Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, R01-CCR-115818, 2003 Jan; :1-21
The purpose of this research program was to address uncertainties in the exposure data previously used to define the relationship between lung cancer and diesel exhaust exposure in a cohort of over 55,000 US railroad workers. It was hypothesized that historical information regarding railroad-specific diesel locomotive use in the US railroad industry could be used to estimate lung cancer risk. Overall, a 20% to 50% excess lung cancer risk has been observed in occupations where diesel exhaust exposure was likely, including in studies where it was possible to adjust for cigarette smoking. Since miners in dieselized underground mines are exposed to levels of diesel particles 10 to 20 times greater than other occupational groups, MSHA is concerned about the assessment of risk in underground miners and potential lung cancer risk in other occupational cohorts. Comprehensive mortality studies assessing the risk attributable to diesel exhaust in miners are not yet available. Therefore, there has been interest in assessing lung cancer risk by examining other existing diesel exhaust exposed occupational cohorts Previous risk assessments, including work done by NIOSH, relied heavily on a cohort originally studied by our research group. Lung cancer mortality was assessed for 55,407 U.S. railroad workers from 1959 through 1980. Whether or' not a meaningful quantitative risk assessment can be done to estimate the potential lung cancer risk attributable to diesel exhaust is controversial due to inadequate information regarding historical exposure measurements and an ill-defined linkage between job title and personal exposure. The railroad industry changed from steam to diesel locomotives starting in the late 1940's, such that by 1959 based on aggregate data, 95% of the locomotives in service were powered by diesel. However, the date that an individual worker started working with diesel equipment is uncertain because there was wide variation in the acquisition of diesel locomotives by individual railroads. Based on locomotive testing and on differences in engine design and maintenance, it is likely that earlier generations of diesel locomotives (used in the 1950's and 1960's) had greater particulate emissions than later locomotives. No information about job-specific historical exposure to diesel exhaust is available. It was proposed that extensive information about the number and design of older locomotives used by each railroad could be linked with emission data to estimate relative differences in historical and current exposures. This information would permit the development of a profile of exposure for jobs in each railroad which could be assigned to each worker and possibly permit improved estimates of lung cancer risk per unit of exposure. The yearly locomotive roster of each railroad was obtained, which identified the make, model, and horsepower of each locomotive in service. From this, the number and type of locomotives that were diesel for each year could be calculated for each railroad. USEPA emission inventory guidance documentation provides emission factors for many locomotives. The emission factor times the locomotive's horsepower gives an estimate of total emissions. It was possible to estimate an average annual emission factor for each railroad weighted by the proportion of diesel locomotives. The assumption is that train crews' exposures will be roughly proportional to an average emission factor. Prior to the development of the historical profile of diesel locomotive use in the US railroad industry it was necessary to update the mortality experience of the railroad worker cohort between 1959 and 1980 since mortality was not completely ascertained in the original analysis. With the cooperation of the US Railroad Retirement Board, it was possible to obtain additional mortality information on 54,973(99.2%) subjects. Although there were over 500 railroads listed in this railroad worker database, the sample studied was limited to 93 railroads that contributed at least 0.1 % of the subjects to the cohort. This sample captured 96% of the eligible cohort (52,812 subjects), with 22 larger railroads (each contributing >1% of the cohort) accounting for 76%. Diesel exhaust exposure weighting factors were determined by examination of railroad equipment rosters between 1945-1980 by calculating a measure of the magnitude of exposure (an emission adjustment factor) and a measure of exposure probability (diesel fraction). The average emission adjustment factor for each railroad was based on railroad-specific number and type of locomotives. Since there have been a large number of mergers among the railroads, to account for service prior to a railroad merger, a weighted average of the emission adjustment factors of precursor railroads was calculated. The last railroad employer for each worker was available and provided the linkage between the railroad roster derived factors and work history. Examination of more detailed work histories for a subset of workers indicated that it was unusual for workers to change railroad employers. A yearly index exposure was defined based on the product of months of service for subjects who worked on operating trains (train riders: brakemen, conductors, engineers, firemen, or hostlers), the average railroad emission adjustment factor, and the diesel fraction. Analyses were also based on duration of work as a train rider weighted by the probability of diesel exposure. Yearly job code was available between 1959 and 1980, and job title before 1959 was based on 1959 job. Mortality from lung cancer was assessed using Cox proportional hazard regression methods. Between 1959-1980 there were 21,639 deaths out of 54,973 subjects, and it was possible to identify underlying cause of death in 21,116 (98%). Since lung cancer is usually fatal within several years following diagnosis, lung cancer cases were identified based on both underlying cause of death and if lung cancer was noted elsewhere on the death report (either death certificate or National Death Index report). There were 2,123 lung cancer cases identified overall, and 2,050 lung cancer cases were available for analysis using historical railroad information. In analyses not using railroad specific emissions data where diesel exposure was assumed to start in 1959, subjects who had the greatest duration of work as train riders had the lowest risk of lung cancer mortality. In efforts to consider the healthy-worker survivor effect, a condition where subjects who remain employed tend to be healthier than those who leave employment, analyses were repeated adjusting mortality for years of employment and time after leaving work. Regression models excluding exposure in the year of death and the preceding 5 and 10 years before death were also explored. The relative risk of lung cancer mortality for subjects with <5, 5 to <10, 10 to <15, and >15 years of work in diesel exposed jobs ranged from 1.55 to 1.18 without evidence of an increased risk with greater years of work. There was no increase in lung cancer risk in shop workers. Based on the distribution of railroad-specific diesel fraction, the duration of exposure to diesel exhaust for the train riders was estimated starting in 1945. Adjusting for years of employment and time after leaving work, there was no evidence for an increase in lung cancer with increasing years of exposure. With exposure in the 5, 10, and 15 years before death excluded, the risk of dying of lung cancer ranged from 1.13 to approximately 1.40 to 1.45, with a modest increase in risk with increasing years of exposure in some models. Based on quintiles of a cumulative exposure index (the summation of the product of railroad specific average emission adjustment factor multiplied by months of yearly work and diesel fraction), a consistent relationship between cumulative exposure and lung cancer risk was not noted, even after adjustment for the healthy worker survivor effect. The use of railroad locomotive rosters to estimate historical exposure to diesel exhaust did not significantly improve the estimation of lung cancer risk associated with work as a train rider when compared to analyses based on exposure assumed to start in 1959. There was no consistent relationship between railroad specific indices of cumulative exposure and lung cancer risk, although train-riders in diesel exhaust exposed jobs had a greater risk of lung cancer compared to workers in unexposed jobs, and in some models, workers with a greater duration of exposure had a greater risk. Workers in shop-related jobs did not have an increased risk of lung cancer. When examined closely, the codes selected for inclusion as shop workers also included workers in non-diesel shops, so the extent that all workers in the shop worker category had significant exposure to diesel was uncertain. The relationship between intensity and duration of exposure remains uncertain using the exposure weighting estimates developed in this study. Limitations to this approach include: the .uncertainty in a single railroad-specific value that was used to weight exposure for all train riders on a single railroad; limited source data for emission factor values; and the uncertainty in personal exposures estimated by emission factors obtained from testing a few engines and expecting them to represent emissions from locomotives operated over a variety of real-life conditions. Although exposures to diesel particulate were predicted to increase with engine emissions in this exposure model, there was no validation of the model. Nor was there information regarding how the characteristics of emissions (such as particle size, organic content) changed over time, or how previous exposure to coal fired locomotive emissions influenced lung cancer risk. Additional follow-up of the mortality experience of this cohort through 1996 is underway that may provide additional insight into lung cancer risk.