Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Apr 1.
Published in final edited form as: Environ Res. 2017 Jan 3;154:120–125. doi: 10.1016/j.envres.2016.12.030

Low-level arsenic exposure via drinking water consumption and female fecundity - a preliminary investigation

Michele L Susko a, Michael S Bloom a,b,*, Iulia A Neamtiu c,d, Allison A Appleton a, Simona Surdu b, Cristian Pop e,g, Edward F Fitzgerald a,b, Doru Anastasiu h,i, Eugen S Gurzau c,d,j
PMCID: PMC5328919  NIHMSID: NIHMS840792  PMID: 28061371

Abstract

High level arsenic exposure is associated with reproductive toxicity in experimental and observational studies; however, few data exist to assess risks at low levels. Even less data are available to evaluate the impact of low level arsenic exposure on human fecundity. Our aim in this pilot study was a preliminary evaluation of associations between low level drinking water arsenic contamination and female fecundity. This retrospective study was conducted among women previously recruited to a hospital-based case-control study of spontaneous pregnancy loss in Timiş County, Romania. Women (n = 94) with planned pregnancies of 5–20 weeks gestation completed a comprehensive physician-administered study questionnaire and reported the number of menstrual cycles attempting to conceive as the time to pregnancy (TTP). Drinking water samples were collected from residential drinking water sources and we determined arsenic levels using hydride generation-atomic absorption spectrometry (HG-AAS). Multivariable Cox-proportional hazards regression with Efron approximation was employed to evaluate TTP as a function of drinking water arsenic concentrations among planned pregnancies, adjusted for covariates. There was no main effect for drinking water arsenic exposure, yet the conditional probability for pregnancy was modestly lower among arsenic exposed women with longer TTPs, relative to women with shorter TTPs, and relative to unexposed women For example, 1 μg/L average drinking water arsenic conferred 5%, 8%, and 10% lower likelihoods for pregnancy in the 6th, 9th, and 12th cycles, respectively (P = 0.01). While preliminary, our results suggest that low level arsenic contamination in residential drinking water sources may further impair fecundity among women with longer waiting times; however, this hypothesis requires confirmation by a future, more definitive study.

Keywords: arsenic, fecundity, pregnancy, Romania, reproduction, water

1. Introduction

Chronic arsenic exposure is linked to myriad human health outcomes, affecting virtually every biologic system, including reproduction (ATSDR, 2007; Naujokas et al., 2013). Millions worldwide are exposed to arsenic contaminated drinking water exceeding 10 μg/L, the maximum contaminant level (MCL) set by the World Health Organization (WHO) (Smedley and Kinniburgh, 2002; WHO, 2006). Arsenic levels exceeding 10 μg/L or even 50 μg/L and higher, have been found in groundwater aquifers in Argentina, Bangladesh, Chile, China, Hungary, India, Mexico, and the U.S. among other locations (Smedley and Kinniburgh, 2002; Welch et al., 2000). At these levels of exposure, drinking water arsenic contamination has been linked to spontaneous abortion, stillbirth, lower birth weight, and smaller birth size (Bloom et al., 2010; Bloom et al., 2014b; Quansah et al., 2015). However, many reproductive-aged women worldwide are also likely to be exposed to drinking water arsenic contamination below 10 μg/L (van Halem et al., 2009; Welch et al., 2000), low levels according to the U.S. National Research Council (NRC, 2001). Given its impact on pregnancy outcomes at high doses and a growing concern for adverse non-cancer health effects at low doses (Carlin et al., 2016), drinking water arsenic contamination less than 10 μg/L may also impact women’s fecundity; however, virtually no data exist to characterize the effect.

To begin to address this data gap, we conducted a pilot study of low level arsenic exposure through residential drinking water consumption and female fecundity, defined as the biologic capacity to reproduce, and measured by waiting time until a recognized pregnancy was conceived (Larsen, 2005). Our study population comprised women residing in Timiş County, Romania, an area that falls within the Great Hungarian Plain, which is recognized for groundwater sources naturally contaminated by arsenic, and which are frequently used for private and municipal supplies (Neamtiu et al., 2015; Rowland et al., 2011). These preliminary data will be used to further our knowledge and to guide the design of a future, more comprehensive study to characterize the reproductive toxicity of low level drinking water arsenic contamination in a more definitive fashion.

2. Materials and Methods

2.1 Study population

Our study population comprised pregnant women, of 5–20 weeks completed gestation, originally recruited to a case-control investigation of spontaneous pregnancy loss in Timiş County, Romania (Bloom et al., 2014a). Prenatal care is compulsory and provided to women at no cost in Romania. This includes registering the pregnancy in early gestation (typically within one month of recognition) and follow-up evaluations conducted during the 2nd and 3rd trimesters. We estimate compliance rates at approximately 85%. We enrolled clinically-confirmed spontaneous abortions (n = 150), of not more than 20 weeks completed gestation, as cases, between December 2011 and January 2013. Women with ongoing pregnancies of similar duration to each case (within 1 week) were recruited as control participants (n = 150). All women received care at the Obstetrics and Gynecology Department of the County Emergency Hospital (Bega Hospital), in the city of Timişoara, obstetrical provider to the vast majority of Timiş County residents. Once enrolled, participants completed a physician administered questionnaire to query demographic, socioeconomic, and lifestyle factors, medical, gynecologic, and occupational histories, as well as residential and drinking water consumption histories. We excluded n = 2 planned pregnancies with missing ‘time to pregnancy’ data (TTP), n = 8 pregnancies conceived using fertility treatments, and n = 3 ‘duplicate’ pregnancies (i.e., women from the parent case-control study, who had first participated as a control and later participated as a case). Women reported planned (n = 94) or unplanned (n = 193) pregnancies, and pregnancy planners reported the number of menstrual cycles of intercourse without contraception that it took to conceive. The self-reported number of cycles trying to conceive, or TTP, is a validated and frequently used measure of couple-based fecundity (Joffe, 1997). Clinical data was abstracted from a medical intake form and we collected blood and urine specimens for use in a future analysis. All participants provided informed consent prior to enrolling in the study and the study protocol was approved by the Institutional Review Boards of the County Emergency Hospital in Timişoara and the University at Albany, State University of New York.

2.2 Exposure assessment

Our trained study staff collected drinking water samples from up to two residential sources reported by study participants, as previously described in detail (Bloom et al., 2014a). Briefly, we collected a 50 mL water sample into screw-top containers previously demineralized with nitric acid (HNO3) and then rinsed. Concentrated analytical grade HNO3 (100 μL) was added as a preservative immediately following collection. The water samples were stored on ice and delivered to the Environmental Health Center (Cluj-Napoca, Romania) within 72 hours, for analysis. In the lab, 25 mL of sample water was mixed with 10 mL HCl, and 2 mL of aqueous 5% KI and 5% ascorbic acid (m/m). The sample was heated gently to 50 ºC to reduce As (V) to As (III), and then cooled at room temperature after 15 minutes. Following transfer to a 50 mL volumetric flask, the samples were diluted with de-ionized water to volume. NaBH4 (0.3%) dissolved in NaOH (0.1%) and HCL (3%) (m/v) was added to the sample, which was transported to a quartz cell, heated to 960 ºC to generate hydrides, and analyzed using a Zeenit 700p atomic absorption spectrometer (Analytikjena, Jena, Germany). All laboratory reagents were purchased from Merck (Darmstadt, Germany), Sigma-Aldrich (Steinheim, Germany), and Chem Lab (Zedelgem, Belgium). The method detection limit (MDL) was 0.5 μg/L. Calibration standards and quality control procedures were previously described in detail (Bloom et al., 2014a). We censored negative values to zero for characterizing the exposure distributions and reporting summary measures; however, we did not impute values below the MDL during regression analysis, but rather, we used instrument-reported values to avoid introducing bias (Richardson and Ciampi, 2003; Schisterman et al., 2006).

We operationalized three arsenic exposure variables, to maximize sensitivity for detecting modest associations. Average arsenic level (μg/L) was defined as the arithmetic mean level measured in up to two residential drinking water sources. Peak arsenic level (μg/L) was defined as the highest arsenic level measured in up to two residential drinking water sources. Daily arsenic dose (μg/day) was calculated as the product of average arsenic level multiplied by total average reported daily residential drinking water consumption from non-bottled sources during pregnancy, including teas, coffees, other water-mixed beverages, and soups. For women who resided at the study address for the entire duration of follow-up (i.e., time trying to conceive prior to the pregnancy), we calculated “duration of exposure” by subtracting the week of gestation at study enrollment from the reported TTP in month equivalents (i.e., 1 cycle = 1 month), to target pre-conception exposure. For n = 4 women with incomplete follow-up (e.g., moving to the study residence in the midst of trying to conceive prior to the pregnancy), we left censored “duration of exposure” by subtracting the week of gestation at study enrollment from the date relocating to the study address.

2.3 Statistical analysis

2.3.1 Univariate and bivariate analysis

We examined distributions and assessed potential outliers for arsenic exposure variables and covariates. We used Chi-square and Wilcoxon signed rank tests to compare the distribution of demographic factors and reproductive histories, as well as residential drinking water arsenic exposure for planned pregnancies to unplanned pregnancies. TTP was defined as the number of self-reported menstrual cycles trying to conceive prior to a recognized pregnancy. For planned pregnancies, we used Kaplan-Meier survival curves with log-rank test to evaluate unadjusted associations with TTP, individually for each arsenic exposure variable, and for all covariates. We graphed the log of the cumulative hazard function by log TTP [(log (−log) S(t)) vs. (log TTP)] to visually assess the tenability of proportional hazards assumptions for arsenic exposure variables and confounding covariates (Collett, 1994).

2.3.2 Multivariable analysis

We used Cox proportional hazards regression models to evaluate associations between arsenic exposure variables and TTP among pregnancy planners (n = 94), adjusted for confounding covariates (Collett, 1994). The discrete-time analog to the Cox proportional hazards model generated unstable effect estimates, and so we implemented an Efron approximation for tied event times with traditional Cox-proportional hazards modeling. The Efron approximation more accurately estimates regression coefficients than other methods when ties are frequent but the sample size is limited (Hertz-Picciotto and Rockhill, 1997). In a first step, we used regression models, incorporating cross-product terms between log TTP and arsenic exposure variables or covariates, to further assess violations of the proportional hazards assumption. In a second step, we simultaneously included arsenic exposure, confounding covariates, and all cross-product terms having P < 0.10 in the first step, using a backwards stepwise elimination procedure, and retaining in final models only cross-product terms maintaining P < 0.10. We constructed separate models using average arsenic level, peak arsenic level, and daily arsenic dose as the predictor of interest. Exponentiation of regression coefficients provides fecundability hazards ratios (FHRs), which indicate a higher conditional likelihood for pregnancy when > 1.0 and a lower conditional likelihood for pregnancy when < 1.0.

We used directed acyclic graphs (DAGs) to select self-reported covariates for inclusion in regression models as confounders (Greenland et al., 1999). We identified relevant variables for inclusion in DAGs as those predictive of arsenic exposure and female fecundity in the literature, including: maternal age in years (ACOG and ASRM, 2014; Howe et al., 1985); cigarette smoking, as having smoked at least 100 lifetime cigarettes (Howe et al., 1985; Hughes and Brennan, 1996; Jain, 2015; Leffondre et al., 2002); body mass index (BMI), as kg/m2 (Bolúmar et al., 2000; Grodstein et al., 1994; Yilmaz et al., 2009); urban residence, as a proxy for air pollution exposure (Nieuwenhuijsen et al., 2014; Popescu et al., 2011); education, as a proxy for socioeconomic status (no high school diploma (reference) vs. high school diploma/some college/technical school vs. university degree) (Huber et al., 2010); currently married or living as married (Huber et al., 2010); prior pregnancies and prior pregnancy losses (Axmon et al., 2006); and a history of pelvic inflammatory disease (PID) (Westrom et al., 1992). Based on the DAG we included maternal age, cigarette smoking, and education as confounders in multivariable models.

We used SAS v.9.3 (SAS Institute Inc., Cary NC) for the analysis. Statistical significance was defined as P < 0.05 for main effects and P < 0.10 for interactions, using two-tailed tests.

2.3.3 Sensitivity analyses

In a first sensitivity analysis we included women with ‘unplanned’ pregnancies (n = 193) as having conceived in the ‘zero cycle’ (i.e., TTP = 0), which addresses left truncation associated with unplanned pregnancies (Joffe et al., 2005). In a second sensitivity analysis, we excluded clinically infertile women (n = 13), those reporting TTP > 12 months (ASRM, 2013), as clinically infertile women may comprise a distinct population.

3. Results

Table 1 compares sociodemographic, clinical, and lifestyle factors between planned (n = 94) and unplanned pregnancies (n = 193) included in the current analysis. Among planners, n = 29 (30.9%) conceived in the 1st cycle, half conceived by the 3rd cycle, and the maximum TTP was 72 cycles. Planned and unplanned pregnancies were mostly similar, although the former were more likely to have married (P = 0.02) and to have completed university (P = 0.001). Planners also had a slightly higher BMI than non-planners (median = 22.21 vs. 21.20 kg/m2), although not statistically significant (P = 0.07). Table 1 also shows that arsenic exposure was similarly distributed for planned and unplanned pregnancies. Pregnancy planners were exposed to an average of 4.63 μg/L arsenic in residential drinking water sources (4.11 μg/L for non-planners), with a range of 0 to 175.10 μg/L; however, 90% of women were exposed to ≤ 8.74 μg/L on average. Still, n = 2 were exposed to water sources with an average arsenic concentration > 50 μg/L.

Table 1.

Demographic factors and drinking water arsenic exposures among planned pregnancies and unplanned pregnancies.

Factor Planned pregnancies Unplanned pregnancies P-value a
n Mean Median/% SD Min Max n Mean Median/% SD Min Max
Time to pregnancy (cycles) 94 8 3 13.5 1 72 - - - - - - -
Age (years) 94 29.57 28.96 5.01 18.50 42.17 193 28.78 28.08 6.28 18.08 44.25 0.20
BMI (kg/m2) 93 22.92 22.21 3.74 16.14 36.73 192 22.28 21.20 3.89 15.06 46.61 0.07
Urban residence 68 -- 72% -- -- -- 127 -- 66% -- -- -- 0.28
Married, living as married 85 -- 90% -- -- -- 152 -- 79% -- -- -- 0.02
Education
 University diploma 55 -- 59% -- -- -- 66 -- 34% -- -- -- 0.001
 High school diploma/some college/technical school 21 -- 22% -- -- -- 71 -- 37% -- -- -- --
 No high school diploma 18 -- 19% -- -- -- 56 -- 29% -- -- -- --
Smoked > 100 lifetime cigarettes 38 -- 40% -- -- -- 74 -- 38% -- -- -- 0.80
Reproductive history
 Prior pregnancies 60 -- 64% -- 0 6 109 -- 56% -- 0 14 0.79
 Prior pregnancy losses 48 -- 51% -- 0 3 74 -- 38% -- 0 8 0.09
 History of PID 16 -- 17% -- -- -- 30 -- 16% -- -- -- 0.86
Drinking water arsenic level
 Average (μg/L) 94 4.63 0.46 18.79 0 175.10 193 4.11 0.29 12.19 0 130.30 0.80
 Peak (μg/L) 94 5.55 1.11 19.04 0 175.10 193 4.59 0.53 12.44 0 130.30 0.56
 Daily dose (μg) 94 7.29 0.77 36.42 0 350.20 193 5.75 0.44 13.44 0 90.45 0.45
Open in a new tab

NOTE: Bold typeface indicates P < 0.05.

a

Wilcoxon rank sum tests test for continuous variables and chi-square test for categorical variables.

We conducted bivariate analysis for all potential confounding variables (i.e., those included in multivariate models based on DAGs and those not included in multvariate models) using Kaplan-Meier survival curves to evaluate unadjusted associations between TTP, arsenic exposure variables, and covariates, including age, education, cigarette smoking, BMI, urban vs. rural residence, marital status, prior pregnancies, prior pregnancy losses, and a history of PID (Figure S1). We categorized each arsenic exposure variable, age, and BMI as tertiles for this purpose. We detected a difference only for a history of prior pregnancy loss; TTP was shorter among those women with a prior loss relative to those without (P = 0.03).

We generated multivariable Cox proportional hazard models to estimate the effects of residential drinking water arsenic exposure on TTP, among (n = 94) women with planned pregnancies. We used (log (−log S(t)) vs. (log TTP) graphs in a preliminary unadjusted assessment of the proportional hazards assumption for confounding variables identified by our DAG, including age, education, and cigarette smoking (Figure S2). We further investigated proportional hazards assumption violations in the multivariable context by incorporating log TTP cross-product terms with arsenic exposure variables, maternal age, or education, adjusted for age, education, and smoking; the arsenic exposure (P = 0.002), age (P < 0.0001), and education (P < 0.0001) interactions were statistically significant. Because the proportional hazards assumption was violated in models assessing arsenic exposure, maternal age, and education, we used stepwise backwards elimination to simultaneously evaluate arsenic exposure, maternal age, and education x log TTP cross-product terms, retaining only cross-products for arsenic exposure (P ≤ 0.02) and maternal age (P < 0.0001) in the final Cox-proportional hazards regression models, adjusted for maternal age, cigarette smoking, and education. Including the cross-product terms allowed for the effects of arsenic and maternal age to vary over time. We created a separate model for each arsenic exposure variable - average arsenic level, peak arsenic level, and daily arsenic dose. In all cases, higher levels of arsenic were associated with a lower probability for pregnancy among women with longer TTPs (Table 2). For example, relative to the likelihood for a first cycle pregnancy in an unexposed woman (i.e. average drinking water arsenic = 0 μg/L), a woman with average drinking water arsenic = 1.0 μg/L had a slightly higher risk for pregnancy in the 1st cycle (FHR = 1.08), yet a 5% lower risk in the 6th cycle given the joint effects of arsenic and time (FHR = 0.95), an 8% lower risk in the 9th cycle given the joint effects of arsenic and time (FHR = 0.92), and a 10% lower risk in the 12th cycle given the joint effects of arsenic and time (FHR = 0.90). We detected a similar pattern for more advanced maternal age.

Table 2.

Fecundability hazard ratios (FHR) and 95% confidence intervals (CI) for time to pregnancy (TTP) associated with drinking water arsenic exposure (n = 94).

Predictor a FHR 95% CI P-value
Average arsenic level (μg/L) 1.08 1.02, 1.13 0.01
Maternal age (years) 1.34 1.25, 1.44 < 0.0001
High school diploma/some b college/technical school 1.17 0.56, 2.46 0.68
University diploma b 1.28 0.65, 2.51 0.47
Smoked > 100 lifetime cigarettes c 0.93 0.56, 1.55 0.78
Maternal age (years) x log TTP (cycles) 0.80 0.77, 0.84 < 0.0001
Average arsenic level (μg/L) x log TTP (cycles) 0.93 0.89, 0.98 0.01
Peak arsenic level (μg/L) 1.06 1.01, 1.11 0.02
Maternal age (years) 1.34 1.25, 1.44 < 0.0001
High school diploma/some college/technical school b 1.14 0.54, 2.41 0.74
University diploma b 1.35 0.68, 2.65 0.39
Smoked > 100 lifetime cigarettes c 0.97 0.58, 1.61 0.90
Maternal age (years) x log TTP (cycles) 0.81 0.77, 0.84 < 0.0001
Peak arsenic level (μg/L) x log TTP (cycles) 0.95 0.91, 0.99 0.02
Daily arsenic dose (μg) 1.06 1.01, 1.11 0.01
Maternal age (years) 1.34 1.24, 1.44 < 0.0001
High school diploma/some college/technical school b 1.12 0.53, 2.38 0.77
University diploma b 1.33 0.67, 2.63 0.42
Smoked > 100 lifetime cigarettes c 0.98 0.59, 1.63 0.95
Maternal age (years) x log TTP (cycles) 0.81 0.77, 0.84 < 0.0001
Daily arsenic dose (μg) x log TTP (cycles) 0.95 0.91, 0.99 0.02
Open in a new tab

NOTE: Bold typeface indicates P < 0.05. FHR > 1.0 indicates a higher conditional likelihood for pregnancy and FHR < 1.0 indicates a lower conditional likelihood for pregnancy.

a

Effect estimates adjusted for all other variables in the table;

b

‘No high school diploma’ is reference category;

c

‘Smoked < 100 lifetime cigarettes’ is reference category.

The impacts of pregnancy planning and clinical infertility was of concern. To address the issue of left truncation associated with unplanned pregnancies, we included (n = 193) women with unplanned pregnancies in a sensitivity analysis (Table 3). Setting TTP = 0 for unplanned pregnancies, the average arsenic exposure similarly increased TTP, although to a lesser degree than for planned pregnancies only. For example, relative to the likelihood for a 1st cycle pregnancy in women with 0 μg/L drinking water arsenic, women with exposure to 1 μg/L average drinking water arsenic had FHRs of 0.96 (4% lower risk), 0.94 (6% lower risk), and 0.94 (6% lower risk) for the joint effects of arsenic and time in the 6th, 9th, and 12th cycles, respectively. To evaluate potential bias due to clinical infertility we excluded n = 13 with TTP > 12 months in a second sensitivity analysis (Table 4). The effect of arsenic exposure on TTP was greater than for all planned pregnancies, as we detected even larger reductions in FHRs with longer TTPs.

Table 3.

Fecundability hazard ratios (FHR) and 95% confidence intervals (CI) for time to pregnancy (TTP) associated with drinking water arsenic exposure among planned and unplanned pregnancies (n = 287).

Predictor a FHR 95% CI P-value
Average arsenic level (μg/L) 1.01 1.00, 1.01 0.30
Maternal age (years) 1.04 1.02, 1.06 0.0004
High school diploma/some college/technical school b 0.96 0.69, 1.33 0.79
University diploma b 1.07 0.78, 1.47 0.68
Smoked > 100 lifetime cigarettes c 0.91 0.71, 1.17 0.45
Maternal age (years) x log TTP (cycles) 0.87 0.85, 0.98 < 0.0001
Average arsenic level (μg/L) x log TTP (cycles) 0.97 0.94, 0.98 0.01
Open in a new tab

NOTE: Bold typeface indicates P < 0.05. FHR > 1.0 indicates a higher conditional likelihood for pregnancy and FHR < 1.0 indicates a lower conditional likelihood for pregnancy.

a

Effect estimates adjusted for all other variables in the table;

b

‘No high school diploma’ is reference category;

c

‘Smoked < 100 lifetime cigarettes’ is reference category.

Table 4.

Fecundability hazard ratios (FHR) and 95% confidence intervals (CI) for time to pregnancy (TTP) associated with drinking water arsenic exposure among planned pregnancies with TTP ≤ 12 cycles (n = 82).

Predictor a FHR 95% CI P-value
Average arsenic level (μg/L) 1.12 1.06, 1.19 < 0.0001
Maternal age (years) 1.34 1.23, 1.45 < 0.0001
High school diploma/some college/technical school b 0.84 0.39, 1.77 0.64
University diploma b 0.94 0.49, 1.79 0.85
Smoked > 100 lifetime cigarettes c 0.76 0.44, 1.31 0.32
Maternal age (years) x log TTP (cycles) 0.79 0.75, 0.83 < 0.0001
Average arsenic level (μg/L) x log TTP (cycles) 0.89 0.83, 0.94 0.0001
Open in a new tab

NOTE: Bold typeface indicates P < 0.05. FHR > 1.0 indicates a higher conditional likelihood for pregnancy and FHR < 1.0 indicates a lower conditional likelihood for pregnancy.

a

Effect estimates adjusted for all other variables in the table;

b

‘No high school diploma’ is reference category;

c

‘Smoked < 100 lifetime cigarettes’ is reference category.

4. Discussion

In this retrospective cohort study we report no main effects for low level drinking water arsenic contamination on women’s fecundity, although we found modestly lower conditional pregnancy probabilities for women with higher arsenic exposures coupled to longer TTPs. The association between drinking water arsenic exposure and pregnancy among women with longer waiting times was consistent after adjustment for maternal age, education level, and cigarette smoking, and after conducting sensitivity analyses to assess the impacts of planning bias and clinical infertility. These data suggest that women with longer TTPs may be susceptible to arsenic-associated reproductive toxicity, even at low levels of exposure, although with no main effect. It is tempting to speculate that vulnerability to arsenic-associated reproductive toxicity is enhanced among less fecund women, those taking longer to conceive. To the best of our knowledge, this is the first human study to examine associations between drinking water arsenic exposure and female fecundity.

Our study examined drinking water arsenic exposures primarily in the 1 to 10 μg/L range, mostly at or below the 10 μg/L MCL recommended by the WHO, and enforced by the U.S. and the European Union (EU), including in Romania (Neamtiu et al., 2015; WHO, 2006). This limit, however, is predicated on cancer risk (ATSDR, 2007), and so may not be protective for reproductive endpoints. There has been very little experimental and observational research conducted to investigate the impact of arsenic exposure on female fecundity. An experimental study of pregnant mice exposed to 1, 10, 20, 37.5, 75, or 150 parts per million (ppm) arsenic in drinking water beginning at conception, reported a dose-dependent decrease in fecundity rate, defined as the product of the birth rate and litter size normalized to the unexposed control group, and a dose-dependent increase in the spontaneous abortion rate, beginning with the 20 ppm dose group (He et al., 2007). However, a prior observational study of women residing in western New York State and consuming high levels of sport fish reported no association between blood arsenic and TTP (Bloom et al., 2011). Yet, exposure in that study was believed to be primarily to non-toxic organic arsenic species, from the consumption of seafood. The parent case-control study for the current analysis reported a 1.75 higher odds (95% CI = 0.75–4.10) for spontaneous clinical pregnancy loss with 10 μg/L higher average drinking water arsenic, although among cigarette smokers only (Bloom et al., 2014a). The outcome in that study was clinically recognized loss, whereas our outcome in the current analysis was TTP as a measure of fecundity; together these studies suggest that low level drinking water arsenic contamination may be a reproductive toxicant for specific subpopulations of women. Unfortunately, the limited number of pregnancy planners who were also smokers (n = 38) in our study did not allow for stratification in the context of adjustment for the aforementioned violations of the proportional hazards assumption (i.e., arsenic exposure and maternal age).

The results of this pilot study are limited by several important issues. The retrospective nature of our study design did not allow for collection of detailed information regarding ovulation and timing of intercourse. Women appear to have difficulty recalling TTP over long time intervals (Cooney et al., 2009); however, recall accuracy appears to be high within several months of a pregnancy (Zielhuis et al., 1992), and our data were collected within 4.5 months of conception. Furthermore, women in our study were unaware of their arsenic exposure and so any recall error is unlikely to have varied by TTP. Still, our retrospective study design included only clinically recognized pregnancies. This undersampling for sterile and subfecund couples (i.e., left truncation) may bias results if related to exposure, in particular if exposed women had conceived but suffered unrecognized early losses (Joffe et al., 2005); however, given the modest nature of the detected effect we do not believe there was a substantial impact. Yet, TTP data was available only for planned pregnancies, which may bias study results away from the null if behaviors associated with arsenic exposure differ from unplanned pregnancies (Weinberg et al., 1994). Our study sample included a high proportion of unplanned pregnancies, approximately 66% compared to approximately 50% on average in the U.S. (Finer and Zolna, 2011) However, given the nature of the Romanian health care system and dissimilarities in contraception practices, we do not believe this difference reflects a sampling bias, but rather a difference in cultural norms and health-related behaviors (Johnson et al., 2004). Accordingly, we found similar results when including unplanned pregnancies as TTP = 0, and so we believe any impact to have been minimal. We also excluded women reporting use of fertility services to conceive the study pregnancy and conducted a sensitivity analysis in which women who took more than 12 cycles to conceive a pregnancy were excluded, producing similar results.

Our exposure assessment strategy comprised environmental sampling coupled to questionnaire data. The absence of an exposure biomarker may have misclassified some women, given inter-individual variability in arsenic metabolism (Engstrom et al., 2009), and possible exposure sources in addition to residential drinking water, such as diet (EFSA, 2014). In an earlier study, we reported a modest correlation (r = 0.35) between inorganic urine arsenic and average drinking water arsenic levels for a subsample (n = 20) of our overall study population (Neamtiu et al., 2015). However, arsenic has a short half-life in vivo, approximately 10 hours, and so the impact of exposure misclassification is difficult to ascertain. We cannot rule out confounding by other reproductive toxicants that might track with drinking water arsenic and also impact fecundity. Air pollution, for example, has been linked to lower fecundity (Nieuwenhuijsen et al., 2014), and levels of CO and NOx that exceeded EU air quality standards occurred within limited areas of Timişoara during 2008 (Popescu et al., 2011). Still, we used urban vs. rural residence as a proxy for air pollution exposure and there was little impact on the study results. Future studies should include a more comprehensive exposure assessment, using urinary biomarkers in addition to environmental sampling, in order to integrate arsenic exposure from all sources.

Reproduction is a couple-level function, yet we were only able to collect maternal exposure and covariate data. Evidence suggests that arsenic exposure may also impact male fecundity (Meeker et al., 2008; Xu et al., 1993), and so the absence of paternal exposure data may have misclassified exposure. Still, more than 90% of our participants with planned pregnancies reported being married or living as married, thus similar residential drinking water arsenic exposures can be anticipated. We also generated similar effect estimates when including unplanned pregnancies as TTP = 0 conceptions in a sensitivity analysis, and so we suspect any impact to have been modest.

5. Conclusions

While the results of this pilot study are preliminary, they suggest that low level arsenic exposure through consumption of arsenic contaminated residential drinking water may impair fecundity among women with longer waiting times to pregnancy. Among less fecund couples, arsenic exposures, even at levels below 10 μg/L, might further impair fecundity.

Our study is an important first step to understanding the impact of low level drinking water arsenic exposure on female fecundity. These findings are particularly concerning given the widespread distribution of low level drinking water arsenic contamination worldwide, with millions of reproductively-aged women likely to be exposed. A future prospective study that includes home testing in order to capture ovulation, timing of intercourse, and pregnancy prior to clinical recognition, will allow for inclusion of sterile and subfecund couples, and longitudinal urine collection by the female and male partners will reduce exposure misclassification. Until more definitive data are available, the feasibility for use of alternate water sources might be considered by couples exposed to drinking water arsenic and who are experiencing difficulty conceiving a pregnancy.

Supplementary Material

Supplemental Figures

Highlights.

  • We assessed low level drinking water arsenic as a predictor of fecundability.

  • Arsenic did not affect time to pregnancy among women conceiving quickly.

  • Arsenic increased time to pregnancy among women taking longer to conceive.

  • Low level drinking water arsenic may adversely impact women with lower fecundity.

Acknowledgments

We would like to thank Dr. Ramona C. Anculia, Ms. Liliana Grigore, and Ms. Ioana Trofin for their tireless efforts in recruiting study participants, conducting interviews and collecting and processing biospecimens. We would also like to thank the participants, whose generous time and effort made this study possible.

Financial Support: This work was supported by Grant #R03ES020446 from the National Institute of Environmental Health Sciences (NIEHS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIEHS or the NIH. The NIEHS played no role in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit this article for publication.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. American College of Obstetricians and Gynecologists (ACOG) and American Society for Reproductive Medicine (ASRM) Female age-related fertility decline. Fertility and Sterility. 2014;101:633–634. doi: 10.1016/j.fertnstert.2013.12.032. [DOI] [PubMed] [Google Scholar]
  2. American Society for Reproductive Medicine (ASRM) Definitions of infertility and recurrent pregnancy loss: a committee opinion. Fertility and Sterility. 2013;99:63. doi: 10.1016/j.fertnstert.2012.09.023. [DOI] [PubMed] [Google Scholar]
  3. Agency for Toxic Substances and Disease Registry (ATSDR) Toxicological profile for arsenic. U.S. Agency for Toxic Substances and Disease Registry; Atlanta: 2007. [PubMed] [Google Scholar]
  4. Axmon A, Rylander L, Albin M, Hagmar L. Factors affecting time to pregnancy. Human Reproduction. 2006;21:1279–1284. doi: 10.1093/humrep/dei469. [DOI] [PubMed] [Google Scholar]
  5. Bloom MS, Fitzgerald EF, Kim K, Neamtiu I, Gurzau ES. Spontaneous pregnancy loss in humans and exposure to arsenic in drinking water. International Journal of Hygiene and Environmental Health. 2010;213:401–413. doi: 10.1016/j.ijheh.2010.09.001. [DOI] [PubMed] [Google Scholar]
  6. Bloom MS, Louis GMB, Sundaram R, Kostyniak PJ, Jain J. Associations between blood metals and fecundity among women residing in New York State. Reproductive Toxicology. 2011;31:158–163. doi: 10.1016/j.reprotox.2010.09.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bloom MS, Neamtiu IA, Surdu S, Pop C, Lupsa IR, Anastasiu D, Fitzgerald EF, Gurzau ES. Consumption of low-moderate level arsenic contaminated water does not increase spontaneous pregnancy loss: A case control study. Environmental Health. 2014a;13:81. doi: 10.1186/1476-069X-13-81. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bloom MS, Surdu S, Neamtiu IA, Gurzau ES. Maternal arsenic exposure and birth outcomes: A comprehensive review of the epidemiologic literature focused on drinking water. International Journal of Hygiene and Environmental Health. 2014b;217:709–19. doi: 10.1016/j.ijheh.2014.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bolúmar F, Olsen J, Rebagliato M, Sáez-Lloret I, Bisanti L. Body mass index and delayed conception: A European multicenter study on infertility and subfecundity. American Journal of Epidemiology. 2000;151:1072–1079. doi: 10.1093/oxfordjournals.aje.a010150. [DOI] [PubMed] [Google Scholar]
  10. Carlin DJ, Naujokas MF, Bradham KD, Cowden J, Heacock M, Henry HF, Lee JS, Thomas DJ, Thompson C, Tokar EJ, Waalkes MP, Birnbaum LS, Suk WA. Arsenic and environmental health: State of the science and future research opportunities. Environmental Health Perspectives. 2016;124:890–899. doi: 10.1289/ehp.1510209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Collett D. Modelling Survival Data in Medical Research. Chapman & Hall; London; New York: 1994. [Google Scholar]
  12. Cooney MA, Buck Louis GM, Sundaram R, McGuiness BM, Lynch CD. Validity of self-reported time to pregnancy. Epidemiology. 2009;20:56–59. doi: 10.1097/EDE.0b013e31818ef47e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. European Food Safety Authority (EFSA) Dietary exposure to inorganic arsenic in the European population. EFSA Journal. 2014;12:3597. doi: 10.2903/j.efsa.2021.6380. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Engstrom KS, Nermell B, Concha G, Stromberg U, Vahter M, Broberg K. Arsenic metabolism is influenced by polymorphisms in genes involved in one-carbon metabolism and reduction reactions. Mutation Research-Fundamental and Molecular Mechanisms of Mutagenesis. 2009;667:4–14. doi: 10.1016/j.mrfmmm.2008.07.003. [DOI] [PubMed] [Google Scholar]
  15. Finer LB, Zolna MR. Unintended pregnancy in the United States: Incidence and disparities, 2006. Contraception. 2011;84:478–485. doi: 10.1016/j.contraception.2011.07.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999;10:37–48. [PubMed] [Google Scholar]
  17. Grodstein F, Goldman MB, Cramer DW. Body-mass index and ovulatory infertility. Epidemiology. 1994;5:247–250. doi: 10.1097/00001648-199403000-00016. [DOI] [PubMed] [Google Scholar]
  18. He W, Greenwell RJ, Brooks DM, Calderon-Garciduenas L, Beall HD, Coffin JD. Arsenic exposure in pregnant mice disrupts placental vasculogenesis and causes spontaneous abortion. Toxicological Sciences. 2007;99:244–253. doi: 10.1093/toxsci/kfm162. [DOI] [PubMed] [Google Scholar]
  19. Hertz-Picciotto I, Rockhill B. Validity and efficiency of approximation methods for tied survival times in Cox regression. Biometrics. 1997;53:1151–1156. [PubMed] [Google Scholar]
  20. Howe G, Westhoff C, Vessey M, Yeates D. Effects of age, cigarette smoking and other factors on fertility findings in a large prospective study. British Medical Journal. 1985;290:1697–1700. doi: 10.1136/bmj.290.6483.1697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Huber S, Bookstein FL, Fieder M. Socioeconomic status, education, and reproduction in modern women: An evolutionary perspective. American Journal of Human Biology. 2010;22:578–587. doi: 10.1002/ajhb.21048. [DOI] [PubMed] [Google Scholar]
  22. Hughes EG, Brennan BG. Does cigarette smoking impair natural or assisted fecundity? Fertility and Sterility. 1996;66:679–689. doi: 10.1016/s0015-0282(16)58618-x. [DOI] [PubMed] [Google Scholar]
  23. Jain RB. Association of arsenic exposure with smoking, alcohol, and caffeine consumption: Data from NHANES 2005 2010. Environmental Toxicology and Pharmacology. 2015;39:651–658. doi: 10.1016/j.etap.2015.01.011. [DOI] [PubMed] [Google Scholar]
  24. Joffe M. Time to pregnancy: A measure of reproductive function in either sex. Occupational and Environmental Medicine. 1997;54:289–295. doi: 10.1136/oem.54.5.289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Joffe M, Key J, Best N, Keiding N, Scheike T, Jensen TK. Studying time to pregnancy by use of a retrospective design. American Journal of Epidemiology. 2005;162:115–124. doi: 10.1093/aje/kwi172. [DOI] [PubMed] [Google Scholar]
  26. Johnson BR, Horga M, Fajans P. A Strategic assessment of abortion and contraception in Romania. Reproductive Health Matters. 2004;12:184–194. doi: 10.1016/s0968-8080(04)24023-x. [DOI] [PubMed] [Google Scholar]
  27. Larsen U. Research on infertility: Which definition should we use? Fertility and Sterility. 2005;83:846–852. doi: 10.1016/j.fertnstert.2004.11.033. [DOI] [PubMed] [Google Scholar]
  28. Leffondre K, Abrahamowicz M, Siemiatycki J, Rachet B. Modeling smoking history: A comparison of different approaches. American Journal of Epidemiology. 2002;156:813–823. doi: 10.1093/aje/kwf122. [DOI] [PubMed] [Google Scholar]
  29. Meeker JD, Rossano MG, Protas B, Diamond MP, Puscheck E, Daly D, Paneth N, Wirth JJ. Cadmium, lead, and other metals in relation to semen quality: Human evidence for molybdenum as a male reproductive toxicant. Environ Health Perspect. 2008;116:1473–9. doi: 10.1289/ehp.11490. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Naujokas MF, Anderson B, Ahsan H, Aposhian VH, Graziano JH, Thompson C, Suk WA. The broad scope of health effects from chronic arsenic exposure: update on a worldwide public health problem. Environmental Health Perspectives. 2013;121:295–302. doi: 10.1289/ehp.1205875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Neamtiu I, Bloom MS, Gati G, Goessler W, Surdu S, Pop C, Braeuer S, Fitzgerald EF, Baciu C, Lupsa IR, Anastasiu D, Gurzau E. Pregnant women in Timis County, Romania are exposed primarily to low-level (< 10 μg/l) arsenic through residential drinking water consumption. International Journal of Hygiene and Environmental Health. 2015;218:371–379. doi: 10.1016/j.ijheh.2015.01.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Nieuwenhuijsen MJ, Basagana X, Dadvand P, Martinez D, Cirach M, Beelen R, Jacquemin B. Air pollution and human fertility rates. Environment International. 2014;70:9–14. doi: 10.1016/j.envint.2014.05.005. [DOI] [PubMed] [Google Scholar]
  33. National Research Council (NRC) US National Research Council Subcommittee on Arsenic in Drinking Water: Arsenic in Drinking Water: 2001 Update. National Academy Press; Washington, DC: 2001. [Google Scholar]
  34. Popescu F, Ionel I, Lontis N, Calin L, Dungan IL. Air quality monitoring in an urban aglomeration. Romanian Journal of Physics. 2011;56:495–506. [Google Scholar]
  35. Quansah R, Armah FA, Essumang DK, Luginaah I, Clarke E, Marfoh K, Cobbina SJ, Nketiah-Amponsah E, Namujju PB, Obiri S, Dzodzomenyo M. Association of arsenic with adverse pregnancy outcomes/infant mortality: A systematic review and meta-analysis. Environ Health Perspect. 2015;123:412–21. doi: 10.1289/ehp.1307894. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Richardson DB, Ciampi A. Effects of exposure measurement error when an exposure variable is constrained by a lower limit. American Journal of Epidemiology. 2003;157:355–363. doi: 10.1093/aje/kwf217. [DOI] [PubMed] [Google Scholar]
  37. Rowland HAL, Omoregie EO, Millot R, Jimenez C, Mertens J, Baciu C, Hug SJ, Berg M. Geochemistry and arsenic behaviour in groundwater resources of the Pannonian Basin (Hungary and Romania) Applied Geochemistry. 2011;26:1–17. [Google Scholar]
  38. Schisterman EF, Vexler A, Whitcomb BW, Liu A. The limitations due to exposure detection limits for regression models. American Journal of Epidemiology. 2006;163:374–383. doi: 10.1093/aje/kwj039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Smedley PL, Kinniburgh DG. A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geochemistry. 2002;17:517–568. [Google Scholar]
  40. van Halem D, Bakker SA, Amy GL, van Dijk JC. Arsenic in drinking water: a worldwide water quality concern for water supply companies. Drinking Water Engineering and Science. 2009;2:29–34. [Google Scholar]
  41. Weinberg CR, Baird DD, Wilcox AJ. Sources of bias in studies of time to pregnancy. Statistics in Medicine. 1994;13:671–681. doi: 10.1002/sim.4780130528. [DOI] [PubMed] [Google Scholar]
  42. Welch AH, Westjohn DB, Helsel DR, Wanty RB. Arsenic in ground water of the United States: occurrence and geochemistry. Ground Water. 2000;38:589–604. [Google Scholar]
  43. Westrom L, Joesoef R, Reynolds G, Hagdu A, Thompson SE. Pelvic inflammatory disease and fertility. A cohort study of 1,844 women with laparoscopically verified disease and 657 control women with normal laparoscopic results. Sexually Transmitted Diseases. 1992;19:185–92. [PubMed] [Google Scholar]
  44. World Health Organization (WHO) First addendum to third edition, volume 1, Recommendations. World Health Organization; Geneva: 2006. Guidelines for drinking-water quality. [Google Scholar]
  45. Xu B, Chia Sin E, Tsakok M, Ong Choon N. Trace elements in blood and seminal plasma and their relationship to sperm quality. Reproductive Toxicology. 1993;7:613–618. doi: 10.1016/0890-6238(93)90038-9. [DOI] [PubMed] [Google Scholar]
  46. Yilmaz N, Kilic S, Kanat-Pektas M, Gulerman C, Mollamahmutoglu L. The relationship between obesity and fecundity. Journal of Womens Health. 2009;18:633–636. doi: 10.1089/jwh.2008.1057. [DOI] [PubMed] [Google Scholar]
  47. Zielhuis GA, Hulscher MEJL, Florack EIM. Validity and reliability of a questionnaire on fecundability. International Journal of Epidemiology. 1992;21:1151–1156. doi: 10.1093/ije/21.6.1151. [DOI] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supplemental Figures
NIHMS840792-supplement-Supplemental_Figures.docx (672.1KB, docx)

RESOURCES