Author + information
- Received December 5, 2018
- Revision received April 26, 2019
- Accepted June 24, 2019
- Published online September 2, 2019.
- Charline Warembourg, PhDa,b,c,
- Léa Maitre, PhDa,b,c,
- Ibon Tamayo-Uria, PhDa,b,c,
- Serena Fossati, MD, PhDa,b,c,
- Theano Roumeliotaki, MScd,
- Gunn Marit Aasvang, PhDe,
- Sandra Andrusaityte, PhDf,
- Maribel Casas, PhDa,b,c,
- Enrique Cequier, PhDe,
- Lida Chatzi, MD, PhDd,g,h,
- Audrius Dedele, PhDf,
- Juan-Ramon Gonzalez, PhDa,b,c,
- Regina Gražulevičienė, PhDf,
- Line Smastuen Haug, PhDe,
- Carles Hernandez-Ferrer, PhDa,b,c,
- Barbara Heude, PhDi,
- Marianna Karachaliou, MD, PhDd,
- Norun Hjertager Krog, PhDe,
- Rosemary McEachan, PhDj,
- Mark Nieuwenhuijsen, PhDa,b,c,
- Inga Petraviciene, PhDf,
- Joane Quentink,l,
- Oliver Robinson, PhDm,
- Amrit Kaur Sakhi, PhDe,
- Rémy Slama, PhDk,
- Cathrine Thomsen, PhDe,
- Jose Urquiza, PhDa,b,c,
- Marina Vafeiadi, PhDd,
- Jane West, PhDj,
- John Wright, MBChBj,
- Martine Vrijheid, PhDa,b,c and
- Xavier Basagaña, PhDa,b,c,∗ (, )@ISGlobalorg@Cha_Warembourg
- aISGlobal, Barcelona, Spain
- bUniversitat Pompeu Fabra (UPF), Barcelona, Spain
- cCIBER Epidemiologa y Salud Pública, Madrid, Spain
- dDepartment of Social Medicine, Faculty of Medicine, University of Crete, Heraklion, Greece
- eNorwegian Institute of Public Health, Oslo, Norway
- fVytauto Didziojo Universitetas, Kaunus, Lithuania
- gDepartment of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, California
- hDepartment of Genetics and Cell Biology, Faculty of Health, Medicine and Life Sciences, Maastricht University, Maastricht, the Netherlands
- iINSERM, UMR1153 Epidemiology and Biostatistics Sorbonne Paris Cité Center (CRESS), Early ORigins of the Child's Health and Development Team (ORCHAD), Paris Descartes University, Paris, France
- jBradford Institute for Health Research, Bradford Teaching Hospitals NHS Foundation Trust, Bradford, United Kingdom
- kInserm, Université Grenoble Alpes, CNRS, Institute of Advanced Biosciences, Team of Environmental Epidemiology applied to Reproduction and Respiratory Health, Grenoble, France
- lCHU Grenoble Alpes, Grenoble, France
- mMRC-PHE Centre for Environment and Health, School of Public Health, Imperial College London, London, United Kingdom
- ↵∗Address for correspondence:
Dr. Xavier Basagaña, ISGlobal, Doctor Aiguader, 88 - 08003 Barcelona, Spain.
Background Growing evidence exists about the fetal and environmental origins of hypertension, but mainly limited to single-exposure studies. The exposome has been proposed as a more holistic approach by studying many exposures simultaneously.
Objectives This study aims to evaluate the association between a wide range of prenatal and postnatal exposures and blood pressure (BP) in children.
Methods Systolic and diastolic BP were measured among 1,277 children from the European HELIX (Human Early-Life Exposome) cohort aged 6 to 11 years. Prenatal (n = 89) and postnatal (n = 128) exposures include air pollution, built environment, meteorology, natural spaces, traffic, noise, chemicals, and lifestyles. Two methods adjusted for confounders were applied: an exposome-wide association study considering the exposures independently, and the deletion-substitution-addition algorithm considering all the exposures simultaneously.
Results Decreases in systolic BP were observed with facility density (β change for an interquartile-range increase in exposure: −1.7 mm Hg [95% confidence interval (CI): −2.5 to −0.8 mm Hg]), maternal concentrations of polychlorinated biphenyl 118 (−1.4 mm Hg [95% CI: −2.6 to −0.2 mm Hg]) and child concentrations of dichlorodiphenyldichloroethylene (DDE: −1.6 mm Hg [95% CI: −2.4 to −0.7 mm Hg]), hexachlorobenzene (−1.5 mm Hg [95% CI: −2.4 to −0.6 mm Hg]), and mono−benzyl phthalate (−0.7 mm Hg [95% CI: −1.3 to −0.1 mm Hg]), whereas increases in systolic BP were observed with outdoor temperature during pregnancy (1.6 mm Hg [95% CI: 0.2 to 2.9 mm Hg]), high fish intake during pregnancy (2.0 mm Hg [95% CI: 0.4 to 3.5 mm Hg]), maternal cotinine concentrations (1.2 mm Hg [95% CI: -0.3 to 2.8 mm Hg]), and child perfluorooctanoate concentrations (0.9 mm Hg [95% CI: 0.1 to 1.6 mm Hg]). Decreases in diastolic BP were observed with outdoor temperature at examination (−1.4 mm Hg [95% CI: −2.3 to −0.5 mm Hg]) and child DDE concentrations (−1.1 mm Hg [95% CI: −1.9 to −0.3 mm Hg]), whereas increases in diastolic BP were observed with maternal bisphenol-A concentrations (0.7 mm Hg [95% CI: 0.1 to 1.4 mm Hg]), high fish intake during pregnancy (1.2 mm Hg [95% CI: −0.2 to 2.7 mm Hg]), and child copper concentrations (0.9 mm Hg [95% CI: 0.3 to 1.6 mm Hg]).
Conclusions This study suggests that early-life exposure to several chemicals, as well as built environment and meteorological factors, may affect BP in children.
Cardiovascular diseases are one of the leading causes of death, and high blood pressure (BP) is a major contributing factor (1). Until recently, prevention and control of hypertension mainly concerned adults, but studies now report that children with elevated BP are more likely to become hypertensive adults, showing the importance of BP control earlier in life (2).
There are a number of well-known risk factors for high BP and hypertension, such as obesity, physical inactivity, high sodium intake, tobacco and alcohol consumption, low socioeconomic status, and psychosocial stressors (1). Also, there is emerging evidence that environmental exposures may be important risk factors for high BP (3). There is moderate to strong evidence that exposure to air pollution, cold temperature, and noise increase BP in adults (4–6). On the other hand, the beneficial effect of the built environment, such as living in a walkable environment or close to green spaces, have been associated to lower BP (7). The evidence regarding exposure to chemicals is weaker.
According to recent reviews, environmental risk factors in children are similar to those in adults: moderate to strong evidence exists regarding the increase in BP with exposure to tobacco smoke during pregnancy (8) and to air pollution and noise during childhood (5,9), whereas the evidence about chemical exposure (9) is weaker. Existing studies were limited by considering single or a restricted number of environmental exposures at a time, without taking into account the wide range of other exposures to which an individual is simultaneously exposed.
In recent years, a more comprehensive approach has been advocated in the field of environmental epidemiology through the development of the exposome concept (10). It aims to evaluate all environmental exposures of an individual to better understand the role of environmental factors—which may act in synergy—on multifactorial and chronic diseases (11). This approach limits publication bias resulting from selective reporting of associations and limits the finding of false-positive associations when appropriate methods are used (11).
The fetal life is the starting point of the life course exposome and is of particular interest, because it corresponds to a period of rapid development that, if disturbed, may have long-term health effects (11). Characterizing the early-life exposome is particularly relevant to study BP in children regarding the growing evidence on the developmental origins of hypertension (12). To date, this innovative exposome concept has not been applied to BP.
This study aims to evaluate the association between a wide range of prenatal and postnatal environmental exposures and BP in children.
The study population is based on the HELIX (Human Early-Life Exposome) project, which pooled data from 6 longitudinal-based European birth cohorts: BIB (Born in Bradford) (United Kingdom), EDEN (Étude des Déterminants pré et postnatals du développement et de la santé de l’ENfant) (France), INMA (INfancia y Medio Ambiente) (Spain), KANC (Kaunus Cohort) (Lithuania), MoBa (Norwegian Mother and Child Cohort Study) (Norway) (13), and Rhea (Mother-Child Cohort in Crete) (Greece) (14). They included a total of 31,472 mother-child pairs, among whom a subcohort of 1,301 children (approximately 200 children in each cohort) was selected according to the following criteria of eligibility: 1) age 6 to 11 years; 2) stored pregnancy blood and urine samples available; 3) complete address history; and 4) no serious health problems that may affect the clinical testing or the child safety. These 1,301 mother-child pairs were followed in 2014 to 2015 for a clinical examination, a computer-assisted interview with the mother, and the collection of additional biological samples. Data collection was standardized and performed by trained staff. A full description of the subcohort methods and participants can be found elsewhere (15).
Assessment of exposures
A wide range of exposures were evaluated in HELIX to define the early-life exposome during 2 time periods: the prenatal pregnancy period and childhood (age 6 to 11 years). We assessed exposures in 3 main parts of the exposome: outdoor exposures, chemical exposures measured through biomarkers, and lifestyle factors. In the present study, 89 prenatal exposures and 128 postnatal exposures were included (Table 1).
Details on exposure assessment are provided in Online Appendix 1. In summary, assessment of outdoor exposures was based on the home address for the pregnancy period and on the home and school addresses for the childhood period (age 6 to 11 years) (Online Appendix 1, Part 1). It includes air pollution, measures of the built environment, meteorological conditions, natural spaces, road traffic, and noise. Additional exposures assessed by predictive modeling from home address and questionnaire include water disinfection by products (for pregnancy only) and indoor pollutants (for childhood period only) (Online Appendix 1, Part 2).
Exposure to environmental chemical contaminants was assessed through determination of concentrations in biological samples from the mother and the child (Online Appendix 1, Part 3) (16). The chemicals list was based on concern for child health and include organochlorine compounds (polychlorinated biphenyls [PCBs] and organochlorine pesticides), polybrominated diphenyl ethers, per- and polyfluoroalkyl substances (PFAS), metals, phthalate metabolites, phenols, organophosphate pesticide metabolites, and cotinine. When appropriate, the concentrations were adjusted for lipids or creatinine.
Lifestyle factors were collected by questionnaire and included smoking habits, diet, physical activity, allergens, sleep, and socioeconomic status (Online Appendix 1, Part 4).
The correlations between exposures are described elsewhere (17).
BP was measured during the clinical examination using a standardized protocol: after 5 min of rest in sitting position, 3 consecutive measurements were taken by oscillometric device (OMRON 705-CPII, Omron, Kyoto, Japan) with 1-min time intervals between them, in a pre-defined posture and in preference in the right arm. Adequate cuff sizes were chosen with respect to each child’s arm length and circumference. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) from each measurement were recorded, and the mean of the second and the third measurements was calculated and used in further analysis. Because of some device errors leading to unreliable values, we excluded 24 subjects; thus, the final population was n = 1,277 (98%) children.
For each variable, the optimal transformation to approach normality was applied. Missing data for all confounders and exposures were imputed using the method of chained equations (18). A total of 20 imputed datasets were generated and used in all of the analyses mentioned hereafter; Rubin’s rules were used to aggregate the results (18).
The statistical methods were identified a priori through a series of simulation studies mimicking as closely as possible the situation expected with Helix data (19); 2 approaches were retained. First, exposome-wide association study (ExWAS) analyses were used to perform exposure-by-exposure estimation of the association with SBP and with DBP, using multivariate linear regressions and applying a Bonferroni-type correction to control for multiple testing (20). Then, the iterative model search “deletion-substitution-addition” (DSA) algorithm—a variable selection method—was used to identify the exposures that jointly affect BP (19,21). DSA uses cross-validation and was shown to provide a lower proportion of false-positive associations than ExWAS (19,21). The DSA was applied 50 times, and exposures that were selected in at least 5% of them were included in a multiexposure linear regression model to obtain estimates of effect for each exposure adjusted for the others. We consider the multiexposure model as our main analysis.
Both statistical methods (ExWAS and DSA) were performed separately for the prenatal and the postnatal exposomes and were adjusted for a common set of confounders identified from a Directed Acyclic Graph (Online Appendix 2): the cohort of inclusion, maternal age (continuous in years), maternal educational level (low, middle, or high), self-reported maternal pre-pregnancy body mass index (continuous in kg/m2), parity (nulliparous, primiparous, or multiparous), parental country of birth (none, 1, or both parents born in the country of inclusion), child age (continuous in years), child sex (boy, girl), and child height (continuous in cm).
More details on data pre-processing, imputation process, and statistical analyses are provided in Online Appendix 2. Description of exposure levels, missing rate, and transformation used are provided in Online Appendix 3. All analyses were run under R version 3.4.0 (R Foundation, Vienna, Austria).
Five sets of sensitivity analyses were performed. The first was performed for the full exposome (ExWAS), and the others were performed on the exposures included in the multiexposures model only. They included: 1) using age-, sex-, and height-specific z-score of BP determined by existing charts (22); 2) refitting the multiexposure models by excluding subjects with missing data for a specific exposure; 3) stratifying the multiexposure models by cohort and testing for heterogeneity (i.e., I-square); 4) adjusting models for birth weight as a potential mediator; and 5) exploring nonlinearity of the associations using restrictive cubic splines.
At the time of delivery, pregnant women were age 31 years on average, and primarily had a high educational level (52%), were a native of the cohort country (84%), and already had a child (54%) (Table 2). At birth, the newborns’ birthweight was on average 3,380 g, and <5% of them were born pre-term. At the time of examination, children were around age 8 years on average (range: 6.5 to 11 years), 72% of them were of normal weight, and around 10% could be classified as pre-hypertensive or hypertensive according to existing charts (22).
Associations between single exposures and BP in children
Exposures studied individually in association with SBP and DBP through the ExWAS method are visually presented in Figure 1, and the corresponding estimates and 95% confidence intervals (CIs) are provided in Online Appendix 4.
A total of 9 of the 89 prenatal exposures showed associations with SBP, and 2—facility density and facility richness (both markers of the built environment)—remained negatively associated with SBP after correction for multiple testing. In the childhood exposome, 12 of the 128 exposures showed associations with SBP, and 5 of these—dichlorodiphenyldichloroethylene (DDE), hexachlorobenzene (HCB), PCB 153, PCB 170, and the sum of PCBs (all organochlorine compounds)—remained negatively associated with SBP after correction for multiple testing.
Two prenatal exposures and 11 postnatal exposures were associated with DBP, but none of them remained associated after correction for multiple testing.
Associations between simultaneous exposures and BP in children
After applying the DSA method to study the associations between multiple exposures simultaneously and SBP, 5 prenatal and 4 postnatal exposures were selected in the prenatal and postnatal multiple-exposure models, respectively (Table 3). A lower SBP was observed for higher facility density during pregnancy (β for an interquartile-range increase in exposure = −1.7 mm Hg [95% CI: −2.5 to −0.8]; that is, a 1.7-mm Hg decrease in mean SBP is observed for a 49.5 unit/km2 increase in the number of facilities), maternal concentrations of polychlorinated biphenyl 118 (−1.4 mm Hg [95% CI: −2.6 to −0.2 mm Hg]) and child concentrations of DDE (−1.6 mm Hg [95% CI: −2.4 to −0.7 mm Hg]), HCB (−1.5 mm Hg [95% CI: −2.4 to −0.6 mm Hg]), and mono-benzyl phthalate (−0.7 mm Hg [95% CI: −1.3 to −0.1 mm Hg]). A higher SBP was observed with high fish intake during pregnancy (2.0 mm Hg [95% CI: 0.4 to 3.5 mm Hg]) and with higher outdoor temperature during pregnancy (1.6 mm Hg [95% CI: 0.2 to 2.9 mm Hg]), maternal cotinine concentrations (1.2 mm Hg [95% CI: −0.3 to 2.8 mm Hg]), and child perfluorooctanoate (PFOA) concentrations (0.9 mm Hg [95% CI: 0.1 to 1.6 mm Hg]).
In association with DBP, 2 prenatal and 3 postnatal exposures were selected in the multiexposure models (Table 3). We observed a lower DBP for higher outdoor temperature at examination (−1.4 mm Hg [95% CI: −2.3 to −0.5 mm Hg]) and child DDE concentrations (−1.1 mm Hg [95% CI: −1.9 to −0.3]), whereas we observed higher DBP for higher maternal bisphenol-A (BPA) concentrations (0.7 mm Hg [95% CI: 0.1 to 1.4 mm Hg]), low and high fish intake during pregnancy (1.4 mm Hg [95% CI: −0.1 to 2.8 mm Hg] and 1.2 mm Hg [95% CI: −0.2 to 2.7 mm Hg], respectively), and higher child copper concentrations (0.9 mm Hg [95% CI: 0.3 to 1.6 mm Hg]).
Overall, the estimates obtained in the multiple-exposure models were similar to the ones obtained in the single-exposure models (i.e., ExWAS) (Online Appendix 4). However, changes in estimates and narrower confidence intervals were observed in the multiple-exposure model for childhood concentrations of PFOA (β single-exposure vs. multiple-exposure model = 0.4 vs. 0.9), DDE (2.1 vs. 1.6), and HCB (2.1 vs. 1.5) in association with SBP, suggesting residual confounding in the single-exposure models.
Similar findings were observed using z-score of BP, adjusting models for birth weight or restricting analyses to complete cases (Online Appendix 5, Parts 1 to 3). Most of the results were consistent across cohorts, but some heterogeneity can be highlighted, even if the number of subjects per cohort is quite small: 1) the association between the average temperature during pregnancy and child SBP was driven by the BIB cohort; 2) the association between prenatal BPA and DBP was mainly driven by BIB; 3) the association between child PFOA and SBP was not observed in KANC nor in Rhea; and 4) the effect of child mono-benzyl phthalate level was associated with an increase in SBP in EDEN and Rhea, but with a decrease in BIB, INMA, KANC, and MoBa (Online Appendix 5, Part 4). Linearity of the risk was observed for all exposures except for prenatal levels of BPA (Online Appendix 5, Part 5).
This study is the first to simultaneously consider the possible effects of exposure to hundreds of environmental factors during early life on BP in children (Central Illustration). Several environmental exposures were associated with an increase (fish intake, bisphenol-A, cotinine, and temperature during pregnancy; perfluorooctanoate and copper at age 6 to 11 years) or a decrease (facility density, polychlorinated biphenyl 118 during pregnancy; dichlorodiphenyldichloroethylene, hexachlorobenzene, mono-benzyl phthalate, and temperature at age 6 to 11 years) in BP.
BP is regulated by various mechanisms involving several biological systems as potential targets, in particular, the renal, cardiovascular, and endocrine systems. Evidence exists regarding the long-term effect of fetal events on BP (fetal programming; e.g., maternal undernutrition) while BP is also affected by current factors (acute effect; e.g., temperature) (3,12). From the extensive experimental studies available on the fetal origins of hypertension—mainly concerning maternal undernutrition—several hypotheses have been advanced to explain this delayed effect, including renal dysfunction (e.g., nephron number and the renal renin-angiotensin system), vascular dysfunction, placental dysfunction, hypothalamic–pituitary–adrenal axis programming, oxidative stress, sodium homeostasis, and epigenetic modifications (12).
We observed that measures of the built environment, notably the facility density where the mother was living during pregnancy, were associated with lower SBP in offspring. Urban design factors, such as the density of facilities in a particular area (e.g., shops, restaurants, parks, or transportation hubs), determine how people use and move around the city and promote physical activity and social contact (7). Good evidence exists regarding the effects of urban and transport planning on cardiovascular health, and highly walkable environments were associated with lower BP in adult populations (7). To our knowledge, no study on the effect of the built environment exists in pregnant women and their offspring. We could hypothesize that the beneficial effect observed results from higher physical activity during pregnancy, a condition known to optimize fetal weight (i.e., reduces fat accumulation and maintains muscle mass) (23).
We observed that an increase in outdoor temperature (average of the day before the clinical examination) was associated with a decrease in SBP. Outdoor temperature is among the known environmental factors that affect BP in adults, as reported in a recent meta-analysis (6), but also in children (24).
For organochlorine compounds, we observed that exposure to PCBs during pregnancy and to PCBs, DDE, and HCB during childhood were inversely associated with SBP. PCBs and organochlorine pesticides are persistent organic pollutants that accumulate through the food chain and to which people are chronically exposed, especially through diet, despite the fact that they were banned decades ago (25). One study performed in 427 children from the Rhea cohort reported an increase in SBP and DBP at 4 years in association with prenatal exposure to DDE and HCB, but not PCBs (26). Another study reported higher DBP but not SBP at age 7 to 9 years with exposure to PCBs (but not DDE nor HCB) measured 1 year earlier (27). We suspect that reverse causality may explain the associations we observed, especially for the associations observed cross-sectionally in childhood. Indeed, these compounds are highly lipophilic, and children with a higher fat mass are expected to store more of these compounds in fat tissues and thus have lower circulating blood levels for a given exposure (28). Therefore, circulating levels of POPs may imperfectly reflect the total body burden, to an extent varying according to the child adiposity or BMI, which may confound the associations with BMI-related health outcomes such as BP.
Although we did not observe an association between exposure to PFAS (either during pregnancy or childhood) in the single-exposure models, exposure to PFOA during childhood was associated with higher SBP after coadjustment for organochlorine compounds (i.e., DDE and HCB). PFAS are chemicals present in various consumer products (e.g., cookware coatings, food packing materials, waterproof textiles) to which humans are exposed though diet, but also indoor air and dust (25). Few studies on the effects of PFAS on BP exist. One study performed in children from the INMA cohort reported no association between prenatal exposure to PFAS and BP at age 4 and 7 years (29), as did 1 cross-sectional study performed during childhood (age 12 to 18 years) (30). However, none of them have taken into account the level of exposure to organochlorine pesticides.
In our study population, exposure to some phthalates during childhood was associated with lower SBP and, to a lesser extent, with lower DBP. Phthalates are chemicals mainly used as plasticizers and to which people are exposed through diet but also through cosmetics use or medical devices (25). They are suspected to adversely affect cardiovascular health, but evidence from previous studies is limited, in particular regarding BP in children (9). One study performed among 379 children enrolled in the INMA cohort also reported a decrease in SBP (but not DBP) at age 4 and 7 years with exposure during pregnancy (31). In contrast, 2 cross-sectional studies performed in 8- to 19-year-old U.S. children reported higher SBP and DBP with higher phthalates exposure (32,33). Since phthalates are a PPARγ agonist, they may act through the modulation of the renin-angiotensin system (34,35).
Exposure to BPA during pregnancy, but not during childhood, was associated with higher DBP and, to a lesser extent, with higher SBP. BPA is a chemical present in various consumer products (e.g., plastic and canned food containers, toys, and cashier’s receipts) in countries with no restrictions on use (25). In children, BPA exposure during pregnancy but not during childhood was associated with higher DBP at age 4 years in 486 Korean children (36), but not in 500 children from the Rhea cohort (37). Animal studies have reported that BPA alters nephrogenesis during embryonic development, has endocrine disturbance properties (has estrogenic activity and disrupts the thyroid axis), induces oxidative stress and inflammation, and implies epigenetic changes (38).
Regarding exposure to metals, higher DBP was observed with increasing circulating copper levels in children; other metals were not statistically significantly associated with BP. Copper is a trace element essential for the body and for which deficiency is known to adversely affect cardiovascular health, but findings are inconsistent regarding higher copper levels (39). Epidemiological studies between copper exposure and BP are very limited in number. One study performed among 581 children (age 6 to 19 years) enrolled in the NHANES study also reported higher SBP and a trend for higher DBP with increasing serum copper level (40).
Low and high fish intake during pregnancy (less than and more than 2 to 4 times a week) were associated with an increase in SBP in children. To our knowledge, no previous studies were published on the effect of fish intake during pregnancy and offspring BP. It is well known that omega-3 fatty acids in fish are beneficial for cardiovascular health, but there is a debate regarding a possible opposite effect in the case of contamination of fish by chemicals that could reduce any positive effect of omega-3 fatty acids, or entail nonlinear associations (41). Indeed, fish intake is a major source of exposure to metals and lipophilic compounds, and may thus be a proxy of exposure to a mixture of compounds suspected to affect BP (e.g., mercury, arsenic, cadmium, lead), which, when taken individually, were positively but not statistically significantly associated with BP in children (Online Appendix 4).
We did not observe statistically significant associations for some of the environmental exposures known to affect BP, such as air pollution, noise, and tobacco exposures (3,5,9). We do note that higher BP supporting previous findings were suggested in the ExWAS analyses (e.g., p < 0.10 for NO2 and PM absorbance during childhood) and the DSA selection (prenatal cotinine selected and positively associated with SBP). Similarly, we did not find statistically significant associations between individual behaviors, such as unhealthy diet component and physical activity, and BP in children, even if some point estimates are in the expected direction (i.e., higher SBP and DBP with fast-food intake and sedentary behavior of the child).
Study strengths and limitations
Our study had several strengths, including its multicentric design with mother-child pairs of 6 European countries, the use of standardized protocols to measure both BP and the majority of exposures, and the use of advanced statistical methods to deal with the exposome context. The broad number of exposures simultaneously assessed both during pregnancy and the childhood period represents a step forward over previous approaches and is also a strength of the study, although we note that only a very small part of the exposome is covered. Limitations of this study include exposure misclassification, which is expected to be differential across exposures (e.g., stronger for the least persistent chemicals and for exposures based on modeling of outdoor levels), the relatively small sample size given the large number of exposures investigated, and the cross-sectional design for the postnatal exposome. Although we made efforts to balance sensitivity and specificity by basing our statistical analysis on a previous simulation study (19), we acknowledge that the present study remains at risk of false positives due to its agnostic approach and false negatives that may result from a lack of power.
Using a unique and systematic exposome approach that avoids the issue of selective reporting and reduces coexposure confounding, this study highlights that early life exposures, such as measures of the built environment, meteorological conditions, fish intake, and chemicals, are potentially important in the development of BP in children. Although some of the reported associations go in the same direction than previous research, the novel findings should be replicated in future studies. The clinical impact of the associations reported remains to be evaluated, but the unique strength of the longitudinal cohorts should be fostered to implement follow-ups that allow the investigation of long-term health implications.
COMPETENCY IN MEDICAL KNOWLEDGE: Environmental exposures early in life, including during fetal life, have potentially important effects on BP in children.
TRANSLATIONAL OUTLOOK: Prevention strategies that reduce exposure to modifiable risk factors early in life might prevent cardiovascular disease in adulthood.
ISGlobal is a member of the Agency for the Research Centres of Catalonia (CERCA) Programme, Generalitat de Catalunya. The authors are grateful to all of the participating children, parents, practitioners, and researchers in the 6 countries who took part in this study. The authors further thank Muireann Coen, Sonia Brishoual, Angelique Serre, Michele Grosdenier, Prof. Frederic Millot, Elodie Migault, Manuela Boue, Sandy Bertin, Veronique Ferrand-Rigalleau, Céline Leger, Noella Gorry, Silvia Fochs, Nuria Pey, Cecilia Persavente, Susana Gross, Georgia Chalkiadaki, Danai Feida, Eirini Michalaki, Mariza Kampouri, Anny Kyriklaki, Minas Iakovidis, Maria Fasoulaki, Ingvild Essen, Heidi Marie Nordheim, and the Yorkshire Water.
The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-206) under grant agreement no. 308333—the HELIX project. INMA data collections were supported by grants from the Instituto de Salud Carlos III, CIBERESP, and the Generalitat de Catalunya-CIRIT. KANC was funded by the grant of the Lithuanian Agency for Science Innovation and Technology (6-04-2014_31V-66). The Norwegian Mother and Child Cohort Study (MoBa) is supported by the Norwegian Ministry of Health and the Ministry of Education and Research, National Institutes of Health/National Institute of Environmental Health Sciences (contract no N01-ES-75558), and National Institutes of Health/National Institute of Neurological Disorders and Stroke (grant no. 1 UO1 NS 047537-01 and grant no. 2 UO1 NS 047537-06A1). The Rhea project was financially supported by European projects and the Greek Ministry of Health (Program of Prevention of obesity and neurodevelopmental disorders in preschool children, in Heraklion district, Crete, Greece: 2011 to 2014; “Rhea Plus”: Primary Prevention Program of Environmental Risk Factors for Reproductive Health, and Child Health: 2012 to 2015). The work was also supported by MICINN (MTM2015-68140-R) and Centro Nacional de Genotipado-CEGEN-PRB2-ISCIII (Spain). This paper presents independent research funded by the National Institute for Health Research (NIHR) under its Collaboration for Applied Health Research and Care (CLAHRC) for Yorkshire and Humber. Core support for Born in Bradford is also provided by the Wellcome Trust (WT101597MA, UK). Dr. Warembourg has received funding from the Fondation de France (00069251, France). Dr. Casas has received funding from Instituto de Salud Carlos III (Ministry of Economy and Competitiveness) (MS16/00128). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Listen to this manuscript's audio summary by Editor-in-Chief Dr. Valentin Fuster on JACC.org.
- Abbreviations and Acronyms
- blood pressure
- diastolic blood pressure
- deletion-substitution-addition algorithm
- exposome-wide association study
- systolic blood pressure
- Received December 5, 2018.
- Revision received April 26, 2019.
- Accepted June 24, 2019.
- 2019 American College of Cardiology Foundation
- Yang B.-Y.,
- Qian Z.,
- Howard S.W.,
- et al.
- Wang Q.,
- Li C.,
- Guo Y.,
- et al.
- Nieuwenhuijsen M.J.
- Sanders A.P.,
- Saland J.M.,
- Wright R.O.,
- Satlin L.
- Siroux V.,
- Agier L.,
- Slama R.
- Maitre L.,
- de Bont J.,
- Casas M.,
- et al.
- Haug L.S.,
- Sakhi A.K.,
- Cequier E.,
- et al.
- Tamayo-Uria I.,
- Maitre L.,
- Thomsen C.,
- et al.
- Agier L.,
- Portengen L.,
- Chadeau-Hyam M.,
- et al.
- Sinisi S.E.,
- van der Laan M.J.
- Reyes L.M.,
- Davenport M.H.
- Vafeiadi M.,
- Georgiou V.,
- Chalkiadaki G.,
- et al.
- Lee H.A.,
- Park S.H.,
- Hong Y.S.,
- Ha E.H.,
- Park H.
- Manzano-Salgado C.B.,
- Casas M.,
- Lopez-Espinosa M.-J.,
- et al.
- Geiger S.D.,
- Xiao J.,
- Ducatman A.,
- Frisbee S.,
- Innes K.,
- Shankar A.
- Rőszer T.,
- Ricote M.
- Bae S.,
- Lim Y.-H.,
- Lee Y.A.,
- Shin C.H.,
- Oh S.-Y.,
- Hong Y.-C.
- Vafeiadi M.,
- Roumeliotaki T.,
- Myridakis A.,
- et al.
- Nuñez P.,
- Fernandez T.,
- García-Arévalo M.,
- et al.
- Bost M.,
- Houdart S.,
- Oberli M.,
- Kalonji E.,
- Huneau J.-F.,
- Margaritis I.
- Zang X.,
- Huang H.,
- Zhuang Z.,
- et al.
- World Health Organization. Growth reference 5-19 years. Available at: https://www.who.int/growthref/who2007_bmi_for_age/en/. Accessed July 23, 2019.