The associations using the mediation model weakened and became non-significant, as hypothesized . Of the constructs of pesticide exposure, AChE activity accounted for most of the weakening of the association in the mediation model. Model 2 further adjusted for AChE and hemoglobin yielded an OR per SD decrease in time after the harvest of 1.83 , whereas model 2 further adjusted for residential distance to flower crops yielded an OR of 1.88 which is closer to that of model 2 alone. Our findings suggest that pesticide exposures, as a result of a peak pesticide use period in floriculture, may result in short-term alterations in both systolic and diastolic blood pressure among children living in agricultural communities. We observed that children examined sooner after the harvest had lower AChE activity and higher SBP and DBP than children examined later during a period of decreased flower production. These inverse associations between time and blood pressure continued through 100 days after the Mother’s Day harvest. In this setting, these findings expand the body of evidence documenting physiological and neurobehavioral alterations in children associated a peak pesticide-use period . Peak pesticide spray seasons have been found to be associated with statistically significantly lower cholinesterase activity and higher urinary metabolites of pesticides among agricultural worker and non-agricultural worker children . Potential pathways of exposure of non-worker children associated with pesticide spray seasons include increased off-target drift of pesticides onto nearby homes, schools and playgrounds, in addition to increased introduction of pesticides into homes by agricultural workers. In our study population, we previously observed that the children’s pesticide exposure related to the Mother’s Day harvest did not appear to be influenced by cohabitation with a flower plantation worker but rather by off-target pesticide drift onto nearby homes,square plastic pot particularly if they live near a flower crop .

While we did not observe effect modification by residential distance to flower crops or cohabitation with a flower plantation worker on the associations between time after the harvest and blood pressure, we did observe that the reported associations were partially mediated by residential distance to crops. For instance, most of the attenuation of the associations between time after the harvest and OR for hypertension or prehypertension, resulted from adjusting for residential distance to plantations and AChE activity , whereas cohabitation with a flower worker had a negligible effect on the estimate . The present findings concur with our previous observations that residential proximity to flower plantations and higher areas of flower plantations near homes were associated with higher SBP in children . Interestingly, we have also reported positive associations between AChE activity and blood pressure in children of the ESPINA study . This, in addition to the positive association between time after the harvest and AChE activity, possibly explain why adjustment for AChE activity in the mediation analysis using the linear variables of time and blood pressure resulted in a stronger association. This divergence of associations with blood pressure across exposure constructs within the same study population may be due to the influence of different classes of pesticides. AChE activity is a marker of exposure to cholinesterase inhibitors, whereas time after a peak-spray season and residential distance to flower plantations are indicators of pesticide drift for many classes of pesticides. Floriculture in Pedro Moncayo county actively uses many classes of fungicides, insecticides and herbicides . In fact, 23 different cholinesterase inhibitor insecticides and over 50 fungicides, in addition to pyrethroid and neonicotinoid insecticides, were reported to be used .

Although information of other pesticides are non-existent, the cholinesterase inhibitor exposure levels that children in Pedro Moncayo County are subjected to may not be very different from those of similar communities in the USA and other developed countries . Our findings agree with the few existing studies on the relationship between pesticide exposures and blood pressure. In rats, increased doses of organophosphate insecticides led to prolonged systolic and diastolic pressure elevation . In humans, persistently exposed male workers involved in mosquito control were found to have an increased prevalence of higher brachial and aortic blood pressure, increased heart rate and increased oxidative stress markers compared to unexposed workers . Similarly, adults with detectable levels of the organophosphate metabolites had higher diastolic blood pressure, lower high-density lipoprotein cholesterol and higher triglyceride levels compared to people with non-detectable levels in a representative survey of the United States of America . Among pregnant women in Italy, the incidence of gestational hypertension was twice as high among women occupationally exposed to pesticides compared to those who had indirect or domestic exposures , and four times greater than the incidence among unexposed women . Furthermore, statistically significant increased odds ratios for gestational hypertension were observed with self-reports of exposure to diazinon and malathion compared to unexposed women. In the same source population as our study, a pilot study found that maternal occupational pesticide exposure while pregnant was associated with significantly increased systolic blood pressure in children ages 6-8 . However, concurrent exposures to pesticides were not associated with blood pressure in that study. Research is needed to understand how various classes of pesticides such as fungicides, herbicides, neonicotinoids, pyrethroids and others may affect the cardiovascular system. From the toxicological perspective of cholinesterase inhibitors, the inhibition of AChE activity by organophosphates results in heightened cholinergic stimulation of muscarinic and nicotinic receptors in the central and peripheral nervous systems . The overstimulation of nicotinic receptors can result in increases in blood pressure and heart rate .

Neonicotinoid insecticide exposures also lead to overstimulation of nicotine receptors; however, human studies are lacking on their associations with hemodynamic constructs such as blood pressure. The potential mechanisms of increased blood pressure of these and other classes of pesticides is a topic for further research. A limitation of our study is its cross-sectional design in which we measured blood pressure at a single point in time for each child. The ESPINA study in 2008 was not designed to assess health effects of pesticide spray periods; however, the short examination period in 2008, which occurred relatively soon after the Mother’s Day harvest, made these analyses possible. Conducting blood pressure assessments before, during and after a peak pesticide spray season is difficult in this population because the heightened pesticide-use periods start at approximately October/November and remain relatively high until early May. Nonetheless, studies assessing blood pressure measures before, during and after a pesticide spray season are needed. Limitations of this study also include our inability to account for factors plausibly related to blood pressure such perinatal health history,square pot concurrent and previous dietary patterns , etc. We do, however, account for heart rate, age, height and BMI. Another limitation is our inability to discern which pesticides may be driving the observed associations. Although a growing number of pesticides can be quantified in urine, accurate methods to quantify many other pesticides, including most fungicides and herbicides, have not been developed. AChE activity is a biomarker of exposure that reflects a physiological change associated with cholinesterase inhibitor exposures. It has lower sensitivity to exposure than measurements of pesticides or pesticide metabolites in bio-specimens; however, it is also a more reliable indicator of cholinesterase inhibitor exposure and provides a much wider exposure window than bio-specimen quantification . The concomitant use of a variety of chemicals in agriculture, including substances for stimulating plant growth or pesticides is a challenge for understanding of the role of specific environmental pollutants on health. In this regard, pesticide exposure constructs such as time after the end of a known exposure period or residential proximity to plantations provide valuable exposure information that crudely accounts for a mixture of exposures. As a result, this study prompts the need to further study the role of agrochemicals on hemodynamic markers in children. This is one of the largest studies to date to estimate the associations between a known pesticide exposure period and physiological changes in children. Our logistical approach to participant examination allowed us to measure blood pressure on 20 different days within 100 days from Mother’s Day harvest, averaging 15 participants per time point. The well-established partnerships and ongoing community involvement that we have developed in Pedro Moncayo County resulted high participation rates, fast recruitment and examination of participants.

This allowed us to have approximations of seasonal changes in blood pressure, although with cross-sectional data. Further research is needed to assess the associations of pesticide spray seasons on the health of children living in agricultural communities. Two features have robustly characterized the growth experiences of most countries of the world: reallocation of labor out of agriculture and increase in human capital, at least as measured by years of schooling. Take Brazil, for example. In 1960 more than half of its male population was employed in agriculture, and the average working age male spent a dismal two years in school. By 2010, those numbers changed significantly: less than fifteen percent of working age males were employed in agriculture, and they spent on average eight years in school. In this paper, we ask whether the increase in human capital contributes to the reallocation of labor out of agriculture, and if so, by how much. To answer this question, we develop an accounting framework that leverages novel empirical evidence on labor reallocation within and across birth cohorts, which we document using micro-level data from 52 countries. We conclude that, for our set of countries, human capital accounts on average for approximately one third of total reallocation out of agriculture, but, at the same time, it has only a minor role in explaining why some countries had faster reallocation than others. Our research question is motivated by a recent literature that documents – using micro level data from countries of all income level – that high skilled workers sort out of agriculture, and interprets this empirical pattern as evidence that non-agricultural production is more skill intensive.1 It is then natural to conjecture that, as the average level of human capital grows over time, an increasing number of workers becomes endowed with skills that are valued more by the non-agricultural sector, thus triggering labor reallocation out of agriculture. Our main challenge is to devise an empirical approach that allows us to disentangle the role of human capital relative to the traditional view on structural change, which treats labor as an homogenous input and argues that the decline in relative agricultural productivity is the driving force behind labor reallocation.2 The development accounting literature offers one possible path to directly measure the contribution of human capital. This literature, starting from the seminal work of Klenow and RodriguezClare and Hall and Jones , has used Mincerian returns to school to create human capital stocks from educational attainments. While appealing, this approach entails a number of drawbacks that makes it unattractive in our context. In particular, it would require to use the cross-sectional relationship between years of schooling and individual probability of agricultural employment to forecast the effect, in the time-series, of an increase in average number years of schooling on the share of the population employed in agriculture. Unfortunately, through this methodology, we may either overestimate or underestimate the role of human capital. The cross sectional estimates may provide an overestimate of the effect of an aggregate change in schooling since they capture omitted variables – such as unobserved skills – which are unlikely to vary over time. At the same time, they could also underestimate the role of human by restricting our focus only to changes in schooling years, thus omitting other sources of human capital increase over time, such as health improvements or better school quality. In order to circumvent these concerns, in this paper we follow a new path. We develop a framework to indirectly back-out the role of human capital from individual choices, rather than from observed quantities of schooling and prices, as in the Mincer approach. We build on one simple insight: following a birth cohort of adult individuals over time keeps mostly constant the level of human capital,3 thus allowing to identify the effect of aggregate changes, such as those in relative agricultural prices/productivity; instead, comparing different cohorts at one point in time, which are exposed to the same aggregate conditions, can pin down the role of human capital. As an example of our approach, consider Figure 1a.