Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Jan 1.
Published in final edited form as: Epidemiology. 2018 Jan;29(1):8–21. doi: 10.1097/EDE.0000000000000757

Residential agricultural pesticide exposures and risks of spontaneous preterm birth

Gary M Shaw 1, Wei Yang 1, Eric M Roberts 2, Susan E Kegley 3, David K Stevenson 1, Suzan L Carmichael 1, Paul B English 4
PMCID: PMC5718919  NIHMSID: NIHMS906506  PMID: 28926371

Abstract

Background

Pesticides exposures are aspects of the human exposome that have not been sufficiently studied for their contribution to risk for preterm birth. We investigated risks of spontaneous preterm birth from potential residential exposures to 543 individual chemicals and 69 physicochemical groupings that were applied in the San Joaquin Valley of California during the study period, 1998–2011.

Methods

The study population was derived from birth certificate data linked with Office of Statewide Health Planning and Development maternal and infant hospital discharge data. After exclusions, the analytic study base included 197,461 term control births and 27,913 preterm case births. Preterm cases were more narrowly defined as 20–23 weeks (n=515), 24–27 weeks (n=1792), 28–31 weeks (n=3098), or 32–36 weeks (n=22,508).

Results

The frequency of any (versus none) pesticide exposure was uniformly lower in each preterm case group relative to the frequency in term controls, irrespective of gestational month of exposure. All odds ratios were below 1.0 for these any vs no exposure comparisons. The majority of odds ratios were below 1.0, many of them statistically precise, for preterm birth and exposures to specific chemical groups or chemicals.

Conclusions

This study showed a general lack of increased risk of preterm birth associated with a range of agriculture pesticide exposures near women’s residences.

Keywords: pesticides, environment, prematurity, endocrine disruptors, pregnancy

INTRODUCTION

Preterm birth (delivery before 37 weeks of gestation) is associated with substantial morbidity and mortality with >15 million babies born preterm every year in the world.1 In the US, preterm birth occurs in ~12% of live births.2 There are iatrogenic explanations of preterm birth, most of which can be attributed to maternal or fetal conditions requiring medical intervention to facilitate earlier delivery. Risk factors for spontaneous preterm birth, however, remain largely unexplained. Factors associated with spontaneous preterm birth have included race, infection, stress, genetics, and environmental toxicants.3 Each of these broad factors has either explained only a small fraction of the population burden of spontaneous preterm birth or has been insufficiently studied to derive meaningful inferences.25

Despite some pesticides being known reproductive toxicants6 and substantial public concern about pesticide exposures, few studies have investigated relations between specific pesticide exposures and pregnancy outcomes including spontaneous preterm birth. The few studies that have explored pesticide exposures and preterm birth suggest some associations, but extant data are insufficient to draw clear inferences.79 Studies specifically investigating residential pesticide exposures and preterm birth risks are nearly non-existent.8 In general, studies of preterm birth and pesticides have been limited by the exposure assessments employed, been small in size, or have not investigated finer gestational ages to define preterm birth other than simply <37 weeks. To overcome many of these limitations, we investigated population-based data on >200,000 births and proximal residential exposures to more than 500 commercial agricultural pesticide active ingredients and adjuvants during multiple gestational time points, to extend the limited extant literature on pesticides and spontaneous preterm birth. The study population derived from the San Joaquin Valley of California, one of the highest agricultural pesticide use areas in the US.

MATERIALS AND METHODS

Study population

This study was approved by the Stanford University Institutional Review Board and the California State Committee for the Protection of Human Subjects.

Data for this case–control study come from 1998–2011 California births to women residing in the San Joaquin Valley (Fresno, Kern, Kings, Madera, Merced, San Joaquin, Stanislaus, and Tulare counties). In this region and time period there were 892,088 livebirths delivered in non-military hospitals. We restricted the study to those with gestational ages 20–41 weeks (determined by obstetric estimate for 2007–11 and by last menstrual period for 1998–2006), whose birth weights were between 500 and 5000 grams, and were singleton births. Among 771,416 eligible births, there were 78,421 preterm, i.e., <37 weeks gestation (cases) and 692,995 term, i.e., ≥37 weeks gestation. From the term group, we randomly selected 235,263 births (controls) in a 3:1 control to case ratio.

For each case and control we extracted from the birth certificate the residential address at the time of birth. The California Environmental Health Tracking Program (CEHTP) Geocoding Service was used to geocode these addresses.10,11 The CEHTP Geocoding Service standardizes, verifies, and corrects addresses before matching against multiple address-attributed reference databases. Successful geocoding was achieved for 73,736 (94%) preterm cases and for 221,651 (94%) term controls.

We further linked the 73,736 cases and 221,651 controls derived from birth certificate data with Office of Statewide Health and Planning (OSHPD) maternal and infant hospital discharge data. This linkage allowed for information on a range of maternal and pregnancy characteristics found on the birth certificate paired with clinical detail from the delivery hospitalization for practically all inpatient live births. The algorithm employed for this linkage is accurate and previously described.12,13 Successful linkage was achieved for 72,907 (99%) preterm cases and 220,137 (99%) term controls.

Our analytic goal was to specifically investigate spontaneous preterm birth. Thus, the case group was further restricted to spontaneous preterm birth events based on information coded on hospital discharge or birth certificate records. Spontaneous preterm birth was identified as those births <37 weeks with preterm premature rupture of membranes (ICD-9-CM code 658.1 or birth certificate complication of labor/delivery code 10), premature labor (ICD-9-CM code 644), or the use of tocolytics (birth certificate complication/procedure of pregnancy code 28). This reduced the preterm cases to n=36,758 (excluded from the total were 368 deliveries at 20–23 weeks, 715 at 24–27 weeks, 2366 at 28–31 weeks, and 32,700 at 32–36 weeks). Further, we excluded women with the selected comorbidities of pregestational diabetes (n=888), gestational diabetes (n=2908), gestational hypertension (n=820), pre-eclampsia/eclampsia (n=4739), and chronic hypertension (n=1386) from each case group (except the 20–23 week group for which gestational diabetes was not considered an exclusion criterion because delivery occurred prior to gestational diabetes being typically diagnosed) and from controls (n=22676). These exclusions were motivated by our goal to determine whether “pesticide exposures” alone, i.e., not mediated by or through these comorbid conditions, contributed to spontaneous preterm birth risks. These comorbidities were identified from codes pertinent to the birth hospitalization in the form of ICD-9-CM diagnoses. Specifically, we applied methods similar to those used elsewhere14 to assess maternal morbidity in pregnancy as follows: diabetes (250 and 648.0), gestational diabetes (648.8), chronic hypertension (401–405, 642.0, 642.1, 642.2, 642.7, and 642.9), gestational hypertension (642.3), and preeclampsia/eclampsia (642.4, 642.5, and 642.6). These refinements to the preterm case phenotype resulted in 197,461 term control births and 27,913 preterm birth case births serving as the analytic base. Preterm cases were more narrowly defined as, 20–23 weeks (n=515), 24–27 weeks (n=1792), 28–31 weeks (n=3098), or 32–36 weeks (n=22,508).

Pesticide and adjuvant compounds studied

We assessed exposure to 543 individual chemicals used as pesticides or as adjuvants in pesticide products or application mixtures and 69 physicochemical groupings having the same chemical classification and proven or putative mechanism of action (e.g., organophosphates) that were applied at >100 lb in any of the 8 San Joaquin Valley counties in any year during the study period (1998–2011).15 Low-toxicity chemicals such as biopesticides (e.g., microbial pesticides, soaps, essential oils), low-toxicity inorganic compounds (e.g., sulfur, kaolin clay), and other compounds determined by the US Environmental Protection Agency (EPA) to have low toxicity, as described in US EPA Risk Assessment documents for each chemical16 were excluded. In addition, compounds were flagged as having reproductive or developmental toxicity based on the California Proposition 65 list17 or as endocrine disruptors.1820 Chemicals with a US EPA-determined Reference Dose based on a toxicologic study with a reproductive or developmental endpoint as described in EPA risk assessment documents were included.16

Pesticide exposure assessment

To estimate pesticide exposures, we assigned a time window of exposure for each case or control mother from one month before conception (B1) to date of delivery by every 4 weeks of pregnancy (P1–P9).

To estimate pesticide applications, we obtained statewide Pesticide Use Reporting (PUR) records from the California Department of Pesticide Regulation describing agricultural pesticide applications occurring between 1 January 1998 and 31 December 2011.15 These data are submitted by county agriculture commissioners and are spatially referenced to public land survey sections (PLSS). During the study period, the total number of active ingredient daily production agricultural use records with a PLSS specified, and for the 543 chemicals that were present in PUR records, exceeded 24 million. Following the method of Rull and Ritz,21 we spatially refined PLSS polygons through overlay of matched land-use survey field polygons provided by the California Department of Water Resources. We matched each PUR record to the land-use survey conducted closest in time to the application date (surveys are conducted roughly every 5–7 years in each California county). Matching is based on PLSS and crop type as specified in records. Infrequently rotated crops, such as orchard crops and vineyards, were matched one-to-one, while frequently rotated crops, such as field and truck crops, were grouped together in a single category, and non-agricultural land uses were subtracted from PLSS polygons when no crop types were matched to available polygons. Of the total applications (and active-ingredient poundage) recorded spanning 1998–2011 for the 543 chemicals of interest, >90% were successfully linked to polygons. For those where no field polygon was specified, no spatial refinement was possible. We determined temporal proximity by comparing recorded dates of applications (which are believed to be accurate within a few days) to the time window of exposure for each study subject.

To assign exposure, we utilized the CEHTP Pesticide Linkage Tool, a custom-developed Java (Oracle, Redwood Shores, CA) application that incorporates the PostGIS spatial and geographic objects library for PostgreSQL (http://www.postgis.net/) and the GeoTools Java GIS Toolkit, version Release 12 (open source, http://www.geotools.org/) for GIS data management and spatial analysis.10,11 We calculated pounds of pesticides used during the relevant time window within a 500 m radius of a geocoded point,22 intersecting polygons with the buffer, and assuming homogeneous distribution of pesticides within each polygon.

Statistical analysis

Risks associated with pesticide exposures were estimated using logistic regression. Univariate analyses were conducted to estimate crude odds ratios and 95% confidence intervals (CI) reflecting associations between pesticide exposures and spontaneous preterm birth. We performed multivariable analyses adjusting for maternal age (years), race/ethnicity (non-Hispanic white, U.S.-born Hispanic, foreign-born Hispanic, other), education (less than high school, high school, more than high school), parity (1 or ≥2), prenatal care initiated by fifth month (yes vs no), payer source for care (Medi-Cal, private, or other). Additional analyses based on the availability of data beginning with 2007 births were performed adjusting for pre-pregnancy body mass index (BMI in kg/m2, continuous) and neighborhood poverty derived from US Census data for census block groups. Given that the source of potential covariate information was derived from the birth certificate we determined that women’s cigarette smoking was too incomplete to include in analyses.

To focus on comparisons likely to have the most precise estimates and to fully utilize available data, we did the following. For pesticides that had five or more exposed cases and controls, risks were estimated that compared any versus no exposure. Risks were not estimated for pesticides that had fewer than five exposed cases and controls. We also created overall exposure “scores.” These scores were created by flagging studied chemicals as having reproductive or developmental toxicity based on the California Proposition 65 list17 or as endocrine disruptors.1820 Chemicals with an EPA-determined reference dose based on an acute toxicological study with a reproductive or developmental endpoint as described in EPA risk assessment documents were also included.16 We created overall exposure scores by summing the total number of chemical groups, endocrine disruptors, Proposition 65 chemicals, or chemicals in EPA lists. For the exposure scores, we examined the associations of specific preterm birth phenotypes with these scores specified as categorical variables; that is, exposed subjects were divided into tertiles based on the control distributions.

Analyses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC, 2015–2016).

RESULTS

Compared to term controls, mothers of preterm infants were more likely to be Black, initiate prenatal care after the fifth month of pregnancy, have their delivery paid under MediCal benefits, be nulliparous, or more likely to deliver male infants (Table 1).

Table 1.

Descriptive characteristics (percentages)a of preterm case and term control infants, California, 1998–2011 (n=225,374)

Characteristic Gestational age in weeks at delivery

CASES CONTROLS

20–23
n=515
24–27
n=1,792
28–31
n=3,098
32–36
n=22,508
37–41
n=197,261
Maternal age (years)
  <20 19.2 19.4 20.1 16.2 14.2
  20–24 24.7 27.2 28.3 28.8 29.9
  25–29 23.1 23.8 23.2 25.7 28.0
  30–34 18.8 17.4 16.3 18.2 18.4
  >35 14.2 12.2 12.1 11.1 9.5
  Missing 0 0 0.1 <0.1 <0.1
Maternal race/ethnicity
  White, nonHispanic 19.8 22.7 24.1 28.4 29.4
  White, Hispanic 60.6 57.1 54.9 54.8 57.0
  Black 9.5 9.4 9.6 6.3 4.5
  Asian 7.0 8.7 8.8 8.3 7.1
  Other 1.9 1.6 1.9 1.8 1.5
  Missing 1.2 0.4 0.6 0.6 0.6
Maternal education
  Less than high school 36.9 34.0 37.1 33.7 32.6
  High school 32.8 34.2 33.8 31.0 31.8
  More than high school 27.8 29.4 26.8 33.6 34.0
  Missing 2.5 2.5 2.3 1.8 1.6
Prenatal care initiation by fifth month of gestation
  Yes 85.2 88.2 85.9 88.4 91.8
  No 8.9 7.2 10.3 8.6 6.4
  Missing 5.8 4.6 3.8 3.0 1.8
Parity
  1 41.8 45.1 38.1 35.8 34.8
  ≥2 57.9 54.9 61.8 64.1 65.2
  Missing 0.4 <0.1 0.2 <0.1 0.1
Payer type for delivery
  Medi-Cal 63.5 61.9 63.7 60.0 56.9
  Private 28.9 30.5 30.2 35.5 39.7
  Uninsured 6.2 6.2 4.5 2.9 1.7
  Other 1.0 1.0 1.2 1.3 1.5
  Missing 0.4 0.4 0.4 0.3 0.2
Infant sex
  Male 56.5 57.1 58.5 55.1 50.6
  Female 43.5 42.8 41.5 44.9 49.4
  Missing 0 <0.1 0 <0.1 <0.1
Infant Birth Year
  1998 5.8 5.7 6.3 4.9 6.7
  1999 6.4 5.5 6.3 4.7 6.7
  2000 7.4 6.8 6.4 5.1 6.9
  2001 5.6 6.3 7.0 5.7 6.9
  2002 5.6 5.7 6.6 5.9 7.1
  2003 5.1 5.9 6.3 6.2 7.4
  2004 9.5 6.9 6.9 6.4 7.9
  2005 8.7 7.5 9.0 6.9 9.1
  2006 8.0 7.9 8.4 7.7 9.5
  2007 8.0 8.5 8.8 9.4 6.8
  2008 5.4 8.8 7.6 10.0 6.7
  2009 8.4 7.8 7.1 9.1 6.3
  2010 7.4 7.6 6.5 9.0 6.1
  2011 8.7 9.2 7.0 9.1 6.0
Years 2007–2011 195 751 1141 10452 63021
Prepregnancy body mass index (kg/m2) (2007–2011)
  Underweight (BMI<18.5) 2.1 5.5 5.0 4.7 3.1
  Normal (18.5≤BMI<25) 30.3 30.8 38.9 42.6 40.2
  Overweight (25≤BMI<30) 19.5 26.1 22.8 23.0 24.7
  Obese (BMI≥30) 27.2 24.1 20.0 18.0 20.3
  Missing 21.0 13.6 13.3 11.8 11.8
Poverty (2007–2011)b
  ≤107.25 20.5 16.1 16.5 17.5 19.4
  107.26 – ≤180.14 13.3 17.7 17.3 18.2 19.4
  180.15 – ≤260.29 20.0 19.4 18.6 19.0 19.5
  260.30 – ≤365.66 21.0 21.0 22.4 20.8 19.6
  >365.66 21.5 23.0 22.4 21.6 19.3
  Missing 3.6 2.7 3.0 2.9 2.7
a

Percentages may not equal 100 owing to rounding

b

Quintile cutoffs were determined among term births. The highest quintile reflects the highest degree of poverty.

Frequencies of preterm cases and term controls with any vs no exposure assignments for the B1–P9 month periods are shown in Table 2. The frequency of any exposure was uniformly lower in each preterm case group, and month time period, relative to the frequency in term controls. The corresponding odds ratios (crude and adjusted) are shown in Table 3. All odds ratios were below 1.0 for these any vs no exposure comparisons. Stratum-specific analyses by male and female births did not reveal substantially different results.

Table 2.

Any (as percentage of total) gestational pesticide exposure per month among women of preterm cases and term controls

Count and percentage of any exposure (% yes vs no) per montha
Gestational
Weeks at
Delivery
(total No.)
B1 P1 P2 P3 P4 P5 P6 P7 P8 P9
Pre-term 20–23 (n=515) 129 (25.0) 112 (21.7) 139 (27.0) 127 (24.7) 133 (25.8) 124 (24.1)
24–27 (n=1,792) 413 (23.0) 407 (22.7) 432 (24.1) 416 (23.2) 439 (24.5) 430 (24.0) 415 (23.2)
28–31 (n=3,098) 790 (25.5) 789 (25.5) 815 (26.3) 806 (26.0) 816 (26.3) 792 (25.6) 764 (24.7) 772 (24.9)
32–36 (n=22,508) 5850 (26.0) 5900 (26.2) 5937 (26.4) 5911 (26.3) 5966 (26.5) 5976 (26.6) 5871 (26.1) 5921 (26.3) 5782 (25.7)
Term 37–42 (n=197,461) 55136 (27.9) 55507 (28.1) 55770 (28.2) 56019 (28.4) 55834 (28.3) 56029 (28.4) 56079 (28.4) 56000 (28.4) 55662 (28.2) 54960 (27.8)
a

B1=month before conception, P1–P9=each successive month from first to ninth month of pregnancy,

Table 3.

Risks (odds ratios) for any vs no gestational pesticide exposure (per month) among women of preterm cases and term controls

Gestational
Weeks at
Delivery
Month of
Exposurea
Case (Preterm)
(exposed/not-
exposed)
Control (Term)
(exposed/not-
exposed)
Crude OR
(95%CI)
Adjustedb OR
(95%CI)
20–23 B1 129/386 55136/142325 0.86 (0.71–1.05) 0.86 (0.70–1.06)
P1 112/403 55507/141954 0.71 (0.58–0.88) 0.73 (0.59–0.91)
P2 139/376 55770/141691 0.94 (0.77–1.14) 0.94 (0.77–1.16)
P3 127/388 56019/141442 0.83 (0.68–1.01) 0.84 (0.68–1.04)
P4 133/382 55834/141627 0.88 (0.72–1.08) 0.89 (0.72–1.10)
P5 124/391 56029/141432 0.80 (0.65–0.98) 0.83 (0.67–1.03)
24–27 B1 413/1379 55136/142325 0.77 (0.69–0.86) 0.82 (0.73–0.92)
P1 407/1385 55507/141954 0.75 (0.67–0.84) 0.78 (0.70–0.88)
P2 432/1360 55770/141691 0.81 (0.72–0.90) 0.84 (0.75–0.94)
P3 416/1376 56019/141442 0.76 (0.68–0.85) 0.80 (0.71–0.90)
P4 439/1353 55834/141627 0.82 (0.74–0.92) 0.86 (0.77–0.96)
P5 430/1362 56029/141432 0.80 (0.71–0.89) 0.82 (0.73–0.92)
P6 415/1377 56079/141382 0.76 (0.68–0.85) 0.79 (0.71–0.89)
28–31 B1 790/2308 55136/142325 0.88 (0.81–0.96) 0.93 (0.86–1.01)
P1 789/2309 55507/141954 0.87 (0.81–0.95) 0.91 (0.83–0.99)
P2 815/2283 55770/141691 0.91 (0.84–0.98) 0.95 (0.87–1.03)
P3 806/2292 56019/141442 0.89 (0.82–0.96) 0.92 (0.84–1.00)
P4 816/2282 55834/141627 0.91 (0.84–0.98) 0.95 (0.87–1.03)
P5 792/2306 56029/141432 0.87 (0.80–0.94) 0.90 (0.83–0.98)
P6 764/2334 56079/141382 0.83 (0.76–0.90) 0.87 (0.80–0.95)
P7 772/2326 56000/141461 0.84 (0.77–0.91) 0.88 (0.81–0.96)
32–36 B1 5850/16658 55136/142325 0.91 (0.88–0.94) 0.92 (0.89–0.95)
P1 5900/16608 55507/141954 0.91 (0.88–0.94) 0.92 (0.89–0.95)
P2 5937/16571 55770/141691 0.91 (0.88–0.94) 0.93 (0.90–0.96)
P3 5911/16597 56019/141442 0.90 (0.87–0.93) 0.91 (0.88–0.94)
P4 5966/16542 55834/141627 0.91 (0.89–0.94) 0.93 (0.90–0.96)
P5 5976/16532 56029/141432 0.91 (0.88–0.94) 0.93 (0.90–0.96)
P6 5871/16637 56079/141382 0.89 (0.86–0.92) 0.91 (0.88–0.93)
P7 5921/16587 56000/141461 0.90 (0.87–0.93) 0.92 (0.89–0.95)
P8 5782/16726 55662/141799 0.88 (0.85–0.91) 0.90 (0.87–0.93)
a

B1=month before conception, P1–P8=each successive month from first to ninth month of pregnancy.

b

Odds ratio adjusted for maternal age (years), race/ethnicity(non-Hispanic white, U.S.-born Hispanic, foreign-born Hispanic, other), education (less than high school, high school, more than high school),parity (1 or ≥2), prenatal care initiated by fifth month (yes vs no), payer source for care (Medi-Cal, private, or other).

As noted in the Methods, we employed a minimum sample size criterion for risk estimation, i.e., pesticides (groups or specific chemicals) that had five or more exposed cases and controls for each phenotype. This produced upwards of 54,000 comparisons based on four preterm case groups (20–23, 24–27, 28–31, and 32–36 gestational weeks), as many as 9 exposure months (i.e., B1–P9), 313 chemical groups with exposure, 61 chemical classes of exposure, crude and adjusted odds ratios, and stratification by sex of the infant (this latter stratification motivated by higher frequency of males among preterm births and some pesticides having endocrine disruptor mechanisms). Owing to this large number of comparisons (not easily conveyed in journal tables), we have limited our presentation of results as follows, but summarize in text the general pattern of findings not specifically shown. We show adjusted odds ratios for chemical groups and specific chemicals for which 1) there were at least five cases exposed (this criterion biases toward identifying elevated risks) and 2) only for the exposure month closest to the time of delivery (e.g., for preterm cases 20–23 weeks at delivery the odds ratios shown are for month P5). These results are displayed in Table 4 for chemical groups, and eTable 1 for specific chemicals.

Table 4.

Risks (odds ratios) for any vs no gestational exposures (per month) for specific pesticide chemical groups among women of preterm cases and term controls. Shown are adjusted odds ratios for chemical groups where there were ≥5 cases exposed and for the exposure month closest to the time of delivery (e.g., for preterm cases 20–23 weeks at delivery the odds ratios shown are for month P5).

Chemical Class Gestation
Weeks at
Delivery
Month of
Exposurea
Case
Preterm
(exp/not-
exp)
Control Term
(exp/not-exp)
Adjusted ORb
(95%CI)
2,4 - Dichlorophenoxy acid or ester 20–23 P5 15/500 5792/191669 0.94 (0.54–1.64)
2,4 - Dichlorophenoxy acid or ester 24–27 P6 30/1762 5754/191707 0.60 (0.42–0.87)
2,4 - Dichlorophenoxy acid or ester 28–31 P7 63/3035 5683/191778 0.76 (0.59–0.97)
2,4 - Dichlorophenoxy acid or ester 32–36 P8 586/21922 5710/191751 0.92 (0.84–1.01)
2,6-Dinitroaniline 20–23 P5 15/500 8944/188517 0.65 (0.38–1.11)
2,6-Dinitroaniline 24–27 P6 66/1726 8657/188804 0.88 (0.68–1.13)
2,6-Dinitroaniline 28–31 P7 114/2984 8588/188873 0.89 (0.73–1.07)
2,6-Dinitroaniline 32–36 P8 909/21599 8442/189019 0.97 (0.90–1.04)
Alcohol/Ether 20–23 P5 12/503 6610/190851 0.70 (0.38–1.27)
Alcohol/Ether 24–27 P6 41/1751 6340/191121 0.72 (0.52–0.99)
Alcohol/Ether 28–31 P7 81/3017 6272/191189 0.86 (0.68–1.07)
Alcohol/Ether 32–36 P8 438/22070 6273/191188 0.63 (0.57–0.69)
Alkyl Phthalate 32–36 P8 11/22497 282/197179 0.33 (0.18–0.62)
Amide 24–27 P6 5/1787 1214/196247 0.51 (0.21–1.23)
Amide 28–31 P7 19/3079 1243/196218 1.13 (0.72–1.78)
Amide 32–36 P8 117/22391 1232/196229 0.83 (0.69–1.02)
Amine 28–31 P7 7/3091 698/196763 0.67 (0.32–1.41)
Amine 32–36 P8 46/22462 679/196782 0.62 (0.46–0.83)
Anthranilic diamide 24–27 P6 8/1784 547/196914 1.74 (0.86–3.52)
Anthranilic diamide 28–31 P7 11/3087 645/196816 1.25 (0.69–2.28)
Anthranilic diamide 32–36 P8 106/22402 675/196786 1.45 (1.17–1.78)
Antibiotic 32–36 P8 9/22499 44/197417 1.78 (0.84–3.80)
Aryloxyphenoxy propionic acid 28–31 P7 11/3087 301/197160 2.38 (1.26–4.48)
Aryloxyphenoxy propionic acid 32–36 P8 24/22484 296/197165 0.73 (0.48–1.12)
Avermectin 20–23 P5 11/504 5029/192432 0.83 (0.45–1.56)
Avermectin 24–27 P6 34/1758 5249/192212 0.77 (0.55–1.09)
Avermectin 28–31 P7 87/3011 5196/192265 1.14 (0.91–1.42)
Avermectin 32–36 P8 576/21932 5232/192229 0.97 (0.89–1.06)
Azole 20–23 P5 16/499 8873/188588 0.64 (0.38–1.09)
Azole 24–27 P6 54/1738 8914/188547 0.67 (0.51–0.89)
Azole 28–31 P7 128/2970 8554/188907 1.00 (0.84–1.20)
Azole 32–36 P8 945/21563 8588/188873 0.99 (0.92–1.06)
Benzoic acid 24–27 P6 6/1786 1453/196008 0.50 (0.22–1.11)
Benzoic acid 28–31 P7 20/3078 1425/196036 0.98 (0.63–1.52)
Benzoic acid 32–36 P8 125/22383 1425/196036 0.80 (0.66–0.96)
Bipyridylium 20–23 P5 23/492 9778/187683 0.90 (0.58–1.40)
Bipyridylium 24–27 P6 56/1736 9839/187622 0.65 (0.49–0.85)
Bipyridylium 28–31 P7 120/2978 9784/187677 0.83 (0.69–1.00)
Bipyridylium 32–36 P8 908/21600 9717/187744 0.84 (0.79–0.90)
Bis-Carbamate 32–36 P8 5/22503 52/197409 0.82 (0.33–2.07)
Botanical 20–23 P5 6/509 3408/194053 0.71 (0.32–1.58)
Botanical 24–27 P6 19/1773 3393/194068 0.61 (0.38–0.97)
Botanical 28–31 P7 54/3044 3315/194146 1.06 (0.80–1.40)
Botanical 32–36 P8 341/22167 3304/194157 0.91 (0.81–1.02)
Chloroacetanilide 24–27 P6 6/1786 953/196508 0.64 (0.27–1.55)
Chloroacetanilide 28–31 P7 13/3085 904/196557 0.92 (0.52–1.62)
Chloroacetanilide 32–36 P8 83/22425 890/196571 0.83 (0.66–1.05)
Copper compound 20–23 P5 21/494 11732/185729 0.71 (0.45–1.11)
Copper compound 24–27 P6 80/1712 11537/185924 0.80 (0.63–1.00)
Copper compound 28–31 P7 137/2961 11510/185951 0.78 (0.65–0.93)
Copper compound 32–36 P8 1158/21350 11486/185975 0.90 (0.84–0.96)
Cyclohexenone derivative 24–27 P6 6/1786 1357/196104 0.43 (0.18–1.05)
Cyclohexenone derivative 28–31 P7 17/3081 1335/196126 0.83 (0.51–1.36)
Cyclohexenone derivative 32–36 P8 120/22388 1375/196086 0.76 (0.63–0.92)
Diacylhydrazine 20–23 P5 6/509 2801/194660 0.88 (0.39–1.98)
Diacylhydrazine 24–27 P6 22/1770 2906/194555 0.88 (0.57–1.35)
Diacylhydrazine 28–31 P7 47/3051 2901/194560 1.11 (0.83–1.49)
Diacylhydrazine 32–36 P8 357/22151 2907/194554 1.12 (1.00–1.25)
Dicarboximide 20–23 P5 7/508 4047/193414 0.63 (0.28–1.41)
Dicarboximide 24–27 P6 18/1774 4085/193376 0.47 (0.29–0.77)
Dicarboximide 28–31 P7 44/3054 3868/193593 0.77 (0.57–1.04)
Dicarboximide 32–36 P8 316/22192 3825/193636 0.72 (0.63–0.81)
Dithiocarbamate-ETU 20–23 P5 13/502 4509/192952 1.21 (0.70–2.11)
Dithiocarbamate-ETU 24–27 P6 31/1761 4364/193097 0.80 (0.56–1.16)
Dithiocarbamate-ETU 28–31 P7 54/3044 4214/193247 0.85 (0.64–1.12)
Dithiocarbamate-ETU 32–36 P8 387/22121 4290/193171 0.79 (0.71–0.88)
Dithiocarbamate-MITC 24–27 P6 7/1785 1116/196345 0.78 (0.37–1.64)
Dithiocarbamate-MITC 28–31 P7 19/3079 1135/196326 1.18 (0.74–1.88)
Dithiocarbamate-MITC 32–36 P8 109/22399 1142/196319 0.85 (0.69–1.05)
Endothall 32–36 P8 16/22492 226/197235 0.67 (0.40–1.12)
Glyco Ether 32–36 P8 18/22490 205/197256 0.81 (0.50–1.32)
Glycol 20–23 P5 8/507 3276/194185 0.92 (0.44–1.94)
Glycol 24–27 P6 21/1771 3201/194260 0.75 (0.48–1.17)
Glycol 28–31 P7 54/3044 3219/194242 1.16 (0.88–1.52)
Glycol 32–36 P8 210/22298 3308/194153 0.57 (0.50–0.66)
Halogenated organic 24–27 P6 11/1781 2054/195407 0.65 (0.36–1.17)
Halogenated organic 28–31 P7 17/3081 1965/195496 0.58 (0.35–0.94)
Halogenated organic 32–36 P8 158/22350 1939/195522 0.73 (0.61–0.86)
Hydroxybenzonitrile 24–27 P6 7/1785 1186/196275 0.72 (0.34–1.52)
Hydroxybenzonitrile 28–31 P7 11/3087 1210/196251 0.65 (0.36–1.18)
Hydroxybenzonitrile 32–36 P8 116/22392 1117/196344 0.93 (0.76–1.13)
Imidazolinone 28–31 P7 6/3092 606/196855 0.69 (0.31–1.54)
Imidazolinone 32–36 P8 79/22429 664/196797 1.09 (0.86–1.38)
Insect growth regulator (Chitin) 20–23 P5 6/509 1433/196028 1.72 (0.77–3.86)
Insect growth regulator (Chitin) 24–27 P6 5/1787 1444/196017 0.42 (0.17–1.00)
Insect growth regulator (Chitin) 28–31 P7 20/3078 1477/195984 0.85 (0.54–1.36)
Insect growth regulator (Chitin) 32–36 P8 169/22339 1496/195965 1.01 (0.85–1.18)
Monochlorophenoxy acid or ester 24–27 P6 11/1781 1518/195943 0.89 (0.49–1.62)
Monochlorophenoxy acid or ester 28–31 P7 18/3080 1515/195946 0.85 (0.54–1.36)
Monochlorophenoxy acid or ester 32–36 P8 133/22375 1506/195955 0.78 (0.65–0.93)
N-Methyl Carbamate 20–23 P5 8/507 5496/191965 0.58 (0.29–1.17)
N-Methyl Carbamate 24–27 P6 39/1753 5436/192025 0.81 (0.59–1.12)
N-Methyl Carbamate 28–31 P7 80/3018 5428/192033 0.96 (0.76–1.20)
N-Methyl Carbamate 32–36 P8 526/21982 5382/192079 0.87 (0.79–0.95)
Naphthalene acetic acid derivative 32–36 P8 6/22502 45/197416 1.22 (0.52–2.87)
Neonicotinoid 20–23 P5 8/507 5141/192320 0.47 (0.21–1.06)
Neonicotinoid 24–27 P6 36/1756 5197/192264 0.77 (0.55–1.08)
Neonicotinoid 28–31 P7 83/3015 5342/192119 1.05 (0.84–1.32)
Neonicotinoid 32–36 P8 596/21912 5317/192144 1.01 (0.92–1.10)
Organoarsenic 32–36 P8 22/22486 310/197151 0.64 (0.42–0.99)
Organochlorine 24–27 P6 9/1783 1326/196135 0.80 (0.42–1.55)
Organochlorine 28–31 P7 18/3080 1310/196151 0.92 (0.58–1.47)
Organochlorine 32–36 P8 108/22400 1309/196152 0.74 (0.61–0.90)
Organophosphate 20–23 P5 41/474 16712/180749 0.91 (0.65–1.28)
Organophosphate 24–27 P6 123/1669 16763/180698 0.86 (0.71–1.03)
Organophosphate 28–31 P7 217/2881 16993/180468 0.86 (0.75–0.99)
Organophosphate 32–36 P8 1622/20886 17033/180428 0.85 (0.80–0.89)
Petroleum derivative-Aromatic 20–23 P5 31/484 11522/185939 1.02 (0.70–1.50)
Petroleum derivative-Aromatic 24–27 P6 84/1708 11398/186063 0.87 (0.70–1.08)
Petroleum derivative-Aromatic 28–31 P7 148/2950 11373/186088 0.90 (0.76–1.06)
Petroleum derivative-Aromatic 32–36 P8 1073/21435 11470/185991 0.83 (0.78–0.89)
Phosphine 24–27 P6 5/1787 997/196464 0.64 (0.26–1.53)
Phosphine 28–31 P7 8/3090 986/196475 0.51 (0.24–1.08)
Phosphine 32–36 P8 87/22421 974/196487 0.82 (0.66–1.03)
Phosphonoglycine 20–23 P5 53/462 25745/171716 0.79 (0.59–1.06)
Phosphonoglycine 24–27 P6 158/1634 25816/171645 0.68 (0.58–0.81)
Phosphonoglycine 28–31 P7 309/2789 25733/171728 0.78 (0.69–0.88)
Phosphonoglycine 32–36 P8 2612/19896 25416/172045 0.91 (0.87–0.95)
Piperonyl 32–36 P8 48/22460 465/196996 0.94 (0.69–1.28)
Polyalkyloxy Compound 20–23 P5 31/484 12148/185313 1.05 (0.73–1.53)
Polyalkyloxy Compound 24–27 P6 72/1720 11826/185635 0.68 (0.54–0.87)
Polyalkyloxy Compound 28–31 P7 154/2944 11677/185784 0.86 (0.73–1.02)
Polyalkyloxy Compound 32–36 P8 840/21668 11641/185820 0.64 (0.59–0.69)
Pyrethroid 20–23 P5 27/488 12509/184952 0.79 (0.52–1.19)
Pyrethroid 24–27 P6 91/1701 12452/185009 0.83 (0.66–1.03)
Pyrethroid 28–31 P7 161/2937 12582/184879 0.88 (0.75–1.03)
Pyrethroid 32–36 P8 1295/21213 12541/184920 0.91 (0.86–0.97)
Pyridazinone 20–23 P5 7/508 2053/195408 1.44 (0.68–3.05)
Pyridazinone 24–27 P6 6/1786 1994/195467 0.36 (0.16–0.81)
Pyridazinone 28–31 P7 22/3076 2012/195449 0.74 (0.48–1.14)
Pyridazinone 32–36 P8 148/22360 2029/195432 0.66 (0.56–0.78)
Quaternary Ammonium Compound 24–27 P6 13/1779 1522/195939 0.93 (0.53–1.65)
Quaternary Ammonium Compound 28–31 P7 22/3076 1560/195901 0.94 (0.62–1.44)
Quaternary Ammonium Compound 32–36 P8 164/22344 1570/195891 0.92 (0.78–1.08)
Silicone 20–23 P5 13/502 4358/193103 1.28 (0.74–2.23)
Silicone 24–27 P6 34/1758 4322/193139 0.88 (0.62–1.25)
Silicone 28–31 P7 61/3037 4297/193164 0.92 (0.71–1.20)
Silicone 32–36 P8 305/22203 4358/193103 0.62 (0.55–0.70)
Spinosyn 20–23 P5 8/507 4160/193301 0.78 (0.39–1.58)
Spinosyn 24–27 P6 44/1748 4304/193157 1.13 (0.83–1.55)
Spinosyn 28–31 P7 63/3035 4307/193154 1.01 (0.78–1.30)
Spinosyn 32–36 P8 503/22005 4287/193174 1.05 (0.95–1.15)
Streptomycin 32–36 P8 36/22472 473/196988 0.69 (0.49–0.98)
Strobin 20–23 P5 13/502 6184/191277 0.81 (0.46–1.44)
Strobin 24–27 P6 48/1744 6165/191296 0.86 (0.64–1.16)
Strobin 28–31 P7 90/3008 6140/191321 1.00 (0.80–1.24)
Strobin 32–36 P8 617/21891 6117/191344 0.89 (0.82–0.98)
Sulfonylurea 20–23 P5 5/510 1365/196096 1.59 (0.66–3.85)
Sulfonylurea 24–27 P6 7/1785 1468/195993 0.58 (0.27–1.21)
Sulfonylurea 28–31 P7 21/3077 1363/196098 1.07 (0.70–1.66)
Sulfonylurea 32–36 P8 145/22363 1346/196115 0.96 (0.80–1.14)
Thiocarbamate 24–27 P6 5/1787 756/196705 0.82 (0.34–1.97)
Thiocarbamate 28–31 P7 15/3083 761/196700 1.36 (0.80–2.30)
Thiocarbamate 32–36 P8 55/22453 776/196685 0.66 (0.50–0.87)
Thiophanate, benzimidazole precursor 28–31 P7 5/3093 761/196700 0.48 (0.20–1.15)
Thiophanate, benzimidazole precursor 32–36 P8 61/22447 783/196678 0.70 (0.53–0.91)
Thiophthalimide 28–31 P7 12/3086 1259/196202 0.69 (0.39–1.22)
Thiophthalimide 32–36 P8 100/22408 1203/196258 0.76 (0.62–0.94)
Triazine 20–23 P5 12/503 6948/190513 0.73 (0.41–1.29)
Triazine 24–27 P6 38/1754 6676/190785 0.65 (0.47–0.91)
Triazine 28–31 P7 87/3011 6831/190630 0.87 (0.70–1.08)
Triazine 32–36 P8 632/21876 6626/190835 0.86 (0.79–0.93)
Urea 20–23 P5 8/507 5284/192177 0.61 (0.31–1.24)
Urea 24–27 P6 31/1761 5243/192218 0.66 (0.46–0.96)
Urea 28–31 P7 71/3027 5201/192260 0.92 (0.72–1.17)
Urea 32–36 P8 507/22001 5139/192322 0.88 (0.80–0.96)
Xylylalanine 24–27 P6 10/1782 1664/195797 0.74 (0.40–1.38)
Xylylalanine 28–31 P7 26/3072 1597/195864 1.01 (0.67–1.54)
Xylylalanine 32–36 P8 147/22361 1591/195870 0.81 (0.68–0.97)
Zinc,inorganic 24–27 P6 6/1786 984/196477 0.74 (0.33–1.65)
Zinc,inorganic 28–31 P7 8/3090 981/196480 0.43 (0.19–0.97)
Zinc,inorganic 32–36 P8 80/22428 971/196490 0.75 (0.60–0.95)
a

B1=month before conception, P1–P8=each successive month from first to ninth month of pregnancy.

b

Odds ratio adjusted for maternal age (years), race/ethnicity(non-Hispanic white, U.S.-born Hispanic, foreign-born Hispanic, other), education (less than high school, high school, more than high school), parity (1 or ≥2), prenatal care initiated by fifth month (yes vs no), payer source for care (Medi-Cal, private, or other).

As shown in Table 4, there was only a single comparison (aryloxyphenoxy proprionic acid) that showed a statistically precise (confidence interval did not include 1.0) increased risk. Indeed, the majority of adjusted odds ratios were below 1.0 (crude estimates were similar), many of them statistically precise. Results for the “months of exposure” not shown were not substantially different than those that are shown. Stratum specific analyses by male and female births did not reveal substantially different results than those that appear in Table 4.

In eTable 1 are the adjusted odds ratios associated with specific chemicals. Similar to results for chemical groups, only a small number of elevated risks was observed with the majority of adjusted odds ratios observed to be below 1.0 (crude estimates were similar). The 18 comparisons observed to have elevated odds ratios ranged in magnitude from 1.17 (diacylhydrazine) to 2.94 (silicone). The observed elevated odds ratios were associated with a variety of chemicals and reflected the spectrum of preterm case phenotypes. Stratum specific analyses by male and female births did not reveal substantially different results with but a few exceptions. That is, specific chemicals that were associated with more than a 2-fold (only an elevated risk direction) observed OR differential between males and females were: 1) for males, polyalkyloxy compound exposure for gestational age at delivery 20–23 weeks, OR=3.09 (1.27–7.53), pyrethroids (cypermethrin) exposure for gestational age at delivery of 24–27 weeks, OR=2.35 (1.04–5.29), quaternary ammonium compound (dimethylbenzyl ammonium chloride) exposure for gestational age at delivery of 32–36 weeks, OR=4.65 (1.72–12.59); urea (thidiazuron) exposure for gestational age at delivery of 24–27 weeks, OR=2.15 (1.14–4.04) and 2) for females dithiocarbamate-MITC (potassium n-methyldithiocarbamate) exposure for gestational age at delivery of 28–31 weeks, OR=3.13 (1.38–7.09), thiocarbamate (cycloate) for gestational age at delivery of 28–31 weeks, OR=4.02 (1.88–8.60), alkyl phthalate (chlorthal-dimethyl) exposure for gestational age at delivery of 28–31 weeks, OR=3.57 (1.57–8.12), and spirotetramat exposure for gestational age at delivery of 28–31 weeks, OR=2.55 (1.04–6.22).

To estimate potential effects associated with a sum of multiple exposures we explored “scores” to various chemical classifications, including number of chemical groups, endocrine disruptors, Proposition 65-listed reproductive toxicants, or EPA listed reproductive or developmental toxicants. Increasing numbers of exposures to any of these classifications did not show elevated risks, but rather decreasing risks of preterm birth with increasing sums of exposures (Table 5).

Table 5.

Adjusted Odds Ratios (ORs) for sums of specific classifications of pesticide exposures and preterm birth.

20–23 weeks
(Exposure month=P5)
24–27 weeks
(Exposure month=P6)
28–31 weeks
(Exposure month=P7)
32–36 weeks
(Exposure month=P8)
Case/Control OR (95%CI)a Case/Control OR (95%CI)a Case/Control OR (95%CI)a Case/Control OR (95%CI)a
No. of chemical groups with any exposureb
0 393/142670 Reference 1384/142691 Reference 2343/142786 Reference 16871/143068 Reference
1–2 49/22574 0.80 (0.58–1.09) 192/22771 0.89 (0.76–1.04) 332/22811 0.92 (0.81–1.03) 2523/22572 0.96 (0.92–1.00)
3–5 46/18168 1.01 (0.74–1.38) 137/18121 0.83 (0.69–0.99) 261/18185 0.92 (0.81–1.05) 1868/18122 0.90 (0.85–0.95)
>5 27/14049 0.68 (0.45–1.03) 79/13878 0.61 (0.48–0.77) 162/13679 0.78 (0.67–0.92) 1246/13699 0.79 (0.75–0.84)
Continuous 515/197461 0.97 (0.93–1.01) 1792/197461 0.95 (0.92–0.97) 3098/197461 0.98 (0.96–0.99) 22508/197461 0.97 (0.97–0.98)
No. of endocrine disruptors with any exposure
0 428/159266 Reference 1508/159413 Reference 2604/159562 Reference 18768/159666 Reference
1 42/17198 0.95 (0.69–1.32) 154/17351 0.98 (0.83–1.17) 222/17196 0.82 (0.71–0.95) 1798/17083 0.92 (0.87–0.96)
2 25/9561 1.02 (0.68–1.55) 68/9470 0.78 (0.61–1.01) 128/9490 0.86 (0.72–1.04) 974/9532 0.88 (0.82–0.95)
>2 20/11436 0.61 (0.37–0.99) 62/11227 0.61 (0.47–0.79) 144/11213 0.86 (0.72–1.02) 968/11180 0.76 (0.71–0.81)
Continuous 515/197461 0.93 (0.84–1.02) 1792/197461 0.90 (0.85–0.95) 3098/197461 0.95 (0.92–0.99) 22508/197461 0.94 (0.92–0.95)
No. of Prop. 65 reproductive toxicants with any exposure
0 485/180295 Reference 1705/179906 Reference 2864/180147 Reference 20857/180276 Reference
1 23/13352 0.64 (0.41–1.00) 67/13588 0.56 (0.44–0.71) 167/13513 0.80 (0.68–0.94) 1316/13311 0.87 (0.82–0.92)
>1 7/3814 0.76 (0.36–1.60) 20/3967 0.58 (0.38–0.91) 67/3801 1.18 (0.92–1.52) 335/3874 0.76 (0.68–0.86)
Continuous 515/197461 0.77 (0.58–1.02) 1792/197461 0.69 (0.58–0.81) 3098/197461 0.97 (0.88–1.07) 22508/197461 0.88 (0.85–0.92)
No. of reproductive or developmental toxicants with any exposure
0 397/144777 Reference 1403/144662 Reference 2371/144756 Reference 17083/145090 Reference
1–2 54/24235 0.85 (0.63–1.14) 204/24504 0.89 (0.77–1.04) 346/24455 0.89 (0.79–1.00) 2586/24139 0.93 (0.89–0.97)
3–4 41/13611 1.18 (0.85–1.63) 102/13451 0.83 (0.67–1.02) 194/13420 0.93 (0.80–1.08) 1432/13459 0.92 (0.87–0.98)
>4 23/14838 0.57 (0.36–0.88) 83/14844 0.61 (0.48–0.76) 187/14830 0.83 (0.71–0.97) 1407/14773 0.83 (0.78–0.88)
Continuous 515/197461 0.95 (0.91–1.00) 1792/197461 0.94 (0.92–0.97) 3098/197461 0.98 (0.96–1.00) 22508/197461 0.98 (0.97–0.99)
a

Odds ratio adjusted for maternal age (years), race/ethnicity (non-Hispanic white, U.S.-born Hispanic, foreign-born Hispanic, other), education (less than high school, high school, more than high school), parity (1 or ≥2), prenatal care initiated by fifth month (yes vs no), payer source for care (Medi-Cal, private, or other).

b

This categorization reflects the total number of chemical groups (total possible=67) that an individual was co

For a subset (2007–11) of the overall data (1998–2011) we had information about body mass index and poverty (see Table 1 for description and frequency). These additional variables were added as covariates to adjusted models. Results of these additional analyses did not show substantially different findings from those displayed in Tables 24.

As sensitivity analyses, we re-analyzed data without the exclusions of women with diabetes or pre-eclampsia as well as re-analyzed data excluding women with a prior history of preterm birth. Results of these additional sets of analyses did not show substantially different findings from those displayed in Tables 24 (not shown).

DISCUSSION

We examined associations between women’s residential proximity to agriculture-related pesticide applications in the San Joaquin Valley of California during pregnancy and risk of spontaneous preterm birth. Despite a very large study population, consideration of the preterm phenotype into narrower categories than simply <37 gestational weeks, and consideration of a variety of gestational exposure definitions such as chemical groups, specific chemicals, and number of pesticides, there was a notable lack of association between pesticide exposures and elevated risk of spontaneous preterm birth. Indeed, owing to the large number of comparisons we would have expected many more elevated risks to emerge by chance alone. Results were not materially influenced by presented potential confounders (e.g., maternal age and race/ethnicity) or by those not presented (e.g., year of birth).

Previous research on residential proximity to pesticide applications and risks of preterm birth phenotypes is nearly non-existent. To our knowledge there has been one study that has investigated residential proximity to pesticides. Willis et al.,23 in a small cohort study of 535 women, observed women who reported living near land used for agricultural purposes were not at increased risk to deliver preterm.

Although there has been limited investigation of residential proximity and preterm birth, there have been observed associations between other measures of pesticide exposures and preterm birth. These other measures of exposure have included individual-level estimations such as serum measures of DDE24 or chlordecone,25 as well as more ecologic-level estimations such as county-level pesticide use.26,27 However, findings from these study designs are not directly comparable to those in the current study.

The explanation of our overall observed results is not obvious. Many of the analytic comparisons indicated reduced risks of spontaneous preterm birth and various pesticide exposure estimations. Clearly, it is difficult to believe that such exposures, given the manifold toxicities these various compounds have, could be beneficial to reducing the likelihood of preterm birth. It also seems unlikely that these observations arose from a lack of control for suspected confounders, namely, cigarette smoking, proximity to greenspace,28 and better air quality. That is, 60% of our study population was Hispanic women, a population subgroup known to have very low use of cigarettes, the study region is one of extensive agricultural land use (not greenspace) and the study region has very poor air quality with noted increased risks observed by us previously for preterm birth from exposures to selected air pollutants.29

Given the unexpected direction of our overall results, one might posit that what is being observed here is that pesticide exposures in pregnancy before 20 weeks (the earliest a birth would be identified in vital statistics files) selectively increase the odds of earlier loss or stillbirth of fetuses destined to be born preterm and therefore not observable when only live birth data are the analytic substrate. Others have described the construct of live-birth selection bias.30 Although this selection bias proposition seems plausible, and has been advanced for studies of congenital anomalies,31,32 the extant data investigating potential associations between miscarriage and residential pesticide exposures is also quite sparse.8 Thus, further investigation to understand whether such a selective culling of pregnancies, from the overall population cohort of pregnancies, that would otherwise be born preterm based on pesticide exposures could lead to potentially biased reduced risks (among births), seems clearly warranted.

Among the strengths of our study is the specificity with which we defined the outcome of spontaneous preterm birth, although this is also a source of nuance when considering our primary findings. As noted, we required delivery prior to 37 complete weeks with spontaneous onset of labor and absence of co-morbidities such as maternal pre-diabetes or diabetes, hypertension, or pre-eclampsia. We did conduct sensitivity analyses by re-analyzing data without the exclusions of women with diabetes or pre-eclampsia. These additional sets of analyses did not show substantially different findings. Although the outcome definition we employed maximized the specificity of our findings, further studies might examine other clinical scenarios such as indicated preterm birth with associated co-morbidities not specifically investigated here.

Our study has several other strengths as well, including its population-based design, large sample size, and an exposure assessment that was highly detailed and spatially and temporally specific (to multiple gestational periods), and captured a broad spectrum of pesticide compounds.

Our study also had challenges. Our assessment of residential proximity to agricultural pesticide applications was thorough, but it did not take into account other factors such as qualities of the pesticides and individuals’ behaviors that could affect actual exposures (e.g., chemical half-lives and vapor pressure, wind patterns, accumulative exposures over time a woman may have had before pregnancy, and other sources of pesticide exposure such as occupation or home use). That is, the basis of pesticide exposure considered here was proximity to pesticide applications associated with women’s residence at delivery. Exposure attribution was based on residence address at delivery rather than at other points in pregnancy. Misclassification of exposure could have occurred for women who moved during gestation. If moving was unrelated to preterm vs term birth status, results would be biased toward the null; if not, the direction cannot be predicted. Further, duration of time spent at the given address is unknown and likely reflective of only a portion of what a woman may encounter in her broader environment. Although many pesticides are prone to drift and detectable in air samples at locations beyond the application site,33 and residential proximity to pesticide-treated fields has been associated with household dust and urine levels,34,35 there are certainly other exposure sources such as in food or water that were not considered here, whereby individuals could be exposed. For example, atrazine levels in drinking water have been associated with small increased risks of preterm birth.26 These various sources of misclassification would be expected to be non-differential, reducing our precision to estimate potential associations. Our analyses, although extensive, did not investigate risks to specific pesticides independent of other pesticides nor did it investigate specific pesticide combinations. Exploration of such independent and combinatorial exposures may be the focus of future unsupervised analytic queries.

In addition, an individual’s ability to metabolize the various types of chemicals in pesticides would certainly affect actual tissue exposures. It has been demonstrated that genetic polymorphisms in detoxification pathways for many pesticides may contribute to preterm birth risks.36,37 Such genetic inquiries were beyond the scope of this initial investigation.

Our study rigorously adds to the scant literature on this topic, particularly in its effort to investigate numerous pesticide compounds.

Supplementary Material

eTable

Acknowledgments

Sources of Funding: This project was supported by NIH (R01HD075761) with additional support from the March of Dimes Prematurity Research Center at Stanford University (MOD PR625253) and the Stanford Child Health Research Institute.

Footnotes

Conflict of Interest: None to declare.

Data Sharing: The data are publicly available from the Office of Statewide Health Planning and Development (OSHPD). The data are not available for replication because specific approvals from OSHPD and the California Committee for the Protection of Human Subjects must be obtained in order to access them.

References

  • 1.Born Too Soon: The Global Action Report on Preterm Birth. New York, NY: March of Dimes and the World Health Organization; 2012. [Google Scholar]
  • 2.Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet. 2008;371(9606):75–84. doi: 10.1016/S0140-6736(08)60074-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Muglia LJ, Katz M. The enigma of spontaneous preterm birth. N Engl J Med. 2010;362(6):529–35. doi: 10.1056/NEJMra0904308. [DOI] [PubMed] [Google Scholar]
  • 4.Behrman RE, Butler AS, editors. Preterm Birth: Causes, Consequences, and Prevention. The National Academies Collection: Reports funded by National Institutes of Health; Washington (DC): 2007. [PubMed] [Google Scholar]
  • 5.Slattery MM, Morrison JJ. Preterm delivery. Lancet. 2002;360(9344):1489–97. doi: 10.1016/S0140-6736(02)11476-0. [DOI] [PubMed] [Google Scholar]
  • 6.Shepard TH. Catalog of Teratogenic Agents (Shepherd’s Catalog) 8. Baltimore, MD: The Johns Hopkins University Press; 1995. (last updated through Micromedex 2009) [Google Scholar]
  • 7.Ferguson KK, O'Neill MS, Meeker JD. Environmental contaminant exposures and preterm birth: a comprehensive review. J Toxicol Environ Health B Crit Rev. 2013;16(2):69–113. doi: 10.1080/10937404.2013.775048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Shirangi A, Nieuwenhuijsen M, Vienneau D, Holman CD. Living near agricultural pesticide applications and the risk of adverse reproductive outcomes: a review of the literature. Paediatr Perinat Epidemiol. 2011;25(2):172–91. doi: 10.1111/j.1365-3016.2010.01165.x. [DOI] [PubMed] [Google Scholar]
  • 9.Wigle DT, Arbuckle TE, Turner MC, et al. Epidemiologic evidence of relationships between reproductive and child health outcomes and environmental chemical contaminants. J Toxicol Environ Health B Crit Rev. 2008;11(5–6):373–517. doi: 10.1080/10937400801921320. [DOI] [PubMed] [Google Scholar]
  • 10.California-Environmental-Health-Tracking-Program. Geocoding Service. [Accessed October 2012]; http://www.cehtp.org/page/tools/geocoding_service.
  • 11.California-Environmental-Health-Tracking-Program. Agricultural Pesticide Mapping Tool. [Accessed October, 2012]; http://www.cehtp.org/page/pesticides/agricultural_pesticide_use_in_california.
  • 12.Herrchen B, Gould JB, Nesbitt TS. Vital statistics linked birth/infant death and hospital discharge record linkage for epidemiological studies. Comput Biomed Res. 1997;30(4):290–305. doi: 10.1006/cbmr.1997.1448. [DOI] [PubMed] [Google Scholar]
  • 13.Lyndon A, Lee HC, Gilbert WM, Gould JB, Lee KA. Maternal morbidity during childbirth hospitalization in California. J Matern Fetal Neonatal Med. 2012;25(12):2529–35. doi: 10.3109/14767058.2012.710280. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Kuklina EV, Ayala C, Callaghan WM. Hypertensive disorders and severe obstetric morbidity in the United States. Obstet Gynecol. 2009;113(6):1299–306. doi: 10.1097/AOG.0b013e3181a45b25. [DOI] [PubMed] [Google Scholar]
  • 15.California-Department-of-Pesticide-Regulation. Pesticide Use Reporting. [Accessed in 2012]; http://www.cdpr.ca.gov/docs/pur/purmain.htm.
  • 16.EPA-(U.S.-Environmental-Protection-Agency) Pesticide Chemical Search. [Accessed in 2012]; http://www.epa.gov/pesticides/chemicalsearch.
  • 17.California-Office-of-Environmental-Health-Hazard-Assessment. Proposition 65. [Accessed in 2012]; http://www.oehha.ca.gov/prop65.html.
  • 18.Colborn T. Our Stolen Future: Widespread pollutants with reproductive and endocrine-disrupting effects. [Accessed in 2012]; http://www.ourstolenfuture.org/basics/chemlist.htm.
  • 19.European-Commission. Towards the Establishment of a Priority List of Substances for Further Evaluation of Their Role in Endocrine Disruption, Appendix 1. [Accessed October 30, 2012]; http://ec.europa.eu/environment/endocrine/strategy/substances_en.htm.
  • 20.Keith LH. Environmental endocrine disruptors : a handbook of property data. New York: Wiley; 1997. [Google Scholar]
  • 21.Rull RP, Ritz B. Historical pesticide exposure in California using pesticide use reports and land-use surveys: an assessment of misclassification error and bias. Environ Health Perspect. 2003;111(13):1582–9. doi: 10.1289/ehp.6118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Roberts EM, English PB, Grether JK, Windham GC, Somberg L, Wolff C. Maternal residence near agricultural pesticide applications and autism spectrum disorders among children in the California Central Valley. Environ Health Perspect. 2007;115(10):1482–9. doi: 10.1289/ehp.10168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Willis WO, de Peyster A, Molgaard CA, Walker C, MacKendrick T. Pregnancy outcome among women exposed to pesticides through work or residence in an agricultural area. J Occup Med. 1993;35(9):943–9. doi: 10.1097/00043764-199309000-00019. [DOI] [PubMed] [Google Scholar]
  • 24.Longnecker MP, Klebanoff MA, Zhou H, Brock JW. Association between maternal serum concentration of the DDT metabolite DDE and preterm and small-for-gestational-age babies at birth. Lancet. 2001;358(9276):110–4. doi: 10.1016/S0140-6736(01)05329-6. [DOI] [PubMed] [Google Scholar]
  • 25.Kadhel P, Monfort C, Costet N, et al. Chlordecone exposure, length of gestation, and risk of preterm birth. Am J Epidemiol. 2014;179(5):536–44. doi: 10.1093/aje/kwt313. [DOI] [PubMed] [Google Scholar]
  • 26.Rinsky JL, Hopenhayn C, Golla V, Browning S, Bush HM. Atrazine exposure in public drinking water and preterm birth. Public Health Rep. 2012;127(1):72–80. doi: 10.1177/003335491212700108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Winchester P, Proctor C, Ying J. County-level pesticide use and risk of shortened gestation and preterm birth. Acta Paediatr. 2016;105(3):e107–15. doi: 10.1111/apa.13288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Hystad P, Davies HW, Frank L, et al. Residential greenness and birth outcomes: evaluating the influence of spatially correlated built-environment factors. Environ Health Perspect. 2014;122(10):1095–102. doi: 10.1289/ehp.1308049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Padula AM, Mortimer KM, Tager IB, et al. Traffic-related air pollution and risk of preterm birth in the San Joaquin Valley of California. Ann Epidemiol. 2014;24(12):888–95e4. doi: 10.1016/j.annepidem.2014.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Liew Z, Olsen J, Cui X, Ritz B, Arah OA. Bias from conditioning on live birth in pregnancy cohorts: an illustration based on neurodevelopment in children after prenatal exposure to organic pollutants. Int J Epidemiol. 2015;44(1):345–54. doi: 10.1093/ije/dyu249. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Khoury MJ, Flanders WD, James LM, Erickson JD. Human teratogens, prenatal mortality, and selection bias. Am J Epidemiol. 1989;130(2):361–70. doi: 10.1093/oxfordjournals.aje.a115342. [DOI] [PubMed] [Google Scholar]
  • 32.Stein Z, Susser M, Warburton D, Wittes J, Kline J. Spontaneous abortion as a screening device. The effect of fetal survival on the incidence of birth defects. Am J Epidemiol. 1975;102(4):275–90. doi: 10.1093/oxfordjournals.aje.a112163. [DOI] [PubMed] [Google Scholar]
  • 33.Kegley SEHB, Orme S, Choi AH. PAN Pesticide Database. [Accessed December 3, 2012]; http://www.pesticideinfo.org/
  • 34.Fenske RA, Kissel JC, Lu C, et al. Biologically based pesticide dose estimates for children in an agricultural community. Environ Health Perspect. 2000;108(6):515–20. doi: 10.1289/ehp.00108515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Simcox NJ, Fenske RA, Wolz SA, Lee IC, Kalman DA. Pesticides in household dust and soil: exposure pathways for children of agricultural families. Environ Health Perspect. 1995;103(12):1126–34. doi: 10.1289/ehp.951031126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Harley KG, Huen K, Aguilar Schall R, et al. Association of organophosphate pesticide exposure and paraoxonase with birth outcome in Mexican-American women. PLoS One. 2011;6(8):e23923. doi: 10.1371/journal.pone.0023923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Mustafa MD, Banerjee BD, Ahmed RS, Tripathi AK, Guleria K. Gene-environment interaction in preterm delivery with special reference to organochlorine pesticides. Mol Hum Reprod. 2013;19(1):35–42. doi: 10.1093/molehr/gas039. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

eTable

RESOURCES