and 667.99 expected deaths. The cohort had elevated rates of lung cancer deaths after 5 years of follow-up, and the trend increased throughout the follow-up period at a rate of 1.44 percent per year. This study did not have detailed exposure data on individual workers and measured individual exposure (yes/no) by whether or not the individual worked at least 30 days in a job or department that had contact with radioactive materials. In the multivariate analysis, the strongest predictor of lung cancer mortality was socioeconomic status; professional workers had a substantially lower rate of lung cancer mortality. Mortality due to lung cancer was not significantly associated with the exposure measure (whether or not there was exposure to radiation for 30 days or more), and the risk decreased with increasing duration of employment (this trend was significant for all cancers but not significant for lung cancer). However, the exposure measure was of limited validity since it reduced variation in workers’ exposure by classifying them according to exposure in one site rather than their entire work history. Detecting an effect of an agent is more difficult when anything reduces variation in individuals’ exposure.
The multivariate analysis suggests that exposure to radiation was not the explanation for the increase in lung cancer mortality since the coefficient for exposure to radiation was not significantly different from 1. Furthermore, the analysis suggests that other factors related to socioeconomic status (SES) may account for the association with lung cancer deaths. In particular, being a professional worker rather than an unskilled worker reduced the likelihood of dying from lung cancer, because this predictor variable had a large negative coefficient that was three times its standard deviation. There are a number of factors that could account for the differences in the socioeconomic variable including different rates of smoking or jobs with higher levels of exposure in lower SES individuals.
The salient feature of this case-control study was a thorough dose–response analysis (Table 4.11). Overall, there was no apparent relationship between internal radiation dose, lagged for 10 years, and lung cancer mortality. The only suggestion of an increased risk was for a cumulative internal dose of 25 cGy or more; the relative risk was 2.05, but this figure had extremely wide confidence intervals (0.20–20.70) owing to the very small number of cases in this group. Notably, in the next highest category of exposure, 5 to <25 cGy, the relative risk was 0.64 (CI 0.37–1.12). Because smoking data were not available for all persons, it was not possible to adjust for smoking when examining the relation between exposure at 25 cGy or more and lung cancer risk. There was also no overall relationship between external radiation exposure and lung cancer deaths except when the data were restricted to the small number of persons hired at 45 years of age or older (n = 64 pairs). Even then, the confidence interval was very wide (95% CI 31–1,411), and there was no suggestion of any trend with increasing dose.