after local excision of cutaneous melanoma

National studies examining the service patterns and epidemiology of melanoma have been primarily based on data from the National Cancer Institute's (NCI) Surveillance, Epidemiology, and End Results (SEER) tumor registry. However, SEER comprises a non-random, 14% geographical sample of the U.S. limiting its ability to examine melanoma in all states and counties. SEER data also does not have treatment beyond 4 months following diagnosis. Medicare claims databases are an alternative source of melanoma patient data and enable the study of this disease throughout the entire U.S. Few studies have employed Medicare data for melanoma research, and no study has validated Medicare claims for the study of melanoma. We use a linked database with both SEER and Medicare claims information for patients diagnosed with melanoma in SEER regions to perform three melanoma studies as described below.; In the first section (chapter 2), we examine the completeness of Medicare claims for the identification of patients with melanoma. Using SEER as the gold standard, we assess the proportion of patients diagnosed in SEER that were also appropriately diagnosed with melanoma in Medicare claims. We found approximately 90% of patients diagnosed with melanoma can be identified for studies using Medicare claims alone.; In the second section (chapter 3), we examine the sensitivity and specificity of Medicare radiation and surgical treatments at least as invasive as excision or biopsy. Here again SEER was considered the gold standard. We found moderately high agreement for radiation therapy (70% sensitivity, and 99% specificity) and fair agreement for surgery codes (70–86% sensitivity and 35–60% specificity).; In the final section (chapter 4), we examine patient, tumor, and geographic variation in the use of follow-up surveillance following diagnosis of invasive melanoma. We found significant variation in surveillance by sex (male greater than female), and geographic area, with up to two-fold differences in surveillance practice. There was also a trend for the older aged to receive less surveillance. As expected, more advanced stage disease was also associated with greater surveillance intensity.