Objectives To investigate a proposed association between in vitro fertilisation (IVF) and child asthma.
Design The risk for asthma after IVF was estimated as ORs using Mantel–Haenszel analysis.
Setting The Swedish Medical Birth Register.
Patients Of the 2 628 728 children born in 1982–2007 and surviving the perinatal period, 31 918 were conceived by IVF. Presence of asthma was defined as at least five prescriptions of antiasthmatic drugs during the period 1 July 2005–31 December 2009 according to the Swedish Prescribed Drug Register (115 767 children, 2323 of whom were born after IVF).
Results A significantly increased risk for asthma, albeit small, was found in children conceived by IVF (aOR 1.28, 95% CI 1.23 to 1.34), increasing the absolute risk from 4.4% to 5.6%. The risk increase for asthma was the same in boys and girls, in singletons and twins, and after caesarean section and vaginal delivery. The risk was higher for preterm than term singletons. For children with a low Apgar score, respiratory diagnoses, mechanical ventilation, continuous positive airway pressure or neonatal sepsis, the effect of IVF on asthma risk was low and statistically non-significant. Adjustment for length of involuntary childlessness eliminated the effect, and removal of infants whose mothers had used antiasthmatics in early pregnancy reduced the risk.
Conclusions This study verifies an association between IVF and asthma in children. This can be partly explained by neonatal morbidity and by maternal asthma acting as mediators, but the main risk factor is parental subfertility. The mechanism for this is unclear.
Statistics from Altmetric.com
What is already known on this topic
Only a few previous studies have investigated the possible risk of asthma in children conceived by in vitro fertilisation.
There are two relatively large studies, one of which did and one of which did not find an increased risk for asthma.
What this study adds
This is the largest study yet carried out and demonstrates a statistically significant increase in asthma risk after in vitro fertilisation (IVF).
The increased risk is explained by parental subfertility, but neonatal pathology among IVF infants may be a factor.
It has repeatedly been demonstrated that infants born after in vitro fertilisation (IVF) show an increased risk for preterm birth, low birth weight and neonatal morbidity.1–3 There are more multiple births following IVF than in the general population even though the rate of twins and higher order multiples has declined during the last decade as policy on the number of transferred embryos has changed.4 However, among singletons neonatal complications also occur at an increased rate after IVF and include preterm delivery, low Apgar score and respiratory problems.1 ,2 ,4 These conditions may result in increased morbidity later in life. Some late sequelae are more common than expected among children born after IVF, such as cerebral palsy,5 mainly due to the high rate of multiple births. The risk will therefore probably decline with the decreasing rate of multiple births after IVF. Other long-term morbidity has also been described as a consequence of IVF, for example, a slightly increased risk for cancer.6
In recent years, some epidemiological studies have identified premature birth, neonatal morbidity and caesarean section as risk factors for childhood asthma.7–11 As such neonatal conditions are more common among infants conceived by IVF, these children may have an increased risk of developing asthma. The literature on the subject is limited. Some studies investigating morbidity in children conceived by IVF mentioned asthma and described an approximately 30% risk increase.1 ,12 However, one study found no such risk,13 while other small studies gave inconsistent results.14–16
Material and methods
The Swedish Medical Birth Register contains information on most births in Sweden since 1973 (1–2% are missing).17 Information on year of delivery, maternal age in completed years (5-year groups: <20, 20–24, etc), parity (number of previous infants born: +1, 2, 3, ≥4), smoking in early pregnancy (unknown, none, <10 cigarettes per day, ≥10 cigarettes per day) and prepregnancy weight, height and body mass index (BMI; <19.8, 19.8–24.9, 25–29.9, ≥30) was collected. Pregnancy duration (in completed weeks) was known in nearly all cases and was usually estimated from a second trimester ultrasound. Small for gestational age (SGA) and adequate for gestational age (AGA) infants were identified using sex and parity specific growth graphs based on data from the Medical Birth Register.18 Caesarean section or vaginal delivery was recorded and information on neonatal diagnoses was coded according to the International Classification of Diseases. Register information on cohabitation in early pregnancy was obtained. Information on maternal use of antiasthmatic drugs was acquired from midwife interviews at the first antenatal visit. Records for infants dying in the perinatal period were excluded from analysis.
Identification of births after conception using IVF
In Sweden IVF is performed in only a limited number of clinics, and the National Board of Health and Welfare has collected data on all IVF procedures which resulted in a delivery for the period 1982–2006.19 As all residents of Sweden have a unique personal identification number, the maternal identification number was used for linkage with the Medical Birth Register.
Identification of asthma in children and young adults
Since 1 July 2005 a register of all filled prescriptions with the identification numbers of the patients has been maintained in Sweden.20 From this register, records referring to antiasthmatic drugs (ATC code R03) were extracted up to the prescription filling date of the 31 December 2009 for children born in 1982–2007. For each child, the number of prescriptions filled and the number of events when prescriptions were filled (as more than one drug could be prescribed at each event) were calculated but restricted to prescriptions filled after the age of 2 years as an asthma diagnosis before the age of 2 is regarded as imprecise.9
Children without data in the Medical Birth Register including all immigrant children, were excluded from analysis. To adjust for putative confounders, Mantel–Haenszel analyses were performed to generate ORs, and the approximate 95% CIs were estimated with Miettinen's method. Adjustments were made for year of delivery, maternal age, parity, smoking and BMI. Various subanalyses were carried out after stratification, for example, by infant gender.
Two ORs were compared based on the variance estimates in the Mantel–Haenszel analyses. Weighted linear regression analyses of the log (ORs) were used to detect a linear trend in the ORs.
The study was performed within the responsibilities of the National Board of Health and Welfare and therefore no ethics approval from external ethics committees was required.
During the observation period, 2 628 728 infants surviving the perinatal period were registered in the Medical Birth Register, 31 918 of whom were conceived through IVF. Among all children, 388 328 (15%) had at least one prescription for an antiasthmatic drug, while the corresponding number was 6597 (21%) among IVF children. The OR for having an antiasthmatic drug prescribed after adjustment for year of birth, maternal age, parity, smoking and BMI was 1.18 (95% CI 1.14 to 1.21).
Table 1 shows that the OR for asthma among children born after IVF increases with the number of prescription events for each child, while the number of cases simultaneously decreases. Further analysis will be restricted to children who had at least five prescriptions filled, constituting 7.3% of IVF children and 4.4% of all children in the population who had antiasthmatic drugs. The adjusted OR was 1.28 (95% CI 1.23 to 1.34).
Putative confounders among maternal characteristics
Table 2 shows that there is a higher chance of an infant being born after IVF as maternal age increases, but only a weak increase in the risk that the child will have asthma. There is a strong over-representation of first parity children after IVF and a weaker increase in asthma risk in first parity children. Maternal smoking is lower with IVF but results in a slight increase in asthma risk. With increasing BMI, the chance of IVF and risk of the child having asthma both increase. Subfertility is obviously always a factor for couples who have IVF and table 3 shows that there is a moderately increased risk for childhood asthma with subfertility (excluding cases with known IVF).
Women who immigrated to Sweden have a lower rate of IVF than Swedish-born women. Their children may have other risk factors for asthma compared to the children of Swedish-born women but exclusion of women born outside Sweden did not change the OR which was 1.27 (95% CI 1.21 to 1.33). Nearly all women who undergo IVF are cohabiting. Restricting the analysis to cohabiting women gave an OR of 1.29 (95% CI 1.23 to 1.35).
The reported maternal use of antiasthmatic medicines in early pregnancy was analysed in order to estimate the impact of maternal asthma on the association between IVF and asthma in their offspring. Such data were only available after 1 July 1995, and for this period, the OR for asthma in children conceived by IVF was calculated as 1.28 (95% CI 1.22 to 1.34). Of all infants, 4833 had asthma among the 36 397 whose mother had reported use of antiasthmatic drugs in early pregnancy (13.3%), while among 732 such infants conceived by IVF, 127 had asthma (17.3%). Among women using antiasthmatic drugs, the risk for asthma in children conceived by IVF was 1.15 (95% CI 0.94 to 1.42), representing only about 55% of the risk for all children and not statistically significant.
When adjustment was made for parental subfertility (measured as number of years of unwanted childlessness), the effect of IVF on the risk of childhood asthma disappeared: OR 1.02 (95% CI 0.98 to 1.09).
Infant characteristics as mediators of the effect of IVF on asthma risk
As seen in table 4, there was no difference in the increase in risk for asthma in IVF children according to infant gender. As expected, the sex ratio of children with asthma was significantly increased, both (males/females) after IVF and in the population. The oldest children in the cohort had the highest risk for asthma, but the trend between the three birth periods did not reach statistical significance (z=1.40, p=0.15).
Twins had a higher OR than singletons, but this difference did not reach statistical significance (z=1.94, p=0.06). Among singletons, the OR was higher in children born preterm than in children born term and this difference was statistically significant (z=2.17, p=0.04). There was no significant difference in the risk for asthma associated with IVF between children born after caesarean section and children born vaginally (z=0.53, p=0.35).
For five conditions (5 min Apgar score <7, neonatal respiratory diagnoses, use of mechanical ventilation, use of continuous positive airway pressure (CPAP) and neonatal sepsis), the ORs were low and not statistically significant.
When the analysis was restricted to term singleton AGA children, delivered vaginally, with an Apgar score at 5 min of ≥7, and without a neonatal diagnosis of respiratory problems, mechanical ventilation or CPAP, sepsis or neonatal icterus, the OR was 1.16 (95% CI 1.08 to 1.24). This estimate is based on 14 512 children conceived by IVF of whom 909 had asthma, and 1 940 731 children in the population of whom 79 150 had asthma. Thus, this selected group had about half of the risk increase.
The advantage of the present study is its size (seven times bigger than the previous largest study13) which was made possible by linking two national registers using national personal identification numbers. Both exposure (IVF treatment) and outcome (treatment for asthma) were thus based on information collected independently of this research. This also allowed for adjustment for various putative confounders and examination of mediators, based on prospectively collected register data. As always in epidemiological studies, there may be unidentified confounders which can explain part or even all of the association. When information on a specific variable is missing in a large proportion of cases (such as information on BMI), adjustment will be incomplete. However, exclusion of cases lacking information on smoking and/or BMI usually had little effect on the OR estimates.
Various methods can be used to identify asthma in children including detailed parental questionnaires (eg, the international ISAAC study21) or hospital discharge diagnoses.1–12 We used prescriptions for antiasthmatic drugs as an indicator of asthma.9–11 However, as such drugs can be used for indications other than asthma, resulting in dilution with non-asthma cases, we decided to use five or more prescription events as the criterion for inclusion in the study. Nevertheless, some non-asthmatic cases may remain which will bias the ORs towards 1.0. Selection of the cut-off value is a balance between getting a high proportion of true asthma and sufficient numbers for acceptable statistical power.
A weakness in the study is that there was no information on post-neonatal death or emigration of children. Therefore, some such children will be included in the denominators and this will reduce risk estimates slightly. There is no major difference in risk of post-neonatal death between children born after IVF and other children, but immigrant parents may be more inclined to emigrate again and may also have used IVF less. Exclusion of non-Swedish born women did not markedly change the OR.
There was no difference in the OR for asthma after IVF between boys and girls. The same was true for caesarean section versus vaginal delivery: the ORs for the two groups were similar even though caesarean section is a known risk factor for childhood asthma.7 There is a slight and not quite significant difference in ORs between children born as twins and as singletons. However, among singletons, there is a higher risk of asthma in preterm IVF infants than in term IVF infants. If the effect of IVF on asthma risk was due to an increased rate of preterm birth after IVF, the OR for preterm children should instead be lower than for term births. This phenomenon was seen for six neonatal conditions: SGA in singletons, low Apgar score, respiratory problems, use of mechanical ventilation, use of CPAP and neonatal sepsis. These factors thus behaved as mediators and when stratified for them, the effect of IVF declined and lost statistical significance. However, these factors explain only a small amount of the total effect because of the relatively small proportion of infants with these conditions.
When analysis was restricted to ‘normal’ outcomes (singleton AGA term infants delivered vaginally, with normal Apgar score and no neonatal diagnosis of respiratory problems, no mechanical ventilation or CPAP, and no sepsis or jaundice), a lower but significantly increased risk remained, which was a little more than half of the initial risk. This indicates that the remaining risk may not be due to neonatal effects seen after IVF but may be more directly linked to IVF per se or to the subfertility underlying IVF.
An effect of parental subfertility (estimated as years of unwanted childlessness) on the risk for asthma was demonstrated. Adjustment for the length of unwanted childlessness when known showed that the effect of IVF on asthma nearly disappeared, leaving a non-significant OR of 1.02. This indicates that the main effect is related to the subfertility of the couple, partly via neonatal pathology. It does not seem to be directly related to the IVF methodology.
One can only speculate on possible links between subfertility and asthma in children. There is a genetic component in the risk of asthma and if parental asthma was associated with an increased risk of subfertility, such a link could at least partly explain the association. In a previous study we showed that women who had IVF had used a surfeit of antiasthmatics (OR 1.39, 95% CI 1.22 to 1.58).22 Women using antiasthmatics have an excess of unwanted childlessness of 2 years or more.23 A link has been suggested between the use of asthma medication and the risk of ovulatory infertility.24 On the other hand, no effect on final fertility rates in women with asthma was found.25 When we analysed women who had reported own use of antiasthmatic drugs in early pregnancy, the association between IVF and asthma in the offspring was reduced, but the number was low and the CI large. All mothers with a genetic load for asthma were certainly not identified and the contribution of paternal genetics was not considered. A genetic link between asthma, subfertility and therefore IVF, and childhood asthma may exist but can probably explain only part of the association between IVF and asthma.
In conclusion, a link between IVF and asthma in children was found which seems only partly due to increased neonatal morbidity. The association is mainly caused by underlying fertility problems. The link between subfertility and asthma risk has no obvious explanation so further studies are needed.
Contributors BK carried out the data analysis and drafted the manuscript. KGN and OF took part in the planning of the study, discussed the results, and read and commented on the manuscript draft. POO supervised data collection, took part in the planning of the study, discussed the results, and read and commented on the manuscript draft.
Funding This study was supported by a grant from the Evy and Gunnar Sandberg Foundation, Lund, Sweden to BK.
Competing interests None.
Ethics approval The study was performed within the responsibilities of the National Board of Health and Welfare and therefore no ethics approval from external ethics committees was needed.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.