Associations between exposure to ambient benzene and PM2.5 during pregnancy and the risk of selected birth defects in offspring

Associations between exposure to ambient benzene and PM2.5 during pregnancy and the risk of selected birth defects in offspring

Environmental Research 142 (2015) 345–353 Contents lists available at ScienceDirect Environmental Research journal homepage: www.elsevier.com/locate...

378KB Sizes 0 Downloads 52 Views

Environmental Research 142 (2015) 345–353

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Associations between exposure to ambient benzene and PM2.5 during pregnancy and the risk of selected birth defects in offspring Jean Paul Tanner a,n, Jason L. Salemi a,b, Amy L. Stuart c,d, Haofei Yu c, Melissa M. Jordan e, Chris DuClos e, Philip Cavicchia e, Jane A. Correia e, Sharon M. Watkins e, Russell S. Kirby a a

Birth Defects Surveillance Program, Department of Community and Family Health, College of Public Health, University of South Florida, Tampa, FL, USA Department of Family and Community Medicine, Baylor College of Medicine, Houston, TX, USA c Department of Environmental and Occupational Health, College of Public Health, University of South Florida, Tampa, FL, USA d Department of Civil and Environmental Engineering, College of Engineering, University of South Florida, Tampa, FL, USA e Bureau of Epidemiology, Division of Disease Control and Health Protection, Florida Department of Health, Tallahassee, FL, USA b

art ic l e i nf o

a b s t r a c t

Article history: Received 4 February 2015 Received in revised form 6 July 2015 Accepted 7 July 2015

Objective: A growing number of studies have investigated the association between air pollution and the risk of birth defects, but results are inconsistent. The objective of this study was to examine whether maternal exposure to ambient PM2.5 or benzene increases the risk of selected birth defects in Florida. Methods: We conducted a retrospective cohort study of singleton infants born in Florida from 2000 to 2009. Isolated and non-isolated birth defect cases of critical congenital heart defects, orofacial clefts, and spina bifida were identified from the Florida Birth Defects Registry. Estimates of maternal exposures to PM2.5 and benzene for all case and non-case pregnancies were derived by aggregation of ambient measurement data, obtained from the US Environmental Protection Agency Air Quality System, during etiologically relevant time windows. Multivariable Poisson regression was used to estimate adjusted prevalence ratios (aPRs) and 95% confidence intervals (CIs) for each quartile of air pollutant exposure. Results: Compared to the first quartile of PM2.5 exposure, higher levels of exposure were associated with an increased risk of non-isolated truncus arteriosus (aPR4th Quartile, 8.80; 95% CI, 1.11–69.50), total anomalous pulmonary venous return (aPR2nd Quartile, 5.00; 95% CI, 1.10–22.84), coarctation of the aorta (aPR4th Quartile, 1.72; 95% CI, 1.15–2.57; aPR3rd Quartile, 1.60; 95% CI, 1.07–2.41), interrupted aortic arch (aPR4th Quartile, 5.50; 95% CI, 1.22–24.82), and isolated and non-isolated any critical congenital heart defect (aPR3rd Quartile, 1.13; 95% CI, 1.02–1.25; aPR4th Quartile, 1.33; 95% CI, 1.07–1.65). Mothers with the highest level of exposure to benzene were more likely to deliver an infant with an isolated cleft palate (aPR4th Quartile, 1.52; 95% CI, 1.13–2.04) or any orofacial cleft (aPR4th Quartile, 1.29; 95% CI, 1.08–1.56). An inverse association was observed between exposure to benzene and non-isolated pulmonary atresia (aPR4th Quartile, 0.19; 95% CI, 0.04–0.84). Conclusion: Our results suggest a few associations between exposure to ambient PM2.5 or benzene and specific birth defects in Florida. However, many related comparisons showed no association. Hence, it remains unclear whether associations are clinically significant or can be causally related to air pollution exposures. & 2015 Elsevier Inc. All rights reserved.

Keywords: Air pollution Birth defects Exposure assessment Benzene Particulate matter

1. Introduction One out of every 33 babies in the United States (US) is born with a birth defect, and birth defects collectively account for over

n

Correspondence to: 13201 Bruce B Downs Blvd MDC56, Tampa, FL 33612, USA. E-mail addresses: [email protected] (J.P. Tanner), [email protected] (J.L. Salemi), [email protected] (A.L. Stuart), [email protected] (H. Yu), Melissa.Jordan@flhealth.gov (M.M. Jordan), Chris.Duclos@flhealth.gov (C. DuClos), Philip.Cavicchia@flhealth.gov (P. Cavicchia), Jane.Correia@flhealth.gov (J.A. Correia), Sharon.Watkins@flhealth.gov (S.M. Watkins), [email protected] (R.S. Kirby). http://dx.doi.org/10.1016/j.envres.2015.07.006 0013-9351/& 2015 Elsevier Inc. All rights reserved.

20% of all infant mortality (Canfield et al., 2006; Mathews and MacDorman, 2012). Embryonic and fetal development involves complex cellular processes and any disruptions to these pathways have the potential to cause congenital anomalies (Sadler, 2012). Various characteristics, including genetic makeup, socio-demographic characteristics, and parental behaviors (e.g., maternal smoking or alcohol use) have been posited as risk factors for birth defects, but their roles in teratogenesis are not well understood (Zhu et al., 2009). More recently, national initiatives such as the Centers for Disease Control and Prevention's (CDC) Environmental Public Health Tracking Program, the National Births Defects Prevention Study, and the Birth Defects Study to Evaluate Pregnancy

346

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

Exposures have placed an increasing focus on etiologic investigations regarding maternal-fetal exposure to environmental toxins, including air pollution, and their potential teratogenic effects on the developing fetus (CDC, 2014a, 2014b, 2014c). Despite a scarcity of research that elucidates the biological mechanism(s) by which air pollution may cause birth defects, some specific mechanisms have been proposed, including the inhibition of cell proliferation, differentiation, or migration (Agency for Toxic Substances and Disease Registry, 2007; Lupo et al., 2011; Wang and Yu, 2004); increased or decreased cell death (Abbott et al., 1989; Millicovsky and Johnston, 1981; Wang and Yu, 2004); and oxidative stress and DNA damage (Hozyasz et al., 2004; Kannan et al., 2006; Meng and Liu, 2007; Meng et al., 2003). Furthermore, the established association between maternal smoking during pregnancy and adverse birth outcomes supports the potential for detrimental influences of ambient air pollution on birth defects (Ritz and Wilhelm, 2008). A growing number of studies have investigated the association between air pollution and adverse pregnancy outcomes. Exposure has been reported to impact fetal growth (Le et al., 2012; Rich et al., 2009), birth weight (Morello-Frosch et al., 2010; Wilhelm et al., 2012), preterm birth (Chang et al., 2012; Rudra et al., 2011), as well as pregnancy conditions such as preeclampsia (Dadvand et al., 2014; Wu et al., 2009; Wu et al., 2011). Several studies have also examined the association between exposure to air pollution and select birth defects (Dadvand et al., 2011a, 2011b; Dolk et al., 2010; Gilboa et al., 2005; Hansen et al., 2009; Hwang and Jaakkola, 2008; Rankin et al., 2009; Ritz et al., 2002; Stingone et al., 2014; Strickland et al., 2009); however, only until recently have studies included PM2.5 and benzene as pollutants of interest (Agay-Shay et al., 2013; Lupo et al., 2011; Marshall et al., 2010; Padula et al., 2013a, 2013b, 2013c; Ramakrishnan et al., 2013; Schembari et al., 2014; Vinikoor-Imler et al., 2013). Data from the Florida Birth Defects Registry (FBDR) and air pollution monitoring data from the US Environmental Protection Agency (EPA) Air Quality System (AQS) were used to investigate whether maternal-fetal exposure to ambient benzene and PM2.5 were associated with several birth defects of significant public health importance. This study adds to findings in the existing literature because it is based on 10 years of data from a large statewide, population-based birth defects surveillance program, in a state whose population is ethnically diverse and also the 4th highest in terms of number of live births. The size of the study population permitted investigation of the impact of maternal exposure to these air pollutants during etiologically relevant time windows on the risk of rare but serious critical congenital heart defects (CCHD) for which previous literature is limited (e.g., Ebstein anomaly, single ventricle, truncus arteriosus), in addition to spina bifida and orofacial clefts.

2. Methods 2.1. Study population We conducted a retrospective cohort study of infants born to Florida-resident mothers that were captured by the FBDR. Since 1998, birth defects in Florida have been monitored by the FBDR, a statewide, population-based surveillance system that relies almost exclusively on passive case ascertainment. The construction of the FBDR involves linking data from multiple sources, including the Florida Bureau of Vital Statistics birth and infant death records, the Agency for Health Care Administration's inpatient and outpatient hospital discharge data, and service-related data sets from the Florida Children's Medical Services program (Salemi et al., 2011, 2012). To be included as a case, an infant must be diagnosed with

one or more structural, functional, or biochemical abnormalities within the first year of life (disease codes from the International Classification of Diseases in the range of 740–759.9 for the 9th Edition, Clinical Modification [ICD-9-CM], or with Q-codes for the 10th Edition [ICD-10]). The registry subsequently collects demographic and diagnostic information for all cases by leveraging the aforementioned data sources. Infants born alive between January 1, 2000 and December 31, 2009 were eligible for this study. Cases included singleton births identified from the FBDR with spina bifida, orofacial clefts (cleft lip with cleft palate, cleft lip alone, and cleft palate alone), and infants with the following CCHDs: coarctation of the aorta, double-outlet right ventricle, D-transposition of the great arteries, Ebstein anomaly, hypoplastic left heart syndrome (HLHS), interrupted aortic arch, pulmonary atresia with intact septum, single ventricle, total anomalous pulmonary venous return (TAPVR), tetralogy of Fallot, tricuspid atresia, and truncus arteriosus. Singleton infants born during the study period without a birth defect diagnosis were defined as non-cases; the information for non-cases is also stored in the FBDR database. 2.2. Exposure assessment Exposures were assigned to each mother for all case and noncase pregnancies using ambient air pollution measurements. We assigned exposures to each study subject based on values calculated for the census block group corresponding to the geocoded residential address reported at the time of birth. We calculated exposure concentration values for the time window during gestation corresponding to vulnerability to teratogens for each organ system. We only assigned exposures for subjects with addresses within Florida that had a daily ambient air pollution measurement within an interpolation radius of the census block group and time window. Daily concentration values for PM2.5 and benzene at each included measurement station were aggregated using inversedistance weighted spatiotemporal averaging to produce an exposure estimate for each subject. Details are provided below, with the step-by-step procedure and equations provided as supplementary material (Appendix A). Ambient air quality data are collected routinely at many monitoring stations in Florida and are archived in the EPA AQS database (www.epa.gov/ttn/airs/airsaqs). We obtained daily mean summary values of PM2.5 concentration and raw benzene concentrations during the years 1999–2009 at all stations within Florida or within 200 km of its border (in Georgia and Alabama) that operated at any time during the study period. This included 91 PM2.5 and 41 benzene stations. During periods of activity for each station, the frequency of measurements for each pollutant varied substantially, from almost every day to biweekly. For stations with more than one concentration measurement for the same pollutant on a given day, we averaged the available data on that day to obtain a single value for each date and station. Because the distributions of these daily data are highly skewed, we logtransformed these values prior to aggregation to provide a better measure of central tendency in exposure. The station-specific daily data were interpolated to estimate exposure concentration values for all mothers. Specifically, we used inverse distance-squared weighted spatiotemporal averaging to obtain an exposure for the block group and time window of defect-specific gestational vulnerability for each mother. Our interpolation approach is equivalent to inverse distance-squared weighting in space, but uses a weight proportional to the inverse of the square of the multi-dimensional distance (rather than the spatial distance alone). To calculate the multi-dimensional distance, we used the Euclidean distance formula with each component distance (in km or days) non-dimensionalized (divided) by

Table 1 Maternal and infant characteristics by pollutant and birth defect grouping, Florida 2000–2009. Characteristic

Sample size Median exposure (IQR) Benzene: log10 ppbC PM2.5: log10 mg/m3

Benzene analysis population

PM2.5 analysis population

No malformation

CCHD

Orofacial clefts

Spina bifida

No malformation

CCHD

Orofacial clefts

Spina bifida

973,797

2,028

1,299

271

1,664,770

3,494

2,361

513

0.15 (0.03–0.28)a

0.15 (0.02–0.27)

0.16 (0.04–0.29)

0.14 (0.03–0.30)

0.95 (0.89–1.00)a

0.95 (0.89–1.00)

0.95 (0.89–1.00)

0.95 (0.86–1.02)

NA NA

81.5 18.5

71.1 29.9

57.6 42.4

NA NA

81.1 18.9

73.8 26.2

57.4 42.6

Maternal race/ethnicity (%) Non-hispanic white Non-hispanic black Hispanic Other

38.1 23.2 34.1 4.2

41.9 24.8 30.1 2.8

47.3 17.6 31.4 3.5

36.5 21.8 40.6 0.7

46.3 21.2 28.1 4.0

50.8 22.1 23.8 3.0

55.9 15.8 24.7 3.3

46.9 20.9 30.1 2.0

Maternal nativity (%) US-born Foreign-born

60.4 39.5

66.3 33.6

67.2 32.7

62.4 37.6

67.6 32.4

72.8 27.1

74.1 25.8

71.7 28.3

Maternal smoking status (%) Smoker Nonsmoker

4.7 95.1

6.6 93.2

7.6 92.1

4.1 95.6

7.0 92.9

8.5 91.3

10.7 89.2

5.7 93.9

Maternal education (%) o High School High School only College

18.3 31.4 49.4

19.4 32.1 47.3

20.5 31.4 47.0

21.4 35.1 41.4

19.5 31.9 48.0

20.1 32.7 46.3

21.1 32.6 45.6

23.4 34.8 40.8

Maternal age (years) (%) o20 20–24 25–29 30–34 Z35

9.9 23.7 26.6 24.0 15.7

10.1 23.3 26.0 22.6 18.0

9.5 23.9 27.0 23.8 15.9

8.5 28.0 23.2 21.4 18.8

10.8 25.3 26.7 22.7 14.5

10.3 24.5 26.5 21.6 17.1

10.5 25.5 26.9 22.4 14.7

10.4 27.0 26.2 20.9 15.6

Maternal marital status (%) Married Not married

58.0 42.0

56.1 43.9

58.5 41.5

55.7 44.3

58.1 41.9

56.6 43.4

58.3 41.7

57.6 42.4

Parity (%) Nulliparous Multiparous

43.3 56.4

42.9 56.5

40.4 59.5

34.7 65.3

42.6 57.1

41.2 58.4

40.7 59.1

37.7 62.3

Neighborhood median income (%) o $20,000 $20,000–$49,999 Z $50,000

5.7 66.6 27.7

6.4 67.9 25.7

5.1 68.8 26.1

7.0 68.3 24.7

5.5 69.8 24.7

5.6 71.9 22.5

4.7 72.9 22.4

6.4 74.4 19.1

Infant sex (%) Female Male

49.4 50.6

43.3 56.7

45.3 54.6

53.5 46.5

49.4 50.6

43.3 56.7

45.2 54.8

51.0 49.0

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

Type of defect (%) Isolated Multiple

Percentages may not add to 100 due to missingness; CCHD ¼critical congenital heart defect a

Median exposure from 3 to 12 week temporal scale of analysis 347

348

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

the respective spatial or temporal interpolation limit. We used a maximum interpolation radius of 50 km in space and a maximum temporal distance limit of 3 days for PM2.5 and 7 days for benzene. Distance in space was defined with respect to the block group centroid; distance in time was defined with respect to the date limits of the time window of vulnerability. For stations within a block group, the spatial distance was set to the radius of a circle with the same area as the block group. For dates within the time window of vulnerability, the temporal distance was set to half that window. Due to the sparsity of the available measurement data in both space and time, data were not available within these limits for every mother. Hence, mothers without at least one measurement within these limits were excluded from the analyses. To select the appropriate time window during gestation for each mother, we used the established windows during fetal development in which the maturing organs or organ systems have been found to be particularly vulnerable to teratogens; hence, exposure during these windows may be most likely to result in related fetal malformations. For some pollutants, build-up of concentrations of the pollutant or its metabolites in the body over a longer period of time (i.e., before the start of organ development) may also affect fetal development (Ritz and Wilhelm, 2008). However, toxicological evidence is too limited to support selection of a specific time period for build-up or to discern differences between pollutants. Hence, for this study we used the vulnerability window specific to each birth defect, but consistent between pollutants. Specifically, we used 3–4 weeks gestation for spina bifida because neurulation, the process through which the neural tube is formed, begins during week three and is complete by week four (Copp et al., 2003; Sadler, 2012). For CCHDs, we used 3–8 weeks gestation because the vascular system appears in the middle of the third week and the heart is fully formed by the end of the seventh week (Sadler, 2012). For cleft lip alone, we also used 3–8 weeks gestation because formation of the mandibular prominences, which merge to form the lower and then upper lips, begins during week three, and by week eight, the upper lip is fully formed (Bender, 2000). For cleft palate alone, we used 5–12 weeks because the palate begins to form during the fifth week of development and is complete at the end of the twelfth week (Merritt, 2005). For cleft lip with cleft palate, we used 3–12 weeks to span from the formation of the mandibular prominences to the complete formation of the palate. To evaluate whether our results were sensitive to the time window chosen and to account for the possibility that build-up of pollutant concentrations may be important, we also conducted a sensitivity analysis using a time window that included two additional weeks prior to the established window of vulnerability. Gestational age was determined using the clinical estimate of gestation from each infant's birth certificate. If the clinical estimate was missing or out of the inclusion range (20–44 weeks), gestational age was calculated using the date of last menstrual period. Geocoded residential address (latitude and longitude) for each mother, at the time of the child's birth, were obtained from the infant birth record. We mapped these to their respective census block group using ArcGIS 10.1 (ESRI, Redlands, California). 2.3. Statistical analysis Multivariable Poisson regression models were used to estimate adjusted prevalence ratios (aPRs) and 95% confidence intervals (CI) representing the association between PM2.5 or benzene and selected congenital anomalies (spina bifida, cleft lip alone, cleft palate alone, cleft lip with cleft palate, and CCHDs). Exposure concentrations were divided into quartiles (based on the distribution among non-cases) and the lowest quartile of PM2.5 and benzene concentrations were used as the exposure reference groups. To

ensure consistency in the non-case group throughout all analyses for a given pollutant, non-cases without an estimated exposure for all temporal windows of vulnerability (i.e., 3–4, 3–8, 5–12, and 3– 12 weeks) were excluded from the study. Analyses were then stratified by isolated and non-isolated birth defect cases. Isolated cases included infants with a single selected birth defect or multiple birth defects within the same organ system. Non-isolated cases included infants with a selected birth defect and one or more major structural birth defects outside of the primary defect's organ system. For example, an infant diagnosed with tricuspid atresia and pulmonary atresia (both heart defects) would be classified as isolated; an infant with a cleft palate and HLHS (defects in two different body systems) would be classified as non-isolated. In addition to the pollutant and birth defect of interest, we considered a number of potential confounders for inclusion in each model. These variables were selected a priori based on previous literature (Lupo et al., 2011; Padula et al., 2013b; Ramakrishnan et al., 2013) and final selection was determined by empirical analyses. Data on all covariates except household income were available from the Florida Bureau of Vital Statistics birth certificate. Maternal race/ethnicity was categorized as nonHispanic white, non-Hispanic black, Hispanic, or other. Maternal nativity was dichotomized as U.S. or foreign-born (born outside the 50 U.S. states). Maternal age in years was categorized as o20, 20–24, 25–29, 30–34, and Z 35. Maternal education was classified as less than high school (o 12 years), high school (12 years), or more than high school (4 12 years). Maternal marital status at the time of delivery was categorized as married or unmarried. Parity was classified as nulliparous or multiparous. As a proxy for individual household income, block group median household income (i.e., neighborhood household income) from the 2000 US Census Bureau (www.census.gov) was linked to the mother's residential address at the time of delivery; the block group is the smallest geographic unit with this information. The infant's birth cohort and sex were also considered as covariates. Only covariates that were significantly associated with both the birth defect outcome and the pollutant exposure of interest in bivariate analyses were included in the adjusted model. All statistical tests were two-sided and considered significant at po 0.05. Statistical analyses were performed with SAS software, version 9.3 (SAS Institute, Inc., Cary, NC). Approval for the study was obtained from the Institutional Review Board at the Florida Department of Health and the University of South Florida.

3. Results 3.1. Descriptive statistics There were 2,123,874 singleton live births to Florida residents during the study period and 94.7% had geocoded residential addresses available in the birth certificate record. We were able to map 95.3% of these to their respective census block group (n ¼1,917,155). The final study population of infants with or without a birth defect in which we were able to obtain a maternal exposure estimate for benzene or PM2.5, along with the distribution of important maternal and infant sociodemographic and perinatal factors, are presented in Table 1. For both pollutant study populations, mothers of infants with CCHDs, orofacial clefts, or spina bifida were more likely than mothers of infants without birth defects to be US-born, be 35 years or older, have less than a high school education, have given birth previously, and have a neighborhood median income of $20,000–$49,999. A higher proportion of mothers with an infant born with CCHDs or orofacial clefts smoked during pregnancy and infants with these birth defects were more likely to be male. Mothers of infants with spina

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

bifida were more likely to be Hispanic, whereas mothers of infants with CCHDs or orofacial clefts were more likely to be non-Hispanic white. 3.2. Benzene exposure and the risk of birth defects The results from multivariable Poisson regression models for benzene and each birth defect or birth defect group are presented in Table 2. Among isolated cases, higher levels of benzene exposure were associated with an increased risk of cleft palate alone. Compared with women in the first quartile of benzene exposure, women in the fourth quartile were 1.5 times more likely to deliver an infant with a cleft palate alone (95% CI, 1.13–2.04). Although, the risk of any cleft was increased in the fourth quartile compared with the first quartile of benzene exposure (aPR, 1.29; 95% CI, 1.08– 1.56), we did not find evidence of an overall association between any orofacial cleft and benzene exposure. Among non-isolated cases, we did not observe any overall associations between maternal exposure to benzene and any birth defect. However, when comparing the fourth to the first quartile of exposure, we observed some evidence of a strong inverse association between benzene and the risk of pulmonary atresia (aPR, 0.19; 95% CI, 0.04–0.84). Sensitivity analyses using a time window that included two weeks prior to the established window of vulnerability, allowing for pollutant build-up, were not substantially different from the base case analyses reported in Table 2 (data not shown). 3.3. PM2.5 exposure and the risk of birth defects Table 3 presents the results from multivariable Poisson regression models for PM2.5 and each birth defect or birth defect group. Among isolated cases, women in the third quartile group of PM2.5 exposure were more likely to deliver an infant with any CCHD (aPR, 1.13; 95% CI, 1.02–1.25); however, we did not observe an overall association between maternal PM2.5 exposure and any CCHD. Among non-isolated cases, we observed an increasing likelihood of several birth defects with increasing levels of maternal PM2.5 exposure. Compared to women in the lowest quartile of PM2.5, women in the highest quartile were more likely to deliver an infant with truncus arteriosus (aPR, 8.80; 95% CI, 1.11–69.50), coarctation of the aorta (aPR, 1.72; 95% CI, 1.15–2.57), and interrupted aortic arch (aPR, 5.50; 95% CI, 1.22–24.82). We also observed an increased risk of coarctation of the aorta for offspring of women in the third quartile of PM2.5 exposure (aPR, 1.60; 95% CI, 1.07–2.41). Although women in the highest quartile of exposure, compared to those in the first quartile had evidence of being at increased risk of any CCHD, the overall association between PM2.5 exposure and CCHD was not statistically significant. Our sensitivity analyses that included two weeks prior to the established window of vulnerability resulted in the associations for coarctation of the aorta, interrupted aortic arch, and any CCHD being similar to our base case analyses, but for truncus arteriosus the association was no longer significant. Among non-isolated cases, a borderline decreased risk of cleft lip with or without cleft palate and any orofacial cleft was observed in women in the second quartile, as well as a borderline decreased risk in double-outlet right ventricle in women in the fourth quartile among isolated cases (data not shown). However, the point estimates observed in the sensitivity analysis all fell within the confidence intervals of the results presented in Table 3.

4. Discussion This is one of the largest studies investigating the relationship between PM2.5 and benzene and the risk of CCHDs, orofacial clefts,

349

and spina bifida. In general, we found little evidence that maternal exposure to PM2.5 or benzene during an etiologically-relevant time window was associated with the risk of delivering an infant with one or more of the birth defects included in this study. However, among non-isolated cases, higher levels of exposure to PM2.5 was associated with an increased risk of truncus arteriosus, coarctation of the aorta, and interrupted aortic arch. Among isolated cases, exposure to higher levels of benzene was associated with an increased risk of cleft palate alone and a borderline increased risk of any orofacial cleft. However, because many comparisons of related effects showed no associations here, it remains unclear whether the associations found are actually significant or suggest potential causality. Several studies have examined the association between maternal exposure to ambient air pollution and congenital heart defects, orofacial clefts, and spina bifida (Agay-Shay et al., 2013; Dadvand et al., 2011a, 2011b; Dolk et al., 2010; Gilboa et al., 2005; Hansen et al., 2009; Hwang and Jaakkola, 2008; Lupo et al., 2011; Marshall et al., 2010; Padula et al., 2013a; Padula et al., 2013b, 2013c; Ramakrishnan et al., 2013; Rankin et al., 2009; Ritz et al., 2002; Schembari et al., 2014; Stingone et al., 2014; Strickland et al., 2009; Vinikoor-Imler et al., 2013). However, benzene and (until recently) PM2.5 have been understudied pollutants in fetal development. Most studies examining effects on the risk of congenital heart defects have failed to have the statistical power necessary to conduct defect-specific analyses on all heart defects, specifically CCHDs. By leveraging 10 years of data from a large, statewide birth defects surveillance program, our study had increased statistical power to investigate the impact of maternal exposure to these air pollutants on the risk of rare but serious CCHDs (e.g., TAPVR, interruption of the aortic arch, truncus arteriosus, Ebstein anomaly) in offspring. Using data from National Birth Defects Prevention Study, a large population-based case-control study of birth defects, Stingone et al. reported a positive association between higher levels of PM2.5 and HLHS (Stingone et al., 2014). The study's large sample size allowed the investigation of maternal exposure to criteria air pollutants (excluding lead) and the risk of other rare CCHDs; however, no other significant associations were observed. In a study of residents of San Joaquin Valley, California, investigators found that higher levels of maternal exposure to PM2.5 were associated with increased odds of dextro-transposition of the great arteries (Padula et al., 2013b). They also reported an increased odds of pulmonary valve stenosis and perimembranous ventricular septal defect with higher exposure to PM10. More recently, Schembari et al. evaluated the association between several birth defects and various pollutants, and reported an inverse association between PM2.5 and ventricular septal defect (Schembari et al., 2014). No other significant associations between maternal exposure to PM2.5 and the risk of congenital heart defects or neural tube defects were observed. A study in the Tel-Aviv region of Israel examining the association between congenital heart defects and six common air pollutants observed a negative association between isolated patent ductus arteriosus and exposure to PM2.5 and a positive association between multiple congenital heart defects and exposure to PM10 (Agay-Shay et al., 2013). Vinikoor-Imler et al. investigated the risk of a number of CCHDs, orofacial clefts, and spina bifida, and exposure to PM2.5 and ozone (Vinikoor-Imler et al., 2013). Among all congenital heart defects studied, PM2.5 exposure was only associated with a reduced risk of atrial septal defect. No associations between PM2.5 and the risk of spina bifida or orofacial clefts were reported. Other studies examining the association between PM2.5 and orofacial clefts or spina bifida reported no statistically significant associations (Marshall et al., 2010; Padula et al., 2013a), although Padula et al. (2013a) observed an increased odds of spina bifida with exposure to carbon

350

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

Table 2 Adjusted prevalence ratios and 95% confidence intervals for the association between selected birth defects and exposure to benzene. Birth defect

Isolated

Non-isolated

Cases 2nd Quartile

3rd Quartile

4th Quartile

Cases 2nd Quartile

3rd Quartile

4th Quartile

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

0.87 (0.61, 1.26)

0.83 (0.57, 1.21)

21

1.25 (0.34, 4.66)

1.50 (0.42, 5.31)

1.53 (0.43, 5.43)

185 1.19 (0.79, 1.79) 0.90 (0.58, 1.40) 257 0.84 (0.59, 1.18) 1.00 (0.72, 1.38) 369 0.87 (0.65, 1.16) 1.00 (0.75, 1.32) 75 1.00 (0.51, 1.96) 1.02 (0.52, 2.00) 86 0.84 (0.48, 1.49) 0.81 (0.46, 1.44)

1.34 (0.89, 2.00) 0.77 (0.54, 1.10) 0.97 (0.73, 1.29) 1.47 (0.79, 2.74) 0.67 (0.36, 1.24)

62 45 143 7 13

1.31 (0.63, 2.69) 0.50 (0.23, 1.12) 0.82 (0.52, 1.30) NA 1.25 (0.34, 4.66)

1.36 (0.66, 2.77) 0.50 (0.23, 1.12) 0.90 (0.57, 1.41) 0.67 (0.11, 3.99) 0.75 (0.17, 3.35)

1.01 (0.48, 2.16) 0.49 (0.22, 1.10) 0.87 (0.55, 1.37) 0.68 (0.11, 4.07) 0.26 (0.03, 2.28)

135 109 48 510 53

Congenital heart defects Dextro-transposition of the great arteries Double-outlet right ventricle Hypoplastic left heart syndrome Tetralogy of Fallot Truncus arteriosus Total anomalous pulmonary venous return Pulmonary atresia Tricuspid atresia Ebstein anomaly Coarctation of the aorta Aortic interruption/atresia/ hypoplasia Single ventricle Any critical congenital heart defects

0.87 (0.54, 1.39) 0.83 (0.49, 1.42) 0.67 (0.27, 1.63) 1.04 (0.82, 1.32) 0.93 (0.44, 1.98)

1.04 (0.66, 1.64) 0.97 (0.58, 1.61) 1.50 (0.72, 3.11) 0.98 (0.77, 1.25) 0.92 (0.43, 1.96)

0.78 (0.48, 1.28) 0.85 (0.50, 1.45) 0.85 (0.37, 1.97) 0.85 (0.66, 1.10) 0.92 (0.43, 1.95)

25 21 4 128 13

0.45 (0.16, 1.31) 0.63 (0.20, 1.91) NA 0.94 (0.59, 1.50) 0.33 (0.03, 3.21)

0.64 (0.25, 1.64) 0.75 (0.26, 2.16) NA 0.75 (0.46, 1.23) 1.01 (0.20, 4.99)

0.19 (0.04, 0.84) 0.26 (0.05, 1.20) NA 0.77 (0.47, 1.26) 2.03 (0.51, 8.12)

110 0.74 (0.45, 1.24) 1653 1.02 (0.89, 1.17)

0.69 (0.41, 1.15) 1.00 (0.87, 1.15)

0.73 (0.44, 1.23) 0.95 (0.83, 1.09)

20 375

1.50 (0.42, 5.31) 0.90 (0.68, 1.19)

1.68 (0.49, 5.73) 0.90 (0.68, 1.20)

0.68 (0.15, 3.05) 0.90 (0.67, 1.19)

Orofacial clefts Cleft lip alone Cleft palate alone Cleft lip with cleft palate Cleft lip with or without cleft palate Any orofacial cleft

205 347a 368 574 923

1.28 (0.87, 1.89) 1.07 (0.78, 1.47) 1.06 (0.80, 1.41) 1.18 (0.94, 1.49) 1.17 (0.97, 1.41)

1.27 (0.86, 1.87) 1.12 (0.82, 1.52) 0.91 (0.68, 1.23) 1.07 (0.85, 1.35) 1.12 (0.93, 1.35)

0.98 (0.65, 1.49) 1.52 (1.13, 2.04) 1.11 (0.83, 1.48) 1.07 (0.84, 1.36) 1.29 (1.08, 1.56)

26 185 163 189 376

2.25 (0.69, 7.31) 0.98 (0.65, 1.49) 0.71 (0.46, 1.10) 0.81 (0.54, 1.22) 0.78 (0.58, 1.04)

1.50 (0.42, 5.31) 1.03 (0.68, 1.55) 0.77 (0.50, 1.19) 0.78 (0.52, 1.17) 0.92 (0.70, 1.22)

1.79 (0.52, 6.10) 1.15 (0.77, 1.72) 0.97 (0.64, 1.45) 1.03 (0.70, 1.51) 1.00 (0.76, 1.31)

156

1.19 (0.77, 1.83)

0.84 (0.53, 1.35)

1.08 (0.69, 1.67)

115

1.24 (0.73, 2.10)

1.30 (0.77, 2.19)

1.01 (0.59, 1.76)

Spina bifida without anencephaly

232 1.06 (0.75, 1.50)

Note. aPR¼ adjusted prevalence ratio; CI ¼confidence interval; NA ¼ Insufficient birth defect cases available; Bolded values are for confidence intervals that do not contain 1.00. Potential confounders adjusted for across all models were maternal race/ethnicity, maternal nativity, maternal smoking, maternal education, maternal age group, marital status, neighborhood median income, parity and infant sex. However, only variables significantly associated with the exposure and birth defect of interest were included in each model. a

The p-value representing the overall (type 3) effects between benzene exposure and the birth defect was statistically significant ( o 0.05).

monoxide and nitrogen dioxide, as well as a positive association between traffic density and cleft lip with or without cleft palate. Moreover, Marshall et al. reported an increased odds of orofacial clefts with exposure to sulfur dioxide and carbon monoxide (Marshall et al., 2010). Two previous studies investigated the association between ambient benzene exposure and risk of birth defects. Lupo et al. conducted a study in Texas and found evidence of a positive association between spina bifida and maternal exposure to benzene (Lupo et al., 2011). Another study, also conducted in Texas, found no association between benzene exposure and the risk of orofacial clefts (Ramakrishnan et al., 2013). The lack of consistency in findings across studies may be due, in part, to variable study samples. Due to sampling error, different samples may produce variable results that must be taken into account when making inferences about the effect of an exposure on an outcome in a population (Kirkwood and Sterne, 2003). Although measures of effect and measures of association may have been comparable in a number of studies, the inconsistency in reporting significant findings may be due to small sample sizes and suboptimal statistical power. The lack of consensus across studies on ambient air pollution and congenital anomalies is also attributable to different methods for (1) defining and ascertaining cases of birth defects, (2) conducting statistical analyses, and (3) estimating exposures. The majority of previous studies identified infants with birth defects by leveraging routinely-collected data from surveillance programs that employ case-finding strategies that lie on a continuum ranging from “active” to “passive” (Salemi et al., 2012). Active birth defects case-finding typically involves staff seeking cases through the review of medical records and/or other hospital/

ambulatory care records, which results in high completeness of ascertainment (National Birth Defects Prevention Network, 2004). Comparatively, passive case-finding has poorer completeness since it relies on physicians and/or health care facilities reporting cases of birth defects, or identifies cases using existing administrative data sources (e.g., vital records, hospital discharge data). Our study relies on the FBDR, a passive surveillance system using only administrative databases and without case confirmation. Therefore, identification of each case depends on the interpretation of a clinician's diagnostic write-up, translation into an ICD code, and accurate entry of that code into a database, which are all susceptible to human error. Compared to surveillance programs used in other studies, the FBDR's suboptimal completeness and accuracy may increase case misclassification and dilute our reported measures of association. However, a previous evaluation of the FBDR reported that its completeness of ascertainment was approximately 90%, although its ability to identify specific defects was defect dependent (Salemi et al., 2012). Differences in statistical analysis protocols and model specification impact the comparability and consistency of findings in the literature. First, researchers employ different approaches to analyzing maternal exposure estimates. A number of studies have analyzed exposure as a continuous variable, either on its original scale, or as a log-transformed value. Others have instead discretized exposure into various categories (e.g., above/below median, quantiles). Discretization of continuous measures is often criticized due to a loss of statistical power, an assumption of homogeneity of risk within groups, and difficulty in comparability across studies due to data-driven cut-points (Bennette and Vickers, 2012). However, to improve the interpretability of our findings, as well as comparability with other studies using similar

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

351

Table 3 Adjusted prevalence ratios and 95% confidence intervals for the association between selected birth defects and exposure to PM2.5. Birth defect

Isolated

Non-isolated

Cases 2nd Quartile

3rd Quartile

4th Quartile

Cases 2nd Quartile

3rd Quartile

4th Quartile

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

aPR (95% CI)

415

0.93 (0.71, 1.22)

0.93 (0.71, 1.22)

0.88 (0.67, 1.15)

42

2.46 (0.95, 6.34)

1.60 (0.58, 4.39)

1.67 (0.62, 4.52)

310 450 644 134 136

1.03 0.95 1.24 0.83 1.03

0.98 1.20 1.24 1.20 1.06

0.78 1.22 1.09 0.80 0.80

108 93 251 18a 26

1.67 (0.96, 2.91) 1.41 (0.76, 2.63) 0.99 (0.71, 1.40) 2.96 (0.31, 28.44) 5.00 (1.10, 22.84)

1.53 (0.87, 2.69) 1.27 (0.67, 2.39) 0.74 (0.51, 1.07) 4.93 (0.58, 42.17) 2.50 (0.49, 12.90)

1.28 (0.71, 2.31) 1.67 (0.92, 3.03) 1.12 (0.80, 1.56) 8.80 (1.11, 69.50) 4.50 (0.97, 20.83)

215 185 85 880 87 204 2832

1.35 (0.94, 1.93) 0.92 (0.61, 1.39) 1.19 (0.67, 2.13) 1.02 (0.85, 1.24) 1.28 (0.71, 2.29) 0.93 (0.62, 1.40) 1.05 (0.95, 1.17)

0.93 (0.62, 1.37) 0.95 (0.63, 1.43) 1.00 (0.55, 1.83) 1.16 (0.96, 1.39) 0.96 (0.52, 1.78) 1.03 (0.70, 1.52) 1.13 (1.02, 1.25)

0.89 (0.60, 1.32) 1.04 (0.70, 1.55) 0.86 (0.46, 1.61) 0.98 (0.81, 1.19) 0.96 (0.52, 1.77) 1.17 (0.80, 1.71) 1.00 (0.90, 1.11)

36 28 10 213a 27a 38 662

0.99 (0.37, 2.63) 0.63 (0.20, 1.91) 2.50 (0.49, 12.89) 1.35 (0.89, 2.06) 2.50 (0.49, 12.89) 1.43 (0.61, 3.35) 1.17 (0.94, 1.47)

1.23 (0.49, 3.12) 0.75 (0.26, 2.16) 0.50 (0.05, 5.52) 1.60 (1.07, 2.41) 4.50 (0.97, 20.85) 0.98 (0.39, 2.46) 1.16 (0.93, 1.45)

1.24 (0.49, 3.13) 1.13 (0.43, 2.92) 1.00 (0.14, 7.10) 1.72 (1.15, 2.57) 5.50 (1.22, 24.82) 0.74 (0.28, 2.00) 1.33 (1.07, 1.65)

Orofacial clefts Cleft lip alone Cleft palate alone Cleft lip with cleft palate Cleft lip with or without cleft palate Any orofacial cleft

359 653 730 1089 1742

1.14 (0.84, 1.55) 0.94 (0.75, 1.17) 0.97 (0.79, 1.19) 0.99 (0.83, 1.17) 0.99 (0.86, 1.13)

1.12 (0.83, 1.52) 1.07 (0.86, 1.34) 0.95 (0.78, 1.17) 0.98 (0.83, 1.17) 1.02 (0.89, 1.17)

1.26 (0.94, 1.70) 1.12 (0.90, 1.39) 0.96 (0.79, 1.18) 1.03 (0.87, 1.22) 1.07 (0.94, 1.22)

46 297 276 322 619

0.71 1.08 0.81 0.78 0.86

0.79 0.86 1.01 0.93 0.92

0.79 1.05 0.95 0.91 0.91

Spina bifida without anencephaly

294

1.03 (0.76, 1.40)

0.87 (0.63, 1.19)

0.79 (0.56, 1.09) 218

Congenital heart defects Dextro-transposition of the great arteries Double-outlet right ventricle Hypoplastic left heart syndrome Tetralogy of Fallot Truncus arteriosus Total anomalous pulmonary venous return Pulmonary atresia Tricuspid atresia Ebstein anomaly Coarctation of the aorta Aortic interruption/atresia/hypoplasia Single ventricle Any critical congenital heart defects

(0.76, 1.39) (0.72, 1.26) (0.99, 1.55) (0.51, 1.36) (0.65, 1.64)

(0.72, 1.33) (0.92, 1.56) (0.99, 1.55) (0.77, 1.88) (0.67, 1.68)

(0.56, 1.09) (0.94, 1.59) (0.87, 1.38) (0.49, 1.32) (0.49, 1.32)

(0.32, 1.61) (0.78, 1.48) (0.57, 1.14) (0.57, 1.06) (0.69, 1.07)

0.84 (0.57, 1.24)

(0.36, 1.73) (0.61, 1.20) (0.73, 1.40) (0.69, 1.26) (0.74, 1.14)

0.99 (0.68, 1.43)

(0.36, 1.73) (0.77, 1.44) (0.69, 1.32) (0.67, 1.23) (0.73, 1.14)

0.96 (0.67, 1.40)

Note. aPR¼ adjusted prevalence ratio; CI ¼ confidence interval; Bolded values are for confidence intervals that do not contain 1.00 Potential confounders adjusted for across all models were maternal race/ethnicity, maternal nativity, maternal smoking, maternal education, maternal age group, marital status, neighborhood median income, parity and infant sex. However, only variables significantly associated with the exposure and birth defect of interest were included in each model. a

The p-value representing the overall (type 3) effects between PM2.5 exposure and the birth defect was statistically significant ( o0.05).

categorization decisions, we grouped exposure estimates into quartiles. Second, statistical models used to generate adjusted risk estimates include a myriad of combinations of potential confounders (e.g., alcohol intake, poor diet, indoor pollutants). For example, maternal smoking during pregnancy is a well-known risk factor for orofacial clefts (Vrijheid et al., 2011); however, few studies have been able to control for tobacco smoking, as has our study. Different approaches to assigning maternal-fetal exposure to ambient air pollution may also account for some of the variation in exposure estimates and, ultimately, reported measures of association. Although a few recent studies have reported on the sensitivity of estimated exposures to assignment methods (Ozkaynak et al., 2013; Wong et al., 2004), and on impacts of these methods on associations with health outcomes (Baxter et al., 2013) none have looked specifically at birth defect associations. As part of an ongoing study, we investigated the sensitivity of maternal exposure estimates to exposure assessment decisions. Preliminary results suggest that the exposure estimates are somewhat sensitive to these decisions (data not shown). Hence, impacts of the exposure assignment approach on the effect estimate cannot be excluded. Several methods have been routinely applied in air pollution exposure estimation (from local spatial averaging to land-use regression modeling) (Jerrett et al., 2005). For this study, we used interpolation of ambient monitoring data as a representative approach. Inverse distance weighting is a common interpolation technique used in birth defect studies (Agay-Shay et al., 2013; Hwang and Jaakkola, 2008; Padula et al., 2013a, 2013b, 2013c). However, these studies only allow information nearby in space (e.g., measurements at monitoring stations within 50 km of the maternal residence) to inform the estimate of exposure, but not

measurements taken nearby in time (e.g., within a few days of the window of vulnerability). To our knowledge, this is the first birth defects study using inverse distance-squared weighting that incorporates spatiotemporal multi-dimensional distance. Our other decisions included the use of a defect-specific window of vulnerability to teratogens as the temporal scale of analysis. Windows used in previous studies have ranged from 3 to 8 weeks of gestation (Agay-Shay et al., 2013; Dadvand et al., 2011a, 2011b; Gilboa et al., 2005; Schembari et al., 2014) to the entire first trimester (Rankin et al., 2009). We consider the defect-specific window of vulnerability to be the most etiologically relevant, but also conducted a sensitivity analysis on the impacts of that choice. Specifically, we re-estimated maternal exposures for a time period that included the defect-specific window of vulnerability and the two weeks prior, to allow for the effect of build-up of the pollutant concentrations. We observed no meaningful differences from the results of our base case analyses. Several spatial scales have also been used previously for exposure assignment, from the point location of the mother's residential address to the county of residence (typically recorded at the time of the child's birth). All have weaknesses for representing the mother's location during the exposure window. The smaller scales (e.g., point of residential address) do not capture the breadth of the mother's activity locations over the window of interest (Klepeis et al., 2001; Lupo et al., 2010) and estimate variations in exposure that are not supported by the resolution of the underlying monitoring data (Wong et al., 2004). The larger scales (e.g., county) cannot capture spatial differences in ambient pollution that are known to exist (based on measurement studies with higher spatial resolution than regular monitoring data). Because people who share factors that covary with exposures and birth outcomes tend to live in close proximity (Evans and Kantrowitz,

352

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353

2002; Woodruff et al., 2003), smaller spatial scales with relatively homogenous measures for these factors can better control for confounding during the statistical analyses for association. Here, we used the block group area of the mother's residential address as the spatial scale. Block group sizes depend on population density; in Florida, they range in size from less than 0.5 km2 for the most densely populated locations to 3000 km2 in the Everglades. The block group is an intermediate scale that provides a compromise between capturing exposures that occur over the full spatial movement of the mother during the exposure window, and capturing differences in exposure due to small-scale spatial variations. Another limitation of the exposure estimate is the use of the mother's residential address at the time of delivery to estimate maternal exposures to ambient levels of PM2.5 and benzene that occurred during the first trimester. This can introduce misclassification of exposure because women in our study may have resided at different locations during pregnancy. However, Lupo et al. found good agreement between measures estimated at the mother's residential addresses at the time of conception and delivery (Lupo et al., 2010). Furthermore, although a review of research on residential mobility found approximately 20% of the population moved during pregnancy, the median distance moved was less than 10 km (Bell and Belanger, 2012). Thus, the degree of misclassification of maternal exposure due to this temporally mismatched information may be small. Two case-control studies observed no difference in the rate of mobility between mothers of infants with birth defects compared with mothers of unaffected infants (Lupo et al., 2010; Miller et al., 2010). Due to all of the potential limitations of exposure assessment, exposure misclassification is a substantial concern with all studies that do not measure personal exposure. In our study, misclassification of exposure is most likely to be non-differential according to case status, which may result in bias towards or even away from the null (Fosgate, 2006; Jurek et al., 2008, 2005). Furthermore, the interquartile range of estimated exposure is small in comparison to the median value (Table 1), particularly for PM2.5. Hence, noise in the estimate due to uncertainty in both the measurement data itself and the exposure assignment process could substantially impact the results, indicating the need for cautious interpretation of the associations. Our findings add to a growing literature on associations between ambient air pollution exposures in the periconceptional period and risk for selected birth defects, though caution in interpretation of these findings is warranted. Extant research renders systematic assessment of study findings problematical due to differences in exposure estimation, timing of exposure, case ascertainment methods, and analytical strategies. Additional research is necessary to establish whether associations found suggest causal links between PM2.5 or benzene (or other relatively common air pollutants) exposures and selected birth defects, and to better understand mechanisms of action including direct or indirect exposures, gene-environment interaction, or epigenetic effects.

Conflict of interest Dr. Kirby served as chair of the scientific advisory committee for the NPlate pregnancy exposure registry for Amgen Corp, and has consulted with Allergan Corp. on pregnancy outcomes in fetuses exposed to Botox during pregnancy. Dr. Kirby has also served as a consultant to AcademyHealth, related to training and technical assistance on record linkage for state health departments. All other authors have no actual or potential competing financial interests to declare.

Submission declaration This manuscript has not been published previously, nor is it currently under review by any other journal. All authors have fulfilled the criteria for authorship and have approved this final draft of the manuscript. If accepted, the work will not be published elsewhere in the same form without the written consent of the copyright-holder.

Funding sources This project was supported by an award from the Centers for Disease Control and Prevention (Grant no U38-EH000941, Florida Environmental Public Health Tracking [EPHT] Network Implementation).

Research concerning human subjects Approval for the study was obtained from the Institutional Review Board at the Florida Department of Health and the University of South Florida.

Acknowledgments We thank Sarah G. Burns and Lauren Young for contributions to the early analyses. The authors would like to acknowledge the use of the services provided by Research Computing, University of South Florida.

Appendix A. Supplementary Information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2015.07. 006.

References Abbott, B.D., Diliberto, J.J., Birnbaum, L.S., 1989. 2,3,7,8-Tetrachlorodibenzo-p-dioxin alters embryonic palatal medial epithelial cell differentiation in vitro. Toxicol. Appl. Pharmacol. 100, 119–131. Agay-Shay, K., Friger, M., Linn, S., Peled, A., Amitai, Y., Peretz, C., 2013. Air pollution and congenital heart defects. Environ. Res. 124, 28–34. Agency for Toxic Substances and Disease Registry, 2007. Toxicological Profile for Benzene. U.S. Public Health Service. U.S. Department of Health and Human Services, Atlanta, GA. Baxter, L.K., Dionisio, K.L., Burke, J., Ebelt Sarnat, S., Sarnat, J.A., Hodas, N., Rich, D.Q., Turpin, B.J., Jones, R.R., Mannshardt, E., Kumar, N., Beevers, S.D., Ozkaynak, H., 2013. Exposure prediction approaches used in air pollution epidemiology studies: key findings and future recommendations. J. Expo. Sci. Environ. Epidemiol. 23, 654–659. Bell, M.L., Belanger, K., 2012. Review of research on residential mobility during pregnancy: consequences for assessment of prenatal environmental exposures. J. Expo. Sci. Environ. Epidemiol. 22, 429–438. Bender, P.L., 2000. Genetics of cleft lip and palate. J. Pediatric Nurs. 15, 242–249. Bennette, C., Vickers, A., 2012. Against quantiles: categorization of continuous variables in epidemiologic research, and its discontents. BMC Med. Res. Methodol. 12, 21. Canfield, M.A., Honein, M.A., Yuskiv, N., Xing, J., Mai, C.T., Collins, J.S., Devine, O., Petrini, J., Ramadhani, T.A., Hobbs, C.A., Kirby, R.S., 2006. National estimates and race/ethnic-specific variation of selected birth defects in the United States, 1999–2001. Birth Defects Res. A Clin. Mol. Teratol. 76, 747–756. CDC, 2014a. Birth Defects Study to Evaluate Pregnancy Exposures (BD-STEPS). http://www.cdc.gov/ncbddd/birthdefects/bd-steps.html. CDC, 2014b. National Birth Defects Prevention Study (NBDPS). http://www.cdc.gov/ ncbddd/birthdefects/nbdps.html. CDC, 2014c. National Environmental Public Health Tracking Network. http://eph tracking.cdc.gov/.

J.P. Tanner et al. / Environmental Research 142 (2015) 345–353 Chang, H.H., Reich, B.J., Miranda, M.L., 2012. Time-to-event analysis of fine particle air pollution and preterm birth: results from North Carolina, 2001–2005. Am. J. Epidemiol. 175, 91–98. Copp, A.J., Greene, N.D., Murdoch, J.N., 2003. The genetic basis of mammalian neurulation. Nat. Rev. Genet. 4, 784–793. Dadvand, P., Ostro, B., Amato, F., Figueras, F., Minguillon, M.C., Martinez, D., Basagana, X., Querol, X., Nieuwenhuijsen, M., 2014. Particulate air pollution and preeclampsia: a source-based analysis. Occup. Environ. Med. Dadvand, P., Rankin, J., Rushton, S., Pless-Mulloli, T., 2011a. Ambient air pollution and congenital heart disease: a register-based study. Environ. Res. 111, 435–441. Dadvand, P., Rankin, J., Rushton, S., Pless-Mulloli, T., 2011b. Association between maternal exposure to ambient air pollution and congenital heart disease: a register-based spatiotemporal analysis. Am. J. Epidemiol. 173, 171–182. Dolk, H., Armstrong, B., Lachowycz, K., Vrijheid, M., Rankin, J., Abramsky, L., Boyd, P. A., Wellesley, D., 2010. Ambient air pollution and risk of congenital anomalies in England, 1991–1999. Occup. Environ. Med. 67, 223–227. Evans, G.W., Kantrowitz, E., 2002. Socioeconomic status and health: the potential role of environmental risk exposure. Annu. Rev. Public Health 23, 303–331. Fosgate, G.T., 2006. Non-differential measurement error does not always bias diagnostic likelihood ratios towards the null. Emerg. Themes Epidemiol. 3, 7. Gilboa, S.M., Mendola, P., Olshan, A.F., Langlois, P.H., Savitz, D.A., Loomis, D., Herring, A.H., Fixler, D.E., 2005. Relation between ambient air quality and selected birth defects, seven county study, Texas, 1997–2000. Am. J. Epidemiol. 162, 238–252. Hansen, C.A., Barnett, A.G., Jalaludin, B.B., Morgan, G.G., 2009. Ambient air pollution and birth defects in brisbane, australia. PLoS One 4, e5408. Hozyasz, K.K., Chelchowska, M., Ambroszkiewicz, J., Gajewska, J., Dudkiewicz, Z., Laskowska-Klita, T., 2004. [Oxidative DNA damage in mothers of children with isolated orofacial clefts]. Przegl Lek 61, 1310–1313. Hwang, B.F., Jaakkola, J.J., 2008. Ozone and other air pollutants and the risk of oral clefts. Environ. Health Perspect 116, 1411–1415. Jerrett, M., Arain, A., Kanaroglou, P., Beckerman, B., Potoglou, D., Sahsuvaroglu, T., Morrison, J., Giovis, C., 2005. A review and evaluation of intraurban air pollution exposure models. J. Expo. Anal. Environ. Epidemiol. 15, 185–204. Jurek, A.M., Greenland, S., Maldonado, G., 2008. How far from non-differential does exposure or disease misclassification have to be to bias measures of association away from the null? Int. J. Epidemiol. 37, 382–385. Jurek, A.M., Greenland, S., Maldonado, G., Church, T.R., 2005. Proper interpretation of non-differential misclassification effects: expectations vs observations. Int. J. Epidemiol. 34, 680–687. Kannan, S., Misra, D.P., Dvonch, J.T., Krishnakumar, A., 2006. Exposures to airborne particulate matter and adverse perinatal outcomes: a biologically plausible mechanistic framework for exploring potential effect modification by nutrition. Environ. Health Perspect 114, 1636–1642. Kirkwood, B., Sterne, J., 2003. Essentials Medical Statistics, 2nd Ed. Blackwell Science Ltd.. Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., Behar, J. V., Hern, S.C., Engelmann, W.H., 2001. The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J Expo Anal Environ Epidemiol. 11, 231–252. Le, H.Q., Batterman, S.A., Wirth, J.J., Wahl, R.L., Hoggatt, K.J., Sadeghnejad, A., Hultin, M.L., Depa, M., 2012. Air pollutant exposure and preterm and term small-forgestational-age births in Detroit, Michigan: long-term trends and associations. Environ. Int. 44, 7–17. Lupo, P.J., Symanski, E., Chan, W., Mitchell, L.E., Waller, D.K., Canfield, M.A., Langlois, P.H., 2010. Differences in exposure assignment between conception and delivery: the impact of maternal mobility. Paediatric Perinat. Epidemiol. 24, 200–208. Lupo, P.J., Symanski, E., Waller, D.K., Chan, W., Langlois, P.H., Canfield, M.A., Mitchell, L.E., 2011. Maternal exposure to ambient levels of benzene and neural tube defects among offspring: Texas, 1999–2004. Environ. Health Perspect 119, 397–402. Marshall, E.G., Harris, G., Wartenberg, D., 2010. Oral cleft defects and maternal exposure to ambient air pollutants in New Jersey. Birth Defects Res. A Clin. Mol. Teratol. 88, 205–215. Mathews, T.J., MacDorman, M.F., 2012. Infant mortality statistics from the 2008 period linked birth/infant death data set. Natl. Vital Stat. Rep. 60, 1–27. Meng, Z., Liu, Y., 2007. Cell morphological ultrastructural changes in various organs from mice exposed by inhalation to sulfur dioxide. Inhal. Toxicol. 19, 543–551. Meng, Z., Qin, G., Zhang, B., Geng, H., Bai, Q., Bai, W., Liu, C., 2003. Oxidative damage of sulfur dioxide inhalation on lungs and hearts of mice. Environ. Res. 93, 285–292. Merritt, L., 2005. Understanding the embryology and genetics of cleft lip and palate. Adv. Neonatal. Care 5, 64–71. Miller, A., Siffel, C., Correa, A., 2010. Residential mobility during pregnancy: patterns and correlates. Matern Child Health J. 14, 625–634. Millicovsky, G., Johnston, M.C., 1981. Hyperoxia and hypoxia in pregnancy: simple experimental manipulation alters the incidence of cleft lip and palate in CL/Fr mice. Proc. Natl. Acad. Sci. USA 78, 5722–5723. Morello-Frosch, R., Jesdale, B.M., Sadd, J.L., Pastor, M., 2010. Ambient air pollution exposure and full-term birth weight in California. Environ. Health 9, 44. National Birth Defects Prevention Network (NBDPN), 2004. Guidelines for Conducting Birth Defects Surveillance. Sever, L., ed. Atlanta, GA: National Birth Defects Prevention Network, Inc. http://www.nbdpn.org/docs/NBDPN_Guide lines2008.pdf. Ozkaynak, H., Baxter, L.K., Dionisio, K.L., Burke, J., 2013. Air pollution exposure

353

prediction approaches used in air pollution epidemiology studies. J. Expo. Sci. Environ. Epidemiol. 23, 566–572. Padula, A.M., Tager, I.B., Carmichael, S.L., Hammond, S.K., Lurmann, F., Shaw, G.M., 2013a. The association of ambient air pollution and traffic exposures with selected congenital anomalies in the San Joaquin Valley of California. Am. J. Epidemiol. 177, 1074–1085. Padula, A.M., Tager, I.B., Carmichael, S.L., Hammond, S.K., Yang, W., Lurmann, F., Shaw, G.M., 2013b. Ambient air pollution and traffic exposures and congenital heart defects in the San Joaquin Valley of California. Paediatric Perinat. Epidemiol. 27, 329–339. Padula, A.M., Tager, I.B., Carmichael, S.L., Hammond, S.K., Yang, W., Lurmann, F.W., Shaw, G.M., 2013c. Traffic-related air pollution and selected birth defects in the San Joaquin Valley of California. Birth Defects Res. A Clin. Mol. Teratol. 97, 730–735. Ramakrishnan, A., Lupo, P.J., Agopian, A.J., Linder, S.H., Stock, T.H., Langlois, P.H., Craft, E., 2013. Evaluating the effects of maternal exposure to benzene, toluene, ethyl benzene, and xylene on oral clefts among offspring in Texas: 1999–2008. Birth Defects Res. A Clin. Mol. Teratol. 97, 532–537. Rankin, J., Chadwick, T., Natarajan, M., Howel, D., Pearce, M.S., Pless-Mulloli, T., 2009. Maternal exposure to ambient air pollutants and risk of congenital anomalies. Environ. Res. 109, 181–187. Rich, D.Q., Demissie, K., Lu, S.E., Kamat, L., Wartenberg, D., Rhoads, G.G., 2009. Ambient air pollutant concentrations during pregnancy and the risk of fetal growth restriction. J. Epidemiol. Commun. Health 63, 488–496. Ritz, B., Wilhelm, M., 2008. Ambient air pollution and adverse birth outcomes: methodologic issues in an emerging field. Basic Clin. Pharmacol. Toxicol. 102, 182–190. Ritz, B., Yu, F., Fruin, S., Chapa, G., Shaw, G.M., Harris, J.A., 2002. Ambient air pollution and risk of birth defects in Southern California. Am. J. Epidemiol. 155, 17–25. Rudra, C.B., Williams, M.A., Sheppard, L., Koenig, J.Q., Schiff, M.A., 2011. Ambient carbon monoxide and fine particulate matter in relation to preeclampsia and preterm delivery in western Washington State. Environ. Health Perspect 119, 886–892. Sadler, T.W., 2012. Langman’s Medical Embryology. Lippincott Williams & Wilkins, Philadelphia. Salemi, J.L., Tanner, J.P., Block, S., Bailey, M., Correia, J.A., Watkins, S.M., Kirby, R.S., 2011. The relative contribution of data sources to a birth defects registry utilizing passive multisource ascertainment methods: does a smaller birth defects case ascertainment net lead to overall or disproportionate loss? J. Registry Manag. 38, 30–38. Salemi, J.L., Tanner, J.P., Kennedy, S., Block, S., Bailey, M., Correia, J.A., Watkins, S.M., Kirby, R.S., 2012. A comparison of two surveillance strategies for selected birth defects in Florida. Public Health Rep. 127, 391–400. Schembari, A., Nieuwenhuijsen, M.J., Salvador, J., de Nazelle, A., Cirach, M., Dadvand, P., Beelen, R., Hoek, G., Basagana, X., Vrijheid, M., 2014. Traffic-related air pollution and congenital anomalies in Barcelona. Environ. Health Perspect 122, 317–323. Stingone, J.A., Luben, T.J., Daniels, J.L., Fuentes, M., Richardson, D.B., Aylsworth, A.S., Herring, A.H., Anderka, M., Botto, L., Correa, A., Gilboa, S.M., Langlois, P.H., Mosley, B., Shaw, G.M., Siffel, C., Olshan, A.F., 2014. Maternal exposure to criteria air pollutants and congenital heart defects in offspring: results from the national birth defects prevention study. Environ. Health Perspect 122, 863–872. Strickland, M.J., Klein, M., Correa, A., Reller, M.D., Mahle, W.T., Riehle-Colarusso, T.J., Botto, L.D., Flanders, W.D., Mulholland, J.A., Siffel, C., Marcus, M., Tolbert, P.E., 2009. Ambient air pollution and cardiovascular malformations in Atlanta, Georgia, 1986–2003. Am. J. Epidemiol. 169, 1004–1014. Vinikoor-Imler, L.C., Davis, J.A., Meyer, R.E., Luben, T.J., 2013. Early prenatal exposure to air pollution and its associations with birth defects in a state-wide birth cohort from North Carolina. Birth Defects Res A Clin. Mol. Teratol. 97, 696–701. Vrijheid, M., Martinez, D., Manzanares, S., Dadvand, P., Schembari, A., Rankin, J., Nieuwenhuijsen, M., 2011. Ambient air pollution and risk of congenital anomalies: a systematic review and meta-analysis. Environ. Health Perspect. 119, 598–606. Wang, D., Yu, X., 2004. [Morphological study on the neural tube defects caused by passive smoking]. Wei Sheng Yan Jiu. 33, 147–150. Wilhelm, M., Ghosh, J.K., Su, J., Cockburn, M., Jerrett, M., Ritz, B., 2012. Traffic-related air toxics and term low birth weight in Los Angeles County, California. Environ. Health Perspect. 120, 132–138. Wong, D.W., Yuan, L., Perlin, S.A., 2004. Comparison of spatial interpolation methods for the estimation of air quality data. J. Expo. Anal. Environ. Epidemiol. 14, 404–415. Woodruff, T.J., Parker, J.D., Kyle, A.D., Schoendorf, K.C., 2003. Disparities in exposure to air pollution during pregnancy. Environ. Health Perspect 111, 942–946. Wu, J., Ren, C., Delfino, R.J., Chung, J., Wilhelm, M., Ritz, B., 2009. Association between local traffic-generated air pollution and preeclampsia and preterm delivery in the south coast air basin of California. Environ. Health Perspect 117, 1773–1779. Wu, J., Wilhelm, M., Chung, J., Ritz, B., 2011. Comparing exposure assessment methods for traffic-related air pollution in an adverse pregnancy outcome study. Environ. Res. 111, 685–692. Zhu, H., Kartiko, S., Finnell, R.H., 2009. Importance of gene-environment interactions in the etiology of selected birth defects. Clin. Genet. 75, 409–423.