Objective To demonstrate the benefits of relative survival analysis to historical cohort studies in occupational epidemiology. In historical cohort studies, it is never known whether each death observed during the follow-up is attributable to the risks specific to the cohort or to the general population risk. SMR analysis is the most widely employed technique for quantifying the excess risk in the cohort compared to an external reference population. SMR assumes that the rate in the cohort is proportional to the rate in the reference population, at any time within predefined strata.

Methods As SMR analysis, relative survival analysis can be used to separate the excess mortality rate from the mortality due to the population risk. This approach requires the same data as SMR analysis (a reference life table and the death history of the cohort) and manages the death history much like right censored survival data. The time scaled used in the analysis is defined from a well-defined time origin (employment date, date of the first exposure, birthday) to the occurrence of the event (death day in mortality studies). Contrarily to SMR analysis, the excess rate is not supposed to be proportional to the population death rate but is a function of the time scale independently of the population death rate. Relative survival analysis is based on the estimation of the excess mortality rate function, the cumulative excess rate function and the net survival function. Patterns in these functions may reveal excess mortality during specific periods of the time scale.

Results The proposed approach will be illustrated using data from several mortality studies.

Conclusion Relative survival analysis is a relevant methodology for mortality analysis which allows for a dynamic analysis of the possible excess mortality rate.