The risk of congenital malformations, perinatal mortality and neonatal hospitalisation among pregnant women with asthma: a systematic review and meta-analysis

Authors

Errata

This article is corrected by:

  1. Errata: Corrigendum Volume 121, Issue 5, 652, Article first published online: 18 March 2014

Correspondence: Dr V Murphy, Centre for Asthma and Respiratory Diseases, Level 2 West Wing Hunter Medical Research Institute, c/- University of Newcastle, University Drive, Callaghan, NSW 2308, Australia. Email vanessa.murphy@newcastle.edu.au

Abstract

Background

There is conflicting literature on the effect of maternal asthma on congenital malformations and neonatal outcomes.

Objectives

This review and meta-analysis sought to determine if maternal asthma is associated with an increased risk of adverse neonatal outcomes.

Search strategy

We searched electronic databases for: (asthma or wheeze) and (pregnan* or perinat* or obstet*).

Selection criteria

Cohort studies published between 1975 and March 2012 reporting at least one perinatal outcome of interest (congenital malformations, neonatal complications, perinatal mortality).

Data collection and analysis

In all, 21 studies met inclusion criteria in pregnant women with and without asthma. Further analysis was conducted on 16 studies where asthmatic women were stratified by exacerbation history, corticosteroid use, bronchodilator use or asthma severity.

Main results

Maternal asthma was associated with a significantly increased risk of congenital malformations (relative risk [RR] 1.11, 95% confidence interval [95% CI] 1.02–1.21, I2 = 59.5%), cleft lip with or without cleft palate (RR 1.30, 95% CI 1.01–1.68, I2 = 65.6%), neonatal death (RR 1.49, 95% CI 1.11–2.00, I2 = 0%), and neonatal hospitalisation (RR 1.50, 95% CI 1.03–2.20, I2 = 64.5%). There was no significant effect of asthma on major malformations (RR 1.31, 95% CI 0.57–3.02, I2 = 70.9%) or stillbirth (RR 1.06, 95% CI 0.9–1.25, I2 = 35%). Exacerbations and use of bronchodilators and inhaled corticosteroids were not associated with congenital malformation risk.

Authors’ conclusions

Despite limitations related to the observational nature of the primary studies, this review demonstrates a small increased risk of neonatal complications among pregnant women with asthma. Further investigations into mechanisms and potential preventive interventions to improve infant outcomes are required.

Introduction

Asthma is a commonly reported medical condition affecting pregnant women, and the course of the disease can change during pregnancy. There has been conflicting literature as to the effects of maternal asthma on pregnancy and neonatal outcomes. In 2011, we published the first systematic review and meta-analysis of literature in this area examining the risk of adverse perinatal outcomes in women with asthma, focusing on preterm delivery, low birthweight and pre-eclampsia, some of the more commonly reported outcomes.[1] We found a moderately increased risk of each of these outcomes in pregnant women with asthma compared with pregnant women without asthma.[1] Numerous other important adverse outcomes, particularly those affecting the neonate, have been reported in the literature, but with no previous attempt to summarise the data via meta-analysis. Studies on the effect of asthma on congenital malformations vary in terms of sample size, study design (cohort versus case–control), exposures tested (asthma or asthma medication use), adjustment for confounders and multiple testing, and examination of specific malformations, while studies of perinatal mortality have generally been underpowered. There are also few studies that examine the potential mechanisms contributing to these outcomes, such as disease severity, exacerbations or asthma medication use. This updated systematic review and meta-analysis attempts to overcome some of these limitations in the existing literature, by examining the relative risks of adverse neonatal outcomes among women with asthma compared with women without asthma, and to examine the possible roles of medication use, asthma exacerbations during pregnancy and asthma severity on these outcomes.

Methods

Search strategy and study selection

A review protocol was established by the investigators before commencement. The study protocol has been described in detail elsewhere[1] and further details are given in the online supporting information (see Supplementary material, Appendix S1). We recently completed an update to our original systematic review. The initial search identified English-language studies published between 1975 (when inhaled corticosteroids were introduced) and March 2009 from Medline (n = 1642), Embase (n =1755), CINAHL (n = 417) and the Cochrane Central Register of Controlled Trials (n = 75), using the search terms ([asthma or wheeze] and [pregnan* or perinat* or obstet*]). Identified abstracts were independently assessed by two reviewers. The full text version of each potential article was obtained for assessment by two independent reviewers to establish whether it met the inclusion criteria. In the update, we used the same search terms to identify English language studies published between January 2009 and 11 March 2012 from Medline (n = 681), Embase (n = 624), CINAHL (n = 84) and the Cochrane Central Register of Controlled Trials (n = 14).

Inclusion criteria for analysis of the effect of maternal asthma on neonatal outcomes (Analysis A)

Included articles contained data from a group of pregnant women with asthma and a control group of pregnant women without asthma. Further inclusion criteria were (1) reporting at least one perinatal outcome of interest (congenital malformation, major congenital malformation, perinatal mortality, stillbirth, neonatal death, neonatal hospitalisation, transient tachypnoea of the newborn, respiratory distress syndrome or neonatal sepsis) and (2) cohort study design (prospective or retrospective). Maternal asthma could be defined as physician-diagnosed (whether confirmed or subject self-report), database-coded asthma diagnosis, or asthma fulfilling American Thoracic Society criteria.

In all, 139 articles were identified for possible inclusion in the review (Analysis A); 73 of these were excluded for the following reasons: no control group (n = 38), no clear asthma group (n = 3), women with asthma selected based on exacerbation (n = 3), cross-sectional survey (n = 3), case–control study (n = 3), perinatal outcomes not suitably reported (n = 10), women studied before 1975 (n = 2), paper retracted (n = 1), abstract only (n = 2), not the primary paper/first report of results (n = 2), review (n = 6). Of the remaining 66 publications, 21 studies (eight prospective, 13 retrospective) were identified for inclusion in the analysis of the effect of maternal asthma on neonatal outcomes (see Supplementary material, Table S1).

Inclusion criteria for analysis of the effect of maternal asthma subtypes on neonatal outcomes (Analysis B)

Included articles contained data from pregnant women with asthma which had been subdivided based on asthma medication use (e.g. inhaled corticosteroid use, no inhaled corticosteroid use), asthma exacerbations requiring medical intervention during pregnancy, or asthma severity (mild, moderate/severe; see Supplementary material, Table S2). Cohort studies or randomised controlled trials were included if they reported at least one outcome of interest.

Of the studies identified for Analysis A, and the 38 studies that were excluded from Analysis A because they contained no control group, there were 16 studies (seven prospective, eight retrospective, one randomised controlled trial) identified for inclusion in the analysis of the effect of maternal asthma subtypes on neonatal outcomes (Analysis B).

Data extraction

Data extraction was completed on a standardised form by one reviewer and checked by a second reviewer. Investigators discussed any discrepancies to reach consensus. Studies were considered to have provided active asthma management when the study investigators were involved in the management and treatment of women with asthma and this was described (Table S1).

We used the Newcastle–Ottawa Scale to assess study quality.[2] The Newcastle–Ottawa Scale is a validated tool for assessing the quality of non-randomised studies including cohort and case–control studies and has a maximum score of 9. Quality was assessed and scored by two reviewers and all studies were considered to be of adequate quality for inclusion in the meta-analysis (minimum score 5, mean of all scores 7.8, Table S1).

Meta-analysis

The meta-analyses conformed to standard methodological guidelines for meta-analysis of observational studies.[3] The relative risk of the perinatal outcome was examined in women with asthma compared with women without asthma (analysis A) or in subgroups of women with asthma stratified by severity (mild versus moderate-severe), exacerbations during pregnancy (expressed as a yes/no variable) and exposure to oral corticosteroids during pregnancy (expressed as a yes/no variable, analysis B) using Review Manager software (Review Manager (RevMan) [computer program] Version 4.3.2. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011, available online http://ims.cochrane.org/revman/download/revman-4). Sensitivity analyses were also performed in Analysis A by study design (prospective versus retrospective cohort) and asthma management (active versus no active management). For dichotomous outcomes the relative risk with 95% confidence interval was calculated using a random effects model, unless three or fewer studies were combined, in which case the fixed effects model was used. The difference between relative risks for the active management and no active management subgroups was determined using the method of Altman and Bland[4] and expressed as a relative risk ratio (RRR) with 95% confidence intervals. Data from studies that provided adjusted odds ratios were pooled using the generic inverse variance method. Heterogeneity was examined using the chi-square test (P < 0.1 considered significant heterogeneity), the I-square percentage (I2 ≥ 60% considered significant heterogeneity) and subgroup analyses.

Other analyses

When outcomes were reported in at least ten studies, Funnel plots and the Egger test were used to investigate publication bias (Stata 7, Stata Corporation, College Station, TX, USA, www.stata.com). Power calculations were conducted using the PS Power and Sample size program (version 2.1.30).[5] In reporting results of the meta-analysis we followed recommendations from the PRISMA consensus statement.[6]

Further information is available in Appendix S1.

Results

Congenital malformations

Data on congenital malformations (all/any) were included in 16 publications,[7-22] which represented 14 overall studies. Data from Minerbi-Codish et al.[14] were incomplete and therefore not included in the meta-analysis. Schatz et al.[18] reported major and minor congenital malformations separately, which were added together to give total malformations. Clark et al.[8] reported malformations as those which were the reason for neonatal hospital admissions, so these data were excluded because they did not capture all malformations.[8] Data sources and definitions/confirmation of congenital malformations and major congenital malformations varied (see Supplementary material, Table S3).

There were four prospective cohort studies (all had active asthma management) included in the meta-analysis and eight retrospective cohort studies (none had active management).[7, 9, 10, 12, 13, 20-22] There was a significantly increased risk of congenital malformations in women with asthma compared with control women without asthma (Figure 1A; relative risk [RR] 1.11, 95% confidence interval [95% CI] 1.02–1.21, I2 = 59.5%, P < 0.1). Sensitivity analysis showed the effect was only significant in the subcategory of retrospective cohort studies (without active management, RR 1.12, 95% CI 1.03–1.22), whereas in the prospective cohort studies (with active asthma management) there was no significant effect (RR 1.40, 95% CI 0.54–3.59). However, the prospective cohort studies only had 10% power to detect an RR of 1.11, as observed in the retrospective studies. There was no significant publication bias (Egger test, P = 0.977).

Figure 1.

(A) Meta-analysis of cohort studies for congenital malformations. ‘Increased risk’ indicates that the outcome was more likely for women with asthma. (B) Meta-analysis of cohort studies that adjusted for confounders and presented data as odds ratios (OR) for congenital malformations. ‘Increased odds’ indicates that the outcome was more likely to women with asthma.

A meta-analysis of the adjusted odds ratio from four studies (retrospective, no active management)[9, 12, 13, 22] found an increased odds of congenital malformations among women with asthma (Figure 1B; odds ratio 1.18, 95% CI 1.03–1.35).

The relative risk of congenital malformations among subgroups of women with asthma (Table 2) indicated no increased risk of malformations with bronchodilator use,[23-25] inhaled corticosteroid use[9, 26, 27] or exacerbations[16, 17, 25] (Table 1).

Table 1. Congenital malformations and related outcomes in pregnant women with asthma
Outcome Comparison RR [95% CI] Number of studies Number of women (Group 1) Number of women (Group 2) Heterogeneity
  1. ICS, inhaled corticosteroids; I2, the percentage of the variability in effect estimates that is due to heterogeneity.

Congenital malformationsBronchodilator use versus No bronchodilator useRR 1.04 [0.94–1.16]37518 (Bronchodilator)6316 (No bronchodilator)I2 = 0% (P = 0.82)
ICS use versus No ICS useRR 0.96 [0.89–1.04]317 220 (ICS)21 122 (no ICS)I2 = 0% (P = 0.58)
Exacerbation versus No exacerbationRR 1.18 [0.94–1.47]3717 (Exacerbation)13 093 (No exacerbation)I2 = 0% (P = 0.79)
Major malformationsAsthma versus ControlRR 1.31 [0.57–3.02]43239 (Asthma)1850 (Control)I2 = 70.9% (P = 0.02)
Bronchodilator use versus No bronchodilator useRR 1.00 [0.80–1.26]24809 (Bronchodilator)1875 (No Bronchodilator)I2 = 0% (P = 0.93)
Exacerbation versus No exacerbationRR 1.26 [0.95–1.67]2681 (Exacerbation)12 517 (No exacerbation)I2 = 0 (P = 0.71)
Cleft lip with or without cleft palateAsthma versus ControlRR 1.30 [1.01–1.68]238 029 (Asthma)902 233 (Control)I2 = 65.6% (P = 0.09)

Four prospective studies reported major congenital malformations.[18, 28-30] The meta-analysis indicated no significant risk of maternal asthma (Table 1; RR 1.31, 95% CI 0.57–3.02); however, there was significant heterogeneity between studies (I2 = 70.9%, P < 0.1). The study from Bakhireva et al.[28] reported an unusually low prevalence of major malformations in the control group, which contributed to the vastly different result from this study.

There was no significantly increased risk of major malformations with bronchodilator use when data from two studies[31, 32] were combined (Table 1; RR 1.00, 95% CI 0.80–1.26). Data on major congenital malformations among women with and without asthma exacerbations were extracted from four publications.[25, 32-34] However, there was significant similarity in the cohorts from the Canadian studies[25, 32, 34] and data were included in the meta-analysis from Eltonsy et al.[25] only, as this covered the longest time period (1990–2002) and the largest number of pregnancies in women with asthma (13 117 compared with 4344 and 4561 in Blais et al. 2008[34] and 2007,[32] respectively). There was no significantly increased risk of major malformations among women with asthma exacerbations during pregnancy compared with women with no exacerbations during pregnancy (Table 1; RR 1.26, 95% CI 0.95–1.67), with no heterogeneity between studies (I2 = 0%, P = 0.71).

Specific congenital malformations were described in detail by Kallen et al.[9] and Blais et al.[22] Kallen et al.[9] found a significant increase in cardiac defects and anal atresia among infants of women with asthma compared with population estimates. Blais et al.[22] found that women with asthma were at increased risk of having infants with nervous system, respiratory system and digestive system defects. We combined data on cleft lip and/or cleft palate, which was described in both papers, demonstrating that infants from women with asthma had a significantly increased risk of this malformation compared with infants from women without asthma (Table 1; RR 1.30, 95% CI 1.01–1.68, I2 = 65.6%).

Stillbirth

Data on stillbirth (fetal death in utero either >20 weeks, or >500 g or >28 weeks[35]) were included in eight studies.[13, 15, 18, 21, 24, 35-37] Two publications from Stenius-Aarniala et al.[15, 16] contained the same population of women, but had inconsistencies in the description of perinatal deaths. The first publication contained a detailed description of all perinatal deaths (including separate data on stillbirths and neonatal deaths); therefore data from the original paper were included in this analysis.[15] Overall, there was no significantly increased risk of stillbirth in infants of asthmatic mothers compared with control mothers (Figure 2A; RR 1.06, 95% CI 0.90–1.25), and there was no heterogeneity in the studies (I2 = 35%, P > 0.1). There was also no increased risk in the subanalyses by study design (Figure 2A), asthma management (data not shown), or adjusted odds ratio (Table 2).

Table 2. Neonatal outcomes in pregnant women with asthma
Outcome Comparison Odds ratio or RR [95% CI] Number of studies Number of women (Group 1) Number of women (Group 2) Heterogeneity
  1. ICS, inhaled corticosteroids; I2, the percentage of the variability in effect estimates that is due to heterogeneity.

StillbirthAsthma versus Control (adjusted ORs)OR 1.07 [0.92–1.26]246257 (Asthma)278122 (Control)I2 = 0% (P = 0.51)
Exacerbations versus No exacerbationsRR 2.98 [0.48–18.46]275 (Exacerbations)146 (No Exacerbations)I2 = 22.7% (P = 0.26)
ICS use versus No ICS useRR 0.77 [0.50–1.18]28455 (ICS)15 443 (No ICS)I2 = 0% (P = 0.88)
Perinatal mortalityExacerbations versus No exacerbationsRR 3.22 [0.79–13.18]388 (Exacerbations)686 (No Exacerbations)I2 = 31.7% (P = 0.23)
Moderate/Severe Asthma versus Mild AsthmaRR 1.32 [0.97–1.78]311396 (Moderate/Severe Asthma)14 909 (Mild Asthma)I2 = 0% (P = 1.00)
Transient tachypnoea of the newbornAsthma versus ControlRR 1.24 [0.85–1.80]22033 (Asthma)1175 (Control)I2 = 80.9% (P = 0.02)
Respiratory distress syndromeAsthma versus ControlRR 1.57 [0.88–2.81]22225 (Asthma)1367 (Control)I2 = 0% (P = 0.77)
Neonatal sepsisAsthma versus ControlRR 2.27 [1.12–4.58]22457 (Asthma)1598 (Control)I2 = 0% (P = 0.64)
Figure 2.

(A) Meta-analysis of cohort studies for stillbirth. (B) Meta-analysis of cohort studies for neonatal death. (C) Meta-analysis of cohort studies for perinatal mortality. ‘Increased risk’ indicates that the outcome was more likely in women with asthma.

There were no significant associations between exacerbations[33, 38] or inhaled corticosteroid use[39, 40] and the risk of stillbirth among women with asthma (Table 2).

Neonatal death

Data on neonatal death (death up to 1 month of age) were included in six studies, two prospective cohort studies[15, 18] and four retrospective cohort studies.[8, 24, 37, 41] Overall, there was a significantly increased risk of neonatal death in infants of asthmatic mothers compared with control mothers (Figure 2B; RR 1.49, 95% CI 1.11–2.00, no heterogeneity, I2 = 0%, P = 0.67). Individually, only the retrospective cohort subcategory showed a statistically increased risk which was of similar magnitude to the overall effect size (RR 1.48, 95% CI 1.10–1.99).

In three studies, women had active management of their asthma by the study investigators or local hospital,[15, 18, 24] whereas in three studies no active management of women was given.[8, 37, 41] In the subcategory with active management, there was no significant effect of maternal asthma on neonatal death, but the confidence interval was very wide (RR 2.28, 95% CI 0.46–11.40). These studies were underpowered (6%) to detect the difference observed in the retrospective cohort studies. In the subcategory of studies with no active management of women, there was a significant effect of maternal asthma on neonatal death (RR 1.47, 95% CI 1.09–1.98). The difference between the relative risks of the no active management and active management subgroups was not significant (RRR 1.55, 95% CI 0.30–7.94, P = 0.598).

Perinatal mortality

Data on perinatal mortality (a combination of stillbirths and neonatal deaths) were included in nine studies from ten publications. There were overlapping data presented in two publications from Schatz et al.[18, 19] Only the more recent paper with larger numbers and more definitively matched women was included in the analysis.[18]

There were six prospective cohort studies[11, 15, 17, 18, 29, 42] and three retrospective cohort studies.[10, 24, 37] Overall, there was a significantly increased risk of perinatal mortality in infants of asthmatic mothers compared with control mothers (Figure 2C, RR 1.25, 95% CI 1.05–1.50, I2 = 0%), with the overall effect size being intermediate between that observed for stillbirth and neonatal death. Individually, neither the prospective or retrospective cohort subcategories showed a statistically increased risk of perinatal mortality (Figure 2C). The prospective cohort studies were underpowered to detect an RR of 1.25 (11%).

There were seven studies where women had active management of their asthma[11, 15, 17, 18, 24, 29, 42] and two studies where no active management was given.[10, 37] There was no significant heterogeneity between studies, and in neither subcategory was the risk of perinatal mortality significantly increased in women with asthma compared with control women, although the effect sizes were similar to the overall result (active management: RR 1.17, 95% CI 0.61–2.27]; no active management: RR 1.23, 95% CI 0.95–1.58). The active management studies were underpowered to detect an RR of 1.23 (11%).

Women with exacerbations were not at significantly increased risk of perinatal mortality compared with women without exacerbations[16, 17, 33] (Table 2). Women with moderate/severe asthma were not at significantly increased risk of perinatal mortality compared with women with mild asthma[29, 35, 42] (Table 2).

Neonatal hospitalisation

Data on neonatal hospitalisation (treatment in or admission to the neonatal intensive care unit, or neonatal medical/surgical unit) were included in six publications. There were four prospective cohort studies[16, 29, 42, 43] and two retrospective cohort studies.[8, 20] Overall, there was a significantly increased risk of neonatal hospitalisation among infants of asthmatic mothers (Figure 3; RR 1.50, 95% CI 1.03–2.20) compared with infants of mothers without asthma. There was significant heterogeneity between studies (I2 = 64.5%, P = 0.02), and the risk was significant in the retrospective cohort subcategory only (RR 2.36, 95% CI 1.10–5.04). Three studies had active management,[15, 29, 42] whereas three did not.[8, 20, 43] There was a significant risk of neonatal hospitalisation among studies with no active management (RR 1.97, 95% CI 1.07–3.65), but not among studies with active management (RR 1.13, 95% CI 0.95–1.36). The difference between the relative risks of the no active management and active management subgroups was not significant (RRR 0.57, 95% CI 0.30–1.09, P = 0.088).

Figure 3.

Meta-analysis of cohort studies for neonatal hospitalisation. ‘Increased risk’ indicates that the outcome was more likely in women with asthma.

Two of the studies excluded both asthmatic and control women with earlier deliveries (one excluded women who were <36 weeks of gestation at recruitment,[43] while another excluded all preterm deliveries[8]), which may result in fewer neonatal complications than other studies. When sensitivity analysis was conducted without these two studies, the risk of neonatal hospitalisation among women with asthma was of a similar effect size (RR 1.58, 95% CI 0.84–2.99), but was not significant.

Transient tachypnoea of the newborn

Transient tachypnoea of the newborn (TTN) was evaluated in three studies.[8, 29, 44] The TTN data were reported twice in publications from Schatz et al.,[19, 44] and the data from the publication specifically on TTN were used.[44] Two prospective cohort studies (both with active management) were included in the meta-analysis and there was a significantly increased risk of TTN among infants of asthmatic mothers (Table 2; RR 1.54, 95% CI 1.09–2.18) compared with infants of mothers without asthma and significant heterogeneity between studies (I2 = 84.3%, P = 0.002). The retrospective study from Clark et al.[8] included more severe TTN requiring hospitalisation, and was the only study to individually report a significantly increased risk of TTN in neonates of mothers with asthma compared with mothers without asthma (RR 6.32, 95% CI 1.88–21.28). We included this study in the meta-analysis because it was not clear whether the TTN diagnosis required hospitalisation in the other studies, and because the criteria for the diagnosis of TTN in asthma and control women would be the same in each study, we expected the relative relationships between pregnancies in asthmatic versus control women to hold across studies.

Respiratory distress syndrome

Data on infant respiratory distress syndrome (also called hyaline membrane disease) were included in two studies, which were both prospective cohort studies with active asthma management.[18, 29] The meta-analysis did not identify an increased risk of respiratory distress syndrome among infants of asthmatic mothers (Table 2; RR 1.57, 95% CI 0.88–2.81).

Neonatal sepsis

Neonatal sepsis was defined as a discharge diagnosis of neonatal sepsis,[29] or sepsis resulting in admission to the neonatal medical or surgical unit,[8] and there was a significantly increased risk of sepsis among infants of asthmatic mothers (Table 2; RR 2.27, 95% CI 1.12–4.58) compared with infants of mothers without asthma.

Discussion

Main findings

This meta-analysis indicates that infants of pregnant women with asthma are 11% more likely to manifest congenital malformations compared with infants of non-asthmatic women. Significance was reached in the retrospective studies, and among four studies that controlled for important confounders and reported adjusted odds ratio data. It is likely that some studies reported major malformations only (those that were recognised at birth); however, a separate analysis of studies reporting major malformations showed no increased risk with maternal asthma. One possibility is that the risk of malformations is driven by minor malformations, which would be reassuring, because these are less clinically significant. However, the analysis of major malformations may be inadequately powered, as there were fewer studies and women reporting this outcome. More data are needed to clarify these risks.

Pregnant women with asthma were at 30% increased risk of cleft lip and/or palate compared with pregnant women without asthma. Previous case–control studies demonstrated an association between first-trimester oral corticosteroid exposure and the risk of oral clefts.[45, 46] Our finding is consistent with the possibility that only those women with asthma who used oral steroids in the critical window for lip and palate closure are at risk for these specific defects.

Several cohort studies have specifically examined the effects of corticosteroid treatment on malformations. Two studies, exclusively in women using budesonide, found no increased risk of malformations.[26, 47] Kallen et al.[9] evaluated the risk of malformations in women using oral or inhaled corticosteroids in early pregnancy, but found no significant associations. Another large study found a significantly reduced risk of malformations among first-trimester users of moderate-dose inhaled corticosteroids compared with nonusers.[32] There was no increased risk with the use of high-dose inhaled corticosteroids; however, women who experienced exacerbations requiring medical intervention in the first trimester were at significantly increased risk of congenital malformations compared with women without exacerbations.[34] Our meta-analysis combined these data[34] with those from two other studies and no increased risk with exacerbation was observed.

There was a 25% increased risk of perinatal mortality in women with asthma, which was probably driven by the increase in neonatal deaths. However, this risk was small compared with risks for in vitro fertilisation pregnancies (odds ratio 2.2, 95% CI 1.6–3.0)[48] or pregnancies in women with diabetes (RR 3.01, 95% CI 1.55–5.84).[49] The increase in neonatal mortality may be a result of the increased risk of preterm deliveries,[1] because the increased risk of perinatal mortality reported by one study was eliminated when adjusted for prematurity.[37]

Subgroup analyses did not confirm a relationship between asthma management or severity and adverse neonatal outcomes. However, the primary studies were limited in their ability to define asthma control and successful management in individual women. A recent RCT demonstrated that neonatal outcomes could be modified when asthma therapy was adjusted according to the degree of airway inflammation, rather than symptoms and lung function.[50] This management strategy halved the rate of exacerbations requiring medical intervention during pregnancy, suggesting the potential to improve maternal and neonatal health through active management. Large prospective studies with comprehensive assessment of individual patient control and treatment variables as well as relevant confounders may further support the hypothesis that optimal asthma control during pregnancy mitigates the increased perinatal risks demonstrated in the pregnancies of women with asthma.

Strengths and weaknesses

In this systematic review and meta-analysis, we have identified numerous risks, which, although small in size, are consistent and significant. The major strength of this approach is the large number of pregnancies studied and the large number of rare events reported. There are limitations arising from the observational nature of the included studies, such as risk of bias (including reporting bias), the influence of confounders and the presence of significant heterogeneity in some analyses. For congenital malformations, we were able to combine data that adjusted for confounders and found a similar result to that obtained when combining unadjusted data, suggesting a minimal effect of confounders. Significant heterogeneity was present in several of the analyses. However, analyses for stillbirth, neonatal death and perinatal mortality and asthma subgroup analyses for congenital malformations were not affected by heterogeneity. It is possible that the very large sample sizes in some of the retrospective studies result in heterogeneity that is overstated compared with traditional meta-analyses.

Interpretation

It is plausible that maternal asthma may contribute to adverse neonatal outcomes because women with asthma are at increased risk of low birthweight infants,[1] preterm delivery,[1] gestational diabetes and placental problems,[51] which all increase the risk of neonatal complications and death. Using data from different populations and settings, with various study designs, results indicate that neonates of women with asthma are at a small risk of serious complications including intensive care hospitalisation. Although we did not observe associations between asthma severity or exacerbations and neonatal outcomes, the number of available studies was low. There was a small but inconsistent effect of asthma on congenital malformations and further research is needed to clarify the risk of this outcome, and whether the risks found here could be driven by minor malformations.

Conclusions

Asthma is a common chronic disease among pregnant women, and the risks of adverse outcomes for both mother and baby during the perinatal period make this a significant health issue. Pregnant women with asthma are at increased risk of any congenital malformation and specifically of cleft lip and/or cleft palate. The currently available evidence does not indicate any relationship between malformations and bronchodilator use, inhaled corticosteroid use or exacerbations. Neonates were also at increased risk of intensive care hospitalisation and death, although further studies are needed to elucidate the mechanisms involved. Results should be interpreted with caution because of the small absolute risks for women with this prevalent condition. Pending additional data, early pregnancy care to achieve good asthma control and avoid exacerbations seems warranted, which should improve maternal and neonatal health.

Disclosure of interests

VM, WG and HP have no conflicts of interest to report. JN has links with Genentech. PG has links with Glaxo Smith Kline and Boehringer Ingelheim; CC has links with Amgen, Abbott, Apotex, Sandoz, Barr, Heritage, Kali, Bristol Myers Squibb, Sanofi Aventis and Sanofi Pasteur; and MS has links with Aerocrine, Glaxo Smith Kline, Genentech, Merck and Amgen.

Contribution to authorship

VM contibuted to the conception of the study and to study search and identification, inclusion/exclusion, data extraction, quality assessment, interpretation and writing. WG contributed to study search and identification, inclusion/exclusion, data extraction, quality assessment and manuscript editing. JN contributed to study search and identification, inclusion/exclusion, data extraction, quality assessment, interpretation and manuscript editing. HP contributed to study search and identification, inclusion/exclusion, data extraction, quality assessment, analysis and manuscript editing. PG contributed to study design and conception, interpretation, writing and manuscript editing. CC contributed to interpretation and manuscript editing and MS contributed to study design and conception, interpretation, writing and manuscript editing.

Details of ethics approval

Not applicable.

Funding

Funding was received from the Kaiser Permanente Southern California Regional Research Committee. The funding body played no role in the conduct or reporting of this research.

Acknowledgements

The authors thank Prof John Attia for his advice on the meta-analyses and Dr Patrick McElduff for statistical advice.

Ancillary