Tobacco Documents Online

J. Soc. Occup. Med. (1986) 36. 24 Printed in Great Britain N 0 T I C E T;ns Jaw (7iUr} 1'7 U.S: Code). Respiratory Cancer in a Scottish Industrial Community: A Retrospective Case-control Study O,LI. LLOYD. ELEANOR IRELAND. HELEN TYRRELL and FIONA WILLIAMS Environmental Epidemiology and Cancer Centre, Wolfson Institute of Occupational Health. Dundee S0111111-7 A retrospective case-vontrctl study was undertaken as part of an enquiry into possible causes of an epidemic of lung cancer in an industrial town in central' Scotland. Rel9ti.es oL the cases and con- trols answered a questionnaire which encompassed aspects of the sociul and occupational personal history ofrthe deceased. Despite the length of' time intervening between the period of mortality and this investigation. eriough yuestionnaires Mere completed to allow the histories of the cases and controls to be usefully compared. The resutts indicate that smoking and occupation wntributed little to the aetiology of the outbreak of lung cancer in Armadale. tntroduction In the small town of Armadale in central Scotland, an outbreak of primary lung cancer began in 1968 (Lloyd and Barclav. 1979; Lloyd et al.. 1982). The mean standardized mortality ratio (SMR) for primary re- spiratory cancer from 1969-73 was the highest of all cities. burghs and Iandward areas in Scotland during that time (Lloyd and MacDonald. 1984)!. Within the town itself, many of the deaths from lung cancer formed a cluster near a source of air pollution-a steel foundry (Lloyd. 1978a. 1981). This cluster was statisti- cally significant, with an SMR. based on Scottish rates. of 206 between 1968 and 1977. The cases of primary lung cancer had been identified initially by examining the diagnoses on the death certificates of the local parishes in the Registrar General's Office for Scotland. The validity of most of those diagnoses was confirmed later by obtaining supporting information from other sources of diagnostic data (Tyrrell and Lloyd~. 1983). During the earlier stages of the enquiry, the occupa- tional and social backgrounds of the deceased were investigated on a preliminary basis. through the use of hospital case notes. death certificates and discussions with local' doctors. No unusual features were identified which would have explained so many cases of lung cancer in such a short period (Lloyd. 1978b; Lloyd et al., 1962). Nevertheless, to test the hypothesis that the cluster of lung cancer might be causally linked to occupation and smoking habits, it was considered essential to undertake more detailed enquiries into the social and occupational' backgrounds of the deceased. We decided to construct a questionnaire which would be answered by relatives of the deceased in the form of a retrospective case-control, study. Most of the ques- tions concerned details of the smoking habits and occupational'experientxs of the deceased with addi- tional questions covering personal and familial histories of related' lung disease and' cancer. and the residential histories. Metbods The period' covercd by the study was 1968--74.,when the SMR for the town had been found high. The cases were residents of Armadale who had died during 14Gti-7a'1 with the diagnosis of primary lung cancer on the death certificate. The controls were chosen from a list ot residents of Armadale who had died from anv cause other than lung cancer, during the same period. Pre- liminary work had shown that this width of diagnostic frame was necessary to allow matching for the social characteristics in this small population. The cases were computer-matched consecutively for szx, age at death± 10 years. year of death±5 sears. and by social class l-V. Anticipating the problem of failure to trace some controfs, reserve controls were obtained for as many cases as was possible. Ethical permission was obtained at district and area health board levels for tracing and interviewing next-of. kin or other relatives (hereafter referred to collectively as relatives). The agreement of the local family doctors was also obtained. Using the experience gained from a pilot study of a similar questionnaire within an occupational workforce eisewhere. a final questionnaire was constructed. Since the major areas of interest were the tobacco habit and occupational history of the deceased. most of the ques• tions covered details of those areas. For smoking his- tory. questions included the average. minimum and maximum numbers of cigarettes smoked' daily. the age of starting smoking, the number of years of that habit. the use of pipes and cigars and filter cigarettes. the brand name of the tobacco used (from which the tar content was estimated). the inhalation practice. the habit of smoking at work. and exposure to passive smoking at work and' at home. There was also a ques- tion on the certainty with which this information was given. For occupational history, questions covered occupations since leaving sehool. and exposure to spe- cified chemical and physical factors: details were re- quested of any time spent at particular types of work within the coalmining and steel foundry industries. For the previous medical history of the deceased, questions covered' experience of nun-malignant respiratory dis- eases. For the familial medical history. the questions alsol included cancers and coronary heart disease. The questions on place of residence covered addresses since 1940. These addresses were subsequently assigned to Zones A-E. (see Fig. 1). which were aggregates of enumeration districts of the town at the 1971 census and which had been used previously in epidemiological investigations. Questions also covered residential proximity to in- dustrial sources of environmental air pollution. andthe degree to which that pollution inconvenienced the indi- vidual concerned. The type of fuel used normally for tia ~ 2023382625

a0proa 1km fi;. I. Zone map of Annsdale. Zone A:west of foundry: Zone B: north of foundry: Z.one C: east of foundry: Zone D: intermediate: Zone E: distam: F: foundry. heating and cooking was investigated. FinaUy, various combinations of factors were examined to try to iden- tify signs of interaction between them in promoting the elevated mortality from lung cancer. The questionnaire was administered to the nearest surviving relative of the deceased, by one of two inter- viewers. To avoid interviewer bias, the interviewers were not told whether the person they were interview- ing was a relative of a case or of a control. The relatives were traced mainly with the assistance of the local general practitioners and other staff at the only group practice in the town. Where they were unable to iden- tify surviving relatives still' living in the area. they were often able to supply names and addresses of friends who knew where the relatives could be found. Other sources used for tracing relatives were a local minister, and, for some of the more unusual names, the local telephone directory. For those cases and controls whose relatives could not be contacted. other sources of information were used to ascertain only the location of the last known address. The information derived from the questionnaire was subjected to frequency analysis: and for many expo- sures. the relative risks and their 95 per cent confidence limits were used to test the null hypothesis that the answers of the cases and controls did not significantly differ from each other. :. _ _... J_- Results General The interviews were carried out between November 1982 and March 1984. ....... .. -yc .. uc..u vI cAac1'sn4JCOntron 30-39 40-49 30-59 60-69 70-79 80+ Cases 2(5%) 1(2%) 4(10%) 15 (36%), 14 (33%) 6(14%) Controls 0(0%) 2(3%) 6(149:) 15 (36%) 15 (36Y')~ 4(10'Y,) Of a possible 137 relatives on the lhstof interviewees, 103 (75 per cent) were traced and interviewed. Nine- teen controls were subsequently eliminated: either they were the 'reserve"eontrols of cases for which a matched oontrol had already been obtained, or the cases to which they were matched had no known relatives-the relatives having migrated to an unknown location or abroad, The final total of interviews used for the analy sis of the data was 84: 42 cases, each with one matched control. Of the 42' cases of priniary lung cancer obtained for the analysis. 35 were males. The age at death of the cases ranged from 37 to 86 years; those of the controls were from 47 to 84 (Table f). Because the lapse of time between the year of death and this study taking place was up to 18 yean, it was found that many of the spouses of the cases and controls had died. Hence the largest group of informants was the daughter/son group, the second largest being the husband/wife group (Table !!). Thus, information about 30 of the cases and 32 of the controls was obtained! from a close relative (spouse, sibling, son or daughter). Smoking History There were no statistically significant differences be- tween the answers of the cases and controls in any of the questions (Table 1II). More cases than controls were found for those who had ever smoked, for cigarettes only, cigar/pipe smok- ers and for smokers of plain cigarettes; for small'and large numbers of cigarettes smoked, for high tar con- tent of cigarettes; and for inhaling practice. More controls than cases were found for non- smokers and smokers of filter cigarettes; for medium (15-29) consumption of cigarettes; for being permitted to smoke at work; and for passive smoking both at work and at home. The cases started smoking at an earlier age (18•7 years) and continued the habit for longer (45•6 years). There was less certainty about the smoking habits of the controls than about the habits of the cases. Occuparional history The differences between the number of cases and con- trols employed in the major industries of the town (coal mines, steel foundry, brickworks)!, were not statistically significant (Table M. Slightly more cases than controls Table I/. Relationship between rcspondent and deceased HurbandA Son/ Braherr Ntphtw/ wife dau;hrtr sisrer niecr cran"td Ii,-lewt Other Cases 13 (31%) 13 (319'a) 4(10%) 7(17y.) 1(2%) 3(7%) 1(2%) Controls R(19%)' 17 (41%) 7(17%) 7(17%) 1(2%) 2(5%)', 0(0%)

4 OCCUPATIONAL MEDICINE (1986) VOL. 361NO. 1 Table lll. Characteristics of the smoking history of cases and controls Gtcaory Charocrtrisnc Cases Controlr Refctivt risk• 9S 9'a confidence linua Never smoked 3 8 1•0 Cigarettes only 28 26 2,87 0•68-1!1•94 Cigarettes and pipe or cigar 4 4 2-67 0-39-IR•16 Ever Smoked: Pipe or cigar only 7 4 4•67 0•76-28•47 cigareneJpipe! Smokers 39 34 3.06 0•75-12r44 cigar Filter cigarettes 6 8 l•0 Plain cigarettes 19 12 2,11 0•59-7•61 Plain and filter. 5 4 1•66 0•31J9•01 Mean quantity 1-14 Cigarettes/day 11, 7 1•0 smoked for duration of 15-29 Cigarettes/day 10 13 0-49 0a14-1•73 habit 30+ Cigaretteslday 9 9 o-72 0•19-2-78 Tar content Low•medium tar 5 7 1.0 (estimated from brand name) Inctudes cigu/ High tar Inhaling practice 21 14 2'1 0•55-7•95 pipe smokers Yes 26 24 No Permitted to smoke at work 3 6 Yes 10 0 No 25 1:9 Continual'e:. At work Yes 10 1b 0-56 0-21-1•50 posure to No 25 21 l~o passive At home Yes 32 34 0•84 0a29-2•45 smoking No 9 R l0 ~ (in years) Age started fi 18.7 19.2 ~ smoking SD 9.1 8~9 ~ . (in years) Duntion of A 35,6 43•.7 W habit SD12•7 122 ~..y' Reliability of Very reliable 3 3 ~ information Fairly reliable 22 14 Some idea 6 7 Uncertain 3 4 ~ Guess 3 2 ' Relative risk-1 -0 Klenutks baseline casegory: had worked in the coal' mines at some tinte, but fewer had worked in the local steel foundry. Almost equal numbers of cases and controls had worked in the local brickworks. AII' other occupational groups contained negligible numbers of both cases and controls. Some individ'uals t)ad worked in more than one industry. When oalt''tmning and foundry work were catego- rized according to subgroups of occupation, the biggest difference between cases and controls was for the moulder/coremaker group of foundry workers, with I case and! 5 controls. For exposure tochemical agents, there were no significant differences between cases and controls. Personal and familinl medical histories Considerably more cases than controls were reported to have had a history of bronchitis (Table V). This difference was statistically significant, with a relative risk of 3•71 and 95 per cent confidence Gmits of 1•19- 11-58. Due to the difficulties with recall experienced, by the relatives. it was not possible to obtain a reliable or comprehensive history of chest disease or of all'types of cancer in the families of cases and controls. However, in the data available there were no significant differ- ences between the numbers of relatives of cases and controls for the histories of asthma„ lung tuber¢ulosis, and all cancers. Only slight differences between cases and controls were noted' for the types of fuel used for domestic cooking and heating (Table VI). Residential history Most of the cases and controls had been lifelong resi- dents of Atmadale;,only 5 cases and 8 controls had ever lived outside the town. Of those who had resided outside Scotland, 4 were cases (2' in the USA., I in Australia, 1 in Newfoundland) and 6 were controls (3 in England. I in the USA. 1 in France, 1 in Poland). The modal number of addresses for each person was 2; 16 cases and 17 controls had' that number. No person had lived' in more than 4 addresses. i

include persons who evar worked in the industry) Category Charocterisric Cases ConrroU Relative risA • 95% confidence !lnuu Working history Other occupations 17, 24 1.0 Coal mining 21 17 1•74 0•71-4•25 Steel foundry 9 13 0-98 0•34-2-g2 Briclcworks 6 7 1•21 0•354•26 EspoNet ro Coal dust Yes 20 15 1,•78 0•69-4-60 No 15 20 1 •0 Sand. silica. cotton Yes 4 4 1•0 or mineral No 31 31 1-0 fibres Metal dust/ Yes 7 7 1•0 fumes No 28 28 1•0 Difen and Yes 2 2 1•0 intense heat No 33 33 1•0 Mine workers Surface and/or 5 4 1•0 only gefltral1 . Frceworker 13 11 0-95 010-4•44 Surface and face- 3 2 I•2 0-13-10•99 worker Foundry workers Moulder/coremaker 1 5 only Furnaceman 1 0 Sandblaster 2 0 Smith/forger 0 1 Specific occupation 4 1 unknown ' Rclattvc r»k•1 0 dentu6cst+asclinc category. Table V. Characteristics of the medical history of cases and controls Carrgorv Charanrrlsric Cascs ConrroLr Relative Risk' 95% confidence limiu Personal Bronchitis Yes 14 5 3'•71 1•19-11•58 medical No 27 35 1•0 history Pneumonia Yes 6 5 1•21 0-34-1-47 No 35 35 1.0 Pneumoconiosis Yes 10 7 1•5: 032-4•49 No 31 33 1.0 Unspecified Yes 25 21 1•41 0.SI3-3•40 respiratory No 16 19 1•0 Family medical Coronary heart discasc I1 17 1-0 history Asthma 1' 3 0+51 0.05-5•56. T.B. 2 0 Lung cancer 2 3 1•03 0.15-7•1'9 Other cancers 8 3 4•1'2 O-R9-18•Ril' Pncumoconiosis 4 N 0~64 0• 16-2-6t ' Rctative risk- 1-0 mdems/ies /wcline catevory. Within Armadale. there were no statistically sig& ficant differences between numbers of cases and con, trols who had ever lived in the various zones (Fig. 1). More cases than controls had ever lived in Zone A. the area with the cluster of lung cancer deaths (Table V!/),. When the period 1965-74. (i.e. just before and during the time when the SMRs for lung cancer were abnor- mally high) was examined as a separate unit, the differ- ence between the numbers of cases (12)' and controlt (8) who had lived in Zone A was even greater (Table Vlil). The only other zone where the cases exceeded control!s was Zone Cy directly east of the foundry (Fig. Table VJ. Type of fuel or power used for domestic heating and cooking Gas Coal Eltcnic Coal'and other Heating Cases 2 (5%) 35 (83%) i (=%) 4 (10+fL). Controls 1 (2%) 35 (83%) 3 (7%) 3 (7%) Cooking Cases 32 (76%) 6 (14%) 2 (5%)' 1 (2%) Controls 33 (77%) 4 (109:) 3 (7%') 2 (59"0) 1); during the period 1965-74. 6 cases and 3 controls Fifteen lung cancer cases had no known relatives and had lived there. were therefore norincluded' in the 42 cases in this study..

6 OCCUPATIONAL MEDICINE (1986) VOL. JWNO. I Table Vfl: Ever lived in each zone However, they were matched to 30 controls (some A B Zones C D Caw 14 tt 7 17 Conerotr .~ 11 12 4 21 wn addresses were p lotted b k no y zone, there were no E significant differences in distribution between cases 7 over controls (Table lX)~. Thus there was no bias in the 7 geographical distribution of the cases included in the Table Vlll. Dates of residence in sones A and C 19"'0-6< l%s-69 /9m"74 Zone A Caxs 2 11 1 Controls 3 7 1 Zonr C Cases 0 3 3 Controls t 1 2 other cases nav;ng a reserve control). When their last study. Of those who were conscious of a neighbouring source of environmental air pollution, the only' major difference between numbers of cases and controls was for the combination of steel foundry and brickworks (Table X). But the cases did not complain of resulting inconvenience much more than did the controls. Various combinations of factors, including residence in Zone A were analysed' (Table Xf), but no significant signs of interaction were found which might have con- tributed strongly to the elevated mortality from lung cancer. The combination of residence in Zone A during 1965-74 and a history of bronchitis showed the greatest difference. A similar difference was found for the com- ; bination of heavy smoking (more than 29 cigarettes : daily and middle-to-high tar content) and a history of bronchitis. In all comparisons, however, the numbers i' Table IX. Loattion of last known address of cases and controls whose were too small'to allow a«asonable opportunity of 1. relatives could not be contacted A 8 C D E Cases 3 3 1 6 ~. Controlf S 6 2 9 S Table X. Proximity to pollution source findsng statssttcal stgntfncance. DLscussion Despite the long time between the increased incidence ' of lung cancer and the interviews with the relatives, the Comment Variable Cares Confroft Relative rish' 95% confidener lintits Applies only when residence Steel foundry only Gas works and foundryy Brickworks. ps and 11 I It t was within foundry 1 4 0.30 0•01-2741 'h mile of the industry Brickworks. foundry Steelworks. brickworks 20 13 3•0t1 0•F36-I 1•07 Outwith ',4-mite and coal mine a 3 2•67 0•42-16•83 proximity Awarettess of atmosphetx 5 10 t•0 pollution Yes li 12 No 27 30 ' Retaive rulc- 1•n idem1fies twotine caIeIPury. Table Xl. Combinations of factors (3•way tabks) Factots = 1. Resident in sone A between 1965-74 2. Heavy smoker. i.e. over 29 per day. and middle or high tar 3. Ever worked as miner 4. Ever worked in foundry S. History of bronchitis m and m' mandm mand® mandm m'and m m and ® msndm m and ® and (D Oand ® and 0

-----..__. _... _.....__... ......._ .........v. .,;~F..n.x.s e. u,.r,,- view allowed a comprehensive picture to be gained of the occupational and social backgrounds of cases and controls. This experience was similar to that reported in an investigation of asbestos-related mesothelioma (Fin- layson et al., 1971). In general the occupational experiences of both cases and controls were vM similar. Some epidemiological studies have suggesar44it workers in ferrous indus- tries are at a greatatv*af dying from lung cancer than persons in the poptilation; the risk of lung cancer has been reespecially to exposure to 'hot metal', with moulders particularly at risk (Morrison, 1957; Hueper, 1966; Radford et al., 1976; OPCS, 1978; Wall, 1980). However, this present study confirmed the results of preliminary work (Lloyd et al., 1982) in finding no evidence to support the association of lung cancer with foundry work in general; nor was an asso- ciation: found with hot metal exposure or with moulding in particular. There were no reports of asbestos or radon exposure. No statistically significant differences were found between the numbers of cases and controls exposed to coal, dust, sand, silica, direct and intense heat from industrial furmace,, metal dusts, or fumes from petroleum and its products. Hence, occupational experiences did not appear to have contributed in any important way to the elevated mortality from lung cancer in Armadale. (The similarity between the num- bers of cases and controls having a coal mining history and an exposure to coal dust indicated the reliability of the respondents' answers, at least in that context of occupational' experienee. ) Because the most important cause of lung cancer is known to be cigarette smoking, the questions about smoking habits were very detailed. However„ since this study relied on individuals remembering what their relatives were doing up to 15 years previously: we could not obtain as frdl and comprehensive an account of the smoking habits of the deceased as could be expected in more favourable circumstances. For instance, while it was fairly easy to obtain a figure for the amount of tobacco smoked daily, the respondents found it far more difficult to provide information concerning inhal'~ i!ng practice, and often had difficulty with brand names. However, an examination of the information about the amount, the duration of habit, the types of cigarette smoked, the opportunity to smoke at work, and about passive smoking, showed no significant differences be- tween cases and controls in any factor. For some risk factors, -there was an excess of controls. The biggest difference between the groups (19 cases and 12 controls having smoked plain cigarettes) could have been a consequence of the greater amount of missing and uncertain information found with the control group. When considering familial medical histories irt was noted that a history of lung cancer in the close family was as infrequent amongst the cases as amongst the controls. A family history of all'other cancers was more frequent with the cases than with controls, but the small numbers in both groups make this finding difficult to interpret. The high frequency of a history of bronchitis amongst the cases, which was the only statistically significantt difference between cases and controls. is consistent with evidence that bronchitis and lung cancer are both ! : &,xxaateo witn air pollution, as well as with cigarette smoking.. Indoor air pollution from cooking and heating ap- pliances has been suspected as a pathogenic factor for respiratory disease (Florey et al., 1979; Lende, 1980). No significant difference in the use of such appliances by cases and eontrols was apparent in this study. The relatively small difference between numbers of cases and controls who had ever lived in Zone A, an area which was subject to relatively high air pollution (Yule and Lloyd, 1984; Gailey and Lloyd, 1983, 1985) and' where an excess of lung cancer had been discovered (Lloyd', 1978a), might have resulted from our inability to eliminate from our controls all those whose deaths could have been linked with the air pollution through causes other than lung cancer. However, because of the small size of the town and hence the small number of deaths each year, deaths from all other causes had to be included as potential controls in order to allow the other characteristics of the cases to be matched. Never- theless, despite this difficulty, in the design of the study,. the findings were consistent with the statistically signi- ficant excess of cases observed previously in an area close to the site of a polluting industry (Lloyd et al., 1982; Lloyd, 1982). In summary. this study demonstrated that social and occupational factors were probablly, not of importance in the outbreak of lung cancer in Armadale during 1968-74. Hence the hypothesis that environmental air pollution might have played a significant aetiological role was not invalidated. The study also illustrated the practicability of undertaking a retrospective study covering a wide range of occupational and social factors by means of questionnaires given to relatives of people who had died up to 15 years before the start of the study: Acknowk~dgemests The study was undertaken with the help of a grant by the Scottish Home and' Health Department. Statistical advice was given by Mr Simon Ogston and' helpful criticism by Professor C. du V. Florey. Help with interviews was provided by Mrs Yvonne Holland: The typist was Miss Joyce Langlands. We are indebted to the local family doctors and to those interviewed' for their assistance and patience. Some of this material forms part of an M.P.H. dissertation by Helen Tyrrell. REFERENCES Fintayton A.. McEwen J. and Mair A. (1971) ,Home interviews with relatives of deceased persons: a means of obtaining histories of exposure to hazardous wbstances: Scottish Mrdical Journa/ 16, . W Florey C. du V.. Melia R.1. W., Chinn S.. Goldstein B. D.. Brooks A. G. F.. John HI. H.. Craighead 1. B. and Webster X. (1979)'it+e relation herween respiratory illness in primary tctwol children and the use ot gas for cooking. lnarnationa/ Jot+rnal o/ Epidemioloay a. 347. Gailey F. A. and Lloyd O.U. (19R3) The use of Lrranorr Con- izaroldes as a monitor of the distribution of atmospheric pollution by metab., Ecology of Diurase 2. 215. Gaiky F. A. and Uo,vd O.U. (1985) Grass and surfacs soils as monitors of atmosplleric metal pollution in central Scotland. Wa/rr. Air and Soil Polluwion 24, 1. 1

< 8 CKCUPATIn"v.aL 4tEDICIr.E (14f36) vS)Il. Zh/N(s. I Hueper W' C:, hIFN%f+I Recent Rrsulrs in Cancer Rc+rmrh. B4rlin. Springer Vcrlag. Lende R. V. D. ( I41M)) Health a%peYt5 related tulndotx air pullultun. /nternotiowiflpvnef of Epidentiulugr 1. 1". Lloyd O LM[!M) Respiratory cancer ciu.terin` az:wci•rted with loa{isiiRittrtrisd air pollutiun., Luncrt I. 318. Lloyd O. Lun= atttcer and air pi~llution (letterl• C.ancet I. 1366. ~F ~ Lloyd O.LL'V*M 114ortality in a small industrial town: prohkma of analysis an4 interpretation. In: Ward Gardner A. (ed.) Currntr Approochns ro Occupudonal Health. Btixtol. Wright. p. W. LWyd O. U. I 1ytt2) Prcrgress Repon of tirst founeen mtmths' wurl in West Lothian. sections a.3.,lntern•al Rcpun of Epidemiology Unit for Environmental Cancer. Wolfson Institute. Dundee Univcraitv:. Lloyd O.U. and Barclay R. (1979) A atwn latent pericrd for rc.pra- tory, cancer in a susceptible pes{wfation: Cumnrunity Sfedicinr 1. '_ 10~ Lloyd O.LI: and MacDonaldA. (1984) Cuntinuotts cpidemicdogical mapping^t needed public health watr;hdug. Puhlic Hrallh!'uurnu! fd. 321. Ltiprd 0 Lt.. S<larc G.. Lluvd 41. M. and Yulc F. ,y. ttyL1l. Rr+piratcuy cancer in a Scottixh communitv: wane pathok4Kat. ekcuprtiomal and general cnvir(nmental con.lderatk, a. in: firundmann E. led.) Cuncrr Cumpuign A. Cancer Epidemiufu2x. Stuttgart. Gtt4tav Fi.cher 16'erlag: pp. 111L1'11! Murri.un S. L. (1937) Oetvp+INmaI mortahr,v In, Saaland. Bnteth Juurnul of /ndiutnul .Nrdicinr 14. I `tt: Office of Population Cenxuves and Surveys (1y7M) Orcupay,,,ral .Nnnalirr Iyi0-IV7?: London. HMSO., Radford' E. P.. Wlilham S. and Hlirayama T. (IqM). In: Saffiotti U. and Wagner J. K. /ed)• Occupational' Carcinotenexis. Annulr uJ the New York Academs• of Science 271'_ 'r'A: 243. : ft9: Tyrrell H. K. M. and Lloyd O.LI., (1983) The value of tleath certification for investi`ating the epidemiology of lung cantxr in two Sctcttish towns 1961-1977: Ecology of Disease 2. :.t5. Wall S. ( I4tMl) Survival' and Mortality pattern among Swedi ch smelter workers. lntrrnuNonal Journal of Epideminlu2y !, 7.1. Yule F. A.,and Lloyd O.LC (19N1)!Metal content of an indiEenous mc>.% in Armadale. central Scotland. Water. Air med.Sc»l Poll4rlon 21. 261. Requao for rrprina should he ttddreued to: Dr 0. Ll. Lloyd. Envitonmental Epidemiology and Cancer Centre. Wolfson Institute of Occupational Health. Depanment of Community Mcdieine. Level S. Medical Schcnd. Nincwctls. Dundee. ,