Document Jr0D2oOegGzEgRDVdV2vDaG9Z

PLAINTIFF'S EXHIBIT Use of the Canadian Mortality Data Base for Epidemiological Follow-up Martha e. smith, m.sc. and Howard b. newcombe, p(..d ** Epidemiological follow-up of large cohorts exposed to potentially harmful agents and circumstances, has been facilitated in Canada by the development ofa consolidatedfile of information on all deaths in the country back to 1950. This Mortality Data Base is designed to be conve niently searched by computer, and to yield both dates and underlying causes ofdeaths among members ofa study population. Current andfuture major uses are described. Most ofthe studies relate to groups of persons numbering m the tens to hundreds of thousands. The studies are con cerned with the long-term conse quences of various occupations, medical treatments and diagnostic procedures, and circumstances of other kinds which may contribute to deaths from cancer, cardiovascular disease and other causes. The indus trial groups include uranium miners, nickel workers, hard rock miners, and radiation workers, plus some exposed to asbestos, fibreglass, vinyl chloride and formaldehyde vapour. itii sutisf.es and D'scjnc registries Section. Health Ui'iMon. Siat.'iics Canada. Tunnel's Pasture. Ottawa. Omariu < 1A )Z* Reprint rcu jest* (<t. Miss M Smiin. Head, t^ccupauonai arid fc-.vironnesui Heaiti l mi V nai sumtics and Dt** <j>c -ectmfl :*ih Klnnr. R H C.sats Building. sijii:ka jnjda rurines % I'lNture. Ottawa. Ontario * \"Zf The pi r t prosed jt c inu*c > iJ the author* jmJ do not ncce\an % .-ti'cv.1. in.-** u S;jii*ic> Canada that might otherwise bej'udged imprac The medical patients include persons tical. We will emphasize how the Mor exposed repeatedly to fuoroscopic tality Data Base has facilitated the con X-rays, and others treated with a duct of occupational mortality studies potentially carcinogenic drug, isoni- in Canada. azid. Similarfollow-up isplant,edfor The special features of the Canadian participants in a large nutritional Mortality Data Base are that (a) it con survey, and is in progress for large tains, from each death registration, not numbers of mothers who had their only ample personal identifying infor first offspring at various ages, there mation. but also the coded cause of being special interest, respectively, in death, (b) it exists both as machine- cardiovascular disease and breast readable magnetic tape for computer cancer. Demandsfor such uses ofthe searching, and as compact microfiche Mortality Data Base have increased and microfilm listings for manual rapidly since its establishment and searching, and (c) it is sequenced and show no sign of diminishing. consolidated (over all Canadian pro vinces and'over spans of years) to facili- rate such searching. --| The design and functions of a Nothing comparable to this facility CD ".Mortality Data Base", containing exists in any other major country. The -- information from all Canadian deatrhecently established United StatesCT* registrations back to 1950. have already National Death Index, for example, been described. ' *The value of '.his facil goes back only to 1979; it also lacks the^"' ity is best judged in terms of its actual, coded cause of death, so that manual-*"^ planned, and proposed uses for follow clerical checking of the death registra-""- ing "at risk" groups of various sorts to tions forms will be required, normally at determine their mortality experience the various state registry offices.4 over the years. We describe here the Although most developed countries use number anc nature of such studies, the data processing for published death sizes of the groups under investigation, records and annual mortality statistics, and the use made of various personal the idea of manipulating the data into a identifiers. Until recently, truly large- consolidated form for epidemiological scale follow-up of this kind has oeen searching on a national scale is new. virtually impossible except under spe Without such consolidation it is diffi cial circumstances. It is hoped that cult or impossible to search for possible greater lamiliurity with what has now death registrations pertaining to per become :easibie wtil help to clear the sons whose dead-or-alive status is way for needed studies of delayed risk unknown, or for whom the year and Canadian Journal of Public Health Vol. 73, January/February 1982 39 rtfgistrunon area of ihe death are records carrying the same surname two decades to a high degree of sophis- unknown. Where the numbers in a code. Decisions concerning acceptable ucaiion. In practice the computer has study population are large, what isdilfi- matches, and calculations of the proba been found to be more accurate in link eult for even the individual ease, may bility that the correct death record has ing handicap and death records of become impossible for the group as a been found, are based on the full identi children back to the relevant birth regis whole due to the cost and labour fying information common to the tration than clerical staff carry ing out invoked. record which initiated the search and the same searches routinely.'-' Recently Follow-up using the Canadian Mor the death record with which it is being a generalized record linkage system has tality Data Base is carried out at Statis compared. The personal identifiers been developed by Statistics Canada tics Canada, a central statistical agency provided for on the Mortality Data and the Epidemiology Unit of the whose rr.andjte is to compile, analyse Base magnetic tapes are indicated in National Cancer Institute of Canada'* and publish statistical information.1 * 3 T4a*b*le II. and those available on the (NCIC) and is being used to carry out The work is usually done on a cost- records initiating the searches for the investigations initiated since 1979. recovery contract basis. Table I indi various studies are shown in Table III. cates the scale and demand for the work. Nature of the searching process The searches are mostly carried out The linkage technique, including the logic and practice of the decision making procedure and the means by which the degree of assurance of a cor rect link may be calculated, has been Completed searches. Five major searching operations to obtain Canadian death registration data have been completed. with the records of the Mortality Data developed over a period of more than Infant death-birth linkages, for child - Base first split by sex code and then arrayed in a sequence based on the sur TABLE II name, i.e. in order of a phonetically Personal Identifiers Provided for on the Mortality Data Base Tapes coded form of the name, with finer sub division on the basis of the full alpha betic spelling followed by the first given nameand birth year. The purpose of the phonetic coding of the surname (the "New York State Identification and Intelligence S> stem" known as N YSI IS)*is to get around the effects of common spelling variations which could other wise interfere with finding the relevant death records. Where necessary, com parisons can be made with all death Identifier* Rtlatinf to rh* Dfurawd l. Surname (current) 2. Goen names (first and second) } 8trth date tvear. month, da*) s Birth province or foreign t-ouniry (coded) 5. Se\. mamal status, racial origin Reljtrnt to tor 6. Birth surnames (father, mother, spouse*) 7. Initials (father, mother, spouse-) S. Brth province or lortign country (lather, mother) Rrfattnt to the vrni 9. Place oi event 10 Place of residence tcodrd) ! 1. Date ol death ftccu jrc rjr;;> joiiadte Number of characters 10 9* 7 7 2 4 10 10 - 10 2* 2 * 2 l 2 10 5 6 TABLE I Computer Searches of the "Mortality Data Base" Study______________________________________ Cotnplettuj 1. Infant death-btrth linkages 1197 I births) 2. Omano uranium miners 3 isoniard and cancer in tuberculosis patients 4 Canadian labour force !0r? sample 3 Fluoroscopy and cancer in (ubcrcuto>i pattern** O*rr?tu b. 1NCO nickel workers ?. Filcjnbndkie nickel workers 8 Eldorado urjniun workers U Ontario 'nomina; rolls'* (miners and others) 10. Nutrition Cjr.aUa *ut\cy participant* 11 Ontario mortician* e\po-ed to lormaldehyde 12 CNR railway workers 13 Brea*t ejr.ccr and j*e at Ifrst htrtft 14 A.berta Cancer Rrg:ory Jejih clearance 1195.'?!<) 15. Ontario Cancer Incidence Reporting system lb Newfoundland iluorspat miners ( irt.Vr l 7 8 *A*E1C1I 12ra1J3u1t4io1n5 workers J.nd others 15 Intjr.t death-birth ankace*. annual Number of individuals 6.000 16.000 6a.000 1 700.000 f 100.000 1 62.000 1 11.500 1 16.000 57.000 20.UO<) 1.500 la.ooo j00.000 1*5.000 125.000 2.000 1 1.000 5JAM) Agency Statistics Canada Ontario Rovat Commission National Cancer Institute of Canada McMaster l ni*ers:tv Eldorado Nuclear Limned Ontario Ministrv of Labour and WCB Health and Welfare Canada Chemical Industrv institute of Toucologv National Cancer institute of Canada Universiiv of Wnirrn Ontario Alta Provincial Cancer Hospital Board Ontario Cancer Treatment and Research Founu Atomic Energy Control Board and H*AC Atomic Eneres of Canada Limited Health and VVeltare Canada ( H^'Cl ZLQ`i9 I 01S 40 Canadian Journal of Public Health VoL 73 TABLE III Identifiers Available for Linkage With the Death Records L>nkagc studs Sur- Given namcis) namcui Birth date Ptrsottaf (Jvntifi tn% trt/owation iu?J tn tht linkages Birth Rest- Mother's Parent prov dencc mad. Parent birth entry provname initials place Spouse's given names Soc. Ins. No. See Footnotes Inlint death -- birth |19'I) (t0) w_ 2. Isor.iazid and cancer 3. Canadian labour force a F!uoro>copv studv 5. INCO nickel workers * (yr) (entrv) - - - - - - * - *--- - * * - - - * * (*) - * - - - - - - (> b. Fjlconbndgc nickel workers ___ (*) 7. Eldorado uranium workers 3. Ontario "nominal rolls" 9. Nutrition Canada survev * <-> _ (*) <*) () _ (*) (-> - - - - - (*> - (*) - - - (-) - 10. Ontario morticians 11. CNR railuav workers --- (*) * - - - - * - -- - - - 12. Breast cancer studs (tuo) (mil) lag') - - - - (inti - 12. Alberta Cancer Regtstrs - * * - - - - - 14 One. Cancer Incidence Reporting System * _ __ - - 15. *ECL radiation workers - * * * * (-) (*) <*> 1*) * 16. Infant death -- birth lannual) 1". Fluorspar miners (tuo) - -- - - (yr) * - - - -- a b c d d e r g h i i i k a - ,Nc?/e. Parentheses indicate these identifiers were frequently absent but were used wher n ji(able to assist in the computer match or for manual resolution of doubtful cases. Control The success wuh which the Mortality Data Base mav be searched depends upon the amount of personal identifying information common to the death records and the recoros that inmate the searches. This is true whether the searching .s done manually or by computer. Where some ot this tnt or matron may be incorrectly entered into either sort of record, and especially into those that initiate the searchers, a redundancy of identifiers will serve to increase the assurance of a correct link in spite of the discrepancies created by the incorrect information. a. The two surnames are the birth surnames of the father and mother of the new born child, as reported on the child's birth registration and subsequently on its death registration b. The identifiers called lor on the tuberculosis admission forms included residence, date of birth, marital status, sex. country of btrth, year of arrival m Canada, citizenship, race, occupation and industry. c. The identifiers on the Social Insurance Number Index tapes include a second given name if there was one. and an alternative surname (e.g. married as distinct from maiden surname) plus the mother's maiden surname. d. Full first forenames are present on all 73.000 1NCO and Falconbndgc records, and the Social Insurance Number is on the latter half of the file. e \bout flO per cent of the records contain at least one given name plus the full birth date. Jusi o\er a quarter ot these past workers had Social Insurance Numoers. Parentneses indicate other identifiers that are frequently absent but are used when available. f. The 57.000 searches wul be based on names and birth dates. However, the presence of Social Insurance Numbers for 33,000 of the workers will make ii technically leastble to test for the increased accuracy of searching when full identifying information is extracted from the S.l.N. index prior to searching. g. The personal identifiers recorded during the survey should be ample for accurate searching. Moreover, it manual resolution of doubtful linkages is required, involv mg comparisons with the microfilm images of death registrations, added precision may be gained through use of particulars on the survey forms to do with city or place of birth, exact address if unchanged at the time of death, and the identifiers for parents, spouses and children when entered together on t.ne lis: of household members. h The Unking int or (ration from 1928 onw-ard consists of the name in full, date of birth and place of employment. From 1933 onward, place of birth is also available, and from 1958 the fathers name. i. Identifiers available on the birth records, with which to search for records of the subsequent deaths of the mothers, include two surnames (married and maiden}, initials, province or country in which the mother was born, age at the time of the birth (versus birth date as given on the death record), and place oi birth oi the child (versus ihe place of the mother's death). j. The degree of success in linking the cancer records to the Mortality Data Base depends heavily on the amounts of personal identifying information provided for on. and actually entered into, the provincial cancer registration forms. Although there has been improvement in recent years. Canadian province* still vary widely m this respect. The lull birth date, for example, which is especially important I or searching the Mortality Data Base, is present on less truin ''O per cent ol the 1975 cancer registrations across Canada.although some provinces report it quite consistently. Thus the success rate is likely to be high for some provinces, low for others, and tn genera! better for recent than lor earlier years. k. Tne required personal identifying information for most past employees may be obtained trorr. application forms for employment, but in recent years such terms have been changed by law to remote questions pertaining to birth date, btrth place, the identities ol parents.and even sex. as a precaution against discriminatory hiring practices. Thus special action is required to obtain adequate personal identifiers from current employees after they have been hired. This is viewed as ar. ongoing process to include future employees on a continuing basis. ren born in 1971/were carried oui before the Mortality Data Base had been assembled in us present form. In this eariv >cudy. irlant death registra tions tor 19`l and 1972 were linked to births in I9~l. For 96*7 of the infant deaths, a matching Canadian birth reg istration was found bv computer lin kage Inijnts who die very shortly alter birth commonly lack specific forenames on their birth and or death record, and these account for part of the records which failed to match. The major objec tive of this investigation was to explore the technical feasibility of carrying out this t\pe of linkage by computer on a national scale. The study conducted in 1975 was successful, and it was then decided to initiate a plan to develop the historical Mortality Data Base file in order to enhance the usefulness of the existing data sources. From the linked death-birth "linkage" file (in conjunc tion with the stillbirth file for the same year), perinatal and post-perinatal mor tality statistics have been extracted, broken down by maternal age. parity. ELSn9I 0IS Januan/ February 198 41 gestation and birth weight. Such data are unobtainable from the individual fiies prior to linkage and may be used to derive indicators of inadequate obstet ric and pediatric care. The linked files are also useful for monitoring the impact of the maternal (or paternal) environment or occupation on repro ductive outcomes, such as fetal and infant deaths, and the births of children with congenital malformations or deviant birth weights. honiazid and earn er in tuberculosis patients has been studied by the Epidemiology Unit of NC1C using records from the Canadian National Tuberculosis Register of admissions to treatment institutions over the years 1952-60. These records have been linked to death registrations n the Canadian Mortality Data Base in the years 1952-75 and to the National Cancer Incidence records for 1969-73. Isoniazid is known to cause lung tum ors, leukemias, and lymphomas in mice. In man there were some indications that it might cause bladder and lung cancer10, although most studies in the past hav e yielded negative results. There was thus a need to follow a larger treated population, over a longer period, together with a comparable untreated control group. A total of I08.A56 records from the Tuberculosis Register yielded personal identifying information and histories of drug treatment (or its absence) for 64.037 individuals of whom 41,226 received isoniazid and 22.81 1 did not. A search of the Mortality Data Base yielded 12.398 matching death registrations. This is the most advanced of the investi gations using that facility, in that the results have been analysed ar.d pub lished.11 The data are interpreted as constituting a large body of negative evidence with respect to the carcinoge nicity of isoniazid in man. A Canadian labor force 10 per cent sample provided the base for the large study m which the mortality experience of various occupational groups is com pared. Uneinploy meat insurance "Book Renewal Cards" for the years 1965-69 and for 1971 were used These contain coded information on occupation, industry and province of work, together with the individual's surname and Social Insurance (S.I.) Number. Such cards were filled out each year by con tributors to unemployment insurance whose S.I. Numbers ended in digit 4; from 1971 onward when white-collar workers joined the scheme, the sample size was reduced to 5 per cent (those with S.I. Numbers in which the digit 4 was preceded by an odd number). Because most death registrations lack the S.I. Numbers, an intermediate step was necessary prior to searching the Mortality Data Base. This consisted of extracting, by computer, from the cen tral index of S.I. Numbers, personal identifiers of the kinds contained in the death records (see Tables II and 111). Full details of the procedures to be used in such an investigation had alreadybeen published12 and the actual study was subsequently organized by the National Cancer Institute of Canada in collaboration with Statistics Cana da.A total of some 3.000.000 Unem ployment Insurance records, on mag netic tape, were merged to provide work information for about 700.000 persons. A search of the Mortality Data Base found 30.000 deaths among study group members to the end of 1973. and with subsequent searches this number will, of course, grow. Resuits from the prelimi nary analysis indicate elevated risks of cancer and other fatalities in a surpris ingly wide range of occupations. The implications of these findings are still unclear. Fluoroscopy and cancer in tubercu losis patterns has been studied, using the Mortality Data Base, to determine the relationship between radiation dose and the magnitude of the subsequent risk of cancer. Of particular interest were cancer of the breast, lung and thyroid, and leukemia. An association between breast cancer and fluoroscopy had already been demonstrated through manual \tudie> of limned size in Nova Scotia and Ontario"'' and detailed plans had been proposed for an auto mated study ofa similar nature covering the whole of Canada." The National Cancer Institute of Canada obtained institutional records of patients treated in Canada for tuberculosis during the !9J0's to early 1950's when artificial pneumothorax was being practiced and fluoroscopy was used to determine the degree of collapse.1'':o NC1C has also studied the organ doses that would have been received from the older fluoros copy machines.21 Some 137,000 treat ment institution records have provided the personal identifying information, and the histories of repeated fluorosco pies. for between 90.000 and 100.000 individual patients. These patient histo ries have now been linked with the rele vant death registrations in the MortalityData Base, using the identifiers availa ble on the treatment institution records (Table HI). In the analysis, information will be sought, not only on the doseeffect relationships, but also the latent periods forcancer induction, and possi ble changes in sensitivity to the carcino genic effects of radiation with age. The latter is of special relevance to mammo graphy for which there is believed to be a reduced risk for older women.22 Searches in progress The studies in progress have to do with nickel workers, uranium miners, hardrock miners, an assortment of industrial workers exposed to various hazardous substances, morticians exposed to formaldehyde, railwayworkers, and participants in a survey of family nutrition in Canada. The effect which the age of a mother at her first birth may have on her subsequent risk of breast cancer is also being investi gated. In addition, a number of provin ces have expressed a wish to "clear" their cancer register files of records per taining to deceased registrants, and a pilot study involving the Alberta pro vincial cancer registry file is being used to test the feasibility of doing this lin kage at a national level. The Ontario registry also uses the Mortality Data Base to supplement their incidence data with Ontartocases not previously ascer tained from other kinds of records. \ickel workers in the Sudbury and STOI 6457b 42 Canadian Journal of Public Health Vol. 73 Port Colborne area arc being lollowed because of evidence suggesting that the various sulphide forms of nickel are carcinogenic.;i In Canada the cancer risk from exposures to the atmosphere in sintering plants continued well into the 1950`s.-`:? Since INCO has pro duced in earlier years up to "5^ of the western world's nickel and still accounts for 40^. its potential contribution to the scientific evidence is vast. The initia tive for the INCO study came from a joint union-management committee on occupational health set up by t\CO Metals Limited and the United Steel workers of America. In 1976 the Occu pational Health Program at Me Master University was commissioned by INCO and the United Steelworkers of Amer ica to conduct a mortality study within the company's Ontario operation.:6 Falconbridge Nickel, which likewise operates in the Sudbury area, has initiated a similar study. The Eldorado uranium workers, w ho are being followed in a similar fashion, include those employed in mining, mil ling. the extraction of radium for medi cal purposes, and the refining of ura nium as a nuclear fuel. A nominal roll has been prepared from personnel records, pay records, pension plan forms and medical forms, representing a total of 16.658 employees. The study has been described in some detail byEldorado Nuclear.To ensure that interpretation of the results of thestudy is conducted independent of the ura nium industry, the Epidemiology Unit of NCIC has been asked to do an analy sis and to make the tindings public. The Ontariu "nominal rolls" of min ers and other workers, were developed for a study sponsored jointly by the Ontario Workmen's Compensation Board (WCB) and the Ministry of Labour. Special interest has focused on assessment of the risk of lung cancer in uranium miners. Preliminary investiga tions had already been carried out using an early version of the Mortality Data Base."1' The present study is more comprehensive and will permit compar isons o! uranium miners with other hard-rock miners for whom the radon daughter exposures were much lower, and also with salt miners for whom there was little or no exposure except to diesel fumes.'-" The Xutrition Canada Sun ev partic ipants comprise some 19.500 Canadians for whom detailed nutritional data were collected in 1970-72. along with infor mation on demographic, social, anthrometric and medical characteris tics (e g. age. sex. income, education, stature, obesity, serum cholesterol, blood pressure). Without knowledge of the subsequent mortality experience of these persons, the implications for health of the various descriptive tabula tions that have so far come out of the survcyu,, would necessarily remain a matter of indirect inference and there fore be less clear. The Ontario morticians are of inter est as a study group because of their exposures to formaldehyde vapour, m concentrations that are believed to average around 0.J to 1.2 parts per mil lion. Use is to be made of the records of licences issued to Ontario embalmers and funeraldirectors. A full description of the study has been prepared by R.J. Lev ine of the Chemical Industrial Insti tute of Toxicology.** Railway workers are of interest as a study group because of earlier indica tions that an excess of bladder cancers may possibly occur among them. The present study cohort consists of all of the male Canadian National Railway pensioners in the period 1965 to 1977. The breast cancer study will com pare the ages of women bearing their first children, as indicated on the British Columbia birth registration tapes, with the causes of their eventual deaths.'' It has been claimed that the breast cancer risk of women, vv ho bear their first child over the age of 35 is three times the risk for women whose first child is born betore age 20. but not all investigations support this conclusion. The present study will be based on more than 848.000 British Columbia birth registra tions in the years 1950-73. The Alberta cancer file death clear ance. using the Mortality Data Base, was proposed by the Alberta Prov incial Cancer Hospital Board. If considered successful, the procedure would almost certainly become routine for the other prov incial cancer registers, and for the centralized National Cancer Incidence Reporting System. Such searches serve also to identify additional cases of cancer not previously ascertained from other sources. Statistics Canada for example, could potentially use the Mor tality Data Base as a source of informa tion with which to update the Ontario Cjncer Incidence file. Fluorspar mining in St. Lawrence. Newfoundland, was carried out from 1933 to 1978. Excess mortality from lung cancers among the miners, appar ent by the late 1950's, was attributed to accumulation of radon and radon daughters in the fluorspar mines.14 and this led to a Royal Commission study.''' The data are currently being updated. Searches currently being negotiated Studies involving searches of the Mortality Data Base which are cur rently being negotiated with Statistics Canada, have to do with (a) a follow-up ofemployees of Atomic Energy of Cana da Limited, including radiation TABLE IV Ontario "Nominal Roll-' Files, Breakdown Occupation 1 Uuniun miners > Non-uranium miners 3. Sail miners i Asoeslos workers 5 Insulation workers tfibreglass) t>. Nickel >;nicr workers 7 Nickel carbon*.\ workers * PoKwnsI chloride workers V Vmsl chloride workers Tout Number ol individuals i e>.s- J J.UJ7 750 4.1SSJ 773 254 ICO :oi in: 56.65) 9LS'i 9 I 0IS January/ February 1982- workers. (b) an annual linkage of infant death records and the appropriate birth records, tor reasons given eariier in connection with the similar study of 1971 births and (c) a study of members of the Canadian Diabetic Association to investigate the use of saccharin and the incidence of bladder cancer. The A ECL radiation worker study relates to some 11,000 past and present employees of Atomic Energy of Canada Limited.'"''1' Details of a possible study of all radiation workers in Canada, including those at AECL. have been published earlier.4' The infant death-birth linkage to be earned out on an annual basis would serve the purposes outlined above for the comparable study pertaining to 1971 births. This would involve records of some 360.000 live births per year and just over 5.000 deaths in the first year of life. Possible uses of the Mortality Data Base in the near future A list of further possible studies is given in Table V, together with the approximate sizes of the population for which searches of the Mortality Data Base might be required. Statistics Cana da has been approached regarding some of these, and some have been the subject of public discussion; for yet others the need for mortality data has seemed obvious enough to justify inclusion in the Table. At least one of the studies listed (i.e. that of Ontario Hydro workers, which has been underway for five years) already monitors mortality occurring in both active and pensioned employees. Follow-up studies of many sorts of population sub-groups are becoming increasingly popular, and those who feel the need to know the mortality experience of such groups will undoubtedly turn to the Mortality Data Base when they learn of its existence and potential. The Sational Dose Registry for radia tion workers deserves special mention in this connection, because if may serve as a model for registries of exposures to other harmful agents with delayed effects, such as asbestos, vinyl chloride, wood dust, benzene, iron oxides, mer cury. lead etc. The data pertaining to the radiation exposures come predomi nantly from la) the national dosimetry service provided for those exposed occupationally to x-rays from medical, dental and industrial radiography, (b) nuclear power generating stations, and (c) uranium mines. The Registry files go back as far as 1945. and relate to some 250.000 workers. These files are cur rently being reorganized to provide more ready access to personal dose his tories.'" *' Such centralization into computerized files represents the only practical means by which lifetime dose histories may be assembled for workers who move from one employer to an other. The feasibility of linking records of the National Dose Registry with those of the Mortality Data Base is being investigated. Other such records, required in Ontario for industrial expo sures to lead, asbestos, silica, mercury, vinyl chloride monomer, isocyanates and noise, could well evolve into similar registries, provided there is sufficient public concern over the possible long term consequences of chronic expo sures. Future potential Forthe moredistant future, use of the Mortality Data Base will continue to depend on the extent of public demands to know the delayed risks associated with conditions in the workplace and in the environments in which wc live. Added to such demands, there may be a continuing interest in the extent to which such killing conditions as cancer and cardiovascular disease can be attributed to yet other circumstances, including diet, food additives, pharma ceuticals. life styles and even the mineral and- or trace element content of drink ing water. It has been repeatedlyemphasized that the most effective way of investigating such problems is not through observations at a point in time, but rather through individual linear follow-up of people on a substantial scale.1*-'4446 As a result of current experience with searchers of the Mortality Data Base, there should in the future be greater assurance that the approach is feasible, not unduly expensive, and that it need present no threat to personal privacy as carried out under the constraints of present federal and provincial legisla tion. and within the walls of an agency which already has custody of the records for the production of other sorts of statistics pertaining to the protection of human health. TABLE V Studies Which Might Make Use of the Mortality Data Base in the Future Study 1. National Dux Reeistrv ol` radiation workers I 1`pilute oi ?Pc Ftdorado Nuclear Studv j Port Hope cohort *tudv - Ontario Hkdro worker* 3. Stuck ot diabetes p KcflDlN J'.OevtO* \iud> ? Manitoba mining jnd >ncltsr worerrs S Icelandic conn-amt*. he!ath>iudv V Si ftcuis health wudi 10 L pdaic ut the Ontario miner'* itudv 11 Oincr provincial cancer registry dejin clearance t 1 *)*} onw jrd i Numoer of indn-id ua is :so.ooo 21.000 30.000 25.000 12.500 l .TOO 30.000 100.000 2.500 58.UOU Agency Hcaith and Welfare Canada Eldorado Nuclear Limited Federal* Provincial Ontario Hvdro National Cancer lnsiiuc ol Canada Stanford Research Initiate Manitoba Cancer Treatment and Research Foundation Ottawa UnixrMiv Workmen* Compeniation Board 9/.9t|9l01S 44 Canadian Journal of Public Health Vol. 73 La poursuite des recherches en eptdemtoiogie sur nombre d'individus exposes aux dangers et aux acci dents au Canada a etcfavorisee par la niise en place d'un systeme consolide de jlches de renseignemenis sur tous les deces survenus au pays depuis 1950. Ces statistiques sur la mortalite sont cortfues pour etre facilement analysees par ordinateur et pour reveler a lafois les dates et les utilisa tions fondamentaies des deces chez les representants d'une population etudiee. Les principals utilisations, presentes et a venir. sont ici decntes. La plupart des etudes portent sur des dizaines et des centaines de milliers d'individus. Elies s'interessent aux consequences a long terme des types de travail, traitements medicaux et methodes de diagnostic, ainsi qu'aux circonsrances de type \arie qui peuvent entrainer le deces par cancer, maladies cardiovasculaires. et autres causes. Les groupes industries comprennent les mineitrs de iuranium. les mineurs au nickel, les mineurs de niineraux durs. les travailleurs exposes aux radiations, ainsi que ceux exposes a I'amiante. au fibre de verre. au chlorure de vinyte et aux vapeurs de formaldehyde. Parmi les patients traites medicalement. on retrouve des gens exposes regulierement aux rayons-x fuorescopiques, et d'autres traites aver des droguespossiblement cancerigenes et a Visoniazide. On en visage d'effectuer un sui vi sembtable chez les participants a de grandes enquetes sur !a nutrition, et on deja commence a le faire chez un grand nombre de meres ayant leur premier accouchement a des ages varies, en accordant tine attention paniculiere d'ahord aux maladies cardiovasculaires et ensuite au cancer dtt sent. Les demandes pour utdiser ces statistiques de base sur la mortalite se son', accrues rapidement depuis lair mtse en place et tie ccssent d'utv'menter REFE3ENCES ( Smith Mf! r he CangJun \'\>rtjiitv IJjfj Base tile. (Pjpcr presented at the NCIC work shop `Computerized Record Linkage in Cancer Epidemiology1 Ottawa 8-10 Aug 1979). 2. Smith ME. The present state or* automated follow-up in Canada. Part l: Methodology and files. J Clinical Computtnf 1980; 9: 1-18 3. Smith ME. Newcombe HB. Automated follow-up facilities in ,___ada for monitoring delaved health eifects. Am J Public Health 1980; 70: 1:61-1268. 4. Patterson J. The establishment of a National Death Index in the United States. In: Cairns J. Lyon JL. Skolnick M eds. Cancer Inci dence m Defined Populations. Banburv Report 4. Cold Spring Harbour Laboratory 1980: 443-451. 5. Statistics Act. 1970-71-72. c. 15. An Act ropccung statistics of Canada. Amended by 1976-77. c.28. Minister of Supply and Servi ces: Ottawa: 1977. 6. Lynch BT. Arends WL. Selection of a sur namecoding procedure forthe S.R.S. record linkage system. U.S. Department of Agricul ture: Washington. D.C.: 1977. 7. Smith ME. Newcombe HB. Methods for computer linkage of hospital admissionseparation records iniocumulativc health his tones. Mesh Inform Med 1975; 4: IIS-125. 8. Smith ME. Neweombe HB. Accuracies of computer versus manual linkages of routine health records. Meth Inform Med 1979; 8- 89-97. 9. Howe GR. Lindsay J. A generalized iterative record linkage system for use in medical follow-up studies. Comput Biomed Res 1981; 14; 327-40. 10. Jansen JD. Cemmcsen J. Sundaram K. Isontazid -- an attempt at retrospective predic tion. Mutation Research 1980;76:85-112. 11. Howe CR. Lindsay J. Coppack E. Miller \B. Isonund exposure in relation to cancer inci dence and mortality in a cohort of tuberculo sis patients, /nr J Epidemiol 1979; S: 305-312. 12. Newcombe HB. A Method of monitoring nationally for possible delayed effects of var ious occupational environments. Publication No. NRCC-136S6. Environmental Secre tariat. National Research Council of Canada. Ottawa. Feb 1974. 13. Howe GR. Lindsay J, Miller AB. A national system for monitoring occupationally related cancer morbidity and mortality \ Paper pre sented at American Society of Preventive Oncology, Symposium on Occupational Cancer. Chicago. March 1980). 14. Howe GR. Miller AB. Cancer incidence and mortality in reiarion to occupation in 700.000 members of the Canadian labour force. (Paper presented at Fourth International Symposium on Prevention and Detection of Cancer. London. England. July 1980). 15. Anon. Pmacv. epidemiology and record link age. Sr Med J 1979; 1: 1013. 16. Mvrcen JA. HihzJE. Breast cancer following multiple fluoroscopies during artificial pneumothorax treatment of pulmonary tuberculosis. Cun Med Assoc J 1969; (00. I0J2-IUM 17. Cook DC. Dent O. Hewitt D. Breast cancer following multiple chesi fluoroscopy The Ontario experience. Cun Med -Usui*./ 1974. Ill 406-410. JX. Newcombe HB. Cancer following multiple fluoroscopies. PubL No. ArCL-5243. Chalk Kivcr Nuclear Laboratories. Chalk River. Ontario Vjg 1975. 19. Howe OR. Miller AB. Sherman GJ Breast cancer mortal;:;, lollowmg flaroscapic irraJ.j* fort tn j cohort ol tuberculous patients. (Paper presented at the Fourth International Sj mpos.um on Prevention and Detection ol Cancer. London. England Juiy 1980). 20. Miller AB. Lindsay J. Cancer following mul tiple fluoroscopies. (Paper presented at the \C*C workshop Computerized Record Link age m Cancer Epidemiology* Ottawa 8-10 kua. 197<> v 21. Sherman GJ. Howe GR. Miller AB. Rosenstem M. Organ doses per unit of exposure resulting from fluoroscopy for artificial pneumothorax. Health Phvs 1978; 35. 259-269. 22. ftoice JO. Land CE. Shore RE. Norman JE, Tokunaga M. Risk of breast cancer following low-dose radiation exposure. Radiology 1979: 131: 589-597. 23. Doll R. Strategy for detection of cancer hazards to man. Mature 1977; 265; 589-596. 24. Sutherland RB. Mortality among sinter workers -- International Nickel Company of Canada Limited -- Copper Cliff smelter. Toronto. Ontario Department of Health. Environmental Health Services Branch 1969. 25 McEwan JC. Cytological monitoring of nickel sinter plant workers. Ann SY Acrid Set 1976: 271: 365-369. 26. Roberts RS. Julian JA. Shannon HS. Muir DCF. Mortality Studies in Ontario nickel workers. The INCO'JOHC project. In: Brown SS. Sunderman Jr., eds. Nickel Toxi cology. Academic Press Londons 1980; 27-50. 27 Abbatt JD. Newcombe H8. Eldorado Nuclear retrospective epidemiology project. A retrospective study of uranium workers for mines, mills, and refinery. In. Gomez M. ed. Radiation Hazards m Mining: Control. Mea surement. and Medical Aspects. Kingsport Press Inc.. Kingsport Tenn. 1981; 369-371. 28. Abbait J Dand EN L Project Team. The Eldo rado epidemiology project: health follow-up of Eldorado uranium workers. Eldorado Nuclear Ltd. Ottawa Feb 1980. 29 Ham JM. Report of the Royal Commission on the Health ar.d Satety of Workers in Mines. Ministry of the Attorney General of Ontario. Toromo June 1976. 30 Hewitt D. Radiogenic lung cancer m Ontario uranium miners 1955-74. In: Ham JR (Com missioner) Report of the Royal Comission on ihe Health and Safety cf Workers in Mines. Ministry of the Attorney General of Ontario Toromo June 1976: 319-329. 31. Hewitt D. Uranium mining and ihc nsk of lung cancer (Paper presented at the NCIC workshop'Computerized Record Linkage in Cancer Epidemiology* Ottawa 8-10 Aug 1979). 32. Muller J. Wheeler WC. Causes of death in Ontario uranium miners. In: Proceedings of Symposium on Radiation Protection in Min ing and Milling. ILO WHO IAEA Occupa tional Safety and Health Series No. 32. 1976: 29-42. 33. Muller J. Wheeler WC. Gentleman JF. Sura- nyi G. Kusiak R. Smith M. The Ontario min ers mortality study. In: Gomez M. ed. Radia tion Hazards in Mining: Control. Measure ment. and Medical Aspects. Kingsport Press Inc.. Kingsport. Tenn. 1981; 359*362. 34. Dept, of.National Health and Welfare. Nutrttion.a National Priority A report by Nutri tion Canada to the Debt, of National' Health and Welfare I Information Canada Ottawa 1973) 35 Verdtcr P Nutrition in relation to mortal.ty. I Paper presented at the NClC workshop 'Computerized Record Linkage \n Cancer Epidermo.ogy'Ottawa cl-10 Aug 1979). 36. Lcune RJ Retrospective cohort mortality ttudv hi Ontario cinbuimcrs and funeral directors ( \ Jocrtpiton of a research proiect. Mimeograpned Jncurr.cnt Chemical Indus trial Institute ol Toxicology. Research Tn- STOI 64577 January/February 19X2 45 angle Park North Carolina 27709 19791 37. Stravaky It. Smitn SIE. L'h SH. Miller JR. A ' cohort study 01 the age at first birth upon the risk ol death Irom breast cancer (Paper pre sented at NCIC workshop 'Computerized Record Linkage in Cancer Epidemiology' Ottawa a-10 Aug I9'9). 33. de ViUiers AJ. W'indish JP Lung cancer in a fluorspar mining community -- radiation, dust and morulitv experience. 8r J tnd \lrj 1964; 21:44.108. 39. Report of the (Newfoundland! Royal Com mission Respecting Radiation. Compen sation and Safety in Fluorspar Mines. St. Lawrence, 'slid.. Queen's Printer Newfound land 1969. 40. Weeks JL. A registry for the study of the health of radiation workers employed by Atomic Energy of Canada Limited. Publ. No. AECL-6914. VVhiteshell Nuclear Research Establishment. Pinawa. Manitoba May 1979 41 Myers DR. Newcombe HB. Marko AM. Long-term foltow-up of radiation workers in Canada. In: Hubner KF. Fry SA. eds. The Medical Basis for Radiation Accident Pre paredness. Elsesicr. Holland New York 1980: 4JI-U0. 42. Newcombe H B. Plan for a continuing follow up of persons exposed to radiation in the nuclear power industry. Publication No AECL-5S38. Chalk River Nuclear Laborator ies. Chalk River. Ontario May 1976. 43. Ashmore P. The National Dose Registry, file content and potential uses. (Paper presented at NCIC workshop 'Computerized Record Linkage in Cancer Epidemiology' Ottawa 8- 10 Aug 1979). 44. Newcombe HB. Design and future use of national dose registers for regulatory control and epidcmiologv. iltulih Plus 1980: 39. 783-796. 45. Achcson ED. Record linkage and the identifi cation of long-term environmental hazards. Pw RSkk Lund I Biol) 1979:205: 165-178. 46 Science Council of Canada. Policies and poi sons: The containment of long-term hazards to human health in the environment and in the workplace. Publication No. SCC-28. Science Council of Canada. Ottawa. Oct 1977. Received Accepted: ST0164578 46 Canadian Journal of Public Health Vol. 73