Mortality among asbestos-exposed workers in a railroad workshop.

C. Mortality among asbestos-exposed workers in a rail- road workshop. Scand J Work Environ Health 10 (1984) 283-291. The mortality experience of a cohort of 3 297 railroad maintenance shopworkers exposed to asbestos was investigated. The study period was 1951-1980, and the vital status was assessed for 99.6 % of the men. Individual estimates of cumulative asbestos exposure were based on detailed records on work tasks and divisions. Dust measurements were scanty in earlier decades, and estimates of average fiber levels were therefore based on information on the amount and kind of asbestos used, job descriptions, and interviews with older workers. The overall mortality was lower than expected from the national death rates (standardized mortality ratio = 82). The mortality from lung cancer increased as cumulative exposure increased in consistent dose-response relationships. Employment times of less than 30 years in workplaces with moderate levels of mainly chrysotile asbestos was not associated with any apparent increase in the risk of lung cancer. A subgroup exposed for more than 30 years in workplaces repairing steam engines, where amphiboles were used as well, had a standardized mortality ratio of 192 for lung cancer. This figure mav be an underestimation due to healthy worker selection and fewer smokers than normal. The "true" standardized mortality ratio was estimated to be about 300. Five cases of mesothelioma were observed.

figure 1 -56 % had been employed for more than 10 years and only 16 % for less than one year. The workers used to stick to a specific division, and changes, eg, from the steam engine division to the passenger-coach division, were not common. Before the middle of the 1950s special permission was required for a person to pass through other divisions. Furthermore, for a long time, the requirements for employment demanded that persons to be employed should be in good health; have a good sense of hearing, straight posture, and military service fulfilled; and be no more than 24 years of age.

Vital status of the cohort members
The study population consisted of 3 442 men, and their vital status was traced through several sources, as displayed in figure 2. Some 2 196 subjects were found through the National Health Service, which keeps records of all living persons in Sweden by a tendigit identification number based on time of birth. The remaining 1 246 were checked through other public population registers, and the death and burial books of the parishes. Three men had emigrated. Information on address was missing or obviously wrong for 205 subjects, but 191 could be traced through local registers. Only five of these men were alive.
In total, the vital status of the study population could be assessed for 99.6 % of the subjects, and the remaining 14 subjects were excluded from the calculations, thus based on 3 297 subjects.
Death certificates and thereby the officially classified underlying cause of death (15) was obtained for 848 of 925 men deceased in 1951-1980, which was chosen as the study period. Valid classification according to rules of the World Health Organization for underlying causes of death were introduced in 1951. This procedure excluded 128 men who had died before 1951. The death certificates could not be found for 38 men, and for them the underlying causes of death were coded from the diagnoses given in the death and burial books according to the same classification rules.
The expected number of deaths was calculated by the multiplication of the person-years of observation within five-year age categories during respective single calendar years of the study period by the cause-, gender-and calendar year-specific national death rates. The relative risks were calculated with a requirement of 20 years' latency time for the total cohort, as well as for the subcohorts of different exposure categories, ie, the time from first exposure to the beginning of the individual observation period under risk. A computer program -EPILINdeveloped at the University of Linkoping for calculating expected numbers and relative risks with 95 % confidence intervals was used (2).
The relative risk, ie, the ratio between observed and expected numbers, or the standardized mortality ratio (SMR), is reported for all causes of death, lung cancer, gastrointestinal cancer, and nonmalignant respiratory diseases. Chronic obstructive lung disease could mainly be assumed to reflect smoking habits and is therefore shown separately.

Steam engines
Asbestos has been used as insulation material at the workshop since 1902. In the 1920s and 1930s special padded insulation, so-called mattresses, containing crocidolite was used. During the Second World War these mattresses were filled with chrysotile. "Magnesia lagging," which was probably a mixture of amosite and magnesia, was periodically used as a substitution for mattresses. This "lagging" was enforced by firm sheets of chrysotile and with wire nets wound around the steam boiler. As there was a lack of imported asbestos during the Second World During the maintenance of the steam boiler the War, mineral wool was used as a substitute.
insulation material was loosened, and some of the The steam boilers were maintenanced regularly, in padded asbestos could be torn and spread around. general every sixth year, and broken up after 25 years After being dismantled, the steam boiler was carried of service.
to the platers' shop, where the remaining asbestos was brushed away with long brushes of steel. According to statements from several old workers this part of the shop was very dusty and caused rays in the light from the windows in the ceiling. It has, however, not been possible to assess the amount of crocidolite from torn mattresses. The asbestos material that was brushed away was probably mostly amosite and chrysotile.
It seems likely that the mattresses containing crocidolite could have appeared in the shop as late as 1950. During the first half of the 1950s the steam engines were gradually replaced by electric ones, and the last steam engine passed through the maintenance shop in 1955.

Passenger coaches
The walls of the passenger coaches were spray-insulated with crocidolite during 1961-1971. Before 1961 the coaches were insulated with cork, or with mineral wool, and from 1971 on with polyurethane foam. The heating equipment of the coaches was insulated with yarns of chrysotile wound around the pipes. The exhaust pipes of the motor coaches were also insulated in this way. The main source of dust was caused by the dismantling of old and dry asbestos, particularly when it was shaken and brushed away. The water tanks at the end of the passenger coachs were insulated with crocidolite. The ventilation pipes contained amosite, and amosite mixed with chrysotile was used in the equipment for air conditioning. shop. Such tasks were the punching of gaskets from electric engines, and insulating with asbestos powder in the casting and insulation of the electrical works of the engines.

Exposure data from the 1970s
Fiber counts from personal and stationary air samples made with standardized techniques were available in Sweden from 1970 on. During the period 1970-1980 the fiber counts were relatively low. Of 64 air samples from this period, only eight yielded counts higher than 1 fiber/ml. However, the measurements at this workshop aimed at giving evidence of improvements initiated by the industrial health organization rather than at being representative of the exposure situation.

Exposure in different divisions
The estimations of the exposure in the various divisions of the workshop were performed in consultation with a reference group of representatives from the staff management, the local trade union, and a former supervisor who had experience of the work practices in the 1940s and 1950s. The estimates were based on information on the intensity of production, the amount and kind of asbestos used during various periods, and a detailed report by the industrial health organization on the handling of asbestos. Furthermore, descriptions and recollections by exemployees were taken into account in estimates of the exposure of various work tasks, the distance between the worker and various dust sources, and the impact on other divisions from air contamination.
With all these factors taken together, an individual classification of the exposure intensity for every year could be achieved. The intensity was graded according to an ordinal scale with scores from 0 to 4 according to the following criteria: 0 -no asbestos exposure 1 -irregular exposure of low intensity 2 -regular, low exposure nore after the first date of employment (latency time) (SMR = stan- This basis for classifying exposure differs from the commonly applied practice of using data from dust samples. As there is no reliable factor of conversion from dust measurements as milligrams per milliliter or millions of particles per cubic foot to fibers per milliliter (6), the air concentrations in different parts of the shop have been related to other trades with known fiber concentrations. American insulators form the basis for comparison with other trades in a n extensive report (11). The workers with the highest exposure in the railroad maintenance shops, ie, those working with the insulation of steam engines and passenger coaches, were estimated in this report to be as heavily exposed as insulators with a historical exposure of around 10-20 fibers/ml, on the average (9).
The exposure in the workshop of the present study was estimated to have been about 3-5 fibers/ml, on the average, during previous decades. This level would correspond to regular low-medium exposure, ie, to a dust score of 2 or 3. The exposure of the men who worked with steam engines and passenger coaches, ie, heavy exposure estimated at 10-20 fibers/ml, would consequently correspond to a dust score of 4. Attempts to reach a crude conversion to fiber counts is displayed in table 1.
Information on the employment times and the dust scores of every division during the respective calendar year for every cohort member form the basis for the exposure categories. Two aspects of the exposure are elucidated, cumulative exposure, eg, the sum of all employment times multiplied by the dust scores and the number of years in highly exposed work, eg, dust scores 3 and 4.

Results
The overall mortality for the whole cohort was lower than expected from the national death rates, SMR = 82 (table 2). The same tendency towards decreased mortality was seen for gastrointestinal cancer and nonmalignant diseases of the respiratory tract, especially for the subgroup chronic obstructive lung disease (SMR = 36). However, the relative risk for lung cancer was increased (SMR = 116).
The mortality pattern was basically the same with a requirement of 20 years of latency and follow-up only to 79 years of age (table 2).
A subcohort of men employed for 30 years displayed an increased risk for lung cancer, SMR = 154 (table 3). Neither the risk of overall mortality nor specific causes of death were influenced to any appreciable extent by the length of employment, except for that of short-term. Among men employed 15-30 years, the overall mortality was higher than within the other strata because of an excess mortality from ischemic heart disease.  Cumulative exposure as the basis for categoriza-sure, but no dose-response relations were found for tion yielded a more marked dose-response relation-other causes of death. ship, the SMR for lung cancer increasing from 27 to Number of years in heavily exposed work (r 3 162 (table 4). The nonmalignant diseases of the respi-dust scores) as a basis for categorization demonratory tract also showed a tendency towards an in-strated an even greater dose-response relationship for creased relative risk from increased cumulative expo-lung cancer, increasing from an SMR of 47 to one of 189 in the highest exposure category (table 5). The same tendency was seen for other diseases of the respiratory tract. Table 6 displays the mortality pattern of a subcohort employed before 1939. The relative risk for lung cancer was relatively high (SMR = 188). The marked undermortality from lung cancer among men employed in 1939-1954 could be explained by the fact that unproportionately many short-term employees entered the cohort during the Second World War.
The mortality pattern of cohort members ever employed in divisions where steam engines were repaired is displayed in table 7. A marked excess mortality from lung cancer (SMR = 192) is shown for the highest exposure category. The relative risk for lung cancer was clearly higher in the two highest exposure categories within this subcohort compared to the same categories within the rest of the cohort (table 8).
In comparison to other asbestos-exposed cohorts there were few cases of mesotheliomas, only five. These have not been included among the lung cancer cases displayed in the tables. Three of the five cases had exposure to amphiboles from steam engine insulation noted in the records. The average length of life of the mesothelioma cases was 76 (range 70-82) years, the average latency time was 44 (range 30-57) years, and the average employment time was 32 (range 26-37) years.

Discussion
This study demonstrates a moderate increase in lung cancer mortality in comparison t o the national death rates. The lung cancer risk increased by dust exposure categories in a consistent way, which can be interpreted as a dose-response relationship for asbestos exposure and the risk for lung cancer.

Expected numbers from national death rates
The overall mortality for the whole cohort was only 80 "lo of that expected from the nationa! death rates.
Such an undermortality for work groups compared to the general population is often seen and has been referred to as the "healthy worker effect" (a), but it signifies mainly a lack of validity in the comparison (14). Good health was required for employment at the workshop in this study, and this circumstance further limits the usefulness of national death rates. However, no other major source was available as a comparison group.
The age-standardized regional incidence of lung cancer was 72 % of the national incidence in 1975-1980 (4). The SMR for lung cancer among the workers with low exposure was around 50. The striking undermortality among the workers employed for less than one year can be attributed to short-term trainees who later became nonexposed salaried employees. All these indications point to the conclusion that this cohort would have experienced a lung cancer mortality distinctly below the national average if it had not been exposed to carcinogens.
The relative risks for the subcohort with the highest exposure should therefore be adjusted upwards in accordance with the low base-line risk from which the cohort started. The true risk for lung cancer in the category with the highest exposure may therefore be estimated to have been at least threefold that of the nonexposed workers.

Smoking habits and lung cancer
Smoking habits are important to consider when lung cancer risk is being studied. Information on the smoking habits of each individual was not available. However, within a sample of workers participating in a medical examination at the end of the 1970s, 36 "lo had never smoked, and this percentage is a high proportion of nonsmokers in comparison to the general Swedish population.
The mortality from chronic obstructive lung disease could be used as another indication of the smoking habits in earlier decades. Tables 2-8 show a marked undermortality from these diseases, a finding which supports the suggestion that the entire cohort had a higher proportion of nonsmokers than the general population also in earlier decades. In earlier decades smoking was not permitted in workplaces. This finding is in agreement with the low risk of lung cancer among the lowly exposed workers.

Exposure standards and dose response
The SMR for lung cancer was higher for the highest exposure category, which was determined on the basis of intensity, ie, dust score 3 or 4, or cumulative exposure, ie, score times years, compared to categorization based on years of employment. Thus the dose-response relationship was stronger when the exposure standard was based on intensity rather than time.
The increased risks for lung cancer were comparatively modest, and this finding can be ascribed to a relatively moderate average exposure, as well as to the low amount of smoking among the workers. However, the prevalence of pleural plaques was about 50 % in a group of 323 active and retired workers that participated in a voluntary, health examination in 1982.

Amphiboles and chrysotile
The impact of different fiber types on lung cancer risk could not be studied in detail due to mixed exposures and lack of historical information. The highest relative risk for lung cancer was determined for a subcohort of men who had been employed in divi- Observed and expected numbers of deaths before the age of 80 years for the subcohort who had worked at any time in the ment times and 20 years or more after the first date of employment (latency time sions where the repair of steam engines took place and who had a total employment time of at least 30 years. This type of work involved exposure to amphiboles, which appear to be stronger carcinogens than chrysotile (1, 3, 7). Five cases of mesothelioma were observed. This is a low number compared to that of other large cohorts exposed to mixed types of asbestos. Three of the cases had exposure to amphiboles from steam engine repairs but, according to the records, two did not. Mesotheliomas have also been reported among men exposed mainly to chrysotile (5).

Miscellaneous
A weak increase in the number of deaths from pancreas cancer was observed among subjects employed 15-30 years. It is possible that some cases of peritoneal mesothelioma might have been misdiagnosed as pancreis cancer (13). However the increased mortality from this disease was not related to the degree or length of exposure, and therefore a causal association is less likely.
In summary, this cohort study of 3 297 male workers exposed to asbestos in a railroad workshop dem-onstrated a comparatively low overall increase in mortality from lung cancer in comparison t o national death rates. However more than 30 years of mixed asbestos exposure was associated with a threefold increase in lung cancer risk. Five cases of mesothelioma were observed. A subcohort of men employed for less than 30 years and without exposure to the highest dust intensities had no apparent increase in the risk of lung cancer. Thus, in this railroad maintenance shop after the Second World War, employment, other than steam engine repairing, was not associated with any apparent increase in the risk of lung cancer. But this disease, and especially mesothelioma, requires long latency periods from the onset of exposure to diagnosis (death). Therefore many asbestos-related cancers may be found in the future follow-up of the cohort, and this possibility requires that the conclusions be interpreted with care.