Populations exposed to risks at work show a deficit in deaths in comparison with the national population, the "healthy worker effect". This effect has been observed among former nuclear workers. The aim of our cross-sectional study was to evaluate the mortality in a metallurgy worker population, and to study the impact of the choice of the reference population on the estimation of the Standardized Mortality Ratio (SMR).
Methods : The studied population was the COGEMA Metallurgy Department workers, aged 25 years to 64, during the period 1980-1995 in two regions of France (North- La Hague; South- Cadarache, Marcoule, Miramas and Pierrelatte). In order to account for geographical variability and the difference in mortality between the working population and the non-working population, we used several reference populations : national population, regional population, "working" national population (farmers not present in COGEMA and non-workers excluded), and "working" regional population.
Results : All SMRs were increased when we used a "working" regional reference compared with the national reference population. Among men, eight pathologies presented a signifiant deficit with a national reference, whereas only two pathologies showed a deficit with the "working" regional reference.
Conclusion: The use of a reference population close to the study population enabled us to reduce the Healthy Worker Effect among metallurgy workers at COGEMA plants.