Occupational Medicine Advance Access originally published online on September 20, 2005
Occupational Medicine 2005 55(8):618-624; doi:10.1093/occmed/kqi145
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A cancer incidence and mortality study of Dow Chemical Canada Inc. manufacturing sites
The Dow Chemical Company—Epidemiology, 1803 Building, Midland, MI 48674, USA
Correspondence to: Carol Burns, The Dow Chemical Company—Epidemiology, 1803 Building, Midland, MI 48674, USA. Tel: +1 989 636 2278; fax: +1 989 636 1875; e-mail: cburns{at}dow.com
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
Background Previously, the mortality was reported in a cohort of male workers at an Ontario chemical plant.
Aims To expand the cohort and to investigate the mortality and cancer incidence risk among chemical manufacturing sites.
Methods We followed 5277 men and 1301 women from 1950 to 1999.
Results Employees experienced lower mortality and cancer incidence rates than the general population for several major causes of death, including heart disease, respiratory cancer and many other cancers. There were no cases of angiosarcoma of the liver. We observed a lower mortality rate of prostate cancer [standardized mortality ratio = 0.79, 95% confidence interval (CI) 0.43–1.32], but a higher incidence rate of prostate cancer [standardized incidence ratio (SIR) = 1.22, 95% CI 1.00–1.48]. Among the Sarnia employees, the incidence of pleural neoplasms was increased (5 observed versus 1.48 expected, SIR = 3.37, 95% CI 1.09–7.86). These cancers were included in the 12 deaths with malignant mesothelioma at Sarnia.
Conclusion Consistent with the earlier report, lower mortality rates were observed for the major classifications of disease and malignant neoplasms. The higher incidence rates of prostate cancer are not readily explainable but may reflect increased screening among current employees and recent retirees. Past asbestos exposure prior to 1980 is probably a contributor to the deaths due to malignant mesothelioma but is not reflected in lung cancer mortality. We find little indication of any other increased rates of mortality or cancer within the overall workforce of these chemical plants.
Keywords Asbestos; cohort; epidemiology; occupation exposure
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
As part of a comprehensive epidemiology programme, The Dow Chemical Company routinely evaluates and reports on the mortality data of its past employees at major manufacturing locations. These studies have evaluated the health status of Dow employees, provided background rates from which internal comparisons could be made [1,2] and have identified priorities for future research [3–5]. Previously, the mortality experience was reported for 3479 male Dow Chemical Canada Inc. (DCCI) employees who worked in Sarnia, Ontario [6]. Fewer deaths than expected were observed for many cancers. The observation of three deaths due to malignant mesothelioma and the category ‘other forms of heart disease’ resulted in statistically significant increased standardized mortality ratios (SMRs). In response to this earlier study, we conducted a cancer incidence and mortality study of DCCI's sites at Sarnia and Fort Saskatchewan.
The Human Studies Review Board of The Dow Chemical Company and the review board of Ethica Clinical Research, Inc. have approved this study. Statistics Canada approved the record linkage. The vital statistics Registrar General in each province/territory and the Cancer Registrar in each province/territory agreed to the study. No external researchers from outside Statistics Canada have had access to the analysis files. To further protect the confidentiality of individuals, observed deaths and cancers are only presented when there are 
3 observations.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
In the 1940s, a styrene plant was constructed in Sarnia, Ontario [7]. Following the end of World War II, the Sarnia operations expanded into a variety of other production units including chlorine, ethylene, sodium hydroxide (caustic), chlorinated solvents, vinyl chloride monomer, epoxy resins, vinyl ester resins, propylene oxide, polyols, latex, plastics (polyethylenes and polystyrene) and glycols. Vinyl chloride monomer, chlorine, caustic, chlorinated solvents and propylene oxide/propylene glycol production units were decommissioned by 1993.
The Fort Saskatchewan, Alberta, manufacturing site began production in 1961. In 1979–1980, world-scale chlor-alkali, vinyl chloride monomer, ethylene oxide/ethylene glycol and a power plant were commissioned. In 1994, a world-scale fractionator and ethylene plant went into production at Fort Saskatchewan.
Eligible workers were male and female employees who worked for at least 12 continuous months during the years 1950–1996 in either Sarnia or Fort Saskatchewan. The cohort of 6578 individuals (5277 males and 1301 females) was submitted to the Canadian Mortality Data Base (CMDB). The CMDB is maintained by Statistics Canada and is a computer-readable file of all deaths registered in Canada, as well as voluntarily reported deaths of Canadians occurring in the USA. The CMDB has captured a single underlying cause of death, which is coded in the International Classification of Diseases (ICD) version in effect at the time of death.
Angiosarcoma of the liver was a disease of a priori interest due to the manufacture of vinyl chloride at both sites [8]. We also wanted to identify all cases of malignant mesothelioma due to the findings of the previous study and community interest in Sarnia. We requested individual death certificate review for selected disease categories because neither disease has a unique cause of death code during the study period. The following categories were selected based on observations of mesothelioma in our US cohorts: non-specified digestive neoplasm (ICD8 152, 158, 159), lung cancer (ICD8 162, 163), non-specified respiratory neoplasms (ICD8 160, 163), non-specified malignant neoplasms (ICD8 171, 173.0–173.4, 173.6–173.9, 195–199), benign neoplasms (ICD8 211, 212, 215, 228, 239), liver cancer (155, 156) and angiosarcoma (171, 197). The death registrations were reviewed by both the resident nosologist at Statistics Canada and an occupational physician (K.F.H.). Any mention of ‘mesothelioma’ or ‘angiosarcoma’ on the record was recorded.
We used the Canadian Cancer Data Base (CCDB) to identify incident cancer events from 1969 to 1999. The provincial/territorial registries include persons diagnosed with a malignant cancer from 1969 onwards. The CCDB covers all incidences of cancer for individuals whose usual place of residence is Canada. Skin cancers and benign, in situ and uncertain lesions are not consistently reported by all provinces and have been omitted from the study.
The mortality analysis included the years from 1 January 1950 to 1 July 1999 whereas the incidence rates were from 1969 to 1999. The comparison populations for both were all provinces and territories of Canada. The analysis files were used internally by Statistics Canada staff to produce SMR and standardized incidence ratio (SIR) table outputs using the Monson routine [9]. Comparisons were also analysed for subcohorts by period of hire (1950–1970 and 1971–1999) and duration of employment (1–4, 5–14, 15 years). The duration cut-points are selected to be uniform across other occupational studies of Dow workers. Five deaths were identified using the tax files that were not identified by the CMDB. No cause of death information was available and these deaths were not included in the analyses.
Person-years for cohort members were accumulated across 5-year age and calendar-year-specific intervals. The first date of follow-up was the day after the employee had accumulated 12 months of full-time employment at any combination of the two study locations. Person-years at risk accumulated from the date of follow-up until the employee died or the end of the study period, whichever was earliest. The employees who were lost to follow-up were considered alive until the end of the study period.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
The majority of the cohort worked in Sarnia, Ontario (Table 1). The men were more equally distributed than the women by period of hire. The duration of employment was also different between the sexes.
|
We observed 558 deaths among the 5277 male workers (Table 2). The (SMRs) were significantly less than 1.0 for all causes (SMR = 0.61), all neoplasms (SMR = 0.70), diseases of the circulatory system (SMR = 0.69), respiratory system (SMR = 0.51), digestive system (SMR = 0.30) and external causes of death, including accidents (SMR = 0.39). Cancers of the respiratory system were also significantly reduced in these subjects (SMR = 0.53).
|
There were no causes of death for which the rates were statistically significantly higher than expected. Cancer of the oesophagus (SMR = 1.28) and malignant melanoma (SMR = 1.10) were the only cancers with more deaths than expected. Risk estimates for non-cancerous diseases of the musculoskeletal system (SMR = 2.57, n = 5), multiple sclerosis (SMR = 2.58, n = 4) and motor neurone disease (SMR = 1.78, n = 4) were also higher than expected. These were observed with low frequency and the confidence intervals (CIs) were very wide. When limited to the 3483 men with
20 years of follow-up we observed 422 deaths. Most of the SMRs were slightly lower than the 0 latency category and did not change the interpretation from the entire cohort.
There was no indication of higher mortality rates among short-term workers (1–4 years employment) or among long-term workers (15 years employment) with presumably greater opportunity for occupational exposures to carcinogens. Period of hire may also influence mortality because of years of follow-up and changes in opportunity for exposure to harmful substances. These changes may be due to improvements in plant environmental and industrial hygiene practices. Eighty-seven percent of the total deaths occurred among the men first employed prior to 1971. The mortality rates were not dissimilar from those of the entire cohort (Table 3). There were significant deficits for several causes of death including circulatory disease, accidents/poisonings/violence and cancer of the respiratory system. There were 70 deaths among the men who were hired during or after 1971 with a resulting SMR of 0.45 (95% CI 0.35–0.57) for all causes. The 22 observed cancer deaths were also significantly less than expected (SMR = 0.59, 95% CI 0.37–0.89) (data not shown).
|
The majority of the deaths occurred among the Sarnia subjects (497 deaths). Therefore, the mortality rates for Sarnia mirror those for the entire cohort as discussed above. We observed 72 deaths among the 1679 Fort Saskatchewan male workers. The Fort Saskatchewan group was also an extremely healthy workforce with a significant deficit in all-cause mortality (SMR = 0.66, 95% CI 0.51–0.83). About 32% of the deaths (n = 23) were due to cancer (SMR = 0.77, 95% CI 0.49–1.15). The most common cancers were those of the digestive system. Respiratory cancer, which accounted for only three deaths, was statistically less than expected (SMR = 0.30, 95% CI 0.06–0.86).
We observed 341 new cancers between 1969 and 1999 among the Sarnia and Fort Saskatchewan men (Table 4). This represents significantly fewer cancers than the 487 expected. Several other categories of cancers were significantly less than expected and included buccal cavity and pharynx, digestive, respiratory, urinary, ill-defined, non-Hodgkins lymphoma and lymphoid. We observed six cases of pleural cancer, which was significantly greater than expected (SIR = 3.59). Five of the cases of pleural cancer occurred among the Sarnia employees. Other cancers with higher rates included malignant melanoma (SIR = 1.07), prostate (SIR = 1.22) and reticulum cell sarcoma (SIR = 1.34). As with mortality, the site-specific results were similar to the overall results.
|
When evaluating cancers by duration of employment, all cancers were significantly less than expected across all three subcohorts. Five of the six cancers of the pleura occurred in men who worked at least 15 years (SIR = 5.87, 95% CI 1.89–13.70). Prostate cancer was less frequently observed in the short-term workers (SIR = 0.82, 95% CI 0.49–1.30) and rates were similar among the mid-term workers (SIR = 1.38, 95% CI 0.88–2.05) and long-term workers (SIR = 1.35, 95% CI 1.03–1.79). Conversely, leukaemia rates were higher than expected among the short-term workers (n = 6, SMR = 1.62, 95% CI 0.59–3.52) than among the long-term workers (n = 4, SIR = 0.56, 95% CI 0.15–1.42). Malignant melanoma was evenly distributed across the groups by duration of employment.
With respect to period of hire, all the cancers of the pleura and 13 of the 15 malignant melanomas occurred among men hired between 1950 and 1970 (Table 5). Prostate cancer was not clearly predicted by period of hire. Reticulum cell sarcoma was elevated among the recent hires, between 1971 and 1996.
|
We observed 39 deaths among the 1301 women employed at Sarnia and Fort Saskatchewan (data not shown). Circulatory disease (n = 15, SMR = 0.49, 95% CI 0.28–0.81) and breast cancer (n = 8, SMR = 0.97, 95% CI 0.45–2.07) were the most frequently observed causes of death. Thirty-five deaths were observed among the women who worked at Sarnia. Even with few observed deaths, significant deficits were observed for deaths due to all causes (SMR = 0.40, 95% CI 0.28–0.56), all malignant neoplasms (SMR = 0.58, 95% CI 0.34–0.92) and circulatory disease (SMR = 0.51, 95% CI 0.29–0.84) among the Sarnia women. Cancer of the breast was the most commonly occurring cancer death with seven observations (SMR = 1.02, 95% CI 0.41–2.10). Four deaths were observed among the 321 women at Fort Saskatchewan.
As with mortality, cancer rates were low among the female employees. We observed 42 cancers overall (SIR = 0.53, 95% CI 0.38–0.71). The most common sites were cancer of the breast (n = 22, SIR = 0.80, 95% CI 0.50–1.21) and respiratory system (n = 6, SIR = 0.75, 95% CI 0.27–1.62). There were no remarkable patterns by site, duration of employment or period of hire for mortality or incidence (data not shown).
On review of selected death certificates no cases identifying angiosarcoma of the liver were discovered. There were 13 cases of malignant mesothelioma noted either on the death certificate or the ICD-O morphology coding from the CCDB. All six incident cancers of the pleura were included in these deaths of malignant mesothelioma. One of the deaths from malignant mesothelioma (and incident pleura cancer) was that of an employee at Fort Saskatchewan.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
Consistent with the earlier report of Sarnia workers, we observed significantly lower mortality rates for the major classifications of disease and malignant neoplasms at both locations [6]. Other mortality studies of large manufacturing sites have reported SMRs that were closer to unity [10–12]. The current cohort is similar in size and employment period as another vinyl chloride and polyethylene facility located in Louisiana (all-cause SMR = 0.56) [13]. We observed no errors in data collection upon review of all incident cases and deaths due to lung, liver, pancreas and oesophagus cancers. We also recalculated the overall SMR changing the assumptions for workers who were lost to follow-up (i.e. assume they are lost at the time of loss; assume they are dead at the time of loss). The all-cause SMR was still very low at 0.76. The very low mortality rates may be a reflection of a younger cohort with less follow-up.
The rates for ‘other forms of heart disease’ were significantly elevated in the previous study (8 observed versus 2.7 expected) but were no longer elevated in the current study. Other notable causes of death in the previous study were malignant melanoma, multiple sclerosis and musculoskeletal disease. These causes remain uncommon, with three to four observed deaths each. The wide CIs make it difficult to make causal inferences.
We observed 13 deaths from malignant mesothelioma, of which 12 occurred among the Sarnia employees. Asbestos-related deaths are not specifically coded in older revisions of the ICD and therefore, comparison rates cannot be computed. Only three mesothelioma deaths were observed in the previous study of Sarnia employees. The opportunity for exposure to asbestos was discussed by Egedahl et al. [3]. Additional cases of mesothelioma in this study period were not unexpected due to the long latency period from possible asbestos exposure to manifestation of disease. The latency period is reported as being as high as 56 years [14], as low as 10 years [15] and more recently evaluated as 
25 years [16]. A latent period of 30 years is consistent with the observed appearance of malignant mesothelioma from 1986 to 1999 within the Sarnia workforce compared with the date of hire or start work dates between 1948 and 1962. Control measures to minimize potential asbestos exposure that were implemented in the later 1970s may result in significantly fewer malignant mesothelioma (and other asbestos-related disease) being reported in future studies. Health surveillance of the current workforce, reviewing retiree medical benefit information, linking with union health committee representatives and the retiree association, and monitoring workers' compensation claims shows that since the end of the study (1999) to date, only four additional deaths due to mesothelioma have occurred (2001 and 2004). As with all deaths in this study, we are limited by a lack of information on jobs and avocations held before and after the subjects' Dow employment. The lack of any excess in lung cancer suggests that occupational exposure to asbestos at this site was not pervasive.
We observed 14 deaths due to prostate cancer (17.78 expected) and 104 incident cases of prostate cancer (85.25 expected). This is the only cancer for which the mortality rates are not comparable to the incidence rates. Another occupational study by MacLennan and co-workers [17] found 11 prostate cancers compared to 6.3 expected (95% CI 0.9–3.1). The authors attributed this increased risk to early detection from prostatic-specific antigen (PSA) testing offered to active employees of the company. The increased prostate incidence among the DCCI employees may be due to heightened awareness, including PSA testing being used as a screening tool both in active employee and retiree populations. There are no confirmatory data at this time.
Occupational epidemiology was formulated at a time when exposure levels were high but declining. Large studies of employees with multiple exposures at low doses are considered obsolete by some [18]. The current study is limited by its generic nature in which specific chemical exposures are not addressed. Furthermore, we are limited by a lack of data on smoking, lifestyle and other potential confounders. However, this study is not conducted in a vacuum. Periodic physicals are offered to all DCCI employees as part of a health assessment programme. Personal counselling on lifestyle factors is offered. Both operations are state-of-the-art facilities which meet Canadian and global requirements to protect employees from exposure and injury. The current study has included both men and women employed at two large manufacturing facilities. Quantifying and tracking their health is an important public health process.
In conclusion, we evaluated the all-cause mortality experience and cancer incidence of 6578 male and female Canadian employees of a major chemical manufacturing company. Collaboration with Statistics Canada was successful. As a surveillance tool, the principal objective of this analysis was to identify potential hazards of employment rather than of specific chemical exposures. It is beyond the scope of this study to evaluate the heterogeneous exposures to the cohort. We observed a strong healthy worker effect with highly significant deficits for all causes, all malignant neoplasms and many other causes of death. We identified no cause for which a subsequent case–control follow-up study is warranted. The observed deaths due to malignant mesothelioma reflect historical exposures to asbestos. Due to the long latency of mesothelioma we may see future deaths due to mesothelioma despite asbestos exposure control measures which have been in place for several decades.
|
Conflicts of interest |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
Carol Burns, Ken Harrison and Brenda Jammer are employees of The Dow Chemical Company and hold stock in the company.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConflicts of interestReferences |
|---|
- Burns CJ, Beard KK, Cartmill JB. Mortality in chemical workers potentially exposed to 2,4-dichlorophnoxyactiec acid (2,4-D) 1945-1994: an update. Occup Environ Med 2001;58:24–30.
[Abstract/Free Full Text] - Bodner KM, Collins JJ, Bloemen LJ, Carson ML. Cancer risk for chemical workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Occup Environ Med 2003;60:672–675.
[Abstract/Free Full Text] - Bond GG, Shellenberger RJ, Fishbeck WA et al. Mortality among a large cohort of chemical manufacturing employees. J Natl Cancer Inst 1985;75:859–869.
- Bond GG, Fores GH, Stafford BA, Olsen GW. Lung cancer and hydrogen chloride exposure: result from a nested case-control study of chemical workers. J Occup Med 1991;33:958–961.[Web of Science][Medline]
- Bond GG, Cook RR, Shellenberger RJ, Daniel RL, Fishbeck WA. Case-control study of brain tumor deaths among employees at a chemical manufacturing plant. Ann NY Acad Sci 1982;381:73–82.[Web of Science][Medline]
- Egedahl RD, Olsen GW, Coppock E, Young ML, Arnold IMF. An historical prospective mortality study of the Sarnia Division of Dow Chemical Canada Inc., Sarnia, Ontario (1950–1984). Can J Public Health 1989;80:441–446.[Web of Science][Medline]
- Dudek V. In: The first 50 years: Dow Canada 1942–1992. Samia: Grafiks Marketing and Communications, 1991;11–12.
- Mundt KA, Dell LD, Austin RP, Luippold RS, Noess R, Bigelow C. Historical cohort study of 10,109 men in the North American vinyl chloride industry, 1942–72: update of cancer mortality to 31 December 1995. Occup Environ Med 2000;57:774–781.
[Abstract/Free Full Text] - Monson R. Analysis of relative survival and proportionate mortality. Comput Biomed Res 1974;7:325–332.[CrossRef][Web of Science][Medline]
- Olsen GW, Lacy SE, Cartmill JB et al. Half-century of cause-specific mortality experience of chemical manufacturing employees. Am J Ind Med 1996;26:203–219.
- Burns CJ, Cartmill JB, Chau M. Cause-specific mortality among Michigan employees of a chemical company: 1940–1994. J Occup Environ Med 2002;44:168–175.[Web of Science][Medline]
- Rinsky RA, Ott G, Ward E, Greenberg H, Halperin W, Leet T. Study of mortality among chemical workers in the Kanawha Valley of West Virginia. Am J Ind Med 1988;13:429–438.[Web of Science][Medline]
- Burns CJ, McLaury RL, Jammer BL, Cartmill JB. Mortality experience for Louisiana chemical workers (1956–1992). J La State Med Soc 1999;151:42–46.[Medline]
- Bianchi C, Giarelli L, Grandi G, Brollo A, Ramani L, Zuch C. Latency periods in asbestos-related mesothelioma of the pleura. Eur J Cancer Prev 1997;6:162–166.[Web of Science][Medline]
- Hauptmann M, Pohlabeln H, Lubin JH et al. The exposure-time-response relationship between occupational asbestos exposure and lung cancer in two German case-control studies. Am J Ind Med 2002;41:89–97.[CrossRef][Web of Science][Medline]
- Nurminen M, Karjalainen A, Takahashi K. Estimating the induction period of pleural mesothelioma from aggregate data on asbestos consumption. J Occup Environ Med 2003;45:1107–1115.[Web of Science][Medline]
- MacLennan PA, Delzell E, Sathiakumar N. Cancer incidence among triazine herbicide manufacturing workers. J Occup Environ Med 2002;44:1048–1058.[Web of Science][Medline]
- Guidotti TL. Occupational epidemiology. Occup Med (Lond) 2000;50:141–145.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||