McLaughlin, Joseph K. PhD; Mumma, Michael T. MS; Sonderman, Jennifer S. MHS; Farnsworth, Eli P. BA; Lipworth, Loren ScD
Manufacture and testing of space satellites and related equipment began in 1961 at the Valley Forge complex in King of Prussia, Pennsylvania, currently operated by the Lockheed Martin Corporation. The Valley Forge complex was originally owned by the General Electric Company, and has transitioned through multiple owners (General Electric, Martin Marietta, and Lockheed Martin). Lockheed Martin took over operation of the Valley Forge complex in 1995. Operations within the Valley Forge complex have entailed highly specialized prototype satellite manufacturing in cleanrooms, including manufacturing of Defense Satellite Communications System (DSCS), LANDSAT satellites, and Global Positioning System (GPS), re-entry systems testing, and computer-based research and software development. Over time the work force has varied in size, with as many as 8000 to 10,000 employees active during the 1960s through the early 1980s.
Within the past decade, there have been self-reports of kidney cancer diagnoses among employees who worked at the Valley Forge complex, raising concern about the possibility of an increased risk of renal cell cancer resulting from possible occupational exposures. To assess the risk of cancer overall, and renal cell cancer in particular, we performed a long-term retrospective mortality follow-up study of 29,504 workers employed during the years 1962 through 2008 at the Valley Forge complex.
The study was reviewed and approved by the institutional review board of Independent Review Consulting, Inc (www.irb-irc.com).
The study cohort comprises all aerospace workers employed for at least 1 year at the Valley Forge satellite manufacturing complex in King of Prussia, Pennsylvania between January 1, 1962, and December 31, 2008. Workers were defined as having been active in the General Electric Aerospace business unit prior to 1993, or its equivalent after facility ownership was transferred to Martin Marietta in 1993 and Lockheed Martin in 1995. Overall, 35,392 workers were identified using six independent and overlapping data sources consisting of pension, benefits, payroll, and human resources records, which provided name, date of birth, Social Security Number (SSN), sex, work facility location, and dates of employment. After excluding 5888 workers employed for less than 1 year, the final study cohort comprises 29,504 workers employed at the Valley Forge complex. In the process of examining the various databases used to identify the study cohort, 34,346 aerospace workers were identified who never worked at the Valley Forge complex. As these workers likely performed similar work at smaller aerospace facilities in Florida, California, and Virginia and had similar socioeconomic characteristics as Valley Forge workers, they were included in the follow-up as an additional population to examine if mortality patterns were similar to those of Valley Forge workers.
Vital Status Determination
Vital status was sought for all 29,504 Valley Forge workers and all 34,346 other aerospace workers through December 31, 2008. Mortality and vital status were determined by linkage with the Social Security Administration (SSA) Death Master File (1960–2008), the SSA Epidemiologic Research Service for alive status (1960–2008) (http://www.ssa.gov/policy/about/epidemiology.html), the National Death Index (NDI) (1979–2008) (http://www.cdc.gov/nchs/ndi.htm), the California Death Statistical Master File (1960–2008), Comserv, a computer services firm specializing in locating persons (www.comserv-inc.com), and LexisNexis, an online information service provider (www.lexisnexis.com), as well as company pension records. Cohort members were matched against these registers using SSN, name, date of birth, and sex to identify those who died and to determine date and cause of death. For matching with the SSA Death Master File and the California Death Statistical Master File, we used the Centers for Disease Control and Prevention matching program, LinkPlus, which incorporates a probabilistic scoring system and does not require exact matches on all variables.1
Underlying cause of death was obtained from the NDI or from the California Death Statistical Master File for all deaths occurring in the United States in 1979 through 2008. For deaths prior to 1979, death certificates were obtained from company benefits claim records or state vital statistics departments. A trained nosologist coded causes of death from the death certificates according to the International Classification of Disease (ICD) revision in use at the time of death.
Excluding 20 deaths that occurred outside the United States, a total of 6583 (22.3%) workers in the study cohort were known to have died by December 31, 2008; among these, cause of death was known for 6499 (98.7%). Using SSA files, LexisNexis residential records of last known address, credit bureau records, voter registration records, property tax records, and motor vehicle records, we confirmed that 22,498 workers not identified as having died were alive as of December 31, 2008. For the remaining 403 workers (1.4% of workers), vital status as of December 31, 2008 was unknown.
Standardized mortality ratios (SMRs), the ratios of the observed to the expected numbers of deaths, and 95% confidence intervals (CI) were calculated to determine whether the mortality experience of the Valley Forge workers differed from that of the general population of the United States. Observed numbers of deaths from cancers and other diseases were counted by sex, age, and calendar year for Valley Forge workers. Complete information on race was available only after 2000 for members of the cohort; almost 90% of workers employed after 2000 were white, as were 91% (2591 of 2843) of employees who terminated employment prior to 2000 and for whom hard-copy personnel folders providing race information were available, confirming that the overwhelming majority of employees at the Valley Forge complex in all decades of operation were white. Therefore, expected numbers of deaths were computed based on age-, calendar year-, and sex-specific rates in the white United States general population. Expected values were also calculated based on the mortality experience of the white Pennsylvania population, but because more than 40% of deaths among the workers were recorded in states other than Pennsylvania and results did not materially differ whether national or Pennsylvania rates were used, only SMRs for the cohort relative to United States mortality rates are presented.
Person-years of follow-up began after accruing 1 year of employment at the Valley Forge complex or January 1, 1963, whichever came later. Follow-up ended on the date of death, date lost to follow-up, December 31, 2008 or age 95, whichever occurred first. Workers with unknown vital status (n = 403) or those who died outside the United States (n = 20) were considered lost to follow-up and assumed to be alive until their date of last company employment or date of last known residential address in the United States, whichever occurred later. Standardized mortality ratios and 95% CIs were calculated using OCMAP-Plus software,2 assuming the observed number of deaths followed a Poisson distribution. Standardized mortality ratios were calculated for total mortality and for mortality from cancer overall and at specific sites, including renal cell cancer, and for noncancer causes of death. Observed and expected numbers of deaths were also distributed over categories of calendar year of hire and duration of employment at the Valley Forge complex and trend analyses were conducted using methods based on the Poisson distribution.3
To account for the favorable mortality experience among newly hired workers, SMRs were also conducted excluding the first 10 years of follow-up (data not shown). As there were no material differences in the SMRs for cancer, and in particular renal cell carcinoma, the tables in this article include all follow-up periods.
Table 1 shows the demographic and employment characteristics of the 29,504 workers in the study cohort. At the end of follow-up, December 31, 2008, 3791 (13%) workers were actively employed at the Valley Forge complex. Most workers were men (76%), born before 1950 (64%), and hired before 1990 (82%), with 51% hired before 1970. Almost 70% of workers were employed for less than 10 years, and 73% were followed for 20 or more years.
Standardized mortality ratios for cancer overall and at specific sites, as well as for other selected causes of death, are presented in Table 2. The total person-years of observation was 829,744, and the average follow-up time was 28.1 years (SD = 13.1 years). The SMRs for all causes of death (SMR = 0.70; 95% CI = 0.68–0.71) and for all cancer deaths (SMR = 0.79; 95% CI = 0.76–0.83) were significantly lower than 1.0. Forty-three deaths from kidney cancer were observed, which was significantly below the 67.2 expected based on rates prevailing in the general population (SMR = 0.64; 95% CI = 0.46–0.86). The workers had significantly low risks of death from several other types of cancer, including cancers of the stomach (SMR = 0.72; 95% CI = 0.53–0.95; n = 47), liver (SMR = 0.59; 95% CI = 0.42–0.80; n = 40), lung (SMR = 0.65; 95% CI = 0.60–0.70; n = 587), and non-Hodgkin lymphoma (SMR = 0.77; 95% CI = 0.61–0.96; n = 78). Of the 25 site-specific cancer SMRs presented in Table 2, 19 were below 1.0 and six were higher than 1.0. The only statistically significantly elevated SMR was that for brain and other central nervous system cancer (SMR = 1.34; 95% CI = 1.08–1.63), based on 95 observed deaths compared with 71.2 expected. Significantly low death rates were observed for virtually all noncancer causes of death, including heart and cerebrovascular diseases, diabetes, nonmalignant respiratory disease, nervous system diseases, and external causes, which contributed to the overall mortality deficit in the cohort.
Standardized mortality ratios are also presented by duration of employment at the Valley Forge complex (Table 3). Overall mortality, as well as risk of death from all cancers and at specific sites, was generally similar among the workers with the longest duration of employment compared with shorter-term employees. The all-cause SMR was 0.69 (95% CI = 0.66–0.72) both among workers employed for 10 or more years and among those employed less than 5 years; the corresponding SMRs for cancer overall were 0.80 (95% CI = 0.74–0.86) and 0.78 (95% CI = 0.72–0.84), respectively (Table 3). The number of observed deaths from kidney cancer was similarly below expectation across all categories of duration of work, with SMRs of 0.80 (95% CI = 0.47–1.26), 0.48 (95% CI = 0.23–0.89) and 0.63 (95% CI = 0.35–1.03) among those who worked for less than 5 years, 5–10 years, and 10 or more years, respectively. Standardized mortality ratios for lung cancer were significantly low across all categories of duration of employment. There were no statistically significant elevations of any cause of death for any duration of employment period with the exception of a SMR of 1.52 (95% CI = 1.06–2.11) for brain cancer among those employed for 10 or more years, based on 35 observed deaths. The SMRs for brain cancer among employees who worked for less than 5 years (SMR = 1.10; 95% CI = 0.74–1.58) or 5 to 10 years (SMR = 1.42; 95% CI = 0.97–2.02) were also elevated, although not significantly; the test for trend of increasing SMRs across duration of employment was not statistically significant (P for trend = 0.20).
Results of analyses by calendar year of hire are presented in Table 4. Overall, 5824 of the 6583 (88.5%) observed deaths occurred among workers hired before 1970, whereas only 2% of deaths occurred among workers hired after 1986. The all-cause SMRs were 0.73 (95% CI = 0.71–0.75), 0.58 (95% CI = 0.54–0.63), and 0.38 (95% CI = 0.32–0.45) among those hired before 1970, between 1970 and 1985, and after 1986, respectively. The all-cancer SMRs, 0.81 (95% CI = 0.78–0.85), 0.75 (95% CI = 0.66–0.85), and 0.45 (95% CI = 0.33–0.62), were also significantly reduced among those hired in all three categories of year of hire, respectively, as were the SMRs for lung cancer. For most other types of cancer examined, including kidney cancer, SMRs were below 1.0, although generally not significantly so, regardless of calendar time period of hire. Thirty-six of the 43 observed kidney cancer deaths occurred among those hired in the earliest calendar time period (before 1970), yielding a statistically significantly reduced SMR of 0.64 (95% CI = 0.45–0.88) among these early workers. Seventy-nine (83%) of the 95 observed deaths from brain and other central nervous system cancer occurred among those hired before 1970, yielding a statistically significantly increased SMR of 1.39 (95% CI = 1.10–1.73); the SMR for brain cancer was similarly but not significantly increased among those hired between 1970 and 1985 (SMR = 1.36; 95% CI = 0.74–2.28), and decreased to 0.51 (95% CI = 0.06–1.84) among those hired in the most recent calendar period, 1986–2007, although the latter estimate was based on only two deaths.
Mortality overall and from cancer was substantially decreased among 29,504 workers involved in the manufacture and testing of space satellites and related equipment at the Valley Forge complex. The follow-up spanned nearly 50 years and resulted in 829,744 person-years of observation. The SMRs for all causes of death and for cancer overall were 0.70 (95% CI = 0.68–0.71) and 0.79 (95% CI = 0.76–0.83), respectively. Such substantially decreased mortality rates, as well as the low rates of death observed for heart disease, cerebrovascular disease, nonmalignant respiratory disease, and external causes such as suicides, compared with the general population suggest that our cohort of highly skilled workers employed in satellite manufacturing was relatively healthy, of higher socioeconomic status, and less likely to smoke, a pattern commonly observed in skilled occupational cohorts.4
Risk for renal cell cancer, the cancer site of a priori interest due to recent self-reported cases at the Valley Forge complex, was significantly decreased by about 35% among workers in our cohort compared with the general population. Examination of SMRs according to duration of employment showed no increases in renal cancer risk for any subgroup. Similarly, no excess risk for kidney cancer was seen among workers employed at the Valley Forge complex during any calendar time period.
In the United States in 2010, 58,240 new cases and 13,040 deaths were expected from kidney cancer.5 Approximately, 85% of kidney cancers are renal cell cancers, whereas the remainder are mainly urothelial cancers of the renal pelvis.6 Renal cell cancer is about twice as common among men as among women,5 with the mean age at diagnosis in the early 60s. Cigarette smoking, obesity, and hypertension are the most consistently established risk factors for renal cell cancer, but taken together cannot account for the majority of these cancers.6–8 Although the total rate of new cancer diagnoses overall dropped 1.8% among men and 0.5% among women in the United States between 2001 and 2005,9 incidence rates specific for renal cell cancer have been rising by about 2% each year in the United States over the past several decades. Increases in incidence have been more rapid among women than men and, in particular, among blacks than whites.6,10 Despite increasing incidence rates, kidney cancer mortality rates in the United States have been relatively stable over the past 25 years, suggesting a benefit of early detection and surgical treatment. Given the relatively high kidney cancer survival rates, with 5-year relative survival reaching almost 70% overall and 90% for localized renal cell cancer today, an increased incidence of renal cell cancer among Valley Forge workers cannot be ruled out but is unlikely because kidney cancer survival rates were substantially lower during much of the follow-up of our cohort.
Detailed employment histories, such as departments or job titles, or information on specific occupational exposures for the individual workers were not available. Limited data were available on the primary satellite manufacturing and testing processes used at the Valley Forge complex only from 1980 on, including mechanical assembly, electrical assembly, machining, electroplating, and laser operations. On the basis of extensive reviews of these processes, including the DSCS, LANDSAT, and GPS satellite programs, there was no known widespread workplace exposure among the Valley Forge workers suspected a priori to be associated with renal cell cancer. Renal cell cancer is not generally considered to be an occupationally associated tumor,6,11 although a few sporadic associations have been reported between exposures or jobs/industries and renal cell cancer. Most attention has been focused on asbestos, gasoline and, more recently, the solvent trichloroethylene (TCE). However, comprehensive reviews and meta-analyses of occupational cohort studies have failed to confirm suspicions of excess kidney cancers among workers exposed to asbestos or gasoline;12–15 and the weight of the evidence to date does not provide consistent, credible support for the hypothesis that TCE is a cause of renal cell cancer in humans.11,16 Several other occupational associations, including among workers exposed to heavy metals or employed in metal-related industries, have been sporadically reported in the literature, although none has been consistently associated with an increase in renal cell cancer risk. Arsenic, beryllium, and cadmium and their compounds, as well as nickel compounds and chromium [VI] compounds are classified by the International Agency for Research on Cancer (IARC) as established human carcinogens; however, for none of these agents was the classification based on evidence of increased risk of renal cell cancer in humans.17–19
Our analyses of deaths from more than 20 different cancer types revealed a significant 30% excess of deaths due to brain and other central nervous system cancer among the Valley Forge satellite workers. When conducting multiple statistical analyses of numerous causes of death, it is to be expected that some elevated risks would occur by chance alone. Furthermore, we relied in this study on death certificates for the ascertainment of brain cancer deaths, which may reflect metastases from other sites and may be of questionable diagnostic accuracy during our almost 50-year study period.20–24 We also cannot rule out the possibility that the observed excess of brain tumors among the Valley Forge workers compared with the general population reflects a diagnostic sensitivity bias,25 which is plausible if these highly skilled workers had access to enhanced medical surveillance and diagnostic testing over the years, resulting in more accurate diagnoses of brain cancers26 and therefore increased likelihood of recording brain cancer as a cause of death compared with persons in the general population. Similar increased risks for brain cancer have also been reported in other skilled occupational groups of high socioeconomic status, including professional/technical aerospace workers in California, as well as physicians, dentists, and other health professionals,27–29 but whether these reflect diagnostic bias or a true etiologic association is unknown. Therefore, to investigate further the elevated SMR for brain cancer observed among the Valley Forge workers, and to determine whether it represents a real increase or an artifact due to death certificate-based reporting for this cancer, we are currently conducting an incidence study to assess whether the occurrence of brain cancer in this population of workers is higher than expected in the general population.
In light of the a priori interest in renal cell cancer and the observed excess of brain cancer among the Valley Forge workers, we also examined mortality from these two cancers through December 31, 2008 among the 34,346 aerospace workers who likely performed similar or related work at smaller aerospace facilities primarily in Florida, California, and Virginia, but did not work at the Valley Forge complex. Somewhat similar mortality patterns were observed among these non-Valley Forge aerospace workers, who had a slightly reduced SMR for kidney cancer (SMR = 0.98; 95% CI = 0.66–1.41), based on 30 deaths, and an elevated SMR for brain cancer (SMR = 1.16; 95% CI = 0.83–1.58), based on 39 deaths. In comparison with the Valley Forge workers, the 34,346 non-Valley Forge workers were much younger (67% were born after 1950 compared with 36%) and were hired in later years (two-thirds were hired after 1990 compared with 18%), so their comparability to the Valley Forge study cohort is uncertain. Nonetheless, the general similarity of the kidney and brain cancer mortality patterns may be worth noting.
To our knowledge, this is the first epidemiologic study of workers in the space satellite manufacturing and testing industry. This long-term retrospective mortality study spanned five decades, during which ownership of the Valley Forge facility passed through three different companies. Using numerous electronic and hard copy sources from all three companies, we are confident that we were able to create a complete roster of workers at this complex during the period under study. Ascertainment of vital status for cohort members was virtually complete (vital status was known for 99% of cohort members), and we obtained cause of death for 99% of the workers who had died. A limitation of the study is the absence of job descriptions, including departments or job titles, for individual workers, which would have allowed us to characterize cancer mortality patterns by specific occupational categories. However, there was no workplace exposure postulated a priori to be associated with renal cancer risk in this satellite manufacturing and testing complex. Because 78% of workers were alive at last follow-up in 2008, and almost 90% of deaths occurred among workers who were hired prior to 1970, the first decade of operation, further follow-up of this cohort may be warranted to evaluate fully the long-term cancer mortality patterns of employees at the Valley Forge complex.
In summary, the study found no evidence of increased cancer mortality overall or from renal cancer in particular among a cohort of workers employed at a satellite manufacturing and testing complex. The finding of an increased risk of death from brain cancer warrants further investigation in a cancer incidence study in this worker population.
1. Campbell KM, Deck D, Krupski A. Record linkage software in the public domain: a comparison of Link Plus, The Link King, and a ‘basic’ deterministic algorithm. Health Informatics J. 2008;14:5–15.
2. Marsh GM, Youk AO, Stone RA, et al.. OCMAP-PLUS: a program for the comprehensive analysis of occupational cohort data. J Occup Environ Med. 1998;40:351–362.
3. Breslow NE, Day NE. Statistical methods in cancer research. Volume 2. The design and analysis of cohort studies. IARC Sci Publ. 1987;82.
4. Checkoway H, Pearce N, Crawford-Brown DJ. Research Methods in Occupational Epidemiology. New York, NY: Oxford University Press; 1989.
5. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. Ca Cancer J Clin. 2010;60:277–300.
6. Lipworth L, Tarone RE, Lund L, McLaughlin JK. Epidemiologic characteristics and risk factors for renal cell cancer. Clin Epidemiol. 2009;1:33–43.
7. Benichou J, Chow WH, McLaughlin JK, et al.. Population attributable risk of renal cell cancer in Minnesota. Am J Epidemiol. 1998;148:424–430.
8. Calle EE, Kaaks R. Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nature Rev Cancer. 2004;4:579–591.
9. Jemal A, Thun MJ, Ries LAG, et al.. Annual report to the nation on the status of cancer, 1975–2005, featuring trends in lung cancer, tobacco use, and tobacco control. J Natl Cancer Inst. 2008;100:1672–1694.
10. Kosary CL, McLaughlin JK. Miller BA, Ries LAG, Hankey BE, eds. Kidney and renal pelvis, In: Cancer Statistics Review: 1973–1990. National Cancer Institute. NIH Pub. No. 93–2789; 1993:Xl–X22.
11. McLaughlin JK, Lipworth L, Tarone RE, Blot WJ. Schottenfeld D, Fraumeni JF Jr, eds. Renal cancer. In: Cancer Epidemiology and Prevention. 3rd ed. New York, NY: Oxford University Press; 2006: 1087–1100.
12. Sali D, Boffetta P. Kidney cancer and occupational exposure to asbestos: a meta-analysis of occupational cohort studies. Cancer Causes Control. 2000;11:37–1147.
13. McLaughlin JK. Renal cell cancer and exposure to gasoline: a review. Environ Health Perspect Suppl. 1993;101(suppl 6):111–114.
14. International Agency for Research on Cancer. Occupational exposures in petroleum refining. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Vol. 45. Occupational exposures in petroleum refining, crude oil, and major petroleum fuels. Lyon: IARC, 1989:39–117.
15. Wong O, Raabe GK. Critical review of cancer epidemiology in petroleum industry employees, with a quantitative meta-analysis by cancer site. Am J Ind Med. 1989;15:283–310.
16. Wong O. Carcinogenicity of trichloroethylene: an epidemiologic assessment. Clin Occup Environ Med. 2004;4:557–589.
17. International Agency for Research on Cancer. Arsenic in drinking-water. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Vol. 84. Some Drinking-water Disinfectants and Contaminants, including Arsenic. Lyon: IARC; 2004:39–267.
18. International Agency for Research on Cancer. Cadmium and cadmium compounds. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Vol. 58. Beryllium, Cadmium, Mercury, and Exposures in the Glass Manufacturing Industry. Lyon: IARC; 1993: 119–237.
19. International Agency for Research on Cancer. Chromium and chromium compounds. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Vol. 49. Chromium, Nickel and Welding. Lyon: IARC; 1990:49–256.
20. Engel LW, Strauchen JA, Chiazze L Jr, Heid M. Accuracy of death certification in an autopsied population with specific attention to malignant neoplasms and vascular diseases. Am J Epidemiol. 1980;111:99–112.
21. Percy C, Stanek E, Gloeckler L. Accuracy of cancer death certificates and its effect on cancer mortality statistics. Am J Public Health. 1981;71:242–250.
22. Percy CL, Miller BA, Ries LAG. Effect of changes in cancer classification and the accuracy of cancer death certificates on trends in cancer mortality. Ann NY Acad Sci. 1990a;609:87–97.
23. Percy C, Ries LG, Van Holten VD. The accuracy of liver cancer as the underlying cause of death on death certificates. Public Health Reports. 1990b;105:361–367.
24. Schoenberg BS. Schottenfeld D, Fraumeni JF Jr, eds. Nervous system. In: Cancer Epidemiology and Prevention. 2nd ed. New York, NY: Oxford Univ Press; 1996:968–983.
25. Greenwald P, Friedlander BR, Lawrence CE. Diagnostic sensitivity bias—an epidemiologic explanation for an apparent brain tumor excess. J Occup Environ Med. 1981;23:690–694.
26. Inskip PD, Hoover RN, Devesa SS. Brain cancer incidence trends in relation to cellular telephone use in the United States. Neuro Oncol. 2010;12:1147–1151.
27. Preston-Martin S, Munir R, Chakrabarti I. Schottenfeld D, Fraumeni JF Jr, eds. Nervous system cancers. In: Cancer Epidemiology and Prevention. 3rd ed. New York, NY: Oxford Univ Press, 2006:1173–1195.
28. Savitz D, Trichopoulos D. Adami HO, Hunter D, Trichopoulos D, eds. Brain cancer. In: Textbook of Cancer Epidemiology. New York, NY: Oxford University Press; 2002:486–503.
29. Thomas TL, Waxweiler RJ. Brain tumors and occupational risk factors: a review. Scand J Work Environ Health. 1986;12:1–15.
©2011The American College of Occupational and Environmental Medicine