Objective: To investigate factors that most influenced survival from bladder cancer in New South Wales, Australia (NSW) and to consider the impact of changes in coding practices on the reporting the of bladder cancer outcomes.
Patients and methods: All NSW cases of bladder cancer diagnosed between 1980 and 2003 were followed to the end of 2004 (17 923 cases). Survival analysis was undertaken using Kaplan-Meier unadjusted disease-specific survival and adjusted disease-specific survival using Cox proportional hazards regression modelling. This analysis was unique in that it modelled the effect of sex, age, country of birth, socio-economic status (SES), histological type, extent of disease and period of diagnosis on survival from bladder cancer in NSW.
Results: After adjusting for sex, age, extent of disease, SES, period of diagnosis and histological type, the likelihood of death was 11% (95% confidence interval, CI 5-18%) higher in females than in males, with case fatality most influenced by age at diagnosis, extent of disease, and histological type. When the analysis was repeated for cases with a method 6 (i.e. coding undertaken in the registry after examination of the pathology report, which would enhance accuracy), the likelihood of death was 13% (95% CI 5-21%) higher in females than in males.
Conclusions: The NSW analysis controls for variability in coding, extent of disease at diagnosis and histological type of cancer. The analysis shows significantly lower survival from bladder cancer in NSW women compared with men, with no improvement in survival from 1980 to 2003. Possible reasons for the lower survivals in women, the lack of improvement in survival and coding differences in jurisdictions are discussed.