MESuRE: Method of estimating relative survival

Outline of reasons and objectives
Relative survival methods make it possible to study the occurrence of events in a specific population relative to a reference population for which rate tables exist. They are mainly used in clinical and epidemiological studies on cancer. They are not used in occupational epidemiology but they are advantageous for studying the relationship between occupational exposures and health.
The object of this study is to transfer methodology to the field of occupational epidemiology and to develop computer tools for using those models.
This study represents the contribution from the division to two projects funded by the ANR (French National Research Agency), the first one, the MESuRE project, as part of the ANR’s 2008 White Programme, and the second, the CENSUR project, as part of the ANR’s 2012 White Programme.
Approach
Two computer tools were developed in consultation with the other partners for implementing the methodological developments of the MESuRE project. For each of them, a validation procedure was put in place. Adaptations for meeting needs more specific to occupational epidemiology were also implemented.
The concepts of analysis of relative survival were transposed to the field of occupational epidemiology. The methods were used in the short study “Mortality among Paris sewage workers”, in parallel with the usual methodology that is based on calculating standardised mortality ratios (SMRs).
Main results
The two tools are operational and were used in the mortality study, making it possible to illustrate the advantage of such methods compared with the usual methodology. While the SMR is an overall measurement, relative survival analyses give dynamic indicators on the relationship between the studied risk and health.
Discussion
Relative survival methods can be used in occupational epidemiology in studies for improving prevention of occupational diseases. Like the calculation of the SMR, they use the rate tables of a reference population. The tables that are usually available correspond to the general population, and biases inherent to studies on populations of workers can ensue (healthy worker effect, profession). Looking for solutions is being addressed in the next stage of the CENSUR project which will continue until October 2017 and is the subject of a study scheduled to start in 2015.