County-Level Atrazine Use and Gastroschisis

Key Points Question What is the association of county-level atrazine use with infant diagnoses of gastroschisis? Findings In this cross-sectional study of nearly 40 million US births between 2009 and 2019, White race, young age, lower body mass index and parity, cigarette smoking, rurality, and Chlamydia infection during pregnancy were associated with gastroschisis. Additionally, higher levels of atrazine use at the county level were associated with an increased incidence of gastroschisis. Meaning These findings suggest that exploring alternatives to atrazine in the US may be warranted.


Introduction
Gastroschisis is a birth defect involving the abdominal wall that results in herniation of the small bowel and, at times, other abdominal organs. 1,2The condition requires immediate surgical attention following birth, resulting in prolonged hospital stays in neonatal intensive care units.Although surgical advancements have dramatically improved outcomes for infants born with the congenital abnormality, gastroschisis may be associated with substantial morbidity, including short bowel syndrome and intestinal atresia. 1,2Data suggest that rates of gastroschisis have increased globally, [3][4][5] increasing from 0.06 to 0.8 per 10 000 births in the 1960s 6 to 4.9 per 10 000 births in 2014. 5These increased rates have predominantly been observed in young White mothers 3,7 ; however, data on the national incidence of gastroschisis at present are sparse.Despite this recorded rise in incidence of gastroschisis, substantial gaps remain in our knowledge about the cause of the defect.Compared with other congenital anomalies, gastroschisis may be the result of the joint interplay of genetic and nongenetic factors. 6Several theories hypothesize the embryologic origins of gastroschisis, but few have successfully connected known risk factors, such as young age (<25 years), smoking, low body mass index (BMI) (as measured by weight in kilograms divided by height in meters squared), and nulliparity, to the anatomic basis of the defect. 6One predominating theory proposes that these various risk factors may contribute to the pathophysiology of gastroschisis via an estrogen-linked hypothesis. 8In standard embryologic development, the right-side umbilical vein involutes, leaving the umbilical cord with 2 umbilical arteries and the remaining left-side umbilical vein. 8,9This area of involution provides a site that is susceptible to clot formation, which is precipitated by increased levels of estrogen. 8,9Thrombus formation disrupts cell signaling and, consequently, the development of the abdominal wall, leading to the external protrusion of abdominal contents. 8,9While the association between gastroschisis and the established risk factors may be explained by the estrogen-linked hypothesis, it does not account for the increased rates of gastroschisis observed in the US over time.
][10] However, studies investigating the association of atrazine exposure with incidence of gastroschisis have been limited to single states 10,14,15 and counties. 16,17 address these knowledge gaps, our objective was to investigate more recent national trends in gastroschisis incidence and evaluate maternal and infant characteristics associated with gastroschisis.To expand on existing research, we also examined the association between atrazine use and gastroschisis at the national level.While the reasons for the increase in rates of gastroschisis may be multifactorial, we hypothesized that county-level exposure to atrazine would be associated with the increased incidence of gastroschisis, potentially explaining the observed trends.

JAMA Network Open | Public Health
County-Level Atrazine Use and Gastroschisis

Cohort and Data
In this analysis of retrospective, repeated cross-sectional data, we included data for all US births from January 1, 2009, to December 31, 2019.Birth data from the National Vital Statistics System were obtained from the National Center for Health Statistics at the Centers for Disease Control and Prevention.Pesticide data were acquired from the US Geological Survey (USGS).A report published in 2013 by the USGS outlines the methods regarding the collection of these data and generation of pesticide use estimates. 18,19The Vanderbilt University Medical Center institutional review board deemed this study exempt from human participant review, and informed consent was not required.
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.

Conceptual Model
The estrogen-linked hypothesis provides a framework for understanding how different risk factors may exert effects at the molecular level to disrupt abdominal wall development in utero, leading to the development of gastroschisis (eFigure 1 in Supplement 1). 8,9]8,9 Although smoking is not associated with estrogen dysregulation, it has been associated with negative vascular outcomes, which aligns with the estrogen-linked hypothesis's anatomic explanation of gastroschisis. 8,9The literature also has suggested an association with the increased risk of estrogen-mediated clotting among White women. 8Finally, some study results have indicated that mothers who present with sexually transmitted infections, most often Chlamydia, or urinary tract infections around the beginning of the first trimester were more likely to deliver babies with gastroschisis. 20,21It has been hypothesized that rather than these infections being responsible for the development of gastroschisis, high estrogen levels, which underlie the pathophysiology of the defect, may predispose mothers to contract these infections. 22

Exposure of Interest
Our exposure of interest was atrazine use.The USGS used 2 different methods to estimate pesticide use when a crop reporting district did not report pesticide use. 18,19The EPest-low method (hereafter referred to as low) assumed zero use, and the EPest-high method (hereafter referred to as high) treated it as missing data and used the rates from nearby crop reporting districts to estimate the rate.
We include both estimates from USGS in our analysis.
In addition, we considered 3 variations of the atrazine use variable to account for timing of the exposure prior to delivery.The first was the mean county-level atrazine use in the year before the birth, which we refer to as the 1-year average.Next, to account for the potential long-term exposure to atrazine at the county level, we also calculated means of county-level atrazine use for 5 and 10 years before the birth.

Outcomes and Covariates
Incidence of gastroschisis, the outcome of interest, was measured using birth certificate data.
Covariates included maternal age, maternal race and ethnicity (Hispanic; non-Hispanic Black; non-Hispanic White; and other [Alaska Native, Asian or Pacific Islander, non-Hispanic American Indian], given the low collective incidence of gastroschisis), BMI, number of previous births, birth payment source (Medicaid, private insurance, self-pay, other), Chlamydia infection during the pregnancy, smoking during the pregnancy, and rurality (urban, rural adjacent, rural remote).
Information regarding covariates, including race and ethnicity, were obtained from birth certificate data from the National Vital Statistics System.As mentioned previously, research has identified

JAMA Network Open | Public Health
County-Level Atrazine Use and Gastroschisis differences in prevalence of gastroschisis among different racial and ethnic groups, with higher prevalence found in non-Hispanic White mothers. 3,7

Statistical Analysis
Analyses were conducted between August 5, 2021, and May 26, 2023.We compared characteristics of maternal-infant dyads with and without a diagnosis of gastroschisis using medians and IQRs for continuous variables and frequencies and percentages for categorical variables.We tested for differences between the groups using Wilcoxon rank sum tests and χ 2 tests.Additionally, we created line plots to examine the rates of gastroschisis over time both overall and stratified by cigarette use, rurality, and payer.
We also evaluated the geographic distribution of gastroschisis and atrazine use by the 4 US census regions (Northeast, Midwest, South, West) and created maps of the county-level data.The χ 2 and Kruskal-Wallis tests were used to test for differences in gastroschisis rates and mean atrazine use, respectively, among the census regions.Atrazine data were missing for 3.6% of county-years.In 1-year lagged models, these counties were not included; however, in the 5-and 10-year average model, the mean exposure over the time period was considered without including missing countyyear data (ie, if 4 years of data were available within a 5-year period, the mean was calculated using the available 4 years).
Our mixed-effects models included fixed effects for years as an indicator variable and random effects for counties to account for correlation within county.Using these mixed-effects logistic regression models, we examined the association between county-level atrazine use and presence of gastroschisis while adjusting for rurality, maternal age, maternal race and ethnicity, maternal BMI, cigarette use, number of previous births, and Chlamydia infection during pregnancy, calendar year, county of residence, and payment source.In these models, all covariates were fixed effects except county of residence, which was a random effect.Separate models were constructed using each of the atrazine county-level exposure variables, ie, 1-year, 5-year, and 10-year averages.We used the high values for the primary analysis.All tests were 2-sided with a threshold of P < .05for statistical significance.The models were run using Stata, version 15.1 (StataCorp LLC), and the maps were generated using the ggplot2 and urbnmapr packages of R, version 4.2.1 (R Foundation for Statistical Computing).
We conducted supplementary analyses to assess the robustness of our findings.In the first supplementary analysis, we fit the same regression models with the low values for the atrazine variables.In the other supplementary analyses, we excluded births in California because the atrazine data in this state were collected differently compared with the other states. 18,19

Association of County-Level Atrazine Use With Gastroschisis
From 2009 to 2019, a higher proportion of infants with gastroschisis were born in the Midwest compared with other US Census regions (Figure 2; eTable 1 in Supplement 1).Notably, atrazine use from 2009 to 2019 has largely been concentrated in the Midwest, with a mean (SD) use of 22 702.1 (1399.1) kg of atrazine per year within this period.Some counties in the South and on the East Coast

Discussion
In this national study of all US live births between 2009 and 2019, we found that county-level atrazine use was associated with infant diagnoses of gastroschisis.Additionally, we found that the incidence of gastroschisis has declined slightly in recent years.Prior estimates of the national incidence of gastroschisis are limited to post 2014.Like previous research, [1][2][3]8,9 we also found that the diagnosis was more common among previously nulliparous women who identified as White, had a lower median BMI, lived in rural areas, smoked cigarettes, and reported a Chlamydia infection during pregnancy.
Our study found that atrazine use was highest in the midwestern region of the US, where rates of gastroschisis were similarly the highest compared with other census regions.Within our model of atrazine use averaged over a 10-year period, a 100 000-kg increase in average atrazine use in a mother's county of residence was associated with a 21% increase in the odds of an infant being born with gastroschisis, after accounting for covariates and year fixed effects and county random effects.
This association was similarly observed when atrazine use was averaged over 1-year and 5-year periods.16][17]23 One study investigating this correlation in Texas found that the probability of a birth complicated by gastroschisis as a result of atrazine exposure may vary by age, with the risk being greater in mothers 25 years or older. 14Future studies investigating the association between national atrazine use and gastroschisis may reveal differences in risk stratified by age.
The Centers for Disease Control and Prevention estimates that each year, more than 70 million pounds of atrazine are used in the US. 12 Aside from gastroschisis, atrazine has also been associated with preterm birth and low birth weight. 24,25While our study did not assess biomarkers of atrazine exposure, other studies have shown an association between atrazine exposure and adverse birth outcomes using urinary biomarkers. 26The findings from the PELAGIE (Perturbateurs endocriniens: Étude Longitudinale sur les Anomalies de la Grossesse, l'Infertilité et l'Enfance) birth cohort in France showed an association between atrazine exposure and fetal growth restriction and small head circumference. 26Future research using urinary biomarkers may find a stronger association between atrazine exposure and gastroschisis incidence.
Although rates of gastroschisis have appeared to decline, the condition is still associated with considerable infant morbidity.Moreover, surgical repair of the defect has been associated with longer hospital stays and additional costs upward of $300 000, placing an enormous financial burden on families. 3Atrazine is the second-most used herbicide in the US, and widespread application of the pesticide has raised several concerns with regard to its effects on biodiversity. 11,12,27Numerous countries, including the European Union, have banned the substance out of concerns for its adverse effects on human health. 28Our study findings suggest that reexamination of policy regarding atrazine use may be warranted.

Limitations
Our study has limitations that are common among studies of this type.First, birth certificate data may be subject to misclassification bias, with errors of omission or commission.Second, some variations in pesticide use data collection were present at the state level, and some data were missing for states at different time points.To address these limitations, we used 3 models and calculated means of county-level atrazine use for 5 and 10 years before birth to minimize the amount of missingness for our exposure of interest.We also made use of supplementary analyses excluding states with missing data to assess the robustness of our findings.Third, atrazine use data obtained from the USGS from 2018 and 2019 were listed as preliminary estimates.Final estimates are projected to be available in 2025.Fourth, aside from the covariates we identified, there may have been other confounding variables unaccounted for in this study, such as alcohol or opioid use. 8,9,29 present, the data on the association between opioid use and gastroschisis are limited; however, shared downstream pathways between the estrogen receptor and opioid receptor may explain how opioid use fits with the estrogen-linked hypothesis. 30Fifth, while our study made use of reported atrazine use data at the county level, we did not examine direct exposure of mothers to atrazine through the use of biomarkers.Finally, our results indicate an association between county-level atrazine use and incidence of gastroschisis but does not necessarily imply causation.

Conclusions
In summary, this cross-sectional study found that higher county levels of atrazine were associated with infant diagnoses of gastroschisis.While atrazine is the second-most used herbicide in the US, numerous countries around the world have banned it out of concern for adverse effects on human health.These findings suggest that exploring alternatives to atrazine in the US may be warranted.

Table 2 .
Unadjusted Association of County-Level Atrazine Use With Gastroschisis Diagnosis, US 2009-2019

Table 3 .
Multivariable Analysis a of Individual and County-Level Characteristics and Exposures Associated With Gastroschisis, US 2009-2019 a Models account for year fixed effects, county random effects, and use of high estimates of atrazine use.b Includes Asian or Pacific Islander, Alaska Native, and non-Hispanic American Indian.c Includes Indian Health Service, TRICARE (formerly, Civilian Health and Medical Program of Uniformed Service), other government (federal, state, local), and other.
4. Mixed-Effects Logistic Regression Model of Individual-and County-Level Characteristics and Exposures Associated With Gastroschisis, US 2009-2019, Excluding Births in California and Using EPest-High eTable 4. Mixed-Effects Logistic Regression Model of Individual-and County-Level Characteristics and Exposures Associated With Gastroschisis, US 2009-2019, Excluding Births in California and Using EPest-Low eFigure 5. Mixed-Effects Logistic Regression Model of Individual-and County-Level Characteristics and Exposures Associated With Gastroschisis, US 2009-2019, Excluding Births in California and Using EPest-Low