Cancer Risk from Diesel Emissions Exposure in Central and Eastern Europe: A Feasibility Study

ABSTRACT

The objectives of this study were to assess the feasibility of enrolling a multicenter historical cohort of workers exposed to diesel emissions in central and eastern Europe (the Czech Republic, Estonia, Hungary, Latvia, Lithuania,Poland, Russia, Slovakia, and Slovenia) and of reconstructing past exposures for the cohort. We developed a company questionnaire aimed at collecting standardized information to determine the number of workers who might be included in the cohort study and their exposures. Our collaborators contacted companies that might participate in the study, provided a contact person with questionnaires to complete, and checked the completed questionnaires. Data from company questionnaires were combined to estimate the size of the expected cohort in terms of exposure to diesel emissions and relevant agents. We determined that it would be possible to enroll 3 multi-center cohorts of workers exposed to diesel exhaust in nonmetal mining, bus transport, and railway transport. A total of 16 companies from 8 countries agreed in principle to participate in the proposed study, yielding an estimated total of 46,500 exposed workers with 295 deaths from lung cancer expected among them. The study would have statistical power to assess a 30% increase in lung cancer risk in each industry-specific cohort and a 15% increase overall. Groups of workers without exposure to diesel exhaust, from the same companies, would be available for internal comparisons. The cohorts would be primarily composed of male workers.



METHODS
FEASIBILITY OF EPIDEMIOLOGIC STUDY
Development of Company Questionnaire

A company questionnaire was developed with the aim of collecting standardized information on the size and exposure of the groups of workers that might be included in the cohort study (Appendix A). Special emphasis was put on the availability of information on exposure to agents other than diesel emissions. The questionnaire was prepared at IARC, translated into the national language in each collaborating center, and back-translated into English at IARC to ensure the integrity of the original translation. 

Contact with Possible Participating Companies 

The feasibility study was conducted in the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Russia, Slovakia, and Slovenia. Study collaborators are listed in the Acknowledgments section. A similar study, not reported here, was conducted in Romania and Israel. The feasibility study focused on 2 broadly defined groups of workers with potentially high exposure to diesel emissions: drivers (ISCO 985), mechanics (ISCO 843), and railway workers (ISIC 71) in land transport; and miners (ISCO 71) in nonmetal mining. (The codes in parentheses are from the International Standard Classification of Occupations [ISCO; International Labour Office 1968] and the International Standard Industrial Classification [ISIC; United Nations 1971])


Each collaborating center identified and contacted companies in its study area that could include a substantial number of workers in the 2 categories defined above. The initial contact aimed at identifying companies available to participate in the study, the approximate size of an exposed cohort and of another cohort not exposed to diesel emissions, the expected completeness of available employment records, and the available information about changes in sources of diesel emissions and the environment where the emissions occurred.

Completion of Questionnaires

The collaborators visited each company and presented the questionnaire to the contact persons. The contact persons, who typically were production managers, technical engineers, or industrial hygienists, were in charge of completing the questionnaire. They also provided additional information such as copies of sample pages of employment rosters and exposure records. The completed questionnaires were returned to the collaborators, who checked the completeness and quality of the information collected. Typically, the collaborators visited the company on several occasions to complete or clarify some aspects of the information collected. The questionnaires were then translated into English, and copies were transferred to IARC and to the Institute of Occupational Medicine (IOM, Edinburgh, UK).

During the feasibility study and after completion of the company questionnaires, IARC epidemiologists visited each collaborating center; they also visited some of the companies and discussed details about all companies with the contact persons. In order to carry out personal exposure monitoring for selected occupations, chosen to represent a diverse range of exposure scenarios, IOM industrial hygienists visited four centers. In addition, the hygienists assessed current work practices and conditions and verified or clarified the scope and quality of preexisting exposure-related information that would be required for the reconstruction of exposure histories.

Estimate of Statistical Power of Cohort Study

We derived from company questionnaires the approximate distribution by gender, age, and calendar period of first employment of workers in the companies retained for a full-scale study. We estimated the person-years of observation specific to gender, age, and calendar year, and applied the relevant incidence of lung cancer (Parkin et al 1997) in order to estimate the number of cases of lung cancer expected in each cohort.

Approximately 80 expected cases of lung cancer among exposed subjects would be necessary to obtain an 80% power to detect as statistically significant (at α level of 0.05) a 30% increase in mortality based on external comparison (Breslow and Day 1987).

OCCUPATIONAL EXPOSURE DATA

Availability of Historical Exposure Data

The availability and quality of historical exposure data was assessed from information gathered through company questionnaires and site visits by IARC epidemiologists and IOM industrial hygienists. Copies of raw results of measurements were obtained whenever available. Information about historical conditions and work practices was also obtained from the company representatives, who had many years of experience in the industry.

Measurement of Current Exposures

At the time of the survey, information about current work practices and conditions was obtained by observation and by discussion with company representatives.  Measurements of personal exposure to respirable dust and diesel exhaust were made using IOM respirable dust samplers fitted with foam inserts size-selective for particulate matter 10 µm or smaller in aerodynamic diameter (PM10). The IOM sampler cassettes were loaded with 25-mm quartz fiber filters, and the cassettes were weighed before the size-selective foam inserts were added. Each sampler was positioned in the breathing zone by clipping it to the worker’s lapel. The sampler was connected to a battery-operated pump by means of polyvinyl chloride tubing, and the sampling flow rate was set to 2.0 L/min. This flow rate was checked at the beginning and end of sampling and, when possible, at regular periods between these times. The sampler was worn for a full work shift to ensure that the measured concentrations represented daily average exposures. 

We anticipated that some workers might consider personal sampling an unacceptable intrusion, in which cases the sampler could be placed at a fixed location within the normal working environment. This was not necessary, however, and we were able to monitor all tasks in the conventional manner.

In railway transport, it was not possible to directly observe the work practices of the train drivers owing to the distances involved and limited space available on board. Sampling was restricted to short-haul locomotive driver - that is, drivers of shunting locomotives and those operating suburban goods trains in a circular route around the city.  At the end of sampling, the filter cassette was removed from the sampling head and the size-selective foam insert was removed and discarded. The cassettes were then analyzed gravimetrically to determine the respirable dust concentrations. 

Each filter sample was analyzed for organic carbon (OC) and EC using evolved gas analysis with a thermal optical sensor in accordance with National Institute for Occupational Safety and Health (NIOSH) method 5040 (NIOSH 1996). This work was carried out by David Dabill of the Health and Safety Laboratory, Sheffield, England. After analysis of the filter samples, the respirable dust concentrations and EC concentrations were calculated. The geometric mean (GM) and geometric standard deviation (GSD)of concentrations were calculated for each set of measurements from similarly exposed groups.


Airborne concentrations of EC are taken to be a reliable marker of diesel exhaust concentrations, and Verma and colleagues (1999) have shown this to be a suitable method for monitoring diesel exhaust exposures in the railroad industry. The American Conference of Governmental Industrial Hygienists (ACGIH) has now proposed a threshold limit value of 0.05 mg/m3 for diesel exhaust (ACGIH 2000), and until further clarification, this value is assumed to apply to the EC content of submicron dust particles.Use of the respirable dust samplers was not practical in the oil-shale mine, so cyclone-type respirable dust samplers (SIMPEDS samplers, Casella, London, UK) were used instead. 

These samplers were prepared for gravimetric determination, but losses from the filter media due to the brittle nature of the quartz fiber filters prevented it. The samples were analyzed for EC and OC, however, as described previously in this report. Ideally, we would have preferred to use a sampler incorporating a preselector with a submicron-sized cutoff for the oil-shale mine, but this type of equipment was not available at the time of survey. The measurements were therefore intended to be indicative only. 

RESULTS
FEASIBILITY OF EPIDEMIOLGIC STUDY

The company questionnaire was translated into the national languages. Selected sections were back-translated into English at IARC, and the back-translations were compared with the original English version. No important discrepancies were identified. A total of 16 companies were retained for analysis at the end of the feasibility study (Table 3): 7 companies were involved in nonmetal mining, 2 in railway transport, and 7 in bus transport. Slovenia was the only country in which no suitable companies were identified.

Table 3 presents also the estimated total number of workers and the number of workers exposed to diesel emissions who could be included in a historical cohort study in each company. Most workers exposed to diesel emissions were men. The total estimated size of the cohort exposed to diesel emissions was 46,500 male workers, of whom 55% were miners.The period of available follow-up and the year of first use of diesel engines were determined by country and type of industry (Table 4). We found that all groups of workers exposed to diesel exhaust could be followed for at least 25 years.The expected numbers of person-years and cases of lung cancer among male workers exposed to diesel emissions were determined (Table 5). The combined cohort of exposed workers would contribute more than 800,000 person-years of observation, yielding an estimated 295 cases of lung cancer. The cohorts of miners and bus workers would meet the target statistical power, while the cohort of railway workers would have a power of 70% to detect a 30% increased risk.

The 7 mines included in the study were involved in mining coal (1 in the Czech Republic and 3 in Slovakia), oil shale (1 in Estonia), bauxite (1 in Poland), and magnesite (1 in Slovakia). Exposure to silica, radon, arsenic, or other lung carcinogens was minimal or absent in all these mines.A number of companies in addition to those listed in Table 3 were contacted during the feasibility study, but they were excluded for different reasons. In particular, a Polish copper mine and a Slovakian iron mine were excluded because of possible coexposure to radon or arsenic. Two small truck transport companies in Latvia were excluded because they would have been the only companies from this industry. One railway transport com- any and 2 bus transport companies in Lithuania, as well as a bus transport company in Slovakia, qualified for the feasibility study, but the size of the cohorts and the number of expected lung cancer cases were small (fewer than 5 in each company). Drivers constituted a large proportion of the workers in bus transport companies, but we considered only garage mechanics as definitely exposed to diesel exhaust. Drivers were therefore excluded from calculation of the expected size of the exposed cohort in bus transport companies. Information on past smoking habits was available only for workers in the cohorts from the Czech Republic and Poland. 

Tables 3 and 5 report the results for male workers only. Although the companies included in the feasibility study employed many women, they were mainly in jobs not exposed to diesel exhaust. The only job with potential expo- sure to diesel exhaust in which a sizable number of women were employed was bus driver. Among bus mechanics, railway conductors and mechanics, and underground miners (the jobs with the highest potential for exposure in our study), the number of women was very small. In parallel to the field work in the 9 countries included in the feasibility study, contacts have been established with a salt mine in Romania (600 workers, of whom 200 currently are exposed to diesel exhaust) and a railway company in Israel (2,000 workers, of whom 500 currently are exposed to diesel exhaust). Although we have not reviewed the employment and exposure records in these companies with the same detail as for the other companies, these data might become available for a full-scale epidemiologic study.

AVAILABILITY OF HISTORICAL EXPOSURE DATA

The availability and quality of historical exposure data were assessed for all 16 companies that agreed in principle to participate in a diesel emissions exposure study. The results are presented here by industry.

Nonmetal Mining

Czech Republic: Coal Mining 

A great deal of exposure data for the Czech coal mining company was included in the feasibility study. Benzene, polycyclic aromatic hydrocarbons, and airborne dust were regularly monitored. The benzene concentration can be considered to be a specific marker of diesel exhaust exposure because there were no other sources of exposure, and these measurements may be used to understand the historical pattern of exposures. Measurements of EC concentrations from at least one of the coal mines will be necessary in order to provide a baseline measurement of current exposure.

Estonia: Oil-Shale Mining 

The available exposure data for the Estonian oil-shale mining company included measurements of carbon monoxide and nitrogen dioxide made in 1986 and repeated in 1998. A limited number of benzene measurements were also available. Measurements of respirable dust, being obtained by regular static monitoring at the mine entrance since 1970, were also available. Although the measurements were quoted as respirable dust, the sampling method was for total respirable dust. The local custom is to assume, on the basis of microscopic evaluation of the particle sizes, that all of the airborne dust is respirable.

The diesel-powered front loaders were mainly of Ukraine origin although some vehicles from Finland had been introduced. These vehicles were well maintained, and their diesel exhaust emissions were not considered to be excessive. The mine had strict rules for monitoring ventilation rates, and a preliminary inspection at the time of our visit indicated that the mine was efficiently ventilated. The ventilation conditions appeared to have been relatively stable since 1970.Work practices in the mine were such that diesel- exposed workers (loader drivers and support operators) were segregated by time from nonexposed groups such as drillers and shot-firers. It may therefore be possible to include the latter occupations in a control cohort.

Before 1995 the number of mine workers exposed to diesel exhaust was greater because approximately 12 manual workers were usually deployed in the vicinity of the diesel-powered loaders.Hungary: Bauxite Mining Regular (monthly) measurements were available only for carbon monoxide and nitrogen oxides in the Hungarian bauxite mine; sporadic measurements of total dust were also available. Although the carbon monoxide results might provide some information on expo-sure to diesel exhaust, an accurate assessment of past and current exposure would require further monitoring.

Slovakia: Coal Mining 

In the 3 Slovakian coal mines, the available data for carbon monoxide and oxides of nitrogen were considerable. These measurements were obtained by area sampling (rather than by personal monitoring). These measures probably will not be useful in assessing exposure levels because only the concentrations within a certain range of a specified limit value were usually recorded. Therefore, the measured carbon monoxide levels will pro-vide a positive bias in any exposure assessment. Further monitoring for EC concentrations would be required to establish current exposure levels.

Slovakia: Magnesite Mining 

The sample analyses available from the Slovakian magnesite mining company were limited to 8 polycyclic aromatic hydrocarbon samples collected in 1999, inside cabins of diesel-powered loaders used within the mine workings. Because of their limited nature, it is not clear whether these can be compared with any historical measurements. An accurate assessment of exposure for these workers will require additional monitoring.

Bus Transport
Estonia 

The Estonian bus company had very little occupational exposure data available for bus mechanics and none for bus drivers. None of the data were relevant in estimating past exposures to diesel exhaust particulate matter.

In 1991 the bus company introduced used buses imported from Sweden (Scania, Volvo). Prior to 1991 onlyHungarian buses (Icarus) were used. Although the Hungarian buses were quite old at the time of this study, they were still highly regarded and had been considered to be reliable vehicles when new. It may be concluded that exposures have been quite stable for the last 30 years or so with only modest changes recently.

In any case, the exposures of bus drivers appeared to be quite low and mainly influenced by the numbers of diesel- powered trucks and buses on the road rather than by the emissions from the bus that they were driving. There are significant differences in environmental pollution inEstonia due to burning of shale oil in power stations, and
the relative health effects may need to be evaluated.
Special measures have been in place since 1993 to control exposure to asbestos during work on brake parts.

Latvia 

The working conditions and usage of diesel engines in the 3 Latvian bus companies appeared to follow the same pattern as in the Estonian company. The majority of buses were of Hungarian production with some Western vehicles being introduced. Since the early 1980s, the maintenance program has included active tuning of the engines to reduce exhaust emissions. If necessary, the diesel fuel can be cleaned using a centrifugal filtering device. This means that low-grade diesel fuel can be improved on site, preventing high emission levels. 

Again, buses were stored outdoors, thereby eliminating high concentrations of airborne diesel exhaust within the garages when the vehicles are started up in the morning.Smoking is prohibited for both drivers and passengers on buses in Latvia and has always been so.

The bus depots included separate departments that covered general maintenance, body repairs, hydraulics, engineering, and so forth.  Consequently, not all of the workers were exposed to diesel exhaust. However, the employment records are sufficiently detailed to identify those workers who were exposed. The exposure data available for the bus companies were limited. Annual checks were carried out on airborne dust, welding exhaust, etc. These data will have no direct relevance in estimating exposure levels for EC but may be useful in further analyzing how conditions have changed over time. There are currently no specific measures for controlling asbestos exposure during maintenance of brakes, and this would have to be evaluated in terms of the potential to cause lung cancer.

Lithuania 

Very few measurements of carbon monoxide had been made at the Lithuanian bus company we studied, and these related only to the year 1998. No other exposure measurements relevant to diesel particulate matter exposure were available. Any future evaluation of exposure in this company would require additional monitoring.

Poland

Detailed exposure data, including regular monitoring of benzene, benzo[a]pyrene, carbon monoxide, and nitrogen oxides, were available since 1987 for 1 of the 2 Polish bus companies. In the other company, regular monitoring has taken place only for benzene and total dust.These measurements will be useful to assess past exposure patterns. It will be necessary to measure EC concentrations to assess current exposure to diesel exhaust.

Railway Transport

Latvia 

The Latvian railway depot we studied included a large repair and maintenance workshop. Locomotives (for shunting) were of Czech origin, while the passenger trains were of Latvian manufacture. The locomotives were highly regarded and the company’s standard of maintenance was high. By observation, there was no evidence of black smoke emissions from the exhausts, and even though shunting locomotives would maneuver rolling stock into the maintenance buildings, there was no noticeable exhaust in the indoor atmosphere. Smoking is allowed for train drivers.

The conditions also appeared to have remained stable for at least 30 years, so exposures would seem to have been generally low. Because the working conditions and work practices adopted by the Latvian railway drivers were very similar to those of the Russian cohort, the exposures to diesel exhaust particles can reasonably be assumed to be similar in magnitude.

Russia

Comprehensive information about work practices, historical uses of diesel engines, and exposure data in the Russian railway industry is available from the company. These data include 18 years’ worth of “soot” measurements from 1976 to 1994 obtained by static sampling in locomotive cabins. The sampling and analytic methods used to obtain these data are unclear, and the measured values seem too low to be simple measurements of total respirable dust. However, the data will be useful in assessing how conditions have changed over time if a year- by-year analysis can be provided.

Suvorov and Shterengarz (1986) found the soot in diesel emissions in Russian railways, at percentage of engine power, to be 83 ± 38, 79 ± 35, 90 ± 36, and 51 ± 14 mg/m3, respectively, for an engine that was idle, at 25% power, at 50% power, and at maximum power. On the basis of 200 to 300 measurements obtained over a 10-year period (1970 to 1980), they found the mean level of soot in the locomotive cabin of trains to be 0.109 (range, 0.00 to 0.169) mg/m3 and the mean level in cabins of shunting locomotives to be 0.133 (range, 0.00 to 0.363) mg/m3. Anecdotal information suggested that operational conditions have been stable from 1970 to present, but little information was available about conditions and work practices prior to this time.

A great many related measurements were available covering the period 1976 to 1994. These included measurements of benzene, hydrocarbons, polycyclic aromatic hydrocarbons, carbon monoxide, oxides of nitrogen, and oxides of sulfur. However, these measurements will be only marginally relevant in estimating exposures in terms of EC concentrations.

There were no indications that drivers or mechanics would be exposed to other agents that may cause lung cancer. Mechanics were exposed to chrysotile asbestos during some maintenance activities, but in attempts to measure the asbestos exposure levels, fiber concentrations were below the limit of detection (R Shterengarz, personal communication, 2000).

MEASUREMENT OF CURRENT EXPOSURES


Current diesel exhaust exposure levels were measured for train drivers in Russia and for bus mechanics, bus drivers, and oil-shale mine workers in Estonia. Approximately 50 samples were collected in total from 3 companies. The sampling was carried out in close cooperation with local collaborators and host companies, and the occupational groups were selected after consultation with all parties. The sampling program provided a consistent set of diesel exhaust exposure data of similarly exposed occupational groups.The data were used in the exposure reconstruction model, which might be applied to a full-scale cohort study. Tables 6 through 10 report the results of measurements in each industry, and Table 11 summarizes the results across industries. We have compared our results with those of other studies based on EC determination (Table 12).

Railway Workers

Personal air monitoring of respirable dust (PM10), OC, and EC was carried out for drivers of short-haul suburban goods trains (Table 6) and for drivers of locomotives carrying out shunting maneuvers (Table 7) in Russia. The local industrial hygienist reported that shunter drivers may have higher exposures than main line train drivers.This higher exposure was anticipated in part because drivers must lean out of the cabin window to see the rail track during coupling and uncoupling operations and when maneuvering rolling stock into sidings or sheds. We therefore distinguished between these shunter drivers and the main line drivers operating short-haul suburban goods trains in a continuous circular route around the city.

The measured range of PM10 concentrations for the train drivers was 0.34 to 1.60 mg/m3 with a GM of 0.8 mg/m3; the EC concentrations ranged from 10.6 to 71.8 µg/m3 with a GM of 18.3 µg/m3 (Table 6). Two of the train drivers had much higher exposures than the rest of the group (samples OM09 and OM11). The ratios of OC to EC were similar for all samples, so it is unlikely that these higher exposures were due to contamination. Since these high-exposure samples were obtained from 2 drivers working together, their work presumably differed in some way from that of the others.

The shunter drivers had exposures similar to those of the local train drivers. Their PM10 concentrations ranged from 0.51 to 1.54 mg/m3 with a GM of 0.80 mg/m3. EC concentra- tions ranged from 5.3 to 30.8 µg/m3 with a GM of 12.3 µg/m3 (Table 7). Several of the sample EC concentrations for the shunter drivers were very low at 5.3, 6.4, and 8.9 µg/m3. The nature of the work that these drivers were performing is unclear, but the exposures indicate general background levels for an urban environment (Zaebst et al 1991).

The quoted concentrations were obtained over 12-hour work shifts and as such are a measure of concentrations across the full work shift. The measured concentrations compare well with the EC concentrations quoted by Verma and colleagues (1999) for Canadian railroad drivers (se eTable 12). This suggests that diesel exhaust exposures of railway workers in Russia currently are the same as exposures in most Western countries.

Bus Mechanics and Drivers

Personal air monitoring of respirable dust (PM10), OC, and EC was carried out for bus mechanics involved with maintenance and repair work (Table 8) and for bus drivers (Table 9) in a major urban bus company in Estonia. The vehicle mechanics were expected to have higher diesel exhaust exposures than the bus drivers. Although local exhaust ventilation systems were provided for static testing of engines, vehicles were regularly being moved around, resulting in a gradual buildup of diesel exhaust inside the workshops. The garages were used only for repair and maintenance. Buses were stored outdoors, a practice that avoids the production of high indoor concentrations of exhaust when buses are started up before being used.

In these garages the mechanics were a separate group from welders and therefore not exposed to welding fumes or other airborne contaminants that would interfere with the exposure assessment. The measured range of PM10 concentrations for the mechanics was 0.45 to 2.21 mg/m3 with a GM of 1.02 mg/m3; the EC concentrations were in the range of 24.8 to 52.1 µg/m3 with a GM of 37.6 µg/m3 (Table 8). The bus drivers had lower exposures than the mechanics. Their PM10 dust concentrations ranged from 0.40 to 1.01 mg/m3 with a GM of 0.58 mg/m3, and their EC concentrations were 5.7 to 10.5 µg/m3 with a GM of 9.3 µg/m3 (Table 9).

All exposure measurements were obtained over 8-hour work shifts and as such are a measure of concentrations across the full work shift.A comparison of these exposures with those of truck drivers and mechanics in the United States as quoted by Zaebst et al (1991) indicated that the exposures for Estonian bus drivers were similar to those of US truck drivers (see Table 12). This suggests that the main source of diesel exhaust exposure was the general urban environment.

The mean EC concentration of 37.6 µg/m3 measured for bus mechanics in Estonia is higher than the mean value of 12.1 µg/m3 quoted for US truck mechanics (Zaebst et al 1991). This is most likely due to the differences in working environments between truck and bus repair depots rather than differences between Estonia and US work practices


Oil-Shale Miners

We visited a large oil-shale mine in Estonia intending to carry out personal sampling, but production had temporarily ceased owing to technical problems. We toured the production areas, however, and the mine safety engineer explained the work practices and production methods.The Estonian collaborators carried out the sampling at a later date under normal production conditions.

Oil shale is deposited in bands separated by layers of limestone. The limestone contains some crystalline silica (quartz). Owing to the nature of the work, current dust and silica exposures appear to be rather high.The mining method in this case was the pillar-and-room technique. The oil-shale seam is drilled and blasted, and the resultant material is transferred to a conveyor by means of diesel-powered front-end loaders.

The diesel-exposed workers (loader drivers and support operators) were segregated by time from nonexposed groups such as drillers and shot-firers (workers preparing the explosive charges). Therefore, the loader drivers were not exposed to airborne dust from any other operation; however, this in itself appears to be a significant source of dust exposure. The mine utilized water sprays on conveyor transfer points, but it was not clear whether any dust suppression was applied to the loose material as it was loaded by the diesel-powered vehicles. The mine workers were supplied with disposable face-piece respirators that filter dust, but anecdotal evidence suggested that they only wore these respirators for the very dustiest of tasks.

The technical problems with sampling for EC as an indicator for diesel exhaust in coal mines are well known. Diesel exhaust particles are largely submicron in size, and larger carbon-based compounds (such as those present in coal dust) interfere with the analysis. Usually, an impactor or a similar preseparator with a submicron cutoff is necessary to separate out the larger particles. This option was not practical at the time of survey, so cyclone-type respirable dust samplers were used to obtain an estimate of exposure levels. Errors are assumed to be inherent in this approach.

Estimates of PM10 were possible from only 2 of the 5 samples; other filters were rejected because of losses from the filter media (Table 10). For the 2 samples for which estimates were possible, the concentrations were 4.07 and 2.73 mg/m3. These measures are likely to underestimate the true concentrations, however, because we suspect losses from these samples as well.

Analyses for EC were possible, however, and the measured concentrations were in the range of 96.5 to 377.2 µg/m3 with a GM of 202.4 µg/m3 (Table 10). This compares well with previous measurements of EC in zinc and potash mines carried out by others (Haney and Fields 1996; Dahmann and Bauer 1997; Stanevich et al 1997).Diesel exhaust exposures by industry are summarized in Table 11 and compared with previous studies in Table 12.


RETROSPECTIVE EXPOSURE ASSESSMENT
Development of Exposure Assessment

Information collected via the company questionnaire and site visits was summarized, and a common strategy for retrospective assessment of exposures was devised. The method is based on information about the work tasks and the work environment. The structure of the method reflects the process of concomitant generation at source, subsequent dispersal into the work environment, and finally, interaction of the worker with the pollutant. For each task, possible sources of the hazardous substance are identified and categorized in relation to the worker as either near-field or far-field sources. Near-field sources are those within 1 m of the worker in any horizontal direction; far-field sources are the others. For each category of source, the active emission is estimated from 3 key factors: the intrinsic emissions, the handling or processing associated with the source, and the effect of any localized engineering controls that were present at the source. These parameters define the concentration of substance emitted to the near field or far field.

The method has been developed and its use validated for different substances known to cause or suspected to cause lung cancer. Diesel particulate matter is not validated, however, although further consideration was given to how the method could be adapted to assess such exposure.

The questionnaire information was evaluated for likely availability of exposure measurements and other relevant information. In particular, we aimed at assessing the type of diesel engine used, engine maintenance schedules, fuel used, and workload demanded of the engine. We also assessed the proximity of workers to the engine exhaust, the size of rooms occupied by the workers, the likely general ventilation conditions, the use of any control measures such as local ventilation or enclosed cabs, the duration of engine running, and whether the engine exhaust vented inside a building.

Quantitative exposure assessments were considered practical in companies with good contextual exposure information, preferably with exposure measurements, although additional measurements of current exposure could be collected. A ranked categorical assessment of exposure level was feasible when the information was of poorer quality. Plants with little or no information about the type of exposure were excluded.

Evaluation of Proposed Exposure Reconstruction Method

The validity of the proposed exposure reconstruction method was evaluated using the exposure data obtained during the visits. The reconstruction method, developed in other studies (Cherrie et al 1996), was adapted by us to deal with the special circumstances of workplace diesel exhaust exposure. Further guidelines will be produced to support the assessments for diesel particulate matter. An experienced industrial hygienist was presented with descriptive information for the sampled occupations and asked to estimate exposure levels using the method. These data were then compared with measured values. Provided the correlation was good (that is, r > 0.7) between the measured and estimated exposure levels, we considered the method suitable for a subsequent study.

Theoretical Model of Diesel Exposure

We have developed a theoretical description of theexposure process from which we plan to reconstruct past exposures in a full-scale epidemiologic study. This builds on work for an earlier epidemiologic study investigating the risks of lung cancer from man-made vitreous fibers and for other research currently underway (Cherrie et al 1996;Cherrie 1999; Cherrie and Schneider 1999).

In this scheme we use EC as a marker for diesel particulate matter. The exposure level is defined as the average concentration of EC inhaled by a worker during a task or some other defined period of time. The exposure level (C) is the sum of contributions from a multiplicity of back- ground sources (CB); from other more local sources, that is, in the far field (CFF); and from sources in close proximity to the workers, that is, in his or her near field (CNF): 

An arbitrary distinction is made between the worker’s near field, which is defined as a cubic space with dimensions 2 × 2 × 2 m centered on the subject’s head, and the far field, which is the remainder of the immediate environment. For indoor spaces the far field is defined by the boundary walls of the space, while outdoors the far field is defined as being within 10 m of the subject. In modeling exposure we have normally assumed that emissions are dispersed uniformly from the source in all directions. However, this is not the case with diesel emissions because of the high initial velocity of the exhaust gases.

The background concentration of EC, which arises from diesel engines and other sources in the general urban environment, depends on many factors, including the number of vehicles actively emitting, the types of engine, and meteorologic conditions. The background concentrations probably are well represented by fixed location monitors situated some distance away from localized sources (such as busy roads) and are relatively constant at any one time across a town or city.

The far-field contribution to diesel exhaust exposure may come from the exhaust plume of the vehicle being driven by the individual or from other vehicles in the general vicinity.

The intensity of any source is determined by 3 factors: the intrinsic emission, the handling or processing (h), and the effectiveness of any localized controls. In the case of diesel particulate matter, the intrinsic emission is determined by the age and type of diesel engine being used and corresponds to the concentration measured about 1 m from the exhaust outlet, downstream, and out of the exhaust plume. The handling would be described by the way in which the engine was being driven, and the effectiveness of local controls would be determined by the presence of any particulate traps on the tailpipe. We have assumed a further reduction of exposure in a vehicle with an enclosed cab.

The 3 parameters, that is, Ei, h, and (1 − n1c), are multiplied together to provide the active emission of the source (Ea). Note that the efficiency of local controls is expressed as a fraction, and that the multiplier used to obtain the reduction in active emission due to the controls is expressed as 1 minus the efficiency of the controls. Three further parameters are incorporated into the basic model: the fractional time the source is active (ta), the efficiency of any personal (respiratory) protection worn by the subject (nppe), and the gravimetric dispersion of the emissions from the source (dgv). So, for a single source close to a worker, the near-field exposure level (CNF) would be as follows: 

The passive or fugitive emission (Ep) generally represents emission from resuspension of settled dust or evaporation of spilled volatile liquids. In the case of diesel exhaust,  would correspond to fugitive emissions from the engine exhaust system that may diffuse into the environment where the worker was located. For example, if the engine were mounted under or in front of the driver’s cab in a truck. In addition, although workers can wear some form of respiratory protection against diesel particulate matter, in practice this is not done.

For a source in the far field of the worker, similar considerations apply, and so the following equation gives the far-field contribution to the exposure level: 

Dispersion of the pollutant depends on the proximity of the source to the person exposed (that is, whether the source is in the person’s near or far field) and on the directional nature of the initial dispersion. In addition, if the source is inside a building, then the volume of the enclosed space and the quantity of general ventilation will also determine exposure level.

Cherrie (1999) has presented information about dispersion from the source, which is reproduced in Table 13. For sources within the worker’s near field in a large, poorly ventilated room, the general ventilation multiplier would be 1.5. Assuming no sources are in the far field, then there would be no far-field component of exposure level (that is, ). If the source were in his far field, then the ventilation multiplier would be 0.5. Also, for a small, poorly ventilated workroom, the general ventilation multiplier would be similar regardless of whether the source was in the near or far field (that is, 15 vs 14).

For outdoor diesel sources in the far field, we have assumed a Gaussian dispersion pattern. This allows the reduction in concentration to be estimated along with an adjustment of the ta to allow for variation in the plume trajectory because of changes in wind direction. For example, moving away from the source is assumed to reduce the particulate concentration approximately according to the inverse square of the distance from the source. In addition, we assume that the direction of the exhaust plume may vary according to the wind direction.The probability that it will blow from the source toward the worker is reflected in the model in ta. If there is no preferred wind direction in relation to the line between the source and worker, then it is assumed that ta,FF would be best estimated as the angle of expansion.


Using Theoretical Model to Predict Exposure
We have used the model to estimate average exposure level for the 5 groups of workers for whom measurements of exposure were undertaken. Estimates were made as EC with no adjustment made to estimate 8-hour time- weighted average level. The exposure assessments were carried out by one of us (JWC), who was unaware of the measurement data at the time of the assessments. Information about the exposure circumstances was obtained partly by observation of work activities (train drivers in Russia) or by descriptions from another investigator who had visited the sites, and partly from the questionnaire information or other descriptive information acquired during the study.

Little objective information was available to determine the magnitude of parameters that should be input to the model, and so most assignments were based on the judgment of the assessor. Many of the samples were obtained for drivers who, because of the nature of their job, were away from the investigators for most of the sampling duration; thus it was not possible to obtain detailed contextual information about the circumstances of the exposure. For this reason the exposure assessments were completed for all of the broad job categories identified, rather than for each individual measurement.

Comparison of Model Predictions with Measured Exposures

The estimated EC exposures ranged from 5 µg/m3 for the bus drivers to 500 µg/m3 for the oil-shale miners (Figure 1). For local train drivers, the main determinant of exposure was judged to be the time that the driver spent in the station while the train was being loaded and unloaded.This activity, although it was judged to take only approximately 25% of the work time, accounted for approximately 70% of the estimated exposure for these drivers. The remainder of their exposure was considered to come from background sources within the city (that is, CB). It was assumed that while the train was in motion, the driver was not exposed because the exhaust plume would be directed away from his cab. 

The mean measured exposure level for the local train drivers was 18 µg/m3, but excluding 2 outlying points reduced the average to 14 µg/m3. These data compare well with our estimated level of 16 µg/m3. We are unsure of the reason for the 2 outliers, but it is likely that these drivers worked part of their shift in an enclosed or semi-enclosed space such as an engine shed or tunnel.

The estimated exposure level for the shunter drivers (61 µg/m3) was higher than the measured level (15 µg/m3).The estimated level for these workers was mostly deter mined by circumstances: when the train was moving so slowly that the exhaust plume might blow toward the driver’s cab and when the driver had his head out of the cab window. From our observations we estimated that the train was either stationary or moving backward about half of the work shift and that the driver had his head out of the window for about a quarter of this time. If he had spent only 25% of his time moving slowly and 12% with his head out of the window, then the estimated exposure would have been 21 µg/m3.

The estimated exposure level for the bus mechanics was 43 µg/m3, and the mean measured exposure level was 39 µg/m3. The main source of exposure for this group was judged to be exhaust from buses moving around the garage (about 8% of the work shift). Drivers were assumed to be exposed for a longer period of time, but their exposure was lower because we assumed that the cab reduces the exhaust entering the driver’s near field. The estimated exposure level for bus drivers was 5 µg/m3, and the average measured exposure level was 9.5 µg/m3.

For the oil-shale miners, the estimated exposure level was 500 µg/m3 and the average measured exposure was 220 µg/m3. The main reason for this group having the highest exposures was their relatively continuous work in a confined space. Part of the reason for this overestimation may have been the difficulty in judging the impact of general ventilation on controlling exposure. We used the available data for large, well-ventilated spaces, shown in Table 13, although the actual ventilation rates were much higher than this.

The association between the average measured and estimated exposure levels for all jobs combined (Figure 2) was similar to that seen in another comparison exercise that we had undertaken (Cherrie and Schneider 1999). Although the association between measurements and estimates was fairly good, there was a tendency to overestimate rather than underestimate exposure.

Use of Model to Reconstruct Exposure

The model provides a scheme for describing exposure in a way that can help in estimating past exposure. The estimates clearly depend on the quality of information about the work and the work environment. Information about the work processes can be obtained from records and by interview with long-service employees or retired workers. This information may then provide the basis for reconstructing exposure levels. 

Combining measurements of current exposure levels with the exposure reconstructions provides an opportunity to minimize bias in the estimates and to refine the magnitude of component factors in the theoretical model. Further experimental investigations may result in refinement of the model parameters. Historical background concentrations of diesel particulate matter could be estimated from information about diesel traffic density obtained either from historical records or from interviews with knowledgeable local people.
