Mortality and disability among granite workers

KOLARI A. Mortality and disabilityamong granite workers. Scand J Work Environ Health I3 (1987)18- 25. The objective of the present study was to investigate the mortality, disability, and long-term morbidity of granite workers. The study included 1 026workers hired between 1940and 197I and followed until the end of 1981.The total number of deaths was 235, and the expected number was 229.7. Excess mortality rates were observed for respiratory dis eases (observed/expected = 28/13 .9). The number of tumor deaths was 46 (expected 44.9). Excess lung cancer mortality was evident at 15to 35 years of latency; the observed number of lung cancer deaths for the follow-up period of 25 to 29 years was 8, while 2.1 were expected. Mortality from cardiovascular diseases and violent deaths was slighty lessthan expected. The results for disability and long-term morbi dity showed elevated incidence and prevalence rates for respiratory diseasesand rheumatoid arthritis. The observed number of disability pensions due to rheumatoid arthritis in 1981 was 10 observed versus 1.8 expected, and the observed number of patients granted free medication was 19 versus 8.1 expected. The results indicate that granite dust exposure per se may be an etiologic and pathogenetic factor for lung cancer, cancer of the gastrointestinal tract, and some extrapulmonary nonmalignant chronic diseases.

The relation between exposure to inorganic dusts such as silica and the development of lung disease is well recognized. Most research, both epidemiologic investigations and studies related to the pathophysiology of silica-induced disease, has been concerned with silicosis. Silicosis may be complicated by pulmonary heart disease, bronchitis, emphysema, and spontaneous pneumothorax and segmental middle lobe collapse (27).
Lung cancer occasionally occurs in silicotics, but there is incomplete evidence of a cause-effect relationship between cancer and silicosis or siliceous dusts (9,12,16,29,35,36). Even fewer epidemiologic studies have been done on workers exposed to "pure" silica (8,33). In addition, methodological difficulties have hindered the detection of possible associations between silica exposure and cancer.
Besides tuberculosis and cancer of the lung, several other diseases, such as progressive systemic sclerosis, rheumatoid arthritis, systemic lupus erythematosus, and nephropathy, have also been ascribed to silica exposure (7, 13, 30 and our own unpublished results).
In 1972 the Finnish Institute o f O ccupational Health carried out a field study on granite workers . The study comprised a health screening program for current workers and a survey of occupational exposures (including dust, vibration, and noi se) (I). In addition, disability and mortality were studied in a cohort of gran- ite wo rkers hired in 1940-1971 (17, 20). This cohort, wh ich was composed of workers with fairl y pure quartz du st exposure, has now been followed for at least ten yea rs and investigated for mortality, disability and chronic d isea ses.

Subjects
The stud y cohort comprised I 026 Finnish granite workers employed in quarries and processing yards from three regions (Vehmaa, Viitasaari, and Kuru) in the middle and southwest parts of Finland for at least three months between 1940 and 1971. The workers' ages ranged from 15 to 72 (median 27) years at the time of entry into the cohort. The mean length of exposure to quartz dust was about 12 years .
Each worker's name, date and place of birth, and work history at the companies were collected from the employers' personnel records . The persons were traced through the Population Data Register, and all of them could be traced.
The mortality of the cohort has been followed up three times, until the end of 1972, 1975, and 1981 The causes of death were ascertained from death certificate s according to the eighth revision of the International Classification of Diseases (IC D) (26). Even though the calculations of the mortality rate s were based on the prima ry cause of dea th , the other (second ar y) causes were also coded for the study of disease combinati ons a nd case analyses. Th e dea th certificates were obtained from the Cen tral Statistical Office of Finland and, for tho se who had emigrated to Sweden, from Swedish Statistics. The disability of the granite workers has been follow ed up twice, until the end of 1971 and 1981. The causes of disabili ty, coded according to the lCD, were ob ta ined from the Social Insurance Institution. Diseases for which medication was free according to the national sickness insurance law were similarly followed until 30 June 1981. Thi s information was also obtained from the Social Insurance Institution.

Basis for comparison
Th e official statistics on cau ses of death in 1965 (fo r mortality from 1940 to 1972), in 1969 (for mortality from 1940 to 1975), and in 1972 (for mortality fro m 1940 to 1981) were selected for the comparison with the general male population (37,38,39). These yea rs repre sented the med ian for th e years of deaths during the respective follow-up period. Period -specific mortality result s were also calculated according to years since ent ry (latency periods), in five-year periods . In this analysis, the mortality figures from 1972 were used for the comparison . The median years of all deaths were selected for comparison becau se in our earlier studies the expected numbers, ba sed on the med ian year of deaths, correlated well with the calendar periodspecific expected numbers. In add ition , we calculated, the med ian year s of canc er deaths and cardiovascular deaths separately, the maximum deviation being three years (1975).
The prevalent num bers (as of 31 December 1981) and incidence rate s (from 1969 to 1981) for the disability pen sions of the granite workers were compared with the corresponding figures of the male population of working capacity for the same period of time (31). The prevalent numbers of persons granted free medication by the end of 1981 were compared with the corresponding expected numbers for the general male populat ion of Finland (32).

Exposures
A sur vey of du st, vibration, and noi se expo sure and the health , disabi lity, and mortality of the workers in the Finnish granite industry (qua rrying, crushing, and block processing) was performed by the Institute of Occup ational Health in 1970-1972. Dust was measured in 28 quarries and processing yards and in four crushing plan ts. Dust co ncentra tions were high in several work phases. The highest concentration s were noted in drilling, the hygienic standard (0.2 rng/rrr') for quartz being exceeded, on the average, to-fold. The concentration of dust varied with the work phase. In drill ing, the total du st concentrations ranged from 12 to 116 mg/ rn", and the geometric mean was 39. 8 mg/rn", In the sur facing of blocks with a pneumatic hammer th e ran ge was 4-94 mg/rn", and the geometric mean 24.9 mg/rrr ' . In othe r wor k phases the range was 0.3-90 mg/rn! and the geometric mean rang ed from 1.7 to 28.1 mg/rn", The ranges of the quartz con centrations were 0.3-4.2 mg/rn! (geometric mean 1.47 mg /rrr') , 0.2-4.9 mg/rn! (geometric mean 0.82 mg/rrr') , and 0.02-3.6 rng/rn! (geometric mean 0.12-1.44 mg/rrr ') for drillin g, block surfacing and other work phases, respect ively (I) .
The location of the granite workplaces probably excluded coexpo sure to carginogens such as radiation , asbestos, or polycyclic aromatic hydrocarbons from organic combustion products (25). Earlier or later occupational exposure to such agents must have also been minor becaus e of the long-term employment in granite wo rk (I ). For instance, two-thirds of the gra nite workers in 1972 were still living in the ir nati ve parish , and onl y 7 010 of the current granite workers had been employed in other industrial work .
Smoking habi ts were clarified by questionnaire during the health screening in 1970-1972. The percentage of non smo kers was 21, exsmo kers 17, and smokers 62. The gra nite wo rkers' smoki ng habits were thu s similar to tho se of other Finnish gro ups of active male wor kers of the same age (3).

Statistical analysis
In the analysis of mortality, the age-specific ob served and expected numbers of dea ths and the standar dized mor tality rat io (SMR) values were co mputed by cause of death. The coho rt was also stratified in five-year periods, according to the length of follow-up , and the mortality rates were calculated for each period separately . The age-specific and cause-specific observed numbers of deaths were tested again st the corre spond- ing expected numbers, based on national figures (SMR) , by the Poisson-distribution model (21).
The age-specific and cause-specific prevalent numbers of disability pensions were calculated at the end of the two follow-up periods, as of 31 December 1971 and 31 December 1981. The incidence rates of disabi-. lity pensions were also calculated for the periods from I January 1969 to 31 December 1981. The expected numbers were computed for the prevalent figures by dividing the observed cases multinomially according to the proportions in the whole population. The Poisson distribution model was used to test the observed number of prevalent disability pensions against the expected number. The incidence densities were tested with the Mantel-Haenszel chi-square test for incidence density data (28).
The use of prescribed free medication was calculated as age-specific and cause-specific prevalences on 30 June 1971 and at the end of 1981. The prevalences were calculated for the most common diseases for which free medication was granted and for diseases which were determined to be excessive in the mortality and /or disability analyses or which could be associated etiologically with these diseases. The expected numbers were calculated in the same manner as those for prevalent figures of disability pensions. The Poisson distribution was used as an approximate method to test the observed number of persons with free medication against the expected number.

Mortality
At the end of 1981 the overall mortality among the granite workers was nearly the same as among the total male population of Finland (SMR 102) (table I) . Violent deaths were slightly less frequent among the granite workers than was expected . The violent deaths had taken place mainly during leisure time (ie, traffic accidents and suicides); only two of 28 such deaths were occupational accidents. Cardiovascular deaths were fewer among the granite workers than among the general male population until the age of 65 years . Coronary deaths were also fewer than expected.
The mortality from tumors was nearly the same as that of the male population [SMR 102, 95 % confidence interval (Cl) 75-137]. Of the 46 tumors, 22 were lung cancers (expected 17.1). In addition lung cancer was mentioned as the secondary cause of death in two cases of coronary heart disease and in one case of suicide. Mortality from cancer of the digestive organs was 1.6 times the expected value (observed 15, expected 9.7). Nine such cases were cancer of the stomach (expected 6.0), and cancer of the stomach was reported as a secondary cause in one case of coronary death. The other tumors reported as primary (N = 9) or secondary (N = 2) causes were not concentrated in any specific category of cancer diagnosis. A closer analysis of the cancer mortality can be found in another paper (19).
There were 28 deaths from respiratory diseases (SMR 201, 95 % CI 134-291). Of these 10 were silicotic and 18 nonsilicotic (emphysema, chronic bronchitis, bronchial asthma, and pneumonia) pulmonary diseases. Silicosis was thus observed to be the main cause of death for 4.3 % and a secondary cause of death for 2.6 % of all the deceased . Silicosis as the primary cause of death was connected with lung cancer only in one case (early lung cancer as a secondary cause), with other respiratory disease in three cases, and with cardiovascular disease in five cases. Rheu - Table 1. Mortality from certain causes-in the cohort in an age-specific comparison with the expectd (E) number of deaths calculated on the basis of death rates for the Finnish male population . (0 = observed, SMR = standardized mortality ratio, 95    ----- matoid arthritis was mentioned as a secondary cause of death for one person who died of silicosis. Silicosis was mentioned as a secondary cause in the three cases of cardiovascular death (in one of them lung tuberculosis was mentioned as an additional secondary cause), in one case of chronic pyelonephritis together with silicosis, and in one case of lung tuberculosis. In all, lung tuberculosis was observed in five cases as the main cause and in two cases as a secondary cause of death .
In addition to being reported as a secondary cause in the case of two silicotic deaths, lung tuberculosis was also mentioned in connection with different cardiovascular deaths . Rheumatoid arthritis was reported as an additional secondary cause of death in one case of lung tuberculosis . Diseases of the urinary tract were observed as the main cause of death in four cases (expected 1.8), three of which were chronic pyelonephritis (expected 0.6), and the other was chronic nephritis . As other than the primary cause of death, renal disease was reported in five cases . Three cases of urinary tract disease were associated with rheumatoid arthritis on the death certificates.
Rheumatoid arthritis was reported on nine death certificates (on two of them as the primary cau se). In addition, one case of polyarteritis nodosa-like rheumatoid arthritis was observed.
Comparison of the results for mortality until the end of the three follow-up periods revealed variations in the cause-specific results. The first follow-up through 1972 (table 2) indicated a slightly higher mortality from respiratory diseases and cancer of the digestive organs than expected on the basis of national rates. The second fo llow-up in 1975 confirmed an excess mortality from both types of disease . Mortality from cardiovascular diseases was clearly lower than expected . The third follow-up through 1981 revealed increasing mortality from lung cancer. Mortality from respiratory and cardiovascular diseases was similar during the first and second follow-up periods.
The analysis of mortality acco rding to years since entry into granite work (latency) revealed excess mortality from tumors in the group of worke rs followed for 20 years or more (observe d 26, expected 15.8, p < 0.05). The greatest excess was found between 25 to 29 years after the commencement of expos ure. The excess was mainly caused by lung cancer (observed 8, expected 2. 1) (table 3). The analysis of the other causes of death according to the length of follow-up did not show an y specific time dependency.

Disability
The disab ility of granite workers was studied twice, until the end of 1971 an d the end of 1981. At the end of the first follow-up in 1971 the total number of disability pensions for granite workers (39 observed) was statistically significantly higher than that for the Finnish male population (27.8 expected, p<O,05). The primary causes of the excess were musculoskeletal diseases (13 observed versus 3.8 expected, p < 0.001), mainly osteoarthritis and rheumatoid arthritis. Disability due to tuberculosis of the lungs was also more frequent thanexpected (5 observed versus 0.9expected, p <0.05).
In the secondjollow-up new disability pensions during 1969 to the end of 1981 were studied (figure 2). A comparison of the granite workers and the Finnish male population of working capacity did not reveal any essential differences with respect to the incidence rates for alI disabilit y pensions. However, a comparison of the cause-specific results showed significantly higher incidences of respiratory disease and rheumatoid arthritis among the granite workers . The incidence results were verified through the calculation of the prevalent numbers of disability at the end of 1981 (table 4). According to these calculations mental disorders were less frequent among granite workers than expected. Respiratory diseases were more frequent than expected (9 observed versus 4.4 expected) . The excess was mainly caused by silicosis and chronic bronchitis. Excessive numbers were also found for rheumatoid arthritis (10 observed versus 1.8 expected). In ci dence oensuyz 10 00 person-years

Granting of free medication
Chronic diseases were also measured by the occurrence of diseases for which free medication was granted by the national sickness insurance. The numbers of persons with free medication at the end of the first followup were few, a slight excess being observed for rheumatoid arthritis and bronchial asthma (I). The prevalence rates for free medication at the end of the second follow-up in 1981 are presented in table 5. At this time both rheumatoid arthritis (19 observed versus 8.1 expected cases) and bronchial asthma (19 observed versus 10.3 expected cases) were more common as causes of free medication among the granite workers than among the general male population. The granting of free medication for congestive heart failure and hypertension was less frequent than expected among the granite workers. This finding was compatible with the mortality and disability results. The other disease groups did not show any essential differences between the granite workers and the general male population of the same age.

Variations in the time-specific rates
Our mortality study of granite workers has given different results for the three periods of follow-up (1940-1972, 1940-1975, and 1940-1981) (table 2). The results for the mortality of even the same cohort depend on whether they are calculated as period-specific or cumulative or according to years since entry. The detection of potential excessive mortality from a certain disease, particularly from cancer , requires the recognition of the .optimum composition of the periods of entry , the periods of follow-up, and the age structure of the cohort (18). The internal structure of our granite cohort showed that excess overall mortality did not become evident until the granite workers had been followed for at least 20 years, but still in 1981 only 15 070 of the granite workers could be followed for this length of time.
A comparison of the mortality results of different cohorts is even more difficult. Inconsistencies in the earlier reported results can be explained by differences in (i) study design, (ii) the internal structure of the cohort [period of entry, period of follow-up, age structure, and occurrence of disease (eg, tumors) in the population], (iii) potential confounders and methods of analysis , and (iv) differences in exposure intensity and duration (18,19).

Violent deaths
The low mortality from violent deaths probably reflects physical and mental health selection before employment, and it was probably also influenced by workers remaining in that occupation. The minimum require-ment of three months of employment before entry into the granite work cohort studied by us may also explain these low rates.

Cardiovascular deaths
The low mortality (in comparison with the expected values) from cardiovascular diseases, and also from coronary heart disease separately, particularly during follow-up periods of less than 15 years, can be explained by the selection of healthy workers for granite work. The healthy worker effect was further demonstrated by the low morbidity rates (incidence and prevalence of disability and free medication for hypertension and cardiac insufficiency). In spite of the low mortality and morbidity rates in the short follow-up periods, the mortality rates indicated an excess of cardiovascular diseases among the older workers after more than 15 years of follow-up.

Cancer
Epidemiologic studies have suggested an excess risk for cancer among workers exposed simultaneously to quartz dust and confounding carcinogens (radon, asbestos, polycyclic aromatic hydrocarbons, some metals) (4,10,15,16,24,34). On the contrary, the few studies available on workers exposed to "pure" silica indicate only slight or no excess mortality from cancer (8,33). This diversity has been seen both in exposure cohorts and in cohorts of silicotic patients, as well as in case-referent studies (9,11,12,14,29,35,40).
The present study showed that the overall mortality from tumors for the whole period of follow-up (1940 to 198I) was similar to that expected . However, a stratified analysis revealed a significant excessin mortality after a minimum latency of 15years, with a maximum excess during the follow-up period of [25][26][27][28][29] years. The excess was entirely caused by cancer of the lungs and cancer of the gastrointestinal tract (mainly stomach cancer).
Our granite workers were exposed to pure silica unconfounded by other occupational and nonoccupational carcinogens . In addition the periods of exposure and follow-up were long. Among workers deceased due to lung cancer, there was only one case of manifest pulmonary silicosis. Furthermore an excess mortality from gastrointestinal cancer was observed. Taken together, these facts would suggest that silica exposure per se influences the development of lung cancer. This conclusion is not contradictory to earlier proposed connections between lung cancer and silicosis. The association between silica exposure and cancer has been discussed in detail in a separate paper (19).

Musculoskeletal diseases
Excessivemorbidity from rheumatoid arthritis was observed in the prevalence numbers for disability and free 23 medication in both 1971 and 1981. The excess in 1981 was not caused only by the cases of 1971, because the incidence rates of rheumatoid arthritis for the period between 1969 to 1981 were also significantly higher than the expected national values. On the contrary, although the prevalent number of cases of osteoarthritis was excessive in 1971, new cases during 1969-1981 were fewer than expected, with a consequent decrease in the prevalent number in 1981. The disappearance of osteoarthritis as a cause of morbidity among granite workers may depend on changes in the physical demands and work loads due to technical development of the work process.
An epidemiologic study of miners and ex-miners in a defined population showed no increase in the prevalence of rheumatoid arthritis (23). The author concluded that exposure to dust was of no etiologic importance for the disease. However, there are studies that have demonstrated an association between silica exposure and connective tissue diseases, eg, rheumatoid arthritis (5), systemic lupus erythematosus (6), and progressive systemic sclerosis (30). In fact the concept of silicoarthritis has appeared in the literature (22). In the present study no case of Caplan's syndrome (rheumatoid pneumoconiosis) was found.
The etiopathogenetic events behind the association between silica exposure and rheumatoid arthritis are not known. Hypothetically, exposure to dust may modify the host's defense against the unknown exogenous agent that causes rheumatoid arthritis or decrease resistance to other etiologic factors starting an immunologic arthritic process. Still, the initiation and perpetuation of the inflammatory joint disease may be influenced by the high prevalence of autoantibodies (rheumatic factor, antinuclear antibodies, etc) observed among silica exposed workers and by the silicamediated cytotoxicity of macrophages (2). The association between exposure to granite dust and rheumatoid arthritis has been discussed in another report that is as yet unpublished.

Respiratory diseases
The association between prolonged silica exposure and lung disease, eg, pneumoconiosis, silicotuberculosis, emphysema, and heart disease caused by pulmonary hypertension, is well recognized. Excess mortality and disability from respiratory diseases was obvious in all the follow-up studies of the present cohort. The excess rates became the most evident 10 years after the first exposure to granite dust.
Inhalation of granite dust has not been clearly established as a cause of chronic obstructive bronchitis (27). The excess number of workers granted free medication due to bronchial asthma can be explained in several ways. It is likely that males with even a slight predisposition to bronchial hyperreactivity would develop symptoms of bronchial obstruction even under nonspecific environmental exposure to dust. The ex-24 cess number granted free medication for bronchial asthma may also be explained by diagnostic difficulties and by the fact that free medication is not granted for chronic bronchitis.

Other diseases
Although the number of persons receiving free medication for diabetes did not differ from that of the reference population, diabetes was mentioned on exceptionally many death certificates as a secondary cause of death, and also on sickness insurance records. This association is a subject for further case-oriented studies. In addition, a number of other probable or possible disease associations were identified, ie, glomerulonephritis, amyloidosis and mesenchymal malignancies (chronic lymphocytic leukemia, lymphosarcoma, osteosarcoma, polycythemia vera). However, the small number of cases hinders any definite conclusions about a possible association with exposure to granite dust.