Document VJeevN4D92y204b1Epg3N7Ybp

Occupation and Risk of Malignant Pleural Mesothelioma: A Case-Control Study in Spain Antonio Agudo,1' Carlos A. Gonzalez,1 Maria J. Bleda,1 Jose Ramirez,2 Santos Hernandez,3 Francisca Lopez,3 Asuncion Calleja,3 Rafael Panades,3 Oomenec Turuguet,4 Antonio Escolar,5 Manuel Beltran,s and Jose E. Gonz&lez-Moyas Background The association ofmesothelioma and asbestos exposure is well known, but some data suggest that probably many people are still being exposed to asbestos without knowing it. Methods Between 1993 and 1996, 132 cases (77% males) of histologically confirmed malignant pleural mesothelioma and 257 controls, residents in two provinces of Spain (Barcelona and Cadiz), were interviewed. They were classified according to their probability and intensity of occupational asbestos exposure by a panel of industrial hygienists, based on a detailed occupational history. Results Age and sex-adjusted odds ratio (OR)for the highest probability of exposure to asbestos was 13.2 (95% confidence interval 6.4--27.3), and 27.1 (9.28--79.3) for high intensity. A dose--response trend was observed for both, probability and intensity. Overall, 61% ofcasesand 42% ofcontrols had ever worked in an occupation with risk of asbestos exposure, with an OR of2.59 (1.60--4.22). In ourpopulation 62 % ofcases could be attributed to occupational asbestos exposure. Conclusion A high risk of pleural mesothelioma due to occupational asbestos expo sure is confirmed, but there is still a sizeable proportion for which no evidence of occupational exposure wasfound. Most of these cases could be due to other sources of asbestos exposure, mainly domestic or environmental. Am. I. Ind. Med. 37:159-168, 2000. <D 2000 Wilty-Uss, Inc. KEY WORDS: mesothelioma; asbestos; case-control studies INTRODUCTION of the pleura in Europe is found in registries from Northern Italy for both sexes, with age-adjusted rates of around 4 per Malignant mesothelioma is a relatively rare tumor 100,000 for men and from 0.5 to 0.9 in women [Parkin et al., normally located in pleura. The highest incidence of cancer 1992], Other European countries with relatively high rates are UK, the Netherlands, Denmark, and Switzerland. The average age-adjusted incidence rate for Spanish registries is 0.41 for men and 0.13 for women. Recent studies in the \RC PiKifejte of EptdanMogy and Ortal Research), (Starts, Spall USA [Price, 1997] and UK [Peto et al., 1995] suggest that *Department of Pathology,Hospital QlnfcBareetaH.Spaln ^Centre tfeSegureW t Conditions da Salut en t Tretat, Baratea, Spain : `Burns DtwarerttaSan Servteos,Instfttsto Nadars! duSeguridad s (Sotos en elTrabajo, Baafcre.S|ir< incidence and mortality are still rising. The overall upward trend is primarily because of the increased rates in men 75 years and over. According to these results, mesothelioma ^Hosptel (Mhrersltarto Puerto delMar, Cddtt,Spain deaths in males will continue to increase for at least 15-25 f Cartraet part spertscrsRS {Health Research ftndj of lf Spanish Mhlslrysf HealDt OenhraE^ number B4fOS5d; Contract grantsponssrrBtOMH) P'rogransme of tne C: years more. I ContradgramiitrniJ)er:Pt3312W. There is strong evidence supporting the causal associa I Torresponsfene to: Antato Acudo, I. {REG (Institute ol EpfdemSohgy and Clinical tion between mesothelioma and occupational exposure to I Retard^, c. Sant PeJsgrfS,08301 Mafcam, Spain E-ra* atxi<tocsniscsas asbestos. Nevertheless, some data suggest that many people I &. uJwc1.' are still being exposed to asbestos without knowing it. ia HWBUI0008873 X Mesothelioma is almost always fatal, and primary preven tion is the only way to reduce mortality. Given the long latency period of the disease, primary prevention today can reduce the frequency of mesothelioma in the future. In recent years, it has been shown that mesothelioma risk is no longer confined to workers in the asbestos industry and there is a small proportion of cases for which no evidence of exposure to asbestos exists (Huncharek, 1992], Although the association of mesothelioma and occupational exposure with asbestos is well known, the proportion of cases caused by such exposure is quite different across populations. Furthermore, accurate assessment of occupational exposure is necessary to undertake research on non-occupalionat causes of the disease. Wa conducted a case-control study in two provinces in Spain, Barcelona and Cadiz, within the framework of a European study on environmental asbestos exposure and mesothelioma [Mollo and Magnani, 1995]. Both areas have known or suspected sources of asbestos exposure and relatively high mortality rates for pleura! cancer within Spain (GEMEBA, 1993; Ldpez-Abente et al., 1996]. In Spain, there has been a decrease in the importation and number of enterprises using asbestos; however, in 1991 the estimated number of workers exposed to asbestos was 60,488 PNHST. 1992]: The permitted level of asbestos in Spain was fixed at 1 fiber/mL or less in 1984, and the industrial use of blue asbestos was prohibited in 1987 (Castejon et a]., 1987]. The concentration of asbestos fibers went down from a mean level of2.4 fiber/mL in 1984 to 0.22 in 1991 [INHST, 1992J. In this paper, we present the results on (he risk associated with occupational asbestos exposure. We also tried to quantify the proportion of cases in our populations that can be attributed to such exposure, and to identify the main occupations and activities related to an increased risk of mesothelioma. SUBJECTS AND METHODS Cases ,, The study base consisted of residents in the provinces of Barcelona and Cidiz between 1/1/1993 and 12/31/1996. Potential cases were all subjects from the study base, newly diagnosed with primary malignant mesothelioma of the pleura, and histologically confirmed. All hospitals with pathology departments in the study area participated in the ascertainment ofcases. They were identified by pathologists and specialists from other departments Involved in the diagnosis of pleural mesothelioma in any of the 30 hospitals in the study area. A referent pathologist obtained a complete protocol of diagnosis procedures and the necessary histological material to perform histochemical and inmunological examinations to confirm the diagnosis. To ensure total ascertainment of cases, a systematic review of clinical documentation and pathology files from each hospital was carried out periodically. Controls Two controls per case were selected according to sex and age of cases (frequency matched), among patients admitted to the participating hospitals in each province. Population controls were discarded because the low participation race observed in the pilot phase (below 50%). Selecting controls from the same hospital where their matched case had been diagnosed was deemed unsuitable when investigating environmental exposures, as the choice of the hospital may be related to the place of residence. In order to avoid the bias that this procedure could have produced, we used a two-step selection. Hist, a control group was selected from a random sample ofthe population; this group determined the age, sex, and municipality of residence for the control series. Then, patients with the same age and sex distribution were selected from the participating hospitals in the study area. The particular hospital where each patient was selected was the nearest to the residential address of its corresponding `population' control. Patients admitted with conditions associated with asbestos exposure were excluded as potential controls. A detailed discussion of this procedure has been presented elsewhere [Agudo and Gonzalez in press]. Information Cases and controls were interviewed at the hospital by trained interviewers (four in Barcelona and two in Cfidiz). For 44% of cases who had died by the time of the interview, information was collected from relatives at home. Out of (he 59 cases where a relative was interviewed, in 29 cases it was the spouse, in 25 a son or daughter, and only for five cases information was gathered from another relative or a friend. All controls but one were interviewed at the hospital, and answers were always provided by the subject A structured questionnaire was used covering the following items: demographic characteristics, smoking habit, past exposures to radiation, occupational history, occupations of parents and other cohabitants, and residential history, including description of dwellings and their environment Assessment of Occupational Exposure to Asbestos Several approaches were used to assess occupational exposure to asbestos. Firstly, a complete occupational history was collected. All the occupations held for at least 6 months by the subject were recorded in chronological order. The age at the beginning and end of each period, as well as the main characteristics of the job and of the HWBUI0008874 Occupation and Risk of Mesothelioma 161 Company, were reported. Occupations and activities were design, terms for sex. age, and center were always included coded a posteriori by an expert according to the Interna in the model. When the risk for a particular occupational tional Standard Classification of Occupations (ISCO) [ILO, or industrial activity group was assessed, the reference 1969] and Classification of Economical Activities of the category was formed by the nonexposed to this group. For European Community (NACE) [EUROSTAT. 1993]. In the assessment of risk associated with levels of occupational addition to the occupational history described above, job- exposure to asbestos (probability and/or intensity), each specific modules were also asked for each job from a list of level of exposure was compared to the baseline category. 33 occupations or industrial activities, potentially associated The OR for occupations at risk of exposure to asbestos with asbestos exposure in Spain, in which the subject could according to the score were estimated by comparing have been employed. They were designed to get a subjects from each occupation with a reference category standardized and more exhaustive description of each job formed by the individuals who had never been employed in and work environment. any of these occupations. Estimation of population attribu Secondly, the exposure to asbestos was evaluated by a table risk proportion (PARP) and the corresponding 95% Cl panel of industrial hygienists, blinded to the case or control were calculated by using the estimates of the adjusted condition of the subject being evaluated. Based on the relative risk obtained by logistic regression and the information collected in the occupational history and proportion of exposed cases (Greenland, 1987]. The specific modules, each occupational period was classified reliability of the occupational exposure to asbestos derived into six categories according to its probability of asbestos from information collected from relatives was assessed by exposure: no exposure, low probability, exposure possible interviewing the spouse or a son or daughter of H of the 59 (intermediate probability), exposure highly probable, sure deceased cases for which information had been previously (or almost sure) exposure, and unknown. It was also obtained directly from the subject Both classifications were classified according to its intensity, taking into account the compared by means of the observed proportion of agree estimated concentration of asbestos fibers in the work ment and the weighted kappa [Fleiss, 1981]. environment and conditions of the job. The lowest or background level was applied to a job or activity with no RESULTS recognizable source of asbestos pollution; otherwise the job was assigned to one of the three levels of increasing Interviews were obtained for 132 of the 134 cases intensity of exposure, or to the unknown category. The eligible for the study (participation 98 5%); of them 117 probability of exposure to asbestos assigned to an individual (88.6%) were identified from 24 hospitals in Barcelona and was the maximum probability recorded in any of the jobs 15 from 6 hospitals in Cadiz. A total of 297 patients were where the subject had been employed; among thosejobs that identified as potential controls of which 257 were determined the subject's probability, the maximum intensity interviewed (participation rate 87%). Only eight subjects recorded was the intensity of exposure assigned to the refused to answer, while the remaining 32 did not participate individual. for other reasons. Finally, both approaches were combined by computing The main characteristics of cases and controls are a score for each occupation based on the probability shown in Thble I. Seventy-seven percent of cases were men, assigned by the panel. For each occupational period (at with a mean age of 65.7 years, ranging from 35 to 92 years. the level of three digits ISCO classification), a score of 0 In 59 cases (44.7%) the information was obtained from was given to the no exposure level; for low probability of relatives, mainly from the spouse or a son or daughter. exposure the score was 1, for possible exposure 2, for high Controls bad sex and age distribution similar to cases, given probability 3, and sure exposure had a score of 4. The score the stratified design. Cases and controls were very similar for an occupation was the average obtained for all the regarding the education level and average number ofJobs occurrences in this occupation. Jobs classified with reported; these characteristics were also similar according to unknown probability were excluded from this procedure. the type of respondent among cases (results not shown). Occupations with an average score of >1 were considered Results on the univariate analysis of risk of mesothe at risk of exposure to asbestos. lioma associated with selected groups of occupation and industrial activities are presented in Table H. Three occu Analysis pational groups had a significantly high risk of mesothe lioma: printers and related workers, material-handling, and Estimation of relative risk was done by means of production and related workers not elsewhere classified [unconditional logistic regression [Breslow and Day, 1980]. (n.e.c.): the latter includes the subgroup of manufacture of jOdds ratios (OR) with corresponding 95% confidence asbestos cement. Four groups of industrial activities pre intervals (CD were calculated for each indicator of exposure sented a significant increased risk: manufacture of rubber analyzed. Talcing into account the stratified sampling of the and plastic products, manufacture of transport equipment HWBUI0008875 162 Agudo et al. TABLE I. Main Characteristics of Cases of Pleura! Malignant Mesothelioma and Controlsih Spain Cases Controls Center Barcelona CSdlt &% 117 88.6 15 114 a% 227 68.3 30 11.7 Bender Males Females m 77.3 30 22.7 202 78.6 55 214 Age group 35-44 years 45-54 years 55-64 years 65-74 years 75^-84 years 85-94years 6 4.5 16 12.1 31 23.5 56 42.4 19 144 4 3.0 10 33 29 113 58 22.6 115 44.7 34 13.2 11 43 Education level1 Primaryschool not completed Primary school completed Secondaryschool or higher 44 364 34 28.1 43 35.5 76 317 83 34.6 81 33,7 type of respondent Subject Spouse Son/daughter Other 73 55.3 29 220 25 183 5 3.8 256 99.6 1 0.4 " Occupational history (fanberolJobs reported mean, standard devtafion 436 2.78 4.45 TarnasesindVniitnlMMiettHmetaltewnntsaisim^tetKsva^irad^owiaia^inifaiOwML 231 (including shipbuilding), manufacturing n.e.c., and manufaccure of non-metallic products, which includes manufac turing of asbestos cement The risk of occupational exposure to asbestos according to the probability and intensity assigned by the panel of hygienists is shown in Table m. The categories of possible exposure and high probability were combined in order to have enough individuals to allow for estimations of the OR. Compared to those who never worked or who were considered as never exposed, all levels of probability and intensity had an increased significant risk, except subjects with low probability of exposure. For exposure classified as sure the OR was 13.2, and for the highest intensity it was 27.1. A dose-response trend was evident for both charac teristics; there was also an increase in risk with increase in probability within each level of intensity, as well as an increase in risk with increase in intensity within each level of probability. The OR for those with sure exposure at the highest level of intensity was 41.6. Subjects with unknown probability or intensity of exposure tended to have an increased risk. Subjects with exposure classified as sure were further examined. There was a remarkable predominance of men: 49 out of 53 cases and all the 29 controls were males. Cases ) HWBUI0008876 Occupation and Risk of Mesothelioma 163 TABLE II. Risk of PteuratMalignant MesotheliomaAccording to Selected* MaJorGroups ofOccupations and IndustrialActivities ISCQcode Job title Cueeicoatroli 011-999 311-399 400490 500699 SOT-649 751-759 831*839 841-849 851459 871474 891499 921929 941949 951959 971979 981989 m Professional, feclmfcaf, and refeted workers Clerical andtelated workers Sales workers Service workers Agrtwltaet animal tius&andry.foresfiy workers, fishermen and hunters Spinners,weavers,testers,dyersami relatedworkers Blacksmiths,toolmakers.andmacNne-taol operators MachinerySilers,machineassemblers, and precision Instrument makeis (exdudng electrical) SecWcal fittersandrelated efec&tafand electronicworkers Plumlieis,wekls,shaetmettiand structuralmetaf preparers anderoctons Gtasslirs,swtferean4retetetlwaters frittersandreteledwaters ProductionandreWedworkersnee. Bdckteyws.carpenters and otherconstruction workers MateriaWaiKffingand related equlpmenloperatois,dockersand freighthandters Transport equipment operators Labourers notelsewhere classified MCEseefSen Group of activity 0/30 24/51 27/46 30/62 41/95 17/34 12/16 17/28 12/15 15/17 13/19 m 13/7 26/42 30/28 16/32 21/32 ,A DA DB 00 OH 01 0J DK m. m f G 1 l AijtaStee,hurtling, and relatedserviceactivities Manufacture otrood products,beverages and tobacco Manufactured textiles and textile products Manufacture otchemicals,chemical products, andman-madefibers Manufacture? rubber and ptesfieproduels Manufacture oiotternon-metaSicndneraf products Manufactureof basicmetalsand fabricated metalproducts Mffliufarriuresfmaclilneryarrdei^lpmemiieii. Manufactures! transportequipment Manufacturing rose Constroclon Wholesaleandretd trade; repairof motor vehicle,motorcycles andpersonalandhousehold goods Transport,storage,aid corranunteita PuWeadmhlsbaSwj.risfence.'sodalsax'rity 42/94 19/26 27/58 12/20 13/10 26/26 18/25 ms. 2V22 16/14 34/65 28/68 tsraa 34/81 OR Ct-S5% 0.82 090 119 091 0.78 095 149 122 164 180 197 119 369 160 2.42 098 134 0.41-164 0.52-1.55 070-202 055-150 049-124 050-184 067-328 063-236 074-365 086-3.78 065-268 141-101 150-100 ~ 074-228 136-428 051-189 0.74--2.44 062 052-130 153 0.80-292 085 0.49-146 U8 056-250 266 til-669 223 122-409 144 0.74-262 127 062-260 208 108-400 2.47 116-524 104 063-173 075 0.45-124 0.81 043-153 0.74 044-123 axupasta far^fiksJdKsjjartiatirfMcs had longer duration than controls in occupations with sure exposure (median of 21.5 vs. 13.5 years) and they started to work at a younger age, although die difference was small. On an average these subjects started to work in jobs with high exposure to asbestos during the 1950s but some of them started during the last decade. The median latency was 40 years. The analysis oa probability of asbestos exposure was restricted to some subgroups: subjects younger than 55 years (11 cases and 4 controls) showed an OR of 10.0 (Cl 1.86-53.9) for the category of sure exposure. Among women, five cases and one control had high or sure levels of exposure, the combined OR being 11.3 with Cl 1.23-103. A total of 39 occupations (defined by 3-digits ISCO codes) were considered as occupations at risk of asbestos exposure according to their probability score (Table IV). Manufacture of non-metallic products showed the highest score and also the highest risk; this job title includes the specific occupation of manufacture of asbestos cement. Launderers, cleaners and pressers, electrical fitters, plum bers, and workers using material handling and related equipment also had a high risk. Other known or suspected occupations involving potential asbestos exposure are also Included in this list, but some of them were held by few subjects thus precluding estimations ofrisk. Overall. 61% of cases and 42% ofcontrols had ever worked in an occupation HWBUI0008877 164 Agudo et at. TABLE III. Risk ot Pleura! Malignant Mesothelioma by Probability and Intensity of Occupational Exposure toAsbestos Probability olwpoiure Intensity of exposure Motexpired (background) low Medium Nswworkedf notexpoiMf t 3m? - - LOW probability -- 1.64 (0.54-4.36) mi - High - - Unknown Overall i 30/127 236 (030-5.68) ms 1.88 (087-4.13) 15753 ORsai^^tyasiasexaMaoshMdf^SSianiiirafiKSeciasesAxMtiiUiypiL Potilble/hlgh probability 3.72 (165-8.37) 1&2B 6.87 (2.12-22.3) 7/9 - 332 (101-152) 4/7 435 (197-0.30) 21/45 Expoiore, sms 9.40 (252-16.3) 13/15 12.9 (56-326) nswo 413 (123-140) 22/4 - 13.2 (6.44-273) 53/23 Exposure, unknown Overall - 1 30427 345 - (172-652) 3570 930 - (438-227) 25/13 27.1 - (928--79.3) 22 17.8 - 3.68 (499-77B) (172-737) 713 20/35 173 (4.10-773) 713 TABLEIV. Rlskot Pleural Malignant Mesothelioma (orOccupations with Risk ot Exposureto Asbestos.Accordingto theExpert's Evaluation* ISCDcoda Jab Ulle Casoi/control* on CI-95% 560 Lauoderers,dry-ctemers, and presses 6/1 1731 (208-155) 841 Sfedrlneryfitters 6/6 359 (198-126) 849 Machineryfittersand assemblersnaa 9/6 457 (144-115) 851 BecWcai fitters 5/2 9.10 (158-494) 8S5 EMIdans 7/9 2S7 (097-8.45) 871 Plumbers 4/2 7.49 (130-433) 872 Welders 6/8 245 (078-733) 873 Steelmetalworkers 5/7 253 (074-8.64) 843 Manufacture ot non-melaffic products 12/2 2117 (4.45-101) 951 Wdayers 20/36 139 (191-335) 874 DriverotmaferlaHtandSng and related equipment 3/1 10.76 (198-107) Anyoccupationwilhhlgh riskofexposure toasbestos* 81/109 259 (130-432) ^irtfe!5eS^e!cs*^te^6ttt^sj>!rasipwt0rt/f)(^te^aM5tas^ia^CKa(pnacftfi4haeiJ!hSe,hs!iSlarili> feMbfo^toOagsiitire^^o>etetosp(WOTl^l^,M8]lh;agigpcrdr^SSCO^iittl^ei8fripafKahwri,Q'<;toftifei: 03ttoi#OTiifi/^.raStse(EwtU-t:0?9fe%r!iedaijafssx&(-/5,3a(^^tre(H^*X)raii4ie!?!a3te('/!}.3OJicifca! 73 l^proOT^STSmSJ^;WMiMpTCj,casino p/-J,rat*t>JtraMSas, tailingandSMftig (V-PflOmtelJtaosws,ctibSioare!raring (2/3J,7UCterta!pfWKuw,saa(WJiaOT,771 Foodpreesssxxrr*r(V-J,633 MatMie4xlBersOT.BMM!dWoi)i!e!airnCWi.estse^^ Statedf8!sonars(27-5,891 Sim terns* (3/5), 831 Stasrater*furnace operate {WJ), 932 PsWtns.veftts f/U, 959 tadnidJonraters n^CViraSti!i^^^rMe<!^xrrtix)OJtopA).9J3ftWd[rw^+fflVW^*teaw^{2H5.9a!S^tt3eMtaiai trrana(l/1k ; HWBUI0008878 Occupation and Risk of Mesothelioma 16S i risk of asbestos exposure; compared to those who had erbeen employed in any of these occupations they had a 'cant OR of 259. We estimated the PARP of mesothelioma for several Vels ofprobability ofoccupational exposure to asbestos. In ; population 62% (Cl 48.4-75.6) of all cases may be 'buted to occupational exposure to asbestos, including all ility categories. The proportion of cases due to sure ure is 37.1% (Cl 28.1-46.1). On the other hand, ,7% (Cl 2.1-53.3) of cases are due to exposure to any of s occupations considered at risk of asbestos exposure. In our study the prevalence of smokers was similar in and controls; the proportion of subjects with past t to radiation was very low and again did not show difference between cases and controls. Relative risk for i factors were close to unity and are not presented. tISCUSSION - We used several approaches to assess occupational risk mesothelioma. Our main results ere based on the classification of occupational exposure to asbestos provided by experts, on the basis of an occupational history and a structured questionnaire with detailed descriptions for a list ofjobs. We found a relative risk of 13.2 for subjects exposed |o asbestos at the maximum level ofprobability and 27.1 for those with the maximum level of intensity according to the evaluation by experts. There was a dose-response relation ship both with probability and intensity. Comparisons with other case-control studies providing similar indicators of occupational exposure to asbestos are complicated because of differences in the method of collecting information, coding schemes, and criteria of assessing exposure. Among those recently published, a hospital-based case-control study in France (Iwatsubo etal., 1998] also used a classification of exposure by experts based on occupational histories. They created a cumulative index of exposure according the probability, intensity and frequency; for the highest level the OR was 8.7, and a clear dose--response relation was observed. One study in UK (Howel et aL, 1997] classified subjects as likely, possible and unlikely exposed according to a list of occupations defined a priori; the risk for likely vs. possible or unlikely was 9.1. In Canada tTesthke et al.. 1997], an OR of 9.3 for exposure to asbestos, queried in an exposure history, was observed. In the USA [Spirtas et al., 1994] an estimation based on a list of defined occupation? had an OR of 13.9, while another based on ajob-exposure matrix had an OR of 21.5. A hospital-based study in USA (Muscat and Wynder, 1991] by using a predefined list of occupations found an overall OR in men of 8.1. In Finland [Tuomi et al., 1991] the btposure was assessed by a panel of experts on the basis of occupational history; the OR of having definite exposure was 17.7, and 3.0 for those with exposure probable. In a study based on a cancer registry from Los Angeles County (USA) [Cicioni et al., 1991] using a list of job titles defined a priori, the ORs for low and high probability of exposure were 1.6 and 6.3, respectively. In Australia [Rogers et al., 1991], the index of exposure was based on fiber content in lung tissue, the OR for a ten-fold increase in fiber concentration was 29.4 for crocidolite, 15.7 for chrysotile, and 2.3 for amosite. Finally, a hospital-based study in Massachusetts [Huncharek et al., 1996] where criteria to define exposure to asbestos are not clearly stated, observed an OR of 25.8. In addition to manufacture of asbestos cement, our analysis based on occupational groups and industrial activities showed a positive association for printers, material-handling workers, laborers n.e.c., those in manu facture of non-metallic products, those in manufacture of transport equipment, and those manufacturing n.e.c. The risk observed for printers is rather surprising, as no significant exposure to asbestos has been described. However, this is a broad group, including many specific occupations and Che estimation is based on only six cases and one control. The possibility that other occupations held by these subjects involved exposure to asbestos can not be ruled out Among the previously cited studies, only one [Teschke et al., 1997] provided detailed results on the analysis by occupational groups; they found a significantly increased risk for sheet metal workers, plumbers and pipefitters, shipbuilding workers, painters, and machinists. A small study in Canada [Finkelstein, 1996], found a high risk (OR 24.5) for maintenance workers of a petroleum refinery in Ontario. The combination of both approaches gives additional help in identifying occupations possibly responsible for the risk observed. In our study, most occupations identified in the areas of study as involving a risk of asbestos exposure had been previously identified; however, no estimations of risk regarding mesothelioma were available in Spain. It must be taken into account that risk estimates for these occupations are probably better than those derived only from the occupational history, since they are compared to a reference category including subjects who never worked in occupations considered at risk of asbestos exposure. There was special interest in the potential risk associated with some occupational or industrial activity groups. We identified 14 subjects which had sometime been employed in manufacture of non metallic products (Thble IV). among which 12 had worked sometime in manufacture of asbestos cement (ISCO 943.30). All of them were male cases and had worked in a big plant of asbestos cement located in the province of Barcelona. These 12 subjects lived in the same municipality where the plant was located or in two neighboring towns. The majority started to work in such employment during the 1960s. On the other hand, 14 subjects (10 cases and 4 controls) had sometime been HWBUI0008879 166 Agudo et al. X employed in companies whose activity was building and repairing ships and boats (NACE 35.1); all of them were males, 9 were from CSdiz (7 cases and 2 controls), and 5 (3 cases and 2 controls) were from Barcelona. Compared to the same reference category used in Table IV they had an OR. of 11.9 (Cl 3.15--44.8). The main results of our study rely on the expert's assessment of exposure. The assessment of risk based on occupational history alone has the major drawback of lacking a clean reference category; those never exposed to a specific group may have worked in other occupations of risk. Another alternative is the utilization ofjob-exposure matrices; they are simpler but (hey are suspected of producing greater non-differencial misctassifcation [Bouyer and Hemon, 1993]. The overall quality of the experts assessment procedure depends as much on the experts as on the information collected in the questionnaire. It also depends on precise rules and criteria of classification, as well as on the standardization of the procedure [Bouyer and Hemon, 1993]. Themain advantage is that it exploits all the available information. Furthermore, experts used die information on the occupational history without taking into account the opinion of the subjects, and blinded to the casecontrol status of the subject being evaluated. In (Ms way we probably minimized both information bias and exposure assessment, two major sources of bias in studies on asbestos exposure and mesothelioma [Siemiatycki and Boffetta, 1998]. Some degree of misclassitication cannot be ruled out. Nevertheless, it seems likely that (he experts only assigned a determined degree of probability or intensity of exposure when they were strongly convinced, as may he suggested by the feet that individuals classified with unknown degree of exposure showed a high relative risk both forprobability and intensity of exposure. Furthermore, the risk in the highest categories of probability and/or intensity of exposure are remarkably higher than for intermediate categories. This is important because in polychotomous exposure variables, misclassificadon, although nondifferential, does not always produce estimates biased away from the null [Dosemeci el al., 1990]. As the exposure depends upon information provided by subjects, its validity must be considered. Information for 44% of cases was provided by a relative; this is a common practice in studies on diseases with short survival after diagnosis. In a study in the UK [Teschke et al., 1997] information for the majority of cases was provided by relatives, as it was in a study in the USA [Ftnkelstein, 1996]. In another study in USA {Spirtas et al., 1994]. information was provided by relatives for 33% of cases and 13.6% of controls. In other studies information on occupation was extracted from death certificates and coroner's records. A validation study showed that information on number of areas and usual work area assignment provided by the spouse performed quite well when compared with the subject itself [Bond et al., 1988]. We had answers from relatives only among cases but not controls, this would produce, if any, a bias toward the null hypothesis. It seems reasonable that a relative might forget to mention some occupation the subject had actually had, but is less likely to assign him a job he had never had. The validity of information provided by relatives was asses sed in a sample of 18 cases. Comparison of classification of probability of exposure by experts based on the information provided by the subject against information provided by a relative showed an agreement of 0.59 measured by the weighted kappa; it was higher when the relative was the spouse (0.79) than for son or daughter (0.46). More Importantly, comparison of prevalence of probability of exposure from two sources produced a small decrease of the highest level of exposure when based on a relative's answers (from 55.5% to 50%), while the proportion for the remain ing three levels of exposure remained unchanged. This sug gests that if some bias may affect the estimates it would be in the direction of underestimation. An additional assess ment on comparability of information is indirectly provided by the fact that the average number ofjobs reported did not differ between cases and controls, and did not differ either by type of respondent among cases. Overall, 37% of cases of mesothelioma in our population are attributed to a sure (or almost sure) occu pational exposure to asbestos and tins proportion comes up to 62% when occupations with any probability of exposure are included. Both are in agreement with estimates from other European countries [Albin et al., 1999]. These estimates rely on the validity of estimates of relative risk as well as on the proportion of exposed cases in the popula tion. In our study, cases may be considered as representative since we included almost all cases diagnosed in the study base. According to mortality rates and the incidence/ mortality ratio [GEMEBA, 1993; Lopez-Abente et al., 1996],for both areas (Barcelona and Cadiz) the number of expected cases were 119 and 13, respectively, while in our study 117 and 15 were included. Additionally, all these cases were histologically confirmed by a referent patholo gist, thus the main findings of our study could hardly be distorted by some substantial degree of misdiagnosis (over or underdiagnosis) bias [Siemiatycki and Boffetta, 1998]. Around 40% of cases of mesothelioma in our popula tion could be due to causes other than occupational exposure to asbestos, although possibility of other occupational exposures in non-Uaditionally hazardous scenarios cannot be ruled out [Pinto et al., 1995]. One study [Muscat and Wynder, 1991] examined the effect ofcigarette smoking but no higher risk was found; the same result was found in our study. History of cancer in parents or first degree relatives has also been considered [Huncharek et al., 1996; Heineman et al., 1996] but no clear pattern was shown. A recent study 1 ha DC * fa : d< pf SI p> p c ( 1 1 1 ( 1 HWBUI0008880 Occupation and Risk of Mesothelioma 167 [Muscat and Huncharek, 1996] suggested that some vegetables may exert a protective effect on the risk of mesothelioma. Other factors such as radiation or virus SV40 have been proposed [Hubbard, 1997]. However, most cases not related to occupational exposure to asbestos could in fact be caused by other sources of asbestos exposure, mainly domestic (including family contacts) or environmental. Proper research on such relationships must be conducted on subjects where the direct occupational exposure has been previously evaluated and discarded. ACKNOWLEDGMENTS Finkelstein MM. 1996. Asbestos-associated cancers in the Ontario refinery and petrochemical sector. Am J Ind Med 30:610-615. Fletss JL. 1981. Statistical methods for rates and proportions. New York: John Wiley. GEMEBA (Study Group of Mesothelioma in Barcelona). 1993. Mortality of pleural mesothelioma in the province of Barcelona. Med Clin (Bare) 101:565-569. (in Spanish). Greenland S. 1987. Variance estimators for attributable fraction estimates consistent in both large strata and sparse data. Stal Med 6:701-708. Heineman EF, Bernstein L. Stark AD, Spinas R. 1996. Mesothelioma, asbestos, and reported history ofcancer in Stst-degree relatives. Cancer 77:549-554. The authors are grateful to the pathologists who contributed to the identification of cases: X. Matias-Guiu Howel D, Arblnster L, Swinburne L, Schweiger M, Renvoize E. Hatton P. 1997. Routes of asbestos exposure and the development of mesothelioma in an English region. Occup Environ Med 54:403-409. (H. Sant Pau), J. Lirola (H. Vail d'Hebron), J. Lioreta (H. de Hubbard R. 1997. The etiology of mesothelioma: are risk factors other I'Esperartga), A. Palacin (Quinta de Salat L'Manga), R. than asbestos exposure important? Thorax 52:496-497. Bemat (H. Bellvitge), L. Camacho (H. Creu Roja), M. Huncharek M. 1992. Changing risk groups for malignant mesothe Vaquero (H. Germans Trias i Pujol), J. Autonell (H. Santa lioma. Cancer 69:2704-2711. Creu de Vic), J. Badal (H. Sant Andreu, Manresa), C. Huncharek M, Kelsey K, Muscat J, Christiani D. 1996. Parental cancer Murcia (H. Sant Jaume, Calella), C. Admella (H. de and genetic predisposition in malignant pleural mesothelioma: a casecontrol study. Cancer Lett 102:205-208. Mataro), A. Serrano (H. General de GranoUers), A. Salas : (H, Mtitua de Tetrassa), M. Culubret (H. de Teirassa), A. ILO. 1969. International Standard Classification of Occupations, revised edition, 1968. Geneva: International Labour Office. J Cabezuelo (H. de Sabadell), I. Moreu (H. Sant Camil). We ~ |also thank I. Ruano, who collaborated in the initial phase of INHST (National Institute for Hygiene and Safety in the Workplace). 1992. Study of incidence and evaluation of a working population this project and C. Mas for her work in the study office. exposed to asbestos in Spanish industry. Madrid: Instituto National dc Hlgiene y Seguridad en el Trabajo. (in Spanish). REFERENCES Iwatsubo Y. Pairon JC, Boutin C, Menard O. Massin N, Caillaud D, Oriowski E. Galateau-Salle E Bignon I, Brochard P. 1998. Pleural Agudo A, Gonzalez. CA. (in press). Secondary matching: A method for mesothelioma: dose-response relation at low levels of asbestos selecting controls in case-control studies on environmental risk exposure in a French population-based case-control study. Am J factors. Ini 3 Epidemiol. Epidemiol 148:133-142. Albin M, Magnani C, Krstev S, Rapid E, Shefer 1.1999. Asbestos and Lopez-Abence G, Pollan M. Escolar A. Ecrezola M, Abraira V. 1996. cancer." An overview of current trends in Europe. Environ Health Adas of cancer mortality and other causes in Spain 1978-1992. Perspect 107(Suppl 2):289-298. Madrid: Fundacidn Ciemffica de la Asoclacifin Espailola Contra el Cincer. On Spanish). Bond GG, Bodner KM, Sobel W. Shelleaberger RJ, Flores OH. 1988. Validation ofwork histories obtained from interviews. Am S Epidemiol Moils F, Magnani C. 1995. European multiceotric case-control study 128:343-331. on risk for mesothelioma after iron-occupational (domestic end environment) exposure to asbestos. Med Lav 86:496-500. Bouyer J, Haaoa D. 1993. Retrospective evaluation of occupational exposure in population-based case-control studies: general overview Muscat JE. Huncharek M. 1996. Dietary intake and the risk of with special attention to job exposure matrices, bit J Epidemiol malignant mesothelioma Br J Cancer 73:1122-1125. 22(SuppI. 2):S57-S64. Muscat JE, Wynder EL. 1991. Cigarette smoking, asbestos exposure, Bteslow EN, Day N. 1980. Statistical methods la cancer research. Vol and malignant mesothelioma. Carreer Res 51:2263-2267. 1. The analysis of case-control studies. IARC Scientific Publications No. 32. Lyon: Interoational Agency for Research on Cancer. Parian DM. Muir CS. Whelan SL, Gao Y-T, Ferlay J. Powell J. 1992. Cancer incidence in five continents. Volume VL IARC Scientific j Castejdn J, Gonzdlez CA, Rodriguez P, Moncada S, Ttiniguel D. 1987. Publications No. 120. Lyon: Internationa! Agency for Research on Carcinogens in the workplace. Book CAPS n" 7. Barcelona; Centre Cancer. d'AohUsis i Programes Sanitaria. Peto J. Hodgson JT, Matthews FE. JonesJR, 1995. Continuing increase Cicioai C, London SJ, Garabrant DH, Bernstein L, Phillips K, Peters in mesothelioma mortality in Britain. Lancet 345:535-539. IM, 1991. Occupational asbestos exposure and mesothelioma risk in Los Angeles County: application of an occupational hazard survey Pinto C, Soffritti M. Maltoni C. 1995. Ignored occupational risks of job-exposure matrix. Ain J Ind Med 20:371-379. asbestos mesotheliomas. Med Lav 86:484--489. \ Dosemeci M, Wacholder S, Lubin JH. 1990- Does nondifferential i misclassification of exposure always bias a true effect toward the null value?. Ads I Epidemiol 132:746-748. EUROSTAT. 1993. Classification of economic activities of the Euro pean community. Etnopea (NACE) Rev. 1. Luxembourg: EUROSTAT. Price B. 1997. Analysis of current trends in United Slates mesothelioma incidence. Am J Epidemiol 145:211-218. Rogers AJ, Leigh J, Berry G, Ferguson DA, Mulder HB, Aekad M. 1991. Relationship between lung asbestos fiber type and concentration and relative risk of mesothelioma. Cancer 67:1912-1920. I t 1 t HWBUI0008881 168 Agudo et al. Siemiatycky 3, Bofeua P. 1998. Invited commentary: Is it possible to investigate (he quantitative relation between asbestos and mesothelioma in a community-based study? Am I Epidemiol 148: 143-147. Spirtas R, Heiutman EF, Bernstein L, Beebe GW. Keehn RJ, Stark A, Harlow BL. Benichou J. 1994. Malignant mesothelioma: attributable risk of asbestos exposure. Occup Environ Med 51:804-813. Teschke K. Morgan MS, Checkoway H, Franklin G, Spinelli JJ, van Belle G, Weiss NS. 1997. Mesothelioma surveillance to locate sources of exposure to asbestos. Can 3 Public Health 88:163-168. Tbomi T. Huuskonen M, Virtano M. Tbssavalneo A. Ibmmilebto L, Mattson K, Lahdensua A, Manila J. Kwhunen P, Lippo K, Tala E. 1991. Relative risk ofmesothelioma associated with different levels of exposure to asbestos. Stand J Work Environ Health 17:404-408. HWBUI0008882