Document ppRGXZEbK5z7DVydvq8YQy4mj

Corporate Occupational Medicine 3M Center, Building 220-3W-05 St. Paul, MN 55144-1000 651 737 4230 Telephone 651 733 9066 Fax 4Rm-oH?o 2. An Epidemiologic Mortality Study of Employees at the Chemolite Plant This is the original retrospective cohort mortality report of employees at the Chemolite Plant in Cottage Grove, Minnesota conducted by Schuman and Mandel of the University of Minnesota. Summary results were cited by Ubel et al (see study # 1). A review of personnel records indicated there were 3688 workers employed for at least 6 months between 1947 and 1978. Vital status was ascertained through 1978. The results of this analysis revealed no significant findings. The authors reported that a number of SMR's, particularly for the cancers, did exceed the null value but were not statistically significant. Therefore they recommended continued updating of the vital status of this cohort. It is important to note that line by line abstraction of work history cards was not done in the original study; rather only the date of first employment at Chemolite, last date of employment at Chemolite, and period of employment in the Chemical Division of the Chemolite plant were abstracted. Individual jobs and departments were not abstracted. This original data file was subsequently used for two updates (see studies # 3 and # 4). However, in the ongoing retrospective cohort mortality study of this Cottage Grove workforce (scheduled for completion by September 2000), the entire work history record has been abstracted and will be used to construct an exposure matrix related to perfluorooctanoic acid. 003126 BESTCOPYAVAII ABLE An Epidemiologic Mortality Study of Employees at the Cheaollte Plant Leonard M. Schunan, M.D. Jack S. Mandel, M.P.H.--N *5>~~i !> Sc c Epidemiology Associates February, 1980 C03127 An Epidemiologic Mortality Study of Employees at the Chemollte Plant Introduction A retrospective cohort mortality study of employees at the Chemollte Plant in Cottage Grove( Minnesota was conducted. All employees who had worked at least six months at the Plant were included in the study. The basic objectives of the study were twofold: 1. To determine whether the mortality experience of employees at the Chemollte Plant was aignlficantly different from that expected in a population group of this demographic composition. 2 . To determine whether the mortality experience of employees in the Chemical Division of the Chemollte Plant was significantly different from that expected. Methods All employee records for all divisions of the Chemollte Plant are stored in a central location (Personnel Office). The personnel recordkeeping sys tem has been designed so that complete records have been maintained on all employees who ever worked at the Chemollte Plant. These records are also maintained for employees who transfer within the 3M Corporation. Thus, in constructing the study cohort it was assumed that a complete roster could be assembled of all employees who ever worked at the Chemollte Plant since its opening in 1947. A careful review of the personnel records indicated that a total of 4218 had been employed at Chemollte for some period of time since 1947. Of these, 3688 were employed for at least 6 months. This group comprised G03128 BESTDOPYAVAIlABLE tiic' bnslc cohort for study. Information was abstracted from personnel records. In many In stances a number of records existed for a given employee and these records wi-re stored In different areas In the personnel office. Nevertheless, all sources of information were ascertained and the appropriate data abstracted. For each individual the following information was coded onto Fortran coding sheets: Employee number Name Address Telephone number Social security number Sex Blrthdate Marital status Date of first employment at Chemolite Date of last employment at Chemolite Period of employment in the Chemical Division Status (current, retired, deceased, disabled, laldoff, transferred, terminated, reason not stated and returned to school) Date of death (if noted on record) Area transferred to (If noted on record) For each of the 3688 employees, two computer cards were punched with the information mentioned above. A computer tape listing the name, social security number and blrthdate of 003129 all but the current employees was developed and sent to the Social Security Administration (SSA) to ascertain those individuals who had died during the period of study (1947 - 1978). A total of 180 deaths were ascertained from SSA records. For each death the state of last residence, Che state of last employment, or the stace from which the benefit was filed along with the date of death was listed. This Information was used to obtain a copy of the death certificate for each decedent. Unfortunately, since the most vital piece of Information - state In which the death occurred - was not pro vided by the SSA, acquisition for all death certificates was a formidable task. For those deaths for which a death certificate was not available from the state of last residence, the state of last employment, or the state from which the benefit was filed, a search was made In Minnesota using the Minnesota Health Department's commulatlve microfiche death Index 1962 - 1978 and the death records prior to 1962. Information (name, social security number, blrthdate, date of death) on all deaths for which the death certificate was not acquired by any of the above mentioned methods was sent to the Wisconsin Health Department to determine if a death certificate was on file In their records. Utilizing these procedures, 168 of the 180 (93.3X) death certificates were acquired. Information on the remaining twelve with a request for a copy of the death certificate was sent to every state health department except Minnesota, Wisconsin and New York. Minnesota and Wisconsin were excluded since their records had already been reviewed. New York was excluded since they require the borough In which the death occurred to 003130 BESTCOPY AVAII ABLE usrorcin the death certificate. However, if any of the twelve death certific.ites remain outstanding after a search by all other state health departments, then an attempt will be made to access New York vital records. The death certificates were coded by a professional nosologist ac cording to the International Classification of Diseases (I.C.D.), eighth Revision. All information in the cause of death section of che death certificate was coded. This Included the underlying cause of death, im mediate cause of death, conditions related to the immediate cause of death (referred to on the death certificate as "as s consequence of' or "due to"), and other significant conditions. These codes along with the individual's name and social security number (to facilitate linking these data to the main file) were transferred onto Fortran coding sheets and keypunched. In addition to Identifying the deaths, SSA repotted that for 85 members of the study cohort no benefits were currently being paid. This meant that SSA could not emphatically state whether these individuals were dead or alive. However, published reports (see for example Wang, H. and MacMahon, 6., J. Occup. Med., 21, 745, 1979) have stated that the ascertainment of mor tality by SSA is virtually complete and that additional follow up on in dividuals for whom no information Is available from SSA records Is u n n e c e s sary. Nevertheless, in order to be absolutely certain that all deaths were ascertained, every effort was made through extensive tracing efforts tn I'-rermine the current status of these 85 unaccounted for study members. The status of virtually all of these people was determined in this way. Method of Analysis The method of analysis Is based on a comparison of the observed to expected death rates specific for cause, age, time, sex and race. The G03131 BEST COPY AVAll ABLE ini'iliod is descrlhe*i in detail in Monson, R., Analysis of relative survival and proportional mortality, Comp. Biomed. Res., 7, 325-332, 1974. The output from this analysis provides the observed and expected number of deaths, the ratio of observed to expected (SKR), and the upper and lower 95Z confidence limits. The 95Z confidence limits are defined as follows: Upper Limit - UL - u n ) ( l - l/9(n) + 1.96/34 l / ( d u ) ) 3 ) /E Where 0 " observed number of deaths E " expected number of deaths Results Records on 4,216 employees were reviewed and abstracted (Table 1). Only those who worked at Chemolite for six or more months (3,688) were included in the mortality followup. i# the 180 known deaths, 168 (93.3Z) death certificates were obtained. There were 159 male deaths and 21 female deaths. Table 2 presents the results for Chemolite Plant male employees. Overall, the number of deatha (149) was significantly less than expected (208.3). This "deficit" was primarily due to the lower than expected number of deaths from diseases of the circulatory system and all external causes of death. The observed-to-expected ratio for cancer deaths was 1.0, indicating no disparity between the number of deatha observed and the number expected. The only 0/E ratios which exceeded 1.0 were for deaths due to asthma (2.8) and motor vehicle accidents (1.1). However, neither even approached statistical significance. 003132 C | c 1. Epidemiologic Mortality Study of Chemolite Plant July, 1978 - December, 1979 Records Abstracted Number Studied Number of Deaths Death Certificates Analyzed <>,218 3,688 180 (A.9Z) 168 CG3133 BEST COPY AVAIIABLE Table 2. Results of Cause-Specific Mortality Analysis Males, Chemollte Plant ( ^ 6 Booths) of Death 140 - Malignant Neoplasms 000 - All Infective and Parasitic Diseases Tuberculosis 240 - Allergic. Endocrine, Metab., Nutr. Diseases Diabetes Mellitus 320 - All Diseases of Nervous Sys. 6 Sense Organs 390 - All Diseases of Circulatory System Chronic Rheumatic Heart Disease Arteriosclerotic H.D., Incl. CHD Ail Vascular Lesions of CHS 460 - All Respiratory Diseases A.11 Pneumonia Emphysema Asthma 520 - All Diseases of Digestive System All Castrie and Duodenal Ulcer Cirrhosis of Liver 800 - All External Causes of Death All Accidents Motor Vehicle Accidents Suicide All Causes of Death 0E 37 37.1 I 2.5 1 1.3 2 3.4 2 2.8 2 2.4 64 83.0 1 2.6 s o 61.4 4 8.5 6 8.9 2 3,5 1 2.0 1 .4 7 12.4 1 1.3 4 7.6 29_ 49.0 24 33.4 20 18.3 5 10.1 0/E 1.0 .4 .8 .9 .7 .9 .8 .4 .8 .5 .7 .6 .5 2.8 .6 .8 .5 .6 .7 1.1 .5 LL UL .7 1.3 .0 2.3 .0 4.5 .2 2.6 .1 2.6 .1 3.1 .6 <1.0 .0 2.2 .6 1.1 .1 1.2 .3 1.5 .1 2.1 .0 2.8 .0 15.8 .2 1.2 .0 4.4 .1 1.4 .4 .8 .4 > 1.0 . 6 1.6 .2 1.2 149 208.3 .7 .6 .8 03134 Tible J- Results of Cause-Specific Mortality Analysis Males. Chemolite Plant 6 months) Cancer Cause of Death All Malignant Neoplasms Cancer of Digestive Organs & Peritoneum 150 - Cancer of Esophagus 151 - Cancer of Stomach 153 - Cancer of Large Intestine 154 - Cancer of Rectum 157 - Cancer of Pancreas Cancer of Respiratory System 161 - Cancer of Larynx 162 - All Cancer of Lung (Prim. 6 Sec.) 172 - Cancer of Skin 185 - Cancer of Prostate 186 - Cancer of Testis 189 - Cancer of Kidney 191 - Cancer of Brain & Other CNS All Lymphopoietic Cancer 201 - Hodgkins Disease 208 - Cancer of Lymphatic Tissue 0 E 0/E 1.1. VI 37 37.1 1.0 _9 9.0 1.0 1 .8 1.3 1 1.6 .6 3 2.9 1.1 1 1.0 1.0 2 1.9 1.1 9 12.5 .7 1 .5 1.9 8 11.8 .7 1 1.1 .9 2 1.0 2.0 1 .7 1.5 1_ 1.0 1.0 2 1.9 1.1 4 5.0 .8 2 1.1 1.9 2 .9 2.2 .7 1.3 .5 1.9 .0 7.1 .0 3.5 .2 3.1 .0 5.6 .1 3.9 .3 1.4 .0 10.7 .3 1.3 .0 5.0 .2 7.3 .0 8.3 .0 5.6 .1 3.9 .2 2.0 .2 6.8 .3 8.0 158 Other Digestive 1 195 111 Defined Sites 1 199 Malig. Neopl. Without Specif, of Site 6 003135 BESTCOPYAVAIIABLE In ruble 3, the 3? cancer deaths are distributed by site. None of the results were significant. Tables A and 5 present results for deaths among men who were employ ed for at least six months in the Chemical Division of the Chemollte Plant. Overall, the observed number of deaths (55) was less than expected (66.1). This was mainly the result of the lower than expected number of deaths from diseases of the circulatory system. The highest 0/E ratio, 9.2, occurred for deaths due to asthma. However, this ratio, based on only a single death in the 21 year study period, was not statistically significant. A total of 13 cancer deaths occurred among the Chemical Division male employees. This was approximately the number expected (11.2). Although the 0/E ratio for a number of cancer sites exceeded one, none were statisti cally significant. A distribution of the 19 observed deaths among female employees is presented in Table 6. No major departures from expected were noted in these results. Discussion The lower than expected mortality from all causes la often found In studies of occupational cohorts. This reflects "the healthy worker" effect. Generally, individuals who are disabled or chronically ill are excluded from certain work situations, particularly those requiring considerable physical activity. Thus, employee cohorts compared to the general population general ly have more favorable mortality outcomes, particularly for younger employees. To compensate for this effect the data could be reviewed with an 0/E ratio of less than 1, (perhaps .9 is appropriate) considered equivalent to no difference between observed and expected deaths. Reviewing the data with 03136 Table 4. Results of Cause Specific Mortality Analysis Males. Chemical Division & 6 Months) 0 E 0/E LL ITL 140 - Malignant Neoplasms 240 - Allergic, Endocrine, Metab., Nutr. Diseases Diabetes Mellitus 320 - All Diseases of Nervous Sys. 6 Sense Organs 390 - All Diseases of Circulatory System Arteriosclerotic H.D. Incl. CHD 460 - All Resplr*tory Diseases All Pneumonia Asthma 520 - All Diseases of Digestive System All Gastric and Duodenal Ulcer Cirrhosis of Liver 800 - All External Causes of Death All Accidents Motor Vehicle Accidents Suicide All Causes of Death 13 11.2 1.2 .6 1.9 2 1.1 1.9 .2 6.8 1 .9 1.1 .0 6.5 1 .8 1.3 .0 7.0 16 24.3 14 17.9 .7 .4 1.1 .8 .4 1.3 2.7 1.5 .4 3.9 2 1.1 1.9 .2 6.7 1 .1 9.2 .1 51.1 5 3.9 1.3 .4 3.0 1 .4 2.6 .0 14.5 3 2.4 1.3 .3 3.7 l i 18.1 .8 .4 1.3 12 12.5 1.0 .5 1.7 9 7.0 1.3 .6 2.4 2 3.6 .6 .1 2.0 55 66.1 .8 .6 1.1 C03137 BEST COPY AVAII ABLE Table 5. Results of Cause-Specific Mortality Analysis Males. Chemical Division (.6 Months) Cancer Cause of Death 40 - Malignant Neoplasms Cancer of Digestive Organs 6 Peritoneum Cancer of Esophagus Cancer of Large Intestine Cancer of Pancreas Cancer of Respiratory System Cancer of Larynx Cancer of Lung (Prim. & Sec.) Cancer of Skin Cancer of Prostate Cancer of Testis 0 E 0/E LL UL 13 11.2 1.2 3 2.6 1.1 1 .2 4.4 1 .9 1.2 1 .6 1.8 4 3.7 1.1 1 .2 6.7 3 3.5 .9 1 .4 2.7 1 .3 3.9 1 .3 4.0 .6 1.9 .2 3.3 .1 24.5 .0 6.6 .0 10.2 .3 2.8 .1 37.3 .2 2.5 .0 15.1 .1 21.8 .1 22.1 199 - Malig. Neopl. Without Specificatlon of Site - 3 003138 BESTCOPY AYAHABU Table 6. Specific Causes of Death Females. Chemollte Fiant (J6 months) 009 Diarrheal disease 162 Mallg. neopl. of lung 172 Mallg. neopl. of skin 174 Kallg. neopl. of breast 183 Hallg. neopl. of ovary 191 Mallg. neopl. of braid 402 Hypertensive H.D. 410 Acute fi.I. 412 Chronic Ischemic H.D. 427 Symptomatic H.D. 472 Influenza with other reap, manifestations 493 Asthma 812 Car accident with other car 819 Car acc. of unspecified nature ' 0 1 1 1 2 1 1 r\ i/ 4( 2\ 1 1 J E .4 1 .2 2.5 .8 .4 7.5 1 .1 1.5 Total 19 I C03139 K S T COPy AVAIIABLE a null hypothesis of .9 would not substantially alter the conclusions. Deaths for U.S. white males and females were used to derive expected values. A more appropriate comparison population might have been deaths for the counties in which the employees resided, namely Dakota, Ramsey and Washington in Minnesota and Pierce and St. Croix in Wisconsin. There are two major problems with using data from these counties to derive expected values. First, age - sex - race and cause specific death rates are not available by county for periods back to the late 1940's. Second, if such rates could be derived from the basic data, the relatively small population bases would result in highly unstable rates. One strategy for considering geographic differences In mortality is to examine available data for the counties of Interest in relation to the United States rates. For cancer, age adjusted death rates are available by county for the period 1950-69. These data were examined for the above mentioned counties and compared to the rates for the United States. For those cancer sites where the rate was based on an average of at least one death per year the only notable difference between the rates for the 5 counties and the U.S.,was for lung cancer where the age-adjusted rate was appreciably lower in 4 of the 5 counties than in the U.S. as a whole (the rate for Ramsey county approximated the U.S. race). The 0/E ratio for lung cancer was less than 1.0. This Is sn underestimate of the SMR if calculated using county rates. However, re-calculating the SMR using county data to derive expected values would probably result in an SMR of about 1.5, which would not be statistically significant. Although the results of this analysis revealed no significant findings,a number of SMR's, particularly for the cancers, did exceed one. Thus, continued surveillance for mortality is warranted. 003140 BESTCOPYAVAIIABLE As of the writing .of this report (February 26, 1980) information on the 12 deceased study members whose death certificates were not available at the time the data was analyzed Is as follows: 2 died in service, one In Vietnam and one in Korea 4 died from cardiovascular disease 2 died from cancer 1 died from injuries sustained In an airplane crash 3 cause of death unknown (2 of these deaths occurred among employees who worked at Chemollte for less than 6 months) From the above distribution of deaths, It Is obvious that the addition of these deaths to the analyses presented In this report does not alter the conclusions. G03141