Document Ow6N7DrxMrgR0a62gBVj2edX

(II be supplied free of variable on the pay' additional reprints ie form sent with the e Editor will not be British Journal ofIndustrial Medicine 1988;45:5-13 Mortality in miners and millers of crocidolite in Western Australia may be subject to ng. ubi , editorials commissioned. The regarding suitable an editorial, how- discussion with the BK ARMSTRONG,1 N H DE KLERK,1 AW MUSK,* M ST HOBBS1 From the National Health and Medical Research Council Research Unit in Epidemiology and Preventive Medicine.' department ofMedicine. University of Western Australia, Perth, the Department ofRespiratory Medicine} Sir Charles Gairdner Hospital, Nedlands, and Unit ofClinical Epidemiology,1 Department of Medicine, University of Western Australia. Perth, Western Australia abstract It is known that 650S men and 411 women were employed in the mining and milling of crocidolite at Witlenoom in the Pilbara region of Western Australia between 1943 and 1966. Employment was usually brief (median duration four months) and exposure intense (median estimated cumulative exposure 6 fibres/cc years). The vital status of 73% of the men and 58% of the women employed in the industry was known at 31 December 1980, providing 95 264 person-years of follow up with 820 deaths in men and 4914 person-years with 23 deaths in women. The standardised mortality ratio (SMR) for all causes 3fmen was 1*53 (95% confidence interval 1*43 to 1*64). Statistically significant excess death rates were observed in men for neoplasms, particularly malignant mesothelioma (32 deaths), neoplasms ofthe trachea, bronchus, and lung (SMR 2-64), and neoplasms of the stomach (SMR-4-90); respiratory diseases, particularly pneumoconiosis (SMR 25*5); infections, particularly tuberculosis (SMR 4*09); mental disorders particularly alcoholism (SMR 4*87); digestive diseases, particularly peptic ulceration (SMR 2*46) and cirrhosis of the liver (SMR 3*94); and injuries and poisonings, particularly non-transport accidents (SMR 2*36). The excess mortality from pneumoconiosis, malignant mesothelioma, and respiratory cancers, but not stomach neoplasms, was dependent on time since first exposure and cumulative exposure. There was no increase in mortality from laryngeal cancer (SMR 1*09) or neoplasms other than those listed. The SMR for all causes in women was 1*47(95% confidence interval 0*98-2*21) and for neoplasms 1*99; there was one death from malignant pleural mesothelioma. Numerous epidemiological studies haveexamined the effects ofexposure toasbestos ofdifferent types.1. Pew, however, have dealt with exposure to crocidolite alone. They have included follow up studies of wartime gas mask workers in Nottingham" and Leyiand* in England, where West Australian crocidolite was used, and Ottawa,' Canada, where the source of the crocidolite was not stated. The propor tion of female workers in these studies ranged from 45%' to 100%.' All studies showed high mortality from mesothelioma and some showed increased mor tality from ovarian cancer.'4The effects ofexposureto South African crocidolite have been examined through a case series' and a cross sectional survey,' both studies indicating probably high mesothelioma mortality rates. A recent study comparing mortality Accepted 4 December IMS rates in crocidolite mining districts with other districts in South Africa also showed high death rates from "asbestosis and/or mesothelioma" as well as excesses of deaths from lung cancer, stomach cancer, and tuberculosis.* Crocidolite was mined at Wittenoom Gorge in Western Australia from 1937 until 1966. Prom 1943 until 1966 the'prindpal leases were mined by a single company, Australian Blue Asbestos (ABA), which employe! over 6000 people, mostly for short periods. This population prorides an opportunity to study the effects of a short period of intense exposure to crocidolite alone as few employees had other asbestos exposure in mining or manufacturing industries and most stayed in Western Australia so that it has been possible to determine their present vital status. Measurements of dust concentrations in the industry in a survey conducted in 1966 permit examination of 5 \ 6 the relation between fibre concentration and mortality in this population. Previous follow up of this cohort to the end of 1977 showed increased mortality from pneumoconiosis, malignant pleural mesothelioma, and bronchogenic carcinoma* The increases were related to duration and intensity of exposure to crocidolite and time since exposure began. At that time there had been no deaths from malignant peritoneal mesothelioma recorded in this population and there was no excess ofdeaths from gastrointestinal cancer. In the current study we have extended follow up to the end of 1980 and undertaken searches of death registries in all states ofAustralia to provide more accurate data on mortality. Subjects Data on employees of ABA were obtained from employment records of the company and supplemen ted by records of subscribers to a mineworkers benevolent fund (The Western Australian Mineworkers Relief Fund). Name, date of birth, age at fust employment, nationality, dates of employment, and occupations at Wittenoom were recorded whenever possible. Additional data were obtained from records ofsubjects presenting themselves forcompulsorychest radiographs taken by the Penh Chest Clinic. A cohort of 6506 male and 411 female employees was assem bled. Methods DETERMINATION OF VITAL STATUS The names ofall workers in the cohon were searched for in death registries in all states of Australia from 1 January 1943 to 31 December 1980, all state and Commonwealth electoral roUs current in 1981 or later, all Australian telephone directories current in 1982 or later, and records ofpermanent departure overseas in the period 1972-8 held by the Australian Department of Immigration. The names were also searched for in records of the pneumoconiosis boards and cancer registries of Western Australia, New South Wales, Victoria, and South Australia, the principal public hospitals in Western Australia, and the motor vehicle drivers licensing authorities in Western Australia, South Australia, and New South Wales. Matches between names of people in the cohort and names in these records were accepted only after clerical inspec tion of all identifying data in potentially matched records. Use of records of hospital and cancer regis tries for the determination of vital status meant that subjects who had been ill were more likely to be traced than those who had not. Armstrong, de Klerk, Musk. Hob ESTIMATION OF CUMULATIVE EXPOSURE TO CROCIDOLITE FIBRES Measurements ofdust concentrations in the air of tr mine and mill at Wittenoom were made periodical between 1948 and 1958 by the Mines Department c Western Australia using a koniometer. Only total du< particles per cubic centimetre (ppcc) were counted an the upper measurement limit of 1000 ppcc was ofte. exceeded. These measurements suggest that there wa little change in total dust concentration in this period at least until a new mill was commissioned in Septern' ber 1957. Anecdotal information indicates that th Mines Department inspections were often preceded bj a shut down ofoperations. The concentrations of airborne respirable fibres of croctdolite greater than S microns in length were measured in various workplaces in s survey under taken forthe Mines Department in 1966.10Samples for counting the number of fibres were taken with a Casella long running thermal precipitator. The dura tion ofmost samples was between four and five hours. A Casella gravimetric dustsamplerand a Hexhlet were used to collect dust to estimate mass. Fibre concentra tions varied from 20 fibres per cubic centimeter (f/cc) in certain parts of the mine to 100 f/cc in the bagging ares of the mill. The ratio of fibres greater than S microns in length to total panicles varied from 1*3% to 4-0% in the same areas. Using the judgement of an ex-superintendent of ; operations at Wittenoom (who had a detailed . knowledge of all jobs on the site throughout the production period) all 87job categories in the various j workplaces were graded on a 0 to 10 scale ofestimated 1 fibre levels for both before and after September 1957 when a less dusty mill started operating. This grading was verified by the industrial hygienist who was j responsible for the 1966 survey and again by the ex- superintendent. By equating the grades of the work places with-fibre levels known from the 1966 survey it j was possible to estimate, by linear interpolation, the likely fibre concentrations applying to otherjob/work- * place combinations. Each subject's cumulative exposure was then calculated in fibre years per cubic ' centimetre (f/ccy) by adding over all their different { jobs the product offibre concentration for thatjob and L the length oftime in thejob. | MORTALITY g Expected numbers of deaths in the cohort were calculated by the man-years method" by use of age, ^ period, cause, and sex specific death rates for Western Australia. These rates were prepared for this study v from annual tabulations of deaths prepared by the - Australian Bureau of Statistics. The calendar periods * adopted were designed to begin or end with years of ^ change from one revision of the International Ctas- Klerk. Musk, Hobbs : exposure to uiora in the air ofthe re made periodically vtines Department of newt. Only totaldust <c)werecounted and 1000 ppce was often uggest that there was tration in this period, missioned in Septemm indicates that the ereoften preceded by ie respirable fibres of rons in length were s in a survey underin 1966."* Samplesfor were taken with a ecipitator. The duran fourandfive hours, er and a Hexhietwere iass. Fibre concentreubic centimeter (f/cc) K)f t the bagging fibres greater than 5 ies varied from 1*3% ix.supenntendent of no bad a detailed site throughout the agones in the various lOscaleofestimated ifter September 1957 crating. This grading hygienist who was jnd again by the ex grades of the workn the 1966 survey it ar interpolation, the -tg to otherjob/workobject's cumulative fibre yean per cubic er all their different ationfor thatjoband in the cohort were hod" by use of age, nh rates for Western pared for this study ths prepared by the rhe calendar periods or e- J with years of : In ttiooal Clas- Mortality in miners andmillers oferocidolite in Western Australia 7 afication ofDisease (ICD) to the next. The first period was 1950--3 and death rates for this period were used to calculate expected numbers of deaths in all personyears accumulated- from 1943 to 1953. SMRs were then calculated as the ratio of observed to expected deaths. Because ofthehigh proportion ofsubjects who were not traced to the end of 1980, most ofwhom were not traced beyond the end ofemployment at Witten* oora (see 'Below), and the substantially complete ascertainment ofdeaths (at least in Australia), conven tional methods ofcalculation will probably give rise to an overestimate of mortality. To demonstrate this likely bias and to provide maximum and minimum estimates of the effects oferocidolite exposure, expec ted numbers of deaths were calculated with two different censoring dates: 31 December 1980 for SMR1 and the date last known to be alive forSMR2. The former assumes that all subjects who were lost to follow up were alive at this date. All subjects were censored at age 85 ifthey were not known to have died before that age. To illustrate possible effects ofcumulative exposure to erocidolite on mortality, death rates for different causes, directly standardised to the world population of Doll,u were calculated within groupings of time since first exposure and total cumulative exposure. Results CHARACTERISTICS OF THE WORKFORCE Staff were first employed by ABA at Wittenoom in 1943 and the last terminated employmentthere in 1967 (table 1). Most were employed after the new mill was started in 1957; the median year of first employment was 1959. Most subjects began work at Wittenoom before age 30 and worked there for less than one year. Indeed 19% worked there for less than a month, a reflection, perhaps, of the poor working and living conditions.1' The median duration ofemployment was fourmonths. Most ofthe workforce was ofAustralian or British origin. The category "Australian or British" (table 1) refers to people who gave their nationality as "British." It was common at the time for those bom in Australia of British origin to consider themselves British and death certificates indicate that some 39% ofthose so classified on work history sheets were bom in Australia. Most men workedin themine or mill and sometimes both (table 1). In total. 31% of those with a known worksite worked in the mill, the site of heaviest croddolite exposure, at some time. Most women, although employed by ABA, did not work in the mine or the mill, or even elsewhere on the site ofoperations, but in the town (80% of those with known worksite). Two women had worked in the mine as surveying assistantsar>/^ two in the mill, one as a millhand and Table l Percentage distributions ofsubjects by demographic and work related characteristics at Wittenoom Chancttrittia Men (n - <5505/ Women (*4111 Year of 6m employment: I94VS2-. 1953-57* '* 22*7 15-6 1955-62 33*7 1963-67 2S-0 Unknown 0-0 At* u 6m employment: 10-24 36-6 25-34 371 >35 25-1 Unknown 1-2 'Nationality: Australian Australian or British 22-3 26-7 Other 35-1 - Unknown 15*9 Worksite: Mill only 26*3 Mine only 32*5 Mill and efaewheret Mine eod eisewberet 15*5 3-3 Neither sine oor mill 12-5 Unknown 9*9 Duration ofemployment:! 19*3 1- Smooths 4-12 months 1- 5 yean > 5 yean 25*5 32-3 19-4 2*7 Unknown 0-5 cumulative exposure to croodoiite: < 10 f/cc yars 55*5 KM00 f/cc years > 100 f/cc years 29-1 4-7 Unknown 104 19-5 17*5 29-4 32-8 0-5 40*1 23-4 26-2 10-2 33-6 ' 32-8 15-6 18-0 0-5 0-5 0*5 00 90-5 6-8 tO-7 27-7 338 23-8 10 2-9 88-1 4-1 00 7-8 *A new null begin operation in September 1957. tMili and elsewhere include* mine, mine end elsewhere exclude* mill. ^Minimum duration for those with some work periods of unknown duration (03%k one as a bagger. Despite the known high fibre concentrations in the mill, most subjects accumulated less than an estimated lOf/ccy of exposure and only 5*5% of men accumulated more than 100 f/ccy. These observations aresubstantially explained by the comparatively short employment periods. The median estimated cumulative exposure to fibres was 6*0 f/ccy in men and 0'S f/ccy in women. FOLLOW UP The vital status at 31 December 1980 was known for 73*2% ofmen and 58*0% ofwomen (table 2). Totals of 135 513 person-years of follow up were recorded in men and 8718 person-years in women if those lost to follow up were considered to be censured at 31 December 1980, and 95 264 person-years in men and 4914 person-years in women ifthey were considered to be censored at the date last known to be alive. To indicate the possible effects of loss to follow up on the results ofthe mortality analyses the percentages 8 Table 2 Results oftracing ofsubjects to 31 December 1980 Status at 31 December 1980 Men No % Dad Alive Permanently emigrated from Australia Lost to follow up: No follow up information since end ofemployment Traced after end of employment but lost before 31 December 1980 820 3833 107 12 6 589 1-6 1629 23-0 116 1-8 Women No % 23 5-6 209 50-9 6 1-3 166 40-3 7 1-7 ofsubjects with some follow up beyond termination of employment are shown within categories ofdemogra phicand work related characteristics in table 3. Follow up was generally greatest in those with complete identifying information and in those of Australian or British nationality who were employed recently and for long periods. Follow up was especially poor in those employed at Wittenoom for less than a month and this deficit is reflected in the low proportion followed up in the lowest category of estimated cumulative exposure to crocidoiite. MORTALITY Table 4 shows the SMRs for all causes of death and categories of cause of death as defined by chapters of the ICD (9th revision) according to the two censoring methods. SMRI (censored at 31 December 1980) for ail causes was just under 1-0 in both sexes whereas SMR2 (censored at the date last known to be alive) suggested about a 50% increase in overall mortality. In men, regardless of the method of estimation of the expected number of deaths, there was increased mortality from infectious and parasitic diseases, neo plasms, mental disorders, respiratory diseases, diges tive diseases, and symptoms, signs, and ill defined oonditions. For circulatory diseases, mortality was substantially less than expected under censoring method 1 (SMRI * 0*67) and near that expected under method 2 (SMR2 * HO). For injury and poisoning SMRI was near unity whereas SMR2 was significantly increased above unity. The few deaths in women limits the informativeness of cause specific mortality analysis. Mortality from neoplasms was increased only under method 2 (SMR2 !*99). One ofthe 10 deaths from neoplasms was due to malignant mesothelioma ofthe pleura (MMP), two to other respiratory cancers (expected numbers of I * 19 and 0*70 under methods I and 2 respectively), two to cancer of the cervix, and one each to cancers of the breast, ovary, skin, tongue, and colon. There was no suggestion, on review of available records, that the cancers of the ovary and colon could have been Armstrong, de Klerk, Musk, Hobbs tfortali peritoneal mesotheliomas. The two deaths from res* 0c4 Mon piratory diseases (certified as due to chronic bronchitis faction of<. and respiratory infection) were consistent with a twofold increase in mortality from these conditions. There was also a high mortality from digestive diseases based on two deaths, one from intestinal obstruction without hernia and one from cirrhosis of the liver. There was no suggestion, on review of available loot* records, that the death due to intestinal obstruction could have been due to peritoneal mesothelioma. ICD fcoowsadp teases chapters with only one or no female deaths were *P** grouped together in the "ail other causes" category. oddisente There were three such deaths, one from a motor vehicle accident, one from senile dementia, and one rwessywen padueas from nephritis. gkuoty dii To obtain more specific indications of the likely Spt.nuxydi: causes ofincreased mortality in men, the ICD chapters which contributed to excess mortality under both post discs methods of calculation of expected numbers were mtoenoary subdivided into smaller categories of cause of death. To avoid artificially raised SMRs due to the poor M0OBL.Si$ I defined e. specificity of cause of death coding for malignant l^tadpo pleural mesothelioma.'* subjects known from other lr Table 3 Percentages ofsubjects with somefollow up beyond Wecuccn: termination ofemployment in categories ofdemographic and Mpcuceu work relaud characteristics at Wittenoom Igotbcrcii. Cheeocumtkt Year of 6m employment: 1943-32 1953-37 1958-62 1963-67 Unknown Age at first employment: 10-24 23-34 >33 Unknown Nationality: Australian Australian or British Other Unknown Worksite: Mill only Mine only Mtacd indudise mill or mine Neither mine nor mill Unknown Duration of employment: < 1 month 4-12 months 1- 3 yon > 5 years Unknown Estimated cumulative exposure to croddoliir < 10 f/ec veers 1<M00 f/oc years > 100 f/ec years Unknown Men fn - 6S0S) 71-6 76-7 772 74-1 78-8 74-0 75-1 8-6 83-1 76-2 71-2 69-8 74-8 73-3 82-5 799 70-1 66-6 73-6 77-1 80-7 92-5 49-1 74-6 78-2 85-9 62-9 Women (* - 411) 53-8 36-2 64-5 61-3 -- 55-8 74-0 72-2 9*3 54 3 39*3 75-0 $6-8 504 394 394 695 53*3 45-5 53 5 73-5 73-3 754 66-7 38-3 82-4 4-- 36-2 Clerk. Musk. Hobbs Mortality in miners and millers ofcrocidolite in Western Australia 9 o deaths from res- L>4 Mortalityfrom allcauses andin selectedchapters ofthe International Classification ofDiseases in men md women who worked in the 5 chronic bronchitis Li(don ofcrocidolite at Wittenoom. Western Australia consistent with a n these conditions, m digestive diseases testinal obstruction rhosis of the liver, erfdeath ICD 9 chapter Men Observed deaths smk / SMRX Women Observed deaths SMR / SMR21 eview of available testinal obstruction mesothelioma. ICD emale deaths were r causes" category, one from a motor dementia, and one tions of the likely n. the ICD chapters rtality under both ted numbers 'were . of cause of death, s due to the poor ling for malignant known from other MM _J-XVIt aim and paraaitk I fim* n nl&ordsi V imiyaionaad rpa discam aduaydimaa pinurydimaa VI vn vm tsneducam XI ioaianrf daeaso X ipuoa. sign*and 1 defined condition* ryaadpoboatag XVI xvn otter aurn m,rv: XI-XV 120 14 206 2fitl 241 83 47 14 14 148 13" 046 (0-90-1-03) 1-77 (1-05-2-99) H3 (049-1-30) 2-30 (I-48-3-S6) 1-02 (057-1-85) 067 (0-59-0-76) 1-46 (H8-I-8I) 1-57 (1-18-2-09) 083 V (049-1-39) 140 (H3-3-22) 144 (088-1-22) 080 (048-1-32) 153 (1-43-1-64) 2-67 (1-58-4-51) 183 (1-60-2-10) 366 (2-36-5-68) 1-63 (040-24$) HO (047-1-25) 2-50 (242-3-10) 246 (1-85-3-27) 1-33 (079-2-24) 343 (1-79-HI) 147 (1-25-173) 1-21 0573-241) 23 --J to ` -t -J 6 2 2 --t -t -I 085 (057-1-29) ** 1-17 (063-2-18) * 060 (027-1-32) 1*72 (043-6-89) 217 (054-4-69) 147 (048- 2-21) 149 (147-3-70) * ' 143 (046- 2-29) 2-99 (075-12-08) 3*81 (045-15-2) ** -- 045 (030-243) 1-64 (053- 547) ftmt, .uow up beyond s ofdemographic and m tbpcaeemondat31 December I9S0. tjeca censored at date fast known to be alh*. I othercausa forwomen tnduda chapters t. HI-VI. and X-XVTt. Women OSI (a m 4tl) 56 2 64$ 61$ $$ 74-0 72-2 9-5 $43 $93 75-0 $6-1 50-0 50-0 503 60'S 53-3 4S-5 33$ 13$ 73$ 75-0 66-7 58$ 324 56-2 sources of information to have died from this disease were excluded from the observed numbers ofdeaths in the causes ofdeath to which they were coded on death certificates. Of the 32 such deaths, 19 were coded as pleural neoplasms (16 malignant, 3 benign), nine were coded as lung cancer (8 malignant, 1 undertermined nature), and two were coded as multiple malignancies without specification of site, one as lymphatic leuk* aemia and one as chronic interstitial pneumonia. Table 5 shows the results of the mortality analyses for these smallersubdivisions ofcauseofdeath in men. Infectious andparasitic diseases--The SMR for tuber* euiosis was large under both censoring schemes (2*7 and 4*1 respectively) and seemed to account for the excess ofdeaths from infectious and parasitic diseases. The other five deaths in this category were from unspecified viral diseases (2 deaths) and septicaemia, meningococcal meningitis, and actinomycosis (one each). Neoplasms--Apart from pleural mesothelioma only neoplasms of the stomach (SMRI * 1*2, SMR2 1-9) and neoplasms ofthe trachea, bronchus, and lung (SMRI - 1*6, SMR2 2-6) showed raised SMRs under'both censoring schemes. Among other neo plasms for which exposure to asbestos has been implicated as a possible cause, them was only one death from peritoneal mesothelioma (0 expected under both schemes) and there were two deaths from laryngeal cancer (2-9 and 1-8 expected under the two censoring schemes). There were half as many deaths from other neoplasms as expected under censoring scheme 1 but near to the number expected under scheme 2. Mental disorders--All 20 deaths from mental disor ders were due to alcoholism (19 coded to alcoholism and ! to alcoholic psychosis) with 6-3 and 4-! deaths expected under the two schemes. Respiratorydiseases--A large proportion ofthe excess deaths from respiratory disease was certified to pneumoconiosis (34 observed and 2-3or l-3expected). There were 22 deaths due to bronchitis or emphysema (26-5 or-14-9 expected) and 26 to other respiratory disease (28 0 or 16-9 expected). These 26 consisted of 17 from pneumonia, five from chronic obstructive airways disease, and four from influenza. Digestive diseases--Then were large increases in mor tality from both cirrhosis ofthe liver (25 observed, 9-9 or 6-3 expected) and peptic ulceration (12 observed, 7-6 or 4-8 expected). The 10 other deaths from digestive diseases included two from other liver dis ease, two from pancreatitis, and one each from intestinal volvulus, gastroenteritis, oesophageal dis ease, intestinal disease nos, disease ofthe stomach nos, and disease of the pancreas nos. Symptoms, signs, and iU defined conditions--Of the 14 deaths in this category, 12 were coded as cause 10 Armstrong, de Klerk, Musk, Hobbs i Table 5 Detailed analysis of mortality from infectious and parasitic diseases, neoplasms, mental disorders, respiratory diseases, digestive system diseases, and injury and poisoning in men Const ofdeath Observed deaths SM/tl sun 2 Infectious end parasitic discus: Tuberculosis Other Neoplasms:* Neoplasms of upper aerodigestive mat Neoplasms of oesophagus Neoplasms of stomach Neoplasms of intestines including rectum Neoplasm* of pancreas Other digestive neoplasms* Neoplasms of larynx Neoplasms of trachea, bronchus, and lung Malignant mesothelioma of the pleura# Neoplasm* of the prostate Lymphoma and myeloma Other Mental disorders: Alcoholism Other Respiratory diseases: PwvonocoruoM* Bronchitis sod emphysema Other Directive diseases; Peptic ulceration Orthosis of the Sver Other Injury sad poisoning Transport accident* OUmt accidentsI Suiddef Homicide** 9 J 5 3 17 M 7 2 2 91 32 5 6 23 20 0 34 22 26 12 10 54 33 37 4 2-72 (M2- 5-24) 199 (0-43- 2-61) 0-85 (0-36- 2*03) 0-72 (0-23- 2-22) 116 <0-72- 1-87) 0-70 (0-41- 1-18) 076 (036- 1-59) 045 (Oil- 1-81) 068 (OI7- 2-74) 1-60 (1-31- 1-97) 057 (024- 1-37) 063 (028- 1-40) 051 (034-077) 3-16 (2-04-4-89) 13-1 (108 -21-1) 083 (055- 1-26) 093 (064- 1-37) t-SS (090- 2-79) 2-53 <1-71- 3-74) 089 (048- 1-64) 076 (058-099) 1-62 (1-23- 212) 096 (069- 1-32) 1-3# (052- 3-68) 4 09 (2-13- 7-86) 1-64 (068- 3-95) 1-55 (056- 3-23) 1-18 (038- 3-66) 1-90 (MS- 396) 1-14 <067- 192) 1-21 (058- 2-55) 071 (018- 2-85) 199 (027-4-35) 264 <2 15-->24) - 199 (043- 261) 097 (044- 2-16) 080 (053- 1-20) 4-87 014- 7-55) . 25-5 (18-2 -33-7) 1-48 (097- 2-25) 1-54 (193- 2-27) 248 (1-41- 437) 3-94 (2-66- 5-83) 1-40 (073- 260) 198 (082- 1-40) 2-36 (1-81- 399) 1-59 (190- 1-91) 1-96 (074- 3-22) lachida benign neoplasm sad neoplasms of undetermined nature. tt^ tongue,tativarygbu^ gum. floor of mouth, other mouth, oropharynx, nasopharynx, feypopharynx.other pharynx, nasal cavities. One death due to malignant mesothelioms of the peritoneum, one to digestive neoplasm doc {Of the 32 known cases of malignant pleural mesothelioma, 19 note coded as pleural neoplasms, nine as other respiratory neoplasm, three as other neoplasms, and one as pneumonia. They have been rtriudad from the lauer three categories in the tabulation, jfndudes late effects ofall accidents. Indudes deaths due to injury not known whether sdf inflicted or not. **Indudes all purposefully inflicted injury except seif inflicted injury. Mor; 4C 3 3C h 1* "IS ic 8 unknown, one u dyspnoea, and one as asphyxia. Injury and poisoning--Most of the excess mortality from injury and poisoning came from the S3 deaths due to non-transport accidents (32-7 or 22-5 expected) spreadover many different causes as follows: acciden tal poisoning 6, falls 1 i. fires 4, cataclysm I, drowning 6, crushing 3, choking I, machinery 4 (2 while employed at Wittenoom), electricity 3, firearms 2 (l while employed at Wittenoom), exposure I, accidents NOS 8 (i while employed at Wittenoom), complica tions in non-therapeutic procedures 2, and late effects of a motor vehicle accktat I. There were 54 deaths from transport accidents <71-1 or 50-0 expected) including 49 involving motor vehicles (1 while at Wittenoom), three involving railways (1 while employed at Wittenoom), and two involving aircraft. There were 37 suicides (38*7 or 26-7 expected), two while still employed at Wittenoom. three violent deaths that were undetermined whether accidentally or purposefully caused, and four homicides (2-9 or 2*0 expected). DOSE-RESPONSE RELATIONS The figure shows-the changes in age standardised death rates in four categories of time since first exposure to crocidolite for four major diseases of interest in three categories of cumulative exposure. The smaller estimates of person-years (censoring scheme 2) were used. There is evidence of an effect of exposure on mortality from lung cancer in both level of exposure and time from first exposure, (o contrast the graph for stomach cancer shows no consistent evidence of dependence of mortality on crocidolite exposure. There were no deaths from this cause in the highest exposure group and only the lowest exposure group showed any increase in rate with time from first exposure. To confirm that the raised death rate for stomach cancer was not due to misdiagnosis of peritonea! mesothelioma, a possibility suggested by Doll and Peto,1 the bases for the death certificate diagnosis of the 17 deaths from stomach cancer were examined. Apart from four deaths that occurred before 1970, Age levet OA-- wh pat sto * onl Th do: Th evi a ext dis de Klerk. Musk. Hobbs tat disorders, respiratory Mortality in miners andmillers ofcrocidolite in Western Australia SM*2 4-09 12-53- 7-86) 164 (0-68- 3-95) l-JS (0-56- 3-23) IIS (0-38- 3-66) 190 (1-18- 3-06) I-14 (0-67- 1-92) 1-21 (058- 2-55) 0-71 (0-18- 2-85) 1-09 (0-27- 4-35) 264 (215--3-24> 1-09 (0-4J- 2-61) 0-97 (0-44- 2 16) 080 (0-55- 1-20) 487 (3-14- 7-55) 25-S <18-2 -3S-7) 1-48 (0-97- 2-25) 1-44 (1-04- 2-27) 2 44 (1-41- 4-3T) 3-94 (2-66- 5 83) 1-40 (0-75- 2-60) (0-82- 140) i-30 (1-41- 3-09) 1-39 (1-00- 1-91) 196 (0-74- 3-22) fcaomx. nuaJ avitia. ratory aeopUtau. ibnett 11 in age standardised i of time since first ur major diseases of cumulative exposure. ;on-years (censoring idence of an effect of g cancer in both level exposure. In contrast shows no consistent tality on crocidolite Tom this cause in the ' the lowest exposure e with'time from first ith rate for stomach :nosis of peritoneal gested by Doll and rtificate diagnosis of icer *c examined, cur. oefore 1970, Yhis sine* first exposed Years since first exposed Age standardiseddeath rates per 100 000person-yearsforspecific causes by timesincefirst exposed to crocidolitefor different kreis oftotal cumuiotm exposure to crocidolite. ------- 0------- 0 A----- A ....... . < 10fleey 10-100flee y > 100.f]cey allexposures combined (including unkrwvn exposure). where the death certificate was the only data source, DhceoiM pathological information confirmed the diagnosis of stomach cancer in each case. This updated and detailed review of mortality in the Mortality from mesothelioma was related to dose cohort ofcrocidolite workers from Wittenoom Gorge only after 20 or more yean since exposure began. has shown a similar pattern of disease to that There were, therefore, indications of a longer indue* previously observed.* There were exposure related don period for mesothelioma than' for lung cancer. increases in mortality from pneumoconiosis, malig The dose effect for pneumoconiosis, however, was nant pleural mesothelioma, and lung cancer. These evident much sooner after first exposure and there was increases were present in all exposure groups, includ a much greater difference between the highest ing the lowest with < 10 f/cc y cumulative exposure to exposure group and the lower ones than for the other crocidolite. There were also increases in mortality diseases. attributed to tuberculosis, malignant neoplasms ofthe -I / 12 stomach, alcoholism, cirrhosis of the liver, peptic ulceration, and non-iransport accidents for which other explanations are required. The best estimate of the increase in overall or cause specific mortality in this cohort probably lies between that suggested by SMRI and that suggested by SMR2. It is most unlikely that all subjects who have not been traced are still alive as implied by SMRI. Many of the Wittenoom workforce were itinerant and many were migrants, newly arrived in Western Australia, who desired more remunerative and less unpleasant work* mg and living conditions11 so that they did not stay in Wittenoom long. This is reflected in the data on duration of employment and the relation between it and the proportion successfully followed up. Their movements after leaving Wittenoom are only vaguely known. A proportion probably returned early to their country of origin, were missed in our search of departure records (available only after 1971) and have since died. In addition anecdotal information indicates that several people worked at Wittenoom under assumed names and changed their names on leaving. While the electoral roll ofthe Cotnmonwealth of Australia is believed to cover some 98% of the population of Australia aged 18 and older (electoral registration is compulsory and the roll is carefully maintained) it is conceivable that the former workers at Wittenoom are over-represented in the 2% not on the roll, if only because of the high proportion of migrants, some of whom may never have become Australian citizens. We have no data that would allow us to estimate the relative contributions of these various factors to the incomplete follow up. It is difficult, therefore, to come to any firm conclusion as to where, between SMRI and SMR2, the truth lies. We believe that it is probably nearer SMR2 than SMRI. Possibly more accurate SMRs could have been derived by assigning estimated follow up times to those lost to follow up on the basis ofaverage follow up times for those with similarcharacteristics who had been traced (J Kaldor, personal communication). This method, however, could lead to further confusion due to confoundingofmortality with whichevercharacter* istics were chosen, especially as the major difference between traced and untraced workers was in their duration ofemployment. In contrast with the previous report on this cohort.* attempts have been made here to relate mortality experience to fibrelevels measured in the industry. The results indicate dose-response effects. Probably, however, cumulative dose was underestimated for most men because the measurements were performed towards the end of the working life of the industry when all dust suppression measures that had been ' instituted were in operation. As a result the estimates Armstrong, de Klerk, Musk. Hobbs ; Mortc of potential of crocidolite fibres for causing disease :: PS. , may be exaggerated. Relative effects ofparticular dose * State levels, however, should be fairly accurate. H cbry; The rarity ofperitoneal mesothelioma in this cohort - SMF in comparison with some others1* ** has been noted before.* Shipyard workers exposed to crocidolite, Wee however, have also been noted seldom to develop Natic peritoneal mesothelioma.17 Although Browne and Jx Aust Smithere showed a longer latent period for peritoneal *' Host than pleural mesothelioma1* others have shown no : with difference.1* Other factors discriminating between i Mr < mesothelioma sites were also discussed elsewhere11 but Bfun no firm conclusions were reached. j Eccle Other studies have noted similar excess numbers of | Mrs cases of malignantmesothelioma and lung cancer in 1 repo cohorts of crocidolite workers.1 In this cohort there *; Kefe was an excess oflung cancer of 34 or SI cases is men, depending on the method used to calculate expected tD numbers, compared with 32 mesotheliomas. The H excess numbers are therefore similar for the two lie conditions although the high prevalence ofsmoking in the cohort10 21 would account for some of the excess of lung cancer. The increases in mortality from tuberculosis and S J' stomach cancer in this cohort may be explained by the fj 4A i!high proportion ofmigrants from continental Europe in the cohort who are known to have a high mortality in Australia from these diseases,2111 indeed SMR2 for migrants for stomach cancer was 3*7 compared with sh 1-7 for Australian or British born workers. The high mortality from alcoholism and alcohol related condi % tions (cirrhosis of the liver and peptic ulcer) is consistent with a known high level of alcohol con sumption in the Wittenoom workforce.u Patterns of mortality simitar to this cohort have recently been noted among a much smaller cohort of former vermiculite miners exposed to tremolite,14 >. i another amphibole fibre. There was, however, a lower proportional mortality for mesothelioma and one of \ these four deaths was peritoneal in origin but a similar proportion of mesothelioma deaths has been seen 'A among British asbestos workers employed before the introduction of their 1969 Control Regulations.21 The present analysis used conventional standard ised mortality techniques and relies on the diagnostic accuracy of death certificates (except with respect to pleural mesothelioma). The expression of dose response relation in terms ofcumulative dose is not as problematical1017 for this cohort as it might be for others because of the relatively short durations of employment compared with duration of follow up. Further analysis of the data will concentrate on the exposure response relation for the separate diseases using within cohort comparisons with separation of dose level and duration and projections for future incidence ofasbestos related disease. ?, de Klerk, Musk, Hobbs bres for causing disease effects ofparticular dose riy accurate. othelioma in this cohorr hers151* has been noted jxposed to croddolite, ted seldom to develop Mihough Browne and nt period for peritoneal others have shown no discriminating between iscussed elsewhere'* but ied. tilar excess numbers of ma and lung cancer is In this cohort there '34 or 51 cases in men, i to calculate expected mesotheliomas. The similar for the two ^valence ofsmoking in * some of the excess of . _ *om tuberculosis and lyb plained by the n c .tentaJ Europe have a high mortality -a indeed SMR2 for s 3-7 compared with n workers. The high Icohol related condind peptic ulcer) is :vel of alcohol con* force.u to this cohort have :h smaller cohort of -sed to tremolite,** is, however, a lower nelioma and one of origin but a similar ths has been seen nployed before the Regulations.3 entional standard* > on the diagnostic :pt with respect to oression of dose ativedosetsnotas :s it might be for aort `durations of ion of follow up. nncentrate on the separate diseases ith separation of vtions for future Mortality in miners andmillers ofcrocidolite in Western Australia 13 PS. A recent study of men retiring from a United States asbestos company and exposed primarily to chrysotiie has, however, reported almost identical SMRs for both stomach cancer and tuberculosis.* We gratefully acknowledge financial support from the National Health and Medical Research Council of Australia, CSR Ltd, and the Sir Charles Gairdner Hospital"Research and Special Purposes Fund. Help with theestimation ofexposure levels was provided by Mr G Major and Mr 0 Allan. The help of Mrs B Blunsdon, Mr D Atkinson, Mr A Lebrasse, Mrs J Eccles, MrS Woodward, DrJ Baker, Ms E Bingle, and Mrs A-M Haynes in collectingdata and preparing this report is gratefully acknowledged. References ~ 1 DbO'KPttoS.Effetamhtoiiko/exposuttoasbestasJjoasiaa: Health and Safety Comnwiea, HMSO, 1985. 2 Jones ISP. Smith PO. Pooiey FD, et aL The cooeequenca of .exposure to etbrstoa dust is a wartime gasmask factory, to: Wagner JC, ad. 8Wogietd effects ef mineral fibres. Lyoe: International Agency for Reoa/ch oo Caaeer, 1980^37-53. (IARC ad puht No 30.) ------ ) WtpaUBK.FoxAJ.MonalityofrcasirfasatsfcesMmbieia.A' SMMrd 19824944-8. 4 Vbeaoe EO, Gardner M3. Pfppeid EC Crime IP. Mortality of two groupe of women who maaofactmd pa masks from Aryaotiieaadcroddoliteesbedec e40-yorfollow up. BrJted Jtfrd 198249:344-8. 5 McDonald AO. McDonald JC. Metothehoma after croddolite apeeur* during gas weak manofactwa. Europe to 1978:17:340-0. 6 Wagner JC, Skgp CA, Marchaad P. Wfm pleerai rao> ii-w, tad aaheetoa espocure in the North Waters Cape Province. BrJhdMed 1960:17:260-71. 7 Talent JM. Karriaoo WO. Solomon A. Webster I. A survey of black minewwker* of the Cepe croddolite tamer*. In: Wagner JC, ed. Bioiogkai effects efmiter*fibres. Lyon: International Agency for (tetewch ea Caaeer, 1980:723-9. (IARC ad publ No 30.) t BothaJU Itwig LM. Strehei PM. Excas mortality from stomach caaeer. hmg cancer aad aabowe and/or mesothelioma in croddolite mining dauioa in South Africa. Am J Epidemiol 1986:123:30-40. 9 Hobbs MST, Woodward SD, Morphy B, Musk AW. Elder JE. The taadenot of pneuaecoaiota, mesothelioma end other tcepiretoty cancer ia mea enpsed ia mining and mitting croddolite in Western Australia, la: Wagner JC ed. Blobgleol tfieett of mineral fibres. Lyoe: International Agency for Research oo Cancer. 1980:615-25.(TARC id publ No 30.) 10 Major G. Aabototdcetexposure. Ik Major G. ed. Ptottedmp ef the first Australian pneunoeoniosts conference. Sydney. 196B. Sydney: Joint Coai Board, 1968:467-74. 1) PetoJ. Maoyearscomputerprogram.Oxford: ICRF Cancrx Uah. 1980. 12 Dofl R. Coaparieoei between registries, agr maderdisad tatee. tm WaterbouaeJ, MdrC Correa P.Powei/J. ed*. Coster Oridmer bt fin emttnenu. VoJ III. Lyoe: International Agency for " Research on Cancer. 1976:453-547. 13 LayaaaL. Work and workers'responsa at Wittenooa 1943-66. Comm*uty Health Stud 1983,71-18. 14 Xo Z. Armstrong BK, Blumdoa BJ. Roger* JM. Musk AW. ShiQua KG. Treads m mortality from malignant mesothelioma of the pteora, aad production aad use ofasbettoa ia Australia. MedJ Aust (985:143:185-7. 15 Newhouae ML, BerTyG. Pattern* ofmortality in aabotoafartory water* in London. Am MY AcadSet 1979430:53-60. 16 Sdikoff U. Hammond EC Scidman H. Mortality experience of aoulatjoowater* in the United SucaandCanada. 19(5-1976. Atm MY Act* Set 197943*91-116. 17 Blot WJ. Freumeai JF. Cucer among shipyard worker*. I&* Peto R.SchneidumaaM, eda. Quattifieation ofoccupationalcancer. (Banbury report No 9.) New York: Cold Spring Harbor Laboratory. 198147-49. 18 Browne K, Senither WJ. Asfaeetoe-reitted meaMheKoma: factor* discriminating between pleural aad peritoneal sites. Be J tnd Med l9S3;48-.MS-52. (9 SdikoffU, Hammond EC,SeidmaoH. Latencyofaabextoadiaeaae among imulatioet worker* ia the Uaited State* aad < Comer 1980.464736-40. - 20 Condon WO. de Klerk NH. Musk AW. Armstrong BK. Clancy JJ, Hobba MST. The prevalence of radiographic aabatod* ia cxocidotite miner* and miller* at Wittenoom. Waters Atm* treks. Be J tnd Med 1986:43:450-7. 21 Baker JE. Ltmg canter incidence amongst preripus employees ofan etbettot men m etlaticrahip to mtidolite exposure and tobacco smoking. Perth: Uotwmty of Western AuMrslia. 1985. (PhD these.) 22 Chest and Tuberculosis Serrica Board. AmmsA report. Wettera Australia. 1984. 23 McMichaet AJ. McCall MQ, Harubom* JM. Woodmp TL. Pattern* of gasro>tuatinal cancer in European migrant* to Australia: the role of dietary change, fat J Cancer 19804$: 431-7. 24 McDonald JC McDonald AD, Armstrong B.SebaadenP. Cohort Nudy ofmortalityofvcrmiculiteatiaenexposed totrenohie.^ JM Utd 1986;43:436-44. 25 Hodgeoa JT. Joaa RD. Mortality ofaabesta worker* ia England aad Wtla 1971-11. Br/MMtd 1986:43:158-64. 26 EaterteePE. PitftHs ia epidcmioiogial roarch. An taammation ofthe asbestos literature. J Ocevp Med 1976:18;150-6. 27 Peto J, Some proMerai ia dose raaponq otimatioo m cancer epidemiology, la: Vouk V8, Butler GC. Hod DG. Peakdl DB. eib. Methodsfor estbnoting risk ofehrmfeef injury: Mom end non human Mote md ecosystems. New York: Wiley and Sons. 1985. 28 Eateriine PE. Hanley J, Henderson V. Atbotot and cancer a cohort followed up to death. Be JM Med 1937;44:396-401. * N