Among farm women in the Agricultural Health Study, growing up on a farm was protective against both atopic and nonatopic asthma, with the greater effect for atopic asthma. Despite this reduced risk, use of pesticides, particularly insecticides, was positively associated with atopic asthma. We saw weak evidence of an interaction between growing up on a farm and use of pesticides with atopic asthma. Although personally mixing or applying pesticides increased the risk of atopic asthma, regardless of whether a woman grew up on a farm or not, the increase in risk was larger among those who grew up on farms. Current farm exposures did not confound the results for pesticides. Growing up on a farm also modified the risk of atopic asthma associated with currently working with animals. Women who did not grow up on farms had higher risk if they currently worked with animals, but women who grew up on farms had a lower risk if they currently worked with animals.
Pesticides, particularly organophosphate and carbamate insecticides, may contribute to airway reactivity and asthma among farmers. Early case reports suggested an association of organophosphate insecticides with exacerbation of asthma (17
); however, few population-based studies have explored this association. Beard and colleagues reported excess asthma mortality among pesticide applicators, particularly among those who applied pesticides during the organophosphate era (after 1962) (3
). In a cross-sectional study of 1,939 farmers in Saskatchewan, carbamate and organophosphate insecticides were associated with current asthma; the effect was greater for carbamates, particularly carbofuran (OR, 2.0; 95% CI, 1.3–3.2) (2
). In the current analysis, we also saw an approximately twofold increase in atopic asthma with carbofuran (OR, 1.92; 95% CI, 0.98–3.77). Among chlorpyrifos manufacturing workers, the prevalence of asthma and other respiratory diseases (as classified by International Classification of Diseases, ninth revision
[ICD-9], codes 490–496) was elevated (OR, 1.41; 95% CI, 0.95–2.09), suggesting a potential association with asthma; however, because all ICD-9 codes were combined, the increased prevalence may be associated with some other respiratory disease in this category (e.g., chronic bronchitis) (19
). Pesticide application was associated with asthma symptoms among Brazilian farmers, but there was no association with organophosphate insecticides or any other specific category of pesticide (20
). We observed ORs greater than 2.0 for the most potent organophosphates (parathion, phorate, and coumaphos) for atopic asthma, and a somewhat weaker association with malathion. Although we did not have information on wheeze among farm women, we previously saw increased wheeze associated with current use of specific pesticides among both the farmers (4
) and commercial pesticide applicators (5
) in the AHS. We saw increased wheeze with current use of specific pesticides, including coumaphos, malathion, parathion, phorate, and permethrin on animals; these same pesticides were also associated with atopic asthma among the farm women studied here. Results for current wheeze and prevalent asthma are not comparable, because the time periods of exposure differ, as do the patterns of pesticide use, between farmers and their wives (9
Recent animal studies have demonstrated that organophosphates induce airway hyperreactivity, potentially at doses below those causing acetylcholinesterase inhibition (6
). Pesticides may also modulate inflammatory responses to other farm exposures, such as endotoxin and allergens. For example, carbaryl, a commonly used carbamate insecticide, may enhance the allergenicity of dust mites (22
). Malathion, an organophosphate insecticide, stimulated inflammatory mediators, resulting in increased macrophage function in mice (23
). In rats treated with the organophosphate insecticide acephate at levels that did not inhibit acetylcholinesterase, Singh and Jiang observed significantly altered endotoxin-induced changes in the levels of proinflammatory cytokines, including tumor necrosis factor-α and inducible nitric oxide synthase mRNA (24
). Rats exposed via inhalation to dichlorvos, another organophosphate insecticide, had increased production of IFN-γ, especially after adrenalectomy, suggesting a role for inflammation in the response to this insecticide (25
). Specific pesticides (carbaryl, 2,4-d
, alachlor, and zineb) modulated endotoxin-induced macrophage activation through inhibition of nuclear factor-κB activation (26
). The primary metabolite of chlorpyrifos, chlorpyrifos oxon, enhanced the endotoxin-induced IFN-γ response in human blood cell cultures, suggesting that chlorpyrifos, in combination with endotoxin, may lead to an increased type 1 immune response (28
). Although limited, these data suggest that pesticides, including those associated with atopic asthma in this analysis, may interact with inflammatory pathways involving endotoxin processing.
Although atopic and nonatopic asthma may have different etiologies (13
), we only observed significant differences between these groups for “growing up on a farm” and for pesticide use, both overall and for carbamates. Associations for other farming exposures were similar between the two groups, except for animal exposures when stratified by “growing up on a farm.” Limited animal and in vitro
data suggest that pesticide exposure may alter the protective effect of early-life endotoxin (26
). Others have observed that use of quaternary ammonia compounds as disinfectants on farms was associated with atopic sensitization in the presence of endotoxin among adult farmers, suggesting that other chemical agents besides pesticides may interact with endotoxin to contribute to allergic asthma (30
). We saw a statistical interaction between pesticide use and “growing up on a farm,” but the interaction was limited to any pesticide use, and not specific chemicals (except maneb, for which the sample size was very small). We also observed that growing up on a farm confounded the pesticide results for atopic asthma; the effect estimates were lower when “growing up on a farm” was not included. This confounding could be the result of women who grew up on farms being more likely to apply pesticides than women who did not (62 vs. 50%), or it may be due to some biological interaction between early-life exposures and pesticide use.
We relied on self-reported cross-sectional data for this analysis. Self-reported doctor-diagnosed asthma is a reliable and valid endpoint (31
). Among farmers, subjects with self-reported asthma have a higher symptom prevalence and lower lung function than subjects without asthma (2
). We studied prevalent disease information; thus, some of our cases will have been diagnosed more recently than others. We did not have information on either respiratory or allergic symptoms, or on duration of disease. To assess potential occupational associations, we limited our analysis to cases of adult-onset asthma diagnosed after age 19 years. The prevalence of adult-onset asthma in our study is similar to that reported for current asthma among California farmers (2.7%) (33
) and Norwegian farmers (3.7%) (34
). Atopy was defined as a history of doctor-diagnosed eczema or hay fever; we (4
) and others (37
) have previously used similar questionnaire-based definitions for atopy. We created the two case groups based on responses to three questions (asthma, eczema, and hay fever) in a manner similar to that of Upton and colleagues, who used asthma with and without hay fever to define atopic and nonatopic asthma among adults (37
). Using this definition, 63% of adult subjects with asthma in Scotland in 1996 were classified as nonatopic, an estimate which is very similar to ours (60%). Among Norwegians, 70% of adult asthma was classified as nonatopic, with atopy defined by IgE (12
). Our atopy definition is based on a doctor's diagnosis of hay fever or eczema; thus, we may overestimate the prevalence of nonatopic asthma if people who experience allergic symptoms do not receive a hay fever diagnosis. Because the prevalence of nonatopic asthma is similar in our sample to that in two other populations, including one with IgE definition, we feel that such overestimation, if present, is minor. A potential explanation for our differential results between atopic and nonatopic asthma could be that factors affecting risk for atopic asthma reflect primarily risk of atopy, and not asthma per se
. Our sensitivity analysis in shows that atopy alone cannot fully explain the observed differences between atopic and nonatopic asthma, because the ORs for individual pesticides were consistently larger among those with atopic asthma than those with atopy alone.
Our exposure information was based on self-reported lifetime pesticide use, both total pesticide use and use of specific pesticides. For individual pesticides, we lacked duration or frequency information and were limited in our ability to assess dose response. Farmers have been demonstrated to provide reliable information regarding their personal pesticide use (38
), and we believe that farm women, particularly those who apply pesticides in agricultural settings, are also likely to do so. We have no reason to believe that pesticide use reporting would differ by asthma or atopy status, and, thus, reporting errors would tend to bias our results toward the null. Because we lacked information on duration of use, we relied on ever-use of a pesticide in a woman's lifetime; thus, women with one use of a particular chemical are combined with women who may have used the chemical for decades, which may dilute the association. Recall may be better for more commonly used pesticides and more recent use, and this feature may lead to underestimates of the association with rarely or historically used chemicals. We had no data on lifetime farming exposures that may contribute to asthma; hence, the findings for some specific pesticides (e.g., coumaphos and permethrin) may be an indicator that these are surrogates for historic animal exposures. However, a majority of the pesticides associated with asthma have not been used in animal production.
Farms represent a complex occupational setting with opportunities for a number of concurrent exposures, including multiple pesticides. We adjusted the pesticide models for other correlated pesticides. Only the commonly used insecticides, malathion and carbaryl, required adjustment. The OR for malathion, but not for carbaryl, remained significantly elevated after mutual adjustment. We controlled for other potential confounders, including smoking, in our base model. To further explore whether smoking was associated with atopic and nonatopic asthma, we limited our analysis to nonsmoking farm women and observed similar associations (data not shown), but the reduced sample size impaired power.
Our study of over 25,000 farm women represents the largest sample of farm women evaluated for adult asthma. Previous work among farmers and agricultural workers has generally been limited to a few types of farming (e.g., livestock or crops), and, except for a few studies, has not had the opportunity to focus on pesticides. In our sample, 57% of farm women applied pesticides at some point in their lives. The number and type of pesticides, as well as the frequency of application, differed among women, thus providing us an opportunity to look at individual pesticides while controlling for correlated exposures. Our findings are consistent with a protective effect of early-life farm exposures on asthma, and suggest that pesticides, particularly organophosphate insecticides, may increase asthma risk.