|Home | About | Journals | Submit | Contact Us | Français|
Organophosphorous insecticides (OPs) are the most commonly used insecticides in US agriculture, but little information is available regarding specific OP use by individual farmers. We describe OP use for licensed private pesticide applicators from Iowa and North Carolina in the Agricultural Health Study (AHS) using lifetime pesticide use data from 701 randomly selected male participants collected at three time periods. Of 27 OPs studied, 20 were used by >1%. Overall, 95% had ever applied at least one OP. The median number of different OPs used was 4 (maximum=13). Malathion was the most commonly used OP (74%) followed by chlorpyrifos (54%). OP use declined over time. At the first interview (1993–1997), 68% of participants had applied OPs in the past year; by the last interview (2005–2007), only 42% had. Similarly, median annual application days of OPs declined from 13.5 to 6 days. While OP use was common, the specific OPs used varied by state, time period, and individual. Much of the variability in OP use was associated with the choice of OP, rather than the frequency or duration of application. Information on farmers’ OP use enhances our ability to characterize and understand the potential health effects of multiple OP exposures.
In 2007, 93 million pounds of insecticides were used in the United States with 70% used in agriculture (USEPA, 2011). Over a third of these were organophosphorous insecticides (OPs), a class of insecticides introduced in the 1950s. While their use has declined over time, OPs are still among the most commonly used insecticides in the U.S. (USEPA, 2011). While national and regional data are available on pesticide application and sales (NASS, 2011; USEPA, 2011; Gianessi and Reigner, 2006), only limited information is available regarding use of specific pesticides by individuals, particularly how many pesticides of a given class an individual uses in a year or lifetime. Information on pesticide use patterns in the U.S. is limited to national surveys of pesticide applicators, state-specific data about use on specific commodity crops, and sales figures (NASS, 2011; USEPA, 2011; Gianessi and Reigner, 2006). These surveys provide information about the relative ranking of specific pesticides used in the U.S., either by pounds of active ingredients applied or dollars spent on specific pesticides (USEPA, 2011). For commodity crops, information is also available regarding the pesticides used on each crop by state and calendar year (NASS, 2011; Gianessi and Reigner, 2006). Since 1990, CA has collected information on agricultural use of pesticides, including the amount of active ingredient and the location where the pesticide was applied (CDPR, 2011). All these sources are helpful in identifying the key pesticides used, but none provides information regarding the frequency of use for an individual applying these pesticides. Of particular interest for health research is understanding how pesticides in a given class are used by individuals.
OPs represent a large class of insecticides. They are often evaluated together based on their ability to inhibit acetylcholinesterase (AChE) (USEPA, 2006). This mechanism may not be related to all health outcomes (Hodgson and Rose, 2006; Mena, Ortega and Estrela, 2009; Proskocil, et al., 2008), nor do all OPs have the same capacity to inhibit AChE (USEPA, 2002). The biomarkers most frequently used to assess exposure to OPs (e.g., dialkylphosphates or AChE inhibition) are non-specific and thus have limited ability to assign exposure to a specific chemical (Barr, et al., 2004). As OPs differ in their toxicities and potential to affect human health, it is critical to understand how use of specific OPs differs among individuals. To characterize OP usage by US farmers, we used data from a random sample of private pesticide applicators in the Agricultural Health Study (AHS), a prospective cohort study of licensed pesticide applicators in Iowa (IA) and North Carolina (NC) (Alavanja, et al., 1996). Using data collected at three time points during 1993–2007, we describe among a well characterized group of farmers: 1) the frequency of use of specific OPs, 2) the number of different OPs used, 3) the number of days spent applying OPs, and 4) temporal changes in use by this group.
Pesticide applicators enrolled between 1993–1997 (Phase 1) when they were receiving or renewing their pesticide licenses (Alavanja, et al., 1996). A total of 82% of eligible private applicators (mainly farmers) enrolled. In Phase 1, all pesticide applicators completed the AHS enrollment questionnaire and approximately 44% also completed the take-home questionnaire (Alavanja, et al., 1996). Approximately five and ten years after enrollment (Phase 2: 1999–2003; Phase 3: 2005–2010), all AHS participants were asked to complete telephone interviews to update their pesticide use history. This analysis focuses on a subset of applicators who completed their Phase 3 interview by 2007.
We assessed lifetime use of OPs among 701 AHS private pesticide applicators recruited for participation in a neurological testing study (Starks, et al. 2011). The OP exposure histories of these individuals were enhanced to include data not currently available in the main AHS data set, specifically frequency and duration of use for pesticides on the take-home questionnaire and resolution of unknown insecticides from the follow-up interviews (see below for more information on exposure assessment). Private pesticide applicators who completed all questionnaires at all three phases were eligible. Participants were selected based on their lifetime days of use of 10 OPs at Phase 1 (Starks, et al. 2011). This selection variable was the sum of the lifetime days of use of each of the 10 OPs for which detailed information was collected at enrollment. As the neurological study required traveling to a fixed site for testing, only applicators living within 150 miles of Iowa City, IA; Dubuque, IA; Greenville, NC; or Wilmington, NC were eligible (Starks, et al. 2011). Women were excluded from the sample, due to their low representation (<3%) among the AHS applicators. We created two strata within each state --a high exposure group and the remainder of the population -- based on lifetime OP days of use; and we oversampled those in the high exposure stratum in each state. In IA, the high exposure stratum included the top 20% of the distribution of lifetime days; in NC, the high exposure stratum included the top 34%. In NC, we selected a larger fraction because the number of NC participants is smaller and they are more geographically dispersed. We randomly sampled equal numbers from all eligible individuals in each of the four state-by-exposure strata and identified empirical sampling weights (i.e., # subjects participating/# subjects eligible) in order to estimate pesticide use histories for the AHS participants (N=3863) who met the eligibility requirements for the neurological study. The overall response rate was approximately 39%.
Information on pesticide use was collected in the AHS during three time periods: Phase 1 (1993–1997), Phase 2 (1999–2003), and Phase 3 (2005–2007 for this subset). Individuals were asked about their personal application of pesticides for crop, animal, and non-crop uses. Phase 1 consisted of two self-administered questionnaires, the enrollment and the take-home questionnaire. Both Phase 1 questionnaires collected information regarding frequency and duration of use of selected pesticides in a closed format, such that all individuals provided responses on pre-determined OPs. At Phase 1, participants provided detailed information on 10 specific OPs (Table 1) including ever use, days used per year, years used, and use in the past year. The take-home questionnaire also included a checklist that collected ever-use information on 12 additional OPs. The pesticide-related questions on the Phase 2 and Phase 3 telephone interviews had an open-ended format and obtained information on recent use of pesticides including specific product used and the number of days of use. The use of an open-ended format allowed reporting of all pesticides used, by either brand name or chemical name, thus permitting identification of all pesticides in a product. We linked this information to the pesticide active ingredients to identify all the OPs used (see online supplement). AHS questionnaires are available at www.aghealth.org\questionnaires.html.
The goal of this analysis was to characterize use of specific OPs with regard to duration and frequency of use. Figure 1 outlines the data collection strategy for each phase including how information on OPs was collected and how key variables were created. We provide more detailed information on data collection and pesticide variable creation in the online supplement. For each OP, we created variables for ever use (yes/no) and days of use (days) for each phase. Days of use for each phase was calculated as the product of the frequency of application (days/year) and the duration of application (years in that reporting period for that individual). From these phase-specific variables we created variables for lifetime ever use (yes/no) and total lifetime days of use (lifetime days), equal to the sum of all days of use in each phase. Results are presented for the 20 specific OPs used by more than 10 participants.
The class of OPs includes chemicals with differing chemical structures and these differences may influence toxicity. We categorized pesticides classified as organophosphorous insecticides (see http://www.alanwood.net/pesticides/index.html (Wood, 2011) (Table 1) into six mutually exclusive and exhaustive subgroups: organophosphate, organothiophosphate, phosphonate, phosphonothioate, phosphoramidate, and phosphoramidothioate. The organothiophosphates were further categorized into aliphatic, heterocyclic, and phenyl organothiophosphates. For each subgroup, as well as for total OPs, we calculated total lifetime days based on the sum of the lifetime days of all OPs within that subgroup.
We computed summary statistics for ever use and annual days of use for specific OPs, overall, by state, and by each time period. We used the empirical sample weights to estimate the results back to the source population, namely, the male AHS participants who met the eligibility criteria for the neurological study (completed all AHS interviews and lived within specific geographic regions of IA and NC). We used Spearman correlation coefficients to assess the correlation between ever-use of specific OPs. All analyses were conducted using SAS 9.1.3 and AHS data releases of P1REL071201, P2REL071202, P3REL090500, and REL090500 for the demographic file.
The 701 participants in this study appeared representative, except for design characteristics, of the source population (N=3863) who had completed all AHS interviews and lived within the geographic regions included in the study (Table 2). Study participants were slightly more likely than non-participants to be still applying pesticides at Phase 3 (78% vs. 69%), Most other pesticide use and demographic characteristics of the sample were similar to all male participants who completed the Phase 3 interview as well as those who completed all AHS interviews (Table 2). By design, the study included equal numbers of participants from IA and NC, whereas two-thirds of the AHS as a whole is from IA. Applicators in the highest quartile of lifetime OP use at Phase 1 were overrepresented by design, consistent with the higher proportion of respondents reporting over 400 days of pesticide application in their lifetime in the sample compared both to their regional cohort (38% vs. 30%) and to individuals who completed the most recent interview (27%). While the results have been weighted to reflect the source population, the data as presented in Table 2 appear to be a good representation of all the male applicators who completed the last interview as well (N=19299). Participants in the eligible region represent 36% of those who had completed all AHS interviews, 17% of those who completed the Phase 3 interview (Figure 2A) and 8% of male applicators who enrolled in the AHS (Figure 2B).
Over 95% of the population represented by this sample of AHS participants had ever used OPs. The median number of OPs used in a lifetime was four and the maximum was 13 (Table 3). The median for IA farmers was four OPs compared to three for NC. Among the participants, 27 different OPs were reported (Table 1). Of the six subgroups, the majority of the OPs were organothiophosphates (n=18) or organophosphates (n=4). Among the organothiophosphates, seven were aliphatic; eight were heterocyclic, and three were phenyl organothiophosphates. Three additional chemicals are classified as phosphonothioate (fonofos) or phosphoramidothioate (acephate and methamidophos) insecticides. The proportion of applicators who used different subgroups of OPs ranged from 20% for organophosphates to 94% for organothiophosphates. Use of OP subgroups was similar in IA and NC, with the exception of organophosphates which had a smaller proportion of users in NC (6%) compared to IA (32%).
Twenty OPs had been used at least once by more than 10 individuals in our sample (Supplementary Table 1). Malathion was the most commonly used OP, with similar prevalence of use in IA and NC (Figure 3). Chlorpyrifos was the second most commonly used OP in NC. After malathion and chlorpyrifos, the relative ranking of chemicals varied between the states. A number of OPs were used only by NC farmers, most notably acephate, which was reported by 46% of NC participants. Acephate was the third most commonly used OP among NC farmers and 7th overall among the cohort. Tebupirimifos was reported by 18% of IA farmers and no NC farmers. All uses of tebupirimifos were associated with the pesticide product Aztec which also contains cyfluthrin, a pyrethroid insecticide. For most of the 20 OPs, pairwise correlations for ever use were low (data not shown). Only 17 of the 190 possible pairs had correlation coefficients greater than 0.2. The highest correlations were between pairs of OPs used more often in NC, acephate and ethoprop (r=0.58), disulfoton and ethoprop (r=0.57), and acephate and disulfoton (r=0.56).
While almost all applicators reported applying OPs sometime during their lifetimes, only 68% reported applying OPs in the year before AHS enrollment (Table 4). This estimate includes use of the OPs reported only on the take-home questionnaire checklist (acephate, azinphos-methyl, dimethoate, disulfoton, phosmet, and tetrachlorvinphos). If we excluded the OPs from the take-home checklist where use in the past year was inferred, we estimated that 50% of the cohort applied OPs in the year prior to enrollment. In Phases 2 and 3, approximately five and ten years after enrollment, 45% and 42% of applicators respectively, reported applying at least one OP. In Phases 2 and 3, IA applicators were slightly more likely to apply OPs than NC farmers (47% vs. 43% Phase 2, 47 vs. 37% Phase 3); however in every phase, NC farmers applied OPs more days per year (medianPhase1=14, medianPhase2=5, medianPhase3=10) than IA farmers (medianPhase1=11, medianPhase2=4, medianPhase3=5). For all phases, the annual frequency of application of any OP ranged from <5 days to more than 150 days, with the majority of users applying OPs fewer than 20 days annually (Supplementary Figure 1). In Phase 1, the median frequency of use of individual OPs was between 2.5 and 7 days, corresponding to the two adjacent questionnaire categories of < 5 days and 5–9 days (Table 4).
Use of specific OPs, on average, changed over time. Note that our results represent population changes, and not changes in use by specific individuals. Changes over time were not the same for all OPs. For example, secular trends were different for two OPs that were both commonly used in Phase 1, chlorpyrifos and malathion. The proportion of applicators using chlorpyrifos declined from 20% in Phase 1 to 11% in Phase 3, while the proportion of applicators using malathion remained relatively constant over all three phases (18–19%). For terbufos, another OP commonly used at enrollment (17%), the proportion of users dropped to 5% by Phase 3. For fonofos, an OP which had its registration cancelled in 1999, use dropped from 5% in Phase 1 to essentially 0% by Phase 2. Conversely, tebupirimfos, a pesticide introduced in 1995, started at 5% prevalence in Phase 2 and increased to 9% in Phase 3.
Lifetime application days for all OPs was more variable than lifetime days for any individual OP (Table 5) because the numbers of OPs used differed among participants. For individual OPs, the minimum number of lifetime days applied ranged from 2.5 to 8.75 days, suggesting infrequent use or use in only one season. The median lifetime days for individual OPs ranged from 15 days for coumaphos to 53 days for acephate. The median lifetime days of all OPs (154 days) was similar to the highest 75th percentile among individual OPs (179 days for dichlorvos).
As seen in Figure 4, the lifetime days for individual OPs varied by state for some OPs, but not others. Of the 20 OPs, acephate, azinphos-methyl, and fenamiphos were used only in NC, while tebupirimfos was used only in IA. Of the remaining 16 OPs used in both states, median lifetime days of use were essentially the same (20 or 25 days) for seven OPs (dimethoate, disulfoton, ethoprop, fonofos, malathion, methyl parathion and tetrachlorvinphos), five had higher lifetime days in NC (diazinon, fenthion, phorate, phosmet, and ronnel), and four were used more among IA farmers (chlorpyrifos, coumaphos, dichlorvos, and terbufos). For some chemicals like malathion with similar median use in both states, differences became apparent at the 75th percentile (109 days for NC and 56 days for IA). While NC farmers had higher lifetime days for 8 of 20 individual OPs, the median lifetime days for all OPs combined was greater in IA (165 days) than NC (139 days), an observation attributable largely to NC farmers typically using fewer individual OPs than IA farmers.
While there is much interest in the combined effects of OPs with regard to health, little is known about patterns of specific OP use by farmers. Using data from the AHS, we found that while almost all farmers had applied OPs sometime during their farming career; at the median, an individual had applied four OPs in his lifetime and at most, 13 OPs. We observed regional differences, with some OPs being used exclusively in one state or another. The median number of days that users applied a specific OP was relatively constant across OPs. Thus, it is possible that the biggest determinant of a farmer’s total annual days of OP exposure is the number of OPs used.
We observed changes over time in the OPs used consistent with nationwide trends. Chlorpyrifos and malathion remain among the most commonly used insecticides in agriculture. In US national estimates based on 2006–2007 data, chlorpyrifos was the most commonly used OP based on pounds of active ingredient used (USEPA, 2011); similarly in CA in 2008, chlorpyrifos was the most commonly used insecticide based both on pounds applied and acres treated (CDPR, 2011). In the US, acephate was the second most commonly used OP in 2007; in CA, malathion was second based on pounds applied while dimethoate was second based on acres treated. In our Phase 3 data, based on use in 2005–2007, malathion was the most commonly used OP with 19% of the cohort applying it; chlorpyrifos was second with 11%. While we lack data on pounds applied or acres treated, our relative ranking based on numbers of users in the AHS is comparable to these other metrics. The differences among data sources probably represent both regional differences in agricultural practices and crops grown as well as data collection.
The Agricultural Health Study is a prospective cohort study of pesticide applicators that provides a unique opportunity to evaluate patterns of OP and other pesticide use over time. At enrollment, 82% of licensed pesticide applicators in NC and IA enrolled in the study; thus, our cohort is representative of private pesticide applicators in these two states. The response rate to the take-home questionnaire was lower, but previous analyses found few differences in demographic or farming characteristics between those who did or did not complete the take-home questionnaire (Tarone, et al., 1997). Participants in this study completed not only the Phase 1 take-home questionnaire, but also the two follow-up interviews (Phases 2 and 3) and lived within specific geographic regions. While our results are weighted to represent all participants who met the eligibility criteria, it is likely that our results are representative of most farmers who completed the most recent follow-up interview (Phase 3) given the similarity of our sample to this larger population with respect to age, farming and pesticide use characteristics. The sampling frame for the neurological testing substudy was a random stratified sample of the cohort residing in specific geographic areas (eastern IA and Eastern NC). Generalizability to the whole cohort is probably very good, with the potential exception of farmers in western NC. Agriculture in NC has more regional heterogeneity in major crops than in IA; consequently, we may have missed exposure patterns unique to Christmas tree or apple production in western NC. However, farmers in western NC make up a small part of the overall cohort (8%), so we believe that the exclusion of farmers from this region would have little influence on the overall estimates.
We analyzed data from a stratified random sample of the cohort to estimate use of 27 different OPs by licensed private applicators in IA and NC. This sample represents a subset of the AHS cohort with enhanced OP assessment. There are two primary differences between these data and the data currently available for the full cohort: 1) information on OPs from the take-home checklist was included and lifetime days of use for these OPs was assigned based on the individual’s use of other insecticides, and 2) unknown pesticides reported at Phases 2 and 3 were resolved to identify any additional OP use (detailed in online supplement). Thus, these data provide a more complete picture of lifetime OP use than is currently available in the AHS as a whole. The main limitation of the AHS for assessment of lifetime OP use is that many of the commonly used OPs were ascertained on the take-home questionnaire (e.g., malathion and diazinon); thus, complete lifetime information is available only for those 44% who completed this questionnaire. In addition to the 10 OPs for which detailed information was collected during Phase 1, the take-home questionnaire also included a checklist of additional pesticides including 12 OPs, some of which are OPs commonly used in NC (i.e., acephate, ethoprop, and disulfoton). While our sample represents a small portion of the AHS as a whole, by using a stratified random sampling method we were able to extrapolate the results of these enhanced OP data to a larger portion of the AHS.
One challenge to compiling lifetime use is that the data collection methods differed over the phases of the study. The Phase 1 questionnaire obtained defined responses about use of predetermined OPs while in Phases 2 and 3, open-ended questions were used to obtain information about all pesticides used. While the open-ended approach in Phases 2 and 3 provided the opportunity to include new OPs, we lost the ability to confirm true non-users (i.e., individuals who did not use that chemical). By having people provide information on all chemicals they were currently using, we captured new OPs (e.g., tebupirimfos).
A major strength of this analysis is that these data reflect personal use of specific OPs by the same group of individuals at three time points. Farmers have been shown to provide reliable and reproducible reports of their pesticide use (Blair and Zahm, 1993; Blair, et al., 2002; Hoppin, Yucel, Dosemeci and Sandler, 2002). National estimates rely on sales figures or proprietary estimates of use (USEPA, 2011) or state specific crop surveys (NASS, 2011). Our data reflect both temporal trends in pesticide use as well as changes in agricultural practices over time. These data, along with national and regional data, provide a better understanding of use of OPs by individuals in agricultural settings. This information is critical not only for health assessments and epidemiology, but also for risk assessment and exposure characterization purposes.
Supplementary Text: Pesticide Variable Assignment
Supplementary Table 1: Proportion of Agricultural Health Study Participants (1993–2007) who ever used specific organophosphorous insecticides during their lifetimes.
Supplementary Figure 1: Annual number of OP application days for 3 time periods for private applicators in the AHS
Footnote: Percentages weighted to reflect those individuals who completed all AHS interviews and lived within the study region (N=3863)
The authors thank Dr. Joe Coble for his efforts to create the AHS pesticide use data base. This work was supported by the intramural research program of the National Institutes of Health, the National Institute of Environmental Health Sciences (Z01-ES049030) and National Cancer Institute (Z01-CP010119) and the National Institute for Occupational Safety and Health. The United States Environmental Protection Agency through its Office of Research and Development collaborated in the research described here; it has been subjected to Agency review and approved for publication.
Disclaimers: The findings and conclusions in this report are those of the authors and do not necessarily reflect the views of the National Institute for Occupational Safety and Health and Health Canada.
Conflict of Interest: The authors declare no conflict of interest.
Supplementary information is available at the Journal of Exposure Science and Environmental Epidemiology’s website.