Organophosphate and Pyrethroid Pesticide Exposures Measured before Conception and Associations with Time to Pregnancy in Chinese Couples Enrolled in the Shanghai Birth Cohort

Abstract

Background:

Pesticides have been associated with reproductive disorders, but there is limited research on pesticide exposures and human fertility.

Objective:

We aimed to investigate the effects of preconception exposure to pesticides on time to pregnancy (TTP) and on infertility in a general population of couples planning to become pregnant in Shanghai, China.

Methods:

A total of 615 women who were planning a pregnancy were enrolled before conception and were prospectively followed for 1 y to observe their TTP. Preconception pesticide exposures were assessed by measuring urinary metabolites of organophosphates (OPs) and pyrethroids (PYRs). Fecundability odds ratios (FORs) and odds ratios (ORs) of infertility were estimated using Cox and logistic regression models, respectively. All analyses were repeated after restricting the sample to nulliparous women ($n=569$).

Results:

After adjusting for age, prepregnancy BMI, current smoking, education, annual household income, age at menarche, and two items from the Perceived Stress Scale (PSS-10), women in the highest quartile of diethylthiophosphate (DETP; an OP metabolite) had significantly longer TTP [adjusted $\text{FOR}=0.68$ (95% CI: 0.51, 0.92)] and increased infertility [adjusted $\text{OR}=2.17$ (95% CI: 1.19, 3.93)] compared with women in the lowest quartile. The highest versus lowest quartile of 3-phenoxybenzoic acid (3PBA; a PYR metabolite) was associated with longer TTP and infertility, with significant associations in nulliparous women [adjusted $\text{FOR}=0.72$ (95% CI: 0.53, 0.98); adjusted OR for $\text{infertility}=2.03$ (95% CI: 1.10, 3.74)].

Conclusion:

Our study provides some of the first evidence that preconception OP and PYR exposures are associated with decreased fertility in Chinese couples. Given that OPs and PYRs are rapidly metabolized in humans, more studies are needed to confirm our findings. https://doi.org/10.1289/EHP2987

Introduction

Growing evidence suggests that human fertility rates are declining in both developed and developing countries (Clementi et al. 2008). This reduced fertility has been assumed to be associated with socioeconomic changes and adverse lifestyle factors (Den Hond et al. 2015; Snijder et al. 2012). However, environmental contaminants such as pesticides have attracted international attention and have recently come to be regarded as possible contributors to reduced human fertility (Mehrpour et al. 2014; Smarr et al. 2016).

China is one of the largest agricultural countries in the world, with $>\mathrm{300,000}\text{tons}$ of agricultural pesticides used annually (Shu et al. 2016). Organophosphates (OPs) and pyrethroids (PYRs) are the most widely used groups of pesticides in agricultural and residential areas in China (Tan et al. 2006; Ye et al. 2015), with 70,000 tons of OPs used in the year of 2015 and 4,000 tons of PYRs used in the year of 2013, respectively (Shu et al. 2014, 2016). Environmental exposure to OPs and PYRs is thought to occur primarily via consumption of food contaminated with pesticide residues and via inhalation or ingestion of contaminated household dust after the application of pesticides indoors (Ding et al. 2015; Ji et al. 2011; Wang et al. 2012). Human exposure to OPs and PYRs is now widespread in many countries and has become a global health issue (Babina et al. 2012; CDC 2017; Imai et al. 2014; Martenies and Perry 2013; Mehrpour et al. 2014; Wang et al. 2012). Epidemiological studies conducted in China have demonstrated relatively high levels of exposure to OPs and PYRs in women, raising concerns about chronic exposures and potential effects on reproductive health (Ding et al. 2015; Ji et al. 2011; Liu et al. 2016; Wang et al. 2012; Xia et al. 2008).

Animal studies have found adverse effects of OP and PYR exposures on female reproductive functions, including inhibited steroid hormones, disordered estrous cycles, and restrained follicle cells, which might ultimately lead to decreased fertility (Fei et al. 2010; Geng et al. 2015; Guerra et al. 2011; Li et al. 2013; Okamura et al. 2009; Rao and Kaliwal 2002; Rastogi et al. 2014). Only a few cross-sectional studies have estimated associations between fertility and occupational pesticide exposures in women, and results have been inconsistent (Bretveld et al. 2006; Greenlee et al. 2003; Idrovo et al. 2005; Lauria et al. 2006). In a study of female flower workers in Colombia, self-reported occupational pesticide exposure was associated with a longer time to pregnancy (TTP) (Idrovo et al. 2005), but no significant association was found in a multicenter study in Italy based on a questionnaire survey (Lauria et al. 2006). None of these previous epidemiological studies assessed pesticide exposures by measuring pesticide metabolites in urine, and to our knowledge, no cohort study of pesticide exposures and infertility has yet been conducted. Therefore, we aimed to prospectively evaluate the associations between preconception pesticide exposure and couple fertility in Shanghai, China, where OPs and PYRs are the most widely used pesticides.

Methods

Study Population

To improve birth outcomes for couples who plan to become pregnant, the Chinese government promotes preconception care at designated clinics that provide health education and physical examinations (Zhang et al. 2016; Yang et al. 2015). The Shanghai Birth Cohort recruited women from two preconception care clinics in Shanghai, China. Detailed information on study recruitment has been described previously (Zhou et al. 2017). Briefly, women were eligible if they were registered Shanghai residents who were not planning to move in the next two years, were $\ge 20\text{years}$ of age, had stopped using contraception recently, and planned to conceive without assisted reproductive technology and to give birth in one of the hospitals participating in the Shanghai Birth Cohort. Women were not eligible for enrollment if they were part of a couple diagnosed with infertility, had tried and failed to conceive spontaneously for $>12\text{mo}$, or had sought reproductive assistance. After recruitment, telephone follow-up was performed every 2 mo for $\le 12\text{mo}$ to collect information on TTP as described below.

Between August 2013 and April 2015, 1,182 eligible women (referred to as the baseline group) were recruited at two preconception care clinics in Shanghai, China (Figure 1). To conserve biological samples for future research, OP and PYR metabolites were measured only in urine samples with a volume $>30\mathrm{mL}$, which were available for 615 women and were included in the TTP analysis (subgroup 1). Of these 615 women, nine women withdrew, and 110 women who used contraception intermittently and did not conceive before the end of the follow-up period were excluded, resulting in 496 women for the analysis of infertility (subgroup 2).

Figure 1.

Flowchart for participant selection.

The protocol was approved by the Medical Ethics Committee of Shanghai Xinhua Hospital, Shanghai Jiao Tong University School of Medicine. All study participants provided written informed consent.

Questionnaires

Trained research staff interviewed the couples separately using a standardized questionnaire, which included demographic characteristics, lifestyle characteristics, perceived stress, occupational pesticide exposure, and reproductive and medical histories. For occupational pesticide exposure, subjects were asked “What is your occupation (clerk, accountant, salesman, worker, farmer…)?” and “Do you produce or use pesticides at work? (Yes or no)”. For reproductive history, women were asked about their history of pregnancies and live births, and their spouses were asked about diseases of the genitourinary system (such as cryptorchidism, hypospadias, and varicocele). Perceived stress has been reported to be a risk factor for infertility (Louis et al. 2011; Lynch et al. 2014); therefore, women were asked to complete a self-administered Perceived Stress Scale (PSS-10) after enrollment. The validity and reliability of the PSS-10 have been reported elsewhere (Wang et al. 2011) (see Table S1 for questions included in the PSS-10).

Outcomes

TTP was defined as the number of months it took for a couple to conceive (Abell et al. 2000; Snijder et al. 2012) and was assessed during the bimonthly telephone follow-ups using the following question: “Have you become pregnant since the last telephone follow-up (or since the recruitment)?”. If the answer was yes, then “When was the first day of your last menstrual period (LMP)?”. If the answer was no, then “Did you take any contraceptive measures since the last telephone follow-up (or since the recruitment)?” and “Do you still plan to become pregnant?”. The couples were prospectively followed until they achieved pregnancy or for a maximum of 12 mo. In our study, TTP was determined based on the number of months of attempted pregnancy before enrollment (self-reported during a face-to-face interview) and the duration of telephone follow-up from enrollment to the LMP (or the end of follow-up). For women who withdrew from the study, TTP was censored upon dropout; for women who used contraception intermittently during follow-up, only the number of months without contraception were counted. In our study, self-reported pregnancies were all confirmed in the Shanghai Birth Cohort–participating hospitals.

Infertility was defined as having a TTP $>12\text{mo}$ (Vélez et al. 2015a). Based on the data from 12 mo of telephone follow-up, participants were divided into a pregnancy group and an infertility group.

OP and PYR Exposure

Preconception urine samples (spot urine) were collected from each participant at enrollment and then aliquoted to polypropylene tubes and stored at $-80°\mathrm{C}$ until further analysis. Six dialkylphosphate (DAP) metabolites of OPs [dimethylphosphate (DMP), dimethylthiophosphate (DMTP), diethylphosphate (DEP), diethylthiophosphate (DETP), dimethyldithiophosphate (DMDTP) and diethyldithiophosphate (DEDTP)] and three metabolites of PYRs [3-phenoxybenzoic acid (3PBA), trans/cis-3-(2,2-dichlorovinyl)-1-methylcyclopropane-1,2-dicarboxylic acid (TDCCA and CDCCA, respectively)] were examined using gas chromatography–mass spectrometry (GC-MS; OP metabolites were analyzed using a PE Clarus 600s/MSD, and PYR metabolites were analyzed using an Agilent 7890B/5977A GC/MSD) based on the modified method reported by Kühn et al. (1996) and Ueyama et al. (2010).

The limits of detection (LODs) of OP metabolites were $0.06\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for DEP and DETP, $0.09\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for DEDTP, $0.18\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for DMP, and $0.3\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for DMTP and DMDTP, respectively. The LODs of PYR metabolites were $0.05\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for 3PBA, $0.23\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for TDCCA, and $0.38\mathrm{\mu }\mathrm{g}/\mathrm{L}$ for CDCCA. Metabolite levels below the LOD were assigned a value equivalent to the $\text{LOD}/\sqrt{2}$ (Hornung and Reed 1990). The molar concentrations of DMP, DMTP, DEP, and DETP were summed to derive total DAPs as a summary measure of environmental OP exposures (Arcury et al. 2006).

Quality control samples prepared by pooling eight urine samples from healthy female adult volunteers were randomly assayed along with the study samples to ensure the quality of the analytical methods. Urine creatinine concentrations were measured using an automated chemistry analyzer (Hitachi 7100).

Statistical Analysis

Descriptive statistical analyses were performed for characteristics of the study population and concentrations of pesticide urinary metabolites. Demographic characteristics were compared between the baseline population and the study population and between women with and without detected pesticides using Student’s t-tests or analyses of variance (ANOVAs) for continuous variables and ${\mathrm{\chi }}^2$ tests for categorical variables. Spearman correlations (for continuous variables) and rank sum tests (for categorical variables) were used to examine associations between demographic characteristics and TTP; Student’s t-tests (for continuous variables) and ${\mathrm{\chi }}^2$ tests (for categorical variables) were used to explore associations between demographic characteristics and infertility. Urine concentrations of pesticide metabolites were adjusted for creatinine (to account for dilution) and were modeled as ${\log }_{10}\text{-transformed}$ continuous variables or categorized into quartiles, using the lowest quartile as the reference. Statistical analysis was limited to metabolites detected in $\ge 75\%$ of the population. Metabolites detected in $<75\%$ of the population (DMDTP, 13.5%; DEDTP, 29.2%; TDCCA, 60.9%; and CDCCA, 14.1%) were excluded from further analyses.

Fecundability odds ratios (FORs) were estimated using Cox models, where $\text{FORs}<1$ reflect a longer TTP or reduced fertility. In addition, we used logistic regression models to estimate odds ratios (ORs) for infertility (Vélez et al. 2015a).

Potential confounders were identified from previous studies of environmental pollutants on TTP or infertility (Greenlee et al. 2003; Lynch et al. 2014; Vélez et al. 2015b), and the following were included as covariates in all Cox and logistic regression models: maternal age (continuous), prepregnancy body mass index (BMI; continuous), current smoking [no (nonsmoking and no smoker in the home), live with smoker (nonsmoking but live with a smoker), yes (current smoking)], education ($<\text{bachelor}’\mathrm{s}\text{degree}$ or $\ge \text{bachelor}’\mathrm{s}\text{degree}$), annual household income ($<\mathrm{150,000}$, 150,000–300,000, or $>\mathrm{300,000}$ CNY), and age at menarche (continuous). In addition, we adjusted for two variables based on questions from the PSS-10 that were associated with TTP or infertility ($p<0.1$): “Have you found that you could not cope with all the things that you had to do?” and “How often have you been able to control irritations in your life?”, which were categorized as never or almost never, sometimes or fairly often, or very often. All covariates were defined based on information provided before conception.

We did not adjust for parity because of the potential for overadjustment bias (Vélez et al. 2015a). However, because parity is influenced by female fertility and is proof of former fertility (Bach et al. 2015; Vélez et al. 2015a, 2015b), we conducted sensitivity analyses restricted to nulliparous women only. In addition, Wald tests were performed to test for trends across quartiles of OP and PYR metabolite concentrations. All statistical analyses were performed using SPSS v.19 (IBM); $p<0.05$ (two-tailed) was considered statistically significant.

Results

In general, demographic characteristics were similar among the 1,182 women in the baseline group and women in subgroup 1 (analysis for TTP, $n=615$) and subgroup 2 (analysis for infertility, $n=496$) (see Table S2). Demographic characteristics were also similar between women with and without urine pesticide metabolite measurements (see Table S3).

Mean values [$\pm \text{standard}\text{deviation}$ (SD)] for maternal age, prepregnancy BMI, and age at menarche were $29.78\pm 3.82$, $21.17\pm 3.07$ and $13.37\pm 1.59$, respectively (Table 1). The majority of women (92.5%) were nulliparous, 81.0% had a bachelor’s degree or above, and 70.7% reported an annual household income $\ge \mathrm{150,000}$ CNY. None of the couples in our study reported producing or using pesticides at work. Only nine women (1.5%) were current smokers at enrollment, but 24.9% lived with smokers.

Table 1.

Characteristics of the study population at enrollment and their associations with time to pregnancy, pregnancy, and infertility.

Characteristic	Study population ($n=615$) $\text{Mean}\pm \text{SD}$ (range) or $n$ (%)	TTP (months) ($n=615$) Median (IQR)	p-Valuea	Pregnancy ($n=365$) $\text{Mean}\pm \text{SD}$ or $n$ (%)	Infertility ($n=131$) $\text{Mean}\pm \text{SD}$ or $n$ (%)	p-Valueb
Maternal age (years)	$29.78\pm 3.82$ (24.00–44.00)	—	0.023	$29.83\pm 3.08$	$30.17\pm 2.90$	0.271
Prepregnancy BMI (${\mathrm{kg}/\mathrm{m}}^2$)	$21.17\pm 3.07$ (14.45–37.65)	—	0.094	$21.24\pm 2.94$	$20.55\pm 2.40$	0.017
Age at menarche (years)	$13.37\pm 1.59$ (10.00–18.00)	—	0.323	$13.37\pm 1.74$	$13.45\pm 1.30$	0.660
Current smoking
No	453 (73.5)	4.5 (2.0–10.0)	0.035	293 (80.3)	96 (73.3)	0.204
Live with smoker	153 (24.9)	5.5 (2.0–12.0)		68 (18.6)	32 (24.4)
Yes	9 (1.5)	10.0 (6.5–12.0)		4 (1.1)	3 (2.3)
Education
$<\text{Bachelor’s degree}$	117 (19.0)	5 (2.0–9.5)	0.988	62 (17.0)	20 (15.3)	0.650
$\ge \text{Bachelor’s degree}$	498 (81.0)	5 (2.0–10.0)		303 (83.0)	111 (84.7)
Annual household income (CNY)
$<\mathrm{150,000}$	140 (22.8)	4.0 (2.0–10.0)	0.363	84 (23.0)	29 (22.1)	0.496
150,000–300,000	329 (53.5)	4.5 (1.5–10.0)		201 (55.1)	65 (49.6)
$>\mathrm{300,000}$	106 (17.2)	6.0 (2.0–12.0)		59 (16.1)	27 (20.6)
Refused to answer	40 (6.5)	6.0 (2.0–12.0)		21 (5.8)	10 (7.6)
Parity
0	569 (92.5)	5.0 (2.0–10.0)	0.062	334 (91.5)	121 (92.4)	0.759
$\ge 1$	46 (7.5)	4.0 (2.0–7.0)		31 (8.5)	10 (7.6)
Have you found that you could not cope with all the things that you had to do?c
Never or almost never	267 (43.4)	5.0 (2.0–9.0)	0.239	175 (48.0)	43 (32.8)	0.013
Sometimes	333 (54.2)	5.5 (2.0–12.0)		183 (50.1)	84 (64.1)
Fairly often or very often	15 (2.4)	3.0 (1.0–12.0)		7 (1.9)	4 (3.1)
How often have you been able to control irritations in your life?c
Never or almost never	27 (4.4)	4.0 (2.0–8.0)	0.078	17 (4.7)	5 (3.8)	0.023
Sometimes	168 (27.3)	4.0 (1.5–8.0)		104 (28.5)	22 (16.8)
Fairly often or very often	420 (68.3)	5.5 (2.0–12.0)		244 (66.8)	104 (79.4)

Note: —, data not available; BMI, body mass index; IQR, interquartile range; SD, standard deviation; TTP, time to pregnancy.

^ap-Values for difference in time to pregnancy according to characteristics at enrollment; Spearman correlation for continuous variables, rank sum test for categorical variables.

^bp-Values for differences in characteristics at enrollment according to pregnancy or infertile status at the end of follow-up; t test for continuous variables, chi-square test for categorical variables.

^cQuestions from the PSS-10 questionnaire.

At the end of follow-up, 131 (26.4%) women who did not get pregnant within a year were classified as infertile, and 365 (73.6%) conceived spontaneously during follow-up, with an average TTP of 4.5 mo. Maternal age and current smoking were significant predictors of TTP, and prepregnancy BMI was inversely associated with infertility (Table 1). In addition, compared with fertile women, infertile women were less likely respond “never or almost never” to “Have you found that you could not cope with all the things that you had to do?” and less likely to respond “fairly often or very often” to “How often have you been able to control irritations in your life?”.

Individual OP metabolites were detected in $>90\%$ of samples, with the exceptions of DMDTP (14.1%) and DEDTP (28.0%) (Table 2). Detection rates for PYR metabolites varied from 99.0% for 3PBA to 13.8% for CDCCA. Creatinine-adjusted median concentrations of urinary DMP, DMTP, DEP, DETP, 3PBA, and TDCCA were 6.99, 2.79, 10.25, 3.47, 0.73, and $0.44\mathrm{\mu }\mathrm{g}/\mathrm{g}$ creatinine, respectively. Median concentrations of DMDTP, DEDTP, and CDCCA were not calculated owing to their low detection rates.

Table 2.

Urinary concentrations of organophosphate (OP) and pyrethroid (PYR) metabolites ($n=615$).

Pesticides	Metabolites	Detection rate	Adjustment	Percentile
		$n$ (%)		25th	50th	75th	95th
OPs	DMP	579 (94.1)	Unadjusteda	2.03	4.83	11.31	45.43
			Adjustedb	2.99	6.99	16.14	53.83
	DMTP	572 (93.0)	Unadjusteda	0.75	1.77	4.67	21.47
			Adjustedb	1.25	2.79	5.97	22.14
	DEP	614 (99.8)	Unadjusteda	3.99	7.22	13.60	31.54
			Adjustedb	5.86	10.25	16.63	35.59
	DETP	615 (100)	Unadjusteda	1.25	2.54	5.13	16.34
			Adjustedb	2.16	3.47	6.43	15.48
	DMDTP	87 (14.1)	Unadjusteda	$<\text{LOD}$	$<\text{LOD}$	$<\text{LOD}$	0.60
			Adjustedb	$<\text{LOD}$	$<\text{LOD}$	$<\text{LOD}$	1.18
	DEDTP	172 (28.0)	Unadjusteda	$<\text{LOD}$	$<\text{LOD}$	0.28	0.65
			Adjustedb	$<\text{LOD}$	$<\text{LOD}$	0.31	1.40
	Total DAPc			108.50	198.43	335.51	766.22
PYRs	3PBA	609 (99.0)	Unadjusteda	0.29	0.51	1.04	3.14
			Adjustedb	0.44	0.73	1.31	3.99
	TDCCA	368 (59.8)	Unadjusteda	$<\text{LOD}$	0.29	0.59	2.22
			Adjustedb	$<\text{LOD}$	0.44	0.82	3.35
	CDCCA	85 (13.8)	Unadjusteda	$<\text{LOD}$	$<\text{LOD}$	$<\text{LOD}$	0.94
			Adjustedb	$<\text{LOD}$	$<\text{LOD}$	$<\text{LOD}$	1.82

Note: 3PBA, 3-phenoxybenzoic acid; CDCCA, cis-3-(2,2-dichlorovinyl)-1-methylcyclopropane-1,2-dicarboxylic acid; DEDTP, diethyldithiophosphate; DEP, diethylphosphate; DETP, diethylthiophosphate; DMDTP, dimethyldithiophosphate; DMP, dimethylphosphate; DMTP, dimethylthiophosphate; LOD, limit of detection; OP, organophosphate; PYR, pyrethroid; TDCCA, trans-3-(2,2-dichlorovinyl)-1-methylcyclopropane-1,2-dicarboxylic acid.

^aNot adjusted for creatinine ($\mathrm{\mu }\mathrm{g}/\mathrm{L}$).

^bAdjusted for creatinine ($\mathrm{\mu }\mathrm{g}/\mathrm{g}$ creatinine).

^cTotal DAP was the summary of molar concentrations of DMP, DMTP, DEP, and DETP ($\text{nmol}/\mathrm{g}$ creatinine).

Compared with women in the lowest quartile of DETP, women in the highest quartile of DETP had significantly lower odds of pregnancy [adjusted $\text{FOR}=0.68$ (95% CI: 0.51, 0.92); $p=0.012$], with a similar estimate when limited to nulliparous women [adjusted $\text{FOR}=0.67$ (95% CI: 0.50, 0.91); $p=0.011$] (Table 3). TTP was not significantly associated with other urinary OP metabolites. Women in the highest quartile of 3PBA had lower odds of pregnancy than women in the lowest quartile [adjusted $\text{FOR}=0.77$ (95% CI: 0.57, 1.03); $p=0.077$], with a slightly stronger association when restricted to nulliparous women [adjusted $\text{FOR}=0.72$ (95% CI: 0.53, 0.98); $p=0.034$].

Table 3.

Fecundability odds ratios and 95% confidence intervals for preconception organophosphate and pyrethroid exposure ($n=615$).

Exposure ($\mathrm{\mu }\mathrm{g}/\mathrm{g}$ creatinine)	FOR (95% CI)
	All women ($n=615$)		Nulliparous women ($n=569$)
	Unadjusted	Adjustedb	Unadjusted	Adjustedb
OPs
DMP
Continuousa	0.97 (0.82, 1.15)	0.97 (0.82, 1.15)	0.97 (0.82, 1.15)	0.99 (0.83, 1.17)
Q1 ($\le 2.99$)	Reference	Reference	Reference	Reference
Q2 (2.99–6.99)	1.04 (0.78, 1.38)	1.03 (0.77, 1.37)	1.04 (0.78, 1.40)	1.02 (0.76, 1.38)
Q3 (6.99–16.14)	0.98 (0.73, 1.30)	0.93 (0.69, 1.24)	0.97 (0.72, 1.30)	0.91 (0.68, 1.24)
Q4 ($>16.14$)	0.88 (0.66, 1.18)	0.85 (0.63, 1.15)	0.90 (0.66, 1.29)	0.86 (0.64, 1.17)
p for trend	0.955	0.883	0.958	0.836
DMTP
Continuousa	1.09 (0.89, 1.33)	1.06 (0.86, 1.29)	1.08 (0.88, 1.33)	1.07 (0.97, 1.32)
Q1 ($\le 1.25$)	Reference	Reference	Reference	Reference
Q2 (1.25–2.79)	0.87 (0.65, 1.17)	0.91 (0.67, 1.23)	0.88 (0.65, 1.20)	0.90 (0.66, 1.23)
Q3 (2.79–5.97)	0.97 (0.73, 1.30)	0.97 (0.73, 1.30)	0.98 (0.72, 1.32)	0.97 (0.71, 1.31)
Q4 ($>5.97$)	1.12 (0.85, 1.49)	1.15 (0.86, 1.53)	1.17 (0.87, 1.32)	1.19 (0.89, 1.60)
p for trend	0.589	0.734	0.670	0.717
DEP
Continuousa	0.90 (0.68, 1.19)	0.90 (0.67, 1.20)	0.89 (0.67, 1.20)	0.91 (0.67, 1.23)
Q1 ($\le 5.86$)	Reference	Reference	Reference	Reference
Q2 (5.86–10.25)	1.04 (0.78, 1.40)	1.05 (0.78, 1.41)	1.06 (0.78, 1.43)	1.02 (0.75, 1.38)
Q3 (10.25–16.63)	1.11 (0.83, 1.47)	1.12 (0.84, 1.50)	1.08 (0.80, 1.46)	1.08 (0.80, 1.47)
Q4 ($>16.63$)	1.05 (0.78, 1.40)	1.08 (0.81, 1.45)	1.03 (0.76, 1.39)	1.04 (0.77, 1.41)
p for trend	0.635	0.601	0.997	0.778
DETP
Continuousa	1.07 (0.81, 1.41)	1.09 (0.82, 1.45)	1.07 (0.80, 1.43)	1.10 (0.82, 1.48)
Q1 ($\le 2.16$)	Reference	Reference	Reference	Reference
Q2 (2.16–3.47)	0.88 (0.66, 1.16)	0.85 (0.64, 1.13)	0.88 (0.66, 1.17)	0.83 (0.61, 1.11)
Q3 (3.47–6.43)	0.82 (0.62, 1.10)	0.84 (0.63, 1.12)	0.77 (0.57, 1.04)	0.78 (0.58, 1.06)
Q4 ($>6.43$)	0.71 (0.53, 0.95)*	0.68 (0.51, 0.92)*	0.71 (0.52, 0.96)*	0.67 (0.50, 0.91)*
p for trend	0.235	0.153	0.231	0.143
Total DAP
Continuousa	0.93 (0.69, 1.27)	0.94 (0.70, 1.26)	0.92 (0.68, 1.25)	0.95 (0.70, 1.28)
Q1 ($\le 108.5$)	Reference	Reference	Reference	Reference
Q2 (108.5–198.43)	1.06 (0.79, 1.42)	1.07 (0.80, 1.43)	1.06 (0.78, 1.43)	1.04 (0.77, 1.41)
Q3 (198.43–335.51)	0.96 (0.72, 1.29)	0.79 (0.58, 1.08)	0.94 (0.70, 1.28)	0.87 (0.64, 1.19)
Q4 ($>335.51$)	1.08 (0.81, 1.44)	0.90 (0.67, 1.21)	1.07 (0.79, 1.44)	1.03 (0.76, 1.39)
p for trend	0.874	0.988	0.914	0.873
PYR
3PBA
Continuousa	1.14 (0.88, 1.48)	1.19 (0.91, 1.54)	1.19 (0.91, 1.56)	1.23 (0.93, 1.61)
Q1 ($\le 0.44$)	Reference	Reference	Reference	Reference
Q2 (0.44–0.73)	0.86 (0.65, 1.14)	0.83 (0.62, 1.11)	0.82 (0.61, 1.11)	0.80 (0.60, 1.08)
Q3 (0.73–1.31)	0.88 (0.66, 1.17)	0.85 (0.63, 1.13)	0.84 (0.63, 1.12)	0.83 (0.62, 1.11)
Q4 ($>1.31$)	0.75 (0.56, 1.00)	0.77 (0.57, 1.03)	0.74 (0.54, 0.99)*	0.72 (0.53, 0.98)*
p for trend	0.216	0.160	0.121	0.089

Note: 3PBA, 3-phenoxybenzoic acid; CI, confidence interval; DAP, dialkyl phosphate; DEP, diethylphosphate; DETP, diethylthiophosphate; DMP, dimethylphosphate; DMTP, dimethylthiophosphate; FOR, fecundability odds ratio; OP, organophosphate; PYR, pyrethroid; Q1–4, quartiles 1–4.

^*$p<0.05$.

^aUrine concentrations of pesticide metabolites were adjusted for creatinine then ${\log }_{10}\text{-transformed}$.

^bAdjusted for age, prepregnancy body mass index (BMI), age at menarche, current smoking, education, annual household income, “Have you found that you could not cope with all the things that you had to do?”, and “How often have you been able to control irritations in your life?” in Cox models.

When compared with women in the lowest quartile of DETP, women in the highest quartile of DETP had higher odds of infertility [adjusted $\text{OR}=2.17$ (95% CI: 1.19, 3.93); $p=0.011$], with a slightly stronger association in nulliparous women [adjusted $\text{OR}=2.30$ (95% CI: 1.23, 4.30); $p=0.009$] (Table 4). Other urinary OP metabolites were not clearly associated with infertility. The OR for infertility was positive but nonsignificant for women in the highest versus lowest quartile of 3PBA [adjusted $\text{OR}=1.55$ (95% CI: 0.86, 2.79); $p=0.148$]; however, the association was stronger and significant when limited to nulliparous women [adjusted $\text{OR}=2.03$ (95% CI: 1.10, 3.74); $p=0.023$].

Table 4.

Odds ratios and 95% confidence intervals for preconception organophosphate and pyrethroid exposure ($n=496$).

Exposure ($\mathrm{\mu }\mathrm{g}/\mathrm{g}$ creatinine)	OR (95% CI)
	All women ($n=496$)				Nulliparous women ($n=455$)
	Pregnancy	Infertility	Unadjusted	Adjustedc	Pregnancy	Infertility	Unadjusted	Adjustedc
OPs
DMP
Continuousa			1.04 (0.75, 1.44)	1.03 (0.74, 1.44)			1.02 (0.72, 1.43)	0.98 (0.69, 1.38)
Q1 ($\le 2.99$)b	96	29	Reference	Reference	89	27	Reference	Reference
Q2 (2.99–6.74)b	84	39	0.92 (0.51, 1.64)	0.94 (0.51, 1.72)	72	36	0.90 (0.50, 1.64)	0.97 (0.52, 1.81)
Q3 (6.74–16.17)b	91	32	1.41 (0.81, 2.46)	1.60 (0.90, 2.84)	84	28	1.48 (0.83, 2.64)	1.74 (0.96, 3.17)
Q4 ($>16.17$) b	94	31	1.07 (0.60, 1.89)	1.13 (0.63, 2.05)	89	30	0.99 (0.55, 1.79)	1.07 (0.58, 1.98)
p for trend			0.710	0.483			0.706	0.415
DMTP
Continuousa			0.81 (0.54, 1.23)	0.83 (0.54, 1.26)			0.77 (0.50, 1.19)	0.79 (0.51, 1.24)
Q1 ($\le 1.31$)b	89	34	Reference	Reference	77	32	Reference	Reference
Q2 (1.31–2.85)b	86	38	1.28 (0.73, 2.28)	1.30 (0.72, 2.35)	82	35	1.40 (0.76, 2.56)	1.42 (0.76, 2.66)
Q3 (2.85–5.92)b	95	29	1.29 (0.74, 2.26)	1.32 (0.74, 2.34)	91	29	1.43 (0.79, 2.60)	1.49 (0.81, 2.77)
Q4 ($>5.92$)b	95	30	0.95 (0.53, 1.70)	0.91 (0.49, 1.66)	84	25	1.07 (0.58, 1.97)	1.07 (0.57, 2.03)
p for trend			0.347	0.332			0.217	0.198
DEP
Continuousa			1.00 (0.56, 1.77)	0.94 (0.52, 1.68)			0.98 (0.53, 1.78)	0.91 (0.49, 1.68)
Q1 ($\le 5.86$)b	90	33	Reference	Reference	84	32	Reference	Reference
Q2 (5.86–10.12)b	92	33	1.01 (0.58, 1.78)	1.10 (0.61, 1.97)	85	30	1.09 (0.61, 1.95)	1.21 (0.66, 2.22)
Q3 (10.12–16.15)b	91	32	0.98 (0.56, 1.72)	1.00 (0.56, 1.77)	79	29	1.01 (0.56, 1.82)	1.04 (0.57, 1.91)
Q4 ($>16.15$) b	92	33	1.01 (0.58, 1.79)	1.02 (0.57, 1.82)	86	30	1.05 (0.58, 1.91)	1.07 (0.58, 1.98)
p for trend			0.996	0.778			0.815	0.590
DETP
Continuousa			0.74 (0.41, 1.32)	0.68 (0.37, 1.26)			0.72 (0.39, 1.32)	0.66 (0.35, 1.26)
Q1 ($\le 2.14$)b	91	33	Reference	Reference	84	32	Reference	Reference
Q2 (2.14–3.49)b	90	33	1.32 (0.73, 2.39)	1.44 (0.77, 2.66)	81	30	1.38 (0.75, 2.55)	1.52 (0.80, 2.88)
Q3 (3.49–6.28)b	91	32	1.53 (0.85, 2.77)	1.56 (0.84, 2.87)	83	29	1.37 (0.73, 2.56)	1.40 (0.73, 2.68)
Q4 ($>6.28$)b	93	33	2.07 (1.17, 3.68)*	2.17 (1.19, 3.93)*	86	30	2.11 (1.16, 3.85)*	2.30 (1.23, 4.30)*
p for trend			0.231	0.146			0.274	0.180
Total DAP
Continuousa			1.00 (0.55, 1.79)	0.91 (0.50, 1.66)			0.96 (0.52, 1.78)	0.85 (0.45, 1.61)
Q1 ($\le 113.42$)b	94	29	Reference	Reference	90	28	Reference	Reference
Q2 (113.42–200.42)b	83	40	0.82 (0.46, 1.45)	0.88 (0.49, 1.59)	71	39	0.79 (0.44, 1.42)	0.90 (0.49, 1.66)
Q3 (200.42–332.46)b	96	28	1.31 (0.76, 2.25)	1.64 (0.93, 2.91)	89	21	1.40 (0.80, 2.45)	1.89 (1.04, 3.43)*
Q4 ($>332.46$)b	92	34	0.77 (0.43, 1.38)	0.86 (0.48, 1.56)	84	33	0.60 (0.32, 1.12)	0.68 (0.36, 1.30)
p for trend			0.957	0.628			0.988	0.526
PYR
3PBA
Continuousa			0.78 (0.46, 1.32)	0.81 (0.47, 1.39)			0.81 (0.47, 1.40)	0.84 (0.48, 1.46)
Q1 ($\le 0.44$)b	92	34	Reference	Reference	86	28	Reference	Reference
Q2 (0.44–0.73)b	92	32	1.29 (0.73, 2.30)	1.22 (0.68, 2.21)	80	30	1.22 (0.66, 2.23)	1.17 (0.63, 2.19)
Q3 (0.73–1.38)b	86	37	1.27 (0.71, 2.27)	1.26 (0.69, 2.29)	76	37	1.40 (0.77, 2.56)	1.39 (0.75, 2.60)
Q4 ($>1.38$)b	95	28	1.56 (0.88, 2.76)	1.55 (0.86, 2.79)	92	26	1.94 (1.08, 3.50)*	2.03 (1.10, 3.74)*
p for trend			0.306	0.403			0.287	0.349

Note: 3PBA, 3-phenoxybenzoic acid; CI, confidence interval; DAP, dialkyl phosphate; DEP, diethylphosphate; DETP, diethylthiophosphate; DMP, dimethylphosphate; DMTP, dimethylthiophosphate; OP, organophosphate; OR, odds ratio; PYR, pyrethroid; Q1–4, quartiles 1–4.

^*$p<0.05$.

^aUrine concentrations of pesticide metabolites were adjusted for creatinine then ${\log }_{10}\text{-transformed}$.

^bThe cut point of OP and PYR metabolites was determined in a sample size of 496 in the analysis for infertility.

^cAdjusted for age, prepregnancy BMI, age at menarche, current smoking, education, annual household income, “Have you found that you could not cope with all the things that you had to do?”, and “How often have you been able to control irritations in your life?” in logistic regression models.

Discussion

Exposures to OPs and PYRs were associated with both TTP and infertility in our study population. Women in the highest quartiles of DETP (one of the major OP urinary metabolites) and 3PBA (the major PYR urinary metabolite) had longer TTP and increased odds of infertility compared with women in the lowest quartile of each exposure.

Our findings are supported by previous epidemiological studies conducted in Colombia (Idrovo et al. 2005), Denmark (Abell et al. 2000), Finland (Sallmén et al. 2003), Canada (Curtis et al. 1999), and France (Thonneau et al. 1999), suggesting that pesticide exposures have adverse effects on fertility. For example, in a cross-sectional study of 492 female flower greenhouse workers in Denmark, women who handled pesticide-treated cultures without wearing gloves had significantly lower fertility [$\text{FOR}=0.697$ (95% CI: 0.46, 0.98)] than women who always used gloves (Abell et al. 2000). Idrovo et al. (2005) studied 2,085 nulliparous agricultural workers in Colombia and reported that women who had been employed in the flower production industry for $\ge 2\mathrm{y}$ had a significantly longer TTP [$\text{FOR}=0.73$ (95% CI: 0.63, 0.84)] than other workers. However, two cross-sectional studies conducted in the Netherlands (Bretveld et al. 2006) and Italy (Lauria et al. 2006) found no associations between pesticide exposures and TTP. It is noteworthy that published studies of the association between pesticide exposures and TTP or fertility have primarily focused on women with occupational exposures or women living in agricultural regions and have classified exposures based on self-reported responses to postal questionnaires or telephone interviews (Abell et al. 2000; Bretveld et al. 2006; Curtis et al. 1999; Idrovo et al. 2005; Lauria et al. 2006; Sallmén et al. 2003; Thonneau et al. 1999). In contrast with previous studies, we measured pesticide metabolites in urine to quantitatively estimate environmental pesticide exposures in women before conception.

In the present study, we found that both DETP and 3PBA were significantly associated with prolonged TTP and infertility, suggesting that exposures to OPs and PYRs have adverse effects on fertility. In addition, when the analysis was limited to nulliparous women, associations for DETP and 3PBA with prolonged TTP and infertility were somewhat stronger, which suggests that parity may have modified the associations between pesticide exposures and fertility, a finding that should be confirmed in future studies.

The prevalence of infertility in our sample was 26.4%, which is higher than the infertility rate in China [15% in 2009 (Ma 2017)]. Several factors may contribute to these differences. First, according to the committee opinion expressed by the American College of Obstetricians and Gynecologists (ACOG) and the American Society for Reproductive Medicine (ASRM), women’s fertility decreases gradually but significantly beginning at approximately 32 y of age (American College of Obstetricians and Gynecologists Committee on Gynecologic Practice and Practice Committee 2014), and 25% of the women in our study were $\ge 32\text{years}$ old. Hou (2011) explored potential risk factors for female infertility in a cross-sectional study of 1,606 women registered with an infertility clinic in Shandong province and reported that the OR for infertility was 2.14 (95% CI: 1.72, 2.65) for women with a bachelor’s degree or higher compared with women who had a middle school education or less, and the OR for infertility was 1.31 (95% CI: 1.04, 1.65) for urban versus rural populations. Because our study population was composed of better-educated urban women, the infertility rate is expected to be higher than the population average. In the Longitudinal Investigation of Fertility and the Environment (LIFE) study of 501 U.S. couples without diagnosed infertility who were seeking to become pregnant, which also enrolled couples before conception and followed up for 12 menstrual cycles, 31% did not achieve pregnancy during the follow-up period (Buck Louis et al. 2014). Women in the LIFE study were highly educated (95% with a college or technical education) and had a mean age of 30 y ($\pm 4$) at enrollment.

Potential biological mechanisms that might explain associations of lower fertility with OP and PYR exposures are unclear. Findings from several animal studies suggest that OP exposures could cause disturbances in estrous cycles, resulting in a decrease in the number of estrous cycles and in the duration of proestrus, estrus, and metestrous with a concomitant prolonged diestrous phase, which may ultimately result in reduced fertility (Nanda and Kaliwal 2003; Rao and Kaliwal 2002; Tello et al. 2013). Another possible mechanism might be through effects on follicular development (Ghodageri and Katti 2013; Mahadevaswami and Kaliwal 2002; Nair et al. 2014; Katti et al. 2012). Animal studies have revealed that PYR exposures can inhibit steroid hormones, restrain the growth of follicles, and damage ovarian corpus luteum cells, which might further contribute to decreased fertility (Fei et al. 2010; Guerra et al. 2011; He et al. 2006; Sangha et al. 2013). However, although findings from animal studies provide supportive evidence that exposure to OPs and PYRs may affect fertility in humans, more research is warranted.

Our study has several strengths. To our knowledge, it is the first prospective study of preconceptional pesticide exposures and TTP and couple fertility in China with quantitative measures of exposures to the most widely used groups of pesticides, OPs and PYRs. In addition, China is one of the few countries to offer free preconception care to any couple who wants to become pregnant, which enabled us to recruit women before conception (Zhang et al. 2016). Unlike women enrolled from infertility clinics, women in our study were recruited from the general population, and none reported occupational exposure to pesticides. Thus, our findings may be more generalizable than those of previous studies. Furthermore, we prospectively ascertained TTP through telephone follow-up, which avoided recall bias, a concern for retrospective studies. Moreover, although many previous studies enrolled pregnant women, we reduced the potential for selection bias by enrolling women from the general population before conception (Tingen et al. 2004). Finally, instead of classifying pesticide exposure based on self-report (Abell et al. 2000; Idrovo et al. 2005; Lauria et al. 2006), we measured urinary concentrations of OPs and PYRs.

Several limitations also need to be acknowledged. First, we did not measure pesticide exposures in the women’s male partners; thus, we were unable to estimate quantitative associations between spousal pesticide exposures and fertility. None of the male participants reported a history of diseases affecting the reproductive system (such as syphilis, cryptorchidism, hypospadias, or varicocele), which should reduce the potential contribution of the male factor to infertility, although misclassification, particularly underreporting of disease history, remains a concern. Second, compared with a large cross-sectional investigation of a general population sample of women in Shanghai ($n=\mathrm{7,310}$) (Xie et al. 2014), where 68% had an annual household income $\ge \mathrm{100,000}$ CNY and 60% had a bachelor’s degree or above, our study population had higher annual household income (87% had annual household income $\ge \mathrm{100,000}$ CNY) and higher educational backgrounds (81% had a bachelor’s degree or above). Third, OP and PYR metabolites that were measured in a single spot urine sample collected upon enrollment might not represent exposure levels during the entire preconception period (Koureas et al. 2012).

Conclusion

In summary, findings from our prospective cohort study suggest that preconceptional exposures to environmental pesticides may adversely affect TTP and couple fertility in Shanghai, China. OP and PYR exposures that were quantified based on urinary DETP and 3PBA concentrations were associated with longer TTP and increased odds of infertility. As pesticide use becomes more widespread in China and in other parts of the world, adverse health effects have emerged as a major public health concern. Further studies are needed to confirm associations between pesticide exposures and couple infertility in other populations and to evaluate potential mechanisms that might be responsible for associations between infertility and the widely used OP and PYR pesticides.