Document e5N226MM4JXrnQ80aDq9nRX6e

Final Report Mortality of Employees of an Ammonium Perfluorooctanoate Production Facility Jessica I. Lundin, MPH Bruce H Alexander, PhD Associate Professor Principal Investigator Division of Environmental Health Sciences University of Minnesota, School of Public Health August 22, 2007 Executive Summary Background: This report presents the results of a cohort mortality study of workers employed at the 3M Ammonium Perfluorooctanoate (APFO) production facility in Cottage Grove Minnesota. The purpose of this study was to examine possible associations between working in jobs with varying exposure to APFO and specific causes of death. Methods: The study population included all employees of the Cottage Grove facility with a minimum of 365 days of cumulative employment prior to 1997. The population was followed from the time they entered the workforce through 2002. Underlying and contributing causes of death were obtained from the National Death Index and death certificates. Work history records were used to determine potential APFO exposure. Each job held was assigned to one of three exposure categories; non-exposed, probable APFO exposure, and definite APFO exposure. Exposure categories were initially established based on duration of employment in each of these jobs. The cohort members were classified as ever or never having a job with probable exposure, probable or definite exposure or for a minimum period of time in these job categories. To estimate cumulative exposure a weighted cumulative exposure model was constructed by multiplying the duration of employment by an exposure weight of 1 for non exposed, 30 for probable exposure and 100 for definite exposure. The weights, though somewhat arbitrary, reflect differences in biological monitoring of the jobs and consider the long half-life of the chemical. The cumulative exposures were then categorized as estimates equivalent to less than one, one to five and more than five years of employment in high exposure jobs. 2 Standardized mortality ratios (SMR) were estimated for all cause and cause specific mortality using mortality rates from the general population of Minnesota as a reference. SMR estimates were made for the sub-cohorts ever exposed to APFO, by exposure category, and for a minimum of one year of exposure. Time dependent Cox regression models were used to estimate the risk of specific causes of death using an internal referent population. Risks were estimated for working in a definite exposure job for a minimum of 6 months (high exposure) and ever in a job with probable exposure or a job with definite exposure for less than 6 months (moderate exposure) in comparison to those who never held a job with definite or probable exposure (low exposure). A similar time dependent analysis for cumulative exposure used weighted estimates equivalent to less than one, one to five and more than five years in a job with definite exposure. All models were adjusted for sex, birth year, year of eligibility, wage type (salary, hourly or both), and an estimate of smoking habit. Causes of death of a priori interest were liver, pancreatic and testicular cancer and cirrhosis of the liver based on animal toxicological data, and prostate cancer and cerebrovascular disease from prior analyses of this cohort. Results: There were 3993 eligible workers, and 807 deaths, in the cohort. The SMR for all causes of death was 0.8 (95% Cl 0.7-0.9) and 0.9 (95% 0=0.8-1.0) for malignant causes of death. No malignant or non-malignant causes of death had significantly elevated for any exposure classification, however the members of the non-exposed cohort had significantly reduced mortality due to prostate cancer (SMR=0.4, 95% 0=0.1-0.9), cerebrovascular disease (SMR=0.5, 95% 0=0.3-0.8), and ischemic heart disease (SMR=0.7, 95% 0=0.3-0.8). The o members of the cohort with definite exposure did have lower than expected rate of death from ischemic heart disease, but not prostate cancer (SMR=2.1, 95% CI=0.4-6.1), or cerebrovascular disease (SMR=1.6, 95% 0=0.5-3.7) The time dependent analysis using the internal referent population produced hazard ratios indicating an association with APFO exposure and prostate cancer and cerebrovascular disease. Those with high or moderate exposure had an elevated risk of dying from cerebrovascular disease (HR=5.1,95% Cl = 1.4-18.6 and HR= 2.1, 95% Cl = 1.0-4.6, respectively) and prostate cancer (HR=7.0, 95% Cl = 1.2-42.2 and HR=3.0, 95% Cl = 0.8-11.0, respectively). The highest cumulative exposure score was associated with an elevated risk of dying from cerebrovascular disease (HR=2.4, 95% CI= 1.1-5.5) and prostate cancer (HR=3.8, 95% CI=1.2-13.2). When stratified by wage type similar associations were observed for prostate cancer and cerebrovascular disease. Lagging exposures by 10 years made little or no difference in the hazard ratio estimates. Conclusion: APFO exposed workers did not have an elevated risk of death when compared to the population of the state of Minnesota, however within the cohort risk of death from prostate cancer and cerebrovascular disease was elevated for workers with higher estimated exposure. Interpreting the somewhat contradictory results requires caution and consideration of several assumptions. /( priori causes of death selected based on animal toxicology studies, liver, pancreatic and testicular cancer and cirrhosis of the liver, were not observed to be associated with APFO exposure. 4 ntroduction Ammonium Perfluorooctanoate (APFO, CF3(CF2)6C02'NFi4>') is a thermally stable synthetic surfactant which is resistant to a large variety of chemicals, and resistant to degradation. These unique properties led to its use in industrial applications, as well as ist use as an ingredient in consumer products such as non-stick coatings for cookware, clothing, automotive products, and paper coatings (sandwich wrappers, popcorn bags, etc) for oil and moisture resistance (1). In the presence of biologic media, APFO dissociates into perfluorooctanoic acid (PFOA, CF3(CF2)6COOH;), an anion of APFO. APFO was produced at the 3M facility in Cottage Grove, Minnesota from 1947 until the year 2000. Laboratory studies have shown PFOA to be absorbed through inhalation, ingestion, and dermal contact and is distributed in the liver and in the blood with notable gender differences varying by species (2). PFOA is metabolically inert; it is not biotransformed or conjugated and is eliminated primarily through urine and feces (2). Biological monitoring data indicate that occupational exposures to APFO in manufacturing workers result in median serum PFOA levels in the range of 0.3 to 5.2 parts per million (ppm), depending on the work area (3). The general population has been shown to have average sera PFOA concentrations of 5 parts per billion (ppb) (4, 5). A comparisons of banked serum from community base samples showed an increase in serum concentrations between 1974 and 1989, but no significant change between 1989 and 2001(6). Preliminary data suggest that the population exposure has decreased following the phase out of the production of these materials (7). PFOA is a peroxisome proliferator with demonstrated effects in laboratory animals. Animal studies have shown effects on the liver, atrophy of lymphoid tissues (spleen and lymph nodes), benign hepatocellular tumors,, pancreatic acinar cell tumors, and testicular leydig cell 5 tumors (2) A previous mortality study of this cohort have shown an association with working in the chemical division of the APFO manufacturing facility and dying from prostate cancer(8), and a updated analysis identified a potential association with high APFO exposure and death from cerebrovascular disease (9). The purpose of this study with an updated mortality follow-up was to further evaluate potential associations between occupational APFO exposure and specific causes of death in this cohort. 6 Methods Study Population The protocol for this study was reviewed and approved by the University of Minnesota Institutional Review Board. This occupational cohort of workers at the APFO production plant in Cottage Grove, Minnesota included all workers with one year (365 days) of cumulative employment between the dates of January 1, 1943 and December 31, 1997. Cohort members meeting eligibility criteria were followed until either December 31, 2002 or their date of death. The human resource records of all eligible employees were abstracted for demographic information, including the worker's name, social security number, employee identification number, date of birth, and details of work history. The latter included job specific start and end dates, department codes, and job classifications. Demographic infonnation and vital status was verified using TRW/Experian (a credit reporting agency) and the Social Security Administration service for epidemiologic research studies. Determination of Vital Status Vital record searches were performed for all cohort members not employed by the company on December 31,2002 and who had not been identified as deceased in previous studies (8, 9). Vital status of these cohort members was detennined through the National Death index (ND1). All potential matches from the NDI search were reviewed by hand to ensure a valid match. The underlying and contributing cause of death was obtained from NDI Plus for deaths after 1979. Death certificates were obtained for decedents who died prior to 1979 and were coded by a certified nosologist for causes of death in the International Classification of Disease (ICD) revision in effect at the time of death. If a death certificate could not be found for an NDI 7 identified decedent, the individual cause of death was classified in the `other' cause of death category. Exposure Assessment The goal of this study was to describe mortality experience in relation to fluorochemical exposure with particular interest in PFOA. Some biological monitoring data are available for more recent years, however they are insufficient to fully characterize exposure to workers over the life of the plant. Therefore, the exposure assessment relied on work history records and expert knowledge of the history of the APFO manufacturing process to first develop a qualitative exposure assessment for each job held by the cohort members. All perfluorochemical development and production occurred in departments affiliated with the chemical division of the plant. The specific areas where APFO and other periluorochemicals were produced changed over the years, as did the work area department codes, thus exposure levels could not be assigned to workers based on work history alone. To remedy this, the department codes were reviewed by year to determine the building and division assigned to each code. The resulting lists were reviewed independently by a panel of veteran workers and plant industrial hygienists to determine where the perfluorochemical production, or the development of the perfluorochemical products, took place over the history of the facility. The individual responses were summarized and the panel met as a group to discuss any discrepancies and confirm the exposure assignments. The available information permitted the panel to classify the jobs in the work histories into three general classifications of exposure. Definite occupational APFO exposure: These jobs were in areas where electrochemical fluorination, drying, shipping, and packaging of APFO occurred and the worker would be 8 exposed to APFO on a regular basis. These jobs also had the potential for higher exposures. Probable occupational APFO exposure: These jobs were in other chemical division areas where exposure to APFO was possible, but the exposures were considered lower or transient. No or minimal occupational APFO exposure: These jobs were primarily in the non chemical division of the plant. These workers may have had the opportunity for some exposure if they passed through an exposed area, but were not exposed specifically as part of their job. They likely had higher exposures than the general population due to contamination at the work site. Hereafter these job exposure subgroups will be referred to as definite APFO exposure, probable APFO exposure, and non-exposed. Exposure Classification for Analysis The employees of this plant changed job classifications frequently, thus they could not be classified into discrete, mutually exclusive exposure groups that mirrored the job exposure subgroups. To accommodate the migration between exposure subgroups we incorporated two approaches for characterizing APFO exposure in the analysis. The primary analysis is based on ever attaining a minimum tenure in jobs with probable or definite exposure. A secondary analysis used a cumulative exposure model developed to explore a weighted exposure distribution based on duration of employment and an assumed exposure intensity. Exposure by .Job Classification. The initial analyses explored the mortality experience of workers compared to that of the general population of Minnesota with respect to their entire 9 work history of ever working in jobs with definite or probable exposure, a minimum of one year in definite or probable exposure jobs, or only working in the non-exposed jobs. Subsequently a more restrictive classification was developed for an analysis using an internal referent population which classified the cohort members as; ljthose working only in jobs not exposed to fluorochemicals (low' exposure), 2) ever working in a probable APFO exposure job or working a definite APFO exposure job for less than 6 months (moderate exposure), or 3) employment in a definite exposure job for 6 months or more (high exposure). Entry into the latter two categories could occur at varying points in the individual's work history. This model assumes that once a person is employed in a high exposure job for a minimum period of time or a moderate exposure job they reach a threshold where their risk is different from individuals working in jobs without direct occupational APFO exposure. Cumulative Exposure. A model estimating cumulative APFO exposure requires that both time and intensity be considered. Biological monitoring data indicate a demarcation of exposure by job. Employees in the type ofjobs classified with definite exposure had median serum PFOA levels ranging from 2.6-5.2 ppm, while employees in jobs classified with probable exposure had levels ranging from 0.3-1.5 ppm (3). The biological half-life of PFOA is believed to be 3.8 years (10), thus short-term peak exposures may equate to longer tenn lower exposures over time. Considering these facts, the initial cumulative exposure assigned exposure weights of 1, 30, and 100 to the time in jobs with no exposure, probable exposure, and definite exposure. These unit less weighting factors, while somewhat arbitrary, were chosen to reflect the relative exposure intensity of jobs and long biological half-life of PFOA. For each worker the weighted exposure level was multiplied by the total days of employment at each level (weighted exposure level*days exposed), which provides a time-dependent exposure metric. The cumulative 10 exposure was categorized at levels representing the equivalent of up to one year (36,499 exposure-days), one to five years (36,500-182-499 exposure-days) and five or more years (1 82,500 exposure-days) of employment in a job with definite exposure. Because these weights are arbitrary we conducted a sensitivity analysis to explore how alternative weighting schemes may affect the results. The alternate weighting schemes were 1,10, 50 and 1, 10, 100, which would minimize migration of workers in jobs with probable exposure being classified with the higher exposed individuals. Detennination of Smokinu Status Smoking was a potentially confounding variable for some diseases of interest. To characterize the smoking habit of the cohort, the occupational medical records of the cohort members were abstracted for infonnation on smoking status; ever smoked regularly, year started smoking, years they smoked, and cigarettes smoked per day. Cohort members were classified by the smoking history and the availability of the records; smoking history available, medical record available but no information about smoking, and medical record not available. Determination of Salary versus Hourly Wage Type Baseline socio-economic status is a likely detenninant of mortality from several diseases. As a proxy measure of socio-economic status, the cohort members were classified by wage type based on the work history records. Workers were classified as hourly, salary, or both. The latter was designated if the job history included earning each type of wage for at least 365 days. A dichotomous version of this covariate was also established in which workers with both hourly and salary experience were classified as hourly or salary based on the predominant wage type. 11 Cause-Specific Mortalities of Interest A priori causes of death of interest were selected from toxicological literature and prior studies of workers exposed to fluorochemicals. Cancer of the liver, pancreas, and testicles and cirrhosis of the liver were selected based on results from toxicological studies (2). Deaths from prostate cancer and cerebrovascular disease (CVD) were previously reported to be associated with potential flurorochemical exposure in this cohort (8, 9, 11) and bladder cancer was associated with fluorochemical perflurooctanesulfonate (PFOS) exposure in another occupational cohort (11). Ischemic heart disease was included as an a priori disease of interest due to the effect of PFOA on lipid metabolism. ICD 9,h revision codes used to classify the underlying causes of death: prostate cancer, 185; pancreatic cancer, 157; liver cancer, 155; bladddr cancer, 188, 189.3-189.9; CVD, 430-438; ischemic heart disease, 410-414; and cirrhosis ofthe liver, 571. Congenital cerebral aneurysm (ICD9 747.81) were not identified in this cohort. Analysis The mortality experience of the cohort was initially compared to the mortality rates for the corresponding population of the state of Minnesota. Reference data were obtained from the Mortality Population Data System (MPDS) center at the University of Pittsburgh, which were derived from National Center for Health Statistics data. The all-cause mortality and malignant neoplasm rates were available from 1940, and the non-malignant cause-specific death rates from 1962. The referent data were available in age (5 year), sex, race, and calendar period (5 year) and were coded using the rules for the ICD revision in effect for the relevant calendar period. The Standardized Mortality Ratios (SMRs) and 95% confidence intervals were computed using the PC Life Table Analysis System (PCLTAS) software developed by the National Institute of Occupational Safety and Health (NIOSH) (12). State specific referent data were only available 12 through 1999 so the referent rates for 1999 were applied for the years 2000-2002. The all-cause and cause-speci fie SMRs w-ere first computed for the full cohort. SMRs were then computed separately for the cohort member ever employed in a job with definite APFO exposure, those who worked jobs with definite or probable exposure for at least one year, and for the cohort members who worked primarily in the non-chemical division. SMRs were also computed for workers by hourly and salary wage type. For this analysis, workers with both hourly and salary jobs were classified by the predominant wage type. To model the risk of the cause-specific mortalities of interest as a function of PFOA exposure using an internal referent population, hazard ratios and 95% confidence intervals were estimated using in multivariable time-dependent Cox regression models (13). Exposure was characterized by job classification and then cumulative exposure. The time variable was the number of days from the first day of employment at the production facility to an event (death) or the end of the study. In the cause-specific mortality models, all other causes of death were censored at the time of death. The models were adjusted for sex, age eligible to be in the study, year of birth, and wage type. To explore potential effects of latency the exposure models were lagged by 10 years prior to end of study or the date of death. The Cox regression analysis was conducted using SAS 9.1(14). Smoking is an important potential confounding variable in this study, but smoking habit data were unavailable for many of the cohort members. An initial Cox regression model was fit with smoking coded as ever smoked, never smoked, no information about smoking on medical record, and no medical record available. In a second approach a multiple imputation model was constructed using individuals with smoking data to predict the smoking status of those without 13 smoking data (15). The predictors used for the imputation process were sex, year of birth, year of first employment at the facility, age eligible to be in the study, and wage type. Eighteen independent, completed data sets were generated and a Cox regression procedure, as described above, was performed on each data set. The results were pooled to give single parameter estimates and 95% confidence intervals. The hazard ratios from the final models are presented to describe the potential effects of confounding by smoking. All imputation procedures were conducted using SAS 9.1(14). 14 Results Of the 6678 individual workers identified at the facility, 3993 employees met the inclusion criteria. Ofthese, 513 workers (12.8%) were ever employed in ajob with definite PFOA exposure, 1688 workers (42.3%) were ever employed in ajob with probable exposure (and never worked ajob with definite exposure), and 1792 workers (44.9%) were employed in jobs where they were never exposed to fluorochemicals (Table 1). The majority of the study cohort was male (80%), particularly for the definite exposure subgroup (92%). The average age at the end of follow-up was slightly younger in the definite exposure subgroup, but the average number of years of employment was longer. The total number of deaths in the cohort was 807, with 68 deaths in the definite, 368 in the probable, and 371 in the non-exposed exposure subgroups. Smoking data was found for 1430 (36%) cohort members, of these, 783 (55%) were found to have ever smoked (Table 2). There was a higher prevalence of smoking in those who ever worked ajob with definite APFO exposure compared to the non-exposed workers, 65% and 47% respectively. However, smoking data was available for 66% of the definite exposure subgroup (338/513), whereas it was only available for a 20% of the non-exposed subgroup (355/1792). The all-cause and cause-specific mortality ratios for the entire cohort, and exposure subgroups, were generally lower than expected compared to the Minnesota referent population (Table 3-6). The all-cause standardized mortality ratio (SMR) for the entire cohort was 0.8 (95% Cl = 0.7-0.9) (Table 3). The results were similar for all deaths from cancer (SMR = 0.9, 95% Cl = 0.8-1.0). The cause-specific mortality ratios for one or more years of definite or probable PFOA exposure were lower than expected for cerebrovascular disease (SMR=0.8, 95% 0=0.51.3) and ischemic heart disease (SMR=0.7, 95% 0=0.6-0.9). The SMRs for prostate cancer 15 (SMR = 1.4, 95% Cl = 0.7-2.4), pancreatic cancer (SMR = 1.2, 95% Cl = 0.5-2.4), bladder cancer (SMR = 1.3, 95% Cl = 0.3-3.9), and diabetes mellitus (SMR=1.7, 95% C = 0.9-2.8) were modestly, albeit imprecisely, above unity (Table 4). The cause-specific mortality ratios for cohort members ever employed in jobs with definite PFOA exposure were greater than expected for prostate cancer (SMR =2.1, 95% C3 = 0.4-6.1) and cerebrovascular disease (SMR =1.6, 95% Cl = 0.5-3.7), although the confidence intervals are quite wide and include the null (Table 5). Similar patterns are present when the analysis is restricted to one year of definite exposure (Table 6) though the number of deaths is small. The cause specific SMRs for all definite exposure strata all have confidence intervals indicating these elevations are not beyond chance. Cohort members who worked in jobs with probable exposure, but never held a job with definite exposure had an elevated risk of death from diabetes mellitus (SMR=2.0, 95% Cl =1.2-3.2), but low risk of death from ischemic heart disease (SMR=0.8, 95% 0=0.7-1.0) and cerebrovascular disease (SMR=0.7, 95% 0 = 0 4-1.1) (Table 7) In contrast, the number of deaths from prostate cancer and cerebrovascular disease were significantly lower among the never exposed members of the cohort; 0.4 (95% Cl 0.1 -0.9) and SMR 0.5 (95% Cl 0.3-0.8)respectively (Table 8). No deaths from testicular cancer were observed and only 3 cases of liver cancer (SMR=0.5, 95% C1=0.1-1.4), of which two held jobs with probable exposure. These small numbers precluded further analysis for these causes of death. Thirteen deaths from cirrhosis of the liver were identified (SMR= 0.7, 95% CT=0.4-1.2) with only 4 occurring in workers with at least one year of definite or probable exposure (SMR=0.5, 95% Cl=0.2-1.4). 16 The SMRs for salaried workers showed a decreased risk of death for all cancers (SMR 0.7, 95% Cl 0.6-0.8), respiratory cancers (SMR 0.6, 95% Cl 0.4-0.9), prostate cancer (SMR=0.5, 95% Cl 0.2-1.2), diabetes (SMR= 0.2, 95% Cl = 0.02-0.74,), cerebrovascular disease (SMR=0.6, 95% CI=0.4-1.0) and heart disease (SMR=0.6, 95% C1=0,5-0.7) (Table 9). The results were somewhat different for hourly employees; all cancers (SMR 1.0, 95% Cl 0.9-1.2), respiratory cancers (SMR 1.2, 95% Cl 0.9-1.6), prostate cancer (SMR=0.9, 95% Cl 0.4-1.6), cerebrovascular disease (SMR=0.7, 95% CI=0.4-1.0) and heart disease (SMR=0.9, 95% 0=0.8, 1.1) (Table 10). The SMR for diabetes (2.1, 95% Cl = 1.3-3.1) was elevated for the hourly workers. Comparing the mortality experience within the cohort in the time-dependent Cox regression models revealed differences for some of the outcomes of interest. A high or moderate exposure work history, compared to only working in low exposure jobs, was associated with an increased risk for cerebrovascular disease (HR=5.3, 95% Cl = 1.5- 19.6 and HR= 2.1, 95% Cl = 1.0-4.5, respectively) and prostate cancer (HR=6.9, 95% Cl = 1.2-41.8 and HR=3.0, 95% Cl = 0.8-11.0, respectively) (Table 11). A moderate exposure work history was also associated, imprecisely, with an elevated risk of dying from diabetes mellitus; HR=1.9, 95% Cl = 0.7-5.4, however no deaths from diabetes were reported in cohort members in the high exposure category. Diabetes mellitus was included in these analyses because of modestly elevated SMRs; specifically in the hourly workers. The inclusion of the imputed values for smoking status derived from the multiple imputation procedure, nor the inclusion of the actual smoking data, made little or no difference in the risks estimates (Table 11). The results for prostate cancer, cerebrovascular disease, ischemic heart disease and diabetes were further explored by stratifying by wage type (Table 12). These analyses suffered from small numbers for prostate cancer, 17 cerebrovascular disease, and diabetes. The risk associated with the high exposure category persisted for cerebrovascular disease. The high exposure prostate cancer deaths were salary workers while most of the moderate exposure prostate cancers were hourly employees. Overall, there was not strong evidence that the effects in the models adjusted for wage type were limited to either hourly or salary workers. Lagging exposures by 10 years also made little or no difference in the hazard ratio estimates (Table 13). Estimated hazard ratios (HR) comparing the highest to lowest weighted exposure category for cerebrovascular disease and prostate cancer were HR=2.4, 95% Cl = 1.1-5.4 and HR=3.6, 95% CI = 1.2-10.6, respectively (Table 14). There was no association between the metrics of weighted exposure and risk of pancreatic or bladder cancer, cirrhosis of the liver, disease and diabetes. The risk of dying from ischemic heart disease was, if anything, lower among those with increased exposure. As with the results for exposure by job classification, there was little evidence that the risks were different between hourly and salary workers (Table 15). The sensitivity analysis using alternate weighting schemes did not change the overall conclusions the analysis of cumulative exposure (Table 16). 18 Discussion This study evaluated the mortality experience of a population of workers at an ammonium pcrfluorooctanoate production facility, with specific attention to exposure to PFOA. No excess mortality was observed for malignant or non-malignant causes of death when compared to the corresponding Minnesota mortality rates. The SMRs for prostate cancer and cerebrovascular disease were slightly elevated for those with a history of working in jobs with definite PFOA exposure, while the SMRs for the members of the cohort never working in a PFOA exposed job were significantly below unity. Analyses using an internal referent category found increased risk of death from prostate cancer and cerebrovascular disease among workers with a history of greater PFOA exposure when compared to those working in jobs with no or minimal exposure. A priori causes of death selected based on animal toxicology studies, liver, pancreatic and testicular cancer and cirrhosis of the liver, were not observed to be associated with PFOA exposure. Interpreting these somewhat contradictory results requires caution and consideration of several assumptions. Some elements of the study design for this study differs from the earlier mortality analyses of this cohort (8, 9). Gilliland and Mandel (8) required six months of cumulative employment for inclusion, while the current study required one year of employment to exclude short-term workers, often summer interns, who might have different underlying risk factors than the long-term workers. In that analysis, job exposures were assigned as working for at least one month in the chemical division compared to working in the non-chemical division (or chemical division for less than one month). To minimize exposure misclassification the current study focused on PFOA and classified jobs in the chemical division as definite or probable, as only certain areas and tasks within the chemical division likely led to high APFO exposure. Lastly, 19 169 additional cohort members were included in the current study that, according to available employment data, were eligible for both studies. This study also differs from the more recent mortality analysis of this cohort (9). The population for the current study was followed through 2002, four years longer than the previous study, increasing the number of deaths from 607 to 807, and included analyses using an internal referent group. The association between prostate cancer and work in a APFO exposed job is similar to the results of Gilliland and Mandel (8) who reported a 3.3-fold increase (95% Cl = 1.0-10.6) in prostate cancer mortality associated with working ten years in the chemical division compared to non-chemical division workers; based on 6 cases. In the current study an association was observed between both metrics of PFOA exposure and prostate cancer when compared to the internal referent category. The biological mechanism for an association between PFOA and prostate cancer is not clear. An effect of PFOA on the endocrine system has been described in the rat which involves regulation of estradiol, testosterone, follicle stimulating hormone, luteinizing hormone and thyroid stimulating hormone through action in the liver (16-18). In this occupationally exposed population PFOA exposure was not clearly associated with changes in circulating levels of reproductive hormones (19), but this cross-sectional assessment was based on single blood samples. The association between APFO exposure and prostate cancer and cerebrovascular disease was most apparent when the internal referent population was used. An internal referent population from within a cohort may provide a more valid comparison, assuming similar social and demographic characteristics; however the inteipretation of these results should also consider how the strata specific prostate cancer deaths compare to the expected deaths based on 20 Minnesota general population. The SMR for the exposed categories were only modestly above unity, while the non-exposed members of the cohort were significantly below. The latter suggests that 3M employees who are not regularly exposed to APFO at work arc different from other men in Minnesota with respect to baseline prostate cancer risk; some of which may be related to socioeconomic status (SES). The extent to which this difference may have influenced the internal analysis is unknown. Our analyses adjusted for wage status as a proxy for socioeconomic status, which may influence several factors associated with prostate cancer death. The proportional hazards analysis stratified by wage type revealed similar patterns of risk with exposure in the hourly and salary workers, though the numbers were very small. Deaths from heart disease and cerebrovascular disease arc almost always below unity in epidemiologic studies of chemical workers (20) , for this reason the increased risk of cerebrovascular disease death associated with higher exposure was unexpected. It is possible that the coding of cerebrovascular disease deaths vary by region, however using mortality from local counties rather than the state made no differences in the earlier analysis (9). The risk of stroke is related to hypertension, diabetes, and life style factors including diet and smoking (21 23). The SMRs for diabetes mellitus and hypertension in the non-exposed subgroup were 0.52 and 0.98, respectively and 1.73 and 1.84, respectively, in the exposed subgroup. The SMR for lung cancer, a potential indicator of smoking in the cohort was 1.01 in the exposed compared to 0.76 in the non-exposed sub-cohort. Though not directly comparable, these patterns may be related to the cerbreovascular disease finding in that the lifestyle characteristics of the cohort members with high PFOA exposure were different than the lower exposed persons. Nevertheless, adjusting for smoking habit and wage type did not alter the association, and similar to prostate cancer, the analysis stratified by wage type indicated higher risk of death from cerebral vascular 21 disease associated with exposure in both hourly and salary workers. Diet is a potential factor in the risk of stroke, though it was not measured in this study. In the same working population, however, body mass index (BMI) of almost 50% of the workers ranged from 25-30 (24), which is considered overweight (25). However, it has also been shown that worker BMIs were similar among varying seoim PFOA levels (26). Any findings in a mortality study related to diabetes should be interpreted with caution as the use of mortality data to evaluate diabetes risk is problematic. In a study of 2,766 decedents with a known history of diabetes it was shown that diabetes was recorded as the underlying cause of death on approximately 10% of the death certificates, and those recorded were related by age, duration of diabetes, and co-morbidities (27). The excess in the hourly workers could indicate lifestyle differences, but future research would require a more comprehensive assessment of diabetes morbidity to fully describe any potential relationship with PFOA exposure. When interpreting this mortality analysis several limitations should be considered. Some exposure misclassification is likely. For example, maintenance and other mobile workers that routinely entered the definite PFOA exposure departments and work areas, but may have not been classified in the definite exposure subgroup. On the other hand, a few workers assigned to the definite exposure subgroup may not have spent much time in those departments or work areas. The extent of exposure misclassification and the effects on the study results remains unknown, as no additional data are available to verify these assumptions. Although several methods of approach were used to find all deaths in this cohort, it is possible that some deaths were not accounted for, which can occur when an individual leaves the country, changes their identity, or due to errors in surveillance systems. Race data were not available for the cohort. 22 The impact of this is likely to be limited as the population of Minnesota has been predominately Caucasian over the decades (97% in the 1980 census, 94% in the 1990 census, and 89% in the 2000 census) (28). The analysis was able to account for potential confounding by age, sex, wage status, and, to some extent, smoking habit. Unfortunately the smoking data were sparse, and though sophisticated methods to impute the missing data were applied, it is debatable whether the assumptions pertaining to this imputation were upheld. Another noteworthy limitation is the power of the study. The mean age at follow-up was 60 years, thus the relatively small number of deaths from the causes of interest limits the ability of the study to examine exposure responses, particularly using an internal reference category. By definition, mortality studies, although cost effective and convenient, miss the cases that do not result in death. Prostate cancer and cerebrovascular disease, the two findings of potential importance, do not always result in death and may not be listed as contributing causes of death on a death certificate unless they had recently been diagnosed or the person was undergoing treatment. While the mortality study does capture the worse case scenario for the diagnosis of these conditions, death from these conditions may be associated with other factors, including access to health care and availability of screening. The potential contribution of these factors are difficult to quantify, however all members of the cohort were, at least for 1 year, employed by 3M, and would have had access to similar health benefits. There are also notable strengths of this study. The cohort was constructed from the production facility employment records, thus allowing for complete enumeration of the cohort. The availability of the work history information in conjunction with detailed review of the 23 production history of PFOA at this plant with veteran workers and industrial hygienists helped reduce exposure uncertainty related to exposure misclassification. A final noteworthy strength to this study is the comprehensive follow-up of the cohort. An underlying cause of death was found for 99.6% of the known deaths (804/807), all unavailable or unfound death certificates were from cohort members who worked in the non-chemical division of the plant. The low SMRs in the non-exposed subgroup are suggestive of a bias contributing to the associations found in the internal analyses. There is also potential bias in the regression models with an internal referent group in that they were used under the assumption that all individuals in the model were the same by all risk factors except exposure. An attempt was made to account for the differences in lifestyle by the inclusion of wage type and smoking status in the models; however we were not able to fully account for the bias from smoking because the amount of available data was so small and the ability of the wage type to fully account for socio-economic differences is unknown. Interestingly, the ischemic heart disease mortality risk did not vary by exposure strata, as would be expected if significant differences in risk factors for heart disease existed between the exposure categories. To the extent that the underlying (unmeasured) risk factors for ischemic heart disease are similar to those for CVD and prostate cancer, potential confounding from these factors is of less concern. In summary, this study evaluated the mortality experience of an occupational cohort exposed to APFO and showed an association within the cohort between both death from cerebrovascular disease and prostate cancer and working in jobs with higher APFO exposure and an estimate of cumulative APFO exposure. No association was observed between APFO exposure and a priori diseases of interest based on toxicology studies. While findings have been 24 suggested in previous cohort studies; the current study had a longer follow-up period, therefore more deaths, increasing the power of the study. Additional follow-up of this cohort is recommended to clarify these associations, and if possible, should account for confounding factors such as BMI, smoking, access to health care, and other lifestyle factors. Studies of disease incidence that capture the cases that do not result in death would be informative, though these types of studies can be difficult to implement. Assessing the incidence of prostate cancer through cancer registries would be helpful in clarifying the association with prostate cancer. Future studies should also consider more sophisticated exposure models to properly assign etiologically relevant estimates of exposure. 25 References 1. Begley TH, White K, Honigfort P, Twaroski ML, Neches R, Walker RA. Perfluorochemicals: potential sources of and migration from food packaging. Food Additives & Contaminants 2005;22( 10): 1023-31. 2. Kennedy GL, Jr., Butenhoff JL, Olsen GW, O'Connor JC, Seacat AM, Perkins RG, Biegel LB, Murphy SR, Farrar DG. The toxicology of perfluorooctanoate. Critical Reviews in Toxicology 2004;34(4):351-84. 3. Olsen GW, Butenhoff JL, Mandel JH. Assessment of lipid, hepatic, and thyroid function in relation to an occupational biologic limit value for perfluorooctanoate. Saint Paul: 3M Company; 2003. Report No.: USEPA Public Docket AR-226-1351. 4. Olsen GW, Church TR, Miller JP, Burris JM, Hansen KJ, Lundberg JK, Armitage JB, Herron RM, Medhdizadehkashi Z, Nobiletti JB, O'Neill EM, Mandel JH, Zobel LR. Perfluorooctanesulfonate and other fluorochemicals in the serum of American Red Cross adult blood donors. Environmental Health Perspectives 2003; 111(16): 1892-901. 5. Olsen GW, Church TR, Larson EB, van Belle G, Lundberg JK, Hansen KJ, Bums JM, Mandel JH, Zobel LR. Serum concentrations of perfluorooctanesulfonate and other fluorochemicals in an elderly population from Seattle, Washington. Chemosphere 2004; 54( 11). 1599-611. 6. Olsen GW, Huang HY, Helzlsouer KJ, Hansen KJ, Butenhoff JL, Mandel JH. Historical comparison of perfluorooctanesulfonate, perfluorooctanoate, and other fluorochemicals in human blood. Environmental Health Perspectives 2005; 113(5):539-45. 7. Olsen GW, Mair DC, Reagen WK, Ellefson ME, Ehresman DJ, Butenhoff JL, Zobel LR. Preliminary evidence of a decline in perfluorooctanesulfonate (PFOS) and perfluorooctanoate (PFOA) concentrations in American Red Cross blood donors. Chemosphere 2007;68(1): 105-1 ]. 8. Gilliland FD, Mandel JS. Mortality among employees of a perfluorooctanoic acid production plant. J Occup Med. 1993;35(9):950-4. 9. Alexander BIT Mortality study of workers employed at the 3M Cottage Grove Facility. Minneapolis: University of Minnesota; 2001. Report No.: U.S. EPA Docket AR-2261030a018. 10. Olsen G, Ehresman D, Froehlich J, Burris J, Butenhoff J. Evaluation of the half-life of elimination of perfluorooctanesulfonate (PFOS), peril uorohexanesulfonate (PFHS) and periluorooctnaoate (PFOA) from human serum. FLUOROS 2005 [cited 2006 August 3]; Available from: http://chem.utoronto.ca/symposium/iluoros/pdfs/TOX017Qlsen.pdf 11. Alexander BH, Olsen GW, Bums JM, Mandel JH, Mandel JS. Mortality of employees of a perfluorooctanesulphonyl fluoride manufacturing facility. Occupational & Environmental Medicine 2003;60(10):722-9. 26 12. National Institute for Occupational Safety and Health. PC LTAS: Life table analysis system for use on the PC. In. Cincinnati: U.S. Department of Health and Human Services; 1998. 13. Breslow NE, Day NE. Statistical Methods in Cancer Research. Lyon: International Agency for Research on Cancer; 1987. 14. SAS Institute Inc. SAS 9.1 for Windows. In. Cary, NC USA; 2003. 15. Allison PD. Missing Data. Thousand Oaks: Sage Publications; 2001. 16. Biegel L, Liu R, Hurtt M, Cook J. Effects of ammonium perfluorooctanoate on Leydig cell function: in vitro, in vivo, and ex vivo studies. Toxicol Appl Pharm 1995; 134:1 8-25. 17. Liu RC, Hurtt ME, Cook JC, Biegel LB. Effect of the peroxisome proliferator, ammonium perfluorooctanoate (C8), on hepatic aromatase activity in adult male CrkCD BR (CD) rats. Fund Appl Toxicol. 1996;30(2):220-8. 18. Biegel LB, Hurtt ME, Frame SR, O'Connor JC, Cook JC. Mechanisms of extrahepatic tumor induction by peroxisome proliferators in male CD rats. Toxicol Sci 2001 ;60:40-55. 19. Olsen GW, Gilliland FD, Burlew MM, Burris JM, Mandel JS, Mandel JH. An epidemiologic investigation of reproductive hormones in men with occupational exposure to perfluorooctanoic acid. J Occup Environ Med 1998;40(7):614-22. 20. Greenberg RS, Mandel JS, Pastides H, Britton NL, Rudenko L, Starr TB. Chemical workers in the U.S. and Western Europe: a meta-analysis of cohort studies describing mortality and cancer incidence. Epidemiology 2001;12:727-740. 21. Gorelick PB, Sacco RL, Smith DB, Alberts M, Mustone-Alexander L, Rader D, Ross JL, Raps E, Ozer MN, Brass LM, Malone ME, Goldberg S, Booss J, Hanley DF, Toole JF, Greengold NL, Rhew DC. Prevention of a first stroke: a review of guidelines and a multidisciplinary consensus statement from the National Stroke Association. Jama. 1999;281(12): 1112-20. 22. Kuller LH. Epidemiology and prevention of stroke, now and in the future. Epidemiol Rev. 2000;22(l):14-7. 23. Desai J, Devlin H. Diabetes in Minnesota. St Paul: Minnesota Department of Health; 2002. 24. Gilliland FD, Mandel JS. Serum perfluorooctanoic acid and hepatic enzymes, lipoproteins and cholesterol: a study of occupationally exposed men. Am J Ind Med 1996;29:560-568. 25. CDC. BMI: Body Mass Index. Atlanta: Centers for Disease Control and Prevention; 2006. 27 26. Olsen GW, Burris JM, Burlew MM, Mandel JH. Plasma cholecystokinin and hepatic enzymes, cholesterol and lipoproteins in ammonium perfluorooctanoate production workers. Drug & Chemical Toxicology 2000;23(4):603-20. 27. Bild DE, Stevenson JM. Frequency of recording of diabetes on U.S. death certificates: analysis of the 1986 National Mortality Followback Survey. Journal o f Clinical Epidemiology>1992;45(3):275-81. 28. US Census Bureau. State and County Quick Facts; 2006. 28 Table 1. Characteristics o f APFO Manufacturing Cohort by Job Exposure Subgroups Total Definite APFO Probable APFO exposure3 exposureb Non-exposed': 513 1688 1792 Total 3993 Gender Male 473 (92%) 1389 (82%) 1323 (74%) 3184 (80%) Female 40 (8%) 298 (18%) 470 (26%) 809 (20%) Age at follow-up (Mean) 55.6 60.0 60.6 59.6 Person years (Mean) 29.3 31.6 31.6 31.3 Year of birth (Mean) 1945 1938 1938 1939 Years of employment (Mean) 17.8 16.4 9.7 13.8 Age at death (Mean) 60.1 65.6 64.9 64.8 Deaths 68 368 371 807 a: Ever employed in job with definite APFO exposure b: Ever employment in a job with probable APFO exposure, but never in a job with definite exposure c: Never held job with definite or probable exposure; primarily non-chemical division 29 Table 2. Characteristics of APFO Manufacturing Cohort Smoking Status by Job Exposure Subgroups Ever smoke cigarettes Yes No MRdreviewed, no data found No MR found Number of packs smoked per day <1 pack per day 1 pack per day >-2 packs per day MRdreviewed, no data found Year began smoking 1928-39 1940-49 1950-59 1960-69 1970-79 1980-89 1990-99 MRdreviewed, no data found Definite APFO exposure3 (n=513) N% 220 42.7% 118 23.0% 92 17.8% 83 16.4% 68 17.7% 84 21.3% 23 5.9% 45 55.1% 2 0.9% 14 6.4% 35 16.4% 40 17.8% 22 10.0% 8 3.7% 1 0.5% 98 44.3% Probable APFO exposure1' 0i=1688) N% 396 23.5% 341 20.2% 414 24.6% 537 31.7% 127 32.0% 125 31.5% 34 8.6% 110 28.0% 10 2.5% 45 11.3% 59 14.6% 48 12.1% 27 6.8% 9 2.3% 0 0.0% 198 50.4% Non-exposedc (n=l792) N 07/0 167 9.3% 188 10.5% 368 20.5% 1069 59.6% 49 28.7% 54 32.9% 12 7.2% 52 31.1% 4 2.4% 5 3.0% 15 9.0% 14 9.0% 9 5.4% 7 4.2% 3 1.8% 110 65.3% Total (n=3993) N% 783 19.6% 647 16.2% 874 21.9% 1689 42.3% 244 31.2% 263 33.6% 69 8.8% 207 26.4% 16 2.0% 64 8.2% 109 13.9% 102 13.0% 58 7.4% 24 3.1% 4 0.5% 406 51.9% 30 Table 2. (Continued) Number of years smoked 1-5 6-10 11-20 21-30 31-40 41-50 51+ MRdreviewed, no data found Definite APFO exposure3 (n=513) N% 13 5.9% 24 11.0% 82 37.0% 32 14.6% 19 9.1% 3 1.4% 0 0.0% 47 21.0% Probable APFO exposureb (n=1688) N% 30 7.6% 29 7.3% 115 29.0% 56 14.1% 49 12.1% 12 3.0% 1 0.3% 104 26.7% b: Ever employment in a job with probable APFO exposure, but never in a job with definite exposure c: Never held job with definite or probable exposure; primarily non-chemical division d: MR = Medical record Non-exposedc (n=1792) N% 11 6.6% 15 9.0% 15 9.6% 18 10.8% 10 6.0% 3 1.8% 0 0.0% 95 56.3% Total (n=3993) N% 54 6.9% 68 8.7% 212 27.1% 106 13.5% 78 10.0% 18 2.3% 1 0.1% 246 31.4% 31 Table 3. Standardized Mortality Ratios (SMRs) for Selected Cause-Specific Mortalities for the Entire APFO Manufacturing Cohort CauseJ All deaths Cancers All cancers Buccal cavity and pharynx Digestive organs and peritoneum Esophagus Stomach Large Intestine Rectum Biliary passages and liver primary Pancreas Other Respiratory System Larynx Trachea, bronchus, and lung Other BreasT Female genital organs Cervix Other Male genital organs Prostate Other Urinary Organs Kidney Bladder and other urinary organs Other and unspecified sites Skin Central nervous system Thyroid gland Lymphatic and hematopoietic tissue Lymphosarcoma and reticulosarcoma Hodgkin's disease Leukemia and aleukemia Other All other cancers Benign neoplasms Observed 807 246 2 61 4 7 28 4 3 13 2 79 3 75 1 6 5 2 3 17 16 1 11 4 7 12 4 7 1 29 3 1 12 13 24 3 Expected 1000.18 286.28 5.54 69.94 7.30 8.49 25.48 5.23 6.36 15.08 1.99 86.42 2.48 83.02 0.92 11.28 6.74 0.90 3.70 23.16 22.20 0.96 14.20 8.53 5.67 15.94 4.18 9.84 0.99 32.75 2.50 2.00 12.46 15.80 20.30 3.11 SMR1' 0.81 95% Cl 0.75-0.86 0.86 0.36 0.86 0.55 0.82 1.10 0.76 0.47 0.86 1.00 0.91 1.21 0.90 1.08 0.53 0.74 2.21 0.81 0.73 0.72 1.04 0.77 0.47 1.23 0.75 0.96 0.71 1.01 0.89 1.20 0.50 0.96 0.82 1.18 0.96 0.76-0.97 0.04-1.30 0.67-1.12 0.15-1.40 0.33-1.70 0.73-1.59 0.21-1.96 0.10-1.38 0.46-1.47 0.12-3.63 0.72-1.14 0.25-3.54 0.71-1.13 0.03-6.01 0.19-1.16 0.24-1.73 0.27-7.99 0.17-2.37 0.43-1.18 0.41-1.17 0.03-5.77 0.39-1.39 0.13-1.20 0.49-2.54 0.39-1.31 0.26-2.45 0.29-1.47 0.03-5.64 0.59-1.27 0.25-3.52 0.01-2.78 0.50-1.68 0.44-1.41 0.76-1.76 0.20-2.82 32 Table 3. (Continued) Cause-1 Observed Non-malignant causes Tuberculosis Diabetes mellitus Cerebrovascular disease 1 23 35 All heart disease 256 Rheumatic heart disease Ischemic heart disease Chronic disease of endocardium 6 201 11 Hypertension with heart disease 5 Other 33 Hypertension without heart disease 6 Diseases of respiratory system 50 Influenza and pneumonia 12 Bronchitis 1 Emphysema 3 Asthma 3 Other 31 Ulcer of stomach 1 Cirrhosis of the liver 13 Nephritis and nephrosis 7 Accidents 49 Motor vehicle accidents 27 Other 22 Violence 22 Suicides 17 Homicides and other 5 All other causes a: Cause not listed if not observed b: Reference rates from state of Minnesota c: All breast cancers observed in female employees 95 Expected SMRb 95% Cl 0.49 19.97 54.39 329.61 4.39 259.23 15.63 4.58 45.78 4.20 73.91 22.53 1.49 8.18 2.11 39.61 3.02 17.87 6.39 55.84 25.74 30.10 26.75 22.02 4.74 114.61 2.05 1.15 0.64 0.78 1.37 0.78 0.70 1.09 0.72 1.43 0.68 0.53 0.67 0.37 1.42 0.78 0.33 0.73 1.10 0.88 1.05 0.73 0.82 0.77 1.06 0.83 0.05-11.40 0.73-1.73 0.45-0.90 0.68-0.88 0.50-2.97 0.67-0.89 0.35-1.26 0.35-2.55 0.50-1.01 0.52-3.11 0.50-0.89 0.28-0.89 0.02-3.74 0.08-1.07 0.29-4.15 0.53-1.11 0.01-1.84 0.39-1.24 0.44-2.26 0.65-1.16 0.69-1.53 0.46-1.11 0.52-1.25 0.45-1.24 0.34-2.47 0.67-1.01 33 Table 4. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for APFO Manufacturing Cohort Members with a Minimum o f One Year o f Employment in Jobs with Definite or Probable APFO Exposure3 Cause1' All deaths Cancers All cancers Digestive organs and peritoneum Esophagus Stomach Large Intestine Rectum Biliary passages and liver Pancreas Respiratory System Larynx Trachea, bronchus, and lung Female genital organs Male genital organs Prostate Urinary Organs Kidney Bladder and other urinary organs Other and unspecified sites Skin Central nervous system Lymphatic and hematopoietic tissue Lymphosarcoma and reticulosarcoma Leukemia and aleukemia All other cancers Benign neoplasms Observed Expected SMRC 95% Cl 309 398.02 0.78 0.69-0.87 109 112.58 0.97 0.80-1.17 30 27.81 1.08 0.73-1.54 2 3.05 0.66 0.08-2.37 4 3.39 1.18 0.32-3.02 12 10.00 1.20 0.62-2.10 3 2.09 1.43 0.30-4.19 2 2.51 0.80 0.10-2.88 7 5.99 1.17 0.47-2.41 34 35.00 0.97 0.67-1.36 2 1.02 1.95 0.24-7.06 32 33.60 0.95 0.65-1.34 3 1.72 1.75 0.36-5.10 13 9.33 1.39 0.74-2.38 12 8.90 1.35 0.70-2.35 4 5.75 0.70 0.19-1.78 1 3.50 0.29 0.01-1.59 3 2.25 1.33 0.28-3.90 7 6.56 1.07 0.43-2.20 2 1.75 1.15 0.14-4.14 5 4.04 1.24 0.40-2.90 11 13.12 0.84 0.42-1.50 1 0.97 1.03 0.03-5.71 6 5.02 1.19 0.44-2.60 7 8.08 0.87 0.35-1.78 2 1.22 1.64 0.20-5.91 Non-malignant causes Diabetes mellitus Cerebrovascular disease 13 7.87 1.65 0.88-2.82 17 20.62 0.82 0.48-1.32 34 Table 4. (Continued) Cause1' Observed Expected SMRC 95% Cl All heart disease 87 131.57 0.66 0.63-0.82 Rheumatic heart disease 1 1.64 0.61 0.02-3.38 Ischemic heart disease 74 103.94 0.71 0.56-0.89 Chronic disease of endocardium 3 5.95 0.50 0.10-1.47 Hypertension with heart disease 3 1.79 1.67 0.34-4.89 Hypertension without heart disease 3 1.63 1.84 0.38-5.38 Diseases of respiratory system 13 28.49 0.46 0.24-0.78 Influenza and pneumonia 4 8.49 0.47 0.13-1.21 Emphysema 1 3.19 0.31 0.01-1.74 Asthma 1 0.80 1.25 0.03-6.94 Ulcer of stomach 1 1.17 0.85 0.02-4.73 Cirrhosis of the liver 4 7.35 0.54 0.15-1.39 Nephritis and nephrosis 4 2.44 1.64 0.45-4.19 Accidents 19 24.20 0.79 0.47-1.23 Motor vehicle accidents 11 11.37 0.97 0.48-1.73 All other accidents 8 12.83 0.62 0.27-1.23 Violence 8 11.86 0.67 0.29-1.33 Suicides 8 9.76 0.82 0.35-1.61 All other causes 29 44.91 0.65 0.43-0.93 a: Includes workers who accrued one year of exposure with definite and probable jobs combined b: Cause not listed if not observed c: Reference rates from state of Minnesota 35 Table 5. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for APFO Manufacturing Cohort Members Ever Employed In Jobs with Definite APFO Exposure Cause"1 All deaths Cancers All cancers Digestive organs and peritoneum Esophagus Large Intestine Pancreas Respiratory System Larynx Trachea, bronchus, and lung Male genital organs Prostate Lymphatic and hematopoietic tissue Leukemia and aleukemia All other cancers Observed Expected SMR1' 95% Cl 68 77.31 0.88 0.68-1.12 19 21.95 0.87 0.52-1.35 4 5.36 0.75 0.20-1.91 1 0.65 1.54 0.04-8.57 2 1.87 1.07 0.13-3.86 1 1.17 0.85 0.02-4.74 9 7.11 1.27 0.58-2.40 1 0.21 4.72 0.12-26.23 8 6.81 1.17 0.51-2.31 3 1.55 1.93 0.40-5.65 3 1.43 2.10 0.43-6.13 1 2.68 0.37 0.01-2.08 1 1.04 0.96 0.02-5.34 2 1.62 1.23 0.15-4.45 Non-malignant causes Cerebrovascular disease All heart disease Ischemic heart disease Hypertension with heart disease Other Hypertension without heart disease Diseases of respiratory system Asthma Other Nephritis and nephrosis Accidents Motor vehicle accidents Other Violence Suicides Homicides and other All other causes a: Cause not listed if not observed b; Reference rates from state of Minnesota 5 3.14 1.59 0.52-3.72 21 24.01 0.87 0.54-1.34 16 19.13 0.84 0.48-1.36 1 0.31 3.27 0.08-18.17 4 3.48 1.15 0.31-2.94 1 0.27 3.73 0.09-20.71 3 4.50 0.67 0.14-1.95 1 0.14 7.32 0.19-40.68 2 2.54 0.79 0.10-2.84 2 0.38 5.22 0.63-18.85 8 6.91 1.16 0.50-2.28 4 3.57 1.12 0.31-2.87 4 3.34 1.2 0.33-3.06 6 3.64 1.65 0.60-3.59 5 2.96 1.69 0.55-3.94 1 0.67 1.49 0.04-8.26 3 8.07 0.37 0.08-1.09 36 Table 6. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for APFO Manufacturing Cohort M embers with a Minimum of One Year o f Employment in a Job with Definite APFO Exposure Cause3 All deaths Cancers All cancers Digestive organs and peritoneum Esophagus Large Intestine Respiratory System Trachea, bronchus, and lung Male genital organs Prostate All other cancers Observed 25 Expected 35.95 SMRb 0.70 95% Cl 0.45-1.03 7 10.29 0.68 0.27-1.40 2 2.54 0.79 0.10-2.85 1 0.31 3.28 0.08-18.21 1 0.89 1.12 0.03-6.24 2 3.36 0.59 0.07-2.15 2 3.23 0.62 0.08-2.24 2 0.80 2.51 0.30-9.04 2 0.75 2.67 0.32-9.65 1 0.76 1.32 0.03-7.32 Non-malignant causes Cerebrovascular disease All heart disease Ischemic heart disease Other Diseases of respiratory system Other Nephritis and nephrosis Accidents Other Violence Suicides All other causes a: Cause not listed if not observed b: Reference rates from state of Minnesota 3 1.55 1.94 0.40-5.66 8 11.51 0.70 0.30-1.37 6 9.16 0.66 0.24-1.43 2 1.64 1.22 0.15-4.40 2 2.27 0.88 0.11-3.18 2 1.29 1.55 0.19-5.61 1 0.19 5.23 0.13-29.08 1 2.77 0.36 0.01-2.01 1 1.39 0.72 0.02-3.99 2 1.43 1.39 0.17-5.03 2 1.18 1.70 0.21-6.13 1 3.83 0.26 0.01-1.45 37 Table 7. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for APFO Manufacturing Cohort Members Ever Employed In Jobs with Probable APFO Exposure, but Did Not Hold Jobs with Definite APFO Exposure Cause'1 All deaths Cancers All cancers Buccal cavity and pharynx Digestive organs and peritoneum Esophagus Stomach Large Intestine Rectum Biliary passages and liver primary Pancreas Respi ratory System Larynx Trachea, bronchus, and lung Breast1"' Female genital organs Other Male genital organs Prostate Other Urinary Organs Kidney Bladder and other urinary organs Other and unspecified sites Skin Central nervous system Lymphatic and hematopoietic tissue Lymphosarcoma and reticulosarcoma Leukemia and aleukemia Other Observed 368 Expected 443.52 SMRb 0.83 95% Cl 0.75-0.92 119 126.63 0.94 0.78-1.12 1 2.47 0.40 0.01-2.25 27 31.19 0.87 0.57-1.26 1 3.26 0.31 0.01-1.70 4 3.78 1.06 0.29-2.71 10 11.35 0.88 0.42-1.62 3 2.33 1.28 0.26-3.76 2 2.84 0.71 0.09-2.55 7 6.74 1.04 0.42-2.14 38 38.37 0.99 0.70-1.36 1 1.10 0.91 0.02-5.03 37 36.86 1.00 0.71-1.38 2 4.72 0.42 0.05-1.53 4 2.89 1.38 0.38-3.54 3 1.59 1.89 0.39-5.52 10 10.12 0.99 0.47-1.82 9 9.69 0.93 0.42-1.76 1 0.43 2.33 0.06-12.96 5 6.30 0.79 0.26-1.85 2 3.80 0.53 0.06-1.90 3 2.50 1.20 0.25-3.50 7 7.01 1.00 0.40-2.06 2 1.83 1.09 0.13-3.95 5 4.32 1.16 0.37-2.70 14 14.55 0.96 0.53-1.61 2 1.11 1.80 0.22-6.51 7 5.52 1.27 0.51-2.61 5 7.04 0.71 0.23-1.66 All other cancers Benign neoplasms 11 9.01 1.22 0.61-2.18 2 1.39 1.44 0.17-5.19 38 Table 7. (Continued) Cause" Observed Noil-malignant causes Diabetes mellitus 18 Cerebrovascular disease 17 All heart disease 110 Rheumatic heart disease 2 Ischemic heart disease 93 Chronic disease of endocardium 7 Hypertension with heart disease 2 Other 6 Hypertension without heart disease 3 Diseases of respiratory system 19 Influenza and pneumonia 7 Bronchitis 1 Emphysema 2 Other 9 Ulcer of stomach 1 Cirrhosis of the liver 6 Nephritis and nephrosis 2 Accidents 23 Motor vehicle accidents 12 Other 11 Violence 8 Suicides 8 a: Cause not listed if not observed b: Reference rates from state of Minnesota c: All breast cancers observed in female employees Expected 8.87 24.15 146.86 1.96 115.52 7.00 2.07 20.31 1.90 32.70 9.82 0.66 3.64 17.65 1.35 7.94 2.82 24.56 11.29 13.27 11.60 9.56 SMRb 2.03 0.70 0.75 1.02 0.81 1.00 0.97 0.30 1.58 0.58 0.71 1.53 0.55 0.51 0.74 0.76 0.71 0.94 1.06 0.83 0.69 0.84 95% Cl 1.20-3.21 0.41-1.13 0.62-0.90 0.12-3.68 0.65 0.99 0.40-2.06 0.12-3.49 0.11-0.64 0.33-4.62 0.35-0.91 0.29-1.47 0.04-8.47 0.07-1.99 0.23-0.97 0.02-4.13 0.28-1.65 0.09-2.56 0.59-1.41 0.55-1.86 0.41-1.48 0.30-1.36 0.36-1.65 39 Table 8. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for APFO Manufacturing Cohort Members Never Employed in Jobs with APFO Exposure Cause" All deaths Cancers All cancers Buccal cavity and pharynx Digestive organs and peritoneum Esophagus Stomach Large Intestine Rectum Biliary passages and liver Pancreas Respiratory System Larynx Trachea, bronchus, and lung Other Breast0 Female genital organs Cervix Male genital organs Prostate Urinary Organs Kidney Bladder and other urinary organs Skin Central nervous system Thyroid gland Lymphatic and hematopoietic tissue Lymphosarcoma and reticulosarcoma Hodgkin's disease Leukemia and aleukemia All other cancers Benign neoplasms Observed Expected SMRb 371 479.34 0.77 95% Cl 0.70-0.86 108 137.70 0.78 0.64-0.95 1 2.59 0.39 0.01-2.14 30 33.39 0.90 0.61-1.28 2 3.39 0.59 0.07-2.13 3 4.08 0.74 0.15-2.15 16 12.26 1.30 0.75-2.12 1 2.50 0.40 0.01-2.22 1 3.04 0.33 0.01-1.83 5 7.17 0.70 0.23-1.63 32 40.94 0.78 0.53-1.10 1 1.16 0.86' 0.02-4.79 30 39.35 0.76 0.51-1.09 1 0.44 2.30 0.06-12.76 4 6.27 0.64 0.17-1.63 1 3.70 0.27 0.01-1.50 1 0.51 1.96 0.05-10.92 4 11.49 0.35 0.09-0.89 4 11.08 0.36 0.10-0.92 6 6.77 0.89 0.32-1.93 2 3.98 0.50 0.06-1.81 4 2.78 1.44 0.39-3.67 2 1.91 1.05 0.13-3.79 2 4.55 0.44 0.05-1.59 1 0.46 2.16 0.05-12.00 14 15.52 0.90 0.49-1.51 1 1.20 0.84 0.02-4.65 1 0.92 1.09 0.03-6.04 4 5.90 0.68 0.18-1.73 11 9.67 1.14 0.57-2.04 1 1.49 0.67 0.02-3.73 40 Table 8. (Continued) Cause3 Non-malignant causes Tuberculosis Diabetes mellitus Cerebrovascular disease All heart disease Rheumatic heart disease Ischemic heart disease Chronic disease of endocardium Hypertension with heart disease Hypertension without heart disease Diseases of respiratory system Influenza and pneumonia Emphysema Asthma Cirrhosis of the liver Nephritis and nephrosis Accidents Motor vehicle accidents All other accidents Violence Suicides Homicides and other All other causes a: Cause not listed if not observed b: Reference rates from state of Minnesota c: All breast cancers observed in female employees Observed 1 5 13 125 4 92 4 2 2 28 5 1 2 7 3 18 11 7 8 4 4 52 Expected 0.23 9.55 27.10 158.74 2.14 124.58 7.82 2.21 2.03 36.71 11.50 4.01 1.04 8.27 3.19 24.38 10.89 13.49 11.52 9.49 2.02 55.57 SMRb 4.28 0.52 0.48 0.79 1.86 0.74 0.51 0.90 0.98 0.76 0.43 0.25 1.93 0.85 0.94 0.74 1.01 0.52 0.69 0.42 1.98 0.94 95% Cl 0.11-23.80 0.17-1.22 0.26-0.82 0.66-0.94 0.51-4.77 0.60-0.91 0.14-1.31 0.11-3.27 0.12-3.55 0.51-1.10 0.14-1.02 0.01-1.38 0.23-6.97 0.34-1.75 0.19-2.75 0.44-1.17 0.50-1.81 0.21-1.07 0.30-1.37 0.11-1.08 0.54-5.05 0.70-1.23 41 Table 9. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for Salaried Workers of the APFO Manufacturing Cohort Cause All deaths Cancers All Cancers Buccal cavity and pharynx Digestive organs and peritoneum Esophagus Stomach Large intestine Rectum Biliary passages and liver Pancreas All other digestive Respiratory system Larynx Bronchus, trachea, and lung All other respiratory Breast Female genital organs All uterine (non-cervix) Cervix Other female genital organs Male genital organs Prostate Testis and other male genital Urinary organs Kidney Bladder and other urinary organs Other and unspecified sites Skin Eye Central nervous system Thyroid gland/other endocrine Bone Deaths 307 Expected 484.92 SMR 0.63 95% Cl 0.56-0.71 97 140.47 0.69 0.56-0.84 0 2.68 0.00 0.00-1.38 27 33.77 0.80 0.53-1.16 2 3.56 0.56 0.07-2.03 2 4.06 0.49 0.06-1.78 17 12.33 1.38 0.80-2.21 2 2.50 0.80 0.10-2.89 0 3.09 0.00 0.00-1.19 4 7.29 0.55 0.15-1.40 0 0.95 0.00 0.00-3.87 27 42.53 0.63 0.42-0.92 1 1.20 0.83 0.02-4.62 25 40.87 0.61 0.40-0.90 1 0.45 2.20 0.06-12.22 2 5.90 0.34 0.04-1.22 3 3.33 0.90 0.19-2.64 0 1.03 0.00 0.00-3.59 2 0.47 4.29 0.52-15.49 1 1.84 0.54 0.01-3.03 6 11.47 0.52 0.19-1.14 6 11.01 0.54 0.20-1.19 0 0.46 0.00 0.00-8.04 5 6.94 0.72 0.23-1.68 2 4.18 0.48 0.06-1.73 3 2.76 1.09 0.22-3.18 6 7.92 0.76 0.28-1.65 2 2.10 0.95 0.12-3.43 0 0.09 0.00 0.00-42.92 3 4.88 0.61 0.13-1.80 1 0.48 2.07 0.05-11.50 0 0.37 0.00 0.00-9.98 42 Table 9. (Continued) Cause Lymphatic and hematopoietic Lymphosarcoma and reticulosarcoma Hodgkin's disease Leukemia and aleukemia Other lymphatic hematologic All other malignant neoplasms Benign neoplasm Non-Malignant Causes Diabetes mellitus Cerebrovascular disease All heart disease Rheumatic heart disease Ischemic heart disease Chronic disease of endocardium Hypertension with heart disease All other heart disease Hypertension without heart disease Nonmalignant respiratory disease Influenza and pneumonia Bronchitis Emphysema Asthma Other nonmalignant respiratory Cirrhosis of liver Nephritis and nephrosis Nephritis and nephrosis Accidents Motor vehicle accidents All other accidents Violence Suicides Homicides All other causes a: Cause not listed if not observed b: Reference rates from state of Minnesota c: All breast cancers observed in female employees Deaths 13 2 1 2 8 8 2 Expected 15.97 1.16 0.97 6.10 7.74 9.96 1.51 SMR 0.81 1.72 1.04 0.33 1.03 0.80 1.33 95% Cl 0.43-1.39 0.21-6.21 0.03-5.75 0.04-1.18 0.45-2.04 0.35-1.58 0.16-4.79 2 9.75 0.21 0.02-0.74 16 25.95 0.62 0.35-1.00 96 157.89 0.61 0.49-0.74 2 2.03 0.99 0.12-3.56 73 123.83 0.59 0.46-0.74 4 7.52 0.53 0.14-1.36 0 2.16 0.00 0.00-1.71 17 22.35 0.76 0.44-1.22 2 2.00 1.00 0.12-3.61 22 35.92 0.61 0.38-0.93 5 11.00 0.45 0.15-1.06 0 0.71 0.00 0.00-5.21 1 3.91 0.26 0.01-1.42 2 1.03 1.94 0.23-7.00 14 19.27 0.73 0.40-1.22 5 8.68 0.58 0.19-1.35 2 3.11 0.64 0.08-2.32 2 3.11 0.64 0.08-2.32 14 27.20 0.51 * 0.28-0.86 8 12.59 0.64 0.27-1.25 6 14.61 0.41 0.15-0.89 7 13.32 0.53 0.21-1.08 5 10.93 0.46 0.15-1.07 2 2.39 0.84 0.10-3.02 42 55.60 0.76 0.54-1.02 43 Table 10. Standardized Mortality Ratios for Selected Cause-Specific Mortalities for Hourly Workers of the APFO Manufacturing Cohort Cause All Deaths Cancers All Cancers Buccal cavity and pharynx Digestive organs and peritoneum Esophagus Stomach Large intestine Rectum Biliary passages and liver Pancreas All other digestive Respiratory system Larynx Bronchus, trachea, and lung All other respiratory Breast Female genital organs All uterine (non-cervix) Cervix Other female genital organs Male genital organs Prostate Testis and other male genital organs Urinary organs Kidney Bladder and other urinary organs Other and unspecified sites Skin Eye Central nervous system Thyroid/other endocrine gland Bone Lymphatic and hematopoietic Lymphosarcoma and reticulosarcoma Hodgkin's disease Leukemia and aleukemia Other lymphatic hematologic Observed 500 149 2 34 2 5 11 2 3 9 2 52 2 50 0 4 2 0 0 2 11 10 1 6 2 4 6 2 0 4 0 0 16 1 0 10 5 Expected 515.26 145.81 2.86 36.17 3.74 4.43 13.15 2.74 3.27 7.79 1.04 43.89 1.27 42.15 0.47 5.38 3.41 1.10 0.44 1.87 11.69 11.18 0.50 7.27 4.35 2.91 8.02 2.07 0.10 4.96 0.50 0.39 16.78 1.33 1.03 6.37 8.05 SMR 0.97 1.02 0.70 0.94 0.53 1.13 0.84 0.73 0.92 1.16 1.93 1.18 1.57 1.19 0.00 0.74 0.59 0.00 0.00 1.07 0.94 0.89 1.99 0.83 0.46 1.37 0.75 0.97 0.00 0.81 0.00 0.00 0.95 0.75 0.00 1.57 0.62 95% Cl 0.89-1.06 0.86-1.20 0.08-2.53 0.65-1.31 0.06-1.93 0.37-2.64 0.42-1.50 0.09-2.64 0.19-2.68 0.53-2.19 0.23-6.95 0.88-1.55 0.19-5.66 0.88-1.56 0.00-7.87 0.20-1.90 0.07-2.12 0.00-3.35 0.00-8.43 0.13-3.87 0.47-1.68 0.43-1.64 0.05-11.03 0.30-1.80 0.06-1.66 0.37-3.51 0.27-1.63 0.12-3.48 0.00-37.85 0.22-2.06 0.00-7.34 0.00-9.48 0.54-1.55 0.02-4.17 0.00-3.59 0.75-2.89 0.20-1.45 44 Tabic 10. (Continued) Cause All other malignant neoplasms Benign neoplasm Non-Malignant Causes Diabetes mellitus Cerebrovascular disease All heart disease Rheumatic heart disease Ischemic heart disease Chronic disease of endocardium Hypertension with heart disease AM other heart disease Hypertension without heart disease Nonmalignant respiratory disease Influenza and pneumonia Bronchitis Emphysema Asthma Other nonmalignant respiratory Ulcer of stomach and dueodenum Cirrhosis of liver Nephritis and nephrosis Accidents Motor vehicle accidents All other accidents Violence Suicides Homicides All other causes a: Cause nol listed if not observed b: Reference rates from state of Minnesota c: All breast cancers observed in female employees Observed Expected 16 10.34 1 1.60 SMR 1.55 0.62 95% Cl 0.88-2.51 0.02-3.46 21 10.22 2.06 1.27-3.14 19 28.43 0.67 0.40-1.04 160 171.72 0.93 0.79-1.09 4 2.36 1.69 0.46-4.33 128 135.40 0.95 0.79-1.12 7 8.11 0.86 0.35-1.78 5 2.42 2.06 0.67-4.82 16 23.43 0.68 0.39-1.11 4 2.20 1.82 0.50-4.65 28 38.00 0.74 0.49-1.07 7 11.53 0.61 0.24-1.25 1 0.78 1.28 0.03-7.13 2 4.26 0.47 0.06-1.69 1 1.08 0.92 0.02-5.14 17 20.34 0.84 0.49-1.34 1 1.59 0.63 0.02-3.49 8 9.19 0.87 0.37-1.72 5 3.28 1.52 0.49-3.56 35 28.64 1.22 0.85-1.70 19 13.15 1.45 0.87-2.26 16 15.49 1.03 0.59-1.68 15 13.43 1.12 0.62-1.84 12 11.09 1.08 0.56-1.89 3 2.34 1.28 0.26-3.74 53 59.01 0.90 0.67-1.17 45 Table 11. Hazard Ratio Estimates and 95% Confidence Intervals from Time-Dependent Cox Regression Analysis to Model the Risk of Cause-Specific Mortalities as a Function of APFO Exposure Characterized by Job Classification Adjusted1' model Adjusted*1 model Cause of Death Job Exposure Classification2 Prostate cancer" No.of deaths 16 Crude HR 95% Cl Adjusted11 HR 95% Cl + actual smoking data HR + imputed smoking 95% Cl data HR 95% Cl High Moderate Low 2 3.9 0.7-21.1 6.9 1.2-41.8 5.6 0.9-35.9 7.0 1.2-42.2 10 2.3 0.7-7.5 3.0 0.8-11.0 2.8 0.8-10.2 3.0 0.8-11.0 41 1 1 1 Pancreatic cancer 13 High 0 Moderate 8 1.5 0.5-4.5 1.5 0.4-5.3 1.2 0.3-4.2 1.5 0.4-5.2 Low 5 1 1 1 1 Bladder cancer 7 High 0 Moderate Low 3 0.7 0.2-3.3 0.4 0.1-1.8 0.3 0.1-1.7 0.4 0.1-1.8 41 1 1 1 Cerebrovascular disease 35 High Moderate Low 3 2.3 0.7-8.2 5.3 1.5-19.6 6.6 1.7-24.5 5.1 1.4-18.6 19 1.5 0.8-3.1 2.1 1.0-4.5 2.2 1.0-4.8 2.1 1.0-4.6 13 1 1 1 1 Ischemic heart disease High Moderate Low 201 6 0.6 0.3-1.4 0.8 0.3-1.8 0.9 0.4-2.1 0.8 0.3-1.8 103 1.1 0.8-1.5 1.0 0.7-1.3 1.0 0.7-1.4 1.0 0.7-1.3 92 1 1 1 1 Cirrhosis of the liver 13 High 0 Moderate Low 6 0.8 0.3-2.6 1.1 0.3-3.7 1.1 0.3-3.6 1.1 0.3-3.7 71 1 1 1 Diabetes mellitus 23 High 0 Moderate Low 18 3.7 1.4-9.9 1.9 0.7-5.4 2.0 0.7-5.6 1.9 0.7-5.3 51 1 1 1 a: Job Classification. High=Worked a job with definite exposure for 6 months or greater; Moderate=Ever worked a job witli probable exposure or worked a job with definite exposure for less than 6 months; Low=Ever worked a job primarily in the non-chemical division of the plant b: Hazard ratio adjusted for sex, age eligible to be in the study, birth year, and wage type c: Men only, n=3184 46 Table 12. Hazard Ratio Estimates and 95% Confidence Intervals to Model the Risk of Prostate Cancer and Cerebrovascular Disease, Ischemic Heart Disease and Diabetes by APFO Job Exposure Classification and Wage Type Wage Type3 Cause of death No.of Job exposure classification13 deaths Hourly HRC 95% Cl No.of deaths Salaiy HRC 95% Cl Prostate cancer0 High 0- 2 9.7 1.3-72.4 Moderate 9 4.3 0.5-35.7 1 1.0 0.1 -9.9 Low 11 31 Cerebrovascular disease High 2 8.9 1.4-54.6 1 4.1 0.5-33.9 Moderate 12 1.6 0.6-4.9 7 2.8 1.0-8.0 Low 51 81 Ischemic heart disease High 5 1.1 0.4-3.0 1 0.42 0.1-3.1 Moderate 83 1.1 0.7-1.6 20 0.8 0.5-1.5 Low 40 1 52 1 Diabetes mellitus High 0- 0- Moderate 17 1.9 0.6-5.9 1 1.7 0.8-33.9 Low 41 11 a: Wage type for persons who worked both hourly and salary jobs is classified by the predominant wage type b: Job Classification. High=Wovked a job with definite exposure for 6 months or greater; Moderate=Ever worked a job with probable exposure or worked a job with definite exposure for less than 6 months; Low=Ever worked a job primarily in the non-chemical division of the plant c: Hazard ratio adjusted for sex, age eligible to be in the study, birth year, and if both wage type jobs were held d: Men only, n=3184 47 Table 13. Hazard Ratio Estimates and 95% Confidence Intervals to Model the Risk of Prostate Cancer and Cerebrovascular Disease Mortalities as a Function of APFO Exposure, by Exposure Lag (0 and 10 years) __________Zero exposure lag________________________ 10 year exposure lag Cause of Death Job Exposure No.of Crude Adjusted1' No.of Crude Adjusted1' Classification1 deaths HR 95% Cl HR 95% Cl deaths HR 95% Cl HR 95% Cl Prostate cancerc High 2 3.9 0.7-21.1 6.9 1.2-41.8 2 5.4 1.1-28.0 8.2 1.5-44.6 Moderate 10 2.3 0.7-7.5 3.0 0.8-11.0 9 3.3 1.1-9.9 3.4 1.1-10.7 Low 4 1 1 51 1 Pancreatic cancer1- High 0 0 Moderate 8 1.5 0.5-4.5 1.5 0.4-5.3 7 2.0 0.7-5.9 2.0 0.6-6.2 Low 5 1 1 61 1 Bladder cancerc High 0 0 Moderate Low 3 0.7 0.2-3.3 0.4 0.1-1.8 3 1.3 0.3-5.8 0.8 0.2-4.0 41 1 41 1 Cerebrovascular disease1-' High Moderate Low 3 2.3 0.7-8.2 5.3 1.5-19.6 3 2.2 0.7-7.6 4.4 1.2-15.5 19 1.5 0.8-3.1 2.1 1.0-4.5 14 1.4 0.7-2.9 1.8 0.9-3.7 13 1 1 18 1 1 ischemic heart disease1- High 6 0.6 0.3-1.4 0.8 0.3-1.8 6 0.6 0.3-1.3 0.7 0.3-1.6 Moderate 103 1.1 0.8-1.5 1.0 0.7-1.3 59 0.8 0.6-1.0 0.7 0.5-0.9 Low 92 1 1 136 1 1 Cirrhosis of the liverc High 0 0 Moderate 6 0.8 0.3-2.6 1.1 0.3-3.7 4 0.8 0.3-2.6 1.0 0.3-3.4 Low 7 1 1 91 1 Diabetes mellitus High 0 0 Moderate 18 3.7 1.4-9.9 1.9 0.7-5.4 12 1.9 0.9-4.4 1.3 0.6-3.1 Low 5 1 1 11 1 1 a: Job Classification. High=Worked ajob with definite exposure for 6 months or greater; Moderate=Ever worked a job with probable exposure or worked a job with definite exposure for less than 6 months; Low=Ever worked a job primarily in the non-chemical division of the plant b: Hazard ratio adjusted for sex, age eligible to be in the study, birth year, and wage type c: Men only, n=3184 48 Table 14. Hazard Ratio Estimates and 95% Confidence Intervals from Time-Dependent Cox Regression Analysis to Model the Risk of Cause-Specific Mortalities as a Function of Cumulative APFO Exposure Cause of Death Equivalent years of exposure" No.of deaths Crude IIR 95% Cl Adjusted11 HR 95% Cl Adjusted15 mode! + actual smoking data IIR 95% Cl Adjusted15 model + imputed smoking data HR 95% Cl Prostate cancerc >5 16 7 3.8 1.4-13.5 3.6 1.2 10.6 3.6 1.0-12.7 3.8 1.2-13.2 1 - <5 1 0.3 0.1-2.9 0.5 0.1-3.6 0.5 0.1-3.9 0.4 0.1-3.6 <1 81 1 1 1 Pancreatic cancer 13 >5 2 1.7 0.3-8.0 1.5 0.3-7.5 1.1 0.2-6.2 1.6 0.3-7.8 1 - <5 4 2.1 0.6-7.2 2.2 0.6-8.1 1.9 0.5-6.9 2.3 0.6-8.2 <1 71 1 1 1 Bladder cancer 7 >5 1 1.3 0.1-11.9 0.7 0.1-7.2 0.6 0.1 -7.1 0.7 0.1-7.3 1 - <5 2 1.9 0.3-10.3 1.4 0.2-8.6 1.4 0.2-8.3 1.5 0.3-8.9 <1 41 1 1 1 Cerebrovascular disease >5 1 - <5 35 9 2.3 1.1-4.9 2.4 1.1-5.4 3.1 1.3-7.3 2.4 1.1-5.5 3 0.5 0.1-1.6 0.7 0.2-2.3 0.7 0.2-2.5 0.7 0.2-2.4 <1 23 1 1 1 1 Ischemic heart disease 201 >5 21 1.0 0.6-1.5 0.7 0.4-1.1 0.8 0.5-1.4 0.7 0.4-1.1 1 - <5 42 l.l 0.8-1.6 1.1 0.8-1.5 1.1 0.8-1.6 1.1 0.8-1.6 <1 138 1 1 1 1 Cirrhosis of the liver >5 1 - <5 13 1 0.7 0.1-5.6 0.7 0.1-5.8 0.7 0.1-5.9 0.7 0.1-5.7 3 1.4 0.4-5.3 1.9 0.5-7.3 1.8 0.5-7.1 1.8 0.5-7.3 <1 91 1 1 1 Diabetes mellitus 23 >5 4 1.6 0.5-4.8 1.1 0.3-3.4 1.4 0.4-4.8 1.1 0.4-3.5 l - <5 5 1.3 0.5-3.7 1.0 0.3-2.7 1.0 0.4-2.9 1.0 0.3-2.8 <1 14 1 1 1 1 a: Weighted exposure days equivalent to 5 years (182,500 weighted exposure days), l-<5 years (36,500-182,499 weighted exposure days), and less than 1 year (<36,500 weighted exposure days) of working in a job with definite exposure, b: Hazard ratio adjusted for sex, age eligible to be in the study, birth year, and wage type c: Men only, n=3184 49 Table 15. Hazard Ratio Estimates and 95% Confidence Intervals to Model the Risk of Prostate Cancer and Cerebrovascular Disease, Ischemic Heart Disease and Diabetes by Weighted Cumulative APFO Exposure and Wage Type Wage Type" Hourly Salaiy Cause of death No.of Weighted exposure Days'1 deaths HRC 95% Cl No.of deaths HRC 95% Cl Prostate cancer11 >5 1 - <5 5 3.9 1.0-15.4 2 3.0 0.5-18.6 1 0.6 0. 7-5.9 0 0.0 <1 41 41 Cerebrovascular disease >5 1 - <5 5 1.7 0.6-5.0 4 3.5 1.0-11.5 1 0.3 0.0-2.4 2 1.6 0.3-7.4 <1 13 1 10 1 Ischemic heart disease >5 1- <5 17 0. 9 31 1.2 0.51.5 0.8-1.8 4 0.4 0. 1- 1.1 11 1.0 0.5-1.9 <1 80 I 58 1 Diabetes mellitus >5 3 0.9 0.3-3.3 1 3.3 0.2-71.7 1 - <5 5 1.0 0.3-2.8 0 - <1 13 1 11 a: Wage type for persons who worked both hourly and salary jobs is classified by the predominant wage type. b: Weighted exposure days equivalent to 5 years (182,500 weighted exposure days), l-<5 years (36,500-182,499 weighted exposure days), and less than 1 year (<36,500 weighted exposure days) of working in a job with definite exposure. c: Hazard ratio adjusted for sex, age eligible to be in the study, birth year, and if both wage type jobs were held d: Men only, n=3184 50 Table 16. Sensitivity Analysis of Hazard Ratio Estimates and 95% Confidence Intervals for Prostate Cancer, Cerebrovascular Disease, Ischemic Heart Disease and Diabetes by Weighted Cumulative APFO Exposure Using Three Exposure Weighting Schemes. Equivalent years of exposureb Prostate cancer11 >5 1-4 <1 Weighting Scheme3: Exposure weights for Non-exposed, probable exposure and definite exposure jobs 1,30, 100 N HRC Cl 1, 10, 50 N HRC Cl 1, 10, 100 N HRC Cl 16 7 4.0 1.1-14.0 5 4.5 1.3-15.1 2 8.8 1.8-42.9 1 0.5 0.1-4.2 3 1.1 0.3-4.3 6 2.5 0.8-7.6 8 1.0 8 1.0 8 1.0 Cerebrovascular disease >5 1-4 <1 35 9 3.3 1.4-7.9 6 3.4 1.3-8.7 3 7.1 2.1-24.3 3 0.7 0.2-2.5 6 1.1 0.4-2.7 8 1.7 0.7-3.8 23 1.0 23 24 1.0 Ischemic heart disease >5 1-4 <1 201 21 0.8 0.5-1.4 13 1.0 0.5-1.8 6 1.5 0.6-3.3 42 1.1 0.8-1.6 44 1.0 0.7-1.4 30 CO o 0.5-1.1 138 1.0 144 165 1.0 Diabetes 23 >5 4 1.4 0.4-4.8 2 1.2 0.3-5.7 0 1-4 5 1.0 0.4-2.9 7 1.7 0.7-4.2 7 1.3 0.4-4.9 <1 14 1.0 14 16 1.0 a: Exposure weights for Non-exposed, probable exposure and definite exposure jobs. Time in each job type multiplied by these weights, b: Exposure equivalents: weighted exposure days working < 1, 1-4, and 5 or more years of in a job with definite exposure, c: Hazard ratio adjusted for sex, age eligible to be in the study, birth year, and if both wage type jobs were held d: Men only, n=3184