Secondary Logo

Mortality Among Hardmetal Production Workers: The Swedish Cohort

Westberg, Håkan Dr Med Sci; Bryngelsson, Ing-Liss BSc; Marsh, Gary PhD; Buchanich, Jeanine MEd, PhD; Zimmerman, Sarah MS; Kennedy, Kathleen MS; Esmen, Nurtan PhD; Svartengren, Magnus MD

Journal of Occupational and Environmental Medicine: December 2017 - Volume 59 - Issue 12 - p e263–e274
doi: 10.1097/JOM.0000000000001054
ORIGINAL ARTICLES
Free

Background: The mortality pattern was determined in a cohort of 16,999 white and blue-collar workers in the Swedish hardmetal industry, particularly for cobalt exposure and lung cancer.

Methods: The mortality follow-up analysis in the Swedish Mortality register covered the period from 1952 to 2012. The exposure measures were ever/never exposed, duration of exposure, cumulative, and mean cobalt concentrations.

Results: The mortality of all causes was significantly increased, highly associated with the short-term employed workers. A negative exposure–response was found for lung cancer and duration of exposure. An exposure–response was determined for cumulative and mean cobalt exposures analyzed by quartiles, but not for exposure classes. Internal comparison analysis using proportional hazard showed no exposure–response.

Conclusions: The cohort lung cancer mortality showed no correlation to cobalt, nickel, or tungsten exposure.

Department of Occupational and Environmental Medicine, Faculty of Science, Örebro University, Örebro, Sweden (Dr Westberg, Ms Bryngelsson); Center for Occupational Biostatistics and Epidemiology, Department of Biostatistics, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania (Dr Marsh, Dr Buchanich, Ms Zimmerman); Division of Environmental and Occupational Health Sciences, School of Public Health, University of Illinois at Chicago, Chicago, Illinois (Ms Kennedy, Dr Esmen); and Department of Medical Science, Occupational and Environmental Medicine, Uppsala University, Uppsala, Sweden (Dr Svartengren).

Address correspondence to: Håkan Westberg, Professor, Dr Med Sci, Department of Occupational and Environmental Medicine, Örebro University Hospital, SE-701 85 Örebro, Sweden (hakan.westberg@regionorebrolan.se).

The study was approved by the Regional Ethical Review Board, Uppsala, Sweden, Dnr 2012/056.

This project was sponsored by the Swedish hardmetal industry, based on a research contract between The Department of Occupational and Environmental Medicine at Örebro University Hospital in Sweden and the hardmetal industry. The design, conduct, and conclusions of the study are exclusively those of the authors.

The authors have no conflicts of interest.

Occupational exposure to cobalt is well established in the hardmetal industry, which produces cutting tools that are mainly used for manufacturing of industrial products and parts. Hardmetal is a group of composite materials that consists predominantly of the hard tungsten carbide (WC) particulate phase tied together with cobalt as a binder.1 Nickel in the metallic state could also be added.

Exposure to cobalt during the production of hardmetal has been associated with several adverse health effects, such as rhinitis, sinusitis, bronchitis,2 asthma,3,4 and other respiratory effects,5 that is, dose-related decreased lung function over time,6 and hardmetal lung disease (HMLD).7 Allergic dermatitis has also been reported,7 as well as cases of cardiomyopathy,8,9 and an increased incidence of ischemic heart disease was determined in a cohort study of hardmetal workers.10

Mortality with special reference to lung cancer has been investigated in a number of epidemiological studies. A French cohort of hardmetal workers based on 709 male workers and mortality followed from 1956 to 1989 showed no overall increased mortality; however, for lung cancer, an standardized mortality ratio (SMR) of 2.1 was determined.11 Another French cohort study that investigated the relationship between hardmetal work and lung cancer based on data from one site with 3398 male and female workers between 1968 and 1998 showed a small rise in total mortality, and an elevated risk for male lung cancer associated with exposure (SMR 1.7).12 Furthermore, a nested case–control study that included data from 10 French hardmetal sites between 1968 and 1991 showed increased total mortality, and determined an odds ratio (OR) of 1.9 from the 68 cases of lung cancer.13 A Swedish cohort study based on 3163 males from three hardmetal plants followed up regarding mortality from 1951 to 1982 showed no increased overall mortality (SMR 0.96), but identified a significant excess lung cancer risk (SMR 2.1) for the overall exposed group (including both high and low exposed workers), workers with a duration of exposure more than 10 years, and with latency of more than 20 years.10

The International Agency for Research on Cancer (IARC) has classified cobalt metal with tungsten as probably carcinogenic to humans (group 2A).14 The harmonized classification for cobalt within EU legislation15 does not address the carcinogenicity; however, on the basis of animal testing of cobalt, the industry self-classification states that cobalt may cause cancer by inhalation.16 Regarding metallic nickel, IARC classifies metallic nickel as possibly carcinogenic to humans (group 2B).17 Results of other country-specific cohort analysis and exposure assessment are presented in the same volume of this journal as a series of companion papers.18–22

The present study investigates how exposure to cobalt, nickel and tungsten for three Swedish hardmetal plants operating from the 1930 s and onwards affects mortality, with a special reference to lung cancer.

Back to Top | Article Outline

METHODS

Study Objects and Processes

Three major Swedish hardmetal production sites were included, and companies A and B were rurally located and company C located in urban area. These companies are currently employing approximately 1340, 1440, and 350 white and blue-collar workers, respectively. They began operating in 1931, 1951, and 1942, respectively, and mainly produce inserts used as cutting tools or drills, and one of the units also produces big parts, such as rolls for hot rolling.

The production of hardmetal tools consists of several steps, the first being formation of tungsten carbide (WC) from tungsten oxide and elementary carbon through carburization to form WC powder. Next, the tungsten carbide is mixed with cobalt powders, followed by granulation. After granulation, the material is pressed, pre-sintered (heated), and then accurately machined into the desired shapes. The pieces are finally sintered at 1400°C to 1500°C to reach the hardness that is close to that of diamond. The products are then sand blasted and covered with a protective layer, and as a last step, the finished products are inspected for quality, stored, and shipped out of the plant.

The hardness and abrasive wear increases when either cobalt content or grain size decreases; a grain particle—diameter between 0.1 and 5 μm—and a cobalt content of 2% to 14% are common today.1 However, titanium-based cemented carbide tools produced by the Cermet method, which uses titanium and molybdenum carbides, also include metallic nickel, as do some other hardmetal products. The nickel concentration levels generally vary between 6% and 14%. Chromium carbide is also used in the production of some hardmetal products.

Back to Top | Article Outline

The Cohort (Study Population) and Cohort Follow-Up

All three Swedish hardmetal sites were included and data were extracted from the personnel files. Data included site, name, complete personal identification numbers, that is, year of birth, and duration of employment, with the year of start and end at defined departments and/or jobs. All blue-collar and white-collar male and female workers potentially exposed to cobalt were included, as were workers who had been employed at more than one site. In total, 16,999 workers were included in the initial cohort and matched in the Swedish Population Register. However, 1366 workers were excluded due to unclear or missing personal ID or duration of employment, leaving 15,633 workers for matching and analysis. The 11,084 male and 4549 female workers were distributed between companies A (3027), B (5481), and C (6999) (Table 1). The cohort included old members, the earliest year of birth ranging from 1889 to 1905, and the corresponding median ranging from 1937 to 1963. The average duration of employment was 9, 12, and 3 years at companies A, B, and C, respectively. The total number of person years was 480,633, unevenly distributed between the different sites. However, 6568 persons of the initial cohort were employed for less than 1 year (42%), at company A 31%, B 17%, and C 67%.

TABLE 1

TABLE 1

The mortality follow-up analyses for comparison with national death rates covered the period 1952 to 2012. A corresponding time period would be 1970 to 2012 for comparison with local rates, defined for the rural sites A and B as the appropriate county with the big cities excluded and for the urban company C the big city county. The vital status for all cohort members was determined by matching the cohort with the Swedish Mortality Register, based on the personal identification numbers. The mortality data were coded according to the International Classification of Disease (ICD 10).

Back to Top | Article Outline

Exposure Assessment and Measurement Data Base

The job or departmental class for each worker and time period were extracted from the personnel files and compared with company classification, which varied between different time periods in resolution. In the international study, some 69 job or departmental classes were identified. All workers were assigned a job class according to classifications from the international study. For the Swedish cohort, aggregated job classes were defined on the basis of similar exposure group (SEG) considerations and measurement data, leaving the following aggregated job classes A to I for exposure–response analysis.23 Job class A was defined as background (unexposed, ie, office workers), B as intermittent low (foremen, engineers, material handler, assembly, mark, pack, inspection, and packing), C as intermittent high (lab R&D, maintenance), D as powder production (weighing, mixing, spray dry, packaging), E as pressing, (pressing, forming, shaping), F as slow moving operations, G as coating, H as rolls (big pieces), and I as grinding.

Personal and area air measurement data were extracted from company records, covering a time period from the early 1970 to 2012. The database constituted included a total of 2693 personal measurements. The majority of the personal samples represented cobalt, tungsten, and nickel, that is, 1230 cobalt, 313 nickel, and 342 tungsten measurements.

Back to Top | Article Outline

Exposure Measures and Modeling

The standard exposure measures used were ever/never exposed, duration of employment, and cumulative and mean exposure. Latency time was defined as the period between first exposure and the observation. In addition, we used data from a log-linear model analysis of air concentrations to calculate cumulative and mean cobalt exposure measures, expressed as mg/m3 • years and mg/m3, respectively. In our model analysis, we used data from our measurement database to determine the exposure concentrations for different time periods, sites, and aggregated job titles. Our model initially included five different time periods (1970 to 1979, 1980 to 1989, 1990 to 1999, 2000 to 2009, 2010+), three different sites, and nine categories of aggregated job titles as independent variables. No data were available for the time period 1950 to 1969, during which a large number of the studied workers were exposed. We used three different extrapolation techniques for the time periods before 1970, that is, a sensitivity analysis to investigate the impact of different assumptions on our exposure measures used in our mortality study. In a standard exposure assessment procedure, the time period before 1970 would have the same modeled exposure as the period 1970 to 1979.However, in this study, we extrapolated regression coefficients (B) values based on average change per 10-year period, and used them to model cobalt exposure for the two time periods 1950 to 1959 and 1960 to 1969. In addition, we also used exponential extrapolation for the time period 1940 to 1969 by 10-year time periods to calculate historical measurement data. Data from the modeled cobalt exposure are shown for blue-collar workers with duration of employment more than 1 year. The exposure measurements were skewed, and necessitated a natural logarithm transformation. Measurement data for nickel and tungsten were sparse and unevenly distributed between jobs and time periods, and required a different approach. On the basis of dichotomized exposure to tungsten 0.1 or less and more than 0.1 mg/m3, all sites were included and analyzed by cumulative exposures as well as by means, classified in quartiles. Measurement data for nickel were only available at one plant, and the exposure measure based on the dichotomized exposure, either 0.01 or less or more than 0.01 mg/m3, to nickel.

The classification of nickel exposure was dichotomized on the basis of median; exposure levels below 0.01 mg/m3 were assigned 1 and exposures above 0.01 mg/m3 assigned 2 for each aggregated job class and time period. A corresponding classification (based on median) was set up for tungsten; exposure levels below 0.1 mg/m3 were assigned 1 and exposures above 0.1 mg/m3 were assigned 2. Cumulative exposures were calculated as level • years.

The results of the log linear modeling were used to determine cobalt concentrations for different time periods, jobs, and companies, and furthermore, cumulative cobalt exposures (mg/m3 • years) and means (mg/m3) were calculated for the exposure response analysis.

Exposure to cobalt was defined as cumulative exposure as mg/m3 • year (ie, exposure level times exposure time) and mean (cumulative exposure divided by duration of exposure) and categorized in two different ways. The categorization represented both standard grouping by quartiles and exposure classes for proper comparison with exposures and doses relevant for risk in comparison with the OELs, and also for comparison of our exposure–response relationship with earlier epidemiological studies. First, the determined cumulative and mean exposures were categorized as quartiles, for both the total cohort and for blue-collar workers. Second, the exposure measures was also categorized into three groups that reflected exposures relevant to a 40-year of working career at the present Swedish Occupational Exposure Limits (SOELs), that is, 0.02 mg/m3, defined as the 8-hour time-weighted average air concentration (8-hour TWA). The cumulative dose values for the three categories were 0.20 or less (low), 0.21 to 0.40 (medium), and at least 0.41 (high) mg/m3 • year. The high exposure group reflects half of the maximum allowed life-time exposure to cobalt, 0.8 mg/m3 • year, which corresponds to 40 years of exposure at the present Swedish OEL for cobalt, 0.02 mg/m3.19 The corresponding classes based on means would be 0.005 or less, 0.0051 to 0.010, and more than 0.010 mg/m3.

We will present exposure–response mortality, in particular that of lung cancer, based on the following exposure measures: ever/never exposed; duration of exposure including latency; cumulative cobalt exposures (mg/m3 • year), and mean cobalt exposure (mg/m3). The quantitative exposure measures will be classified by quartiles and exposure groups in our exposure–response analysis. For nickel and tungsten, a dichotomized exposure classification will be presented and exposure–response analyzed based on cumulative exposure. For nickel and tungsten, a dichotomized exposure classification will be presented and exposure–response analysis based on cumulative exposure performed.

Back to Top | Article Outline

Smoking Data

A questionnaire regarding smoking habits was distributed to cohort workers who were alive and to the next of kin to cohort members who had died. The questionnaire had questions whether the worker was a smoker, nonsmoker, or ex-smoker, as well as the duration and time period of their smoking. In addition, questions regarding jobs other than those in the hardmetal industry, but which were associated with exposures and jobs associated with lung cancer, like quarrying, stainless steel welding, chimney sweepers, coke oven plant workers, iron and steel foundry workers, ship yard workers, insulators, and asbestos cement workers, were included. The postal questionnaire was distributed to 8992 living cohort members and also to the next of kin for 1473 cohort members who had died after 1991. The possibility to trace next of kin was limited to those workers who had died after 1991, because as of 1991, the Swedish Tax Agency implemented a data system for the Swedish Population Register based on complete personal identification information available to us. Smoking data were obtained for 31% of the living workers and 17% for the deceased through next of kin.

Initially, after adding smoking data to our cohort, another matching would be performed and if possible a nested case–control study including smoking habits then performed. However, the response rates and representativeness, in particular for the deceased, were too poor both in number and representativeness to enable a proper nested case–control study. Smoking data were therefore used to discuss smoking habits in the different exposure estimates derived from our Cox proportional hazard regression analysis.

Back to Top | Article Outline

Statistical Methods

Cause-specific mortality data coded in ICD (10) were used to calculate national-based SMRs, which were obtained by comparing the cohort mortality with that of the general population of Sweden between 1952 and 2012 and expressed as ratios of observed and expected deaths. Another comparison was performed using regional rates, defined as either local county or city based on where plants were located. The expected numbers of deaths were calculated by multiplying person-years by gender-, 5-year age group-, calendar year-, and cause-specific mortality rates of the general Swedish population. These calculations were carried out using Stata Statistical Software (version 12.0; Stat Corp College Station, TX). The 95% confidence intervals (95% CIs) for the SMRs were computed assuming a Poisson distribution of the observed numbers of deaths. We have analyzed mortality based on all causes, using selected ICD codes, and paid particular attention for lung cancer. The exposure measures were ever/never exposed, duration of exposure including latency and cumulative and mean exposure for the total workforce, in particular blue-collar workers. The exposures were stratified on three exposure groups and exposure quartiles. Parametric and nonparametric Pearson and Spearman tests were used for correlation analysis.

A Cox proportional hazard regression analysis (Stata Statistical Software, version12) was performed on the basis of cumulative and mean exposure, adjusted for year of birth (categorized in 10-year age bands), duration of employment, and gender to analyze the exposure–response relation between lung cancer and cumulative and mean cobalt.

Back to Top | Article Outline

RESULTS

Exposure Assessment

Measurement Data

The personal and area air measurement data were extracted from company records, covering a time period from early 1970 to 2012. The majority of the samples represented cobalt, tungsten, and nickel, with 1230, 313, and 342 exposure measurements for cobalt, nickel, and tungsten, respectively. The cobalt concentration levels varied between 0.0001 and 2.8 mg/m3, with median and arithmetic mean values of 0.01 and 0.04 mg/m3, respectively. Nickel exposure levels ranged from 0.0001 to 2.8 mg/m3. Of the cobalt concentrations, 37% exceeded 0.02 mg/m3, and 21% exceeded 0.04 mg/m3. On the contrary, only 6% of nickel measurements exceeded 0.1 mg/m3 and only 1% exceeded 0.5 mg/m3. The personal cobalt exposure levels, organized by time period and aggregated job titles, are summarized in Table 2. The Swedish OELs for cobalt and nickel in the metal state are currently 0.02 and 0.5 mg/m3, respectively. Cobalt is considered a carcinogen in Sweden.24

TABLE 2

TABLE 2

Back to Top | Article Outline

Exposure Modeling and Exposure Measures

Log-linear modeling was performed for all aggregated jobs and the time period 1950 to 2012. Estimates for time periods when measurement data were sparse were based on two different assumptions: a) data from 1970 to 1979 were used to assess exposure for earlier time periods (1950 to 1969); b) exposures for the two time periods 1950 to 1959 and 1960 to 1969 were estimated by linear extrapolating for each job class. In our cohort analysis, we have chosen extrapolation according to b); however, as a sensitivity analysis, we have also performed an exponential extrapolation for the time period before 1970 and in short present the data as well as an analysis of SMRs for lung cancer based on these data. Detailed data regarding modeling and exposure measures will be presented elsewhere.25 Here we present regression data based on measurements from 1970 and onwards (Table 3), in addition extrapolation for earlier time periods. The modeling, with 1970 as the reference, mainly reflects declining exposures with time, as well as exposure relations between different job classes related to grinding. The ln-B values calculated for the job codes between 1950 to 1959 and 1960 to 1969 based on our regression model and linear extrapolation were 0.45 and 0.25, respectively, and used in our analysis. The determined cobalt air concentrations were then used to calculate cumulative and mean cobalt exposures in mg/m3 • years and mg/m3, for 3728 blue-collar workers with more than 1 year of exposure presented by job class (Table 4). The average cumulative exposure was 0.084 mg/m3 • years, corresponding to a daily exposure of 0.004 mg/m3 for 20 years of exposure. The jobs showing high cumulative doses were determined at powder, pressing, and rolls production. The same pattern was evident in the mean exposure measure. For all jobs with more than 1 year of exposure (8946 workers), the average cumulative exposure was 0.068 mg/m3 (data not in the table).

TABLE 3

TABLE 3

TABLE 4

TABLE 4

We also performed modeling after adapting data to an exponential extrapolation, resulting in B = 1.9 for 1940 to 1949, B = 1.8 for 1950 to 1959, and B = 1.6 for 1960 to 1969. Our linear extrapolation corresponds to an annual change of 2% to 3%, and our annual change for our exponential extrapolation ranges from 6% to 9% per year for the time period before 1970.When we modeled exponentially decreasing exposures for blue-collar workers more than 1 year of exposure, our cumulative and mean exposures increased. For powder, the average cumulative exposure increased from 0.27 to 0.4 mg/m3 • years, for pressing from 0.1 to 0.15 mg/m3 • years, and for the overall from 0.079 to 0.10 mg/m3 • years.

Back to Top | Article Outline

Cohort Mortality

All Causes

Table 5 presents the overall mortality for the whole cohort using ICD 10 diagnosis coding, which were then compared with national rates. A total of 5279 cases were observed. The SMR of all causes was calculated to 1.31 (95% CI 1.27 to 1.34). However, the overall mortality varied between companies, and company C was the only company to show a significantly increased SMR 1.48 (95% CI 1.44 to 1.53), as the SMRs for companies A and B the SMRs were 1.10 and 1.03, respectively. Using regional mortality data for the rurally located companies, A and B, and urban regional mortality data for company C did not noticeably change the SMRs. For company A, the SMR based on all causes the all cause of death change from 1.10 to 1.00, for company B, the SMR increased from 1.03 to 1.25 (neither increased SMRs were statistically significant), and for company C, the SMR decreased from 1.48 to 1.45; both SMRs were statistically significantly compared with the reference. The SMR for all causes for men was 1.36 (95% CI 1.32 to 1.40) and 1.17 (95% CI 1.10 to 1.23) for women.

TABLE 5

TABLE 5

A negative exposure–response pattern was identified when overall mortality was analyzed based on duration of exposure. The SMR decreased from 1.23 to 0.65 when duration of exposure in five exposure classes was applied (≤1, 1.1 to 5, 5 to 10, 10 to 20 to >20 years). The only statistically significant SMR was found for short-term workers with less than 1 year of exposure.

Back to Top | Article Outline

Circulatory System

For the whole cohort, a significant excess mortality was determined for all heart diseases (SMR 1.22), however, significant only for ischemic heart disease (SMR 1.22) and not for cerebrovascular disease. The significantly increased risk for ischemic heart disease was restricted to men. When total mortality was analyzed and duration of exposure and using the same five durations of exposure classes as for total mortality, a negative exposure–response was determined. Short-term workers, less than 1 year of exposure, were characterized by a statistically significant risk (SMR 1.42). However, workers with 1 to 5 years of exposure also showed a statistically significant risk, which is a more plausible relation to exposure. The SMRs decreased as duration of exposure decreased and were less than 1 for longer durations. The significant excess risk was restricted to company C.

Back to Top | Article Outline

Respiratory System

Overall, a statistically significant excess was determined for nonmalignant respiratory disease (SMR 1.46). The increased risk was more pronounced for men. Pneumonia, bronchitis, and emphysema were the major causes of death, with significant SMRs of 1.37, 1.46, and 2.8, respectively. An exposure–response analysis showed significantly increased mortality (SMR 1.2) for workers with duration of employment less than 1 year (SMR 2.1). For longer exposure time periods, the exposure–response was negative with no statistically significant SMRs. The high SMRs all occurred at company C. Ninety-two cases of bronchitis, 49 cases of emphysema, and 15 cases of asthma were identified.

Back to Top | Article Outline

Malignant Neoplasms

A statistically significant excess risk for overall malignant neoplasms (SMR 1.2) was determined for both men and women. However, no exposure response could be determined based on duration of exposure, and the excess mortality was restricted to company C, with similar exposure–response pattern as to that for lung cancer. Statistically significant overall excess mortality was determined for buccal cavity and pharynx (SMR 2.1), for esophagus (SMR 1.55), lung cancer (SMR1.68), and liver cirrhosis (SMR 2.6). For the buccal cavity, the risk was attributed to men, as for esophagus. For lung cancer and liver cirrhosis, the excess risk was located to both men and women. The increased SMRs for buccal cavity and the pharynx were determined only among the short-term workers, and this finding was influenced by company C, which had an SMR of 3.0 for this category. The same pattern was noted for esophagus. For liver cirrhosis, the excess SMR for workers with less than 1 year of exposure was 3.49, which decreased as the duration of exposure increased. In this case, the findings were attributed to both companies A and C.

Back to Top | Article Outline

External Causes

The excess mortality attributed to all external causes was high, as the SMR reached 1.74 for all external causes and was 2.4 for suicide, both of which were statistically significant. Mortality for all causes and suicide was more pronounced for men than for women. The suicides were also more pronounced for the short-term workers, but an increased SMR was also observed for workers with duration of exposure up to 10 years.

Back to Top | Article Outline

Lung Cancer

Exposure Duration and Latency

When the respiratory cancer mortality data were analyzed according to duration of exposure (years) and by a latency period either more or less than 20 years (Table 6), the highest SMR (2.4, 95% CI 1.7 to 3.2) and number of lung cancer (40) was determined for workers with 1 year of exposure or less and less than 20 years of latency. When more than 20 years of latency was applied, there were in total 105 cases and a significant SMR (1.8, 95% CI 1.5 to 2.2) for short-term workers. In all, short-term workers with less than 1 year of exposure experienced 145 cases of lung cancer, compared with the total group including 298 cases. No exposure–response relationship could be determined for the total group (298 lung cancer cases), regardless of latency period. The data showed an inverse exposure response, if any, when more than 20 years of latency was applied, but the groups with a lower duration of exposure had the highest SMRs. However, it should be noted that the SMR always exceeded 1, and with the significance declining and almost disappeared for workers with more than 10 years of duration of exposure. The SMRs were higher for workers with more than 5 years of exposure when the longer latency period, more than 20 years, was applied.

TABLE 6

TABLE 6

When excess mortality based on lung cancer was analyzed by site, no excess cancer risk was determined for plant A. However, plant B showed statistically significant excess risk was when latency was more than 20 years and duration of exposure was either 5 to 10 years or more than 20 years (SMRs of 2.1 and 1.5, respectively). At this company, almost all SMRs in this group showed increased risk, SMRs more than 1, although not statistically significant. For company C, the main contributor of lung cancer cases, an increased mortality (SMR >1), was determined for all duration of exposure, however, statistically significant only for duration of exposure less than 5 years.

Back to Top | Article Outline

Cumulative and Mean Exposure

All job classes with more than 1 year of exposure showed a significant increased risk for lung cancer, SMR 1.52 (95% CI 1.29 to 1.78) (Table 7). When the cumulative exposure was analyzed by quartiles, three of the highest quartiles showed statistically significant SMRs of 1.49, 1.66, and 1.55, that is, approaching exposure–response. However, when the exposure was stratified on the basis of exposure classes, that is, cumulative doses reflecting a 40-year working life at 0.01 mg/m3, the SMRs increased but were statistically nonsignificant (SMR 1.09, 1.35), respectively, for the highest dose groups. A similar, but weaker pattern was identified when lung cancer mortality was analyzed by mean exposures quartiles; however, a statistically significant increased SMR was determined for the highest exposure class.

TABLE 7

TABLE 7

Blue-collar workers with more than 1 year of exposure showed a significant risk for lung cancer, SMR 1.4 (95% CI 1.03 to 1.86) (Table 8). When the excess mortality was analyzed with cumulative exposures quartiles, the highest exposure group had an increased SMR of 2.1 (95% CI 1.28 to 3.24), which was statistically significant. Furthermore, when the relationship between cumulative exposure class and lung cancer risk was analyzed by exposure class, the exposure grouping showed a significant SMR (1.48, 95% CI 1.1–2.0) for the low exposed group, and the medium and high exposure groups showed SMRs of 0.96 and 1.33, respectively, neither of which was statistically significant (Table 8). A similar but weaker pattern was noted when SMRs were determined for the quartiles of mean exposures. Similar pattern was noted when exponential extrapolation for the time period before 1970 was used and cumulative exposures, classified as quartiles, showed a statistically significant increased SMR of 1.7 for the highest exposure group. When analyzed by exposure class, the exposure grouping showed a significant SMR of 1.48, 95% for the low exposed group.

TABLE 8

TABLE 8

Back to Top | Article Outline

Cox Regression

To adjust for potential differences in socioeconomic status when comparing mortality through the SMRs between the work force and the national comparison group, an internal comparison was performed as Cox proportional hazard regression. The data are presented as hazard ratios for different cumulative and mean exposure groups, as well as by age group, which is adjusted for gender and duration of employment. The hazard ratios ranged from 0.55 to 1.5 for all aggregated jobs more than 1 year of exposure; however, none of the hazard ratios were statistically significant (Table 9). The corresponding analysis for blue-collar workers with more than 1 year of exposure showed hazard ratios ranging from 0.83 to 2.3; these were not statistically significant, and no obvious exposure–response could be determined (Table 10). However, our high exposure estimates as analyzed by quartiles for cumulative exposures and means were rather low, for the cumulative exposure ≥ 0.04 mg/m3 years, for the mean exposure ≥ 0.005 mg/m3, in fact within the low exposure area as we defined our exposure grouping in the standard SMR analysis. In addition to the analysis by quartiles, we performed a Cox analysis with the exposure groups used in our SMR analysis; the high exposure for cumulative exposure was 0.4 mg/m3 years. No significant excess hazard ratio was determined for total or blue collar with more than 1 year of exposure by using our exposure grouping (data not in the table).

TABLE 9

TABLE 9

TABLE 10

TABLE 10

Back to Top | Article Outline

Nickel and Tungsten

On the basis of the dichotomized exposure to tungsten less than 0.1 and more than 0.1 mg/m3, all sites analyzed according to both cumulative exposures and mean exposures and classified in quartiles. No SMRs exceeding 1 were determined from the exposure mean analyses (data not in the table), for the high exposure group, but the cumulative exposure were significant in the high exposed group, SMR (1.94, CI 1.22 to 2.94). On the contrary, tungsten exposure was highly correlated to cobalt, according to parametric (Pearson and) and nonparametric (Spearman) methods. Exposure data regarding nickel only existed for company C in a particular process, the hot rolls production. We analyzed 18 cases of lung cancer based on dichotomized nickel exposure less than 0.01 and more than 0.01 mg/m3. No statistically significant excess lung cancer mortality was identified on the basis of SMR analysis, although almost all the SMRs were exceeding 1 (data not in the table).

Back to Top | Article Outline

Smoking Data

A response rate of 31% (total number of answered questionnaires, 2746; distributed 8992) was achieved from the questionnaires distributed to the living persons in the cohort, and the average percentage of smokers in this group was 45%; 39% ex-smokers were included in the smokers group (data not in the table). The corresponding figures for the deceased members were 17% (245; distributed 1473) and 59% smokers. When smoking habits were analyzed by duration of exposure in five groups, from less than 1 year to more than 20 years, the percentage of smokers in the groups ranged from 41% to 48%, with an average of 46%. Among the short-term workers with less than 1 year of exposure, 48% were smokers, that is, this group did not differ much from workers with more than 1 year of exposure.

When the analysis was restricted to blue-collar workers with more than 1 year of exposure, and evaluated year of birth by decades from the 1930s to 1990, the overall percentage of smokers ranged from 65% to 27%. When job class was included, workers who had worked several jobs had a range of 31% to 73%, powdering from 25% to 63%, pressing 11% to 72%, slow moving operations 23% to 70%, coating 22% to 75%, and grinding 20% to 59%.

When the analysis was focused on smoking incidence data and cumulative exposure, the overall cumulative mean cobalt exposure for ever smokers was 0.054 mg/m3 years and for never smokers was 0.048 mg/m3 years. The difference was small but stable regardless of duration of exposure. However, no statistically significant difference could be determined.

Back to Top | Article Outline

DISCUSSION

Our cohort of Swedish hardmetal showed a significant excess mortality for all causes, all heart diseases (ischemic heart disease), malignant neoplasms (lung cancer, liver cirrhosis), as well as some external causes. The increased mortality was strongly associated with the short -term workers and one of the plants. For lung cancer, a negative exposure–response was determined for duration of exposure and latency. An exposure–response was also determined when cumulative and mean cobalt exposure was analyzed by quartiles, but not for high, medium, or low exposure classes. Internal comparison analysis using proportional hazard showed no exposure–response. No exposure–response through an internal comparison analysis using Cox proportional hazard regression was determined.

Back to Top | Article Outline

Personal Data and Register Qualities

Our cohort encompasses a large number of workers and person years, and thus provides a sufficient number of deaths and lung cancer cases for exposure–response analysis, even though there was a high number of short-term employed workers, reflected by a the low average employment time.

The Swedish Population and National Cause of Death registers are well known for their quality; they employ use of personal identification numbers to enable national and regional tracking.26 We have evaluated mortality patterns in the cohort by time period to ensure the completeness of the company personnel registers. Lost to follow-up due to missing personal identification number or incomplete date of employment was only about 10%; all of the cohort members for mortality analysis were identified in the mortality registers.

Back to Top | Article Outline

Exposure Assessment and Exposure Measures

We have used standard exposure measures to evaluate our cohort (ever/never exposed, duration of exposure including latency, and cumulative and mean cobalt exposure) for exposure–response analysis.27,28 It is important to note that when cumulative dose is strongly related to the duration of exposure or other mechanisms related to high exposure at shorter time periods, it is also useful to include mean as an exposure measure.29 Our aggregated job classes reflect the job titles occurring in the company data files and registers accurately, as do data in our historical measurement database. The aggregation of job titles comes out of necessity, and the resolution of jobs in the company records and the measurement database show poor agreement. The within aggregated job class variability by time period are in the same order of magnitude as for job classes in other cohort studies; clarifying data will be presented in the exposure assessment paper.25 Our quantitative cobalt exposure measures are based on log linear modeling, including company, job class, and time period by decade, from 1970.30 As a large group of our cohort was included in the national mortality registers starting in the early 1950s, we have used our modeling of cobalt exposure to estimate the periods 1950 to 1959 an 1960 to 1969 rather than applying data from 1970 to 1979 for these earlier years. Our model shows decreasing exposure concentration levels at a rate of 2% to 3% per year, in line with trends reported in several other industrial cohorts.31,32 Our cohort consisted of many workers with short duration employments, even when restricted to blue-collar workers with more than 1 year of exposure, implying rather low cumulative and mean exposures. A standard categorization by quartiles would give even the high quartile low cumulative and mean exposures. To compare with risk expressed as a working life at the present OELs for cobalt and to compare the exposure measures with data in earlier mortality lung cancer studies, we also presented data by three exposure groups, the highest representing 40 years of exposure at half the present Swedish OEL for cobalt.

However, we have also provided a sensitivity analysis where we based our model data before 1970 on exponential extrapolation, as suggested by others.33 This provided higher cumulative and mean exposures, but no change in the SMR pattern was found when analyzed by exposure class or quartiles.

Back to Top | Article Outline

Lung Cancer and Exposure to Cobalt

We have presented mortality data using national rates for mortality even though two of the plants are rurally located, and one urban. Our reasons for this were justified by an initial comparison of mortality, all causes as well as lung cancer, where the differences between mortality for national and regional rates were small and insignificant.

Our initial analysis of lung cancer risk based on duration of exposure and latency showed an almost negative exposure–response, even though all the SMRs exceeded 1 (nonsignificant). Given the age of the cohort, we still have a large number of lung cancer cases, characterized by both high duration of exposure and latency to evaluate, that is, 35 lung cancer cases with more than 20 years of exposure and 20 years of latency for the whole cohort. This supports the robustness of our lung cancer mortality data when evaluated for the high exposure group.

To refine the analysis of cobalt and lung cancer, we restricted the duration of employment to more than 1 year. Furthermore, we analyzed blue-collar jobs in addition to all jobs in the exposure–response analyses, and thereby improved our model data, and exposure measures and information regarding the effects of cobalt exposure for different jobs.

We have analyzed dose–response by SMR using two different dose groups for the cumulative and mean exposures measures, one based on quartiles and the other based on exposure grouping (low, medium, and high) related to 40 years exposure at 50% of the current Swedish OEL, 0.02 mg/m3. The latter grouping was performed for reasons of comparison with the earlier studies and OELs, in particular the Swedish study.10

The analysis of all jobs with more than 1 year of exposure gives us significant risk for lung cancer (SMR 1.52), and an exposure–response pattern with significant SMRs for the third and fourth quartiles, of cumulative dose corresponding to more than 0.001 mg/m3 for 40 years. Only the third quartile, however, deviates significantly from the reference group. No exposure–response pattern could be identified in the high exposure group, although all the SMRs were more than 1, significant only for the exposure group characterized by less than 0.005 mg/m3 in 40 years. However, when mean exposure values were used in the analysis, a significant excess risk was determined in the high exposed group. When restricted to blue-collar workers with more than 1 year of exposure, a similar pattern, including an excess SMR of 1.4, was noticed for the blue-collar workers, but the exposure–response was weaker, and the significant findings for mean values have disappeared. We consider cumulative dose measures the most relevant for lung cancer. These findings might appear contradictory, but no adjustments for smoking or lifestyle factors have been performed.

Our data were not adjusted for smoking, nor have we adjusted for differences in socioeconomic or lifestyle factors, that is, we are still comparing our data with the national population. These issues, as well as the data from our questionnaires, are discussed in separate paragraphs. However, the lack of individual smoking data and other potential differences between our hardmetal workers and the general population necessitates an internal comparison, and is why we performed a Cox proportional hazard regression analysis (Tables 9 and 10). No statistically significant increased hazard ratios (hazard ratios >1) were determined for cumulative or mean exposures, when the analysis included all jobs. Our exposure grouping initially used quartiles, the highest cumulative exposure group representing 0.002 mg/m3 for 25 years. For blue-collar workers, the hazard ratio exceeded 1 for the highest quartiles for both exposure measures, cumulative and mean exposure. It could be argued that by using quartiles, we are only internally comparing rather low exposures. For this reason therefore, we have applied Cox analysis using our exposure classes already applied when we analyzed the lung cancer by SMR. These Cox analyses all showed hazard ratios below 1.

The evaluation of exposure–response analyses for the exposure measures ever/never employed, duration of exposure, cumulative and mean exposure to cobalt, and comparing our lung cancer mortality with the general population, an internal comparison failed to provide a consistent exposure–response relationship. Our high exposure group represents cumulative exposures more than 0.4 mg/ m3 years.

Back to Top | Article Outline

Smoking

Tobacco smoke, tobacco smoking, and involuntary smoking are all considered to be carcinogenic (group 1) to humans.34 Any occupational epidemiology study addressing lung cancer as one of the endpoints must investigate, or clarify smoking data and employ to indirect or direct adjustment methods. Initially, lung cancer cases and appropriate number of controls were supposed to be analyzed in a nested case–control study, in which both cases and controls answered an extensive questionnaire on smoking and other lifestyle factors as well as competing occupational exposures by telephonic interviews by our National Bureau of Statistics. However, certain incompleteness of the personal identification numbers during matching our cohort with the Swedish Mortality register restricted our study, and it was not possible to identify cases and controls in our population registers to carry planned telephonic interviews.

As an alternative, we used questionnaires, distributed to living cohort members with a response rate of 31%, and an average percentage of smokers of 45%. The corresponding figures for the deceased cohort members were 17% and 59%, respectively. Among the short-term workers with less than 1 year of exposure, 48% were smokers; this group did not differ much from workers with more than 1 year of exposure. We consider the response rates poor, and our intentions to use questionnaire data for individual smoking data in a case–control study or other adjustments were not possible. We could not establish any statistically significant difference in cumulative exposures when smokers and nonsmokers were compared. If any increased lung cancer risk could be determined in an exposure–response analysis, it could not be attributed to differences in smoking habits between exposure groups.

However, we used our data on smoking habits to compare the reference and the exposure groups regarding smoking habits, in particular, if excess hazard ratios were determined through the Cox regression analysis.

Back to Top | Article Outline

Nickel and Tungsten

Metallic nickel and chromium could be present in certain hardmetal grades during the production of hardmetal products. Our site questionnaire was not specific enough to resolve whether nickel was used, but additional data from the Swedish plants indicated that nickel was used at all three Swedish plants.

As measurement data were restricted to company C, we performed an exposure–response analysis of lung cancer and nickel exposure for that plant. Our analysis included only 18 cases of lung cancer and showed increased SMRs (>1), however not statistically significant. Our SMR analysis and the exposure estimates were based on a very crude dichotomized nickel exposure level for the cumulative and mean exposure estimates. The underlying measurement data are sparse and unevenly distributed between jobs, company, and time period, resulting in residual confounding. Data must be carefully interpreted and considered indicative. No further analysis regarding nickel exposure and lung cancer was performed and the potential Cox regression analysis was left out. Our historical measurement database showed low levels of nickel; only 6% exceeded 0.1 mg/m3 and just 1% exceeded 0.5 mg/m3.

Our historical measurement database showed that only 6% exceeded 0.1 mg/m3 and just 1% exceeded 0.5 mg/m3. The Swedish OEL19 for nickel in the metal state is currently 0.5 mg/m3, and the metal is not classified as carcinogenic; the corresponding ACGIH TWA is 1.5 mg/m3.35 Within the EU, insoluble and soluble compounds are considered carcinogenic and addressed with low recommended OELs; however, metallic nickel is excluded, as neither animal experiment nor human epidemiological studies have indicated any carcinogenicity.36 IARC classifies metallic nickel as possibly carcinogenic to humans (group 1), based on “sufficient evidence in humans for carcinogenicity of mixtures that include nickel and nickel compounds”17 Our evaluation of nickel exposure and lung cancer is restricted to the only plant that has measurement data. Only a few cases of lung cancer were available and the lack of measurement data necessary to perform exposure modeling makes any conclusions limited. However, our finding is somewhat supported by the exposure measurement data for nickel, well below current OELs. Tungsten exposure measurement data were highly correlated to cobalt, and given the sparse number and distribution over time and jobs, any exposure–response findings would be difficult to interpret.

Back to Top | Article Outline

Short-Term Workers

A large proportion of our cohort represents workers with a duration of employment less than 1 year due to the inclusion criteria set in the international study group and having the possibility to use the short-term exposed workers as a reference group. Short-term workers comprised 42% (6568) of our cohort, with skewed representation between company A (31%), company B (17%), and company C (67%). For our analysis, in particular lung cancer, this implies that 153 lung cancer cases with more than 1 year of exposure out of originally 298 cases represented workers with less than 1 year of exposure. The potential exposure factors when analyzing and explaining increased morality in short-term workers have been have been extensively discussed in two previous industrial cohorts.37 In principal, the high mortality for short-term workers could be explained by exposure to the highest concentration levels, a subgroup with higher susceptibility, quitting jobs sooner, and experiencing more hazardous exposure in other occupations. Lifestyle factors, such as not only smoking, which is especially relevant for lung cancer, but also drinking habits and other socioeconomic factors could be involved. In our study, the short-term (<1 year) workers showed excess risk for mortality for all causes when compared with the total cohort, in addition to the increased risk as for malignant neoplasms, buccal cavity, esophagus, larynx, lung, and bladder, cardiovascular disease, heart disease, nonmalignant respiratory disease, cirrhosis, and all external causes of death, including suicide. The very high mortality for the short-term workers, especially liver cirrhosis, indicated an enhanced pattern of alcohol consumption as an indicator of socioeconomic factors that affects health. However, no such difference was noted when we investigated the differences between smoking habits based on data from our self-administered questionnaire. In fact, 49% of the workers with less than 1year of exposure were smokers, only slightly higher than the rate of the total group, 46%. All other groups based on duration of employment showed only slightly lower smoking rates (41% to 47%). These data are in line with findings from other large industrial cohorts.38

Our findings imply other explanations than differences in smoking habits, which were only slightly higher for the short-term workers, to explain the differences in lung cancer outcome.

Back to Top | Article Outline

Other Studies

Our findings show a similar mortality pattern to the earlier published data from an original cohort of Swedish hardmetal workers (3163 male workers), evaluated for the period 1940 to 1982.10 The overall mortality was found to be less than 1. However, when lung cancer mortality was evaluated for members with more than 10 years of exposure and 20 years of latency, a nonsignificant exposure–response based on low and high exposure group was found, SMR (2.3, 3.3, respectively). When more than 10 years of exposure and more than 20 years of latency was applied for our data (Table 6), no consistent exposure–response could be determined, and applying cumulative cobalt exposure estimates for blue-collar workers and internal comparison showed similar results (Tables 8 and 10).

A French cohort of 709 hardmetal workers employed between 1956 and 1989 was analyzed for mortality and exposure–response patterns.11 The high exposed group showed a significant SMR of 5.03, but including smoking habits, the risks among the current smokers, increased the SMRs to 9.2 and 15 for the medium and high exposed group, respectively. Analyses of duration of employment and latency showed no exposure–response trends. Our findings, which did not identify any exposure–response for internal comparison, may reflect the decreased exposure levels in our study, which included the period from 1990 onwards. It should be noted that our high exposure group, more than 0.4 mg/m3 • year, reflects a 40-year exposure to 0.010 mg/m3 of cobalt, to be comparable with exposure group 2 of the French study.

In another French multicenter study, a cohort was formed from 10 different plants, with an inclusion criteria more than 3 months exposure and a mortality follow-up from 1968 to 1991.13 The cohort included 7459 men and women. An increasing trend for the SMR was determined by an increased duration of exposure. The overall analysis in the case–control study showed an excess risk (OR 1.93, statistically significant). However, when adjusting for smoking, smokers (OR 3.62) showed a significantly increased risk, and workers who had never smoked (OR 1.21) showed a nonsignificant risk. The high-dose group in the analysis consisted of more than 164 or more than 299 level ·months; according to the given job exposure levels in the study, it would, for example, correspond to 40 months (3.3 years) at level 4, 0.06 mg/m3, that is, 0.20 mg/ m3 • years. These exposure levels are in the same order of magnitude as the high exposure group in our study. The finding in the French case–control study implicates dose levels about 0.2 mg/ mg/m3 • years, as they potentially generate an increased risk of lung cancer.

Another cohort study from a French plant producing hardmetal and other cobalt products included 2860 workers.12 The company started in 1940, and the follow-up period was 1968 to 1992. The mortality pattern showed excess SMRs for lung cancer in men. Smoking data were retrieved from colleagues. An analysis based on exposure scores and weighted and unweighted cumulative exposures showed significant SMRs (2.36, 2.06) for both the high exposure groups, which had been adjusted for smoking. However, no quantitative data are provided for the different exposure groups. The follow-up period of this study deviates from ours, reflecting different exposure periods.

Back to Top | Article Outline

Other Mortality Causes

Heart Disease

Our data showed overall significant excess risks both for ischemic and cerebrovascular disease. A negative exposure response was also identified when analyzed by duration of exposure. In the original Swedish cohort,10 an excess risk for ischemic heart disease was determined for the high exposure group, but no excess risk was determined in the other studies (Lasfargues, SMR 0.82; Moulin, SMR 0.88 to 0.90; Wild, SMR 0.91). However, the relationship between particle exposure and cardiovascular disease is an emerging area of interest, and the mechanisms for inflammatory response to exposure have been established.39

Back to Top | Article Outline
Nonmalignant Respiratory Disease

We determined excess mortality for certain nonmalignant respiratory disease, such as bronchitis, emphysema, and pneumonia, strongly associated with the urban plant. However, the total mortality showed negative exposure response when analyzed by duration of exposure. As expected, the number of cases of nonmalignant respiratory disease was substantially higher than that of the French studies and earlier Swedish study.

Back to Top | Article Outline

CONCLUSIONS

In our cohort of Swedish hardmetal workers, we have determined a statistically significant excess mortality for all causes, all heart disease (ischemic heart disease), malignant neoplasms (lung cancer, liver cirrhosis) as well as some external causes. The increased mortality was strongly associated with the short-term employed workers and one of the plants. For lung cancer, a negative exposure–response was found for duration of exposure. For cumulative and mean cobalt exposure by quartiles, an exposure–response was determined, but not for high, medium, and low exposure classes. Smoking was not adjusted for. By internal comparison analysis using Cox proportional hazard regression for cumulative and mean exposure, no exposure response was shown. No exposure–response was determined for cobalt, nickel, and tungsten.

Back to Top | Article Outline

Acknowledgments

We would like to acknowledge the cooperation and assistance of the representatives from the industry and their Swedish sites. In particular, we would like to thank all the employees who provided assistance throughout the study.

Back to Top | Article Outline

REFERENCES

1. Sadik M. An Introduction to Cutting Tools Materials and Application, Sandvik. Sandvik Publication. ISBN 978-91-637-4920-9. 2nd ed. 2015.
2. Balmes JR. Respiratory effects of hard-metal dust exposure. Occup Med 1987; 2:327–344.
3. Shirakawa T, Kusaka Y, Fujimura N, et al. Occupational asthma from cobalt sensitivity in workers exposed to hardmetal dust. Chest 1989; 95:29–37.
4. Kusaka Y, Nakano Y, Shirakawa T, Morimoto K. Lymphocyte transformation with cobalt in hardmetal asthma. Ind Health 1989; 27:155–163. 1989.
5. Nemery B, Verbeken EK, Demedts M. Rapidly fatal progression of cobalt lung in a diamond polisher. Am Rev Resp Dis 1990; 141:1373–1378.
6. Rehfisch P, Anderson M, Berg P, et al. Lung function and respiratory symptoms in hardmetal workers exposed to cobalt. J Occup Environ Med 2012; 54:409–413.
7. Nakamura Y, Nishizaka Y, Ariyasu R, et al. Hardmetal lung disease diagnosed on a transbronchial lung biopsy following recurrent contact dermatitis. Intern Med 2014; 53:139–143.
8. Barborik M, Dusek J. Cardiomyopathy accompanying industrial cobalt exposure. Br Heart J 1972; 34:113–116.
9. Kennedy A, Dornan JD, King R. Fatal myocardial disease associated with industrial exposure to cobalt. Lancet 1981; 1:412–414.
10. Hogstedt C, Alexandersson R. Morality for Hardmetal Workers [in Swedish]. ISBN 91 7045 0722. Solna, Sweden: National Institute of Occupational Health; 1990.
11. Lasfargues G, Wild P, Moulin JJ, et al. Lung cancer mortality in a French cohort of hard-metal workers. Am J Ind Med 1994; 26:585–595.
12. Wild P, Perdrix A, Romazini S, Moulin JJ, Pellet F. Lung cancer mortality in a site producing hardmetals. Occup Environ Med 2000; 57:568–573.
13. Moulin JJ, Wild P, Romazini S, et al. Lung cancer risk in hard-metal workers. Am J Epidemiol 1998; 148:241–248.
14. International Agency for Research on Cancer (IARC). IARC monographs on the evaluation of carcinogenic risks to humans. Volume 68. Cobalt in Hardmetals and Cobalt Sulfate, Gallium Arsenide, Indium Phosphide and Vanadium Pentoxide. Lyon, France: IARC Press; 2006.
15. ECHA 2016 Substance Information. Cobalt hazard classification and labelling. Available at: http://echa.europa.eu/substance-information/-/substanceinfo/100.028.325. Accessed March 24, 2017.
16. CoRC 2013 Cobalt REACH consortium. Industry action for responsible assessment of cobalt compounds. Available at: http://www.cobaltreachconsortium.org/web_images//documents/One-page-statement-Industry-Actions.pdf. Accessed December 19, 2013.
17. International Agency for Research on Cancer (IARC). IARC monographs on the evaluation of carcinogenic risks to humans. In: Volume 100 C, 2012. Arsenic, metals, fibres and dust. Lyon, France: IARC Press.
18. Kennedy KJ, Esmen NA, Buchanich JM, Zimmerman S, Sleeuwenhoek AJ, Marsh GM. Mortality among hardmetal production workers: occupational exposures. J Occup Env Med 2017; 59:e297–e305.
19. McElvenny DM, MacCalman LA, Sleeuwenhoek A, et al. Mortality among hardmetal production workers: UK cohort and nested case–control studies. J Occup Env Med 2017; 59:e275–e281.
20. Morfeld P, Groß JV, Erren TC, et al. Mortality among hardmetal production workers: German historical cohort study. J Occup Env Med 2017; 59:e288–e296.
21. Wallner P, Kundi M, Moshammer H, Zimmerman SD, Buchanich JM, Marsh GM. Mortality among hardmetal production workers: a retrospective cohort study in the Austrian hardmetal industry. J Occup Env Med 2017; 59:e282–e287.
22. Marsh GM, Buchanich JM, Zimmerman S, et al. Mortality among hardmetal production workers: US cohort and nested case–control studies. J Occup Env Med 2017; 59:e306–e326.
24. SWEA. Occupational Exposure Limit Values (AFS 2015:7) [In Swedish]. ISBN 978-91-7930-628-1. Solna, Sweden: Swedish Work Environment Authority; 2015.
25. Westberg H, Bryngelsson I-L, Marsh G, et al. Mortality among hardmetal production workers: the Swedish cohort. J Occup Env Med 2017; 59:e263–e274.
26. Ludvigsson JF, Otterblad-Olausson P, Petterson BU, Ekborn A. The Swedish personal identity number: possibilities and pitfalls in health care and medical research. Eur J Epidemiol 2009; 24:659–667.
27. Checkoway H. Methods of treatment of exposure data in occupational epidemiology. Med Lav 1986; 77:48–73.
28. Checkoway H, Dement JM, Fowler DP, Harris RL Jr, Lamm SH, Smith TJ. Industrial hygiene involvement in occupational epidemiology. Am Ind Hyg Assoc J 1987; 48:515–523.
29. Kriebel D, Checoway H, Pearce N. Exposure and dose modelling in occupational epidemiology. Occup Environ Med 2007; 64:492–498.
30. Westberg HB, Bellander T. Epidemiological adaptation of quartz exposure modeling in Swedish aluminum foundries: nested case-control study on lung cancer. Appl Occup Environ Hyg 2003; 18:1006–1013.
31. Symanski E, Kupper L, Hertz-Picotte I, Rappaport S. Comprehensive evaluation of long term trends in occupational exposure: part 2. Predictive models for declining exposures. Occup Environ Med 1998; 55:310–316.
32. Andersson L, Burdorf A, Bryngelsson IlB, Westberg H. Estimating trends in quartz exposure in Swedish iron foundries-predicting past and present exposures. Ann Occup Hyg 2012; 56:362–372.
33. Creely K, Cowie H, van Tongeren M, Kromhout H, Tickner J, Cherrie J. Trends in inhalation exposure: a review of the data in the published scientific literature. Ann Occup Hyg 2007; 51:665–678.
34. International Agency for Research on Cancer (IARC). IARC monographs on the evaluation of carcinogenic risks to humans. Volume 83. Tobacco Smoke and Involuntary Smoking. Lyon, France: IARC Press; 2004.
35. ACGIH. Threshold Limit Values for Chemical Substances and Physical Agents & Biological Exposure Indices. Cincinnati, OH: American Conference of Governmental Industrial Hygienists; 2015.
36. SCOEL. Recommendation from the Scientific Committee on Occupational Exposure Limits for Nickel and Inorganic Nickel Compounds. Brussels, Belgium: SCOEL/SUM European Commission; 2011.
37. Bofetta P, Kolstad H, Olsen J, et al. Mortality of short- term workers in two international cohorts. J Occup Environ Med 1998; 40:1120–1126.
38. Hansen J, Bofetta P, Anderssen A, et al. Comparison of information on occupational and life style habits obtained from European man-made vitreous fibre production workers and their relatives. Int J Epidemiol 1997; 26:1009–1016.
39. Brook RD, Rajagopalan S, Pope CA 3rd, et al. American Heart Association Council on Epidemiology and Prevention, Council on the Kidney in Cardiovascular Disease, and Council on Nutrition, Physical Activity and Metabolism. Particulate matter air pollution and cardiovascular disease: an update to the scientific statement from the American Heart Association. Circulation 2010; 121:2331–2378.
Copyright © 2017 by the American College of Occupational and Environmental Medicine