Send letter to journal:
Re: Estimation of occupational risks from environmental studies?

Since 1986 cancer events in the Dutch population related to nutrition and lifestyle factors are observed in the prospective cohort study “The Netherlands cohort study on diet and cancer”. Now the data of 58.279 male study participants have been evaluated regarding a potential association between the occupational activity and lung cancer. Present information on 5 performed jobs respectively activities of each study participant were condensed to 26 categories and industries respectively. No further details or exposure data were available. The relative risks of affection by lung cancer were evaluated for each of the 26 “industrial categories”.

In a total of 1.920 cases of lung cancer significant increased relative risks were calculated in the categories “electronics/optical instruments”, “construction and homebuilding business” and “railway company” after 15 years of employment. Confounders outside the occupation were considered. A significant decreased risk after 15 years of exposure was observed for “textiles and leather”.

The following questions arise from the view of an occupational physician:

• The creation of the category “construction and homebuilding business” merges such different tasks and their exposure scenery like brick layer, carpenter and painter. Do all these jobs have an increased risk or is the result due to the high risk of one of these occupations?

• Even harder to estimate are the categories, where no significant increased risks were found. Here as well jobs with totally different exposures were put together, e.g. when grouping the branches “Mining, quarrying and offshore” or “farming, forestry, horticulture, hunting, fishing”. In these cases an underestimation of relative risks may occur because occupations without risks are levelling off the results of the risk estimation.

• The problems of this study protocol become also obvious when looking at the significant results of the categories “electronics, optical instruments” and “textiles and leather” for which the authors do not have any explanation.

In general this study leads to the question, if it is justified to use data which originally have been collected for another purpose for issues of occupational medicine, especially when the data base of activities and exposures is obviously rudimentary.