Environment International 144 (2020) 106047 Contents lists available at ScienceDirect Environment International journal homepage: www.elsevier.com/locate/envint Prenatal greenspace exposure and cord blood cortisol levels: A cross- T sectional study in a middle-income country Lilian Marie Bolla,b, Ramezanali Khamirchic, Lucia Alonsoa,d,e, Elisa Llurbaf,g,h, Óscar J Pozoi, ⁎ ⁎ Mohammad Miric, , Payam Dadvanda,d,e, a ISGlobal, Barcelona, Spain b University of Southern Denmark (SDU), Esbjerg, Denmark c Non-communicable Diseases Research Center, Department of Environmental Health, School of Public Health, Sabzevar University of Medical Sciences, Sabzevar, Iran d Universitat Pompeu Fabra (UPF), Barcelona, Spain e CIBER Epidemiología y Salud Pública (CIBERESP), Spain Maternal and Child Health Development Network, Madrid, Spain f Obstetrics and Gynaecology Department, Santa Creu i Sant Pau Hospital, Barcelona, Spain g Universitat Autónoma de Barcelona (UAB), Spain h Maternal and Child Health Development Network II (SAMID II) RD16/0022/0015SEP, Institute of Health Carlos III, 28029 Madrid, Spain i Integrative Pharmacology and Systems Neuroscience Group, IMIM, Hospital del Mar, Barcelona, Spain A R T I C LE I N FO A B S T R A C T Handling Editor: Adrian Covaci Exposure to greenspace has been associated with reduced stress; however, the available evidence on such an Keywords: association for the fetus is still very scarce. We, for the first time, investigated the association between maternal Pregnancy greenspace exposure and the level of cortisol, a stress hormone, in the cord blood. Our study was based on a Psychological distress cohort of 150 pregnant women in Sabzevar, Iran (2018). We comprehensively assessed greenspace exposure for Natural environment each participant through (i) residential surrounding greenspace (using two satellite-derived vegetation indices), Parks (ii) residential proximity to green spaces, (iii) maternal visual access to greenspace, (iv) use of public and private Glucocorticoid green spaces, (v) having a private garden, and (vi) the number of plant pots at home. Linear regression models Newborn were developed to assess the association of each indicator of greenspace exposure with cord blood cortisol levels, controlled for the relevant covariates. We observed that a higher residential surrounding greenspace (100 m buffer), having a window with greenspace view, window greenspace coverage of more than 50%, frequently looking at greenspace through window, residential proximity to large green spaces, and more time spent in green spaces were associated with lower cortisol levels in the cord blood. The findings for residential surrounding greenspace at 300 m and 500 m buffers, residential proximity to any green space regardless of its size, having a private garden, and number of plant pots at home were not conclusive. While about one-third of the association between residential surrounding greenspace (100 m buffer) could be mediated through reduction in exposure to air pollution, we did not observe any strong evidence for such a mediatory role for the visual access to green- space. The findings stratified for parental education and housing type showed mixed patterns. Our findings suggest that more greenspace exposure might reduce cortisol level in the cord blood. 1. Introduction 2017), and heat (Shahmohamadi et al., 2011), which could detrimen- tally influence physical and mental health and wellbeing The ongoing urbanization is projected to lead about two-third of the (Shahmohamadi et al., 2011; Montes-González et al., 2018). In contrast, world population living in urban areas by 2050 (United Nations, 2018). urban greenspace has been associated with improved physical and Urban living is often associated with more stressful and sedentary mental health and wellbeing (Dadvand et al., 2016; Lachowycz and lifestyle and higher exposure to urban-related environmental hazards Jones, 2011; Amoly et al., 2015), partly by reducing exposure to air such as air pollution (Carlsten and Rider, 2017), noise (McMahon et al., pollution (Janhäll, 2015), heat (Aram et al., 2019), and noise Abbreviations: LMIC, low- and middle-income countries; BMI, body mass index; ELISA, enzyme-linked immunosorbent assay; ESPA, earth resources observation and science (EROS) center science processing architecture; NDVI, normalized difference vegetation index; MSAVI2, modified soil-adjusted vegetation index 2; SES, socioeconomic status ⁎ Corresponding authors at: Institute of Global Health of Barcelona, (ISGlobal) Carrer Dr. Aiguader 88, 08003-Barcelona, Spain (P. Dadvand). E-mail addresses: M_miri87@ssu.ac.ir (M. Miri), payam.dadvand@isglobal.org (P. Dadvand). https://doi.org/10.1016/j.envint.2020.106047 Received 26 February 2020; Received in revised form 5 August 2020; Accepted 6 August 2020 Available online 18 August 2020 0160-4120/ © 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). L.M. Boll, et al. Environment International 144 (2020) 106047 perception (Dzhambov and Dimitrova, 2015), increasing physical ac- did not work outside home (i.e. housewives), (vii) spent their entire tivity (Coombes et al., 2010), enhancing social contacts (Dadvand et al., pregnancy in Sabzevar and (viii) did not change their home address 2016; Jennings and Bamkole, 2019), and reducing stress (Gong et al., during their current pregnancy. Of the women who were approached, 2016). A substantial body of experimental evidence has shown that 150 mothers met these criteria and agreed to participate in the study short exposure to greenspace (either real greenspace or its pictures or after being informed about the research aims and steps. Data on de- videos) could have beneficial short-term effects on the stress level (Kim mographic characteristics, socioeconomic status, and lifestyle were et al., 2010; Lee et al., 2011; Tsunetsugu et al., 2013). Moreover, a obtained through questionnaires by face-to-face interviews. The parti- limited body of epidemiological evidence is suggestive for a long-term cipants were enrolled in the study after signing the consent form that beneficial association between greenspace exposure and stress was approved by the Clinical Research Ethical Committee (Triguero-Mas et al., 2017). However, the available evidence on the (IR.MEDSAB.REC.1395.82) of the Sabzevar University of Medical association between prenatal exposure to greenspace and stress level in Science, Sabzevar, Iran. the developing fetus is still very scarce. The corticoid hormone cortisol is essential for our stress response 2.3. Exposure measurement and is produced by the adrenal cortex as an end product of the hy- pothalamic-pituitaryadrenal axis. It regulates the insulin production, We comprehensively characterized six different aspects of exposure activates the catabolic metabolism, and suppresses the immune system to greenspace including: (1) residential surrounding greenspace, (2) to reduce inflammation. However, a long-term overexposure to cortisol visual access to greenspace, (3) the time spent in green spaces, (4) re- can disturb the hormone balance. For the fetus, a sustained excessive sidential access to green space, (5) having a private garden, and (6) the level of cortisol could be associated with higher risk of adverse preg- number of plant pots at home. nancy outcomes such as lower head circumference and birth weight (Cherak et al., 2019; Su et al., 2015), altered brain structure, and 2.3.1. Residential surrounding greenspace postnatal behavioral disorders (Reynolds, 2013). We hypothesized that To measure the residential surrounding greenspace, we used the exposure to greenspace during pregnancy could be inversely associated Normalized Difference Vegetation Index (NDVI). It quantifies the den- with cortisol level in the cord blood, an indicator of the level of this sity of photosynthetically active greenspace by measuring the propor- hormone in the fetal circulation. To our knowledge, there is no study tion of near-infrared light (strongly reflected by vegetation) and visible reporting on such an association. light. Its values range from −1 to 1 with higher values indicating more To date, much of the available studies on health benefits of green- vegetation. To achieve maximum exposure contrast, we looked for space exposure have been conducted in high-income countries. As dif- available cloud-free Landsat 8 images during the spring months (i.e. the ferences in climate, native vegetation types, and cultures constitute maximum vegetation period of the year for our study region) of 2018 different types of green space between high-income and low- and (our study year) from the earth resources observation and science middle-income countries (LMICs), the transfer of evidence from high- (EROS) center science processing architecture (ESPA) (ESPA, 2017). income countries to LMICs could be challenging. Apart from the design Based on this search, we obtained NDVI data for our study region using aspects and vegetation types that could vary among different countries, the image captured by Landsat 8 on June 14th 2018 at 30 m × 30 m maintenance of these spaces could differ between high-income coun- resolution. For each participant, the residential surrounding greenspace tries and LMICs due to variations in available resources. Up to the was abstracted as the average NDVI within a buffer of 100 m, 300 m present, little is known on the health effect of green spaces in LMICs and 500 m (Dadvand et al., 2017) around her geocoded residential where much of the urbanization is predicted to happen. This study address. aimed to investigate the association between a comprehensive array of different indicators of greenspace exposure on the level of the cortisol 2.3.2. Visual access in the cord blood of newborns in Iran, a LMIC. We obtained data on visual access to greenspace through three questions: (i) Can you see plants, trees, grass, flowers, etc. from any 2. Methods and Material window of your home? with possible answers being yes/no; (ii) How often do you see the vegetation through the window (s)? with possible 2.1. Study setting answers being rarely, sometimes, and always. Due to a small number of answers in some categories, we dichotomized the answers to this This study was conducted in Sabzevar, a city of around 240 000 question by having “rarely” together with those who answered “no” to inhabitants (Census 2016 (Iran, 2016)) in the north-east of Iran (co- the previous question as the baseline category and “sometimes” and ordinates: 36°12′ N 57°35′, elevation: 977 m). It has an arid climate “always“ as the other category; and (iii) If yes, what proportion of the (Kottek et al., 2006) with an annual precipitation of 180 mm (mostly window surface is covered by vegetation (If there are multiple win- occurring in winter) and an annual humidity of 43% (Center, 2020). dows, please describe the window most used)? with answers being The annual average temperature is 18 °C, with average monthly 25%, 50% and 100%. To achieve enough number of participants in minimum and maximum temperatures ranging between −2 and 45 °C each category, we dichotomized the answer by considering those who (Center, 2020). Sabzevar has four distinct seasons with maximum ve- did not have a window with green view together with those who an- getation occurring during the spring months. Fig. S1 in the supple- swered 25% as baseline category, and those who answered 50% and mentary materials presents the location of the study area. 100% as the other category. 2.2. Study population 2.3.3. Residential access to green space The residential access to green space was measured as the Euclidean This cross-sectional study was based on a sample of pregnant distance of the participant’s geocoded residential address to (i) the women recruited at their delivery time at the Shahidan Mobini uni- nearest green space of any area and (ii) the nearest green space with an versity hospital (the only tertiary hospital in the Sabzevar) during June area of at least a 5000 m2. The land use map of Sabzevar (2016) pro- and August 2018. We included women who: (i) were non-smokers and vided by the municipality of Sabzevar was used to calculate these did not smoke during pregnancy, (ii) had a normal term pregnancy variables for each participant’s home address. (37–42 weeks), (iii) had a normal vaginal delivery, (iv) had no history of diabetes or hypertension, (v) had no pregnancy complications (i.e. 2.3.4. Having a private garden gestational diabetes, pregnancy-induced hypertensive disorders), (vi) We asked participants whether they have a private garden with 2 L.M. Boll, et al. Environment International 144 (2020) 106047 possible answers being yes/no. surrounding greenspace (100 m buffer) and the cord blood cortisol level could be partly explained through the visual access to greenspace. We 2.3.5. Use of green spaces also hypothesized that reduction in the ambient air pollution levels at Questionnaires already applied in previous studies (Amoly et al., the participant’s residence could partly explain the aforementioned 2015; Dadvand et al., 2018) were used to obtain data on the total time association. The level of air pollution was assessed as the level of par- that each participant spent in green spaces during her pregnancy. We ticular matter pollutant with aerodynamic diameter of 1 µm (PM1), asked participants about the average time (hours per week) spent se- 2.5 μm (PM2.5) and 10 µm (PM10) at the participants’ residential ad- parately in public green spaces (e.g. parks, forest) and in private green dresses estimated using a land use regression (LUR) models developed spaces in a normal week during their pregnancy. for Sabzevar and described elsewhere (Miri et al., 2019). Briefly, these models were developed based on data from 26 monitoring stations that 2.3.6. Number of indoor plant pots at home measured PM1, PM2.5, and PM10 levels during four campaigns/seasons The participants were asked to report the number of indoor plant from April 2017 till March 2018. Five categories of potential predictor pots at home during the pregnancy period. variables were applied to develop these LUR models: urban mor- phology, traffic, population density, geographic location of monitoring 2.4. Cortisol level in the umbilical cord blood station and land use. The final models were able to explain 68%, 71% and 75% of the variation of the annual average of PM1, PM2.5 and PM10 To determine cortisol level, 4 mL blood was drawn from the um- in the study area. Following the approach of Baron and Kenny (Baron bilical cord after delivery. The sample was allowed to clot in a serum- and Kenny, 1986), we fitted separately a model for the outcome given separating tube with clot activator at room temperature for 30 min, the mediator and the exposure, and a model for the mediator given the before being separated from the serum by centrifugation at 3000 rpm exposure to compute the percentage of the association between re- for 15 min. Afterwards, the serum samples were stored at −80 °C till sidential surrounding greenspace and cortisol level that could be the time of analysis. The level of cortisol in the blood samples was then mediated through each indicator of visual access to greenspace and the analysed by an enzyme-linked immunosorbent assay (ELISA) using the air pollutant PM2.5 (one at a time). Bootstrap was used to obtain per- Cortisol Kit (LDN, Nordhorn, Germany). Laboratory assays followed centile-based 95% confidence intervals for the different estimates. manufactorýs instructions. All samples were measured in duplicate and all measured values were in linear range (0.2–800 ng/mL). The coef- 2.7. Further analysis ficient of variation (CV) for intra- and inter-assay variability for the average cortisol level in our analyses (43.9 ng/mL) were 7.9% and 2.7.1. Potential effect modifiers 5.8%, respectively, which agree with the available recommendations We evaluated the statistical significance of the multiplicative in- (EMA, 2012; Food and Drug Administration, 2018). teraction term between each indicator of greenspace exposure and housing type (apartment vs. single-family house) as well as parental 2.5. Main analysis education (one at a time) and fetal sex (girls/boys), using likelihood ratio test, comparing fully adjusted models with and without multi- We developed linear regression models to estimate the change in the plicative interaction term between each indicator of greenspace ex- cord blood cortisol levels associated with a one-interquartile range posure and each of the aforementioned potential effect modifiers. We (IQR) increase in each of the continuous indicators of greenspace ex- then stratified the analyses based on these two factors. For the parental posure (one at a time). We did not include all the indicators of green- education, a new variable was created containing the education status space exposure in one model to prevent collinearity. Given that the first from the parent with the higher educational degree (elementary school/ and third quartiles of the time spent in private green spaces were both high school/university). zero, it was not possible to scale this exposure and hence we conducted the analyses only for the total time spent in green spaces (hour per 2.7.2. Soil-adjusted residential surrounding greenspace week) as the sum of the time spent in public and private green spaces. We applied Modified Soil Adjusted Vegetation Index 2 (MSAVI2) (Qi For the visual access to greenspace, the indicators were included as a et al., 1994) to abstract an alternative set of estimates for the residential binary categorical variable in the models with baseline category in- surrounding greenspace. MSAVI2, compared to NDVI, has the ad- dicating less visual access to greenspace. vantage of controlling for background soil reflectance and has been Based on the available literature, we identified common predictors shown to be a proper measure for characterizing sparse vegetation in of greenness exposure and cord blood cortisol levels as presented in a arid areas of Iran (Barati et al., 2011). We applied the same Landsat direct acyclic graph (DAG) in the supplementary materials (Fig. S2). As image from ESPA (ESPA, 2017) used for calculating NDVI to abstract such, we adjusted all models for the maternal age (years, continuous) MSAVI2 and then characterized residential surrounding greenspace as and body mass index (BMI, kg/m2, continuous), the fetal sex (girls/ the average of MSAVI2 across buffers of 100 m, 300 m and 500 m. We boys), parity (continuous), gestational age at delivery (days, con- repeated our analyses for residential surrounding greenspace using this tinuous), being exposed to tobacco at home (yes/no), three indicators of alternative set of exposure. household socioeconomic status (SES) and two indicators of neighbor- hood SES. We used income (≥15 million Rials per month/15 to 30 2.7.3. Residential surrounding tree cover million Rials per month/30 ≤ million Rials per month), paternal edu- In order to disentangle the association of different types of green- cation (elementary school/high school/university) and maternal edu- space, we used Vegetation Continuous Fields (VCF) as a measure of tree cation (elementary school/high school/university) as indicators of the coverage. The VCF is an estimate of the percentage of land in each household SES and the percentage of people with no completed degree image pixel covered by woody vegetation (i.e. trees) (Sexton et al., and the percentage of unemployment per census tract (based on 2016 2013). We abstracted the VCF data for our study region from Landsat census) as indicators of the neighborhood SES. All statistical analyses images from 2015 provided by the Land Processes Distributed Active were performed using STATA software version 15, considering a sig- Archive Center (LP DAAC) at 30 m resolution (LP DAAC, 2017). For nificance level of 0.05 (StataCorp, 2017). each participant we calculated the average percent of tree coverage across buffers of 100 m, 300 m and 500 m around her residential ad- 2.6. Mediation analysis dress. We repeated our analyses using this alternative exposure. Given that the first and third quartiles of VCF at a buffer of 100 m were both We hypothesized that the association of the immediate residential zero, it was not possible to scale this exposure. Therefore, we reported 3 L.M. Boll, et al. Environment International 144 (2020) 106047 Table 1 visual access to greenspace. Descriptive statistics of the study participants (n = 150).a Variable Description 3.2. Main analysis Age of the mother (years) 27 (8) We observed a significant, inverse association between residential BMI of the mother (kg/m2) 23.6 (5.8) surrounding greenspace (NDVI) at a 100 m buffer and cortisol levels Number of pregnancies 2 (1) Gestational age (days) 278 (13) (Table 2). A one-IQR increase in the residential surrounding greenspace Fetal Sex (girls) 78 (52%) was associated with a decrease of 3.18 ng/ml (95% confident intervals Census tract illiteracy (%) 24.5 (15.2) (CIs): −5.43; −0.93) in the cord blood cortisol level. Additionally, an Census tract unemployment (%) 7.0 (4.3) IQR increase in distance to a green space larger than 5,000 m2 was Exposure to the environmental tobacco smoke at home (Yes) 29 (19.3%) significantly associated with an increase of 3.56 ng/ml (95% CIs: 0.32; Parental education 6.80) in the cord blood cortisol level. Moreover, we observed an inverse Low 63 (42%) association between an IQR increase of the time spent in green spaces Middle 56 (37.3%) High 31 (20.7%) and the cord blood cortisol levels by −2.27 ng/ml (95% CI: −3.88; −0.67). Table 2 further presents the associations between indicators of Maternal education Low 43 (28.7%) visual access to greenspace and cortisol level in the cord blood. Having Middle 65 (43.3%) a window with greenspace view was associated with a 3.67 ng/ml (95% High 42 (28.0%) CI: −7.83; 0.49) lower cortisol levels (p-value = 0.06). Having more Income than half of the window covered by greenspaces was significantly as- Low 104 (69.3%) sociated with lower cortisol levels in the cord blood (β = -7.11 ng/ml Middle 36 (24.0%) (95% CI: −11.90; −2.32). Looking at greenspace through a window at High 10 (6.7%) least sometimes, compared to never or rarely, was associated with a Cortisol (ng/ml) 43.90 (19.00) change of −4.69 ng/ml (95% CI: −9.09; −0.28) in blood cortisol level. Residential surrounding greenspace (NDVI) Having a private garden, residential proximity to a green space of any 100 m buffer 0.066 (0.016) 300 m buffer 0.070 (0.014) size, and the residential surrounding greenspace (NDVI) over buffers of 500 m buffer 0.073 (0.017) 300 m and 500 m were also inversely associated with cortisol levels in Soil-adjusted residential surrounding greenspace (MSAVI2) the cord blood; however, none of these associations attained statistical 100 m buffer 0.06 (0.01) significance. Our observed association for the number of plant pots at 300 m buffer 0.07 (0.01) home was null. 500 m buffer 0.07 (0.01) Residential surrounding tree cover (VCF) 3.3. Mediation analysis 300 m buffer 0.009 (0.025) 500 m buffer 0.017 (0.034) Table 3 presents the percentage of the association between the re- Distance to the nearest green space (m) sidential surrounding greenspace (NDVI) across a 100 m buffer and the of any size 92.3 (112.5) cord blood cortisol levels that could be explained by each indicator of of at least 5.000 m2 190.1 (245.2) visual access as well as air pollution. PM2.5 and PM10 could explain Time spent in total (h/week) 4 (2.5) about one-third (28.6% (95% CIs: −1.6; 128.9) and 32.7 (95% CIs: Number of pots at home 2 (2) −7.0 and 160.9, respectively)) of the aforementioned association, Having a window with greenspace view (Yes) 85 (56.7%) Frequently looking at greenspace (Sometimes/always) 48 (32.0%) while PM1 could only explain about 5.1% (95% Cis: −2.9; 18.8) of this 50% or more of window covered by greenspace (Yes) 35 (23.3%) association. Having a window with greenspace view, frequent viewing Having a private garden (Yes) 64 (42.7%) of greenspace, and more than half of the window(s) covered by PM1 (μg/m3) 39.2 (7.9) greenspace could respectively explain only 5.3% (95% CIs: −23.9; PM2.5 (μg/m3) 44.4 (9.8) PM10 (μg/m3) 49.0 (12.0) 26.3), 8.3% (95% CIs: −63.2; 19.0), and 3.2% (95% CIs: −10.3; 21.1) of the aforementioned association. a Categorical variables are presented as count (%) and continuous variables as median (interquartile range). 3.4. Further analysis the associations only for buffers of 300 m and 500 m. 3.4.1. Potential effect modifiers We tested the statistical significance of the interaction term between 3. Results each indicator of greenspace exposure and the parental education, housing type or fetal sex in association with the cord blood levels of the 3.1. Study population cortisol. As presented in supplementary Table S1, among our evaluated interaction terms between each indicator of greenspace exposure and The descriptive statistics of the sociodemographic characteristics of parental education, housing type, and fetal sex, we only observed a the study participants and their greenspace exposure and cord blood statistically significant interaction between parental education and cortisol levels are presented in Table 1. The median (IQR) of the cord distance to a green space larger than 5,000 m2 (p-value = 0.02). Other blood cortisol levels was 43.9 (19.0) ng/ml. As presented in Fig. S3 (see interaction terms were not statistically significant (p-value > 0.10). supplementary materials) there was a relatively small variation in re- Fig. 2 presents the associations between exposure to greenspace and sidential surrounding greenspace based on NDVI. The correlations be- cord blood cortisol levels stratified for parental education, the partici- tween the exposure variables are presented in Fig. 1. We found a very pants’ housing type, and fetal sex. For having more than half of the strong correlation between the two measurements of residential sur- window covered with a green view and having a private garden we rounding greenspace (NDVI and MSAVI2) at each buffer size. Re- observed some suggestions for a trend with potentially stronger asso- sidential surrounding greenspace, proximity to green spaces (any size ciations for participants with lower parental education. In contrast, for and those larger than 5.000 m2) and having a private garden were the residential surrounding greenspace (NDVI) across buffers of 300 m moderately correlated with the use of green space. We further observed and 500 m, time spent in a green space and residential proximity to a moderate correlation between residential surrounding greenspace and green spaces larger than 5000 m2, there were some indications for 4 L.M. Boll, et al. Environment International 144 (2020) 106047 Fig. 1. Scatterplot matrix and histograms of the continuous Greenspace variables. Bivariate scatterplots of continuous variables below the diagonal; histograms of each variable on the diagonal; Spearman’s correlation coefficient above the diagonal. Note: NDVI = normalized difference vegetation index within buffers of 100 m, 300 m and 500 m; MSAVI2 = modified soil-adjusted vegetation index within buffers of 100 m, 300 m and 500 m; VCF = Vegetation Continuous Fields within buffers of 300 m and 500 m; Time spent in GS = Total time spent in green spaces; Plant Pots = Number of Indoor Plant Pots; Distance to any GS = Distance to a green space of any size; Distance to a large GS = Distance to a green space < 5000 m2. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) potentially stronger associations for those participants with higher based on NDVI. We observed an inverse association between average parental education. No clear trend was observed for other greenspace MSAVI2 across a 100 m buffer around the participant’s residential ad- exposure indicators across strata of parental education. dress and the cord blood cortisol level (Table 2). An increase in average When we stratified the analyses based on the participants’ housing MSAVI2 in this buffer by one IQR was associated with a decrease in the type, we observed some suggestions for potentially stronger associa- cord blood cortisol level by 3.66 ng/ml (95% CIs: −6.04; −1.29). Si- tions between the residential proximity to green spaces (any size and milarly, for the MSAVI2 averages over buffers of 300 m and 500 m, we those larger than 5.000 m2), having a private garden, and cord blood observed inverse associations with the cord blood cortisol levels; cortisol level for participants living in an apartment compared to a however, these associations were not statistically significant. single-family house. In contrast, for the indicators of the visual access to greenspace, we generally observed some indications for a potentially 3.4.3. Residential surrounding tree cover stronger association with cord blood cortisol levels for the participants Similar to our findings for NDVI and MSAVI2, we observed inverse living in a single-family house than those living in an apartment. The associations between residential surrounding tree cover (i.e. VCF results for the residential surrounding greenspace were mixed with averages across 300 m and 500 m buffers around the participant’s re- some suggestions for a stronger association for the residential sur- sidential address) and cord blood cortisol levels; however, none of these rounding greenspace at 100 m buffer for those participants living in a associations attained statistically significance. single-family house and a stronger association for the residential sur- rounding greenspace at 500 m buffer for those living in an apartment. 4. Discussion For other measures of greenspace exposure, we did not observe any clear pattern. To our knowledge, this cross-sectional study is the first to evaluate After stratifying the analyses based on the fetal sex, we observed the association of maternal exposure to greenspace during pregnancy some suggestions for a potentially stronger association for the visual and cortisol level in the cord blood. Furthermore, this study is the first access to greenspace (window area covered by greenspace and looking to evaluate the association of having access to a private garden and the at greenspace frequently) and for residential proximity to large green number of plant pots at home on an indicator of stress. This study also spaces among girls. We did not observe any notable sex difference in adds to the very scarce available epidemiological evidence on health our evaluated associations for other indicators of greenspace exposure. benefits of visual access to greenspace. Moreover, this is the first study to investigate the relationship between pregnancy exposure to green- 3.4.2. Soil-adjusted residential surrounding greenspace space and cortisol levels in a LMIC. The findings of this study also adds The results of the analyses for the residential surrounding green- to the limited number of available reports on health benefits of the space based on MSAVI2 were consistent with those of the main analysis greenspace exposure in LMICs where much of the urbanization is 5 L.M. Boll, et al. Environment International 144 (2020) 106047 Table 2 could be mediated through reduction in exposure to PM2.5 and PM10, Adjusteda associations between prenatal exposure to greenspace and cord blood we did not observe any strong evidence for such a mediatory role for cortisol level. the exposure to PM1 and the visual access to greenspace. We observed Estimate (95%) p mixed patterns of associations across the strata of parental education, housing type, and fetal sex. Residential surrounding greenspace (NDVI)b 100 m buffer −3.18 (−5.43; −0.93) 0.01 4.1. Available evidence 300 m buffer −0.91 (−2.79; 0.98) 0.34 500 m buffer −1.05 (−3.44; 1.35) 0.39 Given that this is the first study looking at the association between Soil-adjusted residential surrounding greenspace (MSAVI2)b 100 m buffer −3.66 (−6.04; −1.29) < 0.01 greenspace exposure and cortisol level in cord blood, no previous stu- 300 m buffer −1.54 (−3.39; 0.31) 0.10 dies can be used as comparison. However, our findings line up with past 500 m buffer −1.84 (−4.45; 0.76) 0.16 studies finding an association between greenspace exposure and re- Residential surrounding tree cover (VCF)b duced stress level in adults. Several experimental studies have ex- 300 m buffer −0.08 (−1.54; 1.38) 0.91 amined the inverse associations between short-term exposure to 500 m buffer −0.45 (−2.81; 1.31) 0.48 greenspace and physiological indicators of stress. Tyrväinen et al., for Distance to green spaceb example, found an accelerated reduction in perceived stress in adults of any size 1.54 (−1.32; 4.39) 0.29 walking in natural areas compared to the city center (Tyrväinen et al., of at least 5,000 m2 3.56 (0.32; 6.80) 0.03 2014). Ulrich et al. (1991) observed an association between short-term Total time spend in green space b −2.27 (−3.88; −0.67) 0.01 exposure to greenspace sceenery and physiological stress indicators, Number of plant pots at home b 0.14 (−1.70; 1.98) 0.88 which has been replicated by several other studies in recent years (Kim Have a window with greenspace view −3.67 (−7.83; 0.49) 0.08 Frequently looking at greenspace −4.69 (−9.09; −0.28) 0.04 et al., 2010; Tsunetsugu et al., 2013; Jiang et al., 2014). These ob- 50% or more of window covered by −7.11 (−11.90; < 0.01 servations are in line with the findings of other experimental studies greenspace −2.32) showing amelioration of the physiological markers of stress such as Having a private garden −0.99 (−5.37; 3.38) 0.65 blood pressure and heart rate as a result of short-term exposure to green a environment (Lee et al., 2011; Tsunetsugu et al., 2013). More recently, Adjusted for the maternal age, body mass index, number of pregnancies, a limited but emerging body of epidemiological evidence has docu- gestational age in days, exposure to tobacco at home, percentage of illiterates and unemployed per census, paternal and maternal education and household mented the positive influence of the long-term exposure to greenspace income. Effect estimates in cortisol (ng/ml) are reported with their 95% con- on stress (Triguero-Mas et al., 2017). A study on mental health with fidence interval (95% CI). over 10,000 participants in UK found an association between the dis- b Effect estimates are reported as increase in cortisol (ng/ml) per one-in- tance to green space and stress (White et al., 2013), which is consistent terquartile range increase with their 95% confidence interval (95% CI). with findings of a Danish study of more than 11,000 participants that observed lower levels of perceived stress among people living less than Table 3 one kilometer from a green space (Stigsdotter et al., 2010). Studies Percentage (%) of the association between residential surrounding greenspace further suggest a greater impact of greenspace exposure on stress levels (100 m buffer) and cord blood cortisol level mediated by visual access and air among people with lower SES (Gov.uk, 2019; Allen and Balfour, 2012). pollution.a. Since people with a low SES are more likely to have poorer health status Mediator Percentage mediated (%, 95% and more stressful lifestyle and live in areas with poorer environmental confidence intervals) conditions, exposure to greenspace could have a stronger effect on them compared to people from high SES groups. Our observed findings for Visual access: Having a window with greenspace view 5.3 (−23.9; 26.3) the stratified analyses based on parental education were partly sup- Frequently looking at greenspace 8.3 (−63.2; 19.0) porting this hypothesis, showing a stronger association for the partici- 50% or more of window covered by 3.18 (−109.3; 21.1) pants with lower parental education for some of the greenspace ex- greenspace posure indicators, but not others. Similarly, we observed a mixed Air pollution: pattern of associations across the strata of housing type, which we PM1 5.1 (−2.9; 18.8) cannot compare with those of the previous study, given lack of avail- PM2.5 28.6 (−1.6; 128.9) able evidence on such an effect modification. The available evidence on PM10 32.7 (−7.0; 160.9) a potential sex difference in the cord blood level of cortisol is not a Adjusted for the maternal age, body mass index, number of pregnancies, consistent. While some studies support such a difference (Giesbrecht gestational age in days, exposure to tobacco at home, percentage of illiterates et al., 2016; Wynne-Edwards et al., 2019), others did not find any and unemployed per census, paternal and maternal education and household differences between the sexes (Bagnoli et al., 2013; Travers et al., income. 2018). The stratification of our analyses by sex suggested a potentially stronger association for maternal visual access and residential proxi- projected to occur (United Nations, 2018). We used data obtained mity to large green spaces among girls. However, the pattern was mix through questionnaires, land cover maps, and satellite imagery to for most of our other evaluated exposure variables. Such an effect comprehensively characterize exposure to greenspace. We observed modification of the health effects of greenspace exposure remains as an inverse associations between higher residential surrounding greenspace open question for future studies. (NDVI and MSVAI2) across a 100 m buffer, residential proximity to a large green space, more visual access to greenspace, and more time 4.2. Potential mechanisms spent in green spaces and the cord blood cortisol levels. We also ob- served an inverse association between residential surrounding green- One potential mechanism on how prenatal exposure to greenspace space (NDVI and MSVAI2) and tree cover (VCF) in larger buffers of may influence the cortisol level in the cord blood is the reduction of 300 m and 500 m and residential proximity to a green space with any maternal stress. Former studies found a positive correlation between size and cortisol levels in the cord blood; however, these associations maternal and fetal cortisol concentrations (Keshavarzi et al., 2014); did not attain statistical significance. The finding for the number of suggesting a linear relation due to placental exchange of cortisol. Al- plant pots at home was not conclusive. While about one-third of the though the placenta enzyme 11β-hydroxysteroid-dehydrogenase (11β- association between residential surrounding greenspace (100 m buffer) HSD-2) partly protects the human fetus from high cortisol levels by 6 L.M. Boll, et al. Environment International 144 (2020) 106047 Fig. 2. Adjusteda associations between prenatal exposure to different indicators of greenspace with cord blood cortisol level stratified for parental education status (A), housing type (B) and fetal sex (C). a Adjusted for the maternal age, body mass index, number of pregnancies, gestational age in days, exposure to tobacco at home, percentage of illiterates and unemployed per census, paternal and maternal education and household income. Effect estimates in cortisol (ng/ml) are reported per one-interquartile range increase with their 95% confidence interval (95% CI). Note: NDVI = normalized difference vegetation index with a buffer of 100 m, 300 m and 500 m; GS = green space. converting 80–90% of it into the biologically inactive cortisone (Duthie Ulrich et al. (1991) proposed that a greener surrounding could result in and Reynolds, 2013), some cortisol still passes the placenta barrier. The physiological recovery in the form of less muscle tension and skin fetal cord blood cortisol level was found to be much lower than the conductance and lower pulse rate and blood pressure, while invoking maternal, thus a small increase in maternal cortisol could have a sub- an emotional response in terms of increased positive feelings and re- stantial effect on fetal cortisol levels (Mulder et al., 2002). Assuming duced negative thoughts and emotions such as fear and anger. In this that the cord blood cortisol level could be associated with maternal context, a study using functional magnetic resonance imaging (fMRI) of stress, we can assume direct and indirect pathways through which ex- brain demonstrated the activation of the areas in the human brain posure to greenspace could reduce maternal stress, and hence cortisol normally responsive to positive emotions (basal ganglia) when parti- levels. cipants were shown pictures of natural environments, while looking at Stress reduction theory offers an explanation for a direct pathway. pictures of urban setting activated the amygdala, which is responsible 7 L.M. Boll, et al. Environment International 144 (2020) 106047 Fig. 2. (continued) for processing of fear and stressful emotions (Kim et al., 2010). These between the residential surrounding greenspace and the cord blood findings could suggest that the effect of greenspace exposure on the cortisol levels could be explained by lower exposure to PM2.5 or PM10. neuronal system is not only through direct immersion in a green en- vironment but also could be through visual access to greenspace. Our 4.3. Limitations analysis of the mediatory role of visual access to greenspace in the as- sociation of residential surrounding greenspace and cord blood cortisol This study faced some limitations. Our relatively small sample size level, however, did not support such a mediation. of 150 participants could have limited our statistical power, especially Physical activity has been suggested as a pathway through which for the stratified analyses. We are further aware that our participants green spaces exert their health benefits (Nichani et al., 2016). Physical represented a specific population, thus generalization of our findings to activity could be a relevant mechanism underlying our finding, given other areas with different climate, vegetation types, and cultures should that it has been shown to decrease stress through increasing endorphin be done with caution. Like any observational study, there was a pos- levels, inhibiting the activation of the hypothalamic-pituitaryadrenal sibility of the residual SES confounding in our study. To minimize this axis (Dishman and O’Connor, 2009) and directly decreasing cortisol possibility, we controlled our analyses for five SES indicators including levels (Beserra et al., 2018). Another potential mechanism of green- three household and two neighborhood-level indicators of SES. We space restoring capacity could be by increasing social cohesion (De tested the mediatory role of PM1, PM2.5, and PM10 in our evaluated Vries et al., 2013). Increased interpersonal interactions could promote association between maternal exposure to greenspace and cord blood mutual respect, meaningful connections and trust, thus positively in- level of cortisol. However, we could not evaluate mediatory roles of fluence the emotional state and decrease levels of stress hormones such other air pollutants such as ozone (Miller et al., 2016) and NO2 as cortisol (Kawachi and Berkman, 2015). Furthermore, urban green (Dadvand et al., 2012) because of unavailability of data on these pol- spaces are shown to mitigate exposure to heat, noise and air pollution in lutants. Exposure to ozone could have been particularly relevant, given urban areas (Bowler et al., 2010; Hartig et al., 2014), which, in turn, the warm climate of our study region. However, our previous study of have been associated to an increased stress response (Niu et al., 2018; the association of residential surrounding greenspace with personal Lederbogen et al., 2011). By reducing these environmental risk factors, exposure of pregnant women to PM2.5 and NO2 based on personal greenspace could therefore have the potential to decrease stress, which monitoring of these pollutants showed that while higher residential can be reflected by lower cortisol levels. Consistently, our mediation surrounding greenspace was associated with lower personal exposure to analysis of air pollution showed that up to one-third of the association PM2.5, it was not associated with personal exposure to NO2 (Dadvand 8 L.M. Boll, et al. Environment International 144 (2020) 106047 Fig. 2. (continued) et al., 2012). The use of the cortisol level in the cord blood as a sur- during pregnancy have been associated with adverse pregnancy out- rogate marker of stress could be influenced by delivery circumstances, comes and postnatal growth and development, our findings, if re- such us labor duration or fetal distress; however, we obtained samples plicated by future studies, could offer evidence base for policymakers to from women with a normal, term vaginal delivery without complica- implement interventions targeted at improving pregnancy outcomes tions at delivery that could have minimized such an influence in our and child health and development in our rapidly urbanizing world. analyses. Moreover, we used ELISA to measure the level of cortisol in More studies are needed to evaluate effects on perinatal outcomes using the cord blood, while High-Performance Liquid Chromatography larger sample sizes in other settings and climates, while considering the (HPLC) could provide more accurate measures. Furthermore, our as- quality of green spaces in their exposure assessment and relying on sessment of the residential surrounding greenspace and proximity to objective measures of visual access to greenspace. green spaces did not address the quality of these spaces, which could have influenced the use of these spaces by our participants. Unsafe or CRediT authorship contribution statement unattractive green spaces, for example, could discourage pregnant women to visit these spaces. Moreover, our characterization of time Lilian Marie Boll: Data curation, Formal analysis, Visualization, spent in greenspace and visual access to greenspace was based on Writing - original draft. Ramezanali Khamirchi: Project administra- questionnaires, which were not validated using objective measures and tion, Writing - review & editing. Lucia Alonso: Data curation, Formal could have been prone to subjectivity and recall bias. Given the rela- analysis, Writing - review & editing. Elisa Llurba: Conceptualization, tively small contrast in VCF levels across our study area and the re- Methodology, Writing - review & editing. Óscar J Pozo: sulting limited statistical power, our non-statistically significant results Conceptualization, Methodology, Writing - review & editing. for VCF should be interpreted with caution. Mohammad Miri: Funding acquisition, Conceptualization, Methodology, Project administration, Writing - review & editing. 5. Conclusion Payam Dadvand: Conceptualization, Methodology, Supervision, Validation, Writing - review & editing. We observed that higher residential surrounding greenspace, re- sidential proximity to green spaces larger than 5,000 m2, more time Declaration of Competing Interest spent in green spaces, having a window with greenspace view, having a window covered at by at least 50% of greenspace view, and frequently The authors declare that they have no known competing financial looking at greenspace through window during pregnancy were asso- interests or personal relationships that could have appeared to influ- ciated with lower cortisol levels in the cord blood. Our findings for the ence the work reported in this paper. association between indoor plants and cortisol levels were not con- clusive. Similarly, we observed some indications for the mediation of Acknowledgment the association between the residential surrounding greenspace and cord blood cortisol levels by air pollution but not by visual access to We thank the Shahidan Mobayani Hospital for their cooperation in greenspace. Given that fetal exposure to sustained high cortisol levels this study. Payam Dadvand is funded by a Ramón y Cajal fellowship 9 L.M. Boll, et al. Environment International 144 (2020) 106047 (RYC-2012-10995) awarded by the Spanish Ministry of Economy and Drugs/GuidanceComplianceRegulatoryInformation/Guidances/default.htmand/ Finance. Oscar J. Pozo is funded by a Miguel Servet II fellowship orhttp://www.fda.gov/AnimalVeterinary/GuidanceComplianceEnforcement/ GuidanceforIndustry/default.htm. (CPII16/00027) awarded by the Health Institute Carlos III (Spain). Giesbrecht, G.F., Rash, J.A., Edwards, H.E., Wynne-Edwards, K.E., 2016. Full-term de- liveries without antecedent labor reveal sex differences in umbilical cord gluco- Appendix A. Supplementary data corticoid concentrations. Psychoneuroendocrinology 74, 121–125. https://doi.org/ 10.1016/j.psyneuen.2016.08.030. Gong, Y., Palmer, S., Gallacher, J., Marsden, T., Fone, D., 2016. A systematic review of the Supplementary data to this article can be found online at https:// relationship between objective measurements of the urban environment and psy- doi.org/10.1016/j.envint.2020.106047. chological distress. Environ. Int. [Internet] 96, 48–57 (cited 2020 Jan 24). Gov.uk., 2019. Marmot Review report – ’Fair Society, Healthy Lives | Local Government Association [Internet]. https://www.local.gov.uk/marmot-review-report-fair- References society-healthy-lives. (cited 2020 Jan 27). Available from: https://www.local.gov. uk/marmot-review-report-fair-society-healthy-lives. Hartig, T., Mitchell, R., de Vries, S., Frumkin, H., 2014 Mar 18. Nature and Health. Annu. Allen, J., Balfour, R., 2012. Natural solutions for tackling health inequalities: realising the Rev. Public Health. 35 (1), 207–228. opportunity. UCL Institute of Health Equity. Janhäll, S., 2015. Review on urban vegetation and particle air pollution - Deposition and Amoly, E., Dadvand, P., Forns, J., López-Vicente, M., Basagaña, X., Julvez, J., et al., 2015. dispersion. Vol. 105, Atmospheric Environment. Elsevier Ltd, pp. 130–137. Green and blue spaces and behavioral development in barcelona schoolchildren: The Jennings, V., Bamkole, O., 2019. The relationship between social cohesion and urban BREATHE project. Environ. Health Perspect. 122 (12), 1351–1358. green space: An avenue for health promotion. Int. J. Environ. Res. Public Health Aram, F., Higueras García, E., Solgi, E., Mansournia, S., 2019. Urban green space cooling [Internet] 16 (3) (cited 2020 Jan 24). effect in cities [Internet]. Vol. 5, Heliyon, p. e01339 (cited 2020 Jan 24). Available Jiang, B., Chang, C.Y., Sullivan, W.C., 2014 Dec Dec. A dose of nature: Tree cover, stress from: https://linkinghub.elsevier.com/retrieve/pii/S2405844019300702. reduction, and gender differences. Landsc Urban Plan. 1 (132), 26–36. Bagnoli, F., Mori, A., Fommei, C., Coriolani, G., Badii, S., Tomasini, B., 2013. ACTH and Kawachi, I., Berkman, L.F., 2015. Social capital, social cohesion, and health. In: Social cortisol cord plasma concentrations in preterm and term infants. J. Perinatol. 33 (7), Epidemiology [Internet]. Oxford University Press (cited 2020 Jan 24), pp. 290–319. 520–524. https://doi.org/10.1038/jp.2012.165. Available from: http://oxfordmedicine.com/view/10.1093/med/9780195377903. Barati, S., Rayegani, B., Saati, M., Sharifi, A., Nasri, M., 2011. Comparison the accuracies 001.0001/med-9780195377903-chapter-8. of different spectral indices for estimation of vegetation cover fraction in sparse ve- Keshavarzi, F., Farnia, V., Yazdchi, K., Najafi, F., Brand, S., Bajoghli, H., et al., 2014 Dec getated areas. Egypt J. Remote Sens. Sp. Sci. 14 (1), 49–56. 1. Effect of maternal anxiety on maternal serum and fetal cord blood cortisol. Asia- Baron, R.M., Kenny, D.A., 2020. The moderator-mediator variable distinction in social Pacific Psychiatry. 6 (4), 435–439. psychological research. Conceptual, strategic, and statistical considerations. J. Pers. Kim, T.H., Jeong, G.W., Baek, H.S., Kim, G.W., Sundaram, T., Kang, H.K., et al., 2010. Soc. Psychol. [Internet] 51(6), 1173–1182 (cited 2020 Jan 24). Available from: Human brain activation in response to visual stimulation with rural and urban http://doi.apa.org/getdoi.cfm? doi: 10.1037/0022-3514.51.6.1173. scenery pictures: A functional magnetic resonance imaging study. Sci. Total Environ. Beserra, A.H.N., Kameda, P., Deslandes, A.C., Schuch, F.B., Laks, J., de Moraes, H.S., [Internet] 408(12), 2600–2607 (cited 2020 Jan 24). Available from: http://www. 2018. Can physical exercise modulate cortisol level in subjects with depression? A ncbi.nlm.nih.gov/pubmed/20299076. systematic review and meta-analysis. Vol. 40, Trends in Psychiatry and Kottek, M., Grieser, J., Beck, C., Rudolf, B., Rubel, F., 2006. World map of the Köppen- Psychotherapy. Sociedade de Psiquiatria do Rio Grande do Sul. p. 360–368. Geiger climate classification updated. Meteorol. Zeitschrift [Internet] 15(3), 259–263 Bowler, D.E., Buyung-Ali, L., Knight, T.M., Pullin, AS., 2010. Urban greening to cool (cited 2020 Jan 24). Available from: http://gpcc.dwd.de. towns and cities: A systematic review of the empirical evidence. Vol. 97, Landscape Lachowycz, K., Jones, A.P., 2011 May May. Greenspace and obesity: A systematic review and Urban Planning. p. 147–155. of the evidence. Obes Rev. 12 (501). Carlsten, C., Rider, C.F., 2017. Traffic-related air pollution and allergic disease: An update Lederbogen, F., Kirsch, P., Haddad, L., Streit, F., Tost, H., Schuch, P., et al., 2011 Jun 23. in the context of global urbanization. Vol. 17, Current Opinion in Allergy and Clinical City living and urban upbringing affect neural social stress processing in humans. Immunology. Lippincott Williams and Wilkins, p. 86–89. Nature. 474 (7352), 498–501. Center, K.R.W., 2020. ‫[ ﺍﻗﻠﯿﻢ ﺍﺳﺘﺎﻥ | ﭘﻮﺭﺗﺎﻝ ﺍﺩﺍﺭﻩ ﮐﻞ ﻫﻮﺍﺷﻨﺎﺳﯽ ﺧﺮﺍﺳﺎﻥ ﺭﺿﻮﯼ‬Internet]. [cited Lee, J., Park, B.J., Tsunetsugu, Y., Ohira, T., Kagawa, T., Miyazaki, Y., 2011. Effect of 2020 Feb 11]. Available from: https://www.razavimet.ir/fa/node/38. forest bathing on physiological and psychological responses in young Japanese male Cherak, S.J., Malebranche, M.E., Wynne-Edwards, K., Williamson, T., Giesbrecht, G.F., subjects. Public Health. 125 (2), 93–100. 2019 Aug. Quantitative meta-analysis of maternal prenatal salivary cortisol and LP DAAC, 2017 (cited 2020 Aug 2); Available from: https://lpdaac.usgs.gov/products/ newborn birthweight does not identify effect of fetal sex. Psychoneuroendocrinology. gfcc30tcv003/. 1 (106), 117–121. McMahon, T.A., Rohr, J.R., Bernal, X.E., 2017. Light and noise pollution interact to Coombes, E., Jones, A.P., Hillsdon, M., 2010 Mar. The relationship of physical activity disrupt interspecific interactions. Ecology [Internet] 98(5), 1290–1299 (cited 2020 and overweight to objectively measured green space accessibility and use. Soc. Sci. Jan 27). Available from: http://www.ncbi.nlm.nih.gov/pubmed/28170099. Med. 70 (6), 816–822. Miller, D.B., Ghio, A.J., Karoly, E.D., Bell, L.N., Snow, S.J., Madden, M.C., et al., 2016. Dadvand, P., Bartoll, X., Basagaña, X., Dalmau-Bueno, A., Martinez, D., Ambros, A., et al., Ozone exposure increases circulating stress hormones and lipid metabolites in hu- 2016 May May. Green spaces and General Health: Roles of mental health status, mans. Am. J. Respir. Crit. Care Med. [Internet] 193(12), 1382–1391 (cited 2020 Jun social support, and physical activity. Environ Int. 1 (91), 161–167. 4). Available from: www.clinicaltrials.gov. Dadvand, P., de Nazelle, A., Triguero-Mas, M., Schembari, A., Cirach, M., Amoly, E., et al., Miri, M., Ghassoun, Y., Dovlatabadi, A., Ebrahimnejad, A., Löwner, M.O., 2019. Estimate 2012. Surrounding greenness and exposure to air pollution during pregnancy: An annual and seasonal PM 1, PM 2.5 and PM 10 concentrations using land use re- analysis of personal monitoring data. Environ. Health Perspect. [Internet] 120(9), gression model. Ecotoxicol. Environ. Saf. [Internet] 174, 137–145 [cited 2020 1286–1290 (cited 2020 Jun 4). Available from: https://ehp.niehs.nih.gov/doi/10. Jan 27]. 1289/ehp.1104609. Montes-González, D., Vílchez-Gómez, R., Barrigón-Morillas, J.M., Atanasio-Moraga, P., Dadvand, P., Tischer, C., Estarlich, M., Llop, S., Dalmau-Bueno, A., López-Vicente, M., Rey-Gozalo, G., Trujillo-Carmona, J., 2018 Oct 18. Noise and air pollution related to et al., 2017. Lifelong residential exposure to green space and attention: A population- health in urban environments. Proceedings 2 (20), 1311. based prospective study. Environ. Health Perspect. 125(9). Mulder, E.J.H, Robles de Medina, P.G, Huizink, A.C, Van den Bergh, B.R.H, Buitelaar, J.K, Dadvand, P., Poursafa, P., Heshmat, R., Motlagh, M.E., Qorbani, M., Basagaña, X., et al. Visser, G.H.A, 2002. Prenatal maternal stress: effects on pregnancy and the (unborn) Use of green spaces and blood glucose in children; a population-based CASPIAN-V child. Early Human Develop. 70 (1-2), 3–14. https://doi.org/10.1016/S0378- study. Environ. Pollut. [Internet] 243, 1134–1140 (cited 2020 Jun 4). Available from: 3782(02)00075-0. https://doi.org/10.1016/j.envpol.2018.09.094. Nichani, V., Dirks, K., Burns, B., Bird, A., Morton, S., Grant, C., 2016. Green space and De Vries, S., van Dillen, S.M.E., Groenewegen, P.P., Spreeuwenberg, P., 2013 Oct Oct. physical activity in pregnant women: Evidence from the growing up in new zealand Streetscape greenery and health: Stress, social cohesion and physical activity as study. J. Phys. Act Health [Internet] 13(12), 1341–1350 (cited 2020 Jan 27). mediators. Soc. Sci. Med. 94, 26–33. Available from: http://www.ncbi.nlm.nih.gov/pubmed/27633620. Dishman, R.K., O’Connor, P.J., 2009 Jun. Lessons in exercise neurobiology: The case of Niu, Y., Chen, R., Xia, Y., Cai, J., Ying, Z., Lin, Z., et al., 2018. Fine particulate matter endorphins. Ment. Health Phys. Act. 2 (1), 4–9. constituents and stress hormones in the hypothalamus–pituitary–adrenal axis. Duthie, L., Reynolds, R.M., 2013. Changes in the Maternal Hypothalamic-Pituitary- Environ. Int. [Internet] 119, 186–192 (cited 2020 Jan 27). Adrenal Axis in Pregnancy and Postpartum: Influences on Maternal and Fetal Qi, J., Chehbouni, A., Huete, A.R., Kerr, Y.H., Sorooshian, S., 1994. A modified soil ad- Outcomes. Neuroendocrinology 98 (2), 106–115. https://doi.org/10.1159/ justed vegetation index. Remote Sens. Environ. 48 (2), 119–126. 000354702. Reynolds, R.M., 2013. Glucocorticoid excess and the developmental origins of disease: Dzhambov, A.M., Dimitrova, D.D., 2015. Green spaces and environmental noise percep- Two decades of testing the hypothesis – 2012 Curt Richter Award Winner. tion. Urban For Urban Green 14 (4), 1000–1008. Psychoneuroendocrinology 38, 1–11. EMA, 2012. Committee for Medicinal Products for Human Use. Guideline on bioanalytical Sexton, J.O., Song, X.P., Feng, M., Noojipady, P., Anand, A., Huang, C., et al., 2013. method validation. EMEA/CHMP/EWP/192217/2009 Rev 1 Corr 2** [Internet] 1–23 Global, 30-m resolution continuous fields of tree cover: Landsat-based rescaling of (cited 2020 Apr 30). Available from: www.ema.europa.eu/contact. MODIS vegetation continuous fields with lidar-based estimates of error. Int. J. Digital ESPA, 2017. ESPA - LSRD [Internet] (cited 2020 Jun 4). Available from: https://espa.cr. Earth 6 (5), 427–448. https://doi.org/10.1080/17538947.2013.786146. usgs.gov/. Shahmohamadi, P., Che-Ani, A.I., Etessam, I., Maulud, K.N.A., Tawil, N.M., 2011. Healthy Food and Drug Administration, 2018. Guidance for Industry Bioanalytical Method environment: The need to mitigate urban heat island effects on human health. Validation Guidance for Industry Bioanalytical Method Validation. Biopharm. Fed. Procedia Eng. 61–70. Regist. [Internet] (100) 66 (cited 2020 Apr 30). Available from: http://www.fda.gov/ StataCorp, 2017. Stata Statistical Software: Release 15 MP [Internet]. College Station: TX: 10 L.M. Boll, et al. Environment International 144 (2020) 106047 Computer Software. Available from: http://www.jstor.org/stable/2234838?origin= Tyrväinen, L., Ojala, A., Korpela, K., Lanki, T., Tsunetsugu, Y., Kagawa, T., 2014 Jun Jun. crossref. The influence of urban green environments on stress relief measures: A field ex- Stigsdotter, U.K., Randrup, T.B., Ekholm, O., Schipperijn, J., Toftager, M., Kamper- periment. J. Environ. Psychol. 38, 1–9. Jørgensen, F., 2010. Health promoting outdoor environments - Associations between Ulrich, R.S., Simons, R.F., Losito, B.D., Fiorito, E., Miles, M.A., Zelson, M., 1991. Stress green space, and health, health-related quality of life and stress based on a Danish recovery during exposure to natural and urban environments. J. Environ. Psychol. 11 national representative survey. Scand J Public Health. 38 (4), 411–417. (3), 201–230. Su, Q., Zhang, H., Zhang, Y., Zhang, H., Ding, D., Zeng, J., et al., 2015. Maternal stress in United Nations. 2018 Revision of World Urbanization Prospects | Multimedia Library - gestation: birth outcomes and stress-related hormone response of the neonates. United Nations Department of Economic and Social Affairs [Internet]. 2018 Revision Pediatr. Neonatol. [Internet] 56(6), 376–381 (cited 2020 Jan 27). Available from: of World Urbanization Prospects | Multimedia Library - United Nations Department https://linkinghub.elsevier.com/retrieve/pii/S1875957215000492. of Economic and Social Affairs. 2018 [cited 2020 Jan 27]. Available from: https:// Travers, S., Martinerie, L., Boileau, P., Xue, Q.Y., Lombès, M., Pussard, E., 2018. www.un.org/development/desa/publications/2018-revision-of-world-urbanization- Comparative profiling of adrenal steroids in maternal and umbilical cord blood. J. prospects.html. Steroid Biochem. Mol. Biol. 178, 127–134. https://doi.org/10.1016/j.jsbmb.2017. White, M.P., Alcock, I., Wheeler, B.W., Depledge, M.H., 2013. Would you be happier 11.012. living in a greener urban area? A fixed-effects analysis of panel data. Psychol. Sci. 24 Triguero-Mas, M., Donaire-Gonzalez, D., Seto, E., Valentín, A., Martínez, D., Smith, G., (6), 920–928. et al., 2017. Natural outdoor environments and mental health: Stress as a possible Wynne-Edwards, K.E., Lee, K., Zhou, R., Edwards, H.E., 2019. Sex differences in sub- mechanism. Environ Res. 159, 629–638. strates and clearance products of cortisol and corticosterone synthesis in full-term Tsunetsugu, Y., Lee, J., Park, B.J., Tyrväinen, L., Kagawa, T., Miyazaki, Y., 2013 May. human umbilical circulation without labor: substrate depletion matches synthesis in Physiological and psychological effects of viewing urban forest landscapes assessed males, but not females. Psychoneuroendocrinology 109. https://doi.org/10.1016/j. by multiple measurements. Landsc Urban Plan. 113, 90–93. psyneuen.2019.104381. 11