Prenatal Urban Environment and Blood Pressure Trajectories From Childhood to Early Adulthood

Background Prenatal urban environmental exposures have been associated with blood pressure in children. The dynamic of these associations across childhood and later ages is unknown. Objectives The purpose of this study was to assess associations of prenatal urban environmental exposures with blood pressure trajectories from childhood to early adulthood. Methods Repeated measures of systolic blood pressure (SBP) and diastolic blood pressure (DBP) were collected in up to 7,454 participants from a UK birth cohort. Prenatal urban exposures (n = 43) covered measures of noise, air pollution, built environment, natural spaces, traffic, meteorology, and food environment. An exposome-wide association study approach was used. Linear spline mixed-effects models were used to model associations of each exposure with trajectories of blood pressure. Replication was sought in 4 independent European cohorts (up to 9,261). Results In discovery analyses, higher humidity was associated with a faster increase (mean yearly change in SBP for an interquartile range increase in humidity: 0.29 mm Hg/y, 95% CI: 0.20-0.39) and higher temperature with a slower increase (mean yearly change in SBP per interquartile range increase in temperature: −0.17 mm Hg/y, 95% CI: −0.28 to −0.07) in SBP in childhood. Higher levels of humidity and air pollution were associated with faster increase in DBP in childhood and slower increase in adolescence. There was little evidence of an association of other exposures with change in SBP or DBP. Results for humidity and temperature, but not for air pollution, were replicated in other cohorts. Conclusions Replicated findings suggest that higher prenatal humidity and temperature could modulate blood pressure changes across childhood.

is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Introduction
Blood pressure is an important modifiable risk factor for cardiovascular disease (CVD) 1,2 and it tracks from childhood into adulthood. 3 Elevated blood pressure in childhood or adolescence is associated with several intermediate markers of CVD and with CVD morbidity and mortality in adulthood. 4 There is a growing body of evidence showing that urban environmental exposures, such as air pollution, 5 noise, 6 temperature, 7 and some characteristics of the built environment 8 are associated with elevated blood pressure/hypertension in adulthood, 9,10 and some associations have also been observed in children. [11][12][13][14] Early life, especially the prenatal and early postnatal periods, is a phase of rapid development and particularly vulnerable to environmental factors, during which adverse exposures may lead to higher risk of CVD. 15,16 Some studies have shown associations of prenatal urban environmental exposures with blood pressure levels in children. [12][13][14] Previous studies have mostly measured blood pressure at a single time point, and none have sought replication of the findings.
Studies exploring early-life urban environmental exposures and later blood pressure have predominantly focused on single exposures, or a specific, related group of exposures, particularly air pollution, 13,17,18 and very few have used a more holistic view of the urban environment. 12, 14 To the best of our knowledge, no study has systematically explored a range of urban environmental exposures and their longitudinal association with blood pressure using repeated measures. Understanding whether associations vary over time is needed to determine the potential impact of the prenatal environment on future CVD.
Our aim was to assess the association of a range of prenatal urban environmental exposures with changes in systolic (SBP) and diastolic blood pressure (DBP) from childhood to early adulthood.

Methods
Data from the Avon Longitudinal Study of Parents and Children (ALSPAC) were used for the discovery analysis. ALSPAC is a prospective population-based cohort (see www.alspac.bris.ac.uk) that recruited pregnant women living in the former Avon area of the UK who were due to give birth between April 1991 and December 1992. [19][20][21] In total, 14,541 pregnancies were enrolled, which resulted in 14,062 live births. Children, mothers and their partners have been followed up repeatedly ever since. [19][20][21] The study website contains details of all the data that is available through a fully searchable data dictionary and variable search . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Blood pressure
Blood pressure was measured in all ALSPAC participants at clinic assessments at average ages 7, 9, 10, 11, 13, 15, 18 and 24 years, and at ages 3, 4 and 5 years in a 10% subsample. 19 In all clinic assessments, SBP and DBP were measured twice with the individual sitting at rest with the arm supported, using a cuff size appropriate for the child's upper arm circumference. The mean of the two measures was recorded and used in the analyses. More details on blood pressure measurement are described in supplementary material.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

Confounders
Analyses were adjusted for a set of confounders defined as known or plausible determinants of urban environment and blood pressure (i.e., maternal education, age at delivery, ethnicity, and area deprivation). Sex, although not a confounder, was also included in the adjustment to reduce residual variance. All confounders, except area deprivation, were obtained through questionnaires assessed during pregnancy. Area deprivation was based on data from the UK Ministry of Housing, Communities & Local Government and linked to the geocoded addresses at birth. More details on the confounders are presented in supplementary material.

Statistical analysis
We included participants who had information on at least one prenatal urban environmental exposure and at least one blood pressure measurement between mean ages 3 and 24 years, and who had complete data on the confounders. This resulted in the inclusion of 5,095 (for noise exposures) to 7,454 (for average greenness exposures) participants with between 28,132 and 41,216 blood pressure observations (Supplementary Figure 1).
An exposome-wide association study (ExWAS) approach 25 was used to assess the association of each exposure individually with SBP and DBP trajectories from childhood to early adulthood. Associations were examined using linear mixed-effects models with linear splines for age (as a fixed effect) to allow for nonlinear change in blood pressure with age. 26 Knots were placed at approximate ages 10 and 18 years to summarise linear change in blood pressure across childhood (3-10 years), adolescence (10-18 years), and young adulthood (18-26 years) (more details in supplementary material). Models included random intercept and random linear slope for age to allow for between-individual differences in blood pressure at baseline and in change with age. Age was centred at 3.1 (mean age at first clinic assessment).
An interaction term between each exposure and age was included in the models to assess the association of each urban environmental exposure with change in blood pressure. All models were adjusted for the confounders defined above.
To limit false positive results, a Bonferroni correction was used to account for multiple testing; an alpha level of 5% was divided by the number of exposures assessed (42 after removing one of highly correlated pairs, i.e. 0.05/42 = p-value threshold 0.0012). For associations that reached our multiple-testing p-value threshold, average predicted means for . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.31.23288002 doi: medRxiv preprint SBP and DBP from childhood to early adulthood in the 25 th and 75 th percentile of the exposure were calculated from the linear mixed-effects models.
Within each exposure-blood pressure trajectory, complete case analyses were undertaken, which gives unbiased results when the chance of being a complete is independent of the outcome after taking the covariates into consideration. 27 To explore potential impact of missing data, we compared characteristics of those included and not included in the analysis due to missing data on one or more confounders. The analyses were performed in the software Stata 17.0 (Statacorp, College Station, TX, USA) and the software MLwiN 3.05 was accessed from within Stata using runmlwin. 28

Sensitivity analyses
To explore the sensitivity of our results to including all participants with at least one blood pressure measurement we also repeated analyses with only participants who had at least three repeated measures (N up to 6,128 with 39,358 observations).
We explored possible sex differences in the associations between prenatal urban environmental exposures and blood pressure trajectories in those exposures found to be associated with blood pressure in the main analysis after multiple-testing correction. For that, we included in the models a 3-way interaction term between exposure, age and sex, along with an interaction between age and sex, and assessed the significance of the 3-way interaction (p<0.05). We also performed sex-stratified analyses.

Replication analysis
To assess robustness in the results found in ALSPAC, we sought replication, for associations that reached our multiple-testing p-value threshold, in four independent . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.31.23288002 doi: medRxiv preprint Linear mixed-effects models with random intercept and a random linear slope for age were used to model blood pressure trajectories in each study. Limited numbers of repeated measurements meant we had to use linear mixed-effects models without splines. In each study, age was centred at the mean age at baseline (which differed in each cohort). The set of available confounders may vary across studies hence, adjustment model may differ from the initial analysis in ALSPAC (more details in supplementary material). Results were considered to be replicated if the direction of association was consistent with the ALSPAC results and if p-value <0.05.

Results
In ALSPAC, up to 7,454 individuals with 41,216 blood pressure observations were included in the analysis. Those included were more likely to have mothers with higher education and older age at delivery, to have White ethnicity and to live in a less deprived area at birth than those not included in the analysis due to missing data on one or more of the confounders (Supplementary Table 2). Up to 11 measures of blood pressure were used (mean 6.7, standard deviation [SD] 2.1), the mean age of the participants ranged from 3.1 (SD 0.02) at baseline to 24.4 (SD 0.79) years at the last follow-up, and the number of individuals with blood pressure measure in each assessment varied from 800 (at age 5 years) to 6,056 (at age 7.5 years) (Supplementary Table 3 Associations of environmental exposures with blood pressure trajectories . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ; The associations of urban environmental exposures (n=42) with change in SBP and DBP from childhood to early adulthood are presented in Figure 1 and Supplementary Tables   6 and 7. After Bonferroni correction, prenatal humidity and PM 10 in the third trimester of pregnancy were associated with change in SBP from childhood to early adulthood ( Figure   1A). Higher relative humidity was associated with a faster increase in SBP in childhood (mean change in SBP for an IQR increase in humidity (which corresponds to 2% increase in humidity [Supplementary Table 4 Higher mean temperature was associated with a slower increase in SBP in childhood (mean change in SBP for an IQR increase in temperature: -0.18 mmHg/year, 95%CI: -0.28, -0.08).
There was little evidence for an association between the other urban environmental exposures and change in SBP from childhood to early adulthood.
The average predicted trajectory of SBP from childhood to early adulthood in the 25 th and 75 th percentiles of humidity, mean temperature and PM 10 at the third trimester of pregnancy is presented in Figure 2 and Supplementary Prenatal humidity, temperature and PM 2.5 , and PM 10 in all trimesters of pregnancy were associated with change in DBP from childhood to early adulthood, after multiple-testing correction ( Figure 1C). Higher humidity was associated with faster increase in DBP in . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Overall associations of humidity, mean temperature, PM 2.5 and PM 10 with blood pressure trajectories were similar in males and females, and there was little evidence for sex differences (Supplementary Table 11). The only two exceptions were for the association between PM 10 in the first trimester of pregnancy and slower increase in SBP in childhood, which was only evident in females, and the association between PM 10 in the third trimester of pregnancy and faster increase in DBP in childhood, which was stronger in males.

Replication in other cohorts
The distribution of the baseline characteristics, environmental exposures and outcomes at each follow-up in GenR, EDEN, PANIC and NFBC1986 are presented in Supplementary Table 12. The association of the confounders with the environmental exposures in all the cohorts is presented in Supplementary Table 13. Whilst some associations between the confounders and the environmental exposures were similar across the cohorts (e.g., those living in the most deprived areas had higher levels of PM 2.5 and PM 10 ), some associations differed (e.g., low maternal education was positively associated with PM 2.5 in ALSPAC and EDEN (Poitiers), whilst no association was evident in EDEN (Nancy) and an inverse association was observed in GenR).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ;

1
The associations of humidity, mean temperature, PM 2.5 and PM 10 with SBP and DBP at baseline in each study are presented in Supplementary Figure 5, and associations with change in SBP and DBP are presented in Figure 4. The association of humidity with faster increase in SBP in childhood replicated in GenR and EDEN (Poitiers), and with faster increase in DBP replicated in EDEN. An association between humidity and change in SBP in adulthood was not observed in NFBC1986. The associations of temperature with slower increase in SBP and DBP in childhood were also observed in GenR and EDEN, but the association of temperature with faster increase in DBP in adolescence did not replicate in PANIC. The association between PM 2.5 and faster increase in DBP in childhood was also observed in EDEN (Nancy), though CI spanned the null, and in GenR associations between PM 2.5 and slower increase in both SBP and DBP in childhood were observed. No data were available for PM 2.5 in adolescence. PM 10 was associated with faster increase in SBP in childhood in EDEN (Nancy) but with slower increase in SBP and DBP in childhood in GenR.
In GenR, the association of PM 10 in the first trimester of pregnancy with slower increase in DBP in childhood was also observed, but associations of PM 10 in the second and third trimesters of pregnancy with blood pressure were observed in the opposite direction to those observed in ALSPAC (Supplementary Figure 6).

Discussion
We assessed the association of 42 prenatal urban environmental exposures with blood pressure trajectories from childhood to early adulthood in ALSPAC, and sought replication of the findings in four independent European cohorts. After accounting for multiple testing, we found that humidity was associated with faster increase in SBP and DBP in childhood, slower decrease in SBP in adulthood and slower increase in DBP in adolescence. Temperature was associated with slower increase in SBP and DBP in childhood and faster increase in DBP in adolescence. PM 2.5 and PM 10 in different trimesters of pregnancy were associated with faster increase in DBP in childhood and slower increase in adolescence. Little evidence for sex differences in these associations was observed. Analyses in independent cohorts replicated results for associations of humidity and temperature with change in blood pressure in childhood. Most of the results for the association of air pollution with blood pressure change did not replicate. The one exception was replication of the association between PM 10 in the first trimester of pregnancy and slower increase in DBP in childhood.
Many studies have assessed the association of outdoor temperature with blood pressure, 7 but less evidence is available for humidity. In adults, humidity has been positively . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.31.23288002 doi: medRxiv preprint associated with blood pressure levels, 34 and a positive association with DBP has also been observed in children. 14 We were able to identify only one study which assessed the association of humidity during pregnancy with offspring blood pressure. This study, which included 1,277 European children did not find an association between prenatal humidity and blood pressure in children aged 6-11 years (mean difference per IQR increase in humidity: -0.76 mmHg, 95%CI -3.3, 1.77 for SBP, and -0.44 mmHg, 95%CI -2,79, 1.92 for DBP). 12 Findings from the same study cited above showed that outdoor temperature during pregnancy was associated with higher SBP and DBP, though CI for the latter spanned the null (mean difference per IQR increase in temperature: 1.6 mmHg, 95%CI: 0.2, 2.9, and 1.2 mmHg, 95%CI: -0.1, 2.4, respectively). 12 In 4,279 European children aged 4-5 years the association of prenatal temperature with SBP and DBP was more imprecise and CIs included zero (mean difference per IQR increase in temperature: 1.16 mmHg, 95%CI: -0.09, 2.41, and 0.45 mmHg, 95%CI: -0.68, 1.58, respectively). 14 Our results corroborate these findings and extend research in those studies by showing that higher prenatal temperature was associated with slower increase in SBP and DBP in childhood. This slower change in blood pressure during childhood could suggest that these associations might no longer be evident in the longterm, or that inverse associations between prenatal temperature and blood pressure might be observed in older ages. A study in China with 6,158 individuals (mean age 40 years, SD 9.0) showed an inverse association between temperature during pregnancy and blood pressure in adulthood. 35 Those born between April and August (and therefore with longer exposure to cold spells during pregnancy) had higher DBP (and also SBP in males only) than those born between September and March. 35 Outdoor temperature and relative humidity are often correlated (r= -0.50 in our study), and associations with blood pressure were similar when these exposures were mutually adjusted (Supplementary Table 14). The mechanisms underlying the association between outdoor temperature and blood pressure are not fully understood, but it is suggested that lower ambient temperature activates the sympathetic nervous system and increases vasoconstriction leading to increase in heart rate and blood pressure. 7,34 However, the mechanisms through which outdoor temperature (and humidity) during pregnancy might influence offspring blood pressure require further investigation.
Although overall prenatal air pollution-related factors were associated with faster increase in DBP in childhood, these findings did not replicate in the independent cohorts, except for the association of PM 10 in the first trimester of pregnancy and slower increase in DBP in childhood. It is possible that these contrasting results are driven by the different . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.31.23288002 doi: medRxiv preprint levels of exposure and different confounding structure across the cohorts; e.g., air pollution levels are higher in GenR than in the other studies, and the association between socioeconomic factors and air pollution differed across cohorts. Previous studies have shown that prenatal exposure to PM 2.5 and PM 10 was associated with higher blood pressure in newborns, 18,36 particularly exposure in the third trimester of pregnancy. 18 Prenatal NO 2 was associated with higher DBP in children aged 4-5 years (mean difference per IQR increase in NO 2 : 0.85 mmHg, 95%CI: 0.38, 1.33), but less evidence was observed for PM 2.5 and PM 10 (mean difference per IQR increase in PM 2.5 and PM 10  In our study, most associations with change in blood pressure were observed in childhood. It is possible that any effects of prenatal environmental exposures on blood pressure trajectory are no longer evident at later ages. However, it is important to highlight that contemporaneous exposure to urban environmental factors can affect blood pressure levels, 5,6,[9][10][11]37 and the long-term exposure to adverse environments might affect blood pressure trajectories across the life course.
Very few studies have explored whether the association between prenatal urban environment and child's health differs by sex. A study with 822 American children aged 4-6 years showed that associations between prenatal exposure to air pollution and child's DBP was stronger in females. 17 In adults, there is suggestion that the association between air pollution and blood pressure is stronger in males. 5 In our study, we found little evidence of sex differences in the associations between prenatal environmental exposures and blood pressure trajectories.

Strengths and limitations
We explored the longitudinal association between urban environment in pregnancy and offspring blood pressure using repeated data spanning from childhood to early adulthood. We used an ExWAS approach to assess the association of a range of urban environmental exposures with blood pressure trajectories, taking account of multiple testing and attempting to replicate findings in independent cohorts were possible, in order to minimise chance findings. Our analyses included all participants with at least one blood pressure measure, . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ; 1 4 under the missing at random assumption (i.e. the probability that an outcome value is missing depends on observed values of the outcome, conditional on the covariates in the model), which may have reduced selection bias due to missing outcome data. Violation of this assumption could bias our results, but when we repeated analyses only including those with at least 3 measures we found similar results. We restricted the analyses to those with complete data on the exposures and confounders, which gives unbiased estimates if having complete data is independent of the outcome, after taking the covariates into consideration. 27 It is not possible to test this assumption in our data, but having complete data is likely to depend on SBP (Supplementary Table 15). Whilst analyses for DBP are likely to be unbiased, analyses for SBP might be overestimated, since those included in the analysis had a higher increase in SBP in childhood. Misclassification is expected to be differential across exposures and different sample sizes might result in different statistical power. The age range of the cohorts included in the replication differed, and although we attempted replication of the results found in ALSPAC, power was limited in some age ranges. The lack of some of our pre-defined confounders in some of the replication cohorts might have influenced the replication results.

Conclusions
Using an ExWAS approach to systematically assess a range of prenatal urban environmental exposures, this study showed that prenatal outdoor temperature and humidity potentially modulate blood pressure trajectories, particularly in childhood. Our study contributes to the growing body of evidence on the longitudinal associations of prenatal environmental exposures with blood pressure later in life.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted March 31, 2023. ; . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  A  s  s  o  c  i  a  t  i  o  n  s  o  f  u  r  b  a  n  e  n  v  i  r  o  n  m  e  n  t  a  l  e  x  p  o  s  u  r  e  s  w  i  t  h  c  h  a  n  g  e  s  i  n  s  y  s  t  o  l  i  c  (  S  B  P  )  a  n  d  d  i  a  s  t  o  l  i  c  b  l  o  o  d  p  r  e  s  s  u  r  e  (  D  B  P  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint Beta corresponds to mean changes in blood pressure per interquartile range (IQR) increase/change in category in the exposure; the IQR for each exposure is presented in Supplementary Table 4. Models are adjusted for maternal education, age at delivery, ethnicity, area deprivation, and sex. The orange line shows the average predicted systolic blood pressure from childhood to early adulthood in the 25 th percentile of exposure, and the navy line in the 75 th percentile of exposure. Shaded areas correspond to 95% confidence intervals. All covariates used in the adjustment were set to the mean value or the reference category: maternal education (high), age at delivery (28.9 years), ethnicity (White), area deprivation (least deprived), and sex (male).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.31.23288002 doi: medRxiv preprint Figure 3. Average predicted population mean for diastolic blood pressure from childhood to early adulthood in the 25 th and 75 th percentiles of relative humidity, temperature, PM 2.5 and PM 10 at different trimesters of pregnancy.
The orange line shows the average predicted systolic blood pressure from childhood to early adulthood in the 25 th percentile of exposure, and the navy line in the 75 th percentile of exposure. Shaded areas correspond to 95% confidence intervals. All covariates used in the adjustment were set to the mean value or the reference category: maternal education (high), age at delivery (28.9 years), ethnicity (White), area deprivation (least deprived), and sex (male).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.