Document eQVooEKrb7v5jmbeyKqewYjG

University of Lowell Department of Work Environment College of Engineering A REPORT TO THE GENERAL ELECTRIC COMPANY CONCERNING A CASE-CONTROL STUDY OF CANCER MORTALITY AT THE GENERAL ELECTRIC PITTSFIELD FACILITY SxcutlY Summary (January 2, 1990) David H. Wegman Professor and Head Department of Work Environment University of Lovell Lovell, MA HONS 028598 One University Avenue Lowell, Masiachuiens 01854-9985 508/452-5000 Fi* 508/452-5711 A REPORT TO THE GENERAL ELECTRIC COMPANY CONCERNING A CASE-CONTROL STUDY OF CANCER MORTALITY AT THE GENERAL ELECTRIC PITTSFIELD FACILITY EHEfiMtiYS Summary (January 2, 1990) David H. Wegman BACKGROUND This study of mortality in Gansral Elsctric vorJcars grew out of a preliminary, unpublished study which was part of a series of worker mortality studies undertaken in the late 1970's. The preliminary study focused on General Electric Company in Pittsfield and found an excess of mortality from leukemia and cancer of the large intestine among those persons where General Electric's name was entered on the death record filed from the town of Pittsfield. The findings from the preliminary study ware described in 1978 to the General Electric Company, which cooperated in the further investigation of the problem. Initially General Electric provided a list of insurance pension records for all employee deaths from 1969 to 1977 which had resulted in death benefit claims, study of thesa data, which included last job at General Electric, did not change earlier findings. No specific last job type or exposure was associated with the proportional mortality excesses noted. In January 1981, it was decided to obtain complete work histories for all pension eligible workers who had died. Subsequently it was decided to undertake a detailed and in depth historical reconstruction of worker exposures to groups of substances considered to be most commonly in use. This reconstruction of exposures for each former employee in the study was attempted for Pyranel (a mixture of polychlorinated biphenyls and trichlorobenzene), trichloroethylene, benzene, other solvents, asbeatos, machining fluids and miscellaneous resin systems. Individual work histories were matched to the exposure data base to generate individual estimates of exposure histories to each group of chemicals. Since not all subject work history records were still avail able, the total number of subjects available for study was lass than the total number of deaths recorded. In order to maximize the study's statistical power, therefore, the mortality data base was extended twice until finally deaths ware included from 1969 through 1984. 1 HONS 028599 The final study population, therefore, includas thosa pension-eligible workers who diad at sona time batwaan 1969 and 19(4 for whoa work history racords could ba racovarad. Tha protocol for tha currant study was aodifiad and adoptad in its final fora in 1985. Tha objactiva of tha study was to dataraina, to tha extent possibla, if thara was avldenca for work ralatad cancar risks, a casa-control approach was followadt exposure historias of cancar casas wars cooparad with those of a control group consisting of workars who diad of causas'not thought to ba ralatad to tha axposura undar study. OVERVIEW or THE 8TUDX Study Subjacts Tha initial study size was 2,914 whita mala amployaas who met "vasting requirements" and whosa last placa of work was at tha GE Pittsfield plant and who diad batwaan January l, 1969 and Dacaabar 31, 1984. Thosa includad nay hava diad aithar while actively employed or after retiring. For all individuals includad in tha study, official death certificates wars obtained. Extensive efforts ware made to collect conplata information on all subjects. After exclusions were made for inadequate or absent work history records and appropriate exclusions made for selection of control subjects, a total of 1,714 subjects were included in the study. Work History Reoorda work history records wars sought for all deceased subjects enrolled in the study but turned out to ba unavailable for a substantial number largely due to elimination of these racords according to standard company policy on record retention. Although there was no evidence that tha losses ware biased thara was no way to evaluate tha losses. occupational Exposures The following exposures thought to be of primary importance wars rated (primarily by systematic overlapping interview* with long-term employees)t Pvranols e transformer oil composed of polychlorinated biphenyls (PCBs) (a mixture of isomers but mostly haxachlorobiphanyl), trichlorobenzene (or a mixture of tri- and tetrachlorobenzane), lass than 0.25t phenoxypropene oxides and trace amounts of dibanzofurans. Tha PCB content in Pyranol could vary from 45 to 80S. 2 HONS 028600 Benzene: a solvent usad in various dapartnants for general cleaning during machining and assambly oparations. Triehloroathvlana fTCE): a solvant usad as a dagraasar Othar solvents: this group inciudas Varsol (patrolaum spirits), CPE 1000 (patrolaum spirits and methylene chlorida), mathylana chlorida, kerosene, paint thinnars (primarily xylene or toluana basad), solvant basad paints, xylana, toluana, and naphtha, soma typa of solvant axposura occurrad in tha majority of plant oparations. Machining fluids; usad for machining and fabrication oparations. Straight minaral oils wars first usad, than solubla oils and finally synthatic oils. Thara was a vary minimal usaga of straight cutting oils. Aabaatoa: Usad as wat insulation blankets during brazing and voiding. Soma insulation piacas wars mads from asbastos. Basin svstams! primarily phanol formaldahyda and polyvinyl formal rasin systams. Exposuras known to bo prasant but not spocifically ratad wars: Minaral oil: (10C oil) usad as transformar oil. Pyranol was usad as a transformar fluid only in 15-29% of tha transformers built and minaral oil was usad in tha ramaindar. A raviaw of tha axposura to Pyranol was conductad in March 198S, and tha following changes ware mada to the existing exposure ratings, in order to improve separation between tha exposures to Pyranol and to minaral oil: a rating of no exposure was assigned to all jobs in which thara was no dielectric fluid usad or in which tha dialaetric fluid usad was certainly minaral oil and not Pyranol. Aa a result, all jobs in buildings 4, 5, 19, 100, 41 and 44 wara assigned a rating of o for Pyranol. Metal fumes and dust: exposure occurrad during welding, brazing and painting with matal basad pigments. Sawdust: in woodworking shops. Adhaslvasi watar, solvant based and epoxy (those using patrolaum solvents are ratad with group 4) Electromagnetic fields: exposure occurrad during testing and development of transformers. 3 MONS 028601 DESIGN AND ANALYSIS Basie Design The study was designed as a case control study of cancer mortality risks. Distribution of demographic characteristics was used to obtain a preliminary description of differences between cases and controls. The characteristics of those with and without work histories were compared to examine bias due to excluding those without work histories. The data ware than divided into various subcategories (stratified analysis) to examine combinations of exposure and other variables (for example,, age, year of death, etc.). This analysis sought to separate exposure effects from affects attributable to other variables. Initially, the less certain cause of death diagnoses were validated by clinical records. Then case and control status was assigned as follows: a) Casas are all types of cancer. b) Controls ware the non-cancer deaths excluding diseases of the digestive system, genitourinary diseases, diseases of the blood and the blood forming organs, mental disorders and ill defined conditions. While the stratified analysis provided an essential description and summary of the data, it was limited when attempting to look at several variables at once. Because of sparse data, special methods were used to examine several variables simultaneously. Careful attention was paid to following a predefined strategy so that the selection of tha models would not be inappropriately affected by the results. Finally special attention was directed at the impact of time on risk. Each exposure was considered first separately and then two or more exposures were placed in the same models in order to obtain estimates of relative risk adjusted for other exposures. Interactions between variables were also considered. Of greatest concern is the lack of information on the smoking history of study subjects. Smoking is a risk factor for several of the cancer sites under study. Therefore it was necessary to assess its potential confounding effect. This was done by hypothesizing several possible distributions of smoking in the study population. 4 HONS 028602 Statistical Powar Statistical powar is a naasura of the probability (given a specified magnitude of association) of obtaining a sat of data such that laval of an affact baing sought is likely. A conventional powar analysis shovad that this study was ganarally charactarizad by low powar, unless tha affacts wara vary strong or the causes of death common (a.g. lung cancer). RESULTS Tha results of tha study are summarized in the attached Chapter III that also' sarvas as tha summary chapter of the datailad report. Tha detailed report includes axtansiva tabular material in support of tha epidemiologic conclusions. In brief, tha epidemiologic analysis did not identify any unequivocal or definitive association of tha General ElectricPittsfield work environment with excess cancer risk, with limited power tha study failed to find an association between pyranol (containing PCBs) and excess cancer risk among employees. In addition cancer of tha large intestine, one of tha two cancer sites which was in excess in the exploratory proportional mortality studies, was not found associated with tha exposures included in the study. Several findings, however, suggest further study vhich may clarify certain of the apparent associations. In particular, the association of lung cancer with operations coded as "rasins" needs to be further evaluated. To do so requires new coding of asbestos exposures in these locations to determine whether tha associations noted are attributable to unestimated asbestos exposure. Other associations deserving further study include: trichloroethylene and leukemia: benzene and cancer of tha esophagus, bladder, kidney, and brain; other solvents and cancers of the kidney and the lymphatic system (especially raticulosarcona); resin components and cancers of the esophagus, large intestine and lung; and machining fluids and kidney cancer. COMMENT AMD RECOMMENDATIONS This study was carried out using state of the art epidemiologic and exposure assessment attribution methods. In some instances new methods were developed to address some of the more difficult data problems or analytical dilemmas. Nonetheless, findings from this study must be tempered by tha fact that a number of important problems existed which limited the ability of the study to examine fully the question of cancer- 5 MQNS 028603 related risks from exposure te toxic agents in this work environment. Ths problems of greatest importance includa: 1. Records of personnel activity had not baan maintained in a manner which sarvas tha naads of an epidemiologic atudy. Thara was no affactiva way to documant employment at tha Pittsfield plant at a tima sufficiantly far in tha past to undartaka a population-basad study using an historical cohort approach; 2. Information on pravious employees was liaitad to pensioneligible amployaas, a assignation which changes many timas ovar tha study intarval. This fact mada interpretation of tha findings much mora difficult, rurtharaora, ths racord of pensioners was raliably availabla only baginning in 1969; this liaitad tha total numbar of subjaets who could ba includad in tha study; 3. Tha naad to raly on daath racords liaitad tha afficiancy of tha study to thosa canears which ara likaly to causa daath (a high case-fatality ratio). Hanca a cancar with ralativaly good survival (for axampla, cancar of tha urinary bladder) is not as wall studiad as for work-ralatad causas sines tha numbar of casas will far exceed tha numbar of daaths attributad to this causa; 4. Evidanea of cancar had to ba basad on information listad on tha daath cartificatc. Although a major affort was mada to varify casas with clinical racords, it was only faasibla to attampt this for thosa canears raportad to bo lass wall documsntad. 5. Vary limitad data wara availabla on historical work-placa axposuras to tha toxic matarials of intarast. For most of tha exposures of intarast there was no information for tha vast majority of tha study intarval. As a rosult, axposura ratings had to ba basad pradominantly on subjactiva assassmant using only a faw catagorias, which limits tha powar to distinguish batwaan axposura lavala. Furtharmora, tha axposura ratings do not raflect individual axposura variation. Thara is a high probability, thsrafors, that avan if olavatad cancar risks axist in this anvironmant thay might not ba found. 6. Work history racords, Ilka tha ganaral parsonnal racords, wara incoaplataly maintalnad. standard policy on racord elimination to raduce voluma of storad paper resulted in loss of a third of the work history racords of potential interest; ?. Thara had baan multiple changes in the work organization, so that operations of intarast wara carried out in different 6 HONS 028604 plant buildings at different times. Tha historical racord of thasa changes was vary incomplata. 8. Healthy-worker salaction effects could not ba corractad for. This affact, although lass important in studias of cancar mortality, rasults from tha fact that workers at incraasad risk (a.g. disablad employees) will hava a tandancy to laava amployment aarly and tharafora ara mora likely not to ba includad in tha study. Mathods to corract for this affact raquira knowladga of tha exposure history and data of tarmination for workars who laft employment. This information is not availabla for this study. 9. Of soma concarn is tha lack of information on tha smoking history of study subjacts. Tharafora, it was nacassary to assass its potantial confounding affact. It was shown, howavar, that smoking was unlikaly to ba a strong confoundar in our study. 10. Tha variaty of problama invariably limit tha statistical powar of tha study to find ralativaly small affects, even if they ara prasant. conventional statistical powar analyses showed that this study is ganerally charactarized by low powar, except for vary strong affaets or for outcomas with a larga number of cases (lung cancar). ***** Baaed on the associations notad in tha apidamiologic study, and on tha absanea of any dafinitiva findings, tha following ara recommended: 1. further study is indicated to attempt to clarify tha findincs for luno cancar and work in resin systems. Because the predominant plastic product manufactured in tha rasin systems category involved the use of asbestos and phenol-formaldehyde, this finding cannot ba clearly interpreted. Asbestos is widely aceeptad as a human carcinogen and OSKA has recently recognized formaldehyde as a carcinogen. Tha jobs in this axposura grouping ware not characterized for exposure to asbestos nor specifically for formaldehyde and the process has not bean operated in the Pittsfield facility for soma time. The possibility for further elucidation of work risks for this area should ba evaluated to determine whether tha sourca of risk can ba battar specified. The feasibility of further study will depend upon the estimate of possible statistical powar (bastd on tha number of parsons who vorkad in tha area and whose work records are available) and tha determination that reasonable job spacifie axposura assignments can ba made for asbestos and formaldehyde. 7 HONS 028605 2. Tha-Workforc.t_-Should ba notified of the findings of this study. A summary of ths rasults should b prsparsd and distributed to all currant amployaas and all ratlraas. In addition tha rasults should ba providad tha union for publication in its newspaper, and to tha local prass for general publication. 3. A_maatlnc should ba -held with tha local physician organization. This meeting should explain tha rasults and answar quastions about tha possibla follow-up which night ba indicatad. 4. fallow-un medical axamlnation should ba offarad in salaetad clrcumstancas. Thara ara no accaptad scraaning tasts for lung cancar which can ba offarad to tha pravious or currant anployaas. It is, howavar, appropriata to provida for an organized rasponsa to raquasts for haalth risk appraisal fron any who night axprass a concam ragarding lung cancar or othar haalth concarns ralatad to this study. It is raconmandad that an outsida contractor ba hirad to develop a progran along tha following linas: a) All currant anployaas and ratlraas who can ba contactad should ba givan tha opportunity to attand a uniform cancar aducation and awaranasa progran on conpany tins on tha plant's pranisas. Tha format should ba anall group sassion that addrass ganaral cancar and cancar of tha lung. Tha contant of tha progran should includa an ovarviaw and introduction on tha study rasults, tha intarvantions or actions which hava takan placa or ara plannad and tha madlcal service program (MSP). Tha ganarie discussion of cancar should includa what it is, how it davalops, how ic is traatad, tha causas of cancar, risk-factors, occupation spacific risks for tha targat group, aarly warning signs and symptoms, aarly detection and diagnosis, traatnant and prognosis and pravantion. b) A panal of physicians (intarnal madicins, occupational medicine specialties) be created to provida assessment and consultation to any employes who has further concarns about the haalth problans identified in tha educational program, specifically lung cancar. Any participant in tha educational program could request this consultstion which should ba an individual, private consultation with a uniform report subnittad to the requestor. 8 HONS 028606 c) A second panel should ba craaead of physicians who art expert in tha diagnosis and treatment of lung cancar and asbastos ralatad disaasa. Tha first panai would rafar any participant who raguiras additional diagnostic workup for cancar of tha lung to this panai. Part of this work-up should ba a targatad examination for thosa judgad to hava had othar than incidental asbastos exposure. (For axampla, chast X-ray scraaning for avidenca of asbastos-lika plaural raaction and/or sputua examination for asbastos fibars could, aftar furthar considaration and raviaw, ba daamad appropriata to offar workars who wara employed in tha rasin systems areas). Advice on tha importance of ceasing cigaratta smoking should ba provided and'an organized program of smoking cassation should ba sponsored for any cigaratta smoker, but especially for any parson with known prior axposura to asbastos. d) Protocols should bo developed to guide decision-making by the two panels to assure uniformity and consistency, these guidelines or protocols should be developed by tha physicians and approved by a group which agrees jointly to represents tha interests of tha community and tha company and who serve as an advisory group to any contractor salaetad to coordinate this overall effort. 9 MQNS 028607 BACKGROUND INFORMATION ON THE CASE-CONTROL STUDY OF CANCER MORTALITY AT THE GENERAL ELECTRIC PITTSFIELD FACILITY January 24, 1990 Prepared by GE HONS 026600 BACKGROUND INFORMATION ON THE CASE-CONTROL STUDY OF CANCER MORTALITY AT THE GENERAL ELECTRIC PITTSFIELD FACILITY January 24, 1990 This document provides background Information on tht Case-Control Study of Cancer Mortality at the General Electric plant In Pittsfield, Massachusetts (the Study) In a question/answer format. The document was prepared by GE scientists based, In part, on discussions with the principal Investigators and the panel of technical advisers. MONS 028609 -2CONTENTS I. STUDY HISTORY A. Why was tha Study Initiated? B. What was thi objective of tha Study? C. What wara tha substancas and cancar typas Includad In tha Study? D. Who wara tha members of tha Study team? E. What was tha rota of tha Scientific Advisory Panel? F. What governmental ravlaw was provldad? 6. What was tha rola of SE management? H. How long did It taka to complete tha Study? I. What ara PCBs? II. EPIDEMIOLOGY A. What Is epidemiology? B. What ara tha limitations of apldamlology? C. What typas of apldamlologlc studlas wara parformad? III. DATA COLLECTION AND ANALYSIS A. How was tha Study population assamblad? 8. How wara tha casas and controls daflnad? C. How was tha axposura data compllad? D. What othar typas of data wara collactad? E. How wara data for tha casas and controls analyzed? Pagt 3 3 4 5 5 6 7 7 8 8 8 9 9 10 10 11 11 HONS 028610 -3- 1. BACKGROUND A. Why was tha Study Initiated? In tha lata 1970s, David H. Wegman, M.D., then occupational hygiene physician with tha Massachusetts Olvlslon of Occupational Hygiene and also at Harvard, conducted a series of mortality studies which focused on selected New England towns. These studies were aimed at determining whether the proportion of deaths due to cancer for specific employees or occupational groups In these towns exceeded that of the U.S. general population. The purpose was to determine whether further In-depth definitive studies were needed. *We chose towns In which the population was relatively stable and In which one or several plants or Industries generally dominated employment," Dr. Wegman noted In a recent interview. "Pittsfield, Massachusetts - site of GE's Transformer, Ordnance and Plastics operations - was such a town." "In Pittsfield, we found an excess of mortality from leukemia and cancer of the large Intestine among people whose death records cited GE as the last employer," he added. In 1978, Dr. Wegman communicated his findings to GE management In Pittsfield. The company then cooperated In an expanded Study by providing a list of Insurance/pension records for all active and retired employee deaths between 196! and 1977 (Including out-of-state, employee deaths) that had resulted in death benefit claims. The analysis was repeated and the results remained the same. There was an Issue whether these excesses were related to chemicals used at GE Pittsfield, Including PC8s. In order to evaluate whether this was the cate, a comprehensive cancer mortality Study was undertaken with Dr. Wegman as principal Investigator. 8. What was the objective of the Studyt The objective of the Study was to determine If there was evidence of work-related cancer risks In the Pittsfield work force. "A case-control approach was followed," Dr. Wegman wrote In his Summary of the Study. "Exposure histories of cancer cases were compared with those of a control group consisting of workers who had died of causes not thought to be related to the chemical exposure under Study." MGNS 028611 -4- Th Study cov*rd the following GE Pittsfield manufacturing divisions; Power Transformers and Distribution Transformers, Ordnance Systems and Plastics Division. The study team Identified seven substances commonly used In the Pittsfield plants. The atm was to examine every kind of link (or association) that might exist between exposure at work and the risk of cancer. C. What were the substances and cancer types Included In the Study? The seven substances as defined by the Study are: e Pyranol: a transformer dielectric fluid composed of PC8s (a mixture of Isomers but mostly hexachloroblphenyl), trlchlorobenzene (a mixture of trl- and tetrachloro- benzene), less than 0.2S percent phenoxypropene oxides and trace amounts of dlbenzofurans. The PC! content In Pyranol could vary from 45 to 80 percent. a Benzene: a solvent used In various departments for general cleaning during machining and assembly operations. e Trichloroethylene (TCE): a solvent used as a degreaser, e Other solvents: this group Includes Varsol (petroleum spirits), CPE 100 (petroleum spirits and methylene chloride), methylene chloride, kerosene, paint thlnners (primarily xylene or toluene based), solvent based paints, xylene, toluene and naphtha. Machining fluids: used for machining and fabricating operations. Straight mineral oils were first used, then soluble oils and finally synthetic oils. e Asbestos: Used as wet Insulation blankets during brazing and welding. Some Insulation pieces were made from asbestos. Resin systems: primarily phenol-formaldehyde and polyvlnyl-formtl resin systems. Initially all cancer types were considered. However, for In-depth analyses, cancer types with nine or more observed deaths (cases) were selected. These were orolaryngeal, esophaous, stomach, colon, rectum, pancreas, liver (liver, biliary and gall bladder), lung, prostate, urinary bladder, kidney, lymphomas, leukemias and brain. MONS 028612 -6- 0. Who ara th nenbart of tho Study taaa? David H. Wagmsn, H.O., Profonor md Hud of tho Departmint of Wort Environment. University of LomiII (Massachusetts), wti tho principal Investlgstor for tho projoet. As sueh, ht Is responsible for overall direction tnd auparvlilon, os will u Interpretation ond coamunleatlon of tho Study findings. With formir potto t Harvard md UCLA, Or. Wegmtn It o well-recognized Investigator In tho field of occupational epidemiology (tot following soetlon on tpldonlology), Ho hoi published numerous resnrch piport In lndlng telintlflc Journals, Stndtr Greenland, Or.P.H., Professor of tpldonlology ot UCLA, was tho co-principal Invootlootor. Ho joined the Study toon in 1964 and aiiumod major responsibility for developing and Implementing analytical methodology. He It noted for hit many contribution! to theoretical epidemiology, tnd hat publlthtd nany rttttreh pipers detllng with thete ttpecti of tpldenlology. Dr. Greenland was milted on this project by Alberto Silvan, H.O., on thlt project. Thomis Smith, Ph.D., Profonor of Induitrltl Hyglone it tho University ef.Mimchuietti, astilted by Htrllyn Hillock, lnduttrlil Hyglonlit, also it the Unlvorilty of Miiuehuiitti, ertitod the eonprihontlvt inployoe ixpoturt profile. Dr. Smith it widely recognized to i leader In exposure meisnent tnd exposuri modeling. I. Whit wit the rote of the Scientific Advisory Penal? ngulthcd group 'expertliVl? the" fleldrof Little* md Induttrlal hygiene. ujKKItt .he panel hu net eeven tlnet tnd hir'STJo lid plant. Tho Panel'i charter Included rovltwlng the Study rntnodelogy, pirtlcipitlng in discussions tnd meatlngi, and providing technical idvlci to md Intirictlng with thi Study tun. Thonai Sinks, Ph.O., Suparvlaory Epidemiologist for NIOSH (the Nttlontl (ntiituta for Occupational Stftty tnd Health) wrote the following to Miiaachuiitti State Representative Sherwood tuomioy (0`Wlllluutown) on Hay 4, 1189 iftor ittending t Scientific Advisory Panel anting on April St, 1(89 in Baltlnorti "1 wit (truck by tha pair ravlew eomnlttee.'.t ooneiro,foc.,the.M?rkiri..nd ***- T'-i--k------------- faien'^ *rwall#by HONS 028613 Members of the Scientific Advisory Panel ere: Leon Gordls, M.D., Dr. P.H. (Chair) Professor and Chairman Department of Epidemiology The Johns Hopkins University Baltimore, MO e Genevieve Matanoskl,'M.D., Dr. P.H. (Assist. Chair) Professor of Epidemiology Department of Epidemiology -The Johns Hopkins University Baltimore, MD e Michael Gerln, Ph.D. Associate Professor Occupational and Environmental Health University of Montreal Montreal, Canada e Jack Mandel, M.P.H., Ph.D. Associate Professor School of Public Health University of Minnesota Minneapolis, MN e Howard Rockette, Ph.O. Professor of Biostatistics School of Public Health University of Pittsburgh Pittsburgh, PA e James Schlesselman, Ph.D. Professor of Biostatistics Uniformed Services University of Health Sciences Bethesda, MO P. What governmental review was provided? Representatives from the National Institute for Occupational Safety and Health (NIOSH) did the following: reviewed and commented favorably on the research protocol in 1985; participated In a series of meetings In Pittsfield with Or. Wcgman, GE and Company retirees; met with Or. Wegman, Or. Greenland and GE at UCLA; and attended several Scientific Advisory Panel meetings In the role of observer. HONS 028614 -7- In addition, Representative Guernsey monitored the progress of the Study. G. What was the role of GE management? GE management provided financial support In the form of research contracts to the universities represented by Drs. Wegman and Greenland, and reviewed progress periodically. It assisted the data collection process by providing work history records and death certificates. It set up a series of plant walk-throughs and Interviews with plant personnel, by which the exposure assessment team observed and evaluated current work practices.' These steps were critical to creating the exposure assessments that formed a primary data base for the Study. GE also established the Scientific Advisory Panel. H. How long did It take to complete the Study? The In-depth case-control Study began In 1981. Data were collected through an Intensive Interview process to recreate environmental conditions and exposures from memory. Assuring that the data were appropriate for In-depth statistical analysis turned out to be a long and complex process. The Study began with worker deaths from 1989 to 1977; the number of deaths proved to be Inadequate for In-depth statistical analysis. ' Ultimately, deaths from 1969 to 1984 were added In order to Improve the reliability of the results. This extension necessitated additional Interviews and exposure assessments. The statistical analysis process consumed several years as the team (headed by Dr. Greenland) developed new methodologies to evaluate a data base significantly more complex than that of most other studies and then subjected the data to an exhaustive series of statistical analyses. In addition, the Scientific Advisory Panel, while making constructive contributions In Its role as advisor, suggested some additional analyses that consumed more time. With respect to the time required, Dr. Wegman noted 'If at any time during the Study I had evidence of an Important finding revealed by the data analysis or new Information of a current risk that could have been controlled, then 1 would have pushed for Immediate, appropriate action.' HONS 020615 8- I. Whit are PCBs? PC8s art a class of compounds that do not conduct electricity, do not catch flra and do not avaporatt or break down easily. For these reasons, PCBs were considered to be "wonder chemicals" that added to the safety of equipment where fire hazard was a concern. They were manufactured In the U.S. from 1929 until 1977, and sold as mixtures for use in a variety of applications, both electrical and non-electrical - for example, In transformers and as hydraulic fluids. GE was one of'about 300 U.S. companies that used PCBs, which were banned In 1977. II. EPIDEMIOLOGY A. What Is epidemiology? Epidemiology Is the science of the occurrence of disease In human populations. Historically, epidemiology evolved while studying great epidemic diseases such as cholera and typhoid. For example, In 1854, John Snow a practicing physician'In London - used simple observations to show that water supply was responsible for the cholera outbreak. Ourlng the last few decades, epidemiologists began focusing on cancer with some remarkible success. For example, epidemiologic studies have contributed greatly to understanding the link between cigarette smoking and lung cancer. Hany epidemiologic studies have demonstrated the association between workplace exposures to certain chemical substances and some kinds of adverse health effects. B. What are the limitations of epidemiology? Epidemiologists often face many practical problems In carrying out their studies. Foremost, the biological process Involved In cancer Is not well understood; this creates difficulty In selecting possible causal factors to study as well as In Interpreting the results. Furthermore, data available to the epidemiologist, such as on death certificates, Is often Incomplete. This Is also true with exposure data when It needs to be reconstructed from memory due to the lack of historical workplace exposure data. Finally, statistical methodology Is often not available for analyzing certain types of complex sets of data. HONS 026616 9 Epidemiologic results can be Inconclusive (unable to establish either the presence 'or absence of a cause/effect relationship) for a variety of reasons. For example, the researcher may not have been able to Identify and study all the potential causal factors for the disease. The results may not seem biologically plausible or consistent with other studies. Also, sample size may be too small to adequately Investigate the relationship of exposures to a specific cancer site. Regardless of what the findings are In any single study, they need.to be confirmed In multiple studies In different populations to be considered conclusive. C. What types of epidemiologic studies were performed? Initially, Or. Wegman performed a proportional mortality study. In this study, the percentage of deceased employees dying from specific causes was compared to the percentage expected based on the U.S. general population. The primary purpose was to Identify cause-specific deaths that might Indicate the need for further In-depth evaluation through other types of studies. A proportional mortality study Is generally considered as screening or exploratory In nature but Is not conclusive. Based upon the findings of the proportional mortality study and other Interim analyses, Drs. Wegman and Greenland conducted a series of case-control studies. They began by Identifying a group of deaths from cancer (cases) and a group of deaths from other causes (controls). Then for each Individual In the two groups, they compared data on exposure, taking Into consideration other factors. The purpose was to determine whether there Is an association between exposure and disease. If cases seemed to have had a greater exposure history than the controls (more Intense or lengthy exposure to the chemical being studied), one could conclude there to be an "association" between the exposure and the disease. However, such an association does not demonstrate causality. III. DATA COLLECTION AND ANALYSIS A. How was the Study population assembled? For this Study, the 6E Insurance carrier provided death certificates on active and retired employees. The following were the criteria for Inclusion In the Study: Died between January 1, 1969 and December 31, 1984 (no pension records were available for employees who had died before 1969). MONS 028617 10- Met certain vesting requirements (these varied over time, but generally required 10 to 15 years employment with GE). The final Study, therefore, totaled 1,714 subjects, which Included 512 cases and 1,202 controls. B. How were the eases and controls defined? Cases are study subjects who died of all types of cancer. Controls are study subjects who died of conditions other than cancer, excluding diseases of the digestive system, genlto-urlnary diseases, of the blood and blood-forming organs, mental disorders and Ill-defined conditions. C. How was the exposure data compiled? The objectives of the exposure assessment were to: 1) Identify the significant exposures; 2} Choose an appropriate categorical rating scale; 3) Rate work history jobs for the selected exposures; 4) Check ratings to the extent possible. The team made a series of week-long plant walk-throughs. During these walk-throughs, they reviewed current operations, and to the extent available, past records on production, chemical use and purchasing, industrial hygiene measurements, and building maps. In addition, they Interviewed 18 plant employees having average length of employment of 38 years. They were selected on the basis of their familiarity with a range of operations. Interviews were aimed at obtaining the following Information: all major and minor chemical and metallurgical exposures within an operation and thair control technology; history of location of the operation (many operations changed location over the years). They also examined work histories for each subject In the Study to determine Job histories and titles that Is, the different assignments etch person held In various buildings over the course of his employment. Ultimately, more than 7,000 separate job titles In various buildings were rated for chemical exposure. 8tsed on these data, the team then determined exposure levels (relative Intensity of exposure) for each of the seven substances. These levels were used later to calculate person-years of exposure. For example, If an employee worked for 10 years at benzene exposure level 1 and 15 years at PCS exposure level 3, then he would have 10 level 1 benzene-years and IS level 3 PCB-years of exposure. HONS 028618 -11- "Thls was a difficult undertaking because of tha large number of materials used In tha plant, tha many types of operations performed, and the limited number of past records," the authors wrote at tha beginning of the Study's Chapter II, which details the exposure assessment process. D. What other types of data were collected? The team collected data on a number of other factors to analyze statistically thalr affects since tha results of epidemiologic studies can be compromised by other risk factors -- environmental or personal that might contribute to the disease under consideration. Other factors for which data were available Included age and year of death, year of hire, duration of employment, foreign birth, and vesting and employment status five years before death. Unfortunately, data were not available on life-style risk factors, In particular, smoking and alcohol consumption. These factors, therefore, could not be fully taken Into account In the analysis. E. How were data for the cases and controls analyzed? After receiving death and exposure data, Drs. Sander Greenland and Alberto Salvan at UCLA set out to analyze the relative cancer risks of the seven substances. "This was a formidable task, as the exposure assessments were somewhat Involved," Or. Greenland, noted. "There were 512 cases and 14 cancer types to deal with. And we were exploring multiple exposures and multiple disease outcomes. We didn't feel any adequate approach existed prior to our assignment to do Justice to something of this complexity. So we tried out some new Ideas." The Greenland team's first challenge was to convert the job histories and exposure ratings Into an exposure profile for each individual subject. From there, the team could compare cases and controls and make first-cut risk analyses for each chemical under study. In essence, Or. Greenland conducted 98 separate case-control studies, (14 cancer types times 7 exposures). The Initial step was to create a series of two-by-two tables (exposed and unexposad) for each of the 14 cancers under study (those that had 9 or more Individual cases). From these relatively simple tables, he was able to estimate the relative risk (odds ratio) of exposure to each substance for the cases versus the controls. HONS 028619 12- Th results of a first cut of this kind, howovtr, do not consldtr i ring* of factors that art tht key to evaluating risk. Thtst factors Include level and duration of exposure, age, year of death, and exposure to more than one chemical. In order to take Into consideration these variables Or. Greenland then employed various types of sophisticated statistical analysts. The product of these grocedures and analyses Is three binders full of tables from which rs. Wtgman and Greenland drew their conclusions and wrote their final report. HONS 028620 GE SUMMARY OF KEY POINTS ON THE CASE-CONTROL STUOY OF CANCER MORTALITY AT GE PITTSFIELD January 24. 1990 STUOT SUMMARY The Study was designed to detact cancer risks - during the period. 1969-1984 -- In the GE Pittsfield work force from workplace expo sures to chemicals. The Study did not identify any definitive association of the GE Pittsfield work environment with excess cancer risk. Specifically, the Study did not establish that.PCSs caused an in crease In cancer at GE Pittsfield. The results are consistent with the seven other studies of PCB-exposed workers worldwide. The authors found several statistical associations between workplace exposures and cancer risks, but. as noted, none were concluded to cause cancer deaths. The authors further state: "Be cause we chose a liberal criterion to screen multiple associations. It Is probable that most (and possibly all).. .do mil represent any real exposure effects.' (Emphasis Is the author's.) The strongest statistical association found was that of lung cancer and exposure to chemicals used In the production of GENAL (which has not been produced since 1983). There was no association between cancer risk and theiute of GENAL. Several weaker statistical asso ciations were Investigated and Interpreted as 'highly Inconclu sive." CURRENT CONDITIONS The results of the Study do not Indicate any known risk of cancer related to current workplace exposures at GE Pittsfield. Of the seven groups of chemicals studied, only two -- solvents and machine fluids -- are currently used at GE Pittsfield. These chemi cals are also commonly used throughout Industry for the machining and cleaning of metal parts. Employees are protected with training regarding the Identification of chemical hazards. Including safe work practices and emergency procedures. Industrial hygiene programs. Including monitoring of certain work areas and employees, are In place to protect the health and safety of the GE work force. HONS 028621 -z- Records of employee exposures to chemicals are maintained in accor dance with guidelines of the U.S. Occupational Safety and Health Agency (OSHA). HISTORY OF THE STUDY* Following preliminary analyses of mortality at GE Pittsfield by Or. David H. Wegman, then with Harvard University, the Study was initi ated In 1981 with GE as sponsor. Or. Wegman, now with the Univer sity of Lowell, was retained as principal Investigator. In the Study, exposure histories of workers who died of cancer (cases) were compared to exposure histories of workers who died of other causes (controls). If cases were found to have greater expo sures to chemicals than the controls, then the statistical associa tion was further Investigated to determine Its significance. Initially the Study period was 1969-1979, but this was extended through 1984 because the population was Insufficient for In-depth analysis. The Study, In Its final form, began In 1985 when Dr. Sander Green land of UCLA Joined the Study team as co-principal Investigator. Independent review of the Study has been provided by a distin guished panel of academic epidemiological experts (see attached list). GE RESPONSES GE will communicate the results of the Study to all GE Pittsfield employees, retirees, union representatives, and the local medical community. All GE Pittsfield employees and retirees will be given an opportu nity to have their questions about the Study answered by a selected medical professional through a toll-free number: 1-800-4GE-3290. Appropriate medical consultations will be available to employees and retirees who have worked In the former GENAL operations. GE's actions -- both current programs and planned measures -- will adequately serve both past and present workers. Because of these GE actions and because no further significant data on past Job-spe cific exposure are available, GE currently has no plan to conduct any further studies. * See Background Information Document. January 24, 1990 HONS 028622 SCIENTIFIC ADVISORY PANEL for THE CASE-CONTROL STUDY OF CANCER MORTALITY AT GE PITTSFIELD Members of the Scientific Advisory Panel are: Leon Gordls, H.O., Dr. P.H. (Chelr) Professor end Chairmen Department.of Epidemiology The Johns Hopkins University Baltimore, HD e Genevieve Matanoskl, H.O., Dr. P.H. (Asstst. Chair) Professor of Epidemiology Department of Epidemiology The Johns Hopkins University Baltimore, HD e Michael Garin, Ph.D. Associate Professor Occupational and Environmental Health University of Montreal Montreal, Canada e Jack Mandel, M.P.H., Ph.O. Associate Professor School of Public Health University of Minnesota Minneapolis, MN e Howard Rockette, Ph.D. Professor of Biostatistics School of Public Health University of Pittsburgh Pittsburgh, PA e James Schlesselraan, Ph.D. Professor of Biostatistics Uniformed Services University of Health Sciences Bethesda, MO HONS 028623 David Howe Wegman, M.D. Department of Work Environment University of Lowell Lowell, MA 01854 Education Swarthmore College Harvard Medical School Degree BA MD. l MS Year 1963 1966 1972 Field History Medical Occupational Health Professional Experience 1987> Professor Sc Head Present Department of Work Environment, College of Engineering, University of Lowall, Lowell, MA. 1983-87 Professor & Director Division of Environmental and Occupational Health Services, UCLA School of Public Health. 1977-83 Assoc. Professor Harvard School of Public Health, Boston. Occupational Health 1972-77 Assistant Professor Harvard School of Public Health, Boston. of Occupational Health 1972-78 Occupational Hyg. Physician 1969-71 Director 1967-69 Medical Epidemiologist Division of Occupational Hygiene, Mass. Dept, of Labor and Industries, Boston. Urban Planning Aid, Inc., Industrial Health and Safety Project, Cambridge, MA National Communicable Disease Center, USPHS, New York City Health Department. 196667 Medical intern Cleveland Metropolitan General Hospital. Adjunct Appointments 1988- Adjunct Professor Department of Family and Community Medicine, University of Massachusetts Medical School, Worcester, MA 19871984- 1983- Visiting Professor Visiting Professor Lecturer UCLA School of Public Health National Board of Occupational Safety and Health, Sweden (May). Harvard School of Public Health, Boston. HONS 028624 Sander Greenland, Dr. P.H. Division of Epidemiology School of Public Health UCLA Lot Angeles, CA 90024-1772 Education U. of California, Berkeley U. of California, Berkeley UCLA UCLA Degree BA MA MS Dr. P.H Year 1972 1973 1976 1978 Field Math, Psychology Math Epidemiology Epidemiology Math Professional Experience 1989- Professor of Present Epidemiology UCLA School of Public Health 1984-89 Associate Professor UCLA School of Public Health of Epidemiology 1985 Visiting Associate Professor of Biostatistics School ofPublic Health University of Sydney 1980-84 Assistant Professor UCLA School of Public Health of Epidemiology 1979 Assistant Research UCLA Public Health Epidemiologist 1979 Assistant Professor Harvard School of Public Health of Biostatistics 1978 Assistant Professor California State University Northridge 1975-78 Statistician School of Nursing & School of Public Health UCLA 1973-74 Research Assistant Department of Biostatistics School of Public Health University of California, Berkeley HONS 028625 Leon Gordis, MJD. Johns Hopkins University Baltimore, MD 21205 Education Degree Columbia University BA Jewish Theological Seminary BA State University of NY, Downstate Medical Center MD. Johns Hopkins University School of Hygiene and Public Hialth MA Johns Hopkins University School of Hygiene and Public Health (Department of Chronic Diseases) PhD. Year Field 1954 1964 Hebrew Literature 1953 Medicine 1965 Public Health 1968 Public Health Professional Experience Current Prof. & Chairman 1971-73 Director 1971 Member 1969-70 Visiting Professor 1963-72 Assistant Professor 1968-79 Chief 1966-69 Associate Director 1967-69 Director 196667 Asst. Medical Dir. 196165 Field Officer Department of Epidemiology, Johns Hopkins School of Hygiene and Public Health, Joint Appointment in Pediatrics, John* Hopkins School of Medicine; Director Clinical Epidemiology, Johns Hopkins School of Medians Epidemiology and Statistics Cantsr, Maryland Regional Medical Program Israel Government .Commission on Mercury Hazards Dspartmsnt of Medical Ecology, Hebrew University Hadassah Medical School Pediatrics - John Hopkins School of Msdicins Department of Community Medicine Sinai, Hospital of Baltimore Pediatric Family Clinic Sinai Hospital of Baltimore Adolescent Center, Sinai Hospital of Baltimore Ambulatory Care, Sinai Hospital of Baltimora Heart Disease Control Program, Division of Chronic Diseases, U.S. Public Health Sarvict MQNS 028626 Genevieve M. Matanoski, M.D., Dr. P.H. John Hopldn* University of Hygiene and Public Health Department of Epidemiology Baltimore, MD 21205 Education Radcllffe College John Hopkins School of Medicine John Hopkins School of Medicine John Hopkins School of Medicine Degree BA MD. MJP.H. Dr.PJL Year 1951 1955 1962 1965 Field Chemistry Medicine Epidemiology Epidemiology Professional Experience 1978- Program Director Present Occupational and Environmental Epidemiology, John Hopkins School of Hygiene and Public Health, Baltimore, MD 1976- Professor Present Epidemiology, Johns Hopkins School of Hygiene and Public Health, Baltimore, MD 1970- Associate Professor University of Maryland Dental School, Present Baltimore, MD 1970- AssociateProfessor Preventive Medians, University of Present Maryland, Baltimore, MD 1969-76 Associate Professor Epidemiology, John Hopkins School of Hygiene & Public Health, Baltimore, MD 1964-69 Assistant Professor Epidemiology, John Hopkins School of Hygiene & Public Health, Baltimore, MD 195964 Instructor Epidemiology, John Hopkins School of Hygiene & Public Health, Baltimore, MD 1957- Pediatrician Out-Patient Department, Johns Hopkins Hospital, Baltimore, MD HONS 028627 Michel Gerin, Ph.D. Dept. of Occupational & Environmental Health Medical Faculty Univereity of Montreal Montreal, Quebec Education Degree Senior School of Chemistry & Industry ofLyon (France) Engineering McGill University, Montreal PhD. Year 1970 1976 Field Chemistry Chemistry Profeaeional Experience Current -Assoc. Professor Dept, of Occupational & Environmental Health, University ofMontreal. 1986-57 Visiting Scientist International Agency for Research on Cancer, Lyon, France 1980-86 Asst Professor Dept of Occupational & Environmental Health, Univereity of Montreal. 1989- Invited Professor Research Center for Epidemiology & Preventive Medicine, Armand Frappier Institute, Laval, Quebec 1979-80 Research Asst. Reserach Center for Epidemiology & Preventive Medicine, Armand Frappier Institute, Laval, Quebec. HONS 028628 Jack S. Mandel, Ph-D., M.P.H Division of Environmental & Occupational Health School of Public Health University of Minnesota Minneapolis, MN Education University of Minnesota University of Minnesota University of Manitoba Degree PhD M.P.H. BSc Year 1981 1973 1966 Field Epidemiology Epidemiology Mathematics Professional Experience 1989- Professor Division of Environmental & Occupational Health, School of Public Health, University of Minnesota 1989- Director of Graduate Study Division of Environmental & Occupational Health, School ofPublic Health, University of Minnesota 1983-88 Instructor Midwest Occupational Health Institute, University of Minnesota 1984-89 Associate Professor Division ofEnvironmental Health, School of Public Health, University of Minnesota 1980-85 Instructor Graduate Summer Session in Epidemiology, University of Minnesota 1979-83 Assistant Director Division of Epidemiology, School of Public Health, University of Minnesota 1982-84 Director of Graduate Division of Epidemiology, University of Study Minnesota 1981-84 Associate Professor Division of Epidemiology, School of Public Health, University of Minnesota 1975-80 Instructor Division of Epidemiology, School of Public Health, University of Minnesota 1974-75 Assistant Profeseor Department of Epidemiology and Community Medicine, University of Ottawa HONS 02B629 Howard E. Rockette, Ph.D. Qraduat* School of Public Health Department of Biostatistics University of Pittsburgh Pittsburgh, PA 15261 Education Degree Franklin & Marshall College BA Pennsylvania State University MA Pennsylvania State University PhD. Year 1965 1972 1972 Field Mathematics & Physics Mathematics Statistics Professional Experience 1987- 88 Director Data Center for the Institute for International Studies in Natural Family Planning 1988- Professor Graduate School of Public Health Department of Biostatistics, University of Pittsburgh 1983- Co-Director National Surgical Adjuvant Breast Protocol, Statistical Center, Pittsburgh 1981- Director Statistical Unit, Otitis Media Center, Children's Hospital, Pittsburgh 1988- Sr. Research Present. Scientist Institute for International Studies in Natural Family Planning 1978-83 Associate Professor University of Pittsburgh Graduate School of Public Health, Department of Biostatistics 1972-78 Research Assistant University of Pittsburgh Graduate School of Professor Public Health HONS 026630 James J. Schlesselman, PhD. 7. Edward Hsbart School of Medians Uniformed Services University of Health Sciences Bethesda, MD 20814 Education Michigan State Princeton Princeton Princeton Degree BS MA MA PhD. Year 1966 1968 1968 1971 Field Psychology Psychology Statistics Statistics 1985- Professor Prasant Department of Preventive Medicine & Biometrics, F. Edward Hebert School of Medicine, Uniformed Services University of the Health Sciences (USUHS). 1981-86 Associate Professor Department of Preventive Medicine & Biometrics, F. Edward Herbert School of Medicine, (USUHS). 197681 Chief Biometry Branch, National Institute of Child Health & Human Development (NICHD). 1973-76 Mathematical Statistican Biometry Branch, NICHD. 1971-73 Senior StaffFellow Biometry Branch, NICHD. 1969-71 Commissioned Officer U.S. Public Health Service, Epidemiology & Biometry Branch, NICHD. MONS 028631 V- GE STATEMENT ON CANCER MORTALITY STUDY `Dr. Negman's Study Is completed. md th news is reassur ing. Despite th* large nunibtr of eaneer typos .that wort Investlgotod. no ooflnltlvo sssoclitlon batwocn workploco chemical expo sures at GE Pittsfield end excess cancer risk was established.' `Specifically, and Importantly. the Study found that past worker exposure to PCBs did not Increase the risk of cancer. This finding Is consistent with seven other studies worldwide. *GE will promptly cooeunlcite the Study results to all Interested pertlas -- employees, retirees, union representatives and the local medical community. 6E will also have medical profes sionals available to answer employee and retiree questions about the Study. *We are grateful to Dr. Hegman, hlsco-lnvestlgator, Dr. Sander Greenland, and their team, as well at to the external Scientific Advisory Panel, for their tremendous effort In complet ing this complex Study.' Herb Coulter, GE Pittsfield, ht media briefing, January 24, 1990 HONS 028632