Researching occupational mortality. The problem of comparison

In a number of countries, studies on occupational mortality have proved to be a useful epidemiological tool for the surveillance of on-the-job health risks. In Scandinavia, studies of this type are facilitated by record linkages made possible by the personal identity number. Health selection of occupational groups makes the problem of comparison one of critical interest in analytical epidemiology of occupational mortality and morbidity. Three generally applied principles in choosing control groups are discussed. Uses of random samples of the whole population will most likely yield under- or overestimates of the true risk of the occupation. It is recommended that comparison groups be introduced from other occupations or from the same occupation if possible making the groups similar with respect to socio-economic factors and especially health selection.