Document qmZK73reqGrRYbjYN9Z2xrNeq

ton, Jamreson, Wrrght Hvlton DB, Bond G G . rcoverv among workers ropane. J Occup Med ,rr al. Differences in the .rs during summer and 3. mnal reproductive out. 1990;4:1-4. i Chicago: Year Book iships in experimenrallv zs 1988;73:335-48. ianging factors, adapted :y conrinuing education , 29th Annual Meeting. Educarion Commirree, 11s. Chemical Institute o/ 5:1-304. Society of Toxicology .ecrure handours). rionships in experimeniin Res 1988;73:335-48. :onthedeveloping brain. -M. The fetal alcohol I. .Vcuroroxicolog.v 1989; and brain development: international neurotoxI989;10:305-6. nen M, Hemminki K . ien exposed to organic -63. C , er rll. Longitudinal lead exposure and early d 1987;316:1037-43. of lead. Environ Healrh st century: lessons from nvtron Heulth Perrpecr mmtssion guidelines on igron, DC: Occupational ireau of National Affairs ,Pogue. Feral prorection nmrnr.d .Medicine Reprr studies, few answers. I 1989;141:974-7. ing pregnancy affect the 1990;4:1-3. I Commitree, Canadian s infection In day-care madian Medical Arrocia- ,uvernement du Quebec 730, boulevard Charesr ienr section. Approaches r listing as reproductive/ osition 65. Los Angcles, Health Hazard Assessdealth Services, March Ight io h o w about toxic :ians. N E@ J Mcd nanr worker- resource ya: Public Affairs Direc- ziarion. Guidelines on cards in the workplace. Bntrsh Journal of Indusrnal Medicine 1992;49:95-103 95 Occupational and other environmental factors and multiple myeloma: a population based case-control study Mikael Eriksson, Maria Karlsson Abstract Epidemiological studies on the aetiology of multiple myeloma are reviewed as a background to this population based case-control study performed in a n area with a high incidence of multiple myeloma. The purpose was to identify a n d evaluate several suspected environmental factors i n relation to this disease. A total of275confirmed casesdiagnosed i n four counties in northern Sweden during four years were c o m p a r e d with the same n u m b e r of control subjects drawn from population registries. T h e controls were matched f o r age, sex, county, a n d vital status. Occupations and work associated exposures to chemicals a n d o t h e r potential carcinogens were assessed by a n extensive questionnaire that also included questions on smoking habits, residential building mater :als,vicinity to electrical power lines, and leisure time contact with animals, electrical equipment, and chemicals. Information obtained f r o m t h e questionnaires was completed by telephone interviews when necessary. Univariate analysis and multivariate logistic regression were performed. The study confirmed a n association found earlier between f a r m i n g and multiple myeloma. Some domestic animals (cattle, horses, and goats) and two types of pesticides (phenoxyacetic acids and DDT) were indicated as risk factors within farming. Exposure to electromagnetic fields, asbestos, and organic solvents were negatively associated with myeloma in this study. Multiple myeloma is an incurable malignant lymphoproliferative disease with an increasing incidence and mortality in several countries. The causes of the Department of Oncology, University Hospital, S-901 85 UmeA, Sweden M Eriksson Department of Occupational Medicine, University Hospital, S-58185 Linkoping, Sweden M Karlsson disease are poorly understood. Some reports on familial multiple myeloma indicate a possible hereditary influence.'-' Exposure to high doses of ionising radiation is now generally accepted as a factor increasing the risk of developing multiple myeloma. Atomic bomb survivors in Hiroshima and Nagasaki have been clearly shown to have a dose related increase in the risk for multiple myeloma, with rather long latency period^.^' Workers at the Hanford atomic plant in Washington State have been followed up regarding different causes of death, and a statistically significant dose-response trend for degree of radiation exposure has been obtained for multiple myeloma, although this was based on few exposed cases.b" A survey of all cohorts of persons exposed to radiation for which data on cancer related mortality were available showed an excess of multiple myeloma in most cohorts.1Z An association between farming and multiple myeloma was first found in a case-control study from Washington State.'' Several studies with different methodologies and from different countries have confirmed this finding,lGZ6but some other recent epidemiological studies do not support the association.n2s In a case-control study from the middle of Sweden a statistically significant increased crude rate ratio (RR) of 1.9 for farming was obtained, but in further analysis with multiple regression the RR decreased to 1.4, which was not significant." Some studies have investigated further the possi- ble connection between farming and multiple myeloma by trying to identify specific risk factors. Thus some American studies with different methodologies indicate an association between exposure to pes- *'ticides and multiple myelomai6 y, 'I but no firm conclusions could be drawn concerning a n y particular class of pesticides responsible for the increase in risk. A study from New Zealand did not find any significant differences between cases and controls regarding potential exposure to phenoxy herbicides, chlorophenols, or any other agricultural chemicals, whereas significantly raised RRs were found for sheep farming and exposure to beef cattle.'' Exposure to asbestos has been connected with % Eriksson, Karlsson multiple myeloma in two recent case-control studies,2' but by contrast a recent American study did not find any such association.u Furthermore, an American multicentre population based case-control study did not find any correlation between jobs with probable exposure to asbestos 7nd multiple myeloma.= Benzene is an established risk factor for acute leukaemia, and multiple myeloma and other haematological malignancies have been associated with exposure to benzene or other organic solvents.p3' 33 Occupational exposure to petroleum products gave an increased risk for myeloma in a case-control s t u d y as well as in a cohort study," both from the United States. Other chemical exposures associated with observed increases in incidence or mortality in multiple myeloma in single studies encompass arseni~,~le"ad," carbon monoxide," and engine exhausts and creosote." Some specific occupations with potential exposure for different chemicals have also been identified as entailing an increased risk of myeloma-for example, hair- dressers and cosmetologist^,^'^^ machinists," and leather workers." Some studies have connected multiple myeloma with the furniture indu~try'"~ or carpentry." Furthermore, a Swedish case-control study found woodworkers in contact with fresh wood to be at risk for myeloma." The possiblecarcinogenic effectof electromagnetic fields has been much debated in recent years as a result of some epidemiological studies. A cohort study from the western United States investigating mortality in amateur radio operators exposed to electromagnetic fields showed a statistically significant increased mortality from malignancies of lymphatic tissues, including multiple myeloma and non-Hodgkin's lymphoma." The present study investigated those suspected risk factors for multiple myeloma, in a high incidence area of Sweden. Special emphasis was placed upon possible hazards in farming, such as contact with pesticides and different domestic animals. Materials and methods CASES Ail persons living in the four most northern counties of Sweden who were diagnosed as having multiple myeloma and were reported to the Swedish Cancer Registry between 1 July 1982and 30 June 1986 were eligible for inclusion in this study. In Sweden all physicians are obliged to report all incident cases of malignant tumours to the Cancer Registry. In this way 293 patients with a diagnosis of myeloma were identified. The medical records of these patients were scrutinised to confirm the diagnosis, and all cases who could not fulfil widely accepted diagnostic criteria proposed for use in therapeutic trialsu were excluded. A total of 18 patients did not meet these criteria. The remaining 275 patients constituted the cases; 156 (57%) of these were alive and 119 (43%) were deceased. No restrictions regarding age or sex were used. Of all cases 141 were men and 134 were women. CONTROLS For every living case two eligible control persons were extracted from the Swedish National Popuia- tion Registry. This was done through a matching procedure taking into consideration age, sex, and county. In Sweden everybody has a unique personal identity number composed of date of birth and four additional figures. The two persons with same sex, living in the same county and being closest in personal identity number-that is, in age-to the respective case were selected, and from these two, one was randomly chosen as the control person. For each deceased case, one deceased control was selected in an equivalent way from the National Registry for Causes of Death. The same matching criteria as for the living cases were used, and, M e m o r e , the deceased controls were matched on year of death. For reasons of humanity, persons who committed suicide were not used as controls. The study thus involved 275 cases and 275 controls, in all 550 persons. ASSESSMENT O F EXPOSURE An extensive questionnaire was mailed to all living subjects in the study. To obtain data concerning deceased persons the questionnaire was sent to the next of kin defined in the order of husband or wife, child, parent, sibling, or other. The specific nature of the investigation was not disclosed, and there was no reference to the special disease under study. Instead, the general information given was that different factors of potential importance for health were studied. T h e questionnairescontained several questions regarding the persons' various jobs over the years, as well as time and place for employment, and work associated exposure to chemicals and other potential carcinogens. It also included questions about smoking habits, residential building materials, Vicinity to electrical power lines, and leisure time contact with animals, electrical equipment, and Chemicals. At the end of the questionnaire were some pages with questions concerning diseases, medications, and other treannents, as well as an inquiry about hereditary circumstances. These pages were detached and filed by a secretary for later analysis when the questionnaires were returned from the subjects, so as not to disclose whether the subject was a case or a control. That particular information is not considered further in this presentation. After this all questionnaireswere handed over to a Occupational anG specially train whether the F interviewer SCI ted them by re tions if any da - subjects, or ne answer the qut interview, whit about disease, avoided interv -the subject wa: All persons were carefully livestock or ust more, regardle persons who ' forestry, or hc pesticides havt were phoned ti ment of exposu all persons w employed in sa7 been used to impregnates, 1 railroad emban and specify an): After the qu the interviewel personal identi removed, thus answers by one Some speciz account when a in earlier epide graphical area working day wa herbicides and I If the inform biocides was lac strong suspicio exposed to pesti type of substant Exposure to I occupations, ar work and hobk done for these t Enquiries ab( ings inhabited analysis, the sc they had ever lii question. W o o plaster and bric According to est building mater concrete, and analysed by age Plaster and bric Only subject Enksson, K a r l s s a A total of 18 The remaining 6 (57%)of these ised. No restricOf all cases 141 control persons 'ational Popula~ g ah matching 1 age, sex, and unique personal f birth and four with same sex, eing closest in in age-to the iom these two, rol person. sed control was n the National same matching ere used, and, .ere matched on ty,persons who i controls. T h e 5 controls, in all led to all living lata concerning was sent to the usband or wife, Jecific nature of nd there was no study. Instead, that different )r health were :d several quesj jobs over the iployment, and cals and other ided questions :ding materials, id leisure time luipment, and :re some pages edications, and inquiry about iges were de- .analysis when n the subjects, :t was a case or In is not con- mded over to a Occupational and other environmenmlfactors and multiple myeloma: a population based case-control study 97 specially trained interviewer who did not know living less than 1 km from electric power lines were whether the persons were cases or controls. T h e regarded as exposed to an electromagnetic field. interviewer scrutinised all answers and supplemen- The above mentioned criteria for different ted them by telephone according to written instruc- exposures were all decided before the coding of tions if any data were unclear or incomplete. Some exposure. subjects, or next of kin, were unable or unwilling to answer the questionnaire, but accepted a telephone STATISTICAL METHODS interview, which did not then include the questions For calculation of the RR the maximum likelihood about diseases, treatments, and heredity. This estimate was used, both in univariate and multi-' avoided interviewer bias by not disclosing whether variate models. The procedures "match" and the subject was a case or control. "logreg" in the statistical computer program All persons who had been working in agriculture "Epilog" was used for the univariate matched were carefully questioned if any answers about analysis and the multivariate logistic regression res- livestock or use of pesticides were missing. Further- pectively. The 90% confidence intervals (90Yo CIS) more, regardless of the answers in that respect all were obtained with the approximative procedure persons who had been working within farming, described by Miettinen" in the "match" procedure, forestry, or horticulture during the period when and with the Newton-Raphson interactive technique pesticides have been used in Sweden (since 1945) in the "logreg". In all analyses the matched pairs were phoned to ensure a good and uniform assess- were kept together. ment of exposure to such chemicals. In the same way All occupations as well as specific exposures were all persons who, during this period, had been analysed in univariate analyses. Multivariate logistic employed in saw mills, where different biocides have regression was then used to evaluate the importance been used to protect against blue stain and as of different exposure factors. Occupations were not impregnates, or employed as roadworkers, or included in the multivariate model, as these were railroad embankment workers were phoned to verify strongly connected with the specific exposure vari- and specify any exposure to pesticides. ables-for example, farmers with domestic animals. After the questionnaires had been completed by The following criteria were used to choose the the interviewer, the front page, including name, exposure variables to be included in the multivariate personal identification number, and address, was model: removed, thus enabling a blinded coding of the The RR in the univariate analysis had to be either answers by one of us (ME). ( a ) statistically significant with a 90% CI not Some special considerations were taken into including unity, or (b) 2 1.5 or G0.67(1/1-5), and account when assessing different exposures. Thus as representing at least 10 exposed subjects. in earlier epidemiological studies in the same geo- The variables should not cover other specific graphical area a minimum exposure of one whole exposure variables. Thus the variable "all pes- working day was used for exposure to phenoxyacetic ticides", which covers different specificpesticides, as herbicides and the related chlorophenols. well as the variaoles with combined types of building If the information about type or trade names of materials mentioned above, were excluded. biocides was lacking or unsure, even if there was a All variables that could meet these criteria were strong suspicion, the subject was only regarded as thus analysed in a multivariate model, in which all exposed to pesticides in general and not to any specific factors with a risk estimate between 0.9 and 1.1 were type of substance. gradually excluded. Also, to check any potential Exposure to organic solvents is common in some confounding from variables not primarily included occupations, and also occurs in some leisure time in the model, the Pearson-correlation coefficient work and hobby activities. Separate analyses were among the referents was used.50Thus factors with an done for these two types of exposure. absolute value of the correlation coefficient 20.25 Enquiries about the building material of all dwell- and with at least 10subjects exposed were considered ings inhabited by the subjects were made. In the as relevant, and they were included in the model after analysis, the subjects were regarded as exposed if the initial exclusion steps (table 5). they had ever lived in a house built of the material in question. Wood, stone, concrete, light concrete, plaster and brick were thereby analysed separately. Results According to estimations of y-radiation from different Of the 312living subjects, 12refused to participate in building materials in Swedish dwellingsa stone, the study, and of the 238 deceased, no answers were concrete, and light concrete houses were also obtained from the relatives of seven persons. T h e analysed by aggregation, and the same was done for refusal rate was consequently 345%. six of the plaster and brick houses. refusers were cases and 13 were controls. Thus 19 Only subjects who reported that they had been subjects and their 19matched counterparts could not 98 Eriksson, Karlsson Table I Occupariuns and exposures with a staristically s i p j i c a n t increased or decreased RR in univariate analysis results in earlier studies on multiple myeloma are presented. Occvpathl exposurefactor No of exposed catu No of exposed controls RR (90% CI) Among occupations only farmers showed an increased risk for myeloma, whereas electricians seemed to be negatively associated with the disease. Exposure to any pesticides resulted in a signifi- Occupation: Fanner 151 121 1.68 (1.23-2.33) cantly increased risk, which was further augmented Elmrician 3 9 0.33 ( 0 ~ 1 1 4 ~ 8 1 ) when only phenoxyacetic herbicides were con- Exposure factor: Power saw Asbestos 40 19 sidered. Exposure to dichlorodiphenylrrichloro- 27 33 1.57 (1.01-2.48) 0.50(0,29482) ethane (DDT) also gave an increased risk for Organic solvents (occupational) Pesticides, all 21 81 myeloma. On the other hand, no increased risks were 37 59 030 ( O . M . 8 0 ) 1.55 (1.15-2.26) noted for exposure to chlorophenols, creosote, Phenoxy acids 20 9 2.22 (1.15-4.66) arsenic, or mercurial seed dressing. Each and every DDT 53 Copy machines 6 Electromagneticfields 32 14 1.75 (1.19-244) 043 (0.19-0.87) other type of pesticide was just mentioned sporadically, and none gave a significant RR. (occupational) (31ftJe . Horse 6 152 137 15 118 101 0.31 (0.124-66) 1.72 (1.27-2.36) Increased risks were also noted for exposure to 1.78 (I .32-2.45) different domestic animals-namely, cattle, horses, Hogs Sheep Gopts 124 99 1.49 (1.11-2.03) hogs, sheep, and goats-but not for poultry or any 60 24 41 11 1.66 (1.12-2.49) 2.18 (1,204.24) other animals. Exposure to asbestos gave, contrary to the results of some other studies, a decreased risk for myeloma. be used in the analysis, which deals with the remain- Occupational exposure to organic solvents also ing 512 subjects or 256 matched pairs. UNIVARIATE ANALYSIS Table 1shows the result of the univariate analysis for occupations and exposures with a statistically significant increased or decreased risk. In tables 2 and 3 other occupations and exposure factors with at least 10 exposed subjects or of a special interest because of Table 2 O c n r p a r h without any sraristically significant differenas between c a u s and controls in univariare analysis. Only variables with at least 10 exposed subjecrs or of special interesr due IO findings in earlier srudies are incluied Occupation No of No of exposed exposed cases controls RR (90% C I ) Railrood worker 7 I 1 0.50 (0.18- 1.22) Rcad worker 19 17 1.14 (0.63- 2.12) Motor vehicle driver 17 17 1.00 (0.51- 1.97) Buildkg metal workcr/mechPnic* 7 11 0.64 (0.29- 1-34) Orher metal workt 4 8 0.50 (0.18- 1.22) Miner 8 8 1.00 (0.44-2.31) Surf.cc 6 4 1.50 (0.52- 4.83) Underground 8 7 1.14 (0.49- 2.77) Gudcner/garden worker 4 7 0.57 (0.20- 1.46) LWM* 82 69 1.48 (0.98- 2.27) AiYornunon worker 34 25 1.41 (0.89- 2.27) Sawmill worker 21 I8 1.19 (0.68- 2.11) Buildkg worker 20 22 0.90 (0.53- 1.53) Furniture industry worker 2 4 0.50 (0.12- 1.72) Pulp industry worker 10 10 1.00 (0.44-2.31) Laundry worker/ laundress Hairdresser/ 6 5 1-20 (0.44-3.44) counetologist Petrol station/ 2 3 0.67 (0.15- 2-71) guagc worker Rdinery worker other chemical work 6 1 8 7 0.86 (0.34-2.12) 1 1.00 (0.1&10~06) 7 1.14 (0.4% 2.77) *hchine-rcpainnan, welder, and plumber included. tBlac)rsmith,turner, and founder included. Table 3 Exposure factors without any statisrically significant differences between cases and conrrols in univariare analysis. Only variables wirh at leasr 10 exposed subjecrs or of a special inrerest due to findings in earlie; studies are Exposure factor No of exposed cases No of exposed controls RR 190% CIj Welding 3 7 0.43 (0-14-1.12) Tractor driving 19 15 1.40 (0-71-2.89) Sulphite pulp 6 8 0.67 (0.23-1.81) Chlorophenols 20 19 1.06 (0.60-1.90) Organic solvents (leisure time use) 43 37 122 (0.8&1-89) Man made mineral fibres 14 13 1.08 10.56-2.13) AncniC 4 3 1.33 io.38-5.21j Creorotc 4 5 0.75 (0.21-2-51) Mercurial seed dressing 5 3 1-67 (0-50-644) ElKnomPgnetic fields: Leisure rime exposure 6 5 1.20 (0.44-344) Amateur radio operator 2 2 1-00 (0.19-5.37) Power line near home 40 42 0-94(0.64-1-40) Display monitor unit 3 5 0.60(o.iai.8o) Poultry 2: 86 74 1-24 (0.91-1.71) 137 140 0.96 (0.72-1.27) I12 106 1.11 (0.82-1.51) Birds 9 15 0.50 (0-22-1.05) Building material (home): Wood 249 248 1.14 (0.49-2.77) Light concrete 20 17 1-20(0.68-2.17) Concrete 48 60 0.76 (0-53-1.07) Brick 45 56 0.77 (0.53-149) Plaster 12 14 0.86 (0.45-1.63) Stone 17 I9 0.89 (0.52-1.54) Light cuncrete, concrete, stone 72 88 0.74 (0.53-1.01) Plaster, brick 51 63 0.77(0-55-1 .OS) Radioactivity in work 2 3 0.67 (0-15-2.71) Smoking 106 117 0.80 (0-57-1.11) Fresh wood' 92 81 1.48 (0-95-2.35) Engine exhaustst 61 50 1-38 (0.92-2-09) ~~ *Working with power saws, at a sawmill, or at a pulp industry. tWorking with tractor, power saws, or as drivers Occuparional and 0, Table 4 Dose-resp< srgntficanr deviated I done I rejer to days o Exposure factor ~ Phenowyacetic acids Ouanic solvents (won Asbestos DDT Cattle Horses Hogs Sherp Goats seemed to be ne; but leisure-time L Exposure to el, which is the case f' some ocher OCCUF whereas residenr produced a risk ar No statistically from any type of from occupational as an amateur rad; cosmetologist, or ! For some expc animals, and resi, information on d dose-response an Table 5 Multivaria +Ernom. Univariat, J S well as 90% CISOJ excluded. The Pearso model (see rext) Cattle Horses Hogs Sheep Goats Poultry (its Dogs Hirds Phenoxyacetic acids (:hlorophenols DDT Organic solvents (work Askstos Sulphite pulp Power saw (>IPS machine Elcctromagneticfields ( . Fresh wood. L g i n e cxhausts' .Man made mineral fibrr Hrickt 'hfined as in table 3 'Building material in dv Enkson, Karkson :iple myeloma are mers showed an ereas electricians with the disease. ilted in a signifiirther augmented cides were coniphenyltrichlorocreased risk for :reased risks were ienols, creosote, . Each and every just mentioned icant RR. for exposure to y, cattle, horses, r poultry or any try to the results sk for myeloma. 2 solvents also itistically wrok in univariate :posed subjects or 'rstudies are RR f 90% CI) 0.43 (0.14-1.12) 1.40 (0.71-2.89) 0.67 (0.23-1.81) 1-06 (0.60-1.90) 1.22 (0.80-1.69) 1.08 (0.56-2.13) 1.33 (0.38-5.21) 0.75 (0.21-2.51) 1-67 (0.50444) I .20 (0.44-344) 1.00 (0.19-5.37) 0.94 (0.64-140) 0-60( o . i a i . 8 o ) 1.24 (0.91-1.71) 0 9 6 (0.72-1.27) 1.11 (0.82-1.51) 0.50 (0.22-1.05) 1.14 (0.49-2.77) 1.20 (0.68-2.17) 0.76 (0.53-1.07) 0.77 (0.53-149) 0.86 (0.45-1.63) 0.89 (0.52-1.54) 0-74 (0.53-1.01) 0-77 (0.55-1 Q8) 0.67 (0.15-2.71) 0.80 (0.57-1.11) 1-48 (0.95-2 35) 1-38 (0.92-249) Ip industry. Occupational and orher enuironmenral factors and multiple myelmna: o population based case-control study 99 Toble 4 Dose-response calcularion on exposure facrors with reliable informarion abour durarion of exposure and statisticnlly significant deviared risks in univariate analysts aspresented in table I . The categories of exposure (decided before analysis was done) refer to days of work with phenoxyaceric acids and D D T and years of exposurefor the rest of rhe factors Exposure factor Phenoxyacetic acids Organic solvents (work) Asbesros DDT cattle Horses Hogs Sheep Goats Duration of exposure ~5 ( R R 190% C I ) ) 3.00 (0.76-1 1.49) 0.37 (0.09-1.48) 0.10 (0.02- 0.66) 1-08 (0.41- 2.81) 1.69 (0.68-4.20) 1.87 (0.63- 5.53) 2-10 (0.65- 6.75) -1-39 (0.55- 3.54) 6-20 IRR (90% C I ) ) 2.00 (0.63-640) 0.59 (0.28-1.25) 0.21 (O~Oa-oT56) 1.54 (0-86-2.77) 1.30 (OK?-2.05) 1.17 (0.73-1.86) 0 9 7 (0.60-1.54) 1.22 (0.63-2.38) 1.00 (0.34-2.59) 2 2 1 IRR 190% G I ) ) 200 (0.7S5.48) - 0.37 (0.16474) 0.58 (0.26-1.31) 1.61 (0.6H.12) I 4 5 (1.3s2.57) 1.86 (1.33-259) 1.56 (1.12-2.19) 2-13 (1.30-394) 3.20 (1-38-7.43) seemed to be negatively associated with myeloma, presents the results of this calculation for exposure but leisure-time use gave a slightly increased risk. factors with a statistically significant increased or Exposure to electromagnetic fields during work, decreased risk in the univariate analysis (table 1). which is the case for electricians and for linemen, and Exposures to several domestic animals showed some other occupations, also gave a decreased RR, statistically significant increased RR only at the whereas residential exposure from power lines highest dose. This indicates a dose-response pattern produced a risk around unity. and further suppons an association between these No statistically significant increased risk resulted animals and multiple myeloma. from any type of residential building material, or from occupational exposure to radioactivity, activity MULTIVARIATE ANALYSIS as an amateur radio operator, work as hairdresser, or In the multivariate analysis several exposure factors cosmetologist, or smoking. were included according to the criteria described. For some exposure factors such as chemicals, Table 5 presents the stepwise reduction with the animals, and residential building materials, where resulting values of RRs. The 90% CISof the R R s in information on duration of exposure was good, a the last step are also given. dose-response analysis was performed. Table 4 Exposure to hogs and sheep were eliminated as risk Table 5 Multivariace analysis of different exposurefacrors chosen according to rhe criteria described in STATISTIW METHODS. Univariate risk rarios (URRs) are given, and a stepwise reduction with resulting values of risk ratios is presented as well as 90% CISof the values in the reduced model. AI1 variables un'th a risk esrimare between 0.9 and 1.1 weregradually excluded. The Pearson-correlarion coegicient w a s used to includc potential confoundingfactors not primarily incIuded in the model (see text) Exposure factor Cattle Horses Hogs Sheep Goats Poultry GCS Dogs Birds Phcnoxyacetic acids Chlorophenols DDT Organic solvents (work) Asbestos Sulphire pulp Power saw Copy machine Elecuomagnctic fields (work) Fresh wood* Engine exhausts' Man made mineral fibres Brickt ' D e h e d as in table 3. t6uilding material in dwelling URR 1.72 1.78 1.49 1.66 2.18 I .24 1.11 0.96 0.50 2.22 1.06 1.75 0.50 0.50 0.67 1.57 0.43 0.3 1 1.48 1.38 1 -08 0-77 Step No I 1.52 1.45 0.71 0.95 1.37 0.73 1.60 I .29 0.61 0.66 0-87 0.88 0.55 0 32 2 1.51 1.45 0.70 1.33 0.73 1.60 1.29 0.6 1 0.67 0.88 0-87 0-54 0.32 3 1.54 1.66 0.84 1.50 0.68 0.91 0.80 0.69 1.87 0.86 1.42 0.68 0.52 0.91 0.78 0.59 0.37 0.90 1.11 2.00 0.91 4 1.48 1.64 0-85 131 0.68 0.79 0.69 1.92 0.85 I 40 0.67 054 0.75 0.55 0.36 1.11 2.00 (90% CI) (0.76-2.91) (0.9S2.91) (0.42-1.69) (0.75-3.03) (04&1.15) (0.56-1.IO) (0-27-1-77) (0.84-4.36) (0-40-1 .80) (0-88-2.25) (0.38-1.21) (0.27-1.08) (0.36-1.58) (0.23-1.33) (0.13499) (060-205) (0834.82) 100 Eriksson, Karlsson Table 6 Some exposurefactors within farming and forestry. Only matched pairs with both subjects being occupied in those industries (n = 94) were included in the analysis Exposure faclor RR 190% C I I Canlc Hones Hogs Shetp Goats Poulvy Phcnoxyacetic acids Chlorophenols DDT Mercurial seed dressing Paricides, orher Tractor driving (fanning) 2.00 (0.954.70) 1.64 (0.94-2.99) 1.00 (0 59-1.71) 1.26 (0.76-2.13) 1.71 (0.78-4.06) 1.04 (0.65-1 68) 2.00 (0.88-5.07) 1.14 (0.49-2.77) 1.86 (1.08-3.35) 1.00 (0.19-5.37) 0.71 (0.27-1.79) 2.25 (0.84-7.11) factors for myeloma in the multivariate setting, whereas exposure to cattle, horses, and goats seemed to remain important although U s were weakened. Phenoxyacetic herbicides and DDT also remained as risk factors, but with slightly decreased RRs. T h e significantly decreased risk Connected with exposure to electromagnetic fields during work remained. Exposure to organic solvents and asbestos was still negatively associated with myeloma but with less pronounced decreased RRs. Some exposure variables were introduced in the model as they were fulfilling the correlation criterion described above, but of these only man made mineral fibres gave an increased risk. For further details on thismultivariate analysis see table 5. In another approach to study the importance of M e r e n t specific factors within farming and related work, a separate analysis was done including only the pairs withboth subjects a l k n i n g work infarming or forestry. A total of 94 pairs were used in this analysis. It confirmed the mentioned domestic animals and pesticides as possible risk factors for multiple myeloma (rable 6). Discussion This study was performed to further evaluate risk factors for multiple myeloma described in other studies. T h e Swedish compulsory reporting system for malignant diseases makes it likely that almost all incident cases in the study area during the four years of inclusion were utilised. A matched study design controlling for age, sex, county, and vital status was used, thereby increasing effcacy in the analysis. To reduce the possible influence of recall bias-for example, a supposed tendency for the cases to remember more hazardous exposures than controls -the aim of the study was unexpressed. Thus no allusion to the disease under study was made in the information to the subjects. Furthermore, the questionnaire asked for information on many different occupations and exposures not specially focussing on any debated risk factors. T o minimise theoretical observational bias during telephone interviews, any data disclosing whether the subject was a case or a control were removed from the questionnaires before they were completed over the phone. The final coding of the answers was also done blinded in this respect. In the analyses univariate as well as multivariate methods were used. To our knowledge this is the first study using multivariate analysis to dissolve different specific risk factors for myeloma among farmers, although some studies have used this method to evaluate farming v other risk factors outside farming"` or e, pesticides specifically." The study confirmed the earlier finding of farmers being at an increased risk of developing multiple myeloma. Contact with different domestic animals, and exposure to some pesticides (phenoxyacetic herbicides and DDT), were associated with multiple myeloma, thus giving support to some theories regarding the specific risk factors at hand in farming. Oncogeniczoonotic viruses have been suggested as possible aetiological agents in haematopoetic and lymphatic malignancies in human subjects, as RNA viruses have been connected with such turnours in several mammalian species5''* It is also interesting to consider a viral disease in mink, Aleutian mink disease, characterised by systemic proliferation of plasma cells associated with marked hyper-;globulinaemia, with some of the affected minks also developing a myeloma-like 7-globulin spike. This disease has been found in all of the mink raising areas of Scandinavia and North America. Furthermore, a case of Aleutian mink disease in man has been describedY5a's well as a case of multiple myeloma in a mink handler who, during many years, had been bitten repeatedly on the hands and arms by mink in a farm where the presence of the viral disease was well documented.%In the present study, one case and one control were working with mink raising, and another case worked on a fox farm. In our study the univariate analysis supported cattle, horses, hogs, sheep, and goats as risk factors for myeloma. This was further supported by doseresponse calculations. T h e multivariate analysis, however, eliminated hogs and sheep, and decreased the RR somewhat for horses, goats, and cattle. For cattle these results are in accordance with the findings in a study from New Zealand," whereas horses and goats have not been reported as risk factors for myeloma in any earlier study. The mechanism responsible for this connection between certain domestic animals and multiple myeloma may well be some unknown viral transmission. Interestingly, a virus has been found to be an imponant aetiological agent in a bovine type of lymphosarcoma prevalent in Occupational and On the othe. support for an: myeloma, an as' sSotamteesu.In6i''var'i'ate The increase supported resul! m which no pal found responsit: both phenoxyacc tically significan orna in the univa decreased somev multivariate ana No earlier studir risk factors for have, however, d risk factors fc Myeloma and ly erative diseases 1 some families` an as already discu: thesis in this stu herbicides might oma. Consequen increase found w phenoxy herbicit and furans, whic The dioxin ison carcinogen in ani1 to have this pro1 study indicated tl cides and chlorop . tissue sarcoma ir. effect by the ph cannot be excludl DDT is a pote: cancelled in the U in 1975, because particularly on p~ is a stable substai tissue where it is enzyme system oi is also establishe neoplasms as we1 tunours have ai Some e] possible associati1 malignant lymph mia in human SUI to the indication o in our study, a f i l pesticides such a connected to my' findings were nor Occupational c organic solvents, studieshave been i E ~ i h s o nK, arlsson ecially focussing on .ational bias during disclosing whether Nere removed from rre completed over e answers was also .ell as multivariate cdge this is the first o dissolve different a among farmers, d this method to tors outside farm- finding of farmers veloping multiple domestic animals, 's (phenoxyacetic ited with multiple 10 some theories c hand in farming. been suggested as iematopoetic and subjects, as RNA such tumours in alsointeresting to , Aleutian mink ' proliferation of larked hyper-;.rected minks also ulin spike. This link raising areas . Furthermore, a man has been ple myeloma in a years, had been rms by mink in a disease was well one case and one ing, and another !lysis supported :s as risk factors ported by doseariate analysis, , and decreased and cattle. For ~ i t thhe findings reas horses and .isk factors for nechanism rescertain domescy well be some tingly, a virus :iological agent a prevalent in Occuporronal and other environmenral factors and multiple myeloma: a population based case-cnntrol stua'y 101 On the other hand, this investigation gave no multiple myeloma, were both negatively associated support for any association between poul&y and with myelo-ka in this study, thereby opposing any myeloma, an association that has been suggested in idea of association between these subsmces and some univariate correlation studies from the United multiple myeloma. States.I6Is yI 57 In the univariate, but not in the multivariate The increased RR for exposure to pesticides analysis, work with power saws gave a statistically supported results of an American case-control study significant increased risk for myeloma. A non-sig- in which no particular class of pesticides could be nificant risk was also seen for tractor driving. These found responsible, however." In our present study risks are reflected by the slightly increased risk for both phenoxyacetic herbicides and D D T gave statis- occupational exposure to engine exhausts found by tically significant increased RRs for multiple myel- univariate analysis. An association between engine oma in the univariate analysis, and although the risks exhausts and multiple myeloma was first suggested in decreased somewhat, and were non-significant in the an earlier Swedish study," in which the possible role multivariate analysis, they both remained increased. of the known mutagenic agents in both diesel and No earlier studies exist indicating these pesticides as gasoline exhausts was emphasised. risk factors for multiple myeloma. Some studies In the same study" woodworkers in contact with have, however, disclosedphenoxyacetic herbicides as fresh wood were found to be at increased risk for risk factors for non-Hodgkin's lymphoma.% myeloma, which prompted us to evaluate the risk for Myeloma and lymphoma are related lymphoprolif- people occupied in forestry, at a sawmill, or at a pulp erative diseases with Occurrence of both diseases in industry combined. The moderately increased risk some families' and a possible common viral aetiology found in the univariate analysis was eliminated when as already discussed. Thus it was an a priori hypo- logistic regression was performed. thesis in this study that exposure to phenoxyacetic A protective effect of being an electrician, or being herbicides might be a risk factor for multiple myel- exposed to electromagnetic fields in any occupation, oma. Consequently, it is less probable that the risk was a constant finding with statistical significance increase found was due to chance alone. Some of the even in the multivariate analysis. Display monitor phenoxy herbicides are contaminated with dioxins units may generate weak electromagnetic fields, and and furans, which can be highly toxic substances. their influence was studied separately. In this case The dioxin isomer 2,3,7,8-TCDD is a recognised also, exposure seemed to lower the risk for multiple carcinogen in animals-'and other isomers also seem myeloma, but the exposed subjects were few. On the to have this A recent epidemiological other hand, the RR was close to unity regarding study indicated that dioxins in phenoxyacetic herbi- exposure to electromagnetic fields from power lines cides and chlorophenols might be a risk factor for soft near the subjects' homes, with exposure defined as tissue sarcoma in human subjects.65A carcinogenic living less than IO00 m from the power line. T h e risk effect by the phenoxyacetic herbicides themselves was not substantially influenced by the distance cannot be excluded, however. between home and power line (100 m and 500 m as D D T is a potent insecticide, the use of which was criteria). No physical or biological explanation is cancelled in the United States in 1972,and in Sweden fonhcoming for the decreased risk seen in relation to in 1975, because of its harmful ecological effects, occupational exposure to electromagnetic fields. particularly on predatory birds and on marine life. It Ongoing studies-for example, in Sweden-will is a stable substance and may accumulate in human hopefully shed more light upon the possible associ- tissue where it is a potent inducer of the microsomal ation between electromagnetic fields and different enzyme system of the liver. Carcinogenicity of D D T types of malignant tumours. is also established in mice, causing liver and lung In conclusion, this case-control study confirmed neoplasms as well as malignant lymphoma.* Liver the association between farming and multiple myel- tumours have also been found in D D T treated oma. Within this occupation the domestic animals, a.t Some epidemiological studies indicate a cattle, goats, and horses, and two types of pesticides, possible association between exposure to D D T and phenoxyacetic acids and DDT, were indicated as risk malignant lymphoma or chronic lymphatic leukae- factors. Occupational exposures to electromagnetic mia in human s u b j e ~ t s . ' ~ 'T~'..IS may lend support fields, asbestos, and organic solvents should be to the indication of D D T as a risk factor for myeloma further considered with regard to whether or not in our study, a finding not described earlier. Other some mechanism could explain the decreased risk for pesticides such as creosote and arsenic have been myeloma for those exposures, which appeared in this connected to myeloma in some studies, but these study. findings were not verified in our present study. Occupational exposure to asbestos as well as organic solvents, which in some, but not all, earlier This study was supported by grants from the studieshave been identified as possible risk factors for Swedish Cancer Fund (project 2683-B90-02X). 102 Erikssa, Karlsson We thank Anna-Lena Zakari for performing ex- cellent telephone interviews. We are also grateful to Monica Sandstrom and Ingemar Majholm for professional assistance in the data collection, and to Professor Olav Axelson for valuable viewpoints and constructive criticism. Requests for reprints to: Mikael Eriksson, MD, Department of Oncology, University Hospital, S-901 85 UmeH, Sweden. 1 Maldonado JE, Kyle RA. Familial myeloma. Report of eight fvnilies and a study of serum proteins in their relatives. A m J Med 1974;57:875-84. 2 Shoenfeld Y, Shaklai M, Berliner S, Gallant LA, Pinkhas J. Familial multiple myeloma: A review of 37 families. Postgrad Med J 1982;58:12-6. 3 Bourguet CC, Grufferman S,Deficll E, DeLong ER, Cohen HJ. Muluple myeloma and family history of cancer. A case-control study. Cancer 1985;562133-9. 4 Eriksson M, Bergstrom I. Familial malignant blood disease in the county of Jiuntland, Sweden. Eur J Haemarol 1987;38:241-5. 5 Schull WJ. Atomic bomb survivors: Patterns of cancer risk. In: Boice Jr JD, Fraumeni Jr JF, eds. Radirrtion carctnogenesis: epidemiology and biological significancr. New York: Raven Press, 1984:21-36. 6 Finch SC. Leukaemia and lymphoma in atomic bomb survivors. In: Boice Jr JD, Fraumvi Jr JF, eds. Radiation carcznogenests: epidemiologv and biological smzficance. New York: Raven Press, 19W.37-44. 7 Preston DL, Kato H, Kopecky KJ, Fujita S. Studies of the mortality of A-bomb survivors. 8. Cancer mortality, 19501982.Radial Res 1987;111:151-78. 8 Gilbm ES,Marks S.An analysis of the mortality of workers in a nuclear facility. Radiat Res 1979;79:12248. 9 Gilbert ES, Marks S. An updated analysis of mortality of workers in a nuclear facility. Radiar Res 1980;83:740-1. 10 Tolley HD, Marks S, Buchanan JA, Gilben ES. A further update of the analysis of mortality of workers in a nuclear facility. Radial Res 1983;95:211-3. 1 1 Gilben ES,Petersen GR, Buchanan JA. Mortality of workers at the Hanford Site: 1945-1981.Health Phys 1989;56:11-25. 12 Cuzick J. Radiation-induced myelomatosis. N Engl J Mcd 1981; 304:204-10. 13 MiIham JrS.Leukaemiaandmultiplemyelomainfarmers.A m J Epidnmol1971;94:307-10. 14 Agu W, Christensen BL, Buffler PA. Geographic patterns of multiple myeloma: Racial and industrial correlates, State of Texas, 1969-71.f Nail Cancer Inst 1980;65:735-8. 15 Burmeister LF. Cancer mortality in Iowa farmers, 1971-78. f NatlCancer Inst 1981;66:4614. 16 Burmeister LF, Everett CD, van Lier SF, Isacson P. Selected cancer mortality and farm practices in Iowa. A m J Epidemtol 1983;118:72-7. 17 Gallagher.RP, Spinelli JJ, Elwood JM, Skippen DH. Allergies and agricultural exposure as risk factors for muluple myeloma. Br J Cancer 1983;48:853-7. 18 Cantor KP, Blair A. Farming and mortality from multiple mveloma: A case-control studv with the use of death cenifioies. J Nail Cancer Insr 1984:72:251-5. 19 P a r c e NE,Smith AH, Howard JK, Sheppard RA, Giles HJ, Teague CA. Case-control study of multiple myeloma and farming. Br J Cancer 1986;54:493-500. 20 Nandakumar A, Armstrong BK, de Klerk NH. Multiple myeloma in Western Australia: A case-control study in relation to occupation, father's occupation, socioeconomic status and country of birth. Int J Cancer 1986;37:223-6. 21 Cuzick J, De Stavola B. Multiple myeloma-A case-control study. Br J Cancer 1988;57:516-20. 22 La Vecchia C,Negri E,D'Avanro B, Franccschi S.Occupation and lymphoid neoplasms. Br J Cancer 1989;60:385-8. 23 Boffetta P, Stellman SD, Garfinkel L. A case-control study of multiple myeloma nested in the American cancer sociery prospective study. In[ J Cilnrer 1989;43:554-9. 24 Alavanja MCR, Blair A, Merkle S,Teske J, Eaton B. Mortality among agricultural extension agents. A m J Ind Med 1988;14. 167-76. 25 Steineck G,Wiklund K. Multiple myeloma in Swedish agricul- tural workers. In[ J Eprdemiol 1986;15:321-5 26 McLaughlin JK, Liner MS, Stone BJ, et a/. Multiple myeloma and occupation in Sweden. Arch Environ Health 1988;43: 1-10. 27 Liner MS, Harlow SD, Mcbughlin JK. A case-control study of multiple myeloma in whites: Chronic antigenic stimulation, occupation, and drug use. Cancer Res 1987;47:2978-8I. 28 Brownson RC, Reif IS. A cancer registry-based study of occupational risk for lymphoma, multiple myeloma and Ieu- kaemia. Int JEpi&miol1988;17:27-32. 29 Flodin U, Fredriksson M, Persson B. Multiple myeloma and engine exhausts, fresh wood, and creosote: A case-referent study. A m J Ind Med 1987;12:519-29. 30 Saftlas AF, Blair A, Cantor KP, Hanrahan L, Andersson HA Cancer and other causes of death among Wisconsin farmers. Am J IndMed 1987;11:119-29. 31 Moms PD, Koepsell TD, D a l i g JR, er al. Toxic substance erDmure and multiole mveloma: A case-control study. J Nut1 Cdncer Inst 1986,7i:987:94. 32 Schwartz DA, Vaughan TL, Hcyer NJ, et al. B cell neoplasms and occuoational asbestos exmsure. A m J Ind Med 1988;14 661-71. 33 Rinsky RA, Smith AB, Hornung R, et al. Benzene and leukaemia. An epidemiologic risk assessment. N Engl J Med 1987;316:1044-50. 34 Lundberg I. Mortality and cancer incidence among Swedish paint indusuy workers with long-term exposure to organic solvents. S c a d J Work Emn'ronHealth 1986;12:108-13. 35 Thomas T L , Decoufle P, Moure-Eraso R. Monality among workers employed in petroleum relining and petrochemical plants. J Occup Med 1980;22:97-103. 36 Ott MG, Holder BB, Gordon HL. Respiratory cmcer and occupational exposure to arsenicals. Arch Environ Health 1974;29250-5. 37 Axelson 0,Dahlgren E, Jansson C-D, Rehnlund SO. Arsenic arposure and mortality: a case-referent study from a Swedish copper smelter. Br J Ind Med 1978;35:b15. 38 G r w e MH, Hoover RN, Eck RL, Fraumeni JF. Cancer mortality among printing plant workers. Environ Res 1979; 2066-73. 39 Guidoni S. Wright WE. P e t m IM. Multiule myeloma in cosmetologists.-Am J I d Med 19iI2;3:169-7i. 40 Spinelli JJ, Gallagher RP, Band PR, Threlfall WJ. Multiple myeloma, leukemia, and cancer of the ovary in cosmetologists and hairdressers. Am J Ind Med 1984;6:97-102. 41 Gallagher RP,Threlfall WJ.Cancer mortality in metal workers. Can Med Assoc J 1983;12911914. 42 von Wrken H, Vollmcr I. Die epidemiologic des multiplen Myeloms. Arch Geschmulsrforsch 1968;3111:1b38. 43 Walrath J,Decoufle P, ThomasT.Mortality among workers in a shoemufacturingcompany. Am J IndMed 1987;12:615-23. 44 Brinton LA, Stone BJ, Blot WJ, Fraumeni JF. Nasal cancer in U.S.furniture industry counties. Lancer 1976ji:628. 45 Tollcrud DJ, Brinron LA, Stone BJ, Tobaanan JK, Blattna WA.Momlity from multiple myeloma among No& Carolina furniture workers. J Natl Cancer Inst 1985;74:79-01. 46 MiIham Jr S.Increpxd mortality in amateur radio operators due to lymphatic and hematopoietic malignanaes. A m J Epidmriol 1988;127,1:50-4. 47 Committee of the chronic leukemia-myeloma task force, NCI: Proposed guidcliics for protocol studies. 11. Plasma cell myeloma. Cancer CIumorher Rep 1968;1:17-39. 48 Swedjemark GA. In-door measurements on natural radio- . anivity in Sweden. Stockholm: National Board of Radiation Protecrion, (SSI). 1979.Repon 1979-026. (In Swedjsh.) 49 Miminen OS. Estimation of relative risk from indwidually matched series. Biometrics 1970;2675-86. 50 Fredriksson.M, Hardell L, Brngtsron NO, Axelson 0. Con- founding in occupational studies. In: Hogstedt C, Reurcrwall C, eds. Progress in occupational epidemiology. Amsterdam: Elsevier Science Publishers, 1988:634. 51 Kaplan HS. From experimental animal models to human lymphoid tissue neoplasia: search for viral etiology. In: Mathc C, Seligmann M, Tubiana M, eds. Recmr results m cancer research. Lymphoid neoplasia. I : Clasnfication, cartgorisation, natural history. Heidelberg: Springer-Verlag, 1978;325-36. 52 Armenian HK, Hamaden RR. Epidemiology of non-Hodgkin's lymphomas. In: Lilienfeld AM, ed. Revinus in cancer cpidemiology.Vol 2.New York: Elsevier, 1983. 53 Chapman I, Jimenez FA. Aleutian-mink disease in man. N Engl JMed 1963;269:11714. 54 Henry LW. Multiple myeloma in a mink handler following exposure to aleutian disease. Cancer 1979;44:27?-5. 55 Ressang AA, Mastenbrook N, Quak J , ct al. Studies on bovine Occuparionai leukemi. lines. Zr 56 Kettmann leukemi;, Acad Scr 57 Priester 'ii' poultry J Nor1 C 58 Hardell L. lymphon solvents, study. Br 59 H o a r S K , t and risk 256:1141 60 Kociba RJ, cdhibreonnzico-1r, 279-303.' 61 Pitot HC,( evaluatior dioxin of 1 cer Res 19 62 van Miller cnhelooproladsihmG 63 National fox bioassay 0 1 Vancoi All manus should cor manuscrip (knownas T h e Br intematior accept arti Vancouver Med I,24 standardisc Referenc in the orde the text by occasion th other repor submitted i Eriksson, Karlsson I 41. Multiple myeloma 'viron Health 1988;43: A case-control study of antigenic stimulation, 987;47:2978-81. cistry-based study of ple myeloma and leu- (ultiple myeloma and )sate: A case-referent In L, Anderson HA. g Wisconsin farmers. t al. Toxic substance -control study. J Natl ' I al. E cell neoplasms n J Ind Med 1988;14: al. Benzene and Ieunent. N Engl J Med a c e among Swedish exposure to organic 1986;lZ:108-1 3. R. Mortality among g and petrochemical pintory cancer and lrch Emiron Htalth hnlund SO. Arsenic rudy from a Swedish .15. P ~ CJFN. Cancer . Emiron Res 1979; ultiplc myeloma in L71. elfall W].Multiple ry in cosmetologists 7-102. ty in metal workers. ogie des multiplen 1:18-38. among workers in a ed 1967;12:615-23. IF. Nasal cancer in 976;ii:628. m a n JK, Blatrner mg North Carolina 8;74:799-801. -adiooperators due es. Am J Epidemiol a cask force, NCI: i. 11. Plasma cell -39. >n natural ndioloard of Radiation :In Swedish.) from individually Axelson 0. Conedt C, Reuterwall ogy. Amsterdam: iodcls to human iology. In: Mathe ' rerultr in cancer m, carcgorisarion, 1,1978; 325-36. jf non-Hodgkin's vwws in cancer 13. e m man. N Engl andler following 273-5. tudies on bovine Occupational and other envtronmentaIjactors and multiple myeloma: a population based case-control study 103 leukernla. I . Establishment of type C v~rusproducing cell lines. Zcntralbl Veiennorrned 1 E ) 197431 302-17. 56 Ketrmann R, Portetelle M, Mammench M, et al. Bovine -leukemia virus: an exouenous RNA oncogenic virus. Proc Nail AcadScr U S A 1976;fi:1014-8. 57 Priester WA, Mason TJ. Human cancer mortality In relation to poultry population, by county, in IO sourheastem states. J Nail Cancer Insi 197453.459- 58 Hardell L, Eriksson~M,Lenner P, Lundgren E. Malignant lymphoma and exposure to chemicals, especially organic solvencs, chlorophcnols and phenoxy aads; A caw-control study. Br J Cancer 1981;43:169-76. 59 Hoar SK, Blair A, Holmes FF, et al. Agricultural herbicide use and risk of lymphoma and soft-tissue sarcoma. J A M A 1986; 256:1141-7. 60 Kociba RJ, Kcycs DG, Beyer J E , c i 01. Results of a two-year chronic toxicity and oncogenicitystudy of 2,3,7,8-tetrachlorodibenzo-p-dioxin in rats. Toxic01 Appl Pharmocol 1978;46: 279-303. 61 Pitor HC, Goldsworthy T, Campbell HA, et al. Quantitative evaluationof the promotion by 2,3,7,8,-terrachIorodiodibenm-pdioxin of hepatwrcinogennis from diethylnitrosamine. Canccr Rcr 1980;40:3616-20. 62 van Miller JP. Lalich JJ, Allen JR. Increased incidence of neoplasm in rats exposed to low levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin. Chosphcrc 1977;6:537+. 63 National toxicology program, National Cancer Institute. NIH bioassay of a rnisiure of 1,2,3,6,7,8 and 1,2,3,7,8,9-hcxachloro dibmzo-p-dioxmr for carnnogeninty (gavage siudy). (Natl Toxicol Program tech rep ser No 198; DHHS pub1 No (NIH)80-I98.) Research Triangle Park, N C Not1 Toxicol Program, 1980. 64 National toxicology program, National Cancer Institute. Toxicology and carcinogmerii studies of two porta-chlorophenol technical-grademixtures B6C3F, mice. (Natl Toxicul Program tech rep ser No 349; DHHS pub1 No (NIH)89-2#4.) Research Triangle Park, NC: Natl Toxicol Program, 1989. 65 Eribson M, Hardell L, Adami H-0. Erposure IO dioxins as a risk factor for soft tissue sarcoma: A population-based m e - -conuol study. J Natl C a w Inrr 1990;82:486-90. 66 Kashyap SK,Nigatn SK,Karnick AB, et al. Carcinogenicityof D D T (dichlorodiphenyluiclomcthPne)in pure inbrsd Swiss mice. Int J Cancer 1977;19:725-29. 67 Rossi L, Rpvera M,Repeni G, el al. Long-term administration of DDT orphenobarbital Na in Wistar rats. Int J C- 1977; 1917945. 68 G b d JRP,Hall RK,Rossi L, et al. Effectsoflong-term intakes of D D T on rats. T ~ r19i82;6%11-7. 69 Penson B, Dahlander AM, Fredriksson M,Noorlind Emge H, Ohlson C-G, Axelson 0. Malignant lymphomu md 0ccupaUonalcxposurcs. Br 1Ind Med 1989;46:516-20. 70 Flodin U, Frcdriksson M, Penson E, Axelson 0. Chronic lymphatic leukaemia and engine exhausts, fresh wood, and DDT: a case-referent study. Br J Ind hied 1988;45:33-8. Accepted 3 June 1991 Vancouver style All manuscripts submitted to the Br J Znd Med should conform to the uniform requirements for manuscripts submitted to biomedical journals (known as the Vancouver style). The Br J Ind Med, together with many other international biomedical journals, has agreed to accept articles prepared in accordance with the Vancouver style. The style (described in full in Br Med J, 24 February 1979, p 532) is intended to standardise requirements for authors. References should be numbered consecutively in the order in which they are first mentioned in the text by Arabic numeralsabove the line on each occasionthe reference iscited (Manson' confirmed other reports*-' . ..).Infuture references to papers submincd to the Br J Znd Med should include: the names of all authors if therc are six or less or, if there are more, the first three followed by et al; the title of journal articles or book chapters; the titles of journals abbreviated according to the style of Index Medicus;and the first and finalpage numbers of the article or chapter. Examples of common forms of references are: 1 International Steering Commince of Medical Editon. Uniform requirements for maauscripu submincd to biomedical journals. BTMcd J 1979;1:532-5. 2 Soter NA, Wasurman SI, Aurta KF. Cold urtiaria: relase into rhe circulation of histamine md coaino-phi1 chanoroccic f u t o r of anaphylaxis during culd challenge. N Enql J Med 1976;294:687-90. 3 Weinscan L.S w a m MN.Pathogenicproperties of invading micm+rganisma. In: sod- W A Jr, Sod- WA, cda. Pasholqicpliyrtology:wchmdnu of&-. Philadel- phip: W B Sound-, 1974457-72.