Document vyjwBKpj232GG5aQBOm3R8Laq

Scand J U'ork Envrron Healih 1992;18126-9 Cancer incidence among creosote-exposed workers by Sigvard Karlehagen, MD,' Aage Andersen, MD,2Carl-Goran Ohlson, MD3 KARLEHAGEN S, XNDERSEN A. OHLSON C-G. Cancer incidence among creosote-exposed workers. Scand J Work Envrron Heolfh 1992;18:26--0. Cancer incidence was studied among 921 creosoteexposed impregnators at 13 piants in Sweden and Vorway. The subjecrs had been impregnating wood (eg. railroad cross-ties and telegraph poles), but no data on individual exposures were available. The study population was resrricted to men employed during the period 1950-197'5, and their cancer morbidity was checked through the cancer registries. The total cancer incidence was somewhat lower than expected, 129 cases versus 137 expected [standardized incidence ratio (SIR) 0.94;. Increased risks in both countries combined were observed for lip cancer (SIR 2.50, 95% confidence intervai (9?0 CI) 0.81-5.83). skin cancer (SIR 2.37. 95% CI 1.08--jSOL and malignant lymphoma (SIR 1.9. 95mo CI 0.53-3.78). Exposure to sunlight may have contributed :O the risk of lip and skin cancer. The small number of cancer cases does not pennit valid conclusions. The findings indicate that impregnating wood with creosote in eariier decades increased the risk of skin cancer. Ke.v ierms: cohort study, impregnators, lip cancer malignant lymphoma, skin cancer. Creosotes consist of hundreds of compounds, and the main components identified are naphthalene and its alkyl homologues (1). Creosote oil is used as a preservative of wood. It is mostly used to impregnate railroad cross-ties and different :ypes o f poles (eg, telegraph poles). The exposure may pose a risk of cancer, direct skin contac: causing skin cancer (2. 3) and inhalation possibly increasing the risk for lung cancer (4). Informa- tion on such risks is available as case reports only. Case-referent studies have indicated increased risks for multiple myeloma (5) and for maiignant lymphomas (6). A Swedish study based on ccnsus data observed an association between urorheiial cancer and creosote (7).A review of the literature concluded, however, that data are too scarce to permit conclusions about this observation (8). Some compounds in creosote can be absorbed via :he lungs and can be recovered from the urine of exposed workers (9). For rats mutagenic activiry nas bem demonstrated in urine after treatment of :he skin with creosote, but no increase in mutagenic acriviry could be observed among zreosote-exposed workers (10). In animal experiments cancer has been caused by mixtures of hydrocarbons similar to creosote (1 1). .A smaIl study of 123 Swedish impregnators exposed io creosote and arsenic showed a small increase in mortaiity from [umors (12). T'ne Swedish Foundation for Occxpational Health and Saie:y ror State Emplovees. Orebro. Sweden. Cancer Registry. Institute for Epidemioiogical Cancer Research, Osio. Yorway. ' Department of Occupational Medicine, Orebro .Medical Center Hospital. Orebro, Swedcn. Reorint requests '0: Dr C-G Ohlson. Department of Occupa:ional Medicine. Orebro Medical Center Hospital. S-701 3 Grebro, Sweden. There has been public concern in both Sweden and Norway about the hazards from environmental exposure to creosote. The need for a comprehensive in- vestigation in the two countries therefore became apparent. The aim of this c o h o n study was to investigaw the cancer incidence and assess relative risks for differ. ent sites among impregnators in both counrries. Subjects and methods The source population consisted of impregnators exposed IO creosote a t all plants known to have used creosote regularly; there were seven planrs in Sweden ana six in Norway. The personnel registers of the 13 plan[! listed all persons employed in 1950 and later. The study base was restricted t o men employed 2: least one year during the period 1950-1975, som: o l whom were hired before 1950. Only subjects wict: 2 verified exposure to creosote were included. Mosr o: the subjects were impregnators, and 36 men empioyec by the Swedish Railway Corporation had been invokei with the repair and mainrenance of railroad car! designed to transport creosote. Two subjects were no: included because of their unclear exposure status. in all. 35.2 impregnators from Sweden and 601 from Svor- way were recruited. Thirty-four subjects (3.6'To) i\:rt excluded from the calculations because of missing data. 3 1 subjects due to missing identification numbers anc three due LO missing information o n the year of hir:.. There was no reason to believe that Ihese subjec:. differed from rhe orhers with regard to risk for ;3ncer. The exclusion wouid therefore not have in1roduc-c an outcome-dependenr selection. Consequenrly. :h: final study population included 346 Swedish and i-6 Norwegian impregnators, totaling 922 subjects. The degree of exposure to volatile carcinogens at :h: plants was not known. Creosotes have about 20 ma- 26 jor constituents with concent: %me of the constituents ha at least t o esperimen: be regarded as sn : j t y . The ievels of naphthr! in two of the S w r C and the average concentrat. 0.07 mg/rn3, respectively. pr d g during the opening o i . h a d i n g of wood before its Gnuations were measured i: "on plants in Finland ( 1 ). Th ofvapors per workshifr varic 11 mg/m3. Naphthalene w a ~ be vapors. The concentration lower than 0.03 pgim'. Co: concentrations of each corn: perally accepted exposure I; carcinogenic effecr of volariie m a t e d in either of the two 10. The exposure levels dur: probably higher. Neither information on .. levels to creosote nor inforr. smoking habits could be obtai: tionnaire on the type of wor management of all the plant. r n c e ~in exposure conditions nation starred when the wooc clave containing creosote oil wood when the air pressure : creased. When the wood was chve, the operator was exposi rhe wood was handled manua. ator was often exposed to ci d n a t i o n of the skin occurand the pipes were cleaned :' However, LWO of [he SRC wood preservatives as well - Table 1. Observed numoers o f 2 2 tlardired incidence ratio iSlRi 4: d workers 'son, MD3 Tong creosote-exposed workitudied among 921 creosote:ad been impregnating wood 'ires were avdable. The studv :d their cancer morbidity was ;hat lower than expected. 129 d risks in both countries cornCI) 0.81-5.83). skin cancer CI 0.83-3.78). Exposure to aumber of cancer cases does d with creosore in earlier de- jkln cancer. -ublic conc-rn in both Sweden hazards from environmental The need for a comprehensive vo countries therefore became :his c o h o n study was :O i : snd assess re!ative risks xegnators in both cou ihods on consisted of impregnator 111plants known :o have used :\.vereseven plants in Sweden -?ersonnelr e g s t e n of the I3 pi xployed in 1950 and h e r . 3s restricted 10 men smploved J :he period 1950- 1975, s :fore 1950. Only subjec:~ :reosote werc inciuded. 4 xegnacors, and 36 men em ay Corporation had been maintenanc: of raiir 1 creosote. Two subjec:s :heir unc!ear cxposure s from Sweden and 604 fr Thirty-four subjects f2.6mo .xiations because of rnissine d s i n g identification numbe information on ihe ;4sar i) n to beiieve :ha[ these su hers with regard :o risk for .uld therefore nor have introdu ~ n ste1ec:ion. Conscquentfy, n included 3-16 Sicedish and . t o r ~ ,totaling 322 subjects. ::re to volatile iarcinogrns at '1. Creosotes have Jbout 30 I Jr ;3z>:;tuents with concentrations of more than 1To. i ,L)I1lt 21' :he constituents have proved :o be carcinoj .cnlc, A less[ to experimental animals. Benzo[a]py- ,- :.,,le i x n be regarded as a n indicator of carcinogeni.!v be ;eveis of naphthalene and diphenyl were .:i,..5ured in :wo of the Swedish impregnation plants, ,rl,j :he average concenrrations were 7 mg/m3 and , 1,1~r7ny. mi, respectively, peak concentrations occur- 7111sdur:ng the opening of cylinders and during the i:;,[1d[inyof wood before its drying (13). Similar con~c!l[r:l[;~wnsere measured in two creosote impregna.,l)~ll! a i i r sin Finiand (1). T h e average concentrations , ) I % , ~ p opresr workshift varied within the range 0.1- 1 I ;ng. m;. Naphthalene was the main component of :nc. ,.apors. The concentrations of benzo[a]pyrene were :C)WC: than 0.03 pg/m3. Consequently, the average , ~ n c t ~ r a t i o nosf each component were well below ieneraily accepted exposure limits (14). The combined x c n o g e n i c effect of volatile carcinogens was not es:irnur:Li in k h e r of the two studies already referred :o. 3 s -a?osure levels during earlier decades were ?robabiy 'nigher. Ncirher information on the individual exposure .rte!s :o creosote ilor information on the individual moiting habits could be obtained in this srudy. X ques:ionnaire on the type of work was completed by the :nanagernenr of all the plants. No significant differi.nw in ;.aposure conditions appeared. The impreg:i.i!!.)n ;;arted when the wood was placed in an auto- -:J'.J ront3ining creosote oil, which penetrated the 1 ,.vhen the air pressure in :he autoclave was in- -:.xwi.LVhen the wood was removed from the auto- .hi. '. :he operator was exposed :o creosote vapor. A s hood '.vas handled manually, :he skin of the oper. w r '.';as often exposed to creosote. Moreover, con:amination of the skin occurred when :he autociave .mu:he pipes were cieaned or r e p i r e d . S[)-.ce.;e:. two of the Swedish P I ~ ~ K uS sed other .r('~'c:'iservacives as we11 - one x e d copper and chromium salts and the ocher, with 29 employees, used arsenic in 25% of the total production volume. In both countries cancer registration is based on compulsory reporting from clinical and pathological departments. The quality and coverage of the registration are considered to be good. .A11 of the subjects in the c o h o n were checked through the registries of the two countries through 79 years of age. The followup period was 1958-1985 in Sweden and 1953-1987 in Norway. The difference in follow-up times was'due to different periods of coverage in the two countries. The persan-years under risk within five-year age groups through 79 years of age were multiplied by the specific national cancer incidence rates for the expected number of cancer cases. The cancer registries provided the incidence rates, specific for gender, age within fiveyear age groups, cdendar years, and cancer site. The person-years were calculated by use of modified life tables. The standardized incidence ratio (SIR), that is, the relative risk, was calculated as the ratio between the observed and expec:ed number of cancer cases. The standardization procedure used for the SIR was the indirect method. The calculations of exact 95% confidence intervals of the SIR values and the statistical sigmficance of increased SIR values (one-tailed P-value c 0.05) were based on the assumption of a Poisson dis- tribution of the observed cases. Results The total cancer incidence was somewhat lower than expected during :he follow-up periods, 129 cases against 137 expecxd (table 1). Table 1 shows some differences in reiative risks between the Swedish and Norwegian subgoups, but the observed numbers are Sweden Sxervec 3iR iumcer 95% c1 004-100 5 23-1 27 0 04-1 30 001-1 73 024-1 JJ 0 03-5 63 ' 53-3 10 0 39-5 55 006-133 021-624 - 3 1 9.71 0.02-113 39 0 3 3 0.59-1 13 . . 3 3asocellular skin tumors not IncludeO. P = 0.02. ' p = a 0 6 Noway Observed SIR iumner c195"1 4 2 72 0.71--.3.97 23 107 o.sa- 1 60 11 1 0 0 0.50-1 79 3 154 0.70-2.93 13 0 74 0.29-1 26 4 2C8 3.57-5.23 3 128 3 25-3.75 5 1w 0.62-4.14 1 I 6 7 0 CJ-3 29 4 197 0 51-5 55 1 042 0.01--234 90 1 0 0 080-1 23 Cbservea lumber Total 950'0 CI 5 2.50. o.ai-5.a3 29 0.93 0.62-1 33 13 0.79 0.42-1.35 10 1 1 1 0.53-2.04 19 0.71 0.43- 1.1 1 5 1 72 0.56-4.01 3 2.375 1 08--J 50 a 190" 0.83-3.78 2 2.00 0.24-7 -13 6 1.d9 0.69-4.12 2 0.64 0.08-2.31 129 0.94 0.78-1 10 Table 2. Observed and eXDeCted numbers O f selected types of cancer 3v time since first employment Site 1- 9 years - -IQ Nonmelanoma skin cancerD Malignant melanoma Lung Observed number -- 1 1 Expected number 0 21 0 21 0 38 136 All sites 12 a 95% conftoence interval 1 21-8 7 i P = 0 0 1 ) Basocellular snin U n o r s l o t included 95O6 confidence interval 0 95-13 5 (P = 0 03) 11 3 - ? n e since first empiovment 10-19 years %served lumber --3 Expected numoer 0 47 06 S 0 75 3 49 19 28 7 - 2 20 years Observed nur-ser 5 6 c '2 98 Expec -numte. 1 ,e. 2 54 17E 1' : \ 97 small, and therefore the differences for some sites may have arisen by chance. Increased risks were observed for lip cancer (SIR 2.50, P = 0.05), nonmelanoma skin cancer (SIR 2.37, P = 0.02). and malignant lyrnphoma (SIR 1.9, P=0.06). Time since first exposure is of importance in most occupational cancer studies. In the group of workers with a latency period of 20 years or more, cancer of the iip showed an excess risk of 3.7 cases (5 and 1.34 observed and expected, respectively) (table 2). For nonmelanoma skin cancer the excess risk was 3.1, and for malignant melanoma it was 2.2, the observed and expected numbers being 6 versus 2.94 and 4 versus 1.78, respectiveiy. The next step in the analysis would be a further sxatification by exposure in addition to the stratification by latency time. However, this analysis was not carried out due to the small numbers and lack of individual exposure data. Discussion In the present study workers exposed to creosote at impregnation plants in Sweden and .Norway were found to have a decreased overall risk of cancer compared with :hat of the national male populations. .An increased risk was however demonstrated for lip cancer and nonmelanoma skin cancer, especially after a iarency period of 20 years or more. Of the cancer sites primariiy assumed to be associated with exposure to creoscte, :he lip and skin had increased cancer incidences, although only the latter was statistically significant. The increases could, at least in part. be explained by the creosote exposure, since skin contact must nave occurred almost daily. The incidence of malignant melanoma was increased in the Norwegian subgroup only, 1 observed versus I .92 expected cases. The numbers of the three different cancer sites together (ie, lip and skin cancer and malignant melanoma) amounted to 19 observed versus 8.72 expec:ed cases, corresponding to an SIR of 2.18 (95mo confidence interval 1.31-3.40. PcO.01). It would have been desirable to include the basocellular skin tumors in the study as well, buc reliable information on this tumor could not be obtained as the cancer repic tries did not have a complete report of these tumor( Repons of this tumor vary between hospitals and ha\? also varied over periods of time. The expected number of cases was based on na. tionai, not regional, rates. It could be questioned : this choice of reference could have introduced bias :c the comparison. The base risk for cancer clearly differ both between various regions of the countries and between urban and rural areas. As to the variation cetween regions the plants were located in the southe:: and middle parts of the countries and not concentrate. in one district. Even though the geographic distribu. tion of ihe plants was not representative of nationb populations. it was considered to be wide enough IC permit a comparison with the national rates. The variation in cancer rates between regions differs betwee: cancer sites, and therefore it was not possible to finc the idea1 reference for the 10 sites examined in : h ~ study. As to :he difference in cancer rates between urba: and rural areas, the use of national rates could w. have introduced a bias because most of the plants we?: located in rural areas. Consequently. the relative ::E! of lung cancer may have been underestimated, and :x reiative risk of lip cancer overestimated by the use 2 national rates. However, the rates of cancer of The iir malignanr melanoma, and skin cancer pooled togethc~ d o not seem to differ between rural and urban areis The impregnators had partly worked outdoors. 3n: a n increased risk of skin diseases caused by sunlie;.' wouid be expected. I t can therefore not be ruled OC' that exposure to sunLeht contributed to the risk o i and skir! cancer. This possibility is supported b! :?. findings of a review of epidemiologic studies on 1'31 cer morbidity among farmers (15). N o increase in the incidence of lung cancer (Si 0.79) was observed although this cancer site is critlL- to exposure to volatile carcinogens. .After a !ateni period of 20 years, 12 cases of lung cancer versus 1 ' expected were observed. There was an elevated risk of malignant lymphop. in both subgroups, corresponding to an SIR of 1 . " Similar findings were reported for two Swedish a': nferent studies in tvhich :! beween esposure to ;rsoq\ myeloma (5)and maiig: wo studies a n 3 the presen: however. the same stud! b:. the same m r area). Therefore :. @&jas independent. Ho\\ yoplegian subgroup sho\i 5 versus 2.6 espected. ~ tarion for this association. No increase in the inciden. be found. This result is cor: Swedish register-based stud. n e probability of detec:in m depends on the magniiL &e number of person-years;. Smprised 16 200 person-ye: so()() person-years. There \\. :. A& of follow-up for the indl: probability of detecting a dc a c e r in this study cannot be but a crude estimate indicz prson-years would be a sui! to detect this or a larger risk a 5 % level (ie, 1-8 = 0.90 an The smoking habits of tht not known. If the proportior popu1a:ion differed conside: population in general, there :he lip and lung cancer risks. fcences required are not iikr Throughout the study period ; creosote have been used, ma:; copper, chromium, and arser.. logic studies indicating that cc m c e r can be caused by inor;: in medications and drinking arsenic in smelting piants ( ! skin cancer was observed ia :: 25% arsenic. In conclusion, this cohor: st: xom&an impregnators sho\b 2 'alCancer morbidity than expt.: A h cancer was however obsc. a d d probably be attributed :c msure to creosote and sunii.. about risks for other cancer si' drawn from the findings 3 : I: Acknowledgments we thank Mr G Jermer at rhc vation Institute for his supgor- Suneborn and Ms x Naess are ?Waration of the cohort dacs % c Soderqvist is acknowied; don and the preparation of thc Svm.son for he: assisrance \v:. 28 J not be obtained as the cancer r 1 complere report of these .:or vary between hospitals a .eriods of time. umber of cases was based on 11, rates. It could be questione -ewecould have introduc-d ;e base risk for cancer cfeariy 21.1s regions of the counrries a - x a l areas. A s to the variation 3lants were located in :he SOU :he countries and nor :oncent :fl though the geographic dist vas not representative of natio :onsidered to be wide enoug with rhe national rates. The :S between regions differs bet erefore it was not posslbie ;O :'or the 10 sires examined in Ice in cancer rates betwecn urb :use O f national rates could 'as because most of rhe plants :j. Consequently, the relative w e been underestimated, and t m e r overestimated by :he use Y e r , the rates oicanc:: :he 1. and skin c a n e : pooled rose ::between rural and urban ar had parIly worked dutdoors, jkin diseases caussd by junlig i t can therefore not be 7uit.d o .!ght contributed to the r:sk of :j possibility is suppor:<i by t of epidemiologic studies on c 4 farmers (15). 2 incidence o i lung cancer (SI :ithou,uh chis cancer site !s zrltic :ile carcinosens. .Ar':er ;1 latent 2 cases o f lung cancer \. ersus 1 I . ed. ,cd risk of ma!ignanr lymphom irresponding :o m SIR i)l' ' reported for two Swedish ,:ase ,~,!:-["dies in which associations were observed :,~posureto creosote a n d the risks of multi- i l , , 2 ~ o( 5~)aand malignant lymphoma (6). These . [ L l u l ca~nd the present Swedish subgroup share, ,,,) . :he same study base (ie, they encompass ap- .r,l,ilTl~[e!tyhe same time period a n d the same geo- ,r3pnic ~ r e 3 ) T. herefore the findings c a n n o t be re- &ir. ;IS independent. However, the fac: that also the .,.,lrs,,cg~ansubgroup showed an increased incidence, L.ih,.y \i'rsus 1.6 expected, supports a causal interpre- ' ,.,!:Llil' j r ;his association. L~ :;:;:ease in the incidence of bladder cancer could .L' : ~ ' ~ i : dT.his result is contrary to the findings of a :,,&h rezister-based study (7). Tile probability of detecting an increased risk of can.c: depends on the magnitude of t h e study base (ie, n-years). The Norwegian subgroup erson-years, and the Swedish one There was a sufficiently long per the induction of lung cancer. The ?rooitiiitl; of detecting a doubled incidence of lung :anc:r :n :his study cannot b e calculated with certainty, hur a crude estimare indicates that 10 000-20 000 xrron-:de3rs would be a sufficiently large study base :o de:sc: this or a larger risk with 9OVo probability at I 5 % ievei (ie. 1-!3=0.90 and a=0.09. Tlic imoking habits of the s t u d y popuiarion were iio[ -nown. If :he proportion o f smokers in the study :Top!:laion differed considerably from :hat in the in?t)ri!lx:on general, there would be an impact on ip and lung cancer risks. iiowever the large dificrcni:~ required are not likely to have existed (16). r'iroughout the study period preservatives other than -xo\ote have been used, mainly mLXKUrcS of salts of m ? x r . chromium, and arsenic. There are epidemioh y studies indicating that both skin cancer and lung :ani:: can be caused by inorganic arsenic compounds :I1 p.sdications and drinking water or by s p o s u r e to c :n sme!ting planKS (17). However, no case of am: was observed in the Swedish plant using in a-dusion, this cohort study of 923 Swedish and Ui)ri.bezianimpregnators showed a somewhat lower to'31 a n e r morbidity than expected. .A doubled risk for ,kin ,:ancer was however observed. a n d :his increase - O 1 i i j ?:obably be attributed to the combination of exW ! C io creosote and sunlight. Valid conclusions risks for other cancer sites can however not be : from the findings or' this study. Acknowledgments ',\'e : h a n k blr C Jermer a t the Swedish W o o d Preser- c. . -'l~~ ~In? stitute for his support of :he project. )IS \L1:l:,pornand Y[s .A Naess are acknowledged for their 'p '''.mt!on of the cohort d a t a and the d a t a analyses. ': hderqvist is acknowledged for the d a t a coliec- :nd :he preparation o f the manuscript and MS.A ' ' m n for her assistance with the d a t a collection. References I . Heikkila PR. Hameila M. P w L. Raunu P. Exposure to creosote in the impregnation and handling of impregnated wood. Scand J Work Environ Health 1987;13: 431-7. 3. Schoket B, Hewer X, Grover PL, Phillips DH. Formation o i DNA adducts in human skin maintained in shonterm organ culture and treated with coal-tar, creosote or bitumen. Int J Cancer 1988;42:622--6. 3. Bos RP, Prinsen WJ. van Rooy !G, Jongeneelen FJ, Theuws JL, Henderson PT. Fluoranthene, a volatile mutagenic compound, present in creosote and coal tar. Mutat Res 1987:187:119-25. 4. Hammond EC. Selikot'f IJ. Lawther PL, et al. Inhala:ion of benzpyrene and cancer in man. Ann NY Acad Sci l976;?7 1:116. 5 . Flodin U,Fredriksson .M, Persson E. blultiple myeloma and engine exhausts. fresh wood, and creosote: a casereferent study. Am J Ind bled 1987;12:519--29. 6. Persson B, Dahlander A-kl, Fredriksson .M, Noorlinc! Brage H, Ohlson C-G. Axelson 0. .Malignant lymphomas and occupational exposures. Br J Ind .Ved 1989; 465 16-20. 7 . Steineck G, Plato Y, Alfredsson L. Norell SE. Industry:elated urothelial carcinogens: application of a jobexposure matrix to census data. Am J Ind .Med 1989: 16:309-21. 8. Steineck G, Plato N, Norell SE, Hogstedt C. L'rothelial cancer 2nd some industry-relared chemicals: an evalua- tion of the eoidemiolo-sic literature. Am J Ind .Med 1990;!7 :371-9 1. 9. Joneeneelen FJ, Anzion RB. Leijdekkers L.M. et al. I-H.fvdroxypyrenein human unne a>terexposure to coal tar and ;t coal tar derived product. Int Arch Occup En- w o n Health 1985;j7:47--55. IO. Bos RP, Joneeneelen FJ, T'neuws JL. Henderson PT Exposure to mutagenic aromatic hydrocarbons of workers sreosoting wood. rn: International Agency for Research on Cancer iIARC). .Monitoringhuman exposure to carcinogenic and mutagenic agents. Lyon: IARC, 1984:379--.38. (IARC Scientific publications; no 59). I I . Rom W N . Environmental and occupational medicine. Boston. .MA; Litt!e, Brown and Company, 1983:535-10. 1.2. Axelson 0, Kiing H. Dodlighetsmonster hos en grupp :rampregerare med kieosot exposition [Mortality among wood preservers with creosote exposure]. (Abstrac:). In: .~betarskyddsstyre!sen. 32nd Nordic occupational hygiene conference. Solna (Sweden,: Arbetarjkyddssryrelsen, 1983:125-6. 13. Rudling J , Rosen G. Kemiska hdsorisker vid trampregnering I1 [Chemicalhealth risks in impregnation of wood 111. Arbetarskpddsstyrelsen. Stockholm 1983. (Undersokningsrapporr 1983: I 1 I. 11. American Conference of Governmental Industrial Hygienists (ACGIH). Threshold limit values and biological exposure indices for 1985-1986. Cincinnati. OH: , .ACGIH, 1985. 15. Blair .A, blalker H, Cantor KP, Burmeiscer L. Wiklund K. Cancer among farmers: a review. Scand J Work Environ Health 1985:11:397--JO-. 16. Axelson 0. Aspects on confounding in occupational health cuidemioloe-y.. S a n d J Work Environ Health 1978:4:9'8--102. 17. International ;\gent:: for Research on Cancer (IARC). Some metals and metallic ;ompounds. Lyon: IARC, 1980. (IARC monographs on the evaluation of carcinogenic risk of chemicals to humans; no 23). Received for publication: 38 January 1991