This article is guided by a model (Figure 1) that contrasts the proximal and the distal determinants of pesticide exposure. Those determinants that are proximal to pesticide exposure—that is, the immediate determinants of exposure—are generally behaviors practiced either by farmworkers in the workplace or by farmworkers or their co-resident household members at home. These determinants include (in the workplace) use of personal protective equipment (PPE) and field sanitation, as well as (at home) laundry practices and child activity patterns. These proximal factors are themselves determined by predictors that are considered more distal to the exposure. These predictors include environmental conditions at work (e.g., safety training), at home (e.g., number of farmworkers in residence), and in the larger community (e.g., total farmland treated with pesticides). These environmental factors affect exposure through behavior; the association of environmental and behavioral factors is moderated by psychosocial factors, including the attitudes, values, beliefs, and knowledge held by farmworkers. For example, farmworker residences with a high residential density might be expected to store soiled work clothing that would present an exposure risk to household residents. This relationship could be positively influenced by beliefs that pesticides are harmless, or negatively influenced by knowledge of recommended laundry practices.

A portion of pesticides to which an individual is exposed is absorbed as the pesticide dose, and this dose can have health effects. According to the model, the amount absorbed is moderated by some of the workplace and household behaviors (e.g., hand washing by workers or household residents) as well as by other factors. The latter moderators include genetic factors, body size, and developmental status; these characteristics are not covered in this review.

This review focuses on the conceptual model (Figure 1) developed by the authors. Components of the model were expanded to produce a list of factors potentially related to pesticide exposure in farmworkers. These factors formed the search terms for our review of the literature that searched the PubMed, (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?DB=pubmed); Science Citation Index and Social Science Citation Index (http://portal.isiknowledge.com/portal.cgi/wos?Init=Yes&SI D=D112jMPBmi56JK4eA1); PsycINFO (http://www.psycinfo.org/psychoinfo/); and AGRICOLA (http://agricola.nal.usda.gov/) databases. We restricted reviews to peer-reviewed publications from studies that involved primary data collection and that were published in 1990 or later. A few earlier studies were included for topics with little research coverage. Articles were graded by the type of evidence for the association of a particular risk factor with pesticide exposure, as follows: 1 = association demonstrated in farmworkers; 2 = association demonstrated in nonfarmworker sample; 3 = plausible association proposed for farmworkers; or 4 = association plausible but not published for farmworkers. To be classified as “1,” the study participants had to be described as migrant or seasonal farmworkers. In most other cases the study participants were described as “growers,” “farmers,” or members of their families, and they were classified as nonfarmworkers. Study participants described as “applicators” were classified as nonfarmworkers. Summaries of articles were compiled by level of evidence and presented in tabular form. Because of space restrictions, only those articles graded “1” or “2” are presented here (Table 1). A minimum list of data to be collected in farmworker pesticide studies was derived from these evidence tables (Table 2).

Wearing PPE is one of the behaviors most widely assumed to protect workers from pesticide exposure. The label PPE can apply to everything from long-sleeve shirts to protective coveralls and respirators. Studies in the United States and abroad show that wearing PPE appropriate to the task results in lower exposure to pesticides (Table 1). Although the studies vary with regard to the types of chemicals investigated, the PPE tested (gloves, overalls), and the types of exposure measured [cholinesterase activity, skin wipes, organochlorine pesticide (OCP) serum levels], they all indicate that PPE is effective in reducing worker exposure to pesticides (Fenske et al. 1990; Gomes et al. 1999; Hernández-Valero et al. 2001; Lander et al. 1991; Ohayo-Mitoko et al. 1999). Studies in farmers (Arbuckle et al. 2002) and applicators (Fenske et al. 2002a; Nigg et al. 1993) lend further support to the effectiveness of PPE, although they also indicate variations because of fabrics and clothing design. In general, fabric less capable of penetration and designs that cover the largest amount of skin provide the greatest protection from pesticide exposure for workers. Despite the indications of efficacy, studies (particularly of farmers and applicators) show that PPE is frequently not used (e.g., Perry et al. 2002).

Other worker behaviors have been suggested as ways to reduce pesticide exposure, and these alternatives are included as recommended practices in the U.S. Environmental Protection Agency Worker Protection Standard (WPS) training (U.S. Environmental Protection Agency 1992). These behaviors include washing hands in the field before eating and after mixing pesticides. The importance of such behavior is demonstrated by studies showing that pesticides can be transferred to the home via automobile (e.g., Curl et al. 2002; Thompson et al. 2003). Curwin et al. (2003) showed that farmworker hand levels of the OP acephate could be reduced 96% by handwashing.

Additional practices have been suggested to reduce exposure. These practices include wearing grower-provided uniforms and showering at the worksite before returning home. There have been no tests to determine if such workplace behaviors would reduce exposure of the farmworker or the farmworker family.

Farmworker children are sometimes taken to the fields either to work or because adequate child care is lacking (Cooper et al. 2001). Such practices are likely to be predictors of pesticide exposure. Hernández-Valero et al. (2003) investigated the possible pathways of OCP exposure among 36 migrant farmworker children whose home base was Baytown, Texas. One-third of the children had previously conducted farmwork, and the farmwork duration significantly increased their exposure levels. Mandel et al. (2005) found that children of Minnesota growers often helped apply chemicals and, therefore, had levels of pesticide exposure closer to those of the parent who applied chemicals than to the other parent.

The application of residential pesticides in the home and yard has been investigated as a source of pesticide exposure among farmworkers and nonfarmworkers (Table 1). The collection of wipe (Quandt et al. 2004) or vacuum samples (Bradman et al. 1997), which allow direct identification of the type of pesticide found, has been used to link pesticides applied to worker dwellings to those pesticides detected. However, not all studies have had positive results (McCauley et al. 2001). Urinary metabolites of OP pesticides have also supported the link between residential pesticide application and worker exposure (Arcury et al. 2005).

Similar results have been found in nonfarmworker populations. Yard and garden pesticides were found to be transferred into homes by residents and by dogs (Lewis et al. 2001, Morgan et al. 2001; Nishioka et al. 2001). Use of OP pesticides in gardens is associated with metabolite levels in children (Fenske 2002b; Lu et al. 2001).

Several household sanitation behaviors are associated with farmworker pesticide exposure. Bradman et al. (1997) found that more frequent mopping and vacuuming was associated with lower pesticide recoveries in dust wipes. Arcury et al. (2005) suggested that having a vacuum cleaner was associated with lower levels of urinary OP metabolites.

A number of studies have documented the high potential for personal exposure to pesticides caused by waiting for extended periods before showering after work, not changing clothes immediately after work, and failure to separate work from household laundry (Alavanja et al. 1999; Curwin et al. 2002; Goldman et al. 2004). However, with the exception of McCauley et al. (2003), there is little direct evidence to support this association.

The organization of work is a subfield of occupational health that is concerned with the way that work processes are structured and managed. Organization of work investigators attend to such factors as the nature of the employment relationship (e.g., permanent versus contingent labor), job design (e.g., complexity of tasks and level of worker control), interpersonal elements of jobs (e.g., worker–supervisor relations), as well as such things as work schedules, job security, and communication with an employing organization. Although it has not been explicitly used in farmworker research, evidence suggests that several aspects of the way farm work is organized contribute to pesticide exposure (Marquart et al. 2003).

Several interrelated processes underlying the nature of the employment relationship suggest that pesticide exposure is likely to be greater among farmworkers in seasonal (e.g., workers with H2A visas) or day labor relationships in contrast to those in more “permanent” positions. Farmworkers in employment relationships that are more permanent may receive more effective safety training and more consistent reinforcement of safety behaviors than seasonal farmworkers or day-laborers. Researchers contend that workers in nonstandard employment relationships, such as seasonal workers or day-laborers, may be given tasks that place them at greater risk of becoming exposed to pesticides compared to permanent workers (Quinlan et al. 2001). Moreover, farmworkers in seasonal and day-labor arrangements may be less likely to request safety equipment or to report potential hazards to owners/operators out of fear that it may jeopardize future opportunities for work (Aronsson 1999; Aronsson et al. 2002; Quinlan et al. 2001). Despite the plausibility of several of these linkages, differences in pesticide exposure among farmworkers in different types of employment relationships have not been studied explicitly.

Different aspects of job design, or the tasks performed on a job and how they are performed, have been linked to pesticide exposure (Table 1). Tasks that are not regulated by the WPS can result in elevated pesticide exposure (Coronado et al. 2004). A great number of tasks or duties that put individuals in contact with pesticides or pesticide residues, such as self-service and repair of application equipment among applicators and a greater number of field activities among workers, are associated with more exposure (Alavanja et al. 1999; Hernández-Valero et al. 2001). Environments that provide farmworkers with little control over how pesticides are applied (e.g., high-exposure application methods), when pesticides are applied (e.g., avoiding windy days), and frequency of application are all associated with increased pesticide exposure among farmworkers (Mage et al. 2000; Martin et al. 2002; Mekonnen and Agonafir 2002). Similarly, environments that provide little personal control over protective behaviors, such as absence of well-maintained PPE or inability to wash or change clothes during the workday, contribute to elevated pesticide exposure (Alavanja et al. 1999; Arcury et al. 2002; Austin et al. 2001; Mekonnen and Agonafir 2002; Parrott et al. 1999).

Although there have been no explicit comparison studies, it is likely that different crops are associated with different levels of pesticide exposure because of the differences in tasks associated with crops. For example, some will involve greater hand labor for cultivation and harvest than others. It is likely that those requiring more hand labor will result in greater exposure.

Interpersonal elements of farm work also contribute to pesticide exposure. Better-quality relationships between workers and farmers/growers are important for identifying potential sources of pesticide exposure as well as for designing and implementing effective strategies for minimizing exposure (Grieshop et al. 1996). Communication difficulties caused by language differences between workers and farmers/growers contribute to greater pesticide exposure through less effective training (McCauley et al. 2002; Rao et al. 2004). Similarly, differences in belief systems about the risks of pesticide exposure and appropriate behaviors for minimizing risk can contribute to elevated exposure by undermining the effectiveness of training and safety programs (Arcury et al. 2001; Quandt et al. 1998; Rao et al. 2004). The psychological demands of the work environment can also contribute to lower adherence to safety regulations (Kidd et al. 1996; Thu 1998; Walter et al. 2002). Despite the strong suggested connection of these work environmental factors to pesticides, no studies have examined pesticide exposure and the organization of work, either in farmworkers or in other populations.

One of the major aspects of the work environment directly related to pesticide exposure is safety training for workers. Minimum content and standards for pesticide safety training are part of the WPS, which mandates training for field workers as well as for applicators. A number of studies have examined safety training in farmworkers, but none of these have examined the association of safety training with pesticide exposure. This work shows that many farmworkers fail to receive training as mandated (Arcury et al. 1999; Elmore and Arcury 2001; U.S. General Accounting Office 2000) but that the rates vary over time (Arcury et al. 2001). Salazar et al. (2004) found that even when safety training is presented, it is sometimes understood poorly because of language barriers. Research with applicators (Martinez et al. 2004) and farmers (Perry and Layde 2003) shows that safety training produces increased knowledge, but it does not necessarily result in appropriate safety behaviors.

Proximity of dwellings to agricultural fields treated with pesticides has been suggested as a dwelling characteristic associated with exposure (Fenske et al. 2000). Studies of dust samples from farmworker residences support this suggestion, both in terms of concentrations of pesticides (McCauley et al. 2001) and in numbers of pesticides found in the home (Quandt et al. 2002, 2004). Curl et al. (2003) found no association between distance to field and levels of metabolites found in children’s urine. However, these metabolite levels were associated with house dust concentrations, which, in turn, were associated with the dust in cars of farmworkers, thereby indicating a pathway from worksite to home. Among nonfarmworkers, distance from dwelling to fields was associated with concentrations in house dust (Fenske et al. 2002b; Lu et al. 2000). This linkage was reflected in higher urine concentrations of metabolites in some (Loewenherz et al. 1997) but not all (Fenske et al. 2002b) studies measuring urinary metabolites.

Various housing quality indicators have been linked to greater pesticide exposure for farmworker families. Older dwelling age (Bradman et al. 1997) and renting rather than owning (Arcury et al. 2005) have been examined. These studies were based on the belief that the greater age of a house as well as a history of different tenants might lead to the accumulation of larger amounts of pesticides, both simply as a matter of time and because there might be more opportunity for pest infestations to which pesticides are applied. Both of these measures have been linked to exposure. Quandt et al. (2004) used an interviewer’s judgment of how difficult or easy a house was to clean, reasoning that houses more difficult to clean would have a less thorough elimination of pesticides. Cleaning difficulty was associated with greater pesticide exposure.

Several aspects of the household social environment related to household composition have been suggested as major influences on pesticide exposure at home. The logic is that more persons in the household, particularly more farmworkers, will increase the volume of take-home pesticides, and this situation might be most extreme in cases of crowding. The simplest measure, total household size, has been linked to pesticides in two studies of farmworkers (Arcury et al. 2005; McCauley et al. 2001). These findings are supported by the study of Goldman et al. (2004) of pesticide-related behaviors. They found that larger household size was associated with fewer in-home safety behaviors. McCauley et al. (2003), in a study of nonfarmworker agricultural households, found weak and nonsignificant associations between household size and OP residues. More specific measures of household social environment (number of adults and number of agricultural workers in the household) have been suggested. However, this association generally has been tested by comparing agricultural and nonagricultural households (Bradman et al. 1997; Lu et al. 2000; Simcox et al. 1995), not by looking at the variation in number of adults within farmworker homes. Exceptions are the work of Arcury et al. (2005) and Quandt et al. (2004), which compared nuclear family households with those that comprised other adult relatives or nonrelatives and appeared to find more pesticides in the latter. This finding may be caused by greater track-in of pesticides with more adults, or by culture-specific issues. The investigators found that women residing in farmworker homes reported difficulty in enforcing standards of household cleanliness when male in-laws lived with the family because gender roles limit the authority of women over the behavior of fathers-in-law and other relatives. Only two studies have used density or crowding (e.g., persons/room and persons/square foot) as measures of the household social environment. McCauley et al. (2001) found no association in homes of farmworkers, and only a slight association in homes of other agricultural workers (McCauley et al. 2003).

Several different measures have been used to associate overall use of pesticides in a community with exposure. None has focused specifically on farmworkers. Fenske et al. (2000) found that a majority of children in an agricultural region from both agricultural and nonagricultural families had urinary metabolites for OPs. Similar results were reported by Koch et al. (2002), who found no differences because of parental occupation or residential proximity to fields. Lee et al. (2002) measured airborne agricultural pesticides at monitoring stations in California communities. They found that the level of exposure exceeded reference values for noncancer health effects for half of the population.

In agricultural communities, historical use of some persistent pesticides may have led to long-term contamination of the soil. In areas where lead arsenate was used extensively, soil samples have demonstrated the persistence of arsenic (Wolz et al. 2003). DDT, an OCP, is still found in soil samples despite its having been removed from use decades ago (Miersma et al. 2003).

While many diverse factors have been proposed to have direct, indirect, or modifying effects on whether or not farmworkers are exposed to pesticides (Table 1; Figure 1), the research connecting characteristics of workers’ environments and behaviors with actual measures of pesticide exposure is meager. Behavioral factors for which the best evidence of a direct relationship with pesticide exposure exists are use of PPE, use of pesticide products in and around the home, and personal hygiene behaviors such as hand washing at work and showering upon returning home from work.

Evidence of environmental factors associated with exposure is lacking for the occupational setting. Aside from clear evidence that job tasks that bring workers into contact with pesticides produce greater exposure, there has been little attempt actually to measure the effect of workplace safety training or the organization of work on exposure. Far more attention has been paid to the effects of the household environment of farmworkers and applicators on the exposure of workers and family members because we have better access to homes than to work sites. With some exceptions, research supports the link between proximity to fields and exposure. While studies use different measures, older houses of poorer quality appear to be linked to exposure. Similarly, different measures of household composition have been used. Most suggest that a greater number of adults and farmworkers in a house leads to greater amounts of pesticide in the dwelling and more pesticide exposure of the residents.

None of the psychosocial or cultural factors proposed as moderators in the association of environment or behavior with exposure has been examined with actual pesticide exposure data. Thus, the role of such factors in farmworker exposure is unknown.

The review of the evidence also highlights the fact that many of the existing studies that identify predictors of pesticide exposure in farmworkers, as well as in nonfarmworkers, have relied on self-reported behaviors rather than on true exposure measures. Among those studies that have included measures of exposure, some have employed environmental samples rather than biological measures. This history suggests that further studies of the association between predictors of exposure and actual biomarkers are warranted.

The evidence provided by this review, encompassing both factors with demonstrable links to exposure and those plausible but not well studied, indicates that a minimum set of concepts should be included in studies of farmworker pesticide exposure. The exact measures for each concept are not entirely clear because of the dearth of research that has actually sought to measure the association of predictors and exposure outcomes. Therefore, the recommendation is to obtain a broad enough group of measures to test for likely pathways of exposure.

This minimum set differs depending on whether the research focus is limited to occupational pesticide exposure of workers or if the focus includes the paraoccupational and environmental pesticide exposure of adults and children who reside with farmworkers. For the latter, some additional measures are included (e.g., child play areas). Measures are presented from proximal to distal determinants (Table 2). Although this review has included a variety of moderators that are likely to be important in the exposure pathway, there is currently insufficient research to recommend any particular set of such measures.

This review suggests that a productive line of research would be to focus on the role of the organization of work with regard to pesticide exposure. This area of research can help identify aspects of the workplace that can be modified to protect workers from pesticide exposure. It is consistent with the approach of much of occupational safety and health, in that it relies less on changing human behavior directly than on “engineering” changes in work and the workplace environment. While the organization of work is a well developed area of research, it has not had widespread application to farmworker pesticide safety research.

The most obvious dearth of data found in this review is in the area of cultural and psychosocial factors that may moderate the effect of household and workplace environments on safety behaviors. Although such factors are clearly not direct influences on exposure, they condition the extent to which behavior or environmental change to protect workers and their families will be accepted, and they are, therefore, necessary components of behavioral interventions. It is premature to list specific data to be collected because such factors do not lend themselves to measurement through simple questions.