Lung cancer mortality and diesel exhaust: reanalysis of a retrospective cohort study of U.S. railroad workers.

A retrospective cohort study of 55,407 U.S. railroad workers has been called the most definitive study linking exposure to diesel exhaust (DE) with lung cancer in humans. However, reanalysis of data from this study suggests caution in interpreting this study as demonstrating such a link. Although workers who rode trains had a significantly elevated lung cancer mortality relative to clerks and signalmen (who were assumed to be unexposed), shop workers did not, despite convincing evidence that these workers had the highest exposures to DE. Mortality from heart disease and cirrhosis of the liver were also significantly elevated among train riders, which suggests that these workers had a substantially different lifestyle from other workers, and raises the possibility that their elevated lung cancer mortality may be related to lifestyle rather than to DE exposure. Smoking information was not available for this cohort. A positive, monotone dose-response trend in lung cancer mortality with increasing duration of exposure found by the original investigators was not present when age was controlled more carefully and years of exposure quantified more accurately. Instead, a negative dose-response trend for lung cancer was seen among exposed workers based on either duration of exposure or quantitative measures of cumulative exposure. Similar negative trends were seen with several broad categories of mortality, including all causes. These negative trends are possibly a result of incomplete follow-up that was most severe among workers with the longest tenures. A sizable fraction of deaths occurring during the last 4 years of follow-up evidently were not identified, and there is evidence that follow-up in earlier years was also incomplete. At the very least, problems with the follow-up should be rectified before any conclusions are drawn about the carcinogenicity of DE in this cohort.