Early life exposure to pollens and increased risks of childhood asthma: a prospective cohort study in Ontario children

To the Editor:

Asthma is a disease characterised by wheeze, cough and shortness of breath, and constitutes the most prevalent chronic disease among children [1]. Various phenotypes have been specifically identified in the paediatric population, and include early transient wheeze, current wheeze/asthma, and mild or moderate asthma [2]. Lifestyle behaviours, genetics, maternal and paternal factors, and environment exposures have been identified as risk factors in the multifactorial aetiology of childhood asthma [3].

A growing body of evidence suggests that early life exposure to vegetation in urban settings might play a role in children's respiratory health and prevent the development of childhood asthma [4, 5]. This can be explained by several potential mechanisms. For instance, greener areas can reduce exposure to environmental factors harmful for respiratory health, including exposure to ambient air pollution, noise exposure/annoyance and exposure to extreme heat. Access to urban vegetation can encourage children to spend more time outside, which may have positive effects on physical activity and exposure to more diverse microbiota [4]. Conversely, exposure to urban vegetation can also co-occur with exposure to aeroallergen sources (e.g. airborne pollen). In fact, studies have found that prenatal and early infancy exposure to different pollen types was associated with an increased likelihood of respiratory symptoms in children as well as an increased risk of developing asthma during childhood [6–8]. However, exposure to airborne pollen has rarely been considered in studies investigating the effects of urban natural environments and childhood asthma.

This study investigated early life exposure to urban vegetation on childhood asthma incidence and whether associations could be modified by prenatal and early life exposure to airborne pollen.

We conducted a retrospective cohort study of singleton live births born between 1 April 2006, and 31 March 2014, in Toronto, Canada. We linked mother–infant pairs using data from the Better Outcomes Registry and Network (BORN) Ontario, a province-wide perinatal registry. Maternal residential address was captured at the postal code level. The exposure data were linked to the study cohort and analysed by the Institute of Clinical Evaluative Science (ICES). ICES is an independent, non-profit research institute funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). As a prescribed entity under Ontario’s privacy legislation, ICES is authorised to collect and use healthcare data for the purposes of health system analysis, evaluation and decision support. Secure access to these data is governed by policies and procedures that are approved by the Information and Privacy Commissioner of Ontario.

We used the Ontario ASTHMA cohort database to identify incident cases of childhood asthma (International Classification of Diseases-10: J45) between birth and age 6 years [9].

Environmental exposures were available for all postal codes and were linked to the BORN dataset using maternal postal code during pregnancy and following birth of the child, capturing residential mobility. Maternal exposure to residential greenness was assigned through a satellite-derived normalised difference vegetation index (NDVI) and using proportion of tree canopy area (also known as crown area) estimates in a 250 m buffer around the central location of the postal code at birth. We assigned residential exposure to airborne pollen types (i.e. total pollen, trees, grasses and weeds) derived from a land use regression (LUR) model that captures exposure at 1 km by 1 km spatial resolution over all weeks of pregnancy and infancy [10].

We conceptualised a directed acyclic graph (DAG) for identifying variables to adjust for in the models: ambient air pollution exposure during pregnancy and during childhood (i.e. particulate matter with aerodynamic diameter <2.5 µm, nitrogen dioxide and ozone), maternal age at delivery, infant sex, parity, breastfeeding status at the time of discharge, maternal smoking during pregnancy, maternal asthma, year of birth, season of birth (i.e. summer, spring, winter or autumn/fall), and an area-level socioeconomic status (SES) indicator based on the Ontario Marginalization Index (ON-Marg Index). The ON-Marg Index tool links individual postal code data to census tract-level data (i.e. small and relatively stable geographic areas that usually have a population between 2500 and 8000 persons) to assign inequalities in health and social wellbeing within Ontario at the neighbourhood level. The ON-Marg Index includes four dimensions of marginalisation: dependency, ethnic concentration, material deprivation and residential instability [11]. We included the ON-Marg Index dimensions as categorical quintiles in our models.

The associations between airborne pollen exposure, vegetation and the incidence of childhood asthma were assessed with Cox proportional hazards models using children's age (in years) as our time scale. We identified incident cases of asthma between birth and any of the following conditions: diagnosis of childhood asthma, death, becoming ineligible for provincial health insurance, moving out of Toronto, or end of follow-up (i.e. 31 March 2018 or attaining 6 years of age, whichever comes first). We used distributed lag models (DLMs) to estimate hazard ratios for childhood asthma incidence, per interquartile range (IQR) increase in weekly ambient pollen concentrations. DLMs incorporated average weekly concentrations of pollen types during gestational weeks 0–36 simultaneously, which combined a linear function for pollen variable and a natural cubic spline function with 3 degrees of freedom for modelling all gestational weeks in the same models. This was done to assess whether there were critical periods of exposure to airborne pollen during pregnancy. We adjusted all models evaluating prenatal exposures to airborne pollen for childhood exposures. Cumulative childhood exposure was ascertained as a time-running average from birth to each year of age of follow-up (average yearly exposure from 0 to 1 for the first year of life, average yearly exposure from birth to age 2 years, etc.). We analogously adjusted models assessing the cumulative effect of childhood exposures for pregnancy exposures. The indicators of vegetation NDVI and tree canopy area were modelled as continuous measures to estimate the hazard ratios on asthma incidence in relation to an increase in the IQR of each value measured during the first year of life. We assessed effect heterogeneity for the associations between vegetation and childhood asthma by tertiles of pollen concentration exposures during pregnancy and childhood. We also stratified associations by birth season and categories of material deprivation to further understand if the effects differ across different levels of SES. We obtained p-values for the interaction terms and for effect modification through likelihood ratio tests.

A total of 28 543 (13.3%) children were diagnosed with childhood asthma, among the 214 211 mother–child pairs included in the study. The mean age of asthma diagnosis was 22 months. Asthma cases had slightly higher prenatal average total pollen exposure when compared to non-asthma cases.

Exposure to NDVI (per 0.08-unit increase) was associated with an increased risk of childhood asthma (HR 1.029, 95% CI 1.008–1.035), while there was a protective effect for the impact of increased tree canopy area (per 35 393 m²) on asthma incidence (HR 0.976, 95% CI 0.960–0.991). A positive association between prenatal weed pollen and childhood asthma onset was found (for an IQR of 8 grains per m³: HR 1.021, 95% CI 1.008–1.035). We found a window of susceptibility for weekly weed pollen concentration during approximately the later part of the first trimester to the beginning of the third trimester. Childhood exposures during the first year of life to pollen concentrations showed that total pollen (HR 1.023, 95% CI 1.010–1.035) and tree pollen (per 81 grains per m³) (HR 1.004, 95% CI 1.002–1.006) were associated with childhood asthma risk. Childhood exposures during the first 3 years of life showed greater magnitude in risks of childhood asthma for total pollen (HR 1.078, 95% CI 1.059–1.098), but effects for total pollen appeared to be driven by weed pollen (HR 1.076, 95% CI 1.055–1.099) during the first 3 years of life.

The protective effect of tree canopy area exposure disappeared by increasing weed pollen concentration during childhood (i.e. HR 0.942, 95% CI 0.913–0.972 for first tertile of weed pollen concentration, versus HR 1.004, 95% CI 0.976–1.033 for the third tertile of weed pollen concentration) (p-value for effect modification=0.01) (figure 1b). The impact of NDVI on childhood asthma development was higher with increasing exposure to weed pollen concentration during childhood (i.e. HR 0.998, 95% CI 0.973–1.024 for first tertile of weed pollen concentration versus HR 1.053, 95% CI 1.029–1.079 for the third tertile of weed pollen concentration) (p-value for effect modification=0.01) (figure 1d).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Hazard ratios and 95% confidence intervals for asthma incidence per interquartile range (IQR) increase in normalised difference vegetation index (NDVI) and tree canopy (i.e. crown area) exposures among all term births, stratified by tertiles of pollen concentrations. Note: The models fitted were adjusted for ambient air pollution exposure during pregnancy and childhood (i.e. NO₂ and O₃), maternal age at delivery, infant sex, parity, breastfeeding status at the time of discharge, maternal smoking during pregnancy, maternal asthma, year of birth, season of birth (i.e. summer, spring, winter or autumn/fall), quintile groups of residential instability (categorical), maternal deprivation (categorical), dependency (categorical) and ethnic concentration (categorical).

Finally, the protective effect of tree canopy area was significant only for births occurring during the fall months (HR 0.949, 95% CI 0.920–0.979). We did not observe any differences in the effects of pollen types and vegetation measures on asthma incidence according to different levels of SES.

In this study, NDVI exposure was associated with a higher risk of childhood asthma development with the highest impact being as childhood weed pollen exposure increased. Exposure to tree canopy area conferred a protective effect on childhood asthma development but its efficacy decreased as childhood weed pollen exposure increased. Prenatal maternal exposure to weed pollen was associated with an increased risk of childhood asthma and exposure to total pollen, tree pollen and weed pollen during the first 3 years of life were associated with increased risk of childhood asthma.

Our findings are consistent with previous epidemiological literature suggesting that findings on early life exposure to residential greenness and urban vegetation and children's respiratory health might provide conflicting results according to the measurement being used [4]. We found that increasing weed pollen exposure decreased the beneficial effects of tree canopy and enhanced the risk of NDVI. This may provide some insights into how weed pollen control programmes should be considered when thinking about urban vegetation designs.

Exposure to green areas in urban environments may offer positive contact to microbiota [12], which may lead to positive development of the immune system through increased activity of natural killer cells as well as greater anti-inflammatory and anti-asthmatic effects [13, 14]. Our findings also suggest that a sensitive window for childhood asthma development is when exposure to weed pollen occurs between the later end of the pseudoglandular period, the canalicular period and the beginning of the saccular period. During this timeframe, cellular differentiation of the conducting airways begins, the air–blood barrier arises, alveolar formation within the parenchyma begins, formation of type I and type II epithelial cells occurs, and surfactant synthesis commences [15]. Another sensitive window identified was for exposure during the first 3 years of life.

Our study also has limitations. First, to account for tree canopy, we used data measured in 2007 and although vegetation has not changed substantially in urban built areas, misclassification of exposure is more likely to occur. Second, we did not have individual-level SES, full address history and other potential important risk factors for childhood asthma (e.g. maternal stress, second-hand smoke during childhood). Third, there is a lack of sensitivity and specificity in the diagnosis of asthma in early childhood as children may experience challenges with their breath coordination when performing pulmonary function tests (i.e. spirometry). Research suggests that diagnostic measurements, which currently use the same cut-off values for both adults and children, should be adjusted for the paediatric population [16].

Early life exposure to pollen, in particular weed pollen, may increase risk of childhood asthma development and may prevent the potential protective effect early life exposure to urban tree canopy confers.

Shareable PDF