Abstract
Iron supplementation in pregnancy may reduce the risk of asthma in the offspring http://bit.ly/2UHmfid
To the Editor:
Evidence for the hypothesis that prenatal nutrition may influence the risk of childhood respiratory and atopic disorders has strengthened in recent years. For example, randomised controlled trials (RCTs) have suggested that maternal supplementation with vitamin D or fish oil in pregnancy could prevent early childhood wheezing and asthma in the offspring [1, 2]. Another nutrient of interest is iron. Animal experiments have implicated iron in fetal airway development [3]. In two UK population-based birth cohorts, using different approaches, we have found evidence suggesting that lower maternal iron levels in pregnancy may increase the risk of wheezing, atopy and lower lung function in the offspring [4–7]. However, the possibility that unmeasured or residual confounding might explain these observational findings cannot be ruled out. A study in the USA reported that maternal anaemia was associated with an increased risk of childhood asthma in offspring of mothers with asthma, but not in those whose mothers did not have asthma [8], although we did not confirm such effect modification [5].
To date, no RCT has been conducted to test whether prenatal iron is causally related to childhood respiratory and atopic outcomes, and whether increasing maternal iron levels in pregnancy could improve the respiratory health of offspring. Such a trial would be time-consuming and expensive. A faster, more efficient, approach is to opportunistically use a previous RCT of iron supplementation in pregnancy, originally undertaken with other outcomes in mind, and to measure respiratory/allergic outcomes in the offspring [9].
We followed up the offspring of mothers who took part in a population-based, multicentre, RCT of iron supplementation in pregnancy in Finland, using health register linkage to investigate whether prenatal iron influenced their risk of asthma. Full details of the trial have been reported previously [10–14]. Pregnant women were recruited during 1985–1986, and 2944 women were initially randomised, in each of 27 centres, to routine versus selective iron supplementation. After subsequent exclusions (e.g. twins and miscarriages), 1336 and 1358 women in each arm, respectively, were included in the original analyses of maternal and perinatal outcomes [10, 14]. The routine iron (RI) group received 100 mg elemental iron per day throughout pregnancy (i.e. from the first prenatal visit until the birth of the child, starting no later than the 17th week), regardless of haemoglobin concentration, which was usual care in Finland at the time. The selective iron (SI) group received 50 mg elemental iron twice a day, but only if women were diagnosed as anaemic; this continued for 2 months or until their haemoglobin increased to 110 g·L−1 [11, 12, 14]. Compliance as reported by the mothers was good [10]. Haematocrit was measured at 12, 20, 28 and 36 weeks gestation, and information was collected on maternal factors, gestational age and weight at birth, and sex of child.
Using unique personal identification numbers assigned to all Finnish residents, the RCT data were linked to national registers to ascertain the asthma status of the offspring of participating mothers. An asthma diagnosis was defined if an individual was found in the register for the right to special reimbursement of medication and in the register of reimbursed medication, maintained by the Social Insurance Institution (SII), and also had purchased asthma medication according to the drug prescription register (Anatomical Therapeutic Chemical R03). To be eligible for the special reimbursement for purchase of asthma medication, a physician had to sign a certificate confirming an asthma diagnosis and the need for long-term treatment. Given that the registration for reimbursed medication started in 1995, an asthma diagnosis was only ascertained between 10 and 32 years of age. Additionally, using the inpatient admissions register, maintained since 1967 by the Finnish Institute for Health and Welfare, Helsinki, we defined first asthma admission (primary or secondary reason for admission) using the International Classification of Diseases (ICD) codes: ICD-9 (4930A, 4930B, 4931B, 4939X) up to 1995; ICD-10 (J45-J46) between 1996 and 2017. Researchers abstracting these data were blind to treatment allocation. We analysed associations between trial arm and asthma diagnosis based on the SII reimbursement data (primary outcome) and asthma admissions (secondary outcome) using random-effects models (multilevel mixed-effects generalised linear model for the former, and multilevel mixed-effects survival model for the latter outcome). This allowed effects to vary between trial centres, given that randomisation was stratified by centre [15]. Risk ratios and hazard ratios, respectively, were adjusted for prognostic factors known to be associated with childhood asthma risk, namely, maternal factors (age, pre-pregnancy body mass index, smoking and asthma) and sex of child, in order to increase statistical power [16]. Analyses were by intention-to-treat, using Stata version 15 (StataCorp, College Station, TX, USA).
Of the 2694 women included in the original trial analysis, 2666 (99%) were included in our follow-up analysis (1320 in the RI arm and 1346 in the SI arm). Figure 1 shows mean haematocrit in pregnancy by gestational age and trial arm. In the last 2 months of pregnancy, haematocrit increased substantially more in the RI group than in the SI group (difference at 36 weeks gestation, p<0.001). There was no imbalance between trial arms in most background characteristics, but mothers in the RI arm were more likely to be underweight.
During the follow-up period, there were 74 individuals with doctor-diagnosed asthma based on the SII data, and 138 first-time hospital admissions for asthma. Compared to offspring in the SI arm, those in the RI arm had a significant reduction in risk of asthma diagnosis (adjusted risk ratio 0.58, 95% CI 0.35–0.96; p=0.033) and a non-significant reduction in risk of asthma admissions (adjusted hazard ratio 0.86, 95% CI 0.61–1.22; p=0.41). As there was a minor reduction in the prevalence of premature birth in the RI group compared to the SI group, we repeated the analyses with further adjustment for gestational age at birth, but this did not alter the risk estimates. When we stratified the analyses by maternal asthma, the reduction in risk of asthma diagnosis associated with routine supplementation was lower in offspring of mothers with asthma (adjusted risk ratio 0.33, 95% CI 0.08–1.28) than in those whose mothers did not have asthma (adjusted risk ratio 0.65, 95% CI 0.38–1.11). However, the very wide confidence intervals around the effect estimates reflect the lack of power to detect significant effect modification, and therefore these exploratory analyses should be viewed with caution.
This study has a number of strengths. First, the association between routine iron supplementation in pregnancy and reduced risk of asthma diagnosis in the offspring is unlikely to be confounded, given that the intervention was randomised. Second, the increase in haematocrit in mothers receiving routine supplementation, which was not seen in the selective supplementation group, confirms compliance with supplementation, suggesting that prenatal iron status is likely to have been significantly higher in the former group during the latter stages of pregnancy. With respect to asthma ascertainment, individuals identified in the reimbursement register are likely to represent the most severe asthma cases, given the strict criteria for reimbursement eligibility, so mild cases may not have been identified. Also, as the reimbursed medication register did not start until 1995, we could not ascertain an asthma diagnosis until the offspring were 10 years old, hence some cases under that age may not have been identified unless their condition persisted beyond that age. This limited our ability to assess the effect of iron supplementation on the incidence of early onset of disease, when most wheeze and asthma begins. However, cases persisting beyond 10 years of age have the greatest impact on the lives of sufferers and the health services. In our experience, those admitted with “asthma” to hospital are likely to include some individuals with acute infective episodes, e.g. acute bronchitis, in whom a firm diagnosis of asthma is not confirmed subsequently. We therefore believe that ascertainment of an asthma diagnosis using the reimbursement register will have led to less misclassification of asthma, compared to asthma ascertained using the admissions register, and this may explain the stronger findings for the former asthma outcome. On the other hand, we acknowledge that the reimbursement register approach will have led to under-ascertainment of mild cases and asthma beginning before 10 years of age. Importantly though, any misclassification of asthma is likely to be random (non-differential) with respect to trial arm, and this would tend to lead to an underestimation of effect. We also acknowledge that a trial comparing routine prenatal iron supplementation versus placebo would have enabled a more powerful test of our hypothesis; if the effects on asthma risk are causal we would expect such a trial to produce a larger effect than we observed in our study comparing two supplementation regimes. Whilst maternal iron deficiency is associated with lower birth weight and prematurity [17], which are both risk factors for asthma, there was little impact of routine versus selective supplementation on these perinatal outcomes in this trial. Hence, they did not appear to mediate the reduction in risk of asthma associated with routine iron supplementation. We speculate that improving prenatal iron status may have reduced the risk of more severe asthma by optimising airway growth.
In conclusion, these findings suggest that routine iron supplementation may reduce the risk of asthma in the offspring, and could therefore offer a potential primary prevention strategy. Additional robust, randomised evidence from a low- or middle-income setting, where iron deficiency anaemia in pregnancy continues to be a major public health problem, would be of interest.
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-02335-2019.Shareable
Footnotes
Conflict of interest: S.O. Shaheen has nothing to disclose.
Conflict of interest: M. Gissler has nothing to disclose.
Conflict of interest: G. Devereux has nothing to disclose.
Conflict of interest: M. Erkkola has nothing to disclose.
Conflict of interest: T.I. Kinnunen has nothing to disclose.
Conflict of interest: H. Mcardle has nothing to disclose.
Conflict of interest: A. Sheikh reports grants from Asthma UK, during the conduct of the study.
Conflict of interest: E. Hemminki has nothing to disclose.
Conflict of interest: B.I. Nwaru has nothing to disclose.
Support statement: This study was funded by the Tampere Tuberculosis Foundation. B.I. Nwaru acknowledges the support of the Knut and Alice Wallenberg Foundation, the Wallenberg Centre for Molecular and Translational Medicine, and the VBG Group Herman Krefting Foundation on Asthma and Allergy. A. Sheikh is supported by the Asthma UK Centre for Applied Research. Ethics approval was obtained from the Finnish Institute for Health and Welfare. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received December 5, 2019.
- Accepted February 4, 2020.
- Copyright ©ERS 2020