|Home | About | Journals | Submit | Contact Us | Français|
Acute organophosphate and carbamate pesticide poisonings result in adverse cardiac outcomes. The cardiac effects of chronic low-level pesticide exposure have not been studied. The authors analyzed self-reported lifetime use of pesticides reported at enrollment (1993–1997) and myocardial infarction mortality through 2006 and self-reported nonfatal myocardial infarction through 2003 among male pesticide applicators in the Agricultural Health Study. Using proportional hazard models, the authors estimated the association between lifetime use of 49 pesticides and fatal and nonfatal myocardial infarction. There were 476 deaths from myocardial infarction among 54,069 men enrolled in the study and 839 nonfatal myocardial infarctions among the 32,024 participants who completed the follow-up interview. Fatal and nonfatal myocardial infarctions were associated with commonly reported risk factors, including age and smoking. There was little evidence of an association between having used pesticides, individually or by class, and myocardial infarction mortality (e.g., insecticide hazard ratio (HR)=0.91, 95% confidence interval (CI): 0.67, 1.24; herbicide HR=0.74, 95% CI: 0.49, 1.10) or nonfatal myocardial infarction incidence (e.g., insecticide HR=0.85, 95% CI: 0.66, 1.09; herbicide HR=0.91, 95% CI: 0.61, 1.36). There was no evidence of a dose response with any pesticide measure. In a population with low risk for myocardial infarction, the authors observed little evidence of increased risk of myocardial infarction mortality or nonfatal myocardial infarction associated with the occupational use of pesticides.
Studies in France, Sweden, and the United States showed lower rates of cardiovascular morbidity and mortality among farmers when compared with the general population (1–5). The reasons for these low rates are not well understood, although some evidence suggests that they could be due to differences in lifestyle risk factors, such as smoking and exercise (6). Although farmers have been shown to have lower myocardial infarction rates when compared with the general population, this would not preclude farmers with a specific exposure from having higher rates than farmers without the exposure. Unfortunately, little is known about whether particular agricultural exposures may be associated with cardiac events.
Farmers are exposed to diesel exhaust and inhaled dust, both of which have been associated with risk for myocardial infarction (7, 8). In addition, acute organophosphate and carbamate pesticide poisoning has been associated with adverse cardiac manifestations, including arrhythmias, hypertension, and sudden death (9–11). An ecologic study of residents of agricultural counties with high wheat production in the midwestern United States, which served as a proxy for phenoxy herbicide exposure, reported an increased risk of myocardial infarction mortality compared with those in low wheat production counties (12).
Chronic pesticide exposure has not been directly assessed as a potential risk factor for myocardial infarction. Although many farmers apply pesticides on a regular basis over the course of their careers, little is known about the implications of such exposures on cardiac health and function. The Agricultural Health Study provides an opportunity to evaluate specific agricultural exposures and risk of myocardial infarction within a cohort of farmers. We analyzed data on male pesticide applicators in the Agricultural Health Study to assess whether pesticides and other agricultural exposures were associated with fatal and nonfatal myocardial infarction.
The Agricultural Health Study is a prospective study of licensed pesticide applicators, primarily farmers with pesticide licenses, in North Carolina and Iowa. Participants (n =57,311) enrolled in 1993–1997 by filling out a self-administered questionnaire about their farming practices, pesticide use, and health when they sought, or renewed, their pesticide license. Forty-four percent of the cohort answered an additional questionnaire by mail. Five years after enrollment, 70% of the original cohort completed a follow-up phone interview conducted in 1999–2003 to obtain information on health conditions diagnosed since enrollment. The cohort is linked annually to state cancer registries and state and national death registries to update cancer incidence, vital status, and losses to follow-up.
We assessed both myocardial infarction mortality and nonfatal myocardial infarction incidence. Deaths from myocardial infarction were recorded from state and national death records starting at enrollment through December 31, 2006. Myocardial infarction mortality was defined as any death attributed primarily to myocardial infarction or with myocardial infarction listed as a contributing cause on the death certificate by International Classification of Diseases, Tenth Revision, codes I21–I22. Nonfatal myocardial infarction incidence was determined on the basis of a positive response on the 5-year follow-up questionnaire to the question, “Has a doctor or other health professional ever told you that you had a heart attack (or myocardial infarction)?” If the respondent answered yes, the age at diagnosis was recorded. Participants who reported a myocardial infarction at enrollment were excluded. First myocardial infarctions occurring at an age greater than or equal to the age at enrollment were counted as incident myocardial infarction. Analyses for myocardial infarction mortality and nonfatal incidence were conducted separately, because longer follow-up was available for the mortality analysis, because mortality information is available for all participants through 2004, and because incidence information was obtained from the first 5-year follow-up questionnaire conducted in 1999–2003 (median mortality follow-up: 11.8 years; median incidence follow-up: 5.0 years). In addition, the groups at risk for the 2 outcomes differed. Mortality information was available for the entire cohort, but only those answering the 5-year follow-up questionnaire (~70% of the cohort) were eligible for the nonfatal incidence analysis. Female study participants were excluded from the analysis because of different myocardial infarction trends and risk factors for females when compared with males. The mortality analysis included all male pesticide applicators (n=54,609) who enrolled in the study and provided complete data on all covariates (Figure 1). The incidence analysis was further limited to those who completed the follow-up telephone interview and who reported no myocardial infarction prior to enrollment (n=32,024).
We investigated associations between pesticide use, as well as other agricultural exposures, and myocardial infarction mortality and nonfatal myocardial infarction. All agricultural exposure information was collected at enrollment, including self-reported ever use of 50 pesticides. In addition, detailed application information was collected for 22 pesticides at enrollment, including years used, days per year used, and use in the last year. Detailed information on the remaining 28 chemicals was collected on a take-home questionnaire completed by approximately 40% of the cohort. We evaluated ever use of the pesticides individually and grouped into functional classes (herbicides, fumigants, fungicides, and insecticides), as well as categorical lifetime days of exposure to any pesticide and to individual pesticides (0, 1–50, 51–250, ≥251 days) defined as the product of years of use and days per year used. Analyses were limited to pesticides with at least 5 cases at each exposure level. We also evaluated past acute exposures for the subset with the take-home questionnaire that asked participants if they had ever had an event resulting in an unusually high pesticide exposure and if a doctor had ever told them that they had pesticide poisoning.
We also considered several other factors to evaluate the potential influence of nonpesticide agricultural exposures, such as time spent driving trucks, combines, diesel tractors, and gasoline tractors, as well as farm size and number of poultry and other livestock on the farm. The frequencies of agricultural practices were obtained at enrollment and were collapsed into categories with greater than 5 cases at each level of exposure.
For both the mortality and incidence analyses, hazard ratios were estimated by using Cox regression adjusting for state, age in 10-year categories, and smoking status defined as whether or not the participant had smoked 100 cigarettes in his lifetime. In addition, the incidence base model included categorical body mass index (<25, 25–<30, ≥30 kg/m2). These inclusions were made on the basis of change in estimate and goodness-of-fit tests. The hours of weekly exercise, race, alcohol consumption, family history of myocardial infarction before age 50 years, and educational level were also considered but did not confound the estimates. All covariate information was obtained from the enrollment and take-home questionnaires. The mortality time variable is defined as the time from enrollment to either death or December 31, 2006, the last date of complete death certificate data. The time from enrollment to either myocardial infarction incidence or the 5-year follow-up interview (5.0-year median follow-up) was used in the nonfatal incidence analysis.
The myocardial infarction mortality analysis used Agricultural Health Study data release AHSREL0612.01, and the incidence analysis used releases P1REL0506 and P2REL0612.01. All statistical modeling was done in SAS, version 9.1.3, statistical software (SAS Institute, Inc., Cary, North Carolina).
Of the 54,069 participants eligible for the mortality analysis, 476 experienced a death from myocardial infarction over a median 11.8-year follow-up period (mortality rate, 78.0 deaths per 100,000 person-years). There were 839 nonfatal myocardial infarctions among the 32,024 participants in the incidence analysis over a median 5.0-year follow-up period (incidence rate, 473.8 events per 100,000 person-years). Characteristics of the Agricultural Health Study male pesticide applicators by cardiovascular disease outcome are presented in Table 1. For both the mortality and incidence study groups, the characteristics shown in prior research to be risk factors for myocardial infarction exhibit the expected trends. For example, participants of advanced age and nonwhite race, those with a family history of myocardial infarction and less education, and those who ever smoked were at increased risk for both fatal and nonfatal myocardial infarction. Obesity, hypertension, and diabetes were also associated with an increased risk of myocardial infarction in the cohort. Those who drank one or more alcoholic drinks per month were less likely to experience myocardial infarction when compared with nondrinkers in both study groups.
Fatal and nonfatal myocardial infarction by ever use of individual pesticides and by functional classes is shown in Table 2. The insecticide trichlorfon did not have a sufficient number of cases to be included. Of the 49 individual pesticides qualifying for both mortality and incidence analysis, 6 estimates had a statistically significant positive association for either mortality or incidence. Ethylene dibromide (hazard ratio (HR)=1.54, 95% confidence interval (CI): 1.05, 2.27), maneb/mancozeb (HR=1.34, 95% CI: 1.01, 1.78), and ziram (HR=2.40, 95% CI: 1.49, 3.86) were associated with myocardial infarction mortality, while aldrin (HR =1.20, 95% CI: 1.01, 1.43), dichlorodiphenyltrichloroethane (DDT) (HR=1.24, 95% CI: 1.04, 1.46), and 2,4,5-trichlorophenoxyacetic acid (2,4,5-T) (HR= 1.21, 95% CI: 1.03, 1.43) were associated with nonfatal myocardial infarction incidence. In addition, 5 pesticides were inversely associated with myocardial infarction mortality: carbaryl, terbufos, imazethapyr, pendimethalin, and petroleum oil. No pesticide was statistically associated with both fatal and nonfatal myocardial infarction.
We observed no associations between myocardial infarction and lifetime days of pesticide use for any individual pesticide (Web Table 1). (This information is described in a supplementary table posted on the Journal's website (http://aje.oxfordjournals.org/).) In addition, no statistically significant dose-response trend was present for overall lifetime days of use of any pesticide, although results were suggestive for nonfatal myocardial infarction (Table 3). Neither high-pesticide-exposure events nor a doctor's diagnosis of pesticide poisoning was associated with myocardial infarction risk.
Other agricultural exposures also were not positively associated with myocardial infarction (Table 4). No significant dose-response trends were observed across frequency of vehicle use or number of animals raised. We observed a nonsignificant inverse trend for number of livestock and myocardial infarction incidence and a similar trend for frequency of truck use and myocardial infarction mortality. Farm size was associated with 40% reduced mortality from myocardial infarction among individuals farming 500 or more acres (≥202.34 hectares); this was not observed for myocardial infarction incidence.
In general, farmers have lower rates of cardiovascular disease than the general population. Among farmers, the differences in risk of myocardial infarction based on agricultural exposures have not been well characterized. We observed no compelling evidence of an association between pesticide use and either fatal or nonfatal myocardial infarction when controlling for myocardial infarction risk factors. Ever use of 6 pesticides was positively associated with either myocardial infarction mortality or incidence, but not both. In addition, the 5 positive pesticides were not in the same functional class, there were no significant dose-response trends, and similar numbers of inverse associations and positive associations were observed.
Prior toxicologic research and epidemiology studies, as well as case reports, have reported cardiac consequences of acute exposure to cholinesterase inhibitors, including organophosphate and carbamate herbicides (2). Although these acute exposures are the most well documented and understood pesticide-related precursors to cardiac disease, there has been some suggestion of associations with chronic exposure to herbicides or persistent organic pollutants, including organochlorines. An ecologic study in Montana, Minnesota, and North and South Dakota found that those living in counties with high wheat production, used as a surrogate for herbicide exposure, were more likely to die from acute myocardial infarction than those living in counties of low wheat production (12). In New York, those living in areas contaminated with persistent organic pollutants, including organochlorine pesticides, had a 20% higher rate of hospital discharges for myocardial infarction than those living in uncontaminated zip codes (13). In addition to analyses by individual chemical, we also assessed lifetime days of application of any pesticide in the event that a general mechanism exists in which frequency of any pesticide exposure would increase myocardial infarction risk. However, we observed no evidence of an increased risk of either mortality or nonfatal incidence as the number of application days increased.
A study in rats suggested that myocardial irregularities can occur for up to 6 months after an acute high-level exposure to a cholinesterase inhibitor (14). We observed no evidence of an increased risk of future myocardial infarction for those who reported pesticide poisoning or a high-pesticide-exposure event prior to enrollment. One limitation of this analysis is that the Agricultural Health Study design does not permit the detection of acute sequelae following pesticide exposure, so myocardial infarctions immediately after these high-exposure events would have occurred before the start of follow-up. We were able to look at use of individual pesticides within the last year before enrollment compared with never and past use (results not shown) and saw no evidence of an association between recent pesticide use and future myocardial infarction mortality or nonfatal incidence. The literature suggests that cardiac outcomes occur within hours after acute high-level-exposure events, and we cannot assess that period in the Agricultural Health Study cohort.
The association between particulate matter exposure and myocardial infarction has been well characterized and is hypothesized to occur through an inflammation pathway leading to increased plasma fibrinogen (15). Farmers are exposed to several types of particulate matter, including diesel exhaust and high levels of dust, containing endotoxin and mold. A Swedish study found an elevated risk of ischemic heart disease mortality among those who experienced occupational exposures to diesel exhaust (16). Occupational dust and livestock exposure have been linked with ischemic heart disease (8, 17). We observed no positive associations between agricultural exposures and myocardial infarction mortality or incidence. The 2 inverse trends that we observed for truck use and mortality and number of livestock and myocardial infarction incidence are not consistent across the 2 myocardial infarction outcome measures. These trends are also not observed for similar exposures. For example, tractor use would expose participants to many of the same hazards as truck use, but the inverse trend with myocardial infarction mortality is not observed for tractor use.
The inverse associations observed for the ever use of some individual pesticides, as well as for some nonpesticide exposures, could be a result of a healthy worker effect within the cohort. Those who are healthy enough to be able to personally apply pesticides and perform other farming tasks may be less likely to suffer from myocardial infarction over the course of a relatively short follow-up. In addition, confounding by age was assessed in 10-year categories, which could lead to residual confounding. To address this possibility, we analyzed the ever use of individual chemicals using age as the time dimension of the life table in the proportional hazards models and saw no substantial differences in the estimates.
Participants in the Agricultural Health Study have a cardiovascular disease standardized mortality ratio of 0.5 (95% CI: 0.5, 0.6) when compared with the general population of North Carolina and Iowa (18). Because farmers generally have lower rates of cardiovascular disease compared with that of the general population, any cardiovascular disease risk factor associated with farming would appear protective if an external control group were used. The Agricultural Health Study has the strength of allowing comparisons of exposed farmers and unexposed farmers, as opposed to comparing farmers with nonfarmers.
We relied on self-reported exposure measures, myocardial infarction risk factors, and nonfatal myocardial infarction history based on questionnaire responses. Agricultural Health Study validation studies have shown that pesticide applicators give plausible responses to questions of pesticide use on the basis of pesticide registration information and have 70%–90% repeatability when asked about ever use of individual chemicals a year after the initial questionnaire (19, 20). Use of self-reported data on myocardial infarction might result in misclassification of disease status. The Minnesota Heart Survey and the Nurses’ Health Study found that 60% and 68%, respectively, of self-reported myocardial infarction could be validated by medical records (21, 22). Often this misclassification is a result of confusing other cardiac conditions for myocardial infarction. To account for this possibility, we also carried out analyses with self-reported angina as the outcome measure and with a combined cardiac outcome measure that counted a positive response to any of the myocardial infarction, arrhythmia, and angina questions as a case, and saw no difference in the results. There is nothing to suggest that the misclassification is differential with respect to pesticide use or agricultural exposures. In addition, the crude hazard ratios of myocardial infarction and self-reported myocardial infarction risk factors in Table 1 show associations that are consistent with findings in prior literature. This suggests that misclassification due to self-report of the outcome and risk factors does not substantially bias the estimates.
In conclusion, the present analyses yield little evidence of an association between chronic exposure to individual pesticides and myocardial mortality or nonfatal myocardial infarction incidence. We also observed no evidence of an association between chronic diesel and gasoline exhaust exposure or livestock farming and either myocardial infarction outcome. Two primary limitations of these analyses are the inability to detect acute sequelae of pesticide exposure and the healthy worker effect, which might limit our ability to see cardiac effects among healthy farmers.
Author affiliations: Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina (Katherine T. Mills); Occupational and Environmental Epidemiology Branch, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Rockville, Maryland (Aaron Blair, Laura E. Beane Freeman); and Epidemiology Branch, National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, North Carolina (Dale P. Sandler, Jane A. Hoppin).
This work was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Environmental Health Sciences (Z01-ES049030) and National Cancer Institute (Z01-CP010119).
The authors greatly appreciate the hard work of Stuart Long for data analysis, as well as the Agricultural Health Study Iowa and North Carolina field stations and coordinating center.
Conflict of interest: none declared.