In this exploratory analysis of a large, population based, cohort of infants born in Stockholm, Sweden, we observed that children whose mothers were exposed to high levels of pollen during late pregnancy had an increased risk of hospitalisation for asthma, while children exposed to high levels of pollen in infancy had a reduced risk, but only in children of mothers who were heavy smokers. The potential mechanisms to explain the protective effect of high pollen exposure in children of mothers who smoked are unclear. We speculate birth during high pollen periods would increase the likelihood of pleasant weather, allowing the mother and newborn to spend substantial periods of time outdoors, and thus reducing the degree of passive smoke exposure in these children. Future research is needed to confirm this effect.
The primary question with regards to these results is if exposure to high levels of pollen in late pregnancy might be causal for increased risk of early life asthma, or is this association due to residual confounding? We have attempted to adjust for a range of potential confounders for which data were available (specifically maternal smoking, sex, gestational age, and season of birth). Interestingly, this effect was only apparent when adjusted for season of birth. That is, the seasonal effects masked this increased risk. As exposure to pollen at various stages of pregnancy and infancy is almost a random event, many factors (including but not limited to pet keeping, infant diet and day care attendance) cannot plausibly be considered to be potential confounders of this association. However, there remains the possibility that other factors that show a correlation with pollen levels may have created these associations. Both low vitamin D levels in the child and the mother, and early life respiratory tract infections[22–24] have a strong seasonal variation, and have been associated with elevated risk of wheeze and asthma in the child. If these events occur at a consistent time following the pollen season, when children are most vulnerable to these exposures, then a spurious association with pollen levels could be created. However, it should be noted that 1) we have adjusted for season of birth and markers of pollution 2) the pattern of results is based on actual pollen exposure rather than season, and 3) the associations appear to be consistent across the years of this study. Furthermore, we have used hospitalisation data for LRTI as a crude proxy for the rate of circulating respiratory pathogens that may cause wheezing illness in the child, and this did not alter the associations with pollen.
If the association between high maternal pollen exposure and increased risk of hospitalisation for respiratory illness is causal, then these results are consistent with a potential priming of either the neonatal immune, or respiratory system, towards a wheeze/asthma like phenotype. Season of birth is associated with a range of differences in cord blood cytokine production profiles[25, 26], and pollen may influence these patterns. It is likely that pollen sensitised mothers exposed to high levels of pollen during pregnancy are at increased risk of symptoms and asthma exacerbations. This may in turn change the intrauterine environment, predisposing the child to a Th2 type response. Alternatively, symptomatic mothers during high pollen periods may have increased risk pregnancy complications, including pre-eclampsia, preterm birth and impaired foetal growth, which could influence the risk of wheezing illness[28, 29]. Unfortunately, we do not have information concerning maternal asthma or pollen allergy in this data set to test this possibility. Further work is required to elucidate exactly how pollen exposure may prime the foetal immune system towards severe respiratory illness in early life.
Comparison with prior literature
The observed associations within this study may appear to be in conflict with prior literature on this topic. Kihlstrom et al. used data from 189 children born during the very high, 1993 pollen season in Sweden (data also included in this study) to argue that post-natal (0–3 months) exposure to pollen is more important than pregnancy exposure for the outcome of sensitisation to birch, and that maternal pollen allergy is more important than pregnancy pollen exposure for the outcome of childhood allergic rhinitis[11, 30]. These results could be seen as demonstrating that pregnancy exposure is less important than infancy exposure. However, the work of Kihlstrom also indicates that a high pregnancy exposure is associated with a small increase in risk of asthma (OR = 1.3, 95% CI-0.8-2.1), when compared with both high neonatal exposure, and low exposure to pollen. The pregnancy exposure group of Kihlstrom et al. included children who had high pollen exposure from approximately 13–31 weeks gestation[11, 30], combining exposures during 2nd and 3rd trimesters. Hurley et al. observed an association with high pollen exposures during infancy and increased risk of wheeze up to two years of age, but did not examine the effect of pollen exposure during pregnancy. It is important to note that these previous studies have examined the outcome of asthma and wheeze, rather than hospitalisation for asthma, which may explain the differences between study findings.
This study has a number of important strengths and limitations that should be considered when interpreting these results. The study’s key strength relates to its large sample size of mother/baby pairs that have been assessed over multiple years. This provides both statistical power, and the ability to assess these associations over multiple pollen seasons. Furthermore, rather than relying on season of birth we were able to assess the associations with actual ambient pollen counts. The pollen count data allowed us to take into account variations in the initiation, duration and intensity of each pollen season between years. The outcome measure of hospitalisation for asthma (ICD-9 code 493) in early life is important for the health care system and the quality of life of affected children and families. Wheezing symptoms are common in infants, but it is very difficult to determine the nature of wheezing illness in young children, and many of the admitted children will not have symptoms that persist into later childhood. As such, it is likely that children in this study who have been hospitalised for “asthma” do not truly have this condition, but another form of respiratory condition. In Sweden, a diagnosis of “asthma” has been recommended for children aged less than 2 years with persistent wheezing symptoms or after at least three wheezing episodes on separate occasions in early childhood. Also, it is more likely that diagnosis of asthma will be given if the child also has eczema or heredity for asthma (parents or siblings with asthma). Conversely, many children with wheeze in early life (that may be early onset asthma) are not hospitalised, but may be seen in outpatient setting. Unfortunately, we did not have access to data on outpatient visits for wheeze or asthma or use of inhalant therapy.
Also, we did not have access to information relating to other allergic disease phenotypes, including specific IgE and allergic rhinitis, pollen allergy and asthma at an age when the diagnosis is more reliable, all of which may have proven to be more informative. The measurement of pollen exposure (total pollen counts per day) was coarse, being captured in only one location, and it did not capture the species of the pollen or details of the smaller allergen carrying molecules, which may have resulted in misclassification of pollen exposure. The maximum distance from the pollen trap for those children included in this study was 17 km. The peak concentrations at this location are typically driven by high levels of birch pollen. Misclassification of pollen exposure will have biased associations towards the null, making it more difficult to observe any true associations. Future studies should attempt to make more personalised pollen exposure assessments. We also did not have data relating to early life respiratory tract infections for the actual child, nor to maternal asthma and allergic sensitisation, both of which have been proposed to interact with pollen exposure to influence risk of allergic disease[9, 30]. We are currently planning a study to address some of these limitations.