Document O30rnx5oBm64OqJeVqQKXmZ2L

1f* American Journal of Epidemiology Copyright S 1969 by Th* Mm Hopkins University School of Hygiene end Public Health All rights reserved VoL 122, No. 4 Printed in U.Sj1. eral decades, a [mortality in t | showed no effe LUNG CANCER MORTALITY AMONG MEN LIVING NEAR AN ARSENIC-EMITTING SMELTER1 aortality froir |(19). The exce Istatistically sig GORAN PERSHAGEN orkers were |dency remained Pershagen, G. (National Institute of Environmental Medicine and Dept of Environmental Hygiene, KsroUnska Institute, 5-10401 Stockholm, Sweden). Lung cancer mortality among men living near an arsenic-emitting smelter. Am J Epi demiol 1985;122*84-94. EtMogic factors for lung cancer ware studied by the case-control technique among 636 man, including 212 with pulmonary carcinoma, who had died between 1881 and 1979 in a county in northern Sweden. Data on smoking habits, occu pation, and residence were obtained from a next of kin to each study subject Validation against data from other sources indicated that the exposure Informa tion was of high quality. A relative risk of 2.0 for lung cancer was seen among men who had lived within approximately 20 km from a large capper smelter. The increased risk, which is statistically significant (p < 0.05), couid not be explained by smoking habits or occupational background. Smelter workers and miners had relative risks for lung cancer of 3.0 and 4.1, respectively. No firm conclusions can be drawn on the cause of excess king cancer risk in the smelter area, but It seems plausible that the vary substantial emissions to air from the smelter, especially of arsenic, may have played a role. effects among si Smelter was gh information wi abits or other e cancer. The purpose c dy etiologic nong men linelter, in parti) i community ex ': it possible 1 bunders in the a l with differen i to evaluate in ors. air pollution; arsenic; lung neoplasms; metallurgy; mining Materiai St Exposure to inorganic arsenic com pounds esn increase the risk of lung cancer among smelter workers and workers en gaged in the production and use of arseniccontaining pesticides (1-7). The data often show positive dose-response relationships. A recent study on experimental animals confirms the pulmonary carcinogenicity of inorganic arsenic (8). Some investigations on populations liv- Received for publication September 17, 1984, and in final form January 22,1985. From the National Institute of Environmental Medicine and Department of Environmental Hygiene. Karolinska Institute, S-10401 Stockholm, Sweden. This study received financial support from Boliden Metall AB, Skelleftehamn, Sweden. The authorgratefullyacknowledges help in thedata collection from Erik Stenlund, Skelleftei Public Health Board; Lena Damber, Lennarth Nystrom, Anita Sandstrom, and Stig Wall, University of Ume&; and Rolf Svedberg, Boliden Metall AB: and from Maud Pannone and Lara Linnuan (who also per formed the computer programming), National Insti tute of Environmental Medicine. ing near copper smelters and other point emission sources of arsenic' to air have re vealed moderate increases in lung cancer mortality (9-12). Other studies have failed to detect an effect in such situations (13, 14). No detailed exposure data were given in the studies, which makes it difficult to evaluate the findings. Furthermore, there was no adequate control of possibly impor tant confounding factors, e.g., smoking habits and various occupational exposures. A major arsenic-producing smelter is lo cated in northern Sweden. Several studies have shown severe health effects among workers, e.g., a high prevalence of occupa tional dermatitis and chronic pulmonary disease as well as an increased mortality from cardiovascular disease and lung can cer (15-18). Inorganic arsenic is probably; of etiologic importance for most of these | effects. Although the emissions from tbe| smelter have been substantial during sev- \ | The RonnskarsA > the tip of a pt hnia in norther ations were start i content of an 1 Production of er metals. The t ) major producer l world for severe The smelter oper wtantial etnissior. |uUur dioxide. D 0, the yearly ar. nounted to several |ent years, the emj jy decreased to le, |). Other metals wl r large amouni l zinc. Measurements of suspended dust hs 1 stations up to 7 : the late 1960s. 1 684 RBBIT1211 Vol. 122. No, 4 Printed in USA. MEAR AN d Dept of iden). Lung . Ant J Epi- I technique sd between ibits, occudy subject, ire intormaeen among melter. The e explained miners had :onclus)ons area, but H he smelter. LUNG CANCER NEAR AN ARSENIC SMELTER 685 eral decades, a pilot study on cause-specific of arsenic and lead were mostly below 0.5 mortality in the area near the smelter Mg/m3 as a monthly average; however, a few showed no effects other than an increased values exceeded 1 Mg/m3. No measurement mortality from lung cancer among men data are available for the time periods when (19) . The excess mortality was no longer the smelter emissions were highest. statistically significant when the smelter Analyses of metals in soil and plants workers were excluded, although a ten show that dust fallout primarily takes place dency remained. Only a crude picture of the within 10-20 km from the smelter (21). effects among subjects not employed at the With this information and the prevailing smelter was given in this study, and no wind directions (north-south), two parishes information was available on smoking near the smelter were chosen as the ex habits or other exposures of importance for posed area. The remaining part of the lung cancer. county to which these parishes belong con The purpose of this investigation was to stituted the reference area. Both the ex study etiologic factors for lung cancer posed and reference areas are sparsely pop among men living near the Swedish ulated with scattered villages, but they also smelter, in particular those effects related contain some urban districts (figure 1). to community exposure. The study design From the 1975 national census, it may be made it possible to control for various con- estimated that about 70-80 per cent of the founders in the assessment of risks associ population in the two areas live in villages ated with different exposure factors as well and towns. as to evaluate interactions between these factors. Source of subjects Materials and methods Study area The study was performed according to the case-control (case-referent) technique with a methodology similar to the one used The Rdnnskarsverken smelter is located by Fershagen (5). The cases constituted and other point | at the tip of a peninsula on the Gulf of men who had died between 1961 and 1979 ic to air have re- i Bothnia in northern Sweden (figure 1). Op in the study area, i.e,, the county containing is in lung cancer I erations were started in 1930. Ores with a both the exposed and reference areas, with tudies have failed [ high content of arsenic have been used in a diagnosis of carcinoma of the bronchus, :h situations (13, / the production of copper, lead, zinc, and lung, or pleura. A total of 215 cases were i data were given I other metals. The smelter has been one of identified from the National Register on kes it difficult to ( the major producers of arsenic products in Causes of Death. This register contains irthermore, there I the world for several decades (20). information from the death certificate and jf possibly impor- . The smelter operations have resulted in is of high validity for most diagnoses (22). s, e.g., smoking I substantial emissions of various metals and An additional six cases were identified from itional exposures. I of sulfur dioxide. During the period 1930 a regional cancer register. These men had :ing smelter is lo- I 1950, the yearly arsenic emissions to air died from causes other than their pulmo n. Several studies ' amounted to several hundred tons (19). In nary carcinoma. Of the 221 cases identified, th effects among I recent years, the emissions of arsenic to air 172 (77.8 per cent) were histologically con alence of occupaironic pulmonary i j have decreased to less than 50 tons a year firmed and 20 (9.1 per cent) cytologically (20) . Other metals which have been emitted confirmed, while in 29 (13.1 per cent), the to air in large amounts include copper, lead, diagnosis was based on clinical and radio creased mortality j and zinc. logic evidence. ase and lung can- | Measurements of metal concentrations Deceased controls were chosen in order rsenic is probably for most of these ; I in suspended dust have been made at sev eral stations up to 7 km from the smelter to obtain comparable exposure information by questionnaire for cases and controls (see missions from the [ since the late 1960s. The ambient air levels below). Controls were extracted from the antial during sev- j J 686 PERSHAGEN For subjeci Rdnnskarsvei patron" accor swers has bet duration of t least five yeai ployees at th were classified five of the dui questionnaire data. t Subjects whc I parishes close | smelter (figure; I or smelter emj * exposed. It was St- from parish regi |isbes until 1961, sification is ba* mation only. The questionn ing habits includ. [daily consumptio |tobacco, and ag [stopped smoking sects who had new who had stopped ' me ofdeath were id the remaindei National Register on Causes of Death among men who had died in the study area during the same time period as the cases but without a diagnosis of carcinoma of the bronchus, lung, or pleura. Two controls were randomly chosen for each case among those who were born the same year as a case. Source of exposure information Information on occupation, smoking habits, and home addresses during the whole life span of each case and control was obtained by a questionnaire mailed in 1982 to a next of kin. Additional informa tion was gathered by telephone interview, e.g., when questionnaire answers were in complete. A similar methodology has been shown to provide data of high quality (23, 24). The questionnaire information on home addresses and employment at the Rdnnskarsverken smelter was validated by comparison with data from parish registers and company records, respectively. It was possible to obtain questionnaire information for 212 (96 per cent) of the 221 cases and for 386 (91 per cent) of their 424 matched controls. Those controls for whom questionnaire data could not be obtained were replaced by new controls, selected ac cording to the original criteria. A detailed comparison of age, cause of death, and par ish at time of death (from the death certif icate) did not reveal, any differences be tween controls who were replaced and new controls. Thus, questionnaire information was available for a total of 212 cases and 424 matched controls. This constituted the study group. Statist The data analy tiding to the met! .id Haenszel (25) tractions between Ben evaluated witl )y Walker (27). Th. vithin the matche< introIs was low, u tposure categories tity exposure effect! 1 birth was of no ; a confounding itching was aband [Tables 1 and 2 laire data on em] tarsverken smeltf ;*i.  Towns and vfflagos -- Border of study area -- Border of exposed area Q Wtnea (predominantly autfide ore) ikOraveriien ell as mines, towns, and villages ionnaire information on and employment at the i smelter was validated by data from parish registers ords, respectively. ; to obtain questionnaire 12 (96 per cent) of the 221 (91 per cent) of their 424 . Those controls for whom ta could not be obtained new controls, selected aciginal criteria. A detailed e, cause of death, and par* ith (from the death certiftveal any differences beho were replaced and new luestionnaire information a total of 212 cases and irols. This constituted the LUNG CANCER NEAR AN ARSENIC SMELTER 687 For subjects other than workers at the . Ronnskarsverken smelter, the "main occu pation" according to the questionnaire an swers has been used in the analysis. The ' duration of the main occupation was at {east five years for all such subjects. Em ployees at the Ronnskarsverken smelter were classified as smelter workers irrespecttive of the duration of employment, using ; questionnaire as well as company record | data. Subjects who had lived in either of two Iparishes close to the Ronnskarsverken femelter (figure 1), and who were not miners smelter employees, were classified as exposed. It was not possible to obtain data |from parish registers for one of these par ishes until 1961, and consequently the clas- ation is based on questionnaire inforation only. The questionnaire information on smok||ng habits included age at start of smoking, aily consumption of cigarettes and/or pipe obacco, and age at which the subject ;ied smoking (when applicable). Subi who had never smoked as well as those vho had stopped at least 16 years before ne of death were classified as nonsmokers [ the remainder as smokers. Table 1 Comparison of data on employment at the Ronnakarsverken smelter questionnaire us. company records Company records Questionnaire* Employed at smidtsr Yes No Total Employed at smelter Yes No 106 6 9 515 112 524 Total 115 521 636 * Sensitivity: 106/115 * 0,92; BpeciGcity: 515/521 -- 0.99. Table 2 Comparison ofdata on residence in a parish in the exposed area: questionnaire us. parish registers Parish registers Questionnaire* Residence in parish Yes No Total Residence in parish Yes No 48 10 58 9 517 526 Total 57 527 584 * Sensitivity: 4S/57 ** 0.ft4; specificity: 517/527 1 0.98. ' Statistical methods | The data analyses were performed sc alding to the methods described by Mantel i Haenszel (25) and Miettinen (26). In actions between exposure factors have en evaluated with the method suggested r Walker (27). The correlation coefficient yithin the matched triplets of cases and ontrols was low, i.e., less than 0.1, for the sure categories used to assess commudty exposure effects. This shows that year ' birth was of no significant importance a confounding factor, and thus the etching was abandoned in further analy- RESULTS Tables 1 and 2 compare the questiondata on employment at the Ronn- sverken smelter and residence in one of the parishes near the smelter with simi lar information from company records and parish registers, respectively. The question naire data on smelter employment had a sensitivity of 92 per cent and a specificity of 99 per cent. There was no difference in the sensitivity or specificity between cases and controls (not shown). In the following analyses, all 121 subjects who were reported as employees at tbe Ronnskarsverken smelter in the questionnaires and/or com pany records were classified as such. There were strong indications that the six men who were reported as Ronnskar employees in the questionnaires, but who were not found in the company records, had actually worked at the smelter on a contract basis. The questionnaire data on residence in one of the parishes near the smelter had a 688 PERSHAGEN Table 3 Distribution of smoking habits, residence in the exposed area, and mining and smelter work among cases and controls Age (years) and subjects 36-65 Cases Controls 66-75 Cases Controls 76-95 Cases Controls Nonsmokers Reference Exposed area* area* Miners Smelter workers Reference area* Smokers Exposed area* Miners Smalttr workers 10 0 3 34 9 43 8 3 13 59 8 52 3 5 36 18 63 10 1 10 56 12 52 47 7 0 0 2 15 11 I 33 10 5 3 4 1 0 1 24 18 24 14 4 3 Total Cases Controls 11 4 3 10 85 38 9 52 153 25 4 24 148 30 5 35 Relative riskt 95% confidence intervals (1) 2.3$ 10.4$ 8.4 8.3 17.5 35.2$ 26.2 0.7-7.6 2.7-40.2 3.2-21.9 4.5-15.4 8.9-34.4 12.2-102.0 12.7-53.9 Relative risk among smoke ret 95% confidence intervals (1) 2.2 3.1$ 1.3-3.9 1.1-9.3 2,5 ltS-4.2 * Miners and smelter workers excluded. t Computed by the Mantel-Hoenssel method (26). $ Aggregation of strata with sero cases or controls. S Estimated according to Miettinen (26). v` sensitivity and specificity of 84 and 98 per cent, respectively (table 2). The comparison has been limited to those men for whom a complete lifetime residence history could be obtained from the parish registers. No meaningful analysis of residence data could be performed for the other parish in the exposed area (see "Source of exposure in formation"). Table 3 shows the distribution of smok ing habits, residence in the exposed area, and mining and smelter work among cases and controls. The smelter workers and miners are analyzed separately, and these two categories are not included in the table among the residents in the exposed and reference areas. There is a relative risk of about 2 for lung cancer among both non smokers and smokers in the exposed area in comparison with nonsmokers and smok ers in the reference area. Only the excess risk among smokers is statistically signifi cant (the lower limit of the 95 per cent confidence interval exceeds 1.0). The rela tive risks for miners and smelter employees are increased both among nonsmokers and smokers. There is close to a multiplicative interaction between residence in the ex posed area and smoking, with a relative risk for lung cancer of 17.5 among smokers in the exposed area. This interaction sig nificantly exceeds an additive effect be tween the two factors upon statistical test ing (p < 0.05). There are differences in the occupational distribution between subjects in the ex posed and reference areas. Figure 2 shows the occupational distribution among con- d smelter work I ( ikers Miners "' _r'1 Smeller workers 5 24 3 18 4 24 1 14 04 13 9 52 5 35 35.2$ 26.2 12.2-102.0 12.7-53.9 LUNG CANCER NEAR AN ARSENIC SMELTER 689 3.1t 2.5 Figure 2. Occupational distribution among controls in the exposed and reference areas. 1.1-9.3 1..5-4.2 smokers and smok>a. Only the excess statistically signifiof the 95 per cent seds 1.0). The rela1 smelter employees ng nonsmokers and to a multiplicative sidence in the ex it g, with a relative 7.5 among smokers his interaction sig- additive effect bepon statistical test- f j j | > ^ [ in the occupational ubjects in the ex;as. Figure 2 shows oution among con- i l Table 4 Age- and smnhing-standardized relative risks associated with residence in the exposed area near the smeller, as well as with mining and smelter work Exposure group Relative 95% confidence risk* intervtflt Residence in exposed area$ Miners Smelter workers 2.0 1.2-3.4 4.15 1.7-9.7 3.0 2.0-4.7 * Computed by the Mantel-Haenszel method (25). t Estimated according to Miettinen (26). t Standardization for occupational groups as in fig ure 2 in addition to standardization for age and smok ing habits. Miners and smelter workers excluded. S Aggregation of strata with zero cases or controls. trols in the two areas. This can be expected to reflect the situation in the population from which the cases originate. The frac tion of smelter workers is higher in the exposed area than in the reference area (38.2 and 7.0 per cent, respectively). The opposite is seen for farmers, forest workers, and fishermen, who are more common in the reference area (44.8 per cent) than in the exposed area (13.5 per cent). No other pronounced differences are obvious be tween the areas. Taken as a whole, the data indicate that occupational distribution should be considered when comparing lung cancer risks in the two areas. Table 4 shows age- and smoking-stan dardized relative risks for lung cancer as sociated with residence in the exposed area as well as for miners and smelter workers. The relative risk for residence in the ex posed area is also standardized for occupa tion according to figure 2. Men in the ex posed area have a relative risk of 2.0 for lung cancer (95 per cent confidence interval 1.2-3.4) in comparison with those who had never lived in the area. There were tenden cies toward increased lung cancer risks within all occupational groups in the ex posed area (not shown). Table 4 shows that the relative risks for lung cancer are in creased both for miners and smelter work ers (4.1 and 3.0, respectively). 690 PER9HAGEN Table 5 Tobacco consumption among smokers in the exposed and reference areas as well as for miners and Table 6 Housing characteristics for controls in the exposed and reference areas as well as for miners and smelter these pc crude in smelter workers workers they expl Tobacco Type of house tor 3$ yean or more Subject Exposure group and subjects Reference area Cases No. of subjects consumption (g/day)* Mean Standard deviation 75 16.0t 11.7 Exposure group No. of sub jects Wooden <*) Mate rial other than wood One or two stories with base ment Three or more stories dents of t regarded been perfi the durati As a rule Controls Exposed area Cases Controls Miners and smelter workers Cases no 11.6 34 13.9 29 12.3 9.2 10.0 7.1 52 12.3 10.1 Reference area 279 86.4 4.3 13.3 2.9 Exposed area 61 82.4 5.9 5.9 3.9 Miners and smelter workers 64 82.8 4.7 31.3* 4.7 indicated; in the exp inconclusi' residence < from the qc with data ft the questio Controls 36 12.6 9.8 * Computed by adding cigarettes (assuming 1 g of * Higher than in the other exposure groups (p < 0.05). quality, at It should be tobacco/cigarette) and pipe tobacco. t Higher than in controls (p < 0.05). Table 5 gives the average daily tobacco consumption among smokers in the ex posed and reference areas as well as in a combined group of miners and smelter workers. It was possible to obtain detailed information on use of cigarettes and pipe tobacco for 84 per cent of the smokers, and these are include^ in the table. There is a higher tobacco consumption among cases than among controls in the exposed (p > 0.05) and reference (p < 0.05) areas. The average daily tobacco consumption among smokers in the study group is about 13 g, including both cigarettes and pipe tobacco. There is a tendency toward a higher daily tobacco consumption among cases in the reference area than in the other groups, while a reverse situation is seen among controls. No meaningful analysis was pos sible of cumulative tobacco consumption per cent of the subjects in all exposure groups had lived in wooden houses for 25 years or more. Similarly, no pronounced differences between the groups are Been either for living in houses of material other than wood or with at least three stories. A higher fraction of miners and smelter work ers had lived in one- or two-story houses with a basement than the subjects in the reference and exposed areas (p< 0.05). The relative risks for lung cancer associated with residence in the exposed area were increased in ail the types of houses (not shown). Discussion This study was initiated as a result of the findings in a pilot study by Pershagen et al. (19) which indicated a relative risk for lung cancer of 1.7 among men living near the Rdnnskarsverken smelter who had not been employed at the smelter. The present in the exj other influe an underest No firm c the cause of in the expos that the sine of important the smelter < genic agents, by inorganic. cer risk in tin risk increase i recently publj posed smelter the limited exj it is difficult t lationships am the occupation cancer risk is i sions, it is dif sponsible agent because of insufficient information on age study shows that the excess risk also re . It is notewor at start of smoking. mained after standardization for age, smok for the relative If Table 6 shows some housing character ing habits, and occupation. Furthermore, it istics for controls which are of importance could not be explained by differences in .very different b Posed area (2.0) for the indoor levels of radon and radon tobacco consumption among smokers or in It should be kep daughters (28). Relevant data for time pe housing characteristics which may be of order to be cer 1 i riods longer than 25 years could be obtained importance for the exposure to radon and between residen in 90 per cent of the subjects. About 80-85 radon daughters. Although the control of jerence areas wai [pational exposur itrain in the exposed ir miners and smelter f I l fur 2fV yearn or more i%\ HCal ner two Three Htoriefl or with more tod base* stories merit I 1 I 1 1 i .3 13.3 2.9 .9 5.9 3.9 .7 31.3* 4.7 iposure groups Ip < in all exposure m houses for 25 no pronounced groups are seen }f material other : three stories. A nd smelter work.wo-story houses i subjects in the s(p< 0.05). The rncer associated posed area were i of houses (not ( * , ( j N as a result of the Pershagen et al. tive risk for lung living near the r who had not Iter. The present ess risk also reon for age, amok. Furthermore, it y differences in ng smokers or in >hich may be of ire to radon and h the control of I ! ` j | | I lung cancer near an arsenic SMELTER 691 these potential confounding factors was smelter, the smelter worker category also crude in some instances, it is unlikely that included workers with very low exposures, they explain the excess risk. e.g., office workers and workers with only Subjects who at any time had been resi a few months of employment. Furthermore, dents of two parishes near the smelter were the selection of deceased controls may have regarded as exposed. Analyses have also resulted in a biased estimate of the relative been performed with restrictions regarding risk if the exposures under study affected the duration of residence in these parishes. the mortality of common causes of death. As a rule, higher lung cancer risks were In view of the observations of an increased indicated in subjects with longer residence mortality from cardiovascular disease in the exposed area, but the results were inconclusive because of small numbers. The residence classification was based on data from the questionnaires only. A comparison with data from parish registers showed that the questionnaire information was of high quality, at least with regard to specificity. It should be pointed out that poor precision in the exposure classification, without other influences on validity, would lead to an underestimation of any risks. No firm conclusions can be reached on the cause of the increased lung cancer risks in the exposed area, but it is not unlikely that the smelter emissions may have been of importance. The emissions to air from the smelter contain a number of carcino genic agents, with quantitative domination by inorganic arsenic. The excess lung can cer risk in the exposed area is close to the risk increase in low exposure groups of two recently published studies on arsenic-ex among smelter workers (18), some under estimation of the relative risk would be expected in this group. On the other hand, it is unlikely that the selection of deceased controls had any effect on the comparison of lung cancer risks between the exposed and reference areas (5). Lung cancer risks were higher among smokers than among nonsmokers in all oc cupational groups. There was also a higher average daily tobacco consumption in the reference area among cases who smoked than among controls who smoked, and a similar tendency was seen in the exposed area. This is to be expected if there is a positive dose-response relationship be tween smoking and lung cancer. The lack of a higher tobacco consumption in cases than in controls among miners and smelter workers may partly be explained by impor tant influences of other factors on the oc posed smelter workers (7, 29). In view of currence of lung cancer in these groups. the limited exposure data from the smelter, A number of epidemiologic studies have it is difficult to estimate dose-response re shown an increased lung cancer incidence lationships and to make comparisons with in air-polluted areas, e.g., in urban environ the occupational data. Furthermore, if lung ments (30). Differences in smoking habits cancer risk is related to the smelter emis partly explain these results, but an excess sions, it is difficult to determine the re risk often remains after standardization for sponsible agent or agents. this factor. Some studies indicate a syner It is noteworthy that the point estimate gism with regard to lung cancer between for the relative risk of lung cancer is not smoking and community exposure to air very different between residents of the ex pollutants (31, 32). In the present study, posed area (2.0) and smelter workers (3.0). there is an interaction between smoking It should be kept in mind, however, that in and residence in the exposed area, which order to be certain that the comparison significantly exceeds an additive effect. Ex between residents in the exposed and ref posure to air pollutants may have been of erence areas was not confounded by occu importance Tor the increased lung cancer pational exposure at the Ronnskarsverken risks in the exposed area and could also 692 PERSHACEN have contributed to the interactions with even preventive (39) effects of smoking smelte smoking. have been reported. These inconsistencies Women were not included in the present can probably be explained to some extent to the should study for several reasons, one of which is by differences in the mining environments. m, sion re the technical difficulty in obtaining valid The mines in the study area are of the eicposure information from a next of kin. sulfide ore type, where exposure to arsenic The average life expectancy was more than and radon may give rise to increased lung l. Hilt five years longer for females than for males during the time period at issue, which made it easier to obtain information from the spouse for males who were deceased than for females who were deceased. A crude analysis of lung cancer mortality from 1961 to 1974 among females in the Rdnnskar area did not reveal any clear effects; how ever, this was based on five cases only and on a lack of data on smoking habits (5,19). A low number of female lung cancer cases is expected because smoking used to be much less common among women than among men. This would also make any interactions with smoking less common among women than among men. At pres ent, no meaningful evaluation is possible cancer risks. An investigation on workers from one of these mines indicated that lung cancer risk was relatively higher among nonsmokers than among smokers (40), which is consistent with the findings in the present study. Swedish studies have indicated an in creased risk of lung cancer associated with living in rural one-family houses made of material other than wood and/or with a basement in comparison with living in rural wooden houses without a basement (41,42). This may be related to exposure to radon emanating from the building material and the ground. An effort was made to classify the dwellings of the subjects in the present study with regard to characteristics of im cane of at; J Inri Z. Roth expot 331:1 3. Ott cance Arch: 4. Tokuc roorta worke 1976;1 5. PeixhE expose ing a i the Co vironm 1978:3'. 6. Pinto S experie viron H 7. Lee-Fel on community exposure effects among women in the smelter area. Exposure to inorganic arsenic is of im portance for the increased lung cancer risks at the Ronnskarsverken smelter (5,17,18). A recent study among workers at this smelter indicates that there is a multipli cative interaction between arsenic exposure and smoking in relationship to lung cancer (33). In the present investigation, there was no clear interaction between smelter work and smoking in excess of an additive effect, portance for the indoor radon and radon daughter levels. No apparent differences in housing characteristics were found between subjects in the exposed and reference areas. Information in the questionnaires was not detailed enough for detecting houses made of highly radioactive aerated concrete, but according to the local public health board, this building material was not more com mon in the exposed than in the reference area. No systematic measurements have yet been made on ground leakage of radon in in humt in Mom 8. Pershag nomas c arsenic puJmoru E 9. Blot WJ and lung |10. Newman Occupati bronchng per-minii Acad Sci |11. Matanosl mortality _ viron Res which indicates that the interaction with smoking is not pronounced among smelter workers without arsenic exposure. This may explain the conflicting results of other studies on smoking interactions in copper smelter workers (6, 34, 35), in which some of the authors did not separately analyze the workers with heavy arsenic exposure. the two areas. In conclusion, the results show that men living in the vicinity of a large copper smelter have an increased risk of lung can cer. The excess risk also remains after stan dardization for smoking habits and occu pational background. It is plausible that the very substantial air emissions from the [12. Cordier S terns in a Environ F j|13, Greaves V tween !un| non-ferrot t 1981;2:15|M. Rom W, \ residents li Med 1982;; ^6. Holmqvist There was also no interaction between smelter, e.g., of arsenic, may have played a study amor mining and smoking beyond an additive role. No firm conclusions can be drawn, effect upon statistical testing. Earlier stud especially in view of the inadequate expo work inciut contact wi k Venereol (S Oil ies on miners have given conflicting results. sure data. A continued follow-up of lung Multiplicative (36, 37), additive (38), and cancer incidence in the area near the Lundgren K workers at. summary),) (39) effects of Bmoking ted. These inconsistencies explained to some extent the mining environments, he study area are of the where exposure to arsenic give rise to increased lung i investigation on workers i mines indicated that lung . relatively higher among in among smokers (40), :nt with the findings in the ies have indicated an inmg cancer associated with ne-family houses made of ;han wood and/or with a parison with living in rural 'ithout a basement (41,42). ated to exposure to radon the building material and ;ffort was made to classify the subjects in the present d to characteristics of im ; indoor radon and radon No apparent differences in eristics were found between cposed and reference areas, he questionnaires was not for detecting houses made ;tive aerated concrete, but local public health board, iterial was not more com>sed than in the reference itic measurements have yet round leakage of radon in the results show that men icinity of a large copper increased risk of lung canisk also remains after btansraoking habits and occuDund. It is plausible that tial air emissions from the irsenic, may have played a onclusions can be drawn, w of the inadequate expontinued follow-up of lung e in the area near the LUNG CANCER NEAR AN ARSENIC SMELTER 693 : smelter is desirable. If there is a causal link ! to the smelter emissions, the excess risk should diminish because of the major emis- 17. Axelson 0, Dahlgren E, Jansson CD, et al. Arsenic exposure and mortality: a case-referent study from a Swedish copper smelter. Br J Ind Med 1978:35:8-15. | gion reductions performed in recent years. 18. Wall S. Survival and mortality pattern among Swedish smelter workers. Int J Epidemiol References 1980;9:73-87. 19. Perahagen G, Elinder C-G, Bolander A-M. Mor ]. Hill AB, Failing EL Studies on the incidence of tality in a region surrounding an arsenic emitting cancer in a factory handling inorganic compounds plant. Environ Health Perspect 1977;19:133-7. of arsenic: mortality experience in the factory. Br 20. Lindau L Emissions of arsenic in Sweden and J Ind Med 1948;5:1-6. their reduction. Environ Health Perspect 2. Roth F. On bronchial carcinoma in arsenic- 1977;19:25-9. exposed vineyard workers. Virchows Arch 1958; 21. Petterason O. Pollution of soil and plants with 331:119-37. heavy metals from the RonnskArsverken smelter. 3. Ott MG, Holder BB, Gordon HL. Respiratory (In Swedish). UKuna: Swedish University of Ag cancer and occupational exposure to arsenicals. ricultural Sciences, 1976. Arch Environ Health 1974;29:250-5. 22. de Faire U, Friberg L, Lorich U, et al. A validation 4. Tokudome S, Kuratsune M. A cohort study on of cause-of-death certification in 1156 deaths. . mortality from cancer and other causes among Acta Med Stand 1976;200:223-8. workers at a metal refinery. Int J Cancer 23. Perahagen G, Axelson O. A validation ofquestion 1976;17:310-17. naire information on occupational exposure and 6. Perahagen G. Lung cancer mortality, occupational : exposure and smoking habits in a region surround smoking. Scand J Work Environ Health 1982;8:24-8. ing a smeltery. In: International Symposium on the Control of Air Pollution in the Working En vironment. Geneva: International Labour Office, 1978:370-87. 6. Pinto SS, Henderason V, Enterline PE. Mortality experience of arsenic exposed workers. Arch En viron Health 1978;33:325-31. 7. Lee-Feldstein A. Arsenic and respiratory cancer in humans: follow-up of copper smelter employees in Montana. JNC1 1983;70:601-10. 8. Perahagen G, Nordberg G, BjOrklund NE. Carci nomas of the respiratory tract in hamsters given arsenic trioxide and/or benxo(o)pyrene by the pulmonary route. Environ Res 1984;34:227-41. 19. Blot.WJ, Fraumeni JF Jr. Arsenical air pollution and rung cancer. Lancet 1975;2:142-4. flO. Newman JA, Archer VE, Saccomanno G, et a). Occupational carcinogenesis: histologic types of bronchogenic carcinoma among members of cop per-mining and smelting communities. Ann NY Acad Sci 1976;271:260-8. |1L Matanoski G, Landau B, Tonascia J, et al. Cancer mortality in an industrial area of Baltimore. En viron Res 1981;25:8-28. |l2. Cordier S, Theriault G, ltuna H. Mortality pat terns in a population living near a copper smelter. Environ Res 1983;31:311-22. Greaves W, Rom W, Lyon J. Relationship be tween lung cancer and distance of residence from non-ferrous smelter stack effluent Am J Ind Med 1981;2:15-23. 24. Dumber L How reliable are data on smoking habits obtained via questionnaires to relatives? (In Swedish). Uraea: Utoei University, 1983, 25. Mantel N, Haensxel W. Statistical aspects of the analysis of data from retrospective studies of dis ease. JNC11959;22:719-48. 26. Miettinen OS. Estimability and estimation in case-referent studies. Am J Epidemiol 1976; 103:226-35. 27. Walker AM. Proportion of disease attributable to the combined effect of two factors. Int J Epidemiol 1981;l(k81-5. 28. Swedjemark G, MjOnes L Exposure of the Swed ish population to radon daughters. In: Proceedings of the 3rd International Conference on Indoor Air and Climate. Stockholm: Swediah Council for Building Research, 1984;2:37-44. 29. Enterline PE, Marsh GM. Cancer among workers exposed to arsenic and other substances in a cop per smelter. Am J Epidemiol 1982;116:895-911. 30. Friberg L, Cederlof R Late effects of air pollution with special reference to lung cancer. Environ Health Perspect 1978;22:45-66. 31. Jedrychowski W. Interaction of smoking and ur ban air pollution in the etiology of lung cancer. Neoplasms 1978^0:603-9. 32. Vena JE. Air pollution as a risk factor in lung cancer. Am J Epidemiol 1982;116:42-56. 33. Perahagen G. Wall S, Taube A, et al. On the interaction between occupational arsenic expo iU, Rom W, Varley G, Lyon J. Lung cancer among sure and smoking and its relationship to lung residents living near the El Paso smelter. Br J Ind cancer. Scand J Work Environ Health 198l;7:302- Med 1982;39:269-72. 9. 8. Holmqviat I. Occupational arsenical dermatitis: a 34. Rancher A, Carter M, McKee D. A retrospective study among employees at a copper ore smelting epidemiological study of mortality at a large west work including investigations of skin reactions to ern copper smelter. J Occup Med 1977;19:754-8. contact with arsenic compounds. Acta Derm 35. Higgins I, Welch K, Ob M, et *1. Influence of Venereal (SuppI) (Stockh) 1951:31:1-214. arsenic exposure and smoking on lung cancer Lundgren K. Lesions in the respiratory organs of among smelter workers: s pitot study. Am J Ind workers at a smeltery. (In Swedish with English Med 1981;2:33-41. summary!. Nord Hyg Tidskr 1954;3:66-82. 36. Archer VE, Wagoner JK, Lundin FE. Uranium Ah Co All 39. AxeTson 6, Sundell L. Mining, lung cancer and smoking. Scand J Work Environ Health 1978;4:46-52. 1979;5:10-16. 42. Axelson O, Edling C, Kling H, et at. Lung cancer and radon in dwellings. Lancet 1981;2:995-6.