Impact of ambient air pollution on lung function in preterm-born school-aged children

Rationale Increased outdoor air pollution worsens lung function in children. However, these associations are less well studied in preterm-born individuals. Objectives We assessed associations between ambient air pollutants and spirometry measures in preterm-born children. Methods The Respiratory Health Outcomes in Neonates study recruited preterm-born children aged 7–12 years who were born at ≤34 week's gestation. We associated four ambient air pollutants (particulate matter with aerodynamic diameter ≤2.5 µm (PM2.5), PM10, nitrogen dioxide (NO2) and sulfur dioxide) at time of birth and spirometry assessment and averaged exposure between these two time points with spirometry measures, using linear regression analyses. Gestational age was banded into 23–28, 29–31 and 32–34 week's. Regression models estimated spirometry values against pollutant levels at birth and at the time of spirometry. Measurements and main results From 565 preterm-born children, 542 (96%) had satisfactory data. After adjustments for early and current life factors, significant detrimental associations were noted between PM10 at birth and per cent predicted forced vital capacity (%FVC) for the 23–28 and 29–31 week's gestation groups and between current PM2.5 and NO2 exposure and %FVC for the 23–28 week's gestation group. No associations with spirometry were noted for the averaged pollution exposure between birth and spirometry. Predictive models showed 5.9% and 7.4% differences in %FVC between the highest and lowest current pollution exposures for PM2.5 and NO2, respectively, in the 23–28 week group. Conclusions Birth and current exposures to road-traffic-associated pollutants detrimentally affected %FVC in preterm-born school-aged children, who already have compromised lung function.


INTRODUCTION
It is well established that both antenatal and postnatal exposures to increased levels of air pollution have significant associations with childhood morbidity and mortality: our previous UK-based study demonstrated that sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ) and particulate matter with aerodynamic diameter ≤10 µm (PM 10 ) were differentially associated with all-cause neonatal and postneonatal mortality. 1Increased postnatal PM 10 exposure has also been linked to higher risk of death from respiratory causes in children and adolescents. 2Aside from mortality, increased air pollution has also been linked to poorer respiratory health and lung function in childhood, although the relationships are not always clear.Antenatal exposure to PM 10 , especially that generated by road traffic, has been linked to decreased forced expiratory volume in 1 s (FEV 1 ) and forced vital capacity (FVC) in 8-year-old offspring in the UK-based Avon Longitudinal Study of Parents and Children (ALSPAC) cohort. 3Postnatal exposure to NO 2 , PM 10 and PM 2.5 (PM with an aerodynamic diameter ≤2.5 µm) has also been differentially linked to significantly reduced FEV 1 and FVC in childhood in several studies, [4][5][6][7] with improvements in air quality over time associated with longitudinal increases in lung function. 8However, other studies have not demonstrated significant links between current air pollution and childhood lung function. 9 10Increased air pollution has been shown to be associated with worse lung function in individuals with underlying respiratory diseases.In children with asthma, exposure to SO 2 has been associated with reduced FEV 1 and FVC; and PM 10 with reduced FEV 1 /FVC ratio and peak expiratory flow. 11In the large multicohort

WHAT IS ALREADY KNOWN ON THIS TOPIC
⇒ Ambient air pollution from industry and road traffic sources has been linked to poorer respiratory health and reduced lung function.However, the impact on preterm born individuals, a significant proportion of who already have compromised lungs, has yet to be studied.

WHAT THIS STUDY ADDS
⇒ We have studied the effect of ambient air pollution at the time of birth and in childhood on spirometry measures in one of the largest contemporary preterm-born cohorts, finding links between elevated PM 10 , PM 2.5 and nitrogen dioxide exposure and reduced per cent predicted forced vital capacity but not on other spirometry measures.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
⇒ Despite pollution levels being within recommended limits, this study highlights the need for further efforts to reduce ambient air pollution, as it has a significant impact on the respiratory function of the most preterm born individuals in later life.

Paediatric lung disease
Evaluation Study of Congestive Heart Failure and Pulmonary Artery Catheterisation Effectiveness (ESCAPE) Study, long-term exposure to higher levels of NO 2 and PM 10 was associated with significant decreases in FEV 1 and FVC in adults 12 with a trend towards development of adult-onset asthma. 13aternal exposure to high levels of outdoor air pollution appears to be a risk factor for preterm birth, especially at extremely preterm gestations. 146][17] Second trimester PM 10 exposure has been shown to impact lung function at 44 week's postconceptional age in preterm-born infants, especially in those born at 32-36 week's gestation. 18For preterm-born infants with established BPD, living closer to major roadways following discharge home, and therefore, higher exposure to traffic-related air pollution, has been associated with increased respiratory morbidity. 19However, the impact of perinatal and childhood exposure to air pollution on the lung function of preterm-born school-aged children has been less well studied.We hypothesised that children born preterm, who were born at an immature stage of lung development with subsequent altered postnatal lung development, 20 were at risk of decreased lung function when exposed to common ambient outdoor pollutants.We examined the relationships between SO 2 , NO 2 , PM 2.5 and PM 10 (a) at the time of birth, (b) current exposure at assessment and (c) averaged pollution exposure between birth and spirometry assessment with spirometry measures in children aged 7-12 years who were born at ≤34 week's gestation.

METHODS
This study was conducted on the cohort of preterm-born children recruited to the Respiratory Health Outcomes in Neonates study (RHiNO; EudraCT: 2015-003712-20), which has been previously described. 16 21Briefly, children recruited as part of a previous questionnaire study 22 were supplemented with additional preterm-born children, sourced from the National Health Service Wales Informatics Service, and sent a respiratory and neurodevelopmental questionnaire, if they were born ≤34 week's gestation and aged 7-12 years old.Children were recruited from South Wales between 2016 and 2019.Children with significant congenital malformations, cardiopulmonary or neuromuscular disease were excluded.
Responders were invited to take part in a home or hospitalbased assessment by two trained research nurses.Neonatal history and current health status were ascertained from questionnaire responses and corroborated with medical records.BPD was defined as supplemental oxygen-dependency for 28 days of age or greater for those born <32 week's gestation and at 56 days of age for those ≥32 week's gestation. 23ntrauterine growth restriction (IUGR) was defined as a birth weight <10th centile adjusted for gestational age and sex (LMS Growth v2.77, Medical Research Council, UK 24 ).To evaluate contribution of non-respiratory neonatal illness, a severity score of zero or one, respectively, was created if none or one or more of the following diagnoses were present in the neonatal period: necrotising enterocolitis, intraventricular haemorrhage, retinopathy of prematurity and/or patent ductus arteriosus.Spirometry (Microloop, Care-Fusion, UK) was performed to American Thoracic Society (ATS)/European Respiratory Society (ERS) guidelines 25 and normalised using Global Lung Initiative references to give per cent predicted lung function measures. 26Spirometry was performed at the time of recruitment.Averaged annual estimated measurements of four common air pollutants, PM 2.5 PM 10 , NO 2 and SO 2 , were obtained from the UK Government's Department for Environment, Food and Rural Affairs (DEFRA) as previously reported, 1 for the respective calendar year in which the child was born and for when spirometry was performed.The UK Government collected this data as part of the European Union's Air Quality Directive from a network of monitoring stations throughout the UK, and verified by modelling to create pollutant levels for 1 km-by-1 km areas for each calendar year. 27To assess socioeconomic status, we used the most contemporaneous geographical assessed Welsh Index of Multiple Deprivation (WIMD) scores for 2014. 28The scores are calculated from composite measures of eight domains of deprivation including wealth, school achievement and home ownership.WIMD scores are given for 1909 geographical areas of similar population size (termed lower layer super output areas (LSOA), containing an average of approximately 1500 residents or 650 households).The LSOAs were supplied with x-coordinates and y-coordinates, on the same scale as the DEFRA 1 km-by-1 km grid, for its centre of mass-based on population distribution within the LSOA.The pollutant level for each combination of LSOA and time point was allocated by determining the closest, that is, minimum Euclidean distance, of the DEFRA grid points to each LSOA's centre of mass to give pollutant exposure and WIMD scores for each participant at year of birth and at the time of spirometry.
Demographic variables were compared between groups using analysis of variance or χ 2 pairwise comparisons as appropriate, both with Benjamini-Hochberg post hoc correction.The differential associations of each pollutant at time of birth, age at spirometry and averaged pollutant exposure scored between birth and current age with per cent predicted lung function measures (including %FEV 1 , %FVC, FEV 1 /FVC ratio and %FEF 25%-75% (forced expiratory flow at 25%-75% of FVC)) were tested through univariable and multivariable linear regression models.Participants were grouped into three gestational age bands: 23-28, 29-31 and 32-34 week's of gestation for analysis.The coefficients to be estimated were the estimands of main scientific interest, so gestation and pollution were included as an interaction term, whereby the beta values generated allowed direct comparison between the impact of pollution on lung function at the three gestation bands.Significant variables identified by univariable linear regression were visualised graphically in a directed acyclic graph.In the multivariable linear regression model, the continuous pollution-level variables and the categorical preterm group variable were included as interaction terms and in such a way as to cause the prematurity group variable to be treated like a dummy variable, consequently producing different pollution level coefficients for each prematurity group without requiring a reference category.This model was designed to include univariable linear regression results with p<0.1 for any spirometry variable, or early life factor including sex, BPD, IUGR and maternal age at birth (banded into six 5-year groups) and assessed the effect of gestational age in bands as an interaction term with pollutant level.
To explore the relationship between how different levels of ambient air pollution exposure at birth and at the current time of assessment would affect lung function values, an additional regression model was created, based on the univariable

Paediatric lung disease
and multivariable linear regression results, to give estimated spirometry values at minimum and maximum pollutant exposures within this cohort.Those variables with p<0.1 for the univariable relationship with any of the lung function measures were combined into a multivariable linear model, including neonatal history of BPD, combined illness, IUGR and maternal age.Gestational band and raw pollutant level were included in a single interaction term to provide a simply interpretable beta coefficient for the effect of the pollutant at each gestation band.These models were used to calculate the predicted influence of the pollutant, at birth and at time of spirometry, over the full range of its measured values at these times on the lung function measures of %FEV 1 , %FVC, FEV 1 /FVC ratio and %FEF 25%- 75% for each of the three gestational age bands.This included predicted spirometry values at minimum, mean and maximum pollutant exposure.All analyses were performed in SPSS V.25 (IBM) and R V.1.1.419.

RESULTS
From 565 preterm-born children taking part in the RHiNO study, 544 (96.3%) had valid spirometry data available as previously described. 16Two additional children were excluded, one for incorrect birth weight and one for incomplete address data.Since pollution data at time of birth was not available for six individuals due to changes in geographical boundaries, they were excluded from the 'pollution exposure at the time of birth' analyses only.The demographics for the included 542 children are shown in table 1.As expected, there was a significant difference between gestation groups for birth weight and rates of BPD, combined neonatal illness and wheeze-ever, with the most preterm-born group being affected the most.All lung function measures were significantly reduced in the most pretermborn group compared with the least preterm-born.Measures of deprivation and pollutant levels were not significantly different between the groups at birth or at time of spirometry (online supplemental tables 1 and 2, respectively).Table 2 shows the range of air pollutant levels within the study compared with contemporary recommended limits.Results of univariable regression analyses are shown in table 3. Significant negative associations were noted between spirometry variables and demographic variables including gestational age band (%FEV 1 , %FVC, FEV 1 /FVC ratio and %FEF 25%-75% ), BPD (%FEV 1 , %FVC, FEV 1 /FVC ratio and %FEF 25%-75% ), combined neonatal illness (%FEV 1 , %FVC, FEV 1 /FVC ratio and %FEF 25%- 75% ), younger maternal age at birth (%FEV 1 , %FVC, FEV 1 /FVC ratio and %FEF 25%-75% ) and history of IUGR (%FEV 1 , %FEF 25%- 75% ).These are represented graphically in a directed acyclic graph in online supplemental figure 1.No associations were noted between lung function measures and antenatal/postnatal smoking or deprivation.A significant association was observed between PM 10 level at birth and %FVC (β=−0.52,p=0.036).
Figure 1 and online supplemental table 3 show the results of the multivariable regressions.Significant negative associations were noted between %FVC for those born at 23-28 (β=−0.66,p=0.016) and at 29-31 week's gestation (−0.58, p=0.027) and PM 10 exposure at birth.The %FVC results for those born at 32-34 week's gestation had β of −0.49 and p value of 0.058 for PM 10 exposure at birth.Current exposure to PM 2.5 and NO 2 had a negative association for %FVC in the 23-28 week's gestation group (−1.19, p=0.031; −0.29, p=0.05, respectively).The associations for the 29-31 week's gestation group and PM 2.5 and NO 2 were β of −0.92 and p value of 0.070 and β of −0.19 and p value of p=0.083, respectively).No significant associations were noted between the pollutants and %FEV 1 or %FEF 25%-75% .The level of ambient air pollution decreased between the two time points measured.We did not observe any significant associations between the calculated mean pollutant exposure between these two time points (birth and current exposure levels) for PM 2.5 , PM 10 , NO 2 or SO 2 and any spirometry measure (online supplemental table 4).
As described in the methods, additional regression models were created based on the univariable and multivariable linear regression results for estimating spirometry values over the total range of pollutant exposures within this cohort.The results from these models for birth and current pollution exposure and %FVC are shown in table 4 and figures 2 and 3; and in online supplemental file 1 for %FEV 1 , FEV 1 /FVC ratio and %FEF 25%-75% (online supplemental tables 5-7 and figures 2-4, respectively).Online supplemental table 8 gives the decile measurement for each pollutant.PM 10 exposure at birth was associated with significant decrease in %FVC for those born 23-28 week's and 29-31 week's gestation, with 6.95% (p=0.02) and 6.1% (p=0.03)reduction in %FVC, respectively, between lowest and highest PM 10 exposures.No other significant associations were noted for %FVC and pollutant levels at time of birth.No significant differences were seen for %FEV 1 , FEV 1 /FVC and %FEF 25%-75% .
There was a difference for %FVC of 5.94% (p=0.03) for current PM 2.5 and 7.44% (p=0.05) for current NO 2 exposure for preterm-born children born at 23-28 week's gestation when the lowest and highest exposures were compared.The predictive models did not demonstrate any significant differences between %FVC and SO 2 at time of spirometry for any gestational group.No significant differences were noted for current exposure levels and predicted differences in %FEV 1 , FEV 1 /FVC and %FEF 25%- 75% between the lowest and highest exposures.

DISCUSSION
In this study of preterm-born children, we have shown that current exposure to increasing levels of two major ambient air pollutants, PM 2.5 and NO 2 , was associated with significantly decreased %FVC in a graded manner with decreasing gestational age.We have also reported that increased PM 10 exposure at the time of birth was associated with reduced %FVC in the same gestationally graded manner as postnatal exposure.The averaged exposure between birth and spirometry assessment was not associated with any of the spirometry measures.However, we did not note any association between pollution levels at any time point with either %FEV 1 or %FEF 25%-75% .Furthermore, birth, current and averaged SO 2 exposures were not associated with any spirometry measures.We generated adjusted regression models which showed significant relationships between %FVC and PM 10 for exposures at birth; and between %FVC and PM 2.5 / NO 2 for current pollution exposures.These effects occurred in a gestationally graded manner with the most prematurely born children affected the most.
Increased ambient outdoor air pollution has been strongly associated with increased respiratory morbidity and mortality in adult populations. 29Our previous study suggested that both antenatal and postnatal exposure of the four pollutants studied here can affect both neonatal and postneonatal mortality possibly via antenatal maternal exposure and postnatal exposure. 1Pretermborn infants have previously been shown to be more susceptible to the detrimental effects of antenatal exposure than term-born infants, with increasing antenatal exposure to PM 10 associated with worse lung function and higher fractional exhaled nitric oxide at 44 week's postmenstrual age. 18In addition, antenatal exposure to increased PM 10 has been demonstrated to have a detrimental impact on placental growth and function, 30 and on birth weight. 31 32Our current data have demonstrated that perinatal exposure to PM 10 could have lasting effects resulting in decreased %FVC in children who were born at <32 week's gestation.These infants are likely to be at higher risk of airway epithelial injury which could potentially explain the detrimental

Paediatric lung disease
effects we observed for pollutants on %FVC via the resultant injury on the lung parenchyma.This association is in keeping with findings from the ALSPAC cohort 3 which noted an association between antenatal PM 10 maternal exposure and reduced FVC in 8-year-old offspring who were, however, born at term.Unlike the ALSPAC study, we did not observe an association between pollutants and %FEV 1 .The reasons for this are unclear.
It is known that larger airways are compromised after preterm birth. 17Whether these airways are already so compromised in early infancy, especially in the lowest gestation group, such that additional epithelial injury from pollutant exposure has little further effect is very speculative and needs further study.Several previous studies, largely of term-born children, have also reported that outdoor air pollution exposure has a negative impact on lung function in paediatric and adult populations, but whether such exposure affects preterm-born children is less known.A Swedish study reported that high exposure to trafficrelated air pollution (nitrogen oxides (NO x ) and PM 10 ) in the first year of life was associated with decreased FEV 1 and FVC 33 and, although NO x and PM 10 in the same cohort were associated with reduced FEV 1 at 8 years of age, no association was noted between pollutant exposure and FVC. 34A Norwegian study examined NO 2 , PM 2.5 and PM 10 exposure in the first 2 years of life and in children up to 9-10 years noting a strong association between increasing air pollution exposure and reduced forced expiratory flows (peak expiratory flow, FEF 25% and FEF 50% ) but no associations were noted for FVC.Longer-term exposure to high air pollution had a stronger association with reduced expiratory flows. 35Another California-based study examined the effects of air pollution on changes of lung function over an 8-year period in children aged 10-18.They noted that exposure to higher levels of NO 2 and acid vapour (a mixture of inorganic and organic acids, primarily produced by motor engines) resulted in slower growth of FVC over the study period, on average being 95 mL and 105 mL, respectively, lower than those exposed to low air pollution.Significant reduction in the growth of FEV 1 was also associated with higher exposure to NO 2 , acid vapour and PM 2.5 , with a near-significant association with PM 10 . 36A recent Chinese study demonstrated that long-term exposure to PM 2.5 had a negative impact on lung function in children aged 7-12 years, with a one unit increase in the average daily dose exposure to PM 2.5 resulting in 14.8 mL decrease in FVC and a 10 mL reduction in FEV 1 . 37Why there are differences between cohorts for spirometry measures for the common pollutants is unclear especially as the pollutants' exposures were similar between the cohorts.We did not see a significant relationship between FEV 1 and pollution exposure in our cohort and the reason for this is unclear from our data.We can hypothesise that the degree of FEV 1 impairment occurring solely as a consequence of preterm birth, which we have previously demonstrated, 17 is so significant that the impact of ambient air pollution on FEV 1 in this population is masked.Additional studies, especially using personal exposure devices, are required to determine if FVC is commonly affected in all preterm-born cohorts.
Road traffic is the primary source of PM 2.5 , PM 10 and NO 2 air pollution especially in urban areas, primarily from petrolengine and diesel-engine exhaust fumes, and from vehicle tyre and brake wear.SO 2 is more associated with industrial activity and power station emissions. 38We noted an improvement in air pollution levels between birth and time of spirometry, with the majority of pollution levels being below European Union legal limits and WHO guidelines. 38However, even within these recommended limits, we have noted air pollution continues to have a detrimental association with FVC.This highlights the Table 3 Continued paucity of evidence on constitutes a 'safe' level of air pollution exposure.As many of the associations were seen between FVC and PM 2.5 , PM 10 and NO 2 , this highlights the importance of concerted efforts to reduce road traffic and fossil-fuel burning vehicle engines in urban areas to prevent further detrimental effects on preterm-born children's respiratory function.We did not see a significant association between deprivation and lung function in univariable modelling for our cohort, however, deprivation has been linked to higher air pollution exposure, 39 respiratory morbidity and decline in lung function 40 41 in other studies.
Infants delivered at extremely preterm gestations (23-28 week's gestation) are born at an immature stage of lung development, at late canalicular and early saccular phases. 42Subsequent ex utero aberrant lung development, partially due to postnatal interventions such as respiratory support and increased supplemental oxygen therapy, then results in structurally and functionally altered lungs for these infants in later life resulting in Figure 1 Plot of beta-values and 95% CIs from multivariable linear regression models of lung function parameters and pollutant exposures by each gestational age band (adjusted for sex, IUGR, BPD, combined neonatal illness and maternal age).Starred values represent betas with a p<0.05.BPD, bronchopulmonary dysplasia; %FEV 1 , per cent predicted forced expiratory volume in 1 s; %FVC, per cent predicted forced vital capacity; %FEF 25%-75% , forced expiratory flow at 25%-75% of forced vital capacity; IUGR, intrauterine growth restriction; NO 2 , nitrogen dioxide; PM 2.5 , particulate matter with aerodynamic diameter ≤ 2.5µm; PM 10 , particulate matter with aerodynamic diameter ≤10 µm; SO 2 , sulfur dioxide.significant respiratory morbidity. 17Our data showed that children born at extremes of gestational immaturity were most susceptible to the negative effects of several ambient outdoor air pollutants and may further impact lung growth as the primary effect we have observed was on FVC.Although the changes in lung function between lowest and highest pollution levels may seem small in childhood, these trajectories are likely to enhance with time given that lung function tracks over time 43 with consequent early development of chronic obstructive pulmonary disease as recently postulated. 44Therefore, the negative effects on lung function of higher levels of ambient outdoor air pollution on this already vulnerable population cannot be ignored.Higher FVC, a proxy measure of lung size, has previously been shown to be positively associated with increased life expectancy in adults, 45 and therefore, this already vulnerable population may have increased early respiratory mortality if exposed to high levels of air pollution.
Strengths of this study include a large preterm-born cohort who represent a wide range of gestational ages at birth, with good representation of extremely preterm-born subjects, which permitted us to delineate the different associations of air pollution and lung function for individuals born at a range of different stages of lung development.We had comprehensive antenatal, neonatal and postnatal medical histories, which allowed us to model our results including important neonatal comorbidities such as BPD and IUGR.We used robustly modelled, publicly available pollution data, which allowed us to accurately ascertain individual participants exposures both at birth at time of current assessment.include that we studied ambient outdoor air pollution but not indoor air pollution exposures, which may further adversely impact on lung function.Our measures of pollution exposures were based on small geographical area exposures and not on personal exposure thus may impact on our findings especially if localised exposures were either enhanced or reduced.Similarly, measures of deprivation, such as WIMD, are proxy measures of socioeconomic status based on geographical location.Even in small areas exposure may differ between close neighbourhoods, for instance, due to proximity to major roads or highways.Smoking status was self-reported by parents and may have been underreported, explaining the lack of association seen in our study

Paediatric lung disease
with spirometry values.Our model provided useful hypothesis generating results but requires validation in another similarly sized cohort which is difficult to source.
In conclusion, our study has demonstrated that exposure to traffic-related air pollutants, both perinatally and postnatally, have detrimental association with %FVC especially for those children born extremely preterm (<28 week's gestation).In our adjusted regression models based on current pollution exposure, we have shown that there may be up to 5.9% and 7.4% differences in %FVC between highest and lowest current pollution exposures for PM 2.5 and NO 2, respectively.These data highlight the importance of air quality strategies to reduce any further detrimental impact on lung growth and function in this group of individuals with likely aberrant lung development, and who are already at heightened risk of long-term respiratory morbidity. 17 Correction notice Since this paper first published, affiliations have been updated.

X Christopher William Course @chriscourse
Acknowledgements We are extremely grateful to all the children and families who took part in the RHiNO study for all their support and enthusiasm and to the nursing staff who were responsible for assessing and collecting the data.We also thank the neonatal consultants and administration teams in the South Wales neonatal units for assisting in identifying the children and for acquiring the medical notes.
Contributors SK conceived and designed the study.MC and KH were responsible for recruiting the children and for data collection.WJW, SJK, CWC and SK were involved in the data analysis and interpretation.CWC, WJW and SK drafted the manuscript.All authors were involved in revising the manuscript and approved the final submitted version.SK accepts full responsibility for the work and/or the conduct of the study, had access to the data, and controlled the decision to publish.

Figure 2 (
Figure 2 (A-D) Per cent predicted FVC values for each gestational age group from adjusted regression model by each pollutant decile at birth.Points represent mean value.Vertical bars represent 95% CI.FVC, forced vital capacity; NO 2 , nitrogen dioxide; PM 2.5 , particulate matter with aerodynamic diameter ≤ 2.5µm; PM 10 , particulate matter with aerodynamic diameter ≤10 µm; SO 2 , sulfur dioxide.

Figure 3 (
Figure 3 (A-D) Per cent predicted FVC values for each gestational age group from adjusted regression model by each pollutant decile at current exposure.Points represent mean value.Vertical bars represent 95% CI.FVC, forced vital capacity; NO 2 , nitrogen dioxide; PM 2.5 , particulate matter with aerodynamic diameter ≤ 2.5µm; PM 10 , particulate matter with aerodynamic diameter ≤10 µm; SO 2 , sulfur dioxide.

Table 2
Ambient air pollution levels in study and recommended annual exposure limits

Table 3
Univariable linear regression analyses of spirometry measures

Table 4
Regression model for FVC values by gestational age band over minimum to maximum exposure to PM 2.5 , PM 10 , SO 2 and NO 2 at birth and current time points Regression model for %FVC values over range of current pollutant exposures measured.Beta and significance level for relationship between difference in lung function and pollutant levels for each gestational age band.%FVC, per cent predicted forced vital capacity; NO 2 , nitrogen dioxide; PM 10 , particulate matter with aerodynamic diameter ≤10 µm; PM 2.5 , particulate matter with aerodynamic diameter ≤2.5 µm; SO 2 , sulfur dioxide.