RESUMEN
OBJECTIVE: To determine the extent to which increases in survival from melanoma are explained by changes in thickness, level, histological type, site of lesion, and sociodemographic characteristics. METHODS: Analyses of changes in survival among 9519 South Australians with melanoma reported to the State's population-based cancer registry during the 1980-2000 diagnostic period, using proportional hazards regression to adjust for thickness, level and other characteristics. RESULTS: Lower survivals applied for thicker lesions, deeper Clark levels, lesions on the trunk and scalp/neck, and for older cases and males. After adjusting for these characteristics, the relative risk (95% confidence limits) of case fatality for the 1994-2000 diagnostic period was 0.79 (0.63, 0.99), when compared with the 1980-1986 baseline. Prior to adjusting, the relative risk for these cases was 0.58 (0.47, 0.72). An unexpected finding was a secular change for deeper Clark levels within Breslow thickness categories. CONCLUSIONS: Approximately half the survival increase was not explained by changes in thickness, level, lesion site, and age and sex. Other possible contributors warranting further study include changes in ulceration, nodal or more distant site involvement, treatment gains and changes in tumour biology. The trend for deeper Clark levels within Breslow thickness categories requires independent confirmation.