Background: The completeness of Medicare claims for identifying patients with melanoma for purposes of conducting population-based studies of melanoma is unknown.
Methods: Using a linked Surveillance, Epidemiology, and End Result (SEER) tumor registry-Medicare database, the sensitivity of Medicare claims for identifying 5372 patients age > or =65 years diagnosed with invasive melanoma between 1992 and 1996 was determined. Sensitivity was calculated as the proportion of incident cases of melanoma reported by SEER that was also captured by Medicare claim diagnostic codes.
Results: The overall sensitivity of combined Part A and Part B Medicare for incident cases of melanoma was 90.1%. Part B Medicare and Part A Medicare alone had 89.5% and 16.5% sensitivity respectively. Sensitivity was lower for patients with unrecorded Breslow depth and for patients with unstaged or distant stage melanoma.
Conclusions: Medicare Part B claims have a high sensitivity for detecting melanoma incidence; Medicare Part A has low sensitivity. This sharply contrasts with published studies of other cancers, for whom Part A rather than Part B Medicare captures the predominant portion of incident cases. Medicare Part B or combined Part A and Part B administrative data is a potentially valuable resource for population-based melanoma research in the elderly. Further research characterizing the specificity and predictive value of Medicare data is needed to assess the potential implications of false positive melanoma diagnostic codes.