Mortality among male farmers in Finland during 1979-1983.

HUSMAN KRH, LAUKKANEN VJ. Mortality among male farmers in Finland in 1979-1983. Scand J Work Environ Health 13(1987) 124-128. In thisstudy the cause-specific mortality of male farmers in Finland was compared with the mortality of all economically active men. Mortality was also examined by size of farm, by type of farm production, and by geographic area. It was shown that, compared to the mortality of all economically active men, that of the farmers was generally low. The onlyexception was mortality due to respiratory diseases, for whichthe mortality rate of the farmers was about 40-50 070 higher than that of all economically active men. The rate ratio, which was higher for thosemen with small farms, wasalmost independent of thecauseof death. The rate ratio was highest for farmersineasternand northern Finland. Assessedbytypeof farmproduction, thedifferencesin mortality were small. It was supposed that the high rate ratio of respiratory deaths among farmers was, at least partly, work-related. The high mortality of the men with a small farm was shown to be associated with the higher proportion of smokers among this group of farmers.

Several occupational epidemiologic studies suggest that the mortality of farmers is generally lower than the average mortality rate for economically active men (4). In earlier studies carr ied out in Finland the same conclusion had been reached . In a study of mortality according to occupation based on official statistics, the total mortality for agr iculture and forestry management [standardized mortality ratio (SMR) 91] in 1971-1975 was lower th an the average rate for economically active men in Finland (15). This result was also confirmed in an epidemiologic study based on data from the "Seven Country Study" (II).
Those few studies in which the morbidity o f farme rs in Finland has been analyzed according to occupation have pointed out that the morbidity of the farming populat ion is higher th an that of other occupat ion al groups (2,7,16).
Although in international studies th e total mortality of farmers has been lower than average, the SMR values of some specific cau ses of death have been high for farmers. Farmers ha ve been shown to ha ve elevated risks o f certain type s o f cancer, ie, cancer of the lymphatic and hematopoietic system, stomach , prostate, brain, lip, skin and connective tissue (4).
In Sauli' s study (15) in Finland fa irly high SMR values were determined for the cau se-spe cific mortality groups of cancer of the prostate (SMR l l S), bronchitis, emphysema and asthma (SMR 135), and nephritic diseases (SMR 129). In the same study the occupational group " farmers" was broken down into two gro ups , ie, those on small and those on large farms; and it was I Kuopio Regional Institute of Occupational Health, Kuopio, Finland. 124 observed that the mortality rate of men with small farms (2-9.9 ha) was generally higher from all specific cau ses o f death than that for farmer s with more land . The gre ate st relative difference in mortality was found for the bronchitis, emphysema and asthma group, in which mortality was approximately 70 % higher for fa rm ers with 2-9.9 ha than for fa rm ers wit h mo re than 10 ha (15). The first aim of th is study was to compare th e causespecific mortality of male farmers in Finland during 1979-1983 to the mortality of all economically active men. Th e second aim was to analyze the variation in the mortality of mal e farmers in relation to the type of farm production, the size of the farm , and the geogr aphic area in whi ch the farm was located.

Subjects and methods
This study was based on the Finni sh farm registe r. The male population included in the farm register on 31 December 1978 was defined as th e population at risk . The Finnish farm register was founded on the ba sis of the information from the general farm calculation o f 1969, to be used in form ing a ba sic register for o fficial farm sta tistics. All those per son s or corporations ownin g farms were included in th e regi ste r , in which a farm refers to the totality of cult ivated land that has a field and garden area whi ch surpasses I ha . Information is collected ea ch year with a po stal questionnai re.
Information about th e own er, occup ation of th e owner, and size of the farm was taken from the 31 December 1978 farm register. Information concerni ng the type of farm was colle cted separately from th e 31 December 1977 farm register . The farm register is ba sically reliable. For example, it was used as part of the 1980 Finnish Population and Housing Census (13).
Data on all deaths during 1979-1983, causes of death, etc, were obtained from death certificate data recorded in the Finnish register on causes of death . For the classification of deaths the International Classification of Diseases (lCD), eighth revision, was used . Only the underlying cause of death was available in the data for causes of death.
Only the active men aged 35-64 years whose occupation was listed in the farm register as farmer were included. The study population thus defined numbered 100435 persons. Women could not be included in the study population because the registered owner of the farm was, in most cases, male.
The data by type of farming were classified as follows: (i) farms, no animals, (ii) dairy farms (fewer than 10 pigs or chickens), (iii) pig farms (no other animals), (iv) poultry farms (egg or chicken production, no other animals), and (v) other animal husbandry.
Farm size was classified by cultivated land owned or rented into the following five different classes: 1-4.9 ha, 5-9.9 ha, 10-14.9 ha, 15-19.9 ha, and 2': 20 ha. The farmers were classified by age into five-year age categories at the beginning of the follow-up on I January 1979.
The geographic location of the farm was classified as follows (11) For each explanatory independent variable, the number of follow-up years was calculated. In the estimation of the rate ratios a log-linear model was used. The aim was to analyze the independent effect of different explanatory categories on the mortality of the farmers, and therefore the effect of possible significant factors such as age and geographic location were taken into account by the use of a log-linear model. The estimation of relative risks was made with the use of the Poisson model (3,5). The statistical significance of the mortality differences between the categories was tested with likelihood ratio statistics (asymptotic chi-squared deviances).
Use of a log-linear model made it possible to adjust the effect of different variables at the same time. It was also possible to analyze the effect of interaction between different variables .
Because of the small numbers of deaths, the agestandardized mortality rates have been calculated with the use of indirect age standardization (I).

Mortality of farmers in comparison to that of all economically active men
The mortality comparison between farmers (1979)(1980)(1981)(1982)(1983) and all economically active men in Finland (1976Finland ( -1980 showed that the mortality of the farmer was clearly lower than that of the latter (table I).
Although, in general , the mortality of farmers was low, those with diseases of the respiratory system were exceptions. In this group the mortality of farmers was about 40-50 0J0 higher than that for all economically active men .

Regional mortality differences among male farmers in Finland /979-/983
As shown in table 2, for the mortality of male farmers, there were considerable differences between the different regions. The risks were highest in eastern and northern Finland. The lowest rate ratios were usually found in western and southern Finland.
In the analysis of the differences in mortality according to the causes of death (table 2), only in the case of cardiovascular diseases were the regional mortality differences among farmers significant. Owing to the regional differences in mortality among farmers, however, the effect of the variable "geographic location" was taken into account when the effects of the variables "size of farm" and "type of farming" were analyzed, and in the consideration of the interaction effects as well.
Mortality differences by size of farm As presented in table 3, the rate ratio was almost always the highest among owners of small farms . The rate ratio due to all diseases was about 30 070 higher among small-farm owners than among farmers having over 20 ha of cultivated land. It also seemed that the relationship between farm size and mortality was linear. The picture was the same for the case of cardiovascular diseases, for accidents, poisonings and violence, and for diseases of the respiratory system, although, owing to the small number of deaths, the dif-    ferences in rat e rat ios du e to diseases of the respirato ry system were not statistically significa nt. The rat e ratio for neoplasms was also the highest among men with sma ll fa rms.

Mortality diff erences by type of farming
When the differences in the rate ratio s by type of farming were compa red to th e differences by size of farm , the differences in mortality were much smaller for type of farming (table 4). Statistically, the di fferen ces by type o f far ming were only significant fo r all diseases, for cardiovascular diseases, and for accidents , poisonings and violence. Although in most cases the differences were no t statistically signific ant, th e ra te ratio was, independently of the cause o f death , almost always th e highest among far mers with no animals.

126
The adjustment for size of farm caused a decline in the di fferences in mo rtality found by type of far ming (table 4, lower part). Th e rate ratio due to cardiovascular diseases was, however, still high among farmers with no animals.

Interaction ef fect s
Accor ding to th e results shown in tabl es 3 and 4, th e interactio n effects were negligible. Mo rta lity differences by size of farm were the same in differen t age groups and in differen t are as. In additio n, there were no interaction effects between the size of the far m and type of farming. Th e mortalit y of men with small farms was highest in all types of farm production indepe nden tly of the cause of death . Onl y in th e case of age and type of farm was there an interactio n effect between the variables.

Discussion
In previous studies in Finland, and also in other countries, it has been sho wn that th e mortality of male far mers is generally low compared to that of all econom ically act ive men . In this stu dy the same conclusion could be drawn. Com pared to the mortali ty of all economically active men, that of male farmers was, on an average, 10-30 070 lower. Th e only except ion was the mortality due to respiratory diseases, for which the mortality rate of male farmers was about 40-50 % higher than that o f all econo mica lly active men .
Altho ugh, in genera l, th e fa rmers' mo rtality was low, the mortality of those with small farms was clearly higher than that of other farmers and higher than the mort ality of all economically active men. As measured by the SMR, also for owners o f larger farms (> 15 ha), the mortal ity due to respiratory diseases was higher (SMR 110) than that for all economically active men.
Th e mortalit y o f men with sma ll farms was high in every type of fa rm production. Almo st independently of the cause of death, the mortality amo ng sma llfar m owne rs in Finland was found to be higher than that among those with larger far ms. Disab ility studies amo ng the farming populat ion ha ve also shown that  the relati ve risk of disability pension is higher among those with small farms than among other farmers, even in the register of the same far m pop ulat ion (12). The pro po rtion o f fa rmer s who have ever smok ed was clear ly higher amo ng those with sma ll farms than among th e rest of th e farm pop ulati on (table 5). Evidentl y the high rela tive risk of death due to respi ratory diseases, neoplasms, and also ca rdiovasc ular diseases among men with small far ms is conn ected with the higher proportion of smokers in this grou p.
The av erag e proportion of sm o kers am ong Finnish farmers was lo wer than a mo ng the who le male populat ion (table 5). For farmers the proport ion o f those who had ever sm o ked (1978)(1979)(1980) was 68.9, but for industrial wor kers the correspo ndi ng figu re was 82.7, and for thos e in the se rvice sec to r it was 75.6 (8).
In addition , even among m en with sm all farms, for whom th e proportion of smo kers wa s highest , the pr opo rt ion o f those who had ever smo ked was lower th an th at o f the tot al population a t the sa me time (table 5) (8) . Smoking there fo re ca nno t explain the high rel ative risk of respira tory deaths a mo ng men with sma ll farms o r a m o ng th e who le fa rm in g population .
Fa rme rs, especiall y on sma ll livesto ck-producing farms, a re exposed to various organic and microbial du st s and endotoxins (9, 10', 14), which , it has been claimed, ca use the high morbidity from respiratory d iseas es (a sthma , fa r m er's lung a nd chro nic b ronchitis) among fa rmers (6). This exposure exp la ins, to some extent, the high relati ve risk of res piratory deaths among male farmers a nd suggests tha t th e high m ortal ity rate due to respiratory disea ses among farmer s is wo rk-re la ted .
In a na lyses o f mo rtalit y di f feren ces by typ e of fa rm p rodu ct ion , th ere is a n ad d itio na l problem , namely, " select ion. " A ltho ug h , in livestock production , farmers ar e probably mor e exposed to di fferent o rga n ic and m icrobial du st s, their mo rtalit y d ue to respirator y d isea ses was no t high er tha n th e mortal ity of farmers in the gro u p with no a nim a ls . This finding might, however , be due to th e fact that the far me rs wh o d id not have animals at th e beginning of th e follo w-up peri od had had a ni ma ls pr eviou sly, but, owing to so me respir atory problems, they had probably changed the type of producti on -fo r example, from dai ry far ming to grain crops . Selec tio n m ight a lso ex pla in th e h igh mortal ity du e to accidents, poisonings and violence in th e group with no a nimals. In an additional a na lysis, it was sho wn th at both mo rta lity du e to transport accide nts an d th at due to suicides wa s higher in this gr o u p .
The interaction effects betw een vari ables wer e small. At thi s stag e of our ana lysis , howe ver , we ha ve to bear in mind tha t , when the di fferent interact ion s were a nalyzed, th e number of de aths was sm a ll.
The mortality of men with sma ll farms wa s sho wn to be a sso ciated with the life-style o f the se men (smo king) . Apparently th e differences in m ortal ity bet ween sm all-farm owners and o ther farmers co uld a lso be partly wo rk-r elated . Gener al soc ioecono mic fac to rs , such as quality of hou sin g, type of d iet , a nd th e general " heavy burd en o f life" m ight a lso par tl y ex plai n the observed d ifference between men with sma ll far ms and othe r fa rmers .
We ar e no w co nd uc ti ng a ser ies of ep ide mi ol ogic studies wit h the ob jective of explain ing the re lation of dust exposures at work (or ganic dusts, mic robial dusts, endotoxins) a nd res pir atory morbidit y am on g Fin n ish farmers.