Stastistical methods for cancer survival analysis.

Abstract

Adequate and complete follow-up is a prerequisite for the conduct of any survival study. Passive follow-up relies on routine availability of mortality data through unique data linkage possibilities, while active follow-up supplements mortality ascertainment, for which there are a variety of methods. Cox proportional-hazard model was employed to test whether censoring was random in presence of loss to follow-up. Absolute survival probability was estimated by the actuarial method following semi-complete approach for all registries, and the period approach was also used wherever possible. Expected survival probability for registries was estimated from the respective country-, age- and sex-specific abridged life tables. Relative survival, as the ratio of absolute to expected survival, was calculated to exclude the effect arising from different background mortalities. To account for the differences in the age structure of the cancer cases, relative survival was adjusted for age and reported as age-standardized relative survival. Estimated incident cancer cases from less-developed countries together for every classified cancer site served as the standard population. Weights were assigned to individual patients, depending on their age, and standardization was carried out using weighted individual data. Analyses were done using the publicly available macros in SAS software.