Asthma and Farm Exposures in a Cohort of Rural Iowa Children

Epidemiologic studies of farm children are of international interest because farm children are less often atopic, have less allergic disease, and often have less asthma than do nonfarm children—findings consistent with the hygiene hypothesis. We studied a cohort of rural Iowa children to determine the association between farm and other environmental risk factors with four asthma outcomes: doctor-diagnosed asthma, doctor-diagnosed asthma/medication for wheeze, current wheeze, and cough with exercise. Doctor-diagnosed asthma prevalence was 12%, but at least one of these four health outcomes was found in more than a third of the cohort. Multivariable models of the four health outcomes found independent associations between male sex (three asthma outcomes), age (three asthma outcomes), a personal history of allergies (four asthma outcomes), family history of allergic disease (two asthma outcomes), premature birth (one asthma outcome), early respiratory infection (three asthma outcomes), high-risk birth (two asthma outcomes), and farm exposure to raising swine and adding antibiotics to feed (two asthma outcomes). The high prevalence of rural childhood asthma and asthma symptoms underscores the need for asthma screening programs and improved asthma diagnosis and treatment. The high prevalence of asthma health outcomes among farm children living on farms that raise swine (44.1%, p = 0.01) and raise swine and add antibiotics to feed (55.8%, p = 0.013), despite lower rates of atopy and personal histories of allergy, suggests the need for awareness and prevention measures and more population-based studies to further assess environmental and genetic determinants of asthma among farm children.

Most epidemiologic studies of childhood asthma have been conducted among inner-city or urban populations and have found asthma prevalence to vary by location, likely attributable to differing environmental exposures [International Study of Asthma and Allergies in Children (ISAAC) Steering Committee 1998]. Studies of rural childhood asthma are of particular interest because they have consistently reported that farm children are less often atopic (Braun-Fahrlander et al. 1999;Downs et al. 2001;Riedler et al. 2000Riedler et al. , 2001, have lower rates of allergic diseases (Braun-Fahrlander et al. 1999;Kilpelainen et al. 2000;Riedler et al. 2000Riedler et al. , 2001Von Ehrenstein et al. 2000;Wickens et al. 2002), and in several reports also have lower rates of asthma (Ernst and Cormier 2000;Kilpelainen et al. 2000;Riedler et al. 2000Riedler et al. , 2001Von Ehrenstein et al. 2000). These findings are consistent with the hygiene hypothesis, which posits that childhood allergy risk is immunologically modulated in early life by exposure to infectious agents. However, several studies have not found positive associations between asthma and asthma symptoms among children and farm exposures, raising questions regarding the influence of unmeasured risk factors and/or selection in these cross-sectional studies (Chrischilles et al. 2004;Downs et al. 2001;Salam et al. 2004;Wickens et al. 2002).
It is recognized that asthma risk is conveyed by a complex interaction of genetic and environmental determinants, which makes the epidemiologic investigation of farmrelated asthma difficult (Douwes et al. 2001;Niven 2003;Schwartz 2001). International studies of childhood asthma among farm children have typically measured atopy to gauge genetic predisposition to asthma but have less consistently described and measured farm environment risk factors, often using endotoxin as an indicator of exposure to infectious agents early in life. Although endotoxin is a ubiquitous exposure in agriculture, its concentration varies within and between farm types, and it is but one of many agricultural respiratory exposures children may encounter (Douwes et al. 2003;Reynolds et al. 1996;Schenker et al. 1998).
Over the last three decades, the development of a vertically integrated livestock industry has significantly reduced the number of U.S. family farms raising hogs, poultry, and cattle but has rapidly increased the number of large animal-feeding operations (AFOs) (National Academy of Sciences 2003). Although inflammatory airway diseases, including asthma, chronic bronchitis, organic dust toxic syndrome, and progressive airway obstruction, are now well documented among AFO workers (Schenker et al. 1998), there has been much less research regarding exposures and health outcomes among AFO-exposed children and community-based residents (Reynolds et al. 1997a;Salam et al. 2004;Thu et al. 1997;Wing and Wolf 2000).
The Keokuk County Rural Health Study (KCRHS) is a large, population-based study of a cohort of rural families living in an intensely agricultural region of southeastern Iowa (Merchant et al. 2002). The aim of the present study was to estimate asthma prevalence and assess whether farm exposures result in less atopy, less allergic disease, and less asthma, while taking into account multiple personal and other environmental risk factors, among this cohort of farm children.

Materials and Methods
The study population. This study reports data on children from birth through 17 years of age collected in round 1 of the KCRHS, which began in 1994 and ended in 1998. Keokuk County was chosen because it is intensely agricultural and entirely rural. A stratified, random sample that identified households from farm, town, and rural nonfarm locations was used. A total of 2,496 eligible households were identified. Details regarding the sampling methodology and survey methods have been reported previously (Merchant et al. 2002). All members of enrolled households were invited to a centrally located research facility for interviews, and all adults and children ≥ 8 years of age were invited for medical examinations. One adult per household was interviewed by a trained interviewer about the health of all of the children (from birth but < 18 years of age) living in the household.
Questionnaire. The childhood respiratory questionnaire chosen for this study was that used in University of Southern California studies of childhood asthma in Los Angeles (Peters et al. 1999). We used four asthma outcomes to estimate asthma prevalencedoctor-diagnosed asthma, asthma/medication for wheeze (doctor-diagnosed asthma and/or medication for wheeze in the last 12 months), current wheeze, and cough with exercise. These four asthma outcomes, severe symptoms consistent with asthma, atopy, an early respiratory illness, and a high-risk birth are fully defined in the definition section of the online version this article. The parent's response to the questionnaire also provided information regarding parental farm exposures, maternal smoking during pregnancy, household exposure to tobacco smoke, parental education, and household income.
Clinical assessment. Children ≥ 8 years of age were invited to complete a medical examination that included skin prick testing (SPT), spirometry, methacholine challenge testing, and height and weight measurements to calculate 95th percentile body mass index (kilograms per square meter) (Must et al. 1991). A total of 18 aeroallergens common to the Midwest, a histamine-positive and normal saline-negative control, were used for SPTs. Common rural aeroallergens included tree pollen mix, grass pollen mix, ragweed pollen, weed pollen mix, cockroach mix, mold mix, insect mix, caddis fly/moth/mayfly mix, cat pelt, dog hair, mouse and rat mix, and dust mite Der f and Der p mix. Farm aeroallergens included grain dust mix or grain smut mix, soybean dust or soybean whole grain, cattle hair, horse hair, chicken feathers, and turkey feathers. Children taking antihistamines and other medications known to affect skin test results, those with histories of past systemic reactions to allergy skin testing, and any participant who might have been pregnant were excluded from skin testing. A wheal ≥ 3 mm in diameter was defined as a positive reaction; subjects were considered atopic by SPT if they had a positive reaction to any two of the allergens tested. Spirometry was completed on a rolling-seal spirometer that conformed to American Thoracic Society (1995) guidelines. Contraindications to methacholine testing included participants with a baseline forced expiratory volume in 1 sec (FEV 1 ) of < 70% of predicted or FEV 1 < 1.5 L, pregnancy or suspected pregnancy, lactation, current use of a β-adrenergic blocking agent, and a decline in FEV 1 of ≥ 15% to the diluent. Methacholine was administered by dosimeter in five serial doses of 0.025, 0.25, 2.5, 10.0, and 25.0 mg/mL, with 3 min between doses (Crapo et al. 2000). Bronchial hyperresponsiveness was defined as having a drop in FEV 1 of ≥ 20% from the postsaline control (PC20), following inhalation of ≤ 8 mg/mL of methacholine (Anto 1998;Crapo et al. 2000).
Serum analysis. Sera were collected from subjects (n = 217) at the time of SPT and analyzed for total and specific IgE. Total IgE was measured by immunoassay using murine monoclonal anti-human IgE as the capture antibody (CLB, Sanguin Blood Supply Foundation, Amsterdam, the Netherlands), rabbit anti-human IgE as the second antibody (Dako, Corp., Carpinteria, CA), and peroxidase-conjugated donkey anti-rabbit IgG as the labeling antibody (Research Diagnostics, Inc., Flanders, NJ) in a TMB substrate system (Pierce Endogen, Rockford, IL). Standard curves were generated using an IgE CAP system standards (Pharmacia Diagnostics, Uppsala, Sweden) with the standard curve from 0.02 to 10 kU/L. Sera were studied at initial dilutions of 1:20, 1:40, 1:80, and 1:160, with higher dilutions run for high IgE sera. Individuals were considered to be atopic by IgE if their total IgE was ≥ 60 kU/L (Contreras et al. 2003).
Environmental assessment. An industrial hygienist visited each household shortly after the clinic visit and completed a home environmental questionnaire and checklist, when applicable a farm environmental questionnaire and farm environmental checklist, and measurement of a limited number of environmental parameters. Details of these environmental assessments have been published previously (Park et al. 2003;Reynolds et al. 1997b). Assessments of specific environmental exposures were taken from these instruments, including several farm operation questions, livestock and antibiotics in animal feed questions, and questions regarding gas stoves, heating with wood, exposure to pesticides, exposure to cats and dogs as pets, and dehumidifier use.
Household type was determined at the time of the child's birth from the biologic mother's reproductive history questionnaire and through follow-up phone interviews with the biologic mother regarding residence type (farm, rural nonfarm, or home) at the time of birth. Children's various farm tasks and the age each task was first performed were determined from a questionnaire on childhood tasks from available KCRHS round 2 data and from follow-up phone administration of this questionnaire to round 1 participants who had not participated in round 2.
Statistical analysis. Chi-squared tests and analysis of variance were used to evaluate any differences among demographic, personal, and environmental risk factors for farm, rural nonfarm, and town households. Univariable logistic regression was used to identify variables that were significant (p < 0.1) for doctordiagnosed asthma, asthma/medication for wheeze, chronic wheeze, and cough with exercise. Multivariable logistic regression was then used to identify significant (p < 0.05) variables in the final models.
Initial data analyses was performed with SAS (version 8; SAS Institute, Inc., Cary, NC) software. SUDAAN software (Research Triangle Institute, Research Triangle Park, NC) was then used to adjust variance estimates for potential intrahousehold correlation resulting from the inclusion of more than one child per household.
The study was approved annually by The University of Iowa institutional review board. A parent or legally authorized representative of each child participant provided written informed consent. Children 8-17 years of age gave their assent.

Cohort description. Of the 2,496 Keokuk
County households eligible for this study, 1,675 households (67.1%) initially contacted by letter and telephone agreed to participate immediately or to be contacted at a later date. Enrollment stopped when the goal of 1,000 households was reached. A total of 1,004 households (336 farm, 206 rural nonfarm, 462 town households) enrolled and participated in round 1 of the study.
The cohort, which consisted of 644 children (224 farm, 155 rural nonfarm, and 265 town), did not differ in age among household types, was somewhat overrepresented by boys in farm and rural nonfarm households, and was 97.7% Caucasian. Of the 336 farms in the cohort, 109 had children. Complete data on all farming characteristics were available on 89 farms with children and on 172 farms without children. These farms produced primarily corn, soybeans, and hogs but very few other livestock. Farms with children were somewhat smaller (434 total acres in production) than farms without children (468 total acres in production) but were otherwise similar, except that farms with children on average raised more hogs (298 vs. 141, p = 0.03). Fifty percent of farm children were reported by a parent to perform tasks around hogs, compared with ≤ 16% for rural nonfarm or town children, whereas 40% of farm children were reported to perform tasks around cows compared with ≤ 13% for rural nonfarm or town children.
Health outcomes. Ninety-five percent of the children's data were provided by the child's biologic mother or female guardian. Complete data on asthma outcomes were available on 610 children. Concordance between the four asthma outcomes varied from strong to weak: doctor-diagnosed asthma (asthma/medication for wheeze κ = 0.81, p < 0.0001; current wheeze κ = 0.31, p < 0.0001; cough with exercise κ = 0.26, p < 0.0001), asthma/medication for wheeze (current wheeze κ = 0.53, p < 0.0001; cough with exercise κ = 0.39, p = 0.11; current wheeze and cough with exercise κ = 0.27, p = 0.73). Only 4.4% of participants were captured by all four asthma outcomes, whereas 33.6% of all 610 participants were captured by at least one asthma outcome. Children with doctor-diagnosed asthma included only a third (8 of 24) of the children with severe symptoms consistent with asthma, whereas children with any one of the four asthma outcomes captured 23 of 24 children with severe symptoms. Of the 394 children 8-17 years of age, 351 (89.1%) had SPT, 347 (88.1%) had pulmonary function tests, and 215 (61.2%) agreed to have blood drawn for sera. Agreement between total individual IgE and SPT results (Aspergillus, cat hair, cockroach, weed mix, tree pollen, Der p, and Der f) ranged from 72.8 to 89.1%.
Children who were born on a farm had a lower prevalence of atopy (IgE), a lower prevalence of diagnosed allergies and a higher forced vital capacity (likely attributable to hyperinflation) (Table 1). Children who currently lived on a farm were somewhat more likely to be boys and somewhat less likely to have diagnosed allergies (Table 1).
A very high proportion of children who lived on a farm at the time of study (currently lives on a farm) were born when their parents lived on a farm (born on a farm) and continued to live on a farm (data for those who lived on a farm during the first year of life or through age 5, or had a parent who continued to work on a farm, were also analyzed but not reported). Because univariable associations were similar for all farm versus nonfarm groups, only comparisons of born on a farm and currently living on a farm exposure results are presented (Table 2). Farm children were consistently exposed to less tobacco smoke but were more often exposed to wood stoves, conditions resulting in dehumidifier use, cats as pets, and application of pesticides outside the home. Farm children's parents were more often better educated and had a household annual income of ≥ $20,000 (Table 2).
Univariable associations among the four asthma outcomes and environmental risk factors are presented in Tables 3 and 4. A weak association was observed between doctor-diagnosed asthma and less parental education. A near significant association was observed between doctor-diagnosed asthma/medication for wheeze and living on a farm raising swine and a significant association with living on a farm that adds antibiotics to feed. No significant association was observed with environmental exposures and current wheeze, but significant negative associations were observed between cough with exercise and exposure to wood smoke and applied pesticides outside home in the last year, significant positive associations were observed with dogs as household pets, and near significant positive associations were observed with living on a swine farm and living on a farm that added antibiotics to feed. Tables 5 and 6 present univariable associations among the four asthma health outcomes and personal and clinical risk factors and health measures, which reveal similar association patterns but a few significant differences.
Multivariable models that included personal or environmental risk factors with univariable significance of p < 0.1 for any of the four asthma outcomes are presented in Table 7. In addition to sex, age, history of allergies, family history of allergies, premature birth, early respiratory infection, and highrisk birth, an interaction term (living on a farm that raised swine and added antibiotics to feed) was independently associated with asthma/medication for wheeze, current wheeze (p = 0.06), and cough with exercise. Of farms that raised swine, 24 of 43 (55.8%) added antibiotics to feed. Of livestock farms that add antibiotics to feed, 24 of 31 farms or 77.4% raise swine. Those farms that add antibiotics to feed were found to have larger mean numbers of livestock than those that did not add antibiotics to feed (750 vs. 392 animals; p = 0.0002). Examination of children who lived on farms raising swine and adding antibiotics to feed found that 55.8% (p = 0.013) reported at least one of the four asthma outcomes (Figure 1).

Discussion
This study reports uniformly high-prevalence estimates of asthma and asthma-related symptoms that are consistent with asthma prevalence observed in studies of U.S. urban populations (Bauer et al. 1999;ISAAC Steering Committee 1998). These high asthma prevalence estimates, 9.6 ± 5.0 (n = 217) 9.6 ± 4.9 (n = 377) 1.00 (0.96-1.04) 1.00 10.0 ± 4.9 (n = 224) 9.5 ± 4.9 (n = 420) 1.02 (0.98-1.06) 0.36 No. of siblings < 18 years of age 1.6 ± 1.2 (n = 217) 1.4 ± 1.0 (n = 377) 1.15 (0.87-1.53) 0.33 1.5 ± 1.2 (n = 224) 1.5 ± 1.0 (n = 420) 1. and our finding of a high proportion (twothirds) of children with severe symptoms consistent with asthma but without a doctor diagnosis of asthma, are consistent with the findings of our Rural Childhood Asthma Study (Chrischilles et al. 2004) and underscore the need for asthma screening programs, for improved rural health care provider asthma diagnostic and management skills, and for health policies that would improve access and insurance coverage for rural children. A history of diagnosed allergies was found to be less common among children who lived on a farm in the first year of life, a finding consistent with many other studies of farm children (Braun-Fahrlander et al. 1999;Kilpelainen et al. 2000;Riedler et al. 2000Riedler et al. , 2001Von Ehrenstein et al. 2000). The three estimates of atopy also tended to be lower among children who lived on a farm in the first year of life, as reported by others (Braun-Fahrlander et al. 1999;Riedler et al. 2000Riedler et al. , 2001. However, asthma and asthmalike symptom prevalences were found to be high and to not differ between children with farm exposures and those without farm exposures, unlike the findings of others (Ernst and Cormier 2000;  Abbreviations: CI, confidence interval; OR, odds ratio. No significant association (p < 0.1) was observed for any asthma outcome for the following variables: farm residence, born on a farm, lived on a farm for at least 3 months while < 1 year of age, lived on farm for at least 3 months while < 5 years of age, parent does farm work, maternal smoking during pregnancy, current household exposure to tobacco smoke, ever household exposure to tobacco smoke, gas stove in home for cooking, burn wood for fuel, current dehumidifier use in home, household income (< $20,000), household pets: cats, household pets: dogs, applied pesticides in home during past year, applied pesticides outside home during past year, or raise livestock.  Kilpelainen et al. 2000;Riedler et al. 2000Riedler et al. , 2001Von Ehrenstein et al. 2000), despite lower rates of allergic disease and atopy and a significantly lower exposure to household tobacco smoke among farm children. However, as depicted in Figure 1, these excesses are found only among children living on farms raising swine, whereas a lower prevalence of asthma was observed among farm children not raising swine compared with nonfarm children, which is consistent with the aforementioned studies.
Children's Health | Merchant et al. 354 VOLUME 113 | NUMBER 3 | March 2005 • Environmental Health Perspectives Abbreviations: BMI, body mass index; CI, confidence interval; FEF, forced expiratory flow; NICU, neonatal intensive care unit; OR, odds ratio. a High-risk birth is defined as premature birth, hospitalization in an NICU, use of oxygen following birth (not including resuscitation at birth), or use of oxygen at home after leaving the hospital. b Adjusted for age, height, and sex.  Farms in Northern Europe tend to be smaller than Iowa farms and to have livestock that are often housed in immediate proximity to living quarters, and these farm families have been described as more traditional in their way of life. Farms in Canada, Australia, and New Zealand are described as larger but typically not as livestock intensive as Iowa farms (Downs et al. 2001;Ernst and Cormier 2000;Wickens et al. 2002). Keokuk County farm families do not live in immediate proximity to livestock buildings but do usually live on the same acreage, typically with many farm family members participating in the farm operation. It is common for young children to be exposed to farming operations, including AFOs, as they accompany a parent or sibling in assisting with farm tasks (Park et al. 2003). Farm children in Keokuk County were reported by their parents to be exposed as bystanders to farm tasks around livestock as early as 1 year of age; however, such tasks around livestock were typically done by male adolescents. Although no environmental measurements of farm task exposures were made, several studies conducted in Iowa document high levels of occupational exposures to respirable and total dust, endotoxin, hydrogen sulfide, and ammonia, which have been associated with asthma, chronic bronchitis, cross-shift declines in lung function, and progressive declines in lung function over time among those working in AFOs (Reynolds et al. 1996;Schenker et al. 1998;Schwartz et al. 1995). It is therefore probable that some swine-farm-exposed children had high exposures to endotoxin and other AFO exposures and that some of the asthma and asthma symptoms observed among these farm youth are attributable to occupational exposures.
Multivariable models for doctor-diagnosed asthma/medication for wheeze and cough with exercise found that raising swine and adding antibiotics to feed were independently associated with these health outcomes. Because farms that add antibiotics to feed were much larger than those that did not add antibiotics to feed, adding antibiotics to feed may serve as an indicator of larger swine operations. However, it is plausible that antibiotic exposures may be playing some causal role because antibiotics have been documented to be components of emissions from AFOs (Hamscher et al. 2003;Svendsen et al. 2003) and, when consumed for medical purposes, have been associated with childhood asthma (Wickens et al. 1999). These high asthma estimates make clear that on-farm exposure to swine production is associated with asthma among children living on these farms and that swine production contributes to the higher prevalence of asthma outcomes in this livestock-intensive rural community. More detailed assessment of the temporal relationships between childhood farm exposures, including measurements of endotoxin-laden dust, irritant gases, and antibiotics in relation to asthma estimates, is needed to further our understanding of these relationships.
Other events early in life, apart from farm exposures, including premature birth, a respiratory infection at ≤ 3 years of age, and highrisk birth, were independently associated with asthma outcomes in this study, also consistent with other studies of childhood asthma (Farooqi and Hopkin 1998;Von Mutius et al. 1993). These early-life risk factors, which did not differ between farm and nonfarm participants in this study, may confound assessment of farm exposures in populations where farm families are poorer and have less adequate prenatal health care.
Two studies of nonfarm infants have evaluated the role of endotoxin exposures early in life and have reported no relationship between endotoxin levels and atopy, allergic disease, and asthma (Bolte et al. 2003;Park et al. 2001), findings inconsistent with the hygiene hypothesis. Another contributing explanation, which has been recognized, but only indirectly assessed (Braun-Fahrlander et al. 1999;Downs et al. 2001;Ernst and Cormier 2000;Leynaert et al. 2001), is the potential unmeasured effect of systematic genetic selection of those susceptible to farmrelated respiratory disease away from farming over successive generations. It is common for farm youth to leave the farm in Keokuk County, so much so that we have reported a significant deficit of asthma among adult farm men compared with other men in Keokuk County (Merchant et al. 2002).
Because indicators of asthma associated with common farm exposures are influenced by genotypic patterns (Arbour et al. 2000;Gilliland et al. 2004), epidemiologic studies of genotype among farm family generations could help define patterns of differential selection of atopic, allergic, and asthmatic members of farm families away from farming.
Limitations of this study include the relatively small numbers of children with clinical data. Also, this study was not designed to address the question of whether exposures to dust, irritant gases, and odors arising from AFOs may be associated with respiratory symptoms or health conditions among rural residents living in proximity to farms with AFOs. However, the few community-based studies of AFO exposures have reported higher rates of airway symptoms (Reynolds et al. 1997a;Thu et al. 1997;Wing and Wolf 2000), and significant peaks in asthma hospital visits have been observed following peak exposures to total reduced sulfur (for children) and to hydrogen sulfide (for adults) from a large animal waste treatment complex (Campagna et al. 2004). As the result of these findings and community complaints about odor, several states now regulate some combination of hydrogen sulfide, total reduced sulfur, ammonia, and odor. Given our finding of a high prevalence of asthma outcomes among farm children living on swine farms, it is clear that farm parents should be aware of this risk and take precautions to reduce childhood respiratory exposures from AFOs. Evaluation of asthma outcomes and environmental exposures among school children and rural residents living proximate to AFOs remains an important research priority.