Skip Navigation


Occupational Medicine Advance Access originally published online on October 18, 2005
Occupational Medicine 2005 55(8):625-630; doi:10.1093/occmed/kqi156
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
55/8/625    most recent
kqi156v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Nichols, L.
Right arrow Articles by Sorahan, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nichols, L.
Right arrow Articles by Sorahan, T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2005. Published by Oxford University Press on behalf of the Society of Occupational Medicine. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Cancer incidence and cancer mortality in a cohort of UK semiconductor workers, 1970–2002

Linda Nichols and Tom Sorahan

Institute of Occupational and Environmental Medicine, University of Birmingham, Edgbaston, Birmingham, West Midlands B15 2TT, UK

Correspondence to: Tom Sorahan, Institute of Occupational and Environmental Medicine, University of Birmingham, Edgbaston, Birmingham, West Midlands B15 2TT, UK. Tel: +44 121 414 3644; fax: +44 121 414 6217; e-mail: t.m.sorahan{at}bham.ac.uk


   Abstract
Top
 Abstract
Introduction
 Methods
 Results
 Discussion
 Conflicts of interest
 References
 
Objective To examine cancer risks in a cohort of workers employed in the manufacture of semiconductors.

Methods The mortality (1970–2002) and cancer morbidity (1971–2001) experienced by a cohort of 1807 male and female workforce employees from a semiconductor factory in the West Midlands (UK) have been investigated. Standardized mortality ratios (SMRs) and standardized registration ratios (SRRs) were used to assess mortality and morbidity, respectively.

Results Overall mortality was close to expectation in males [SMR 99, 95% (confidence interval) CI 79–122] and significantly below expectation in females (SMR 74, 95% CI 65–85). Incidence of all sites of cancer was somewhat elevated in males (SRR 130, 95% CI 95–173) but close to expectation in females (SRR 94, 95% CI 82–109). There were significant deficits of deaths from cancer of the oesophagus in males and females combined and from cancer of the breast in females. Significantly elevated SRRs were found in males for cancer of the rectum [Observed (Obs) 6, SRR 284, 95% CI 104–619], in females for cancer of the pancreas (Obs 10, SRR 226, 95% CI 108–415) and malignant melanoma (Obs 11, SRR 221, 95% CI 110–396) and in males and females combined for cancer of the rectum (Obs 19, SRR 199, 95% CI 120–310) and malignant melanoma (Obs 12, SRR 217, 95% CI 112–379). Detailed work history data were unavailable for analysis. The finding of excess morbidity was not mirrored in the corresponding mortality findings.

Conclusions The study found elevated morbidity for a number of cancer sites that may be unconnected with occupation. Elimination of all possible occupational causes will, however, require more detailed analyses of cancer risks in relation to exposure histories.

Keywords      Cohort study; mortality; semiconductor industry


   Introduction
Top
 Abstract
 Introduction
Methods
 Results
 Discussion
 Conflicts of interest
 References
 
An earlier investigation [1] studied mortality and cancer incidence in 1807 workers employed at a factory manufacturing semiconductors in the West Midlands, UK. This investigation was set up following concerns about a possible cluster of cancers in the workforce. Mortality was investigated for the period 1970–1982, and cancer incidence for the period 1971–1981. The mortality findings were unexceptional, although an excess of borderline statistical significance was reported for the incidence of malignant melanoma [Observed (Obs) 3, Expected (Exp) 0.68, standardized registration ratios (SRR) 441, P = 0.06]. This excess was not associated with duration of employment, either in the plant as a whole or in the photomasking process, a procedure known to involve ultraviolet light exposure. The first update of this study, with a further 7 years of follow-up, found no additional cases of malignant melanoma [2].

In 2003, the Health and Safety Executive (HSE) reported cancer incidence and mortality in a cohort of 4388 semiconductor manufacturing workers based at a factory in Scotland, following concern that there might be a cluster of cancers at the facility [3]. Overall mortality and cancer incidence was below or close to expectation. Based on rates for the general population of Scotland, a statistically significant excess (incidence and mortality) of lung cancer was found in female workers, along with a significant excess of stomach cancer (based on three cases) and a non-significant excess of breast cancer (based on 20 cases). In male workers there was a non-significant excess of deaths from cancer of the brain, based on three deaths.

In the light of the findings of the HSE report, and editorials expressing concern about cancer risks in the semiconductor manufacturing industry [4,5], it was clear that further information was required on this topic. It was decided, therefore, to incorporate a further 13 years of mortality and cancer incidence data into the West Midlands study [2], with the aim of identifying causes of death or sites of cancer that merited further investigation. A longer version of this report is available as a HSE technical report [6].


   Methods
Top
 Abstract
 Introduction
 Methods
Results
 Discussion
 Conflicts of interest
 References
 
In 1983, identification particulars were abstracted from company personnel records for 1807 (281 males and 1526 females) workforce employees from a semiconductor manufacturing facility in the West Midlands, UK. All males were full-time employees whereas 73% of females were part-time employees. The cohort comprised a ‘census’ cohort in employment on 1 January 1970 (758 employees) together with an ‘entry’ cohort that commenced employment between 1 January 1970 and 31 December 1979 (1049 employees). (A previous report noted 759 employees in the census cohort and 1048 in the entry cohort [1]. These figures have been revised after checking the employment dates.) All workers had a minimum period of employment of 1 month. Some 88% of the cohort was employed for 12 months or more.

The site was opened by the parent company in 1941 as a workshop for wartime production, and semiconductor production did not start until 1966. In the 1980s, departments included the plating shop, assembly, wafer plating, diffusion (incorporating photomasking), chemical processing and the test area. A wide variety of chemicals was in use at the plant including hydrochloric acid, sulphuric acid, nitric acid, phosphoric acid, boric acid, acetic acid, chloroauric acid, hydrofluoric acid, nickel sulphate, nickel chloride, sodium hydroxide, potassium hydroxide, gold potassium cyanide, copper cyanide, potassium cyanide, silver cyanide, tin sulphate, chromium trioxide, methylene chloride, trichloroethylene, trichloroethane, cyclohexane, methylated spirits, ethylmethylketone, acetone, ethoxyethanol, lead solder and epoxy resin. Worker exposure to these chemicals is believed to have been well controlled.

Work history data currently available to the study team is limited to dates of hire and leaving employment; full work histories (dates of working in different departments) were not abstracted at the time of the study inception. Date of hire here relates to commencing employment with the parent company and not date of commencing employment in semiconductor production. In addition, duration of employment at the plant was based solely on periods of employment before 31 March 1983.

The National Health Service Central Register (NHSCR) of the Office for National Statistics (ONS) provided information on the vital status of each study subject on the closing date of the survey, 31 December 2002. Vital status information is summarized in Table 1. Mortality data were analysed for the period 1970–2002, with underlying cause of death coded by ONS [International Classification of Diseases (ICD)-8 1970–1978, ICD-9 1979–2000, ICD-10 2001–2002]. Cancer registration particulars were analysed for the period 1971–2001; national cancer registration data are not available before 1971. Site of cancer was also supplied by ONS (ICD-8 1971–1978, ICD-9 1979–1994, ICD-10 1995–2001). At the time of the last update [2], subjects who could not be traced by ONS were traced alive either by the Department of Social Security, by searching the electoral register or by being still employed with the company. Further use has not been made of these latter tracing procedures (see footnote to Table 1 for relevant dates last confirmed to be alive).


View this table:
[in this window]
[in a new window]
  		Table 1. Study population of semiconductor workers, West Midlands, UK, by vital status (as of 31 December 2002) and sex

 
The mortality experience of the cohort was compared with that which might have been expected to occur if mortality rates for the general population of England and Wales had been operating on the study cohort, having due regard to the composition of the study cohort by sex, age (5-year age groups) and calendar period (5-year calendar periods), except that mortality rates for the periods 1966–1970 and 1996–2000 were used to calculate expected numbers for 1970 and 1996–2002, respectively. Expectations based on person-years at risk (p.y.r.) were calculated using the PersonYears computer program [7]. Study subjects entered the p.y.r. on completion of 1 month employment or 1 January 1970, whichever was later. Individuals ceased to contribute to the p.y.r. on the date of death, date of embarkation, the date they were last confirmed to be alive or the closing date of the study, whichever was the earlier. Study subjects made no contribution to observed or expected numbers after their 85th birthday. There are three reasons for this censoring. Firstly, published mortality rates are only available for the ‘open-ended’ age group ≥85 years and the distribution of the cohort p.y.r. by single years of age may be very different to that of the general population; secondly, the reliability of cause of death particulars is probably poorer at later ages; and thirdly, any study subjects incorrectly coded as traced alive at the end of the study would have a disproportionate effect on the expected numbers for the open-ended age group.

In a similar manner, the cancer morbidity experience of the cohort was compared with expectations based on cancer incidence rates for England and Wales for the period 1971–2000. For these analyses, subjects could not enter the p.y.r. before 1 January 1971. Employees traced by sources other than the NHSCR are not included in analyses of cancer incidence.

Standardized mortality ratios (SMRs) and SRRs were calculated as the ratio of observed to expected numbers of deaths expressed as a percentage. In calculating P-values and 95% confidence intervals (95% CIs), deaths and cancer registrations were assumed to occur as a Poisson process. In addition, evidence was sought for any trend (linear component) in the pattern of SMRs or SRRs (e.g. any tendency for SMRs to increase or decrease with time since first employment) [8]. Tests of heterogeneity were also carried out (e.g. could the differences in SMRs by duration of employment represent no more than random variation in sub-groups) [8]. Both tests assume a similar null hypothesis; no trend and homogeneous SMRs or SRRs. Small P-values indicate statistical significance, and either the trend or the amount of heterogeneity is unlikely to have occurred by chance alone. All significance tests were two tailed.


   Results
Top
 Abstract
 Introduction
 Methods
 Results
Discussion
 Conflicts of interest
 References
 
Table 2 shows observed and expected numbers of deaths from main non-cancer disease groupings and individual sites of cancer for male and female employees and for the total cohort. CIs (95%) are only shown for those SMRs based on two or more observed deaths. Compared with national rates, all-causes mortality was close to expectation for males (Obs 86, Exp 87.08, SMR 99) and was significantly below expectation in females (Obs 221, Exp 297.92, SMR 74). Statistically significant deficits of deaths were found in females for breast cancer (Obs 13, Exp 27.38, SMR 47), all neoplasms (Obs 82, Exp 118.70, SMR 69), diseases of the circulatory system (Obs 83, Exp 105.11, SMR 79) and non-malignant diseases of the respiratory system (Obs 18, Exp 29.53, SMR 61). There was also a significant deficit of deaths for cancer of the oesophagus in males and females combined (Obs 0, Exp 4.03). Observed and expected numbers of deaths by year of hire, time since first employment and duration of employment were calculated for stomach cancer, rectal cancer, pancreatic cancer, lung cancer, malignant melanoma, breast cancer and cancer of the brain (not shown in table but available in a more detailed report [8]). The numbers of deaths available were small and no significant trends or significant heterogeneity was found in any set of SMRs.


View this table:
[in this window]
[in a new window]
  		Table 2. Mortality in semiconductor workers, West Midlands, UK, 1970–2002

 
Table 3 shows observed and expected numbers of cancer registrations for male and female employees and for the total cohort. In males, there was a significantly elevated SRR for cancer of the rectum (Obs 6, Exp 2.11, SRR 284). In females, there were significantly elevated SRRs for cancer of the pancreas (Obs 10, Exp 4.43, SRR 226) and malignant melanoma (Obs 11, Exp 4.97, SRR 221). In males and females combined, there were significantly elevated SRRs for cancer of the rectum (Obs 19, Exp 9.57, SRR 199) and malignant melanoma (Obs 12, Exp 5.53, SRR 217).


View this table:
[in this window]
[in a new window]
  		Table 3. Cancer incidence in semiconductor workers, West Midlands, UK, 1971–2001

 
Table 4 shows observed and expected numbers of registrations for rectal cancer, pancreatic cancer and malignant melanoma by year of hire, time since first employment and duration of employment for the total cohort. For cancer of the rectum there was a significantly elevated SRR for workers first employed in the 1970s (Obs 11, Exp 4.45, SRR 247). All cases of pancreatic cancer and malignant melanoma occurred in workers first employed in the period 1960–1979. There was no significant trend or significant heterogeneity in any set of SRRs. Corresponding findings for the incidence of other sites of cancer and for males and females separately are available in a more detailed report [6].


View this table:
[in this window]
[in a new window]
  		Table 4. Incidence of rectal cancer, pancreatic cancer and malignant melanoma among semiconductor workers (males and females combined), West Midlands, UK, 1971–2001

 

   Discussion
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
Conflicts of interest
 References
 
There was low overall mortality in the cohort as a whole compared with national rates, and this may well reflect, at least in part, the influence of a healthy worker effect. In such circumstances it has to be accepted that relying on comparisons with national rates can lead to health effects being missed. The study has, however, found significantly elevated morbidity for cancer of the rectum and malignant melanoma in the total study cohort. In addition, there was significantly elevated morbidity for cancer of the pancreas, but in female employees only. For cancer of the rectum and malignant melanoma the excess cancer morbidity risks were not mirrored in the corresponding mortality findings. Confident explanations for such disparities are not available. The excesses also need to be interpreted with caution because some significant excesses would be expected given the number of significance tests that have been carried out. The overall excess of melanoma incidence incorporates the original reported excess [1], and this cohort study should be viewed as providing a single report of melanoma excess and not two such reports. There was no suggestion of excess risks in this cohort of semiconductor workers for stomach cancer, lung cancer, breast cancer or cancer of the brain. More specifically, there was a significant deficit of deaths for female breast cancer and the corresponding deficit for cancer incidence approached formal levels of statistical significance.

The most important limitation to be attached to this study is the absence of job history information and the consequent absence of exposure histories. (At the time of the original data abstraction in the 1980s, the intention was that work history data, if required, would be considered in nested case–control studies.) It was not possible therefore to make any comparisons between workers employed in different departments. Attempts to identify whether any of the excess risks might be occupationally related were limited to analyses of risks by year of hire, period from first employment and duration of employment. These analyses did not suggest that any of the three cancers in excess (cancer of the rectum, cancer of the pancreas, malignant melanoma) were occupationally related. The absence of work history data means, however, that these latter analyses are of limited value and, if work history records are still available, more informative analyses of occupational exposures need now to be carried out. A further limitation of the study is that workers who ceased employment before 1970 could not be included in the cohort because factory records for these employees had been destroyed.

There is only one other published cohort study of workers engaged in semiconductor manufacture [3]. Excess risks reported by one study are not replicated in the other, and vice versa. There was, for example, no indication of excess morbidity for melanoma in the Scottish study (Obs 4, Exp 4.07). This may indicate that occupational exposures in the two facilities are very different and that different sets of occupational cancer risks are present. Alternatively, it is possible that occupational factors are not involved in the cancer excesses at either facility. Elucidation of possible cancer risks in the semiconductor manufacturing industry will require both more detailed epidemiological analyses of cancer risks in relation to exposure histories and the initiation of further cohort studies. It will be important that such studies are carried out on complete cohorts of workers and not on cohorts of volunteers; the latter option seems to be the epidemiological approach favoured by the present UK authorities.


   Conflicts of interest
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Conflicts of interest
References
 
None declared.


   Acknowledgements
 
We thank the ONS for providing follow-up details. We thank the company management and workforce safety representatives for permission to identify this cohort in the 1980s. We thank M. J. McKiernan and R. H. R. Aston for their earlier assistance with this cohort study and for constructive comment on earlier drafts of the current report. We thank Margaret Williams for word processing. Survey costs of the current analysis were defrayed by a research award from the HSE.


   References
Top
 Abstract
 Introduction
 Methods
 Results
 Discussion
 Conflicts of interest
 References
 

  1. Sorahan T, Waterhouse JAH, McKiernan MJ, Aston RHR. Cancer incidence and cancer mortality in a cohort of semiconductor workers. Br J Ind Med 1985;42:546–550.[ISI][Medline]

  2. Sorahan T, Pope DJ, McKiernan MJ. Cancer incidence and cancer mortality in a cohort of semiconductor workers: an update. Br J Ind Med 1992;49:215–216.[ISI][Medline]

  3. McElvenny DM, Darnton AJ, Hodgson JT, Clarke SD, Elliott RC, Osman J. Investigation of cancer incidence and mortality at a Scottish semiconductor manufacturing facility. Occup Med (Lond) 2003;53:419–430.

  4. Bailar JC, Greenberg M, Farrison R, LaDou J, Richter E, Watterson A. Cancer risk in the semiconductor industry: a call for action. Int J Occup Environ Health 2002;8:163–168.[ISI][Medline]

  5. Fisher J. Cancer in the semiconductor industry. Arch Environ Health 2002;57:95–97.[ISI][Medline]

  6. Nichols L, Sorahan T. Update of cancer incidence and cancer mortality in a cohort of semiconductor workers. HSE Publication RR265. Sudbury: HSE Books, 2004.

  7. Coleman M, Douglas A, Herman C, Peto J. Cohort study analysis with a Fortran computer program. Int J Epidemiol 1986;15:134–137.[Abstract/Free Full Text]

  8. Breslow NE, Day NE. Statistical methods in cancer research. Volume II—the design and analysis of cohort studies. IARC Scientific Publication No. 82. Lyon: IARC, 1987.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
55/8/625    most recent
kqi156v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Nichols, L.
Right arrow Articles by Sorahan, T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Nichols, L.
Right arrow Articles by Sorahan, T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?