This study used the same overall design and a comparable population sample as a previous pilot study (Grandjean et al. 2006
) and replicated the associations observed in that study between prenatal pesticide exposure and visuospatial deficits and increased blood pressure. A joint analysis of the data from both studies confirmed that these outcomes were associated with maternal work during pregnancy, not the father’s work, and not the current exposure to pesticides. In addition, the present study identified lower performance on visual memory and motor tasks as possible effects of prenatal pesticide exposure. A small decrease in BMI was also associated with prenatal exposure. Nutritional parameters such as stunting, current hematocrit levels, number of meals per day, and socioeconomic characteristics did not explain these associations. Neurophysiologic measurements, including evoked potential latencies and heart variability, were not associated with either prenatal or current exposures.
The study design aimed at comparing two groups of children (prenatally exposed vs. nonexposed) with similar backgrounds, except for the maternal history of occupational exposure to pesticides. Three socioeconomic indicators showed better conditions in the exposed group (i.e., higher maternal income, quality of housing, and paternal employment). These differences are meaningful, given the possibility for both parents to be economically active, and they replicate previous observations (Grandjean et al. 2006
). The same applies to maternal education, although the absence of a measure of maternal IQ is a limitation. Any residual differences not accounted for by the covariate adjustment would be expected to bias the study toward the null hypothesis, that is, toward finding no associations between prenatal pesticide exposure and children’s neuropsychological functioning. Otherwise, the similarities between the two groups regarding past medical history and all other respects, including current pesticide exposure, would suggest that confounding bias may be limited in this study.
A study with a cross-sectional design and retrospective assessment of prenatal exposure cannot provide information about dose–response relationships or the time of the impact of the exposures. However, the standardized and blinded techniques applied to the maternal reports support the validity of the employment-based classification of exposures. Although a small degree of misclassification cannot be excluded (e.g., because of domestic uses of pesticides), it would be expected to result in an underestimate of the true neurotoxicity of the exposure. Although selection bias is a common problem in cross-sectional studies, it is unlikely in our case, because all eligible children participated in the examinations in both studies (Grandjean et al. 2006
). Nevertheless, this problem may not be completely resolved, because children not attending school would be missed. The small sample size of the study is also a clear limitation, but the parallel findings in the two studies and the statistically significant findings are remarkable.
The present study included the same neuropsychological instruments used in the previous study in addition to several supplementary tests (Grandjean et al. 2006
). They were carefully selected to be sensitive to developmental neurotoxicity in low doses observed in environmental exposures to pesticides and other pollutants such as methylmercury (Grandjean et al. 1997
). In addition, the instruments used were also validated to avoid cross-cultural influences (Grandjean et al. 1999
; Lezak 1995
; White et al. 1994
), as would be appropriate for indigenous and mixed populations in rural areas of developing countries. Thus, although the battery did not provide a complete assessment of all functional domains, major functions were covered by tests that were feasible and showed the anticipated association with age.
The results support the notion that, in this cohort of children, prenatal exposures to pesticides are more harmful than current exposures, thereby confirming previous results of other environmental studies of neurodevelopmental toxicity and the theory of window of vulnerability of central nervous system during uterine life (Grandjean and Landrigan 2006
). Paternal exposures during pregnancy showed much weaker associations than did maternal exposures during pregnancy, again in accordance with the findings in other studies (Wigle et al. 2009
). Furthermore, the clearest deficits were observed in neuropsychological functions that involved visuospatial scorings, thereby replicating the pilot study that used the same test (Stanford-Binet Copying Test), as well as the findings of other studies of organophosphate exposures during pregnancy (Grandjean et al. 2006
; Guillette et al. 1998
). An association that corresponds to a delay of almost 2 years is large when considering the rapid development in children at early school age.
Moreover, in this case we also observed alterations to memory and motor functions that were in consonance with the findings of previous studies in younger children (Guillette et al. 1998
; Handal et al. 2008
). In addition, a study on long-term neuropsychological dysfunctions of school-age children exposed to organophosphates pesticides during infancy (Ruckart et al. 2004
) reported that both motor inhibition and verbal learning were impaired in the exposed children.
Although nutritional deficiencies may negatively affect neurodevelopment, the associations observed between prenatal pesticide exposure and children’s test scores were independent of stunting and other indicators of nutritional status. Consequently, as previously observed, pesticide exposures and child malnutrition, as common problems in developing countries, may independently increase the risks of long-term neuropsychological impairment (Grandjean et al. 2006
We did not find associations between pesticide exposure and other neurophysiologic outcomes such as visual and auditory brainstem evoked potentials and audiometric and heart variability examinations. Recent scientific literature has shown that these outcomes are sensitive to neurotoxicants, such as lead and methylmercury (Murata et al. 1999
; Rothenberg et al. 2000
). These associations have been documented in fairly large studies, and the absence of an association with pesticide exposure could be due to the small sample size in the present study. At least these outcomes appeared less sensitive to pesticide toxicity at the levels of exposure experienced by the children in this cohort.
Regarding blood pressure, we replicated the results of the pilot study (Grandjean et al. 2006
). Prenatal exposure was associated most closely with increased systolic blood pressure. This association was independent of relevant covariates, such as stunting and maternal smoking during pregnancy, known as risk factors for increased blood pressure in children (Brion et al. 2008
; Morley et al. 1995
; Wilson et al. 1998
). This observation is consistent with the hypothesis that the autonomic nervous system may play a role with organophosphate toxicity, as has been considered in previous poisoning studies (Bardin et al. 1994
A decreased BMI was independently related to prenatal organophosphate exposure, a finding that is consistent with some reports from birth cohort studies of reduced birth weight and birth length as an effect of pesticide exposure (Whyatt et al. 2004
). There appears to be no report on older children, however. The biological mechanism is unknown, but acetylcholine may stimulate contraction of the uterus, thus resulting in a reduced birth length and weight (Whyatt et al. 2004
). BMI in this case may then reflect a long-term consequence. Nevertheless, the percentage of stunting in the study group was higher (38%) than the average percentage estimated for Ecuador in 2004 (25%). Factors other than malnutrition, such as the high altitude of the study area or the Andean genetic background, may contribute to the high calculated proportion of stunted children (Grandjean et al. 2006
; Larrea and Kawachi 2005
The present study suggests that the current level of protection may well be adequate to avoid pesticide toxicity in the worker herself but insufficient to prevent lasting adverse effects in the offspring. Deficits associated with prenatal pesticide exposure may contribute to a “silent pandemic” of developmental neurotoxicity (Grandjean and Landrigan 2006
), and this study therefore adds to the evidence suggesting a need for improved control of occupational exposures that may cause intrauterine neurotoxicity (Julvez and Grandjean 2009
). Regarding pregnant women at work, conventions on maternity protection of the International Labour Organization (ILO) require that a pregnant woman not be obliged to perform work that has been determined to be harmful to her health or that of her child (ILO 2000
). However, the most recent version of this convention has been ratified by only 17 countries so far, and the general practice in Ecuador is for expecting mothers to continue work until the very last day before childbirth; the rules for maternity leave provide no protection against developmental neurotoxicity.