|Home | About | Journals | Submit | Contact Us | Français|
To identify potential risk factors for serum cholinesterase (BuChE) inhibition among agricultural pesticide handlers exposed to organophosphate (OP) and N-methyl-carbamate (CB) insecticides.
We conducted a longitudinal study among 154 agricultural pesticide handlers who participated in the Washington State cholinesterase monitoring program in 2006 and 2007. BuChE inhibition was analyzed in relation to reported exposures before and after adjustment for potential confounders using linear regression. Odds ratios estimating the risk of ‘BuChE depression’ (>20% from baseline) were also calculated for selected exposures based on unconditional logistic regression analyses.
An overall decrease in mean BuChE activity was observed among study participants at the time of follow-up testing during the OP/CB spray season relative to pre-season baseline levels (mean decrease of 5.6%, P < 0.001). Score for estimated cumulative exposure to OP/CB insecticides in the past 30 days was a significant predictor of BuChE inhibition (β = −1.74, P < 0.001). Several specific work practices and workplace conditions were associated with greater BuChE inhibition, including mixing/loading pesticides and cleaning spray equipment. Factors that were protective against BuChE inhibition included full-face respirator use, wearing chemical-resistant boots, and storing personal protective equipment in a locker at work.
Despite existing regulations, agricultural pesticide handlers continue to be exposed to OP/CB insecticides at levels resulting in BuChE inhibition. These findings suggest that modifying certain work practices could potentially reduce BuChE inhibition. Replication from other studies will be valuable.
Organophosphate (OP) and N-methyl-carbamate (CB) insecticides are widely used in agriculture. In Washington State, approximately 589,000 lbs of azinphos-methyl, chlorpyrifos, and carbaryl (three common OP/CB insecticides) were applied in apple orchards in 2007. Other crops grown in Washington State are also frequently treated with OP/CBs including pears, cherries, grapes, and potatoes.
Acute effects of OP/CB exposure have been well documented; inhibition of neuronal acetylcholinesterase (AChE) enzyme activity is the main mechanism of OP/CB toxicity. AChE hydrolyzes the neurotransmitter acetylcholine, and thereby plays a critical role in regulating nerve transmissions in the central and peripheral nervous systems. Cholinesterases (ChE) are found in blood in two different forms; AChE is associated with red blood cell membranes, and butyrylcholinesterase (BuChE) is present in serum. Both AChE and BuChE inhibition are considered to be markers of early biologic effects related to OP/CB exposure. Generally, AChE inhibition is considered to be a better marker of toxicity, whereas BuChE inhibition is a more sensitive marker of exposure because it is inhibited more effectively than AChE by most OP/CBs including chlorpyrifos, diazinon, and malathion. BuChE measurements have been used successfully as endpoints in several previous studies of OP-exposed individuals.[6–8]
Among agricultural workers in the U.S., OP/CBs continue to be responsible for a high proportion of pesticide poisonings, likely due to their high acute toxicity and widespread use in agriculture. In an analysis of acute pesticide poisonings among U.S. agricultural workers from 1998–2005, Calvert et al. found that OP/CBs were implicated more frequently than any other class of pesticides. There is also growing concern about a variety of health endpoints that may be associated with chronic exposure to OP/CB insecticides, including chronic neurologic effects[10, 11] and various cancers.
Agricultural pesticide handlers are workers who are involved in the pesticide application process, which includes applying pesticides and related activities, such as mixing and loading pesticides into spray tanks and repairing application equipment. Handlers are generally considered to have higher levels of pesticide exposure than agricultural workers engaged in other tasks. However, relatively few studies have evaluated specific pesticide handling practices and conditions in relation to biological markers of exposure. Agricultural pesticide handlers may be exposed to OP/CBs as a result of dermal contact with pesticides or spray equipment, inhalation, accidental spills or spray equipment malfunction, inadequate use of personal protective equipment (PPE),[16, 17] and lack of decontamination facilities.
In 2004, the Washington State Department of Labor and Industries initiated a monitoring program among agricultural workers who handle OP/CB insecticides. Workers who participate in this monitoring program are tested for AChE and BuChE activity at an annual baseline (i.e., before the OP/CB spray season), and follow-up tests are conducted throughout the spray season to evaluate ChE inhibition relative to baseline levels. Follow-up tests are only required when workers have handled OP/CBs for 30+ hours in a 30-day period. Generally, most handlers who return for follow-up testing have only one follow-up test each year, though some have multiple follow-up tests during the same spray season. If a worker experiences >20% AChE or BuChE inhibition at follow-up relative to annual baseline levels, the employer must conduct a work practice investigation to determine possible sources of exposure. For ≥30% AChE inhibition or ≥40% BuChE inhibition, the worker is removed from handling activities (with wage protection) until his or her ChE activity returns to within 20% of baseline.
We recruited participants from the statewide ChE monitoring program for a study to identify workplace and behavioral factors associated with BuChE inhibition. This study addresses the need for further epidemiologic research characterizing relations between pesticide use practices and biological markers of exposure, as suggested by Acquavella et al. and Quandt et al. Relatively few studies have evaluated pesticide-related effects among agricultural pesticide handlers due to logistic challenges in accessing and following farmworker populations over time.[10, 22] By recruiting participants from the statewide ChE monitoring program, we were able to investigate potential exposures and their relationship with BuChE inhibition among agricultural workers who handle OP/CB insecticides.
We conducted a longitudinal study among agricultural pesticide handlers in Washington State during the OP/CB spray season (April–July) in 2006 and 2007. To recruit participants, we collaborated with two clinics that conducted ChE monitoring in eastern Washington State. Participants were recruited at the clinic or the worksite at the time of follow-up ChE testing. We used a computer-based survey instrument to collect information from participants. The survey was administered on tablet computers in either Spanish or English. All questions were displayed on the screen and audio-recorded, and icons or photos were used to represent possible responses for most questions.
The final survey consisted of 64 items. We collected information about: 1) OP/CB insecticides used and crops treated; 2) pesticide handling activities performed and spray equipment used; 3) duration and frequency of handling activities; 4) use, condition, and storage of PPE; 5) decontamination practices; 6) acute exposure events; and 7) pesticide safety training. We also collected information about symptoms that may be related to OP/CB exposure, non-occupational risk factors for BuChE inhibition, and demographic characteristics. Questions about potential sources of exposure and pesticide-related symptoms focused on the 30-day period prior to the interview and follow-up ChE test. We considered this 30-day period to be the most etiologically relevant in terms of risk of BuChE inhibition because BuChE activity levels recover naturally over time. Moreover, focusing on relatively recent exposures likely facilitated recall among study participants.
All study procedures were approved by the Institutional Review Board at the University of Washington.
Several algorithms were used to calculate scores for OP/CB toxicity, work activities, and PPE use. The toxicity score was estimated by assigning values to specific OP/CBs based on the relative potency factors used in the USEPA cumulative risk assessments for OP and CB insecticides.[23, 24] Relative potency factors were determined by the USEPA based on the degree of brain AChE inhibition in rat studies. As in the USEPA cumulative risk assessments, we assumed additive effects of exposures to multiple OP/CBs. It should be noted that one participant had an implausibly high toxicity score that was inconsistent with: 1) OP/CB insecticides registered for use on reported crops treated; and 2) OP/CB use by other participants from the same orchard. This record was therefore excluded from all analyses involving OP/CB toxicity score. Scores for work activities and PPE use were based on algorithms developed in the Agricultural Health Study; these algorithms have been validated in several studies in different U.S. regions with various chemicals and crops.[26, 27] Work activity scoring was modified slightly to include tower sprayers and cleaning activities, and PPE scoring was modified to reflect greater use of PPE among participants in this study relative to Agricultural Health Study participants. PPE score was expressed in terms of the estimated likelihood of exposure: handlers wearing full PPE received a score of zero (lowest possible score), and handlers wearing no PPE received a score of 14 (highest possible score).
For all analyses, the exposure score variables were transformed into z-scores (i.e., standardized based on the mean and standard deviation) to allow for meaningful comparisons between these variables. We also calculated a score for cumulative OP/CB exposure in the last 30 days by adding the z-score values for each of the individual exposure score variables:
A detailed description of the scoring system used to estimate values for OP/CB toxicity, work activities, and PPE use is provided in Appendix A.
We obtained participants’ BuChE test results from the participating clinics in the statewide monitoring program. Clinic staff collected and processed serum samples, which were shipped cold overnight for laboratory testing. BuChE assays were performed by the Washington State Public Health Laboratories in 2006 and by Pathology Associates Medical Laboratories in 2007. Both labs measured BuChE activity using the Ellman method with the ChE reagent kit from Roche Diagnostics. The Public Health Laboratories measured BuChE activity using an automated Dade Dimension AR system, and Pathology Associates Medical Laboratories used an Olympus AU5421/AU2700 system. Both labs had high precision for BuChE measurements; the coefficients of variation were 2.5% in 2006 and 2.6% in 2007. The main outcome in our study was BuChE inhibition, which was defined as the percent change in BuChE activity comparing levels at follow-up during the OP/CB spray season against pre-season baseline levels for each handler.
We did not evaluate AChE inhibition because assays performed in 2007 had low precision (16.7% coefficient of variation), and analyses of state monitoring program data found little overall evidence of AChE inhibition. In 2006, mean AChE inhibition among handlers with at least one follow-up test was 1.8%, and only two of the 472 handlers with follow-up tests had >20% AChE inhibition. AChE inhibition may be a more relevant outcome for handlers in developing countries where higher levels of OP/CB exposure are generally observed.[30, 31]
Records for 154 participants with complete surveys and both baseline and follow-up BuChE test results were included in this analysis. This represents 50.7% of the 304 pesticide handlers who were invited to participate in this study during the 2006 and 2007 spray seasons. Study participants were similar to all handlers in the statewide monitoring program in terms of age, ethnicity, and gender. Mean age was 33.6 years for study participants and 32.9 years for all handlers in the state program, and almost all handlers (>99%) in both this study and the state program were Latino males.
For handlers who participated in this study in either 2006 or 2007, we selected the first completed survey and corresponding BuChE data for this analysis. Some handlers participated in this study in both 2006 and 2007 (N=22). For those subjects, we chose only one record per subject to ensure independent results. The record with the larger value of BuChE inhibition was included to enhance the range of values for statistical testing. This would not introduce bias in the results because this choice was made without consideration of determinants of exposure.
We evaluated BuChE inhibition in relation to overall OP/CB exposure during the past 30 days based on the algorithms described above. Cumulative OP/CB exposure score was modeled as a continuous predictor, and percent change in BuChE activity from baseline level was modeled as a continuous outcome (i.e., degree of BuChE inhibition per 1-unit increase in OP/CB exposure score). In another model, we evaluated OP/CB toxicity score, work activity score, and PPE score as separate predictors of BuChE inhibition. Both models included year of participation, days since baseline ChE test, and age in years as covariates. Linear regression with robust standard error estimates was used for each of these analyses.
Specific exposure variables were selected for multivariate analysis based on a priori hypotheses and preliminary bivariate analyses. Several potential confounding factors were included in the statistical models, including: year of participation, days since baseline ChE test, age in years, toxicity score, work activity score, and PPE score. Percent change in BuChE activity from baseline levels (i.e., BuChE inhibition) was used as the main endpoint in these analyses. Due to wide inter-individual variability in BuChE activity, the relative change from baseline levels may be considered to be more biologically meaningful than the absolute level. Additionally, analyses were performed evaluating BuChE activity at follow-up (with baseline BuChE activity included as a covariate) and with log-transformed BuChE values. We also evaluated the risk of ‘BuChE depression’ (>20% inhibition from baseline levels) in relation to specific exposure variables using multiple logistic regression adjusting for year of participation, days since baseline ChE test, and age.
Differences were considered to be statistically significant if P values were < 0.05. Analyses were performed using Intercooled Stata 9.2 (StataCorp, College Station, TX).
All of the participants in this study were male, and all but one participant with reported ethnicity was Hispanic/Latino (Table 1). Almost all participants completed the survey in Spanish (97%). Most participants were younger than 35 years of age (61%), and approximately half had a primary school education or less. Many participants had limited experience handling pesticides; approximately half had been employed as handlers for three years or less. Over three-fourths of our sample had baseline ChE tests within 60 days prior to their follow-up ChE test; longer time since baseline testing was associated with greater BuChE inhibition (β = −0.145; P < 0.001).
Overall, mean BuChE activity at follow-up was significantly lower than BuChE activity at baseline (P<0.001) (Table 2). Mean BuChE inhibition was somewhat greater among handlers who participated in 2007 relative to participants in 2006; however, this difference was not statistically significant [mean (SD) of −4.8% (13.8) and −6.6% (8.6) for 2006 and 2007, respectively; P=0.34]. Approximately 12% of the study sample had >20% BuChE depression, which was consistent with the frequency of BuChE depression in the statewide ChE monitoring program in 2006 and 2007. More cases of BuChE depression were observed in 2006 than in 2007 (P=0.086, chi-square test).
We observed a significant trend toward greater BuChE inhibition with increasing cumulative OP/CB exposure score (Table 3). When analyzed as separate predictors, OP/CB toxicity score and PPE score were significantly associated with BuChE inhibition, and there was a borderline significant association between work activity score and BuChE inhibition. Results were similar after several records with high outlying values for toxicity score or PPE score were excluded (data not shown). There was little evidence of correlation between OP/CB toxicity score, work activity score, and PPE score in this analysis (correlation coefficients ranged from −0.06–0.12, P≥0.16).
Several particular work activities were associated with greater BuChE inhibition (Table 4) and risk of >20% BuChE depression (Table 5). On average, handlers who reported mixing/loading pesticides had 5.25% greater BuChE inhibition than handlers who did not mix/load pesticides after adjusting for covariates (P=0.007). In the adjusted logistic regression analysis, we found that mixer/loaders were approximately twice as likely to experience BuChE depression as other handlers. Handlers who reported cleaning spray equipment had an average of 4.4% greater BuChE inhibition than handlers who did not clean spray equipment (P=0.033), and we observed a nine-fold increased risk of BuChE depression among handlers who cleaned spray equipment. Some other work activities and exposures were moderately, although not significantly associated with BuChE inhibition, including repairing spray equipment, cleaning out pesticide containers, cleaning up after pesticide spills, and reported use of azinphos-methyl, carbaryl, or multiple OP/CBs in the last 30 days. There were no consistent associations of BuChE inhibition and methods of pesticide application, air blast or tower spraying.
Recency of exposure did not appear to be associated with the degree of BuChE inhibition. There was some suggestion of an association between length of spray sessions and BuChE inhibition, with handlers who reported 3–4 spray sessions of eight hours or more having on average 6.9% greater BuChE inhibition than participants who reported no eight hour spray sessions in the last 30 days. However, this association was not statistically significant, and we did not see a consistent trend in the relation between number of 8-hour spray sessions and BuChE inhibition.
Greater BuChE inhibition was observed with increasing age after adjustment for covariates (P=0.048). Self-reported health status was also associated with BuChE inhibition, with participants who reported “poor” or “fair” health having 6.4% greater BuChE inhibition on average relative to participants who reported “excellent” health status (P=0.02).
When analyses were repeated using log-transformed BuChE values, similar associations were observed for each of these exposures (results not shown).
Wearing a full-face respirator appeared to protect against BuChE inhibition. Relative to full-face respirator users, handlers who wore half-face respirators had approximately 7.0% greater BuChE inhibition on average (P=0.034). Half-face respirator users were almost seven times as likely as full-face respirator users to experience BuChE depression. Wearing chemical-resistant footwear was also protective against BuChE inhibition. Handlers who did not wear chemical-resistant footwear had an average of 11.4% greater BuChE inhibition (P=0.041), and an estimated 7.6-fold increased risk of BuChE depression. Relative to handlers who wore nitrile gloves alone, those who wore nitrile gloves with cloth gloves underneath had somewhat less BuChE inhibition, although this difference was not statistically significant (P=0.087). In terms of PPE storage, handlers who reported storing PPE in a locker at work had less BuChE inhibition than handlers who did not use lockers. On average, handlers who did not use lockers for PPE storage had 7.6% greater BuChE inhibition, and were 5.8 times as likely to experience BuChE depression as handlers who did use lockers.
Contrary to expectations, handlers who reported wearing chemical-resistant aprons had somewhat greater BuChE inhibition than handlers who did not wear chemical-resistant aprons (P=0.119). Also, we did not observe any association between hand washing practices before breaks during pesticide applications and BuChE inhibition.
Results were essentially unchanged when these analyses were repeated using log-transformed BuChE values, except that the association between chemical-resistant footwear use and less BuChE inhibition was only borderline significant (P=0.053).
This study identified several work activities that were associated with BuChE inhibition, and some PPE use practices that appeared to prevent BuChE inhibition. Results were generally consistent with the findings of other studies.[21, 33] Handlers who mix/load pesticides are generally considered to have relatively high exposures, and this activity may be particularly hazardous due to potential exposure to OP/CBs in their concentrated forms (i.e., before being diluted for application). We also found that handlers who cleaned spray equipment had significantly greater BuChE inhibition. Similarly, Arbuckle et al. found that washing spray equipment was associated with elevated urinary levels of the herbicide 2,4-D. Although 2,4-D is not a ChE-inhibiting pesticide, the exposure pathway is likely to be similar. There was also some suggestion in our study that handlers may have been exposed while cleaning out pesticide containers or cleaning up after pesticide spills, although these factors were not significantly associated with BuChE inhibition after adjustment.
Use of PPE has been shown to minimize pesticide exposures effectively.[16, 21, 34, 35] In our study, handlers who wore full-face respirators and chemical-resistant footwear had significantly lower levels of BuChE inhibition. Storing PPE in a locker was also protective against BuChE inhibition. Handlers who change into chemical-resistant boots for applications and store PPE in a locker at work may have less “take home” exposure. Although such PPE use and storage practices may afford greater protection, it is also possible that these variables could be surrogates for safer handling practices in general.
Other studies have shown that glove use is associated with lower levels of exposure while mixing or applying pesticides.[34, 35] We did not see any strong associations between glove use and BuChE inhibition in this study, but it should be noted that only one participant did not wear chemical-resistant gloves. As such, we could only evaluate differences in BuChE inhibition between handlers who wore chemical-resistant gloves alone (67%) and handlers who wore chemical-resistant gloves in combination with disposable gloves (18%) or cloth gloves (13%).
We found that handlers who wore chemical-resistant aprons had somewhat greater BuChE inhibition than other handlers; this association was not in the anticipated direction. Chemical-resistant aprons are generally worn by handlers while mixing/loading pesticides, which is an activity with an inherently higher risk of exposure. Although we attempted to control for handling activities in this analysis, there may have been residual confounding due to generally higher exposures among handlers who wore aprons relative to other handlers.
In terms of decontamination practices, Curwin et al. found that hand washing significantly reduced the concentration of acephate residues on the hands of tobacco harvesters. We did not see an association between hand washing practices and BuChE inhibition, which may be due to exposure misclassification. We collected self-reported information about “usual” hand washing practices, but were unable to ascertain the frequency and consistency of such practices after each application. Such misclassification might also explain the lack of associations between decontamination before breaks (i.e., potential contamination from using a cellular phone, eating, or urinating) and BuChE inhibition. More detailed observations of decontamination practices as potential sources of exposure are warranted.
In addition to potential sources of exposure, we evaluated BuChE inhibition in relation to exposure to specific OP/CBs. In the unadjusted analysis, chlorpyrifos users had less BuChE inhibition relative to handlers who were exposed to other OP/CBs, including compounds that were more acutely toxic (e.g., azinphos-methyl). However, the association between BuChE inhibition and chlorpyrifos use was not significant after adjustment for covariates, suggesting that confounding may have been present. In particular, days since baseline test may have been an important confounder in this analysis because chlorpyrifos is typically applied early in the spray season, whereas other OP/CBs (including azinphos-methyl) are usually applied later in the season when there is greater potential for cumulative BuChE inhibition over time. The association between use of multiple OP/CBs and BuChE inhibition is more plausible; recent studies suggest that mixed exposures can potentiate the toxic effects of specific OPs.
We did not observe any association between recency of exposure and degree of BuChE inhibition. However, it should be noted that approximately two-thirds of our study population had handled pesticides within the week preceding their follow-up ChE test, and only 8% of the sample was exposed >30 days previously. This pattern suggests that there may not have been enough heterogeneity in our sample to determine the association between recency of exposure and BuChE inhibition.
Relative to handlers who reported “excellent” health status, handlers who reported “poor” or “fair” health status had significantly greater BuChE inhibition. It is possible that handlers with poorer health were susceptible to BuChE inhibition, or their BuChE activity may recovered more slowly following OP/CB exposure. However, it is also possible that handlers with greater BuChE inhibition may have experienced symptoms of pesticide-related illness, and reported poorer health as a consequence of OP/CB exposure. Because self-reported health status was determined cross-sectionally at the time of follow-up ChE testing, we were unable to characterize the temporal relation between health status and BuChE inhibition. Future studies with prospective data collection may provide additional information about this association.
The implementation of a ChE monitoring program in Washington State provided a valuable opportunity to evaluate potential sources of exposure to OP/CB insecticides among agricultural pesticide handlers. Because participants were unaware of the results of their follow-up ChE tests at the time of the interview, and acute pesticide-related symptoms are relatively uncommon in this population, reporting or healthy worker survivor effect biases on risk estimates in this study were probably minimal. Results were not materially changed after excluding fifteen participants with self-reported symptoms of pesticide-related illness (data not shown). We also repeated the analyses after excluding eight participants with a previous BuChE depression (i.e., >20% BuChE inhibition on the prior follow-up visit). Again, results were essentially unchanged for most exposures, with the following exceptions: 1) cleaning pesticide containers was statistically significantly associated with greater BuChE inhibition (P=0.033); 2) the association between cleaning spray equipment and greater BuChE inhibition was only borderline significant (P=0.053); and 3) the association between wearing a full-face respirator and less BuChE inhibition was no longer significant (P=0.13).
Previous studies have noted the potential for exposure misclassification in self-reported data. In the present study, reliance on self-reported exposure information may have resulted in missing data for some variables (12% for OP/CB insecticides used) and misclassification of other exposures. In particular, exposure misclassification may have been a concern for worker behaviors, which are somewhat more subjective. However, since participants were unaware of their ChE results at the time of data collection, we would expect any misclassification to be non-differential, resulting in under-estimated associations. Future studies may be able to validate self-reported exposures against direct workplace observations.
We evaluated BuChE inhibition in relation to use of specific OP/CBs during the preceding 30 days. However, due to time constraints we were unable to collect detailed information regarding the degree of exposure to specific OP/CBs. It is likely that this limited our ability to characterize the risk of BuChE inhibition associated with individual OP/CBs.
Due to an administrative change in the statewide monitoring program, ChE assays were performed by different laboratories, with differing measurement methods, in 2006 and 2007. This was unlikely to have been an important bias, however. Absolute BuChE levels did differ by year, yet the percent of BuChE inhibition from baseline levels was not substantially different in 2006 and 2007 (means were 4.8% and 6.6%, respectively). Furthermore, year of participation was included as a covariate in all adjusted analyses. Moreover, when we evaluated follow-up BuChE activity as the outcome variable, results were generally consistent with the findings based on BuChE inhibition reported above (results not shown).
Finally, statistical power was limited in this study, particularly for evaluating associations with relatively uncommon exposures (e.g., not wearing chemical-resistant boots). Thus, our findings should be replicated in other populations with greater heterogeneity of PPE use and other exposure-related factors. Although risk estimates from logistic regression analyses were based on a small number of cases of BuChE depression (N=18), the associations were generally in the same direction as those observed in the linear regression analyses, and several strong associations were observed.
Findings from this study suggest that continued efforts are needed to promote and enforce safe pesticide handling practices among agricultural pesticide handlers. We investigated modifiable worker behaviors and workplace conditions, as well as specific high-risk handling activities. These findings may ultimately inform future targeted interventions to reduce pesticide exposures.
Furthermore, evidence of an association between OP/CB exposure and BuChE inhibition in this study suggests that current regulatory exposure assessment models may under-estimate exposure. Estimates of occupational exposure in pesticide risk assessments could be refined based on associations between BuChE inhibition and specific work activities and practices observed in this study. It should be noted that greater use of PPE was reported among participants in this study relative to pesticide handlers in other regions of the U.S. and in developing countries.[25, 30, 31, 40] Nonetheless, the general pattern of consistency of our findings with those from previous studies of pesticide exposure determinants offers some reassurance that our results have relatively broad generalizeability.
We would like to thank all of the workers who participated in this study. Additionally, we gratefully acknowledge the following individuals for their assistance with this project: Ofelio Borges and Flor Servin from the Washington State Department of Agriculture for their collaboration on the development of the survey, Pam Ernst and Joe Cozzetto from Central Washington Occupational Medicine for their assistance with our recruitment efforts, and Maria Negrete and Pablo Palmandez with the Pacific Northwest Agricultural Safety and Health Center for their hard work and dedication during field data collection. Finally, we thank Drs. Fabio Cabarcas and Donald Cole for their thoughtful comments on this article.
Financial support for this project was provided by U.S. CDC/NIOSH grants #1 U50 OH07544 and #1 T42 OH008433-01, and U.S. NIEHS grants #P30 ES07033, #P42 ES04696, and #T32 ES07262.
The authors declare that they have no competing financial interests.
The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive licence on a worldwide basis to the BMJ Publishing Group Ltd and its Licensees to permit this article to be published in Occupational and Environmental Medicine and any other BMJPGL products to exploit all subsidiary rights, as set out in our licence (http://oem.bmjjournals.com/ifora/licence.pdf).