High mortality in the Thule cohort: an unhealthy worker effect

The objective was to study mortality in the Thule cohort in order to clarify whether it is a selected population and to ascertain the possibility of misinterpretation when national mortality rates are used as reference in the analysis of occupational mortality.

The cohort consists of 4322 Danish men who were employed between 1963 and 1971 at the Thule air base in Greenland. One part of the cohort were employed during the clean-up period after the crash, in 1968, of a US bomber carrying nuclear weapons, the other part had been employed only outside the clean-up period. The cohort was followed up until 1992.

After 30 years of follow-up, SMR was 1.38 for all causes (95% confidence interval [CI]: 1.28-1.49), 1.25 for cancer (95% CI: 1.07-1.45), 1.17 for circulatory diseases (95% CI: 1.01-1.34), 1.58 for other natural causes (95% CI: 1.35-1.84), and 1.46 for violent deaths (95% CI: 1.22-1.74). Marked excess mortality measured by SMR was found from lung cancer 1.70, AIDS 3.55, alcoholism 4.04, cirrhosis of the liver 2.45, symptoms and ill-defined conditions 1.93, and suicide 1.63. The SMR was 1.09 for the age group 17-24 at entry, 1.42 for the age group 25-34, and 1.45 for the age group > or = 35.

The high mortality and the mortality pattern in the Thule cohort shows strong evidence for selection and provides strong support for the suggestion that these workers constitute a group in poor health probably caused by lifestyle. The study demonstrates that an incomplete analysis (i.e., one limited to a subgroup of workers involved in the radiation clean-up) could lead to spurious conclusions about hazardous occupational exposures.