Objective
Symptoms of sleep-disordered breathing (SDB) are increased in pregnancy compared to the nongravid state. Maternal SDB may be associated with adverse pregnancy outcomes, but this is still under investigation. We performed a systematic literature review, and where feasible, a metaanalysis, to evaluate whether women with SDB in pregnancy have a higher risk of specific adverse pregnancy outcomes compared with women without SDB.
Study Design
Original studies published until June 2012 evaluating the association between gestational hypertension/preeclampsia, gestational diabetes, low birthweight infants, and maternal SDB, defined either by symptoms or the reference standard polysomnography, were identified from PubMed, EMBASE, and Web of Science. Data were extracted on study design and outcome estimates. When appropriate, effect estimates from each study were pooled using a random-effects model.
Results
Of the 4386 studies identified, 31 met the defined criteria. Twenty-one studies, all observational in design, reported dichotomous outcomes; 9 of these adjusted for potential confounders. Maternal SDB was significantly associated with gestational hypertension/preeclampsia (pooled adjusted odds ratio [aOR], 2.34; 95% confidence interval [CI], 1.60–3.09; 5 studies), and gestational diabetes (pooled aOR, 1.86; 95% CI, 1.30–2.42; 5 studies).
Conclusion
Based on published observational studies to date, maternal SDB is associated with an increased risk of gestational hypertension and gestational diabetes after adjusting for potential confounders. However, large-scale, prospective cohort, and interventional studies are needed to further elucidate the relationship between maternal SDB and adverse pregnancy outcomes.
See related editorial, page 3
See Journal Club, page 87
Sleep-disordered breathing (SDB) refers to a group of disorders characterized by breathing pauses during sleep and is associated with microarousals, oxygen desaturations, and/or hemodynamic changes. In the nonpregnant population, SDB has been associated with several adverse cardiometabolic outcomes, including hypertension, cardiovascular disease, stroke, and altered glucose and lipid metabolism.
In pregnancy, symptoms of SDB are common and worsen with gestational age, with 14-35% of pregnant women reporting habitual snoring in the third trimester. The potential mechanisms contributing to SDB in pregnancy include gestational weight gain, edema, and hormonal influences. However, the true prevalence of SDB in pregnancy is unknown because the majority of studies have relied on symptom-based diagnosis rather than objective polysomnographic (PSG) recordings.
SDB is associated with sleep disruption and intermittent hypoxia-reoxygenation, which may lead to increased sympathetic activity, oxidative stress, and inflammation. During pregnancy, these disturbances could contribute to maternal cardiometabolic derangements such as gestational hypertension and diabetes and to impairment of placental function, resulting in poor fetal outcomes such as intrauterine growth restriction. Establishing a causal link between maternal SDB and adverse pregnancy outcomes could therefore have important implications for prenatal care. Several recent studies have investigated whether symptom-based or PSG-based definitions of SDB are associated with maternofetal complications. However, because the results have been conflicting, there is a clear need for a systematic review and metaanalysis of the available literature in summarizing the current evidence.
Objective
Our objective was to conduct a systematic review and, where feasible, metaanalysis to evaluate whether pregnant women with SDB have a higher risk of adverse pregnancy outcomes, which we defined as gestational hypertension/preeclampsia, gestational diabetes, or the delivery of low-birthweight infants, compared with pregnant women without SDB.
Methods for Review
We used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis statement as the template for reporting the review.
Search strategy and study selection
We searched for citations on the association between SDB and pregnancy in the following 3 electronic databases published up until June 2012: PubMed, EMBASE, and Web of Science. All citations were then combined and the duplicates were excluded.
Search terms in PubMed included the following MeSH terms for SDB exposure: snoring; sleep apnea syndromes; and sleep apnea, obstructive. For adverse pregnancy outcomes, the following MeSH terms were used: pregnancy; hypertension, pregnancy induced; pregnancy complications, cardiovascular; pregnancy outcome; and diabetes, gestational. Text terms used were sleep, sleep apnea, and pregnancy. Terms were combined using the Boolean operator, or, within each category, and the Boolean operator, and, when combining between exposure and outcome.
In EMBASE, the EMTREE search terms for SDB exposure were sleep, sleep apnea syndrome, or snoring. For adverse pregnancy outcomes, the terms used were as follows: pregnancy, pregnancy complications, or text word pregnancy. The search strategy in Web of Science was sleep apnea or snoring and pregnant*. The search was not restricted by language. Studies were excluded if they were conference abstracts, reviews, or case reports. The bibliographies of 3 review papers were manually searched for potentially relevant citations that were not detected by the electronic search.
Studies were included if they assessed for an association between SDB in pregnancy and adverse pregnancy outcomes. We did not narrow our initial search to include the specific adverse pregnancy outcomes of interest (eg, gestational hypertension, gestational diabetes, and low-birthweight infants) because we postulated that some citations may not include these actual outcomes in the title but rather only in the abstract or full text. Instead, we eliminated articles that did not include at least 1 of these 3 outcomes when reviewing citations at the full text article stage of the review.
Studies were included only if there was a comparison group so that we could calculate odds ratios and mean differences in the outcomes between groups. We included all citations in the qualitative review that described an association between SDB and adverse pregnancy outcomes, including those that exclusively reported continuous data.
Data extraction and consensus
Two reviewers (S.P. and L.M.P.) independently screened the titles for potential relevance. Only titles that were mutually agreed upon as being not relevant were excluded. The 2 reviewers then screened abstracts and full-text articles for inclusion/exclusion criteria. At these stages, if there was disagreement between the 2 reviewers, a final decision was made based on consensus.
Once the final articles were identified after the full-text review, the 2 investigators independently reviewed the papers and extracted information on the following information: year and country of the study, study population, study design, number of exposed/unexposed or cases/controls, definitions, and criteria used for SDB and each of the adverse pregnancy outcomes, timing of exposure measurement, effect estimates or continuous data, and where applicable, adjusted variables. Study quality was assessed using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.
Definitions of SDB and adverse pregnancy outcomes
For the purpose of this review, SDB was defined according to the reference diagnostic standard, PSG demonstrating obstructive sleep apnea-hypopnea, or presumed/probable obstructive sleep apnea (OSA) based on reported symptoms of snoring, nocturnal choking/gasping, witnessed apneas, or simplified sleep recordings showing evidence of upper airway obstruction (inspiratory flow limitation) and/or repetitive oxygen desaturations.
In studies that presented more than 1 category of SDB (eg, none, mild, moderate, or severe), if the categories were mutually exclusive, we combined numbers to provide an effect estimate for any SDB compared with none. If the categories were not mutually exclusive, then we chose the more severe category of SDB and compared it with no SDB.
For the outcome of gestational hypertension/preeclampsia, if a study presented numbers for gestational hypertension and preeclampsia separately, we combined them as long as they were mutually exclusive. Gestational diabetes was defined by the presence or absence of the diagnosis by clinical criteria that were specified by the authors of each study. Finally, for infants with low birthweight, we chose the definition as being less than the 10th percentile to ensure consistency between studies.
Data analysis
We classified studies based on whether continuous data or categorical data were reported and also by adverse pregnancy outcome. Within each group of studies classified by adverse pregnancy outcome, heterogeneity of the effect estimates was assessed by calculating the I 2 statistic for both the crude effect estimates and the adjusted effect estimates. Crude and/or adjusted effect estimates were pooled only for studies that did not show significant heterogeneity, defined as I 2 of 75% or greater.
A metaanalysis was performed using a Der Simonian and Laird random-effects model to account for the variability across studies and to derive conservative assessments of the uncertainty in the estimates. The Mantel-Haenszel method was used for pooling crude effect estimates across studies. Studies were excluded if no subjects with the outcome of interest were observed in either exposure group. For the adjusted odds ratios, pooling was performed using the inverse variance method. We pooled adjusted odds ratios, even when these were adjusted for different variables, because adjusted effect estimates are likely to be more representative of the true measure of effect than the crude ratios. Continuous data, weighted by sample size, were pooled if there was no significant heterogeneity among the mean difference in effect size estimates across studies.
Methods for Review
We used the Preferred Reporting Items for Systematic Reviews and Meta-Analysis statement as the template for reporting the review.
Search strategy and study selection
We searched for citations on the association between SDB and pregnancy in the following 3 electronic databases published up until June 2012: PubMed, EMBASE, and Web of Science. All citations were then combined and the duplicates were excluded.
Search terms in PubMed included the following MeSH terms for SDB exposure: snoring; sleep apnea syndromes; and sleep apnea, obstructive. For adverse pregnancy outcomes, the following MeSH terms were used: pregnancy; hypertension, pregnancy induced; pregnancy complications, cardiovascular; pregnancy outcome; and diabetes, gestational. Text terms used were sleep, sleep apnea, and pregnancy. Terms were combined using the Boolean operator, or, within each category, and the Boolean operator, and, when combining between exposure and outcome.
In EMBASE, the EMTREE search terms for SDB exposure were sleep, sleep apnea syndrome, or snoring. For adverse pregnancy outcomes, the terms used were as follows: pregnancy, pregnancy complications, or text word pregnancy. The search strategy in Web of Science was sleep apnea or snoring and pregnant*. The search was not restricted by language. Studies were excluded if they were conference abstracts, reviews, or case reports. The bibliographies of 3 review papers were manually searched for potentially relevant citations that were not detected by the electronic search.
Studies were included if they assessed for an association between SDB in pregnancy and adverse pregnancy outcomes. We did not narrow our initial search to include the specific adverse pregnancy outcomes of interest (eg, gestational hypertension, gestational diabetes, and low-birthweight infants) because we postulated that some citations may not include these actual outcomes in the title but rather only in the abstract or full text. Instead, we eliminated articles that did not include at least 1 of these 3 outcomes when reviewing citations at the full text article stage of the review.
Studies were included only if there was a comparison group so that we could calculate odds ratios and mean differences in the outcomes between groups. We included all citations in the qualitative review that described an association between SDB and adverse pregnancy outcomes, including those that exclusively reported continuous data.
Data extraction and consensus
Two reviewers (S.P. and L.M.P.) independently screened the titles for potential relevance. Only titles that were mutually agreed upon as being not relevant were excluded. The 2 reviewers then screened abstracts and full-text articles for inclusion/exclusion criteria. At these stages, if there was disagreement between the 2 reviewers, a final decision was made based on consensus.
Once the final articles were identified after the full-text review, the 2 investigators independently reviewed the papers and extracted information on the following information: year and country of the study, study population, study design, number of exposed/unexposed or cases/controls, definitions, and criteria used for SDB and each of the adverse pregnancy outcomes, timing of exposure measurement, effect estimates or continuous data, and where applicable, adjusted variables. Study quality was assessed using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.
Definitions of SDB and adverse pregnancy outcomes
For the purpose of this review, SDB was defined according to the reference diagnostic standard, PSG demonstrating obstructive sleep apnea-hypopnea, or presumed/probable obstructive sleep apnea (OSA) based on reported symptoms of snoring, nocturnal choking/gasping, witnessed apneas, or simplified sleep recordings showing evidence of upper airway obstruction (inspiratory flow limitation) and/or repetitive oxygen desaturations.
In studies that presented more than 1 category of SDB (eg, none, mild, moderate, or severe), if the categories were mutually exclusive, we combined numbers to provide an effect estimate for any SDB compared with none. If the categories were not mutually exclusive, then we chose the more severe category of SDB and compared it with no SDB.
For the outcome of gestational hypertension/preeclampsia, if a study presented numbers for gestational hypertension and preeclampsia separately, we combined them as long as they were mutually exclusive. Gestational diabetes was defined by the presence or absence of the diagnosis by clinical criteria that were specified by the authors of each study. Finally, for infants with low birthweight, we chose the definition as being less than the 10th percentile to ensure consistency between studies.
Data analysis
We classified studies based on whether continuous data or categorical data were reported and also by adverse pregnancy outcome. Within each group of studies classified by adverse pregnancy outcome, heterogeneity of the effect estimates was assessed by calculating the I 2 statistic for both the crude effect estimates and the adjusted effect estimates. Crude and/or adjusted effect estimates were pooled only for studies that did not show significant heterogeneity, defined as I 2 of 75% or greater.
A metaanalysis was performed using a Der Simonian and Laird random-effects model to account for the variability across studies and to derive conservative assessments of the uncertainty in the estimates. The Mantel-Haenszel method was used for pooling crude effect estimates across studies. Studies were excluded if no subjects with the outcome of interest were observed in either exposure group. For the adjusted odds ratios, pooling was performed using the inverse variance method. We pooled adjusted odds ratios, even when these were adjusted for different variables, because adjusted effect estimates are likely to be more representative of the true measure of effect than the crude ratios. Continuous data, weighted by sample size, were pooled if there was no significant heterogeneity among the mean difference in effect size estimates across studies.
Results
General characteristics of included studies
Using our search strategy for electronic databases and manual citation searches, we identified and screened 4386 citations. Figure 1 outlines the selection process for the final 31 studies included in the qualitative analysis. Characteristics of the 31 studies are summarized in Table 1 . Twenty-nine of these studies were observational in design and the remaining 2 were small-scale interventional studies in which participants were randomized to continuous positive airway pressure (CPAP) or standard treatment.
| Study | Country, population | Participants, n | Study design | Exposure to SDB | Reporting of adverse pregnancy outcomes | Continuous data reported b | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Measurement and definition | Timing | Crude effect estimates | Adjusted variables | Pooled in metaanalysis a | |||||||||
| GHTN | GDM | LBW | GHTN | GDM | LBW | ||||||||
| Ayrlm et al, 2011 | Turkey; hospital | 41 snorers, 159 nonsnorers | Cross-sectional | Snoring questionnaire | During labor | • | – | – | None | – | – | – | Birthweights |
| Bachour et al, 2008 | Finland; cases: hospital, controls: home | 17 PE, 15 controls | Case-control | 1. Oximetry/nasal pressure cannula 2. Snoring questionnaire (for effect estimate) | Third TM | • | – | – | None | – | – | – | AHI, ODI, flow limitation |
| Blyton et al, 2004 | Australia; hospital | 24 PE; 15 controls | RCT | Full overnight PSGs pre-CPAP | Second/third TM | – | – | – | None | – | – | – | AHI |
| Bourjeily et al, 2010 | United States; hospital | 1000 postpartum women from hospital | Cross-sectional | Index 1 of MAPI questionnaire c | 24-48 h postpartum | • | • | • | Age, DM, chronic HTN, renal disease, BMI at delivery, smoking, multifetal pregnancy | • | • | – | None |
| Calaora-Tournadre et al, 2006 | France; hospital | 20 GHTN/PE, 418 controls | Case-control | Snoring questionnaire | Postpartum | • | – | – | None | – | – | – | None |
| Champagne et al, 2009 | Canada; cases: hospital, cases: clinics | 17 GHTN, 33 frequency-matched controls | Case-control | Overnight unattended portable PSG | >20 wks GA, 1 mo postpartum | • | – | – | Maternal age, gestational age, prepregnancy BMI, previous pregnancy, previous live birth | • | – | – | AHI, ODI |
| Chen et al, 2012 | Taiwan; population-based data sets | 791 with OSA diagnostic codes, 4746 controls | Case-control | ICD diagnostic codes for OSA after PSG | Variable | • | • | • | Maternal education, marital status, GDM, GHTN, other comorbidities, obesity, geographic region, paternal age, infants age, parity | • | • | • | Birthweights |
| Connolly et al, 2001 | Ireland; antenatal clinic and ward | 15 PE, 15 controls from each trimester | Case-control | In-hospital overnight limited PSG | Variable | – | – | – | None | – | – | – | AHI, flattening index, flow limitation |
| Edwards et al, 2000 | Australia; cases: hospital controls: clinics | 25 PE, 17 controls | Case-control | Full overnight PSG | Controls: third TM, Cases: unclear | – | – | – | None | – | – | – | RDI |
| Facco et al, 2010 | United States; clinics | 202 nulliparous women | Prospective cohort | Snoring questionnaire | 6-20 wks and third TM | – | • | – | Age, ethnoracial status, BMI, and short sleep duration | – | • | – | 1 h mean OGTT |
| Facco et al, 2012 | United States; hospital database | 145 pregnant who had PSG | Retrospective (cross-sectional) | Full in-lab PSG | Variable | • | • | – | None | – | – | – | None |
| Franklin et al, 2000 | Sweden; hospital | 518 women | Cross-sectional | Snoring questionnaire | Delivery day | • | – | • | Age at delivery, smoking, weight before delivery | • | – | • | None |
| Higgins et al, 2011 | United States; hospital | 1343 (+) Berlin, 2731 (–) Berlin | Prospective cohort | Berlin Questionnaire | Delivery day | • | – | – | None | – | – | – | Birthweights |
| Izci et al, 2005 | Scotland; clinics | 82 PE, 167 controls | Case-control | Snoring questionnaire | Third TM | • | – | – | None | – | – | – | None |
| Izci et al, 2003 | Scotland; cases: hospital, controls: clinic | 37 PE, 50 controls | Case-control | Snoring questionnaire | Third TM | • | – | – | None | – | – | – | None |
| Jniene et al, 2010 | Morocco; hospital | 144 women after delivery | Cross-sectional | Self-reported snoring | Within 24 after delivery | • | – | – | None | – | – | – | None |
| Koken et al, 2007 | Turkey; prenatal clinics | 40 snorers,43 nonsnorers | Prospective cohort | Snoring: Berlin Questionnaire | Second/third TM | – | – | – | None | – | – | – | Birthweights |
| Loube et al, 1996 | United States; prenatal clinics | 49 frequent snorers, 301 nonsnorers | Prospective cohort | Snoring: Hawaii Scale Questionnaire | Second/third TM | – | – | – | None | – | – | – | Birthweights |
| Louis et al, 2010 | United States; tertiary prenatal clinics | 57 with OSA, 114 obese and normal-weight controls | Retrospective cohort | PSG-confirmed OSA | Before and during pregnancy | • | – | • | None for adverse pregnancy outcomes of interest c | – | – | – | Birthweights |
| Micheli et al, 2011 | Greece; population-based cohort | 48 severe snorers, 151 occasional snorers, 892 nonsnorers | Prospective cohort | Snoring, computer-assisted interview | Third TM | – | – | • | Maternal age, education, prepregnancy BMI, smoking | – | – | • | Birthweights |
| Olivarez et al, 2010 | United States; admitted to antepartum service | 20 OSA, 80 no OSA | Prospective cohort | PSG-confirmed OSA | GA ≥26 wks | • | • | – | None | – | – | – | None |
| Perez-Chada et al, 2007 | Argentina; hospital | 156 snorers, 291 never-snorers | Cross-sectional | Snoring frequency, witnessed sleep apnea, questionnaire | Delivery day | • | – | • | Maternal age, prepregnancy BMI, weight gain, neck circumference, smoking, alcohol (for GHTN/PE outcome only) | • | – | – | Birthweights |
| Poyares et al, 2007 | Brazil; obstetrics clinic with preexisting HTN | All chronic snorers: 7 CPAP, 9 no CPAP | RCT | 8 wks of CPAP; PSG only in those treated with CPAP | Initiated first few weeks of pregnancy | – | – | – | None | – | – | – | Birthweights |
| Qiu et al, 2010 | United States; prenatal clinics | 89 snorers, 1169 nonsnorers | Prospective cohort | Self-reported snoring frequency, Interview | <20 wks | – | • | – | Maternal age, ethnicity, stratified by BMI (overweight vs lean) | – | • | – | None |
| Reid et al, 2011 | Canada; obstetrical ward with GHTN | 34 GHTN, 26 controls with PSG | Cross-sectional | PSG-confirmed OSA | Third TM | • | – | – | None | – | – | – | AHI, RERA index, RDI, ODI |
| Reutrakul, et al, 2011 | United States; women with routine OGTT | 26 GDM, 116 NGT | Case-control | ESS, Berlin, frequent snoring, questionnaires | Second TM | – | • | – | BMI | – | • | – | None |
| Sahin et al, 2008 | Turkey; prenatal clinics | 4 OSA, 31 non-OSA | Prospective cohort | PSG-confirmed OSA | Third TM | – | – | – | None | – | – | – | Birthweights |
| Tauman et al, 2011 | Israel; medical ward | 48 habitual snorers, 74 nonsnorers | Cross-sectional | Self-reported snoring, questionnaire | Delivery room | – | – | – | None | – | – | – | Birthweights |
| Ursavas et al, 2008 | Turkey; third trimester, prenatal clinics | 55 habitual snorers, 414 nonhabitual snorers | Prospective cohort | Self-reported snoring | Third TM | • | – | – | None | – | – | – | None |
| Yin et al, 2008 | United Kingdom; clinics and wards | 150 pregnant women with oximetry obtained | Cross-sectional | 1. Snoring: questionnaires (effect estimates) 2. Overnight oximetry | Third TM | • | – | • | None | – | – | – | None |
| Yinon et al, 2006 | Israel; cases: Department of Obstetrics and Gynecology, controls: advertising | 17 PE, 25 matched controls | Case-control | Nocturnal sleep study: Watch-PAT100 | Third TM | – | – | – | None | – | – | – | RDI by Watch-PAT |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
