incidence of workers in a Finnish sawmill.

S. Cancer incidence of workers in a Finnish sawmill. Scand J Work Environ Health 1989;15:18-23. The cancer incidence of I 223 sawmill workers with continuous employ ment of at least one year between I January 1945 and 31 December 1961 was followed until 31 December 1980. Separate analyses were made for the 801 workers hired after I January 1945, and smoking habits were surveyed. Among the men, 90 cases of primary cancer were detected versus 83.5 expected and among the women 55 cancer cases versus 44.5 expected. Skin cancer was in excess among the men, especially among those employed after I January 1945. Lip, mouth, and pharynx cancer and lymphomas were also slightly in excess among the men, as was leukemia among both sexes. Workplace exposure, especially to chlorophenols, may be associated with the excess skin cancer and the slight excess of lymphomas, but this finding should be further evaluated with special emphasis on well-defined exposure data.

There are limited data on the incidence of cancer among sawmill workers, despite the impo rtance o f the sawmill industr y in many countr ies. For instance, in Finl and th ere wer e 17375 wo rkers employed by the sawmill industr y in 1985 (I), ie, 0.8 070 of the active working population. In addition , in many epidemiologic studies lumber and saw mill wo rkers have been combined , and thus interpreting the results with respect to sawmill wor kers alone is difficult (2). Furthermore, the ep idemiologic data ar e mainl y from surveys o f occupation s on death certificates (2).
Saw mill workers ar e expo sed mainly to wood du st, antista in agents, sa wing vapors, and fungal spores. In a dditi on, many other types of compounds with po ssible health effects can be found in sma ller amounts in their work environment (2).
Various types of can cer excess have been reported for sawmill workers. For exa mple, there have been reports of an increased risk for nasal cancer (3, 4). Nasal adenocarcinomas have been reported to be associated with hardwood dust expo sure, but not with the so ftwood dust exposure typical in Finnish sawmills (5). H owe ver, the re is a report of an association between other type s of na sal a nd sinus paran asal cancer a nd exposur e to softwoo ds (6). Although one death certificate study show ed an increased risk of lun g ca ncer among sa wmill wor kers (7), the re a re other studies showing no excesses of lung cancer (8,9) among t hese wor kers. There ar c reports of increased risk for Hod gkin' s disea se (10) and for leuk em ia and lympho-rna (9) amo ng sawmill workers. Increased risks of softtissue sarcomas and histio cytic lymphomas ha ve been associ at ed with chlorophenol exposure, which also takes place in the sawmill industr y (11)(12)(13).
T he aim of the present study was to assess the incidence of cancer among workers o f a Finnish sawmill. Th e investigation was performed simultaneously with a study o f cancer incidence among Finnish pulp and paper wo rker s (14).

Subjects and methods
Th e study cohort was formed fro m wo rkers in a sawmill in th e province of Kymi in so utheaste rn Finland. The frame sawmill, founded in 1872 , has an annual capacity o f 100000 m' of sawn timber (80 % pine, 20 % spruce). During the 1950s and 1960s the number of workers was over 400, but, as a result of modernization in 1974, the sawmill now employs only 144 per son s. Chlorophenols were used as anti stain agents in th is sawmill from 1945 to 1983. Sawn timber was su bme rged in a 0.5 070 aqueous chlorophenol solution for approxima tely 1 min during the warm seaso n. The commercia l chlorophenol product has co ntai ned sodium sa lt of pent achlorophenol (5-9 070 by weight), sodium salt of 2,3 ,4 ,6-tetrac hlo ro phenol (78-83 070), and 2, 4, 6-trichlo ro p heno l (7-15 % ) (I5).
Th e study population (721 men and 502 wome n, for a total of 1 223 workers) consisted o f all th e workers who had been employed continuo usly at the sa wmill for at least one yea r during the study period (1 January 1945 to 31 December 1961). At the beginning of the study period (I January 1945), 304 men and 198 wo me n of the coho rt were already employe d in the saw mill.
T he basic source of employment dat a was th e co mpan y employment files . Thei r completeness was en- alph ab etical order, without separation for gender or vital status, and a random sample was selected . A total of 144 questionnaires were sent. A repl y was received fro m 74.3 % of the sample.

Results
Th e ob served number of all primary cancers (table 2) among the men was somewhat higher than that expected (90 observed, 83.5 expected, SIR 108, 95 % CI 87-133). Among the women, there was also an excess o f all primary cancers (55 observed, 44.5 expected, SIR 124, 95 % CI 93-161).
There was an excess of skin cancer among the men (6 ob served, 1.9 expected, SIR 313, 95 % CI 115-680). The risk of skin cancer among the men did not differ bet ween the groups with an occupational expo sure of less than five years and at least five years, but the risk increased with follow -up time so that five of the six cases occurred after 15 years or mo re since first employment (versus 1.7 expected). The occupation al histories of th e subjects with skin cancer are presented in table 3.
Cancer of the lip , mouth , and pharynx was in excess among the men (6 observed , 3.4 expected, SIR 175,95 % CI 64-382) . All six cases occurred after 25 years or more since first employment.
There were also excesses of all lymphomas, especially Hodgkin 's disease, among the men, and an excess of leukemia among both sexes. The numbers of the observed and expected cases were low ho wever (table 2). Th e risk of lymphomas among the men was somewhat higher for those with an occupational expo sure of a t least five years (4 ob served, 1.5 expected , SIR 268, 95 % C I 73-687). The job titles and major occup ation al exposures of the subjects with lymphoma o r leukemia are presen ted in table 3.
Th e women had four cases of lung cancer versus 1.2 expected (SIR 325, 95 % CI 89-833), but the number of lung cancer ca ses among th e men was almost the same as that expe cted (table 2).
sured by a check of the material against separately kept employer stati stics of th e mills. No data were available on pos sible work periods in other sawmill s, but a change of emp loyer has been rather rare.
The personal data, including follow-up unt il death or the end of 1981, were confirmed in the National Population Register, and in local population registers when necessary. Reti rement or change of job did not affect the follow-up. Th e cause of death was retrieved through the Central Statist ical Office. Information was also obtained from Statistics Sweden and from authorities of Norway, Denmark , Canada, and Au stralia .
All members of the cohort were identified . The follow-up was complete for all tho se who were still living or had died in Fin lan d and for mo st of tho se who had emigrated. Only three workers (0.2 010) were lost. The y had all emigrated from Finland to places with inadequate registers.
The cancer incidence was followed fro m I January 1953 un til 3 I Decem ber 1980 through the use of the data of the Finn ish Cancer Registry . It has virtually complete data on all primary cancers, and the ir histological types, notified in Finland since 1953 (16). Since there has been emigration from Finland to Sweden and to a less extent to Norway, information on the primary cancers of the emigrated workers was also obtained from the population-based cancer register s of the se countries. It could thus be assumed that the cancer data were practically complete. If no information was obtained on the possible primary cancers of an emigrated person known to be still alive , the date of emigration was used as the end point of the follow-up .
The person-years (table I) were calculated from the end of the minimum exposure period (one year) until death, the closing date of th e follow-up, or (fo r the three lost cases in the follo w-up and for those living emigrants for whom there was no information on pr imary can cers) un til the date of emigration .
The expected numbers of pr imary cancers were calculated by five-year age groups, ten-year calend ar periods, and five-year latency (time since first employment) periods . The age, gender, and calendar-period specific num bers of person-years at risk were multiplied by the corresponding cancer incidence rates . The incidence figures for the population o f Kymi were used as reference. Th e sta ndard ized incidence ratios (SIR) were obtained as rat ios between the observed and expected numbers o f cases. The 95 % confidence int ervals (95 % CI) were estimated on the assumption of a Poisson distribution for the observed number of cases. Separate analyses were made for two cate gorie s acco rding to the duration of employment (1-4 years and 2: 5 years). Furthermore, separate analyses were mad e for the workers who were employed fo r th e first time after I January 1945 in order that a cohort with adequately de fined expo sur e data would be a vailable , especiall y in th e case of chlor opheno ls.
The smoking hab its of a rand om sample of the study population were assessed . Th e wo rkers were listed in  There were no cases o f nasal ca nce r amo ng either th e men (0.2 expected) or the wo me n (0. 1 expected) . Th ere were no soft-tissue sarco mas am ong the men (0.4 expecte d), but there was one amo ng the women (0. 2 expected).
T he find ings for the wor kers employed for the first time after I January 1945 were simila r to those for the whole cohort. The risk o f primar y cancer was increased both among the men (36 observed, 28.1 expected, SIR 128,95 lila CI 90-177) and the wo men (27 ob served, 21.9  20

Discussion
Thi s study showed a slight excess o f primar y cance rs a mo ng both male and femal e saw mill workers, especially amo ng th ose em ployed fo r th e first time after I J anua ry 1945. Becau se of the result s o f previou s studies, we were especially interes ted in certain types of ca ncer , part icul arl y in nasa l can cer, other types o f resp irat ory cancer , soft-tissue sarco ma, leukem ia, and lymphoma .
Th e most interesting finding proved to be the increase d risk of skin cancer among th e male sawmill wo rker s. It had not been reported earl ier . Four of the six cases o f skin cancer among the men occurred among wor kers employed for the first tim e after 1 January 1945 (versu s 0.7 expected) . It is possible that the excess risk of skin cancer is associ ated with wor kplace chemical expos ur e, and espe cially the ro le of chloropheno ls sho uld be car efully eva luat ed . Ho wever, two of the eight subjects with skin ca ncer had not been exposed to ch lo rophenols .
Kau pp inen & Lindroos (19) ha ve measur ed personnel expo sure to chloro phenols in ten Finnish saw mills. Th e co nce ntratio ns o f chlorophen ols in the a ir were usually below the maximal allo wabl e con cent ration of 0.5 mg/m ' . However , rather high (up to 210.9 um ol/l) co nce ntrations were det ected in the urin e of load ers when the th rou gh-dipping meth od was used, as well as in th e ur ine o f wo r kers in the tri mming-gradi ng depar tment (up to 85.0 urnol /l) (20). Th ese wo rke rs had skin contac t with chlorophenols, and skin a bsorp-  tion appeared to be the main route of exposure in the sawmills. Commercial chlorophenol products have been found to contain impurities such as chlorinated phenoxyphenols, chlorinated diphenyl ethers, chlorinated dibenzofurans, and chlorinated dibenzodioxins (21). Some of these compounds are highly toxic and cancerogenic even in low concentrations. It should be noted that not all sawmill workers have been exposed to chlorophenols, and it is therefore difficult to assess the possible carcinogenic risk of these compounds in general sawmill worker cohorts without careful jobbased analyses. The chlorophenol exposure has been the highest among those working in the timber dipping area or handling timber still wet after chlorophenol treatment, but workers in the trimming-grading plant of the sawmill have also been exposed to wood dust containing chlorophenols.
The slightly increased risk of lymphomas among the men was in accordance with the results of some earlier studies (9, 10). One paper has reported excess deaths from leukemia among sawmill workers (9), but another (22) showed no increased risk of leukemia. In the present study, the risk of leukemia was slightly increased, among both the men and the women, but the lower 95 al o confidence interval was below 100.
The risk of lung cancer was not increased among the men in the whole cohort, but was somewhat increased among those who had been employed for the first time after 1 January 1945. There was an increased risk of lung cancer both among the women in the total cohort and among those employed after I January 1945, but the numbers of the observed and expected cases were low.
The use of chlorophenols in the sawmill of this study began in 1945; otherwise workplace exposure has been the same for the whole cohort and for those employed for the first time after 1 January 1945. However, no clear association can be established between exposure to chlorophenols and lung cancer because different smoking habits between the sawmill workers and the general population may partly explain the finding.
Wood dust is typical workplace exposure in sawmills, but in general the exposure levels have not been very high (23,24). According to Kauppinen et al (23), the dust concentrations were generally below I mg/m', but exceeded 3 mg/m' during cleaning operations, the control of chipmaking machinery, and the removal of moldings.
Hernberg et al (6) recently found an association between nasal cancer and exposure to softwoods. The findings of our present study were not consistent with theirs, as no nasal cancer cases were found, but it should be noted that the expected numbers of nasal cancer were very low.
There have been no earlier reports of an increased risk of colon cancer or rectal cancer among sawmill workers. The numbers of the observed and expected cases of these cancer types in the present study were so low that no conclusions can be drawn from the observed statistically nonsignificant excesses. Contrary to the results of an earlier report (25), there were no excesses of laryngeal cancer in this study.
Not much can be said about the possible "healthy worker effect" among sawmill workers, but it certainly exists also in this industry. It would be interesting to study a cohort of sawmill workers without any chlorophenol exposure, but the number of those workers, at least in the sawmill of the present study, would be rather low for a proper cohort.
In conclusion, although only a slight excess of primary cancers was observed among the sawmill workers studied, there was an excess of skin cancer, especially among the men. This excess may be associated with workplace exposure, but the possible association needs to be further evaluated with special emphasis on specific jobs with well-defined exposure data.