Farm residence and exposures and the risk of allergic diseases in New Zealand children

Authors


Dr Kristin Wickens
Wellington Asthma Research Group
Wellington School of Medicine
PO Box 7343
Wellington South
New Zealand

Abstract

Background: Studies in Europe have reported a reduced prevalence of allergy in farmers' children. We aimed to determine if there is a similar reduction in allergy among New Zealand farm children.

Methods: Two hundred and ninety-three children participated (60%) aged 7–10 years, from selected schools in small towns and the surrounding rural area . Skin prick tests (SPT) to eight common allergens were performed. Parents completed questionnaires about allergic and infectious diseases, place of residence, exposure to animals, and diet, and they provided dust from the living-room floor. Endotoxin was measured using an Limulus amoebocyte lysate (LAL) assay and Der p 1 using enzyme-linked immunoassay (ELISA).

Results: Current farm abode was found to increase the risk of having symptoms associated with allergy, but not SPT positivity. Independent inverse associations were found for early-life exposures: at least weekly consumption of yoghurt with hayfever (odds ratio (OR) = 0.3, 95% confidence intervals (CI) 0.1–0.7) and allergic rhinitis (OR = 0.3, 95% CI 0.2–0.7); any unpasteurized milk consumption with atopic eczema/dermatitis syndrome (AEDS) (OR = 0.2, 95% CI 0.1–0.8); cats inside or outside with hayfever (OR = 0.4, 95% CI 0.1–1.0) and AEDS (OR = 0.4, 95% CI 0.2–0.8); dogs inside or outside with asthma (OR = 0.4, 95% CI 0.2–0.8); and pigs with SPT positivity (OR = 0.2, 95% CI 0.1–0.9).

Conclusions: Despite finding a protective effect of early-life animal exposures, we found a greater prevalence of allergic disease on farms.

Increases in allergic disease prevalence have been linked to increased economic development, with reports of higher asthma (1–4) and allergic rhinitis (5–7) prevalence in urban centres compared with rural areas. Similarly, increased allergic disease has been reported following migrations from less economically developed societies to countries with more developed economies (8, 9). There has been an assumption that the movement from a low-risk to a high-risk environment is responsible for the higher prevalence of allergic disease with increasing urbanization or development. For example, higher social class (10), antibiotic use (11, 12), and small family size (13, 14) are characteristics associated with increasing wealth that have been proposed to explain the higher prevalence of allergic disease that occurs with economic development.

The “hygiene hypothesis” (14) proposes that the increase in allergic disease prevalence is due to reduced exposure to microbes that may be more common on farms and in less developed countries. Therefore, the current focus to explain this increase in developed urban centres is to identify the factors in low-risk rural environments that may confer protection against the development of allergic disease. A number of recently published studies in Europe and Canada have shown consistently that being raised on a farm is associated with a lower prevalence of hayfever (15–22), asthma (16–21, 23), and atopy (15, 17, 19, 20, 22) compared with being raised in a nonfarm rural or urban environment. It has been shown that endotoxin levels in Switzerland and Southern Germany are higher in farmhouse kitchens and beds. It has been postulated that through the stimulation of Th1 phenotypic immune responses, endotoxin may protect against the development of atopy (24).

These studies of childhood farm exposures have been conducted predominantly in Europe, within countries that have low humidity, and small farm holdings. The primary aim of our study was to determine whether the protective effect against allergic disease applies to children raised on New Zealand farms, where animals are kept outdoors on large holdings, in a temperate climate. The secondary aim was to determine which factors associated with farming might explain these differences.

Material and methods

We approached 605 children aged 7–10 years from seven selected schools to determine eligibility for this study. Where children had siblings in the eligible age range we randomly selected one child for study. Two schools from the main town in the area (Dannevirke; population 5513) were selected because their school rolls included children transported in from the countryside. A further two schools in smaller towns, and three small schools situated in the countryside were included to ensure participation of enough children from farms.

Children of consenting parents took the questionnaires home for completion by their parents. The International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire (25) was used to collect information on allergic disease history and symptoms.

Hayfever was defined as a positive response to the question “Has your child ever had hayfever?” A positive response to the question “In the past 12 months, has your child ever had a problem with sneezing or a runny or blocked nose, when he/she DID NOT have a cold or flu? defines a condition which is consistent with the EAACI nomenclature for allergic rhinitis proposed in a recent position paper (26). Asthma was defined as a positive response to “Has your child ever had asthma?” and current wheeze as “Has your child had wheezing or whistling in the chest in the last 12 months?” Eczema was defined as a positive response to: “Has your child ever had eczema?” and will be described according to the EAACI nomenclature as the atopic eczema/dermatitis syndrome (AEDS) (26).

We collected information on residence (living in town, or on a farm, or in a rural area but not on a farm), gender, ethnicity, number of years of mother's post-primary education, family history of allergic disease, family size, antibiotic use in the first year, measles and whooping cough infections, animals inside and outside, heating source, and diet, including dairy food consumption. Details of these exposures were collected for the child's first year of life and the past 12 months, except diet information which was collected for the first two years of life and currently. Only 37 children were currently living in rural nonfarm areas, so these children were combined with the town children to form one group.

The children also took home instructions on how to sample dust. They were provided with a nylon sock to insert onto the furniture attachment of their vacuum cleaners, and instructed to sample 1 square meter of dust from the center of their living room floor for 1 minute. Samples were returned to the school for daily collection by the researchers and stored at − 20°C before transportation to Wellington on dry ice.

Sieved dust was stored (0.2 g) for endotoxin analysis and, when sufficient dust was available, for Der p 1 analysis. Endotoxin in dust was determined by a commercial kinetic chromogenic Limulus amoebocyte lysate (LAL) kit (BioWhittaker, Walkersville, MD, USA) on 1 : 500 dilutions of dust extracted with pyrogen-free water containing 0.05% Tween-20 (27). Der p 1 in dust was determined by use of a commercial double-monoclonal antibody ELISA kit (Indoor Biotechnologies, Cardiff, UK) on phosphate-buffered saline extracts of dust extracted at room temperature (28).

Standard methods were used to perform skin prick tests (SPT) to common allergens (Dermatophagoides farinae, D. pteronyssinus, mould mix, cockroach, rye grass, timothy grass, cat, dog) using Bayer allergens and prick lancets. A positive SPT reaction (SPT +) was defined according to the presence of a mean wheal diameter of 3 mm or more to at least one of the allergens.

Statistical analysis was conducted using SAS version 8 (SAS Institute, Cary, NC). We examined differences in the distribution of exposure variables between farm and nonfarm abodes, and for each outcome variable using the Mantel–Haenszel Chi-square statistic. Endotoxin and Der p 1 showed log normal distribution. Differences in distribution among selected exposure variables and disease outcomes were compared using geometric means with t-tests to assess the significance of the differences. Logistic regression was used to assess the independent effects of the exposure variables on each outcome (allergic disease, allergic symptoms, and SPT +). Multicollinearity between exposure variables was estimated by observing changes in the estimates and standard errors (SE) after removal of possible collinear variables from the model.

The Wellington Ethics Committee approved the study.

Results

Response

After the exclusion of 111 siblings, three children who had moved schools, and one child who was outside the age range there were 494 eligible potential respondents. Of these children, 293 (60%) parents completed questionnaires, 286 (58%) collected dust samples, and 275 (56%) children undertook SPT. Four dust samples contained too little dust for analysis of Der p 1 or endotoxin; thus endotoxin analysis was performed on 282 samples. There was sufficient dust for Der p 1 analysis of 214 samples.

Respondent characteristics

Table 1 shows differences in the characteristics of farm and nonfarm children. There were no significant differences in gender, family size, measles, whooping cough, cat exposure, or antibiotic use during the first year of life. However, farm children were more likely to have been exposed to dogs, poultry, pigs, and coal or wood fires during their first year and currently, and to have had diets in infancy that regularly included cheese and unpasteurized milk; but they were less likely to be exposed to maternal smoking during the first year of life and currently. Yoghurt consumption was similar for farm and nonfarm children. Of note is the large difference in distribution of mother's education; the mothers of children on farms had more years of post-primary education. Although not shown in the table, the pattern was similar for the fathers' years of education.

Table 1.  Respondent characteristics by current farm or nonfarm abode
 Current farm (n = 95) n (%)Current nonfarm (n = 198) n (%)P value
  • *

    Question allowed for multiple ethnic identification.

 Male43 (45) 97 (49)0.55
 Female52 (55)101 (51) 
Ethnicity*
 Maori13 (14) 49 (25)0.03
 European88 (93)173 (87)0.18
Mother's post-primary education
 < 4 years20 (21) 75 (40)< 0.0001
 4–6 years43 (45) 84 (45) 
 7+ years32 (34) 29 (15) 
Family history of allergic disease72 (76)125 (63)0.03
Family size
 0–1 siblings30 (32) 79 (40)0.19
 2+ siblings65 (68)119 (60) 
Infections
 Measles infection ever19 (20) 31 (16)0.36
 Whooping cough infection ever 7 (7) 11 (6)0.56
First-year-of-life exposures
 Cats inside or outside73 (77)150 (76)0.84
 Dogs inside or outside67 (71)118 (60)0.07
 Regular contact with poultry19 (20) 17 (9)0.005
 Regular contact with pigs14 (15) 15 (8)0.06
 Antibiotics in the first year54 (57)127 (64)0.23
 Coal or wood fires87 (92)164 (83)0.05
 Mother smoked23 (24) 85 (43)0.002
Diet under 2 years
 Yoghurt once or more a week72 (76)153 (77)0.78
 Pasteurized milk once or more a day53 (56)139 (70)0.02
 Unpasteurized milk ever22 (23) 16 (8)0.0003
 Cheese once or more a week75 (79)125 (63)0.007
Current exposures
 Cats inside or outside79 (83)155 (78)0.33
 Dogs inside or outside85 (89)129 (65)< 0.0001
 Regular contact with poultry27 (28) 18 (9)< 0.0001
 Regular contact with pigs14 (15)  8 (4)0.001
 Coal or wood fires94 (99)179 (90)0.007
 Mother smokes20 (21) 76 (38)0.003

Among current farmers (n = 95), 38% farmed beef cattle, 44% dairy cows, and 45% sheep.

Prevalence of allergic disease and skin prick test positivity

Table 2 shows that living on a farm during the first year of life was positively associated with an increase in hayfever prevalence, but had no effect on SPT positivity, or any other allergic disease or symptom. The effect for hayfever was largely restricted to those that were SPT . Living on a farm currently was associated with a higher prevalence of all health outcomes, with these differences significant for hayfever. The increased risk of current farm abode for hayfever and asthma was largely among SPT + children. Among SPT + children, the mean number of positive SPT was the same for farm (2.9) and nonfarm (2.9) children during the first year of life, and similar for farm (>1.3) and nonfarm children (2.7) currently (P = 0.18).

Table 2.  Prevalence of having skin prick test (SPT) positivity, wheeze in the last 12 months, asthma ever, hayfever ever, allergic rhinitis in the last 12 months, and atopic eczema/dermatitis syndrome (AEDS) ever, in farming and nonfarming children
 FarmNonfarmOdds ratios (95% CI)
  • *

    P  < = 0.05.

  • **

    P =  < 0.01.

First year of life residence
 n = 94 (%) n  = 199 (%)
Current wheeze19 (20)46 (23)0.8 (0.5–1.5)
Asthma ever32 (34)66 (33)1.0 (0.6–1.8)
Hayfever ever24 (26)32 (16)1.8 (1.0–3.3)*
Current allergic rhinitis29 (31)65 (33)0.9 (0.5–1.6)
AEDS ever36 (38)71 (36)1.1 (0.7–1.9)
 n  = 90 (%) n  = 185 (%)
SPT +29 (32)60 (32)1.0 (0.6–1.7)
SPT + to grass20 (22)43 (23)0.9 (0.5–1.7)
SPT + to dust mite22 (24)40 (22)1.2 (0.7–2.1)
Among SPT + childrenn  = 29 n  = 60  
 Asthma ever17 (59)33 (55)1.2 (0.5–2.8)
 Hayfever ever17 (59)20 (33)2.8 (1.1–7.1)*
Among SPT – childrenn  = 61 n  = 125  
 Asthma ever14 (23)29 (23)1.0 (0.5–2.0)
 Hayfever ever 6 (10) 9 (7)1.4 (0.5–4.2)
Current residence
 n  = 95 (%) n  = 198 (%)
Current wheeze24 (25)41 (21)1.3 (0.7–2.3)
Asthma ever37 (39)61 (31)1.4 (0.9–2.4)
Hayfever ever28 (29)28 (14)2.5 (1.4–4.6)**
Current allergic rhinitis37 (39)57 (29)1.6 (0.9–2.6)
AEDS ever39 (41)68 (34)1.3 (0.8–2.2)
n  = 90 n  = 185  
SPT +33 (37)56 (30)1.3 (0.8–2.3)
SPT + to grass25 (28)38 (21)1.5 (0.8–2.7)
SPT + to dust mite25 (28)37 (20)1.5 (0.9–2.8)
Among SPT + childrenn  = 33 n  = 56  
 Asthma ever21 (64)29 (52)1.6 (0.7–3.9)
 Hayfever ever21 (64)16 (29)4.4 (1.8–10.9)**
Among SPT – childrenn  = 57 N  = 129  
 Asthma ever14 (25)29 (22)1.1 (0.5–2.3)
 Hayfever ever 7 (12) 8 (6)2.1 (0.7–6.2)

Endotoxin and Der p 1 levels

Table 3 shows that endotoxin levels were lower on farms, significant for EU/g, but levels of Der p 1 (µg/g and µg/m 2 ) were significantly higher on farms. When endotoxin was defined dichotomously, using the geometric mean levels as cut-off points, the associations with farming remained for EU/g ( P =  0.08) and EU/m 2 ( P =  0.02). However after redefining Der p 1 according to geometric mean cut-off levels, the association between Der p 1 and residence became nonsignificant. We also found nonsignificantly higher ( P =  0.21) geometric mean EU/g in living-rooms when children were regularly exposed to pigs (15 719) compared to homes where children had no pig exposure (9319.10), and significantly higher ( P =  0.03) geometric mean living-room EU/g in association with regular poultry exposure (14 364) compared to homes where children had no poultry exposure (9008). The concentration and the total amount of endotoxin were similar regardless of exposure to dogs, cats, dairy cows or cattle, sheep, or horses. Endotoxin levels (EU/g and Eu/m 2 ) were similar for farmers of sheep, cattle, and dairy cows. However, compared to animal farmers, the few crop farmers ( n  = 4) had lower EU/g (2573 compared to 7798, P = 0.13) and EU/m 2 (977 compared to 3391, P = 0.17).

Table 3.  Geometric mean endotoxin units (EU) Der p 1 and total dust weight by current living abode
 nFarmGSD*nNonfarmGSD*P value
  • *

    Geometric standard deviation.

EU/g947438.34.218811058.23.50.02
EU/m2943216.35.81884506.74.90.11
Der p 1 (µg/g)7312.83.11418.33.80.02
Der p 1 (µg/m2)738.23.61415.54.00.03
Total dust weight (g)940.42.81880.42.90.65

Geometric mean EU/g were marginally lower (P = 0.09) in association with hayfever (7407 compared with 10 340 when hayfever was absent), but the significance of this association was reduced when EU/g was defined dichotomously with geometric means used as cut-off points. There were no differences in EU/g for any other outcome variable. The findings were similar for EU/m2. Level of Der p 1 was not associated with any outcome variable, except a history of AEDS, where a level of Der p 1 per m2 greater than the geometric mean was marginally associated with an increased risk of AEDS (OR = 1.7, 95% CI 0.9–2.9).

Risk factors for allergic disease

Univariate analysis

We found that exposure to pigs in the first year of life was associated with a reduced risk of becoming SPT + later in life (OR = 0.4, 95% CI 0.2–1.1), and exposure to cats at that time was associated with a reduced risk of having a history of hayfever (OR = 0.6, 95% CI 0.3–1.0). Early-life exposure to other animals was not associated with any other outcome variable, but current exposure to poultry increased the risk of becoming SPT + (OR = 1.8, 95% CI 0.9–3.5). Other current animal exposures were not associated with any outcome.

Consuming yoghurt at least once a week during the first two years of life reduced the risk of becoming SPT + (OR = 0.6, 95% CI 0.3–1.1), having allergic rhinitis (OR = 0.4, 95% CI 0.2–0.7), or a history of hayfever (OR = 0.3. 95% CI 0.2–0.6). Consuming unpasteurized milk during the first two years of life was also inversely associated with AEDS (OR = 0.4, 95% CI 0.2–0.8) and allergic rhinitis (OR = 0.3, 95% CI 0.1–0.7). On the other hand, consumption of pasteurized milk and cheese was not associated with any outcome variable.

Exposure to cats early in life was marginally protective (P = 0.07) against developing sensitization to cats, but the association between dogs and sensitization to dogs was weaker and nonsignificant.

Multivariate analysis

Those variables that showed associations (P = < 0.10) with any health outcome in the univariate analysis were combined in a model to assess their independent associations. Other variables, such as exposure to dogs and coal or wood fires, were included if an a priori hypothesis suggested they could be associated with an outcome. Both current and early-life exposures were included in the models. The following potential confounders were also included: gender, ethnicity, mother's education level, mother's smoking during the first year of the child's life and currently, family history of allergic disease, family size, antibiotic use in the first year of life, and having had measles and whooping cough infection. Table 4 presents the results of that model showing factors associated with having a history of hayfever, current allergic rhinitis, asthma, current wheeze, AEDS, and SPT positivity. Living on a farm during the first year of life was not significantly associated with any outcome although there was a tendency for the associations to be inverse. The positive univariate associations for farm abode currently and allergic disease remained for all disease and symptom outcomes, except for SPT positivity. There was a strong inverse association between regular yoghurt consumption by children under 2 years of age and having a history of hayfever and current allergic rhinitis, but the associations were weaker or nonexistent for the other outcome variables. The consumption of unpasteurized milk early in life was inversely associated with all outcome variables except hayfever. In general, exposure to pigs, cats, and dogs in the first year of life were also inversely associated with all outcomes. In contrast, exposure to poultry during the first year was associated with a nonsignificantly increased risk of developing allergic disease, but not SPT positivity—with associations for current poultry exposure less clearly in one direction. Notably, current exposures to animals, that had shown strong inverse associations when exposure occurred early in life, were generally associated with greatly increased risk of having that same outcome. When added to the model, the variables for current or early life exposure to sheep, cows, or cattle showed no association with any outcome variable.

Table 4.  Adjusted † odds ratios for the association between various exposures and having hayfever ever, allergic rhinitis in the last 12 months, asthma ever, wheeze in the last 12 months, atopic eczema/dermatitis syndrome (AEDS) ever, and skin prick test (SPT) positivity
 n  (293) Hayfever everCurrent allergic rhinitisAsthma everCurrent wheezeAEDS everSPT positivity
  • *

    P =  < 0.10.

  • **

    P =  < 0.05.

  • ***

    P =  < 0.01.

  • Adjusted for all variables in table, plus gender, ethnicity, mother's education level, family history of allergic disease, family size, antibiotic use in first year, mother's smoking in the first year and currently, coal and wood fires in the first year and currently, having a history of measles and whooping cough infection, and current dairy food consumption.

  • Per unit increase in endotoxin per gram of dust.

First year of life
 Farm abode 941.3 (0.4–3.9)0.5 (0.2–1.2)0.7 (0.3–1.8)0.5 (0.2–1.4)0.7 (0.3–1.8)1.3 (0.5–3.6)
 Regular poultry 361.8 (0.5–6.6)2.0 (0.7–5.9)2.7 (0.9–7.7)*2.1 (0.7–6.6)3.7 (1.3–10.7)**1.1 (0.4–3.5)
 Regular pig 290.4 (0.1–1.9)0.6 (0.2–2.0)1.0 (0.3–3.3)0.6 (0.2–2.3)0.6 (0.2–1.8)0.2 (0.1–0.9)**
 Cats inside or outside2230.4 (0.1–1.0)**1.4 (0.6–3.1)0.7 (0.3–1.5)1.0 (0.4–2.4)0.4 (0.2–0.8)***0.6 (0.3–1.3)
 Dogs inside or outside1850.5 (0.2–1.3)0.7 (0.4–1.4)0.4 (0.2–0.8)***0.6 (0.3–1.2)*0.8 (0.4–1.5)0.8 (0.4–1.6)
Current exposures
 Farm abode 951.3 (0.4–3.9)2.7 (1.0–6.9)**2.0 (0.8–5.2)1.9 (0.7–5.6)1.7 (0.7–4.1)0.6 (0.2–1.7)
 Regular poultry 452.2 (0.7–7.0)1.5 (0.6–3.8)0.8 (0.3–2.0)1.0 (0.4–2.6)0.5 (0.2–1.2)2.6 (1.0–6.9)**
 Regular pig 222.8 (0.6–12.2)1.0 (0.3–3.6)0.7 (0.2–2.3)1.6 (0.4–5.9)0.7 (0.2–2.2)3.3 (0.9–11.8)*
 Cats inside or outside2340.7 (0.3–1.9)1.0 (0.5–2.2)1.5 (0.7–3.3)0.9 (0.4–2.1)2.8 (1.3–6.1)***1.4 (0.6–3.3)
 Dogs inside or outside2141.5 (0.5–4.0)1.0 (0.5–2.2)1.6 (0.8–3.5)1.5 (0.7–3.4)1.3 (0.6–2.7)2.0 (0.9–4.3)
 Geomean endotoxin 0.9 (0.6–1.2)1.0 (0.8–1.3)0.9 (0.7–1.2)1.2 (0.9–1.5)1.0 (0.8–1.3)1.0 (0.8–1.3)
Diet at < 2 years
 Yoghurt once or more a week2250.3 (0.1–0.7)***0.3 (0.2–0.7)***1.1 (0.6–2.4)1.0 (0.4–2.3)0.6 (0.3–1.2)0.8 (0.4–1.7)
 Unpasteurized milk ever 381.1 (0.2–5.0)0.3 (0.1–1.1)*0.7 (0.2–2.4)0.6 (0.2–0.8)0.2 (0.1–2.2)**0.6 (0.2–1.9)
 Pasteurized milk once or more a day1921.7 (0.7–4.6)1.5 (0.7–3.3)1.3 (0.6–2.7)1.1 (0.5–2.5)1.4 (0.7–3.0)0.8 (0.4–1.7)
 Cheese once or more a week2002.1 (0.8–5.6)1.3 (0.6–2.8)1.1 (0.6–2.4)1.4 (0.5–3.3)1.3 (0.6–2.7)0.7 (0.3–1.4)

Variables for the main farm-produce during the first year of life and first-year farming abode were multicolineal, but after removing the latter from the model we were able to examine the associations of beef cattle, dairy, and sheep farms with each outcome. Beef cattle farming was inversely associated with hayfever (OR = 0.3, 95% CI 0.1–1.3), allergic rhinitis (OR = 0.4, 95% CI 0.1–1.4), current wheeze (OR = 0.2, 95% CI 0.1–1.00), and AEDS (OR = 0.4, 95% CI 0.1–1.1), but not with asthma or SPT positivity. Dairy farming was inversely associated with allergic rhinitis (OR = 0.2, 95% CI 0.0–0.6), asthma (OR = 0.3, 95% CI 0.1–1.1), current wheeze (OR = 0.2, 95% CI 0.1–0.9), AEDS (OR = 0.7, 95% CI 0.2–2.5), and SPT positivity (OR = 0.6, 95% CI 0.1–2.2), but not with hayfever. No effect was found for sheep farming. A similar analysis for main farm-produce currently showed no effects on any outcome variable.

We assessed the presence of multicollinearity between family history of allergic disease and living on a farm, and between first-year and current animal exposures. Since there were only small changes in the estimates and standard errors after removal from models of these potentially collinear variables, we concluded that multicolinearity was not biasing our results.

Discussion

Children in this study who currently live on a farm have more hayfever, allergic rhinitis, asthma, wheeze, and AEDS, but no more SPT positivity. Furthermore, for hayfever and asthma, the risk of farm-living was stronger among SPT + individuals. This may be due to a higher allergen load on farms that selectively provoke symptoms among atopic children. In contrast, farm abode during the first year of life was not associated with an increased risk of developing SPT positivity, or any allergic disease, with odds ratios tending to be less than 1.0. These findings are independent of some farming characteristics that appear to confer protection if exposure occurs early in life. For example, early-life exposure to pigs and dogs showed mainly inverse associations with outcome variables. Consumption of unpasteurized milk as part of the infant diet was also inversely associated with all allergic diseases, except hayfever, and was particularly strong for AEDS and allergic rhinitis.

Our findings contrast with the studies in Europe and Canada (15–19, 21–23) that have shown less allergic disease or atopy associated with childhood farm exposures. Various explanations have been proposed for the apparent protective effect of a farm childhood, including a more traditional lifestyle among farm children. The use of coal or wood fuels has been associated with a reduced prevalence of hayfever, atopy, and bronchial hyperreactivity (29). However, we found that the use of coal and wood was very common in both farm and nonfarm homes; most of our study population burnt these fuels. Our study population also seemed to have a different distribution of socioeconomic status than that found in the European studies, where parental education level was lower among farmers than nonfarmers (15, 16, 18, 30). The parents of farm children in our study were better educated than nonfarm parents, a factor that has been associated with an increase in the prevalence of hayfever (10).

A further difference between New Zealand and Europe may be the different distribution of animal exposures of farm and nonfarm children. For example, we found a high prevalence of pet ownership among all families, regardless of residence. Nonfarm children in Dannevirke may also have high exposure to farm animal effluent, as cattle and sheep are transported through the town to an abattoir just outside the town center. In addition, all the town schools take children from both the town and the surrounding rural areas, suggesting that town children may have contact with farm animals while visiting their friends on farms. Such factors may explain why a protective effect of animals was not reflected in less allergic disease and SPT positivity among the farm children in this study.

Barnes et al. (30) recently showed that within a rural area of Crete, consisting of villages and farms, there was no association between farm animals or milk products and atopy, despite showing protection for rural children compared with urban children. A limitation of the present study may be the lack of an urban comparison group, since urban children may have shown greater divergence in factors that determine the prevalence of allergic disease.

Rural–urban differences are less apparent in studies of Australian children (31, 32); indeed, one of these studies reported a lower prevalence of asthma and wheeze in Sydney compared with less populated regions (32). A study of Australian adults also reported less asthma (but not wheeze) in urban than rural areas (33). Compared to the present study, these studies did not differentiate between farming and small-town populations. However, within rural communities in Australia, livestock farming has been shown to be inversely associated with atopy, whereas crop farming was not (34), demonstrating the importance of the type of farm exposure, rather than farming itself, on determining allergic disease.

It has been postulated (15) that increased microbial exposure at an early age, which occurs in the presence of animals, may promote Th1 immunity, protecting against the development of allergy that is associated with Th2 immunity. In this respect, our findings support the European studies that showed a protective effect on atopic sensitization from regular contact with farm animals (17), and a reduced risk of allergic disease with increasing exposure to livestock (18). In the present study, however, the protective effect of animals only occurred in association with exposure in the first year of life, with no protective effect from current animal exposure. This supports the notion that it is indeed the early priming of the immune system towards a Th1 profile that protects people from developing allergic disease. However the data also suggest that some current animal exposures, perhaps in previously unexposed children, may represent a risk for allergic disease and SPT positivity. These findings were not explained by multicolinearity between first-year and current animal exposures. In contrast, early-life exposure to poultry showed positive associations with all outcome variables, and current exposure was associated with twice the risk of hayfever and SPT positivity. These findings are interesting in light of a study in Nepal (35) where asthma prevalence was two-fold greater if poultry were kept inside.

Interestingly, although we found no association for early-life exposure to sheep, cattle, or cows, we did find some strong protective effects if the main produce of the farm was “dairy cows” or “beef cattle”. We suggest that either the protection afforded by these animals may not result from direct exposure to them, or there is misclassification in the retrospective collection of exposure information, leading to a bias towards the null hypothesis.

It has also been proposed (15, 17) that childhood farm exposures in Europe have resulted in tolerization to the allergens found on farms, an observation first made by Blackley 130 years ago for hayfever (36). Platts-Mills et al. (37) showed that high exposure to cat was associated with a modified (nonallergic) Th2 response, possibly due to a form of tolerance. In the present study, the protective effect of exposure to cats early in life on the development of cat sensitization may be due to this process, but there is currently no evidence that tolerization occurs with high exposure to other allergens.

Endotoxin, found at high levels in association with animals, is a potent stimulant of Th1 responses, and has been proposed as an explanation for the lower prevalence of allergic disease among farm children (24). However, our finding of an increased risk of disease—but not SPT positivity—on farms is consistent with a role for endotoxin in exacerbating symptoms (38). It is difficult to explain why levels of endotoxin in this study were lower in farming homes, but the practice of removing all farming clothes and boots, and washing before entering the home may be effective at preventing high indoor levels. Certainly, living-room endotoxin levels are unlikely to reflect the high levels of exposure that farm children receive around animals, such as in stables. Despite the lack of association between farm abode and endotoxin, our findings are consistent with occupational studies in finding high levels of endotoxin in association with pigs (39) and poultry (40).

There is evidence that the composition of intestinal bacterial flora influences the development of the immune system. It has been shown that the probiotics found in yoghurt, lactobacilli (41) and bifidobacteria (42), influence development of the immune system towards a Th1 response. This is consistent with our finding of an inverse effect of yoghurt consumption at least once a week in infancy on the prevalence of hayfever, allergic rhinitis, and AEDS at age 7–10 years. A possible avoidance effect of dairy products in mothers of infants with allergic symptoms is unlikely to explain this effect because we found no similarly protective effect from consumption of pasteurized milk or cheese in infancy.

A similar mechanism may explain the reduced prevalence of some allergic diseases, especially AEDS and allergic rhinitis, that we found in association with the consumption of unpasteurized cow's milk early in life, because bacteria found in unpasteurized milk (43) would stimulate a Th1 immune response. Furthermore, unpasteurized milk has a high content of animal fat that may provide protection against asthma, eczema, and allergic rhinitis (44). In support of this, von Ehrenstein et al. (18) found that consumption of whole milk, but not skimmed milk, was associated with decreased prevalence of hayfever and asthma.

The low overall response rate of 60% was due to poor response rates (<50%) in two schools from low socioeconomic areas with predominantly nonfarm children. Thus selection bias cannot be excluded as an explanation for the positive associations we found for current farm abode with symptoms and disease.

Although the small size of our study population may limit our ability to generalize the study findings to other New Zealand farming communities, the farmers in this study represented a good cross-section of animal farmers (sheep, cattle, and dairy cows), and the climate was representative of much of New Zealand.

In conclusion, the increased risk of allergic disease among children currently living on farms contrasts with the weak protective effect of farm abode during the first year of life on some outcome variables. However, no similar increased risk of becoming SPT + is associated with farm-living. Nevertheless, we found that exposure to animals and consumption of yoghurt and unpasteurized milk early in life protects against allergic disease prevalence, and these findings are supportive of the “hygiene hypothesis”.

Acknowledgments

We would like to thank Dr Brian Quick, Rose Quick, and Patricia Deane for their local knowledge and help in Dannevirke, Evelyn Finlay and Stephanie Easthope for help with data collection in the schools, and Gordon Purdie for providing statistical advice. We acknowledge the huge contribution of the schools in and around Dannevirke, and the children and parents who participated in the study. The University of Otago provided funding for this project. Julian Crane was supported by a Health Research Council of New Zealand senior research fellowship, and Jeroen Douwes was supported by a research fellowship from the Netherlands Organization for Scientific Research (NWO).

Ancillary