https://orcid.org/0000-0001-6754-7359
Abstract
Subjective recall of supine sleep during pregnancy has been linked to increased risk of stillbirth, but longitudinal, objective data are lacking. We aimed to examine how sleep position and breathing parameters change throughout pregnancy, and investigated associations between maternal supine sleep, assessed objectively in early and late gestation, and fetal growth velocity in high-risk women.
Women with singleton pregnancies and body mass index (BMI) ≥27 kg/m2 underwent level-III sleep apnea testing. Sleep position was assessed by accelerometry. We derived percentiles of estimated fetal weight and birthweight using FetalGPSR software, then calculated growth velocity as change in percentile/week between the second-trimester anatomy scan and birth.
In total, 446 women were included, with N = 126 in the longitudinal sleep pattern analysis and N = 83 in the fetal growth analysis. Sleep-onset position and predominant sleep position were significantly correlated in both early (p = 0.001) and late (p < 0.01) pregnancy. However, supine going-to-bed position predicted predominant supine sleep in only 47% of women. Between early and late pregnancy there was a reduction in predominant supine sleepers (51.6% to 30.2%). Percent of sleep spent supine and oxygen desaturation index, in the third trimester, were significantly associated after BMI adjustment (B = 0.018, p = 0.04). Models did not suggest significant effects of early or late pregnancy supine sleep on growth velocity (p > 0.05).
Going-to-bed position predicts predominant supine sleep in less than half of women with overweight and obesity. Time spent supine throughout pregnancy correlates with measures of sleep-disordered breathing. Maternal sleep position patterns did not affect fetal growth velocity in this high-risk population, but the study was not powered to detect differences.
Previous cross-sectional and retrospective studies have suggested an association between subjectively reported maternal supine sleep and the risk of stillbirth. This study is the first to examine how objectively measured, longitudinal sleep patterns may be associated with respiratory measures and fetal growth in high-risk pregnancies, which can elucidate the mechanisms of adverse neonatal outcomes. Our findings, that predominant sleep position changes over time and that sleep-onset position is often not an accurate proxy, suggest an important role for objective and longitudinal measurements in sleep research. The lack of association between supine sleep and fetal growth adds a new perspective to the body of literature, as women with overweight and obesity are at higher risk of adverse pregnancy outcomes yet are often excluded from research.
Introduction
The supine position during mid and late pregnancy may cause the gravid uterus to compress the mother’s inferior vena cava, potentially reducing maternal venous return and cardiac output. When lower maternal cardiac output endures for hours, as during sleep, uteroplacental perfusion may be affected and may negatively impact fetal oxygenation [1]. The supine position also increases collapsibility of the maternal airways and may lead to obstructive respiratory events, exacerbating sleep-disordered breathing (SDB) [2–4]. Maternal SDB is a risk factor for numerous maternal cardiovascular and metabolic outcomes [5]. Moreover, it is associated with abnormal fetal growth [6, 7], and poor neonatal outcomes at birth including congenital anomalies and an increased risk for neonatal intensive care unit admission [8].
Retrospective studies in the past decade have explored the association between self-reported maternal supine sleep onset during pregnancy and risk of stillbirth. Some results support that the supine and right lateral going-to-sleep positions are risk factors for late stillbirth [9–11]. Notably, the majority of the studies that demonstrated an association between supine going to sleep positions and stillbirth are retrospective case–control studies, and are limited by potential recall bias.
Silver et al. [12], in a large prospective cohort study that objectively collected sleep position as part of the evaluation of sleep-disordered breathing, found no significant association between supine or right lateral sleep position and stillbirth, small-for-gestational-age newborns, or gestational hypertensive disorders, among nulliparous women with singleton pregnancies [12]. Despite these results suggesting that supine sleep may not increase the risk of poor neonatal outcomes in low-risk pregnancies, significant hemodynamic changes in the fetus [13] and the mother [14] associated with maternal sleep position, may occur in high-risk pregnancies.
In most studies demonstrating an association between stillbirth and supine sleep, going to sleep position was used as a surrogate for predominant sleep position, was mainly self-reported, and study design retrospective, or cross-sectional [15]. Furthermore, the majority of past research has only assessed women with uncomplicated pregnancies, with limited inclusion of high-risk pregnancies [12–14]. It is essential that women with overweight or obesity, pre-gestational diabetes, chronic hypertension, or other risk factors be included in these studies because of the higher incidence of adverse pregnancy outcomes in this population, and because of the high prevalence of these disorders in the population at large [12]. Notably, elevated body mass index (BMI) >25 kg/m2 is an independent risk factor for stillbirth [16]. It has also been suggested that supine sleep may only be a risk factor for late stillbirth in high-risk pregnancies, in which the fetus is already at an increased vulnerability [17]. There is also a paucity of data on maternal sleep position across different trimesters of pregnancy and on the effects of sleep position on maternal respiratory parameters and fetal outcomes. Stillbirth is a rare outcome occurring in 5.9 pregnancies for every 1000 live births [18]; however, this rate has shown no improvement in the last decades [16]. In high-income countries, abnormal fetal growth, whether fetal growth restriction due to placental insufficiency or accelerated fetal growth, contributes significantly to an increased risk for stillbirth [16, 19]. Despite that, abnormal fetal growth is often underdiagnosed [20]. Cross-sectional measures of fetal growth contribute to underestimating the risk for impaired growth as they fail to measure growth stunting or reductions that do not meet criteria for the single measure; hence, growth velocity may be a preferred measure [21].
The primary aim of this study is to elucidate whether maternal sleep position patterns, measured by objective methods, change throughout pregnancy among women with overweight or obesity, and whether sleep position is associated with respiratory parameters of SDB (respiratory event index [REI] and oxygen desaturation index [ODI]). The secondary aim is to investigate the association between maternal sleep position, fetal growth velocity, and birth weight.
Methods
Data collection
This study is a secondary analysis of data collected on 446 pregnant women who were recruited as part of two other studies investigating maternal sleep [22]. Participants were recruited from community and hospital-based obstetric practices. Inclusion criteria for both cohorts were: age >18 years, singleton pregnancies, BMI ≥27 kg/m2, gestational age ≤13 weeks, habitual snoring, intention to reside locally for the duration of the pregnancy, and ability to provide informed consent. Exclusion criteria for both cohorts were serious mental illness affecting participation. One of the two cohorts also excluded women with advanced cardiac disease and chronic respiratory illness.
Of the participants who were offered a third-trimester visit (N = 251), 13 withdrew, 22 were lost to contact or did not keep follow-up, 19 had medical complications that necessitated hospitalization or impacted the scheduled study visit, 5 had poor quality follow up Nox data, and 30 study visits could not be scheduled due to COVID restrictions. Sixteen participants could not complete the follow-up visit for multiple, or other reasons. A total of 147 participants completed the third-trimester sleep study and 126 had complete data to be used in this analysis.
Maternal demographics.
Demographic information and past medical and reproductive history were obtained. Medical history relevant to this study included preexisting diabetes mellitus (DM), history of chronic hypertension, and current use of tobacco. Reproductive history included gestational DM (GDM), hypertensive disorders of pregnancy (HDP) (gestational hypertension, preeclampsia, or eclampsia) documented by chart review, and use of in vitro fertilization (IVF) in the current pregnancy.
Sleep study procedures.
Each of the enrolled participants underwent home sleep studies using portable Noxturnal T3 devices (Noxmedical, Swansea GA, USA), a level-III recording sleep apnea testing device. Of the 446 participants who underwent a first-trimester sleep study, 126 completed a second sleep study in the third trimester of pregnancy—as required for longitudinal examination of changes in sleep pattern over time (see Figure 1—“Longitudinal Cohort”). Level-III testing devices are designed for home use and can provide increased efficiency and patient comfort and decreased costs compared to laboratory-based level-I polysomnography [23]. Validation studies have found that Nox T3 has a specificity of 88%–95%, as well as a sensitivity of 85%–100%, when used in the diagnosis of obstructive sleep apnea using apnea–hypopnea index (AHI) criteria, in participants with high BMIs [24] and low BMIs [25]. Level III technology has also been validated compared to polysomnography in pregnancy [26, 27].
Participant inclusion. The early pregnancy cohort was used to examine the relationship between sleep position and breathing measures during sleep in early pregnancy. The longitudinal cohort was used to examine changes in breathing parameters across pregnancy. The fetal growth cohort was used to examine fetal growth.
The following channels were recorded by the Nox devices for each sleep study: snoring channel, nasal and oral airflow (nasal pressure transducer, thermistor), chest and abdominal respiratory movement using respiratory inductance plethysmography, oxygen saturation (SpO2), and body position sensor (SPI, Pro-Tech, USA) attached to the anterior chest wall on the median line. Body position was measured using a 3-D built acceleration sensor for recording of patient’s position and activity. The software automatically analyzed sleep position and differentiated five positions including the supine, left lateral, right lateral, prone, and upright positions. All studies were scored by the same certified sleep technologist with many years’ of experience, using American Academy of Sleep Medicine 2012 recommended criteria [28]. Apneas were defined as a ≥90% reduction in airflow signal compared to baseline, for ≥10 seconds, while hypopnea was defined based on the recommended rule of ≥3% drop in SpO2 [28].
Data collected objectively from the Nox T3 devices included time spent in bed, estimated time spent asleep, percent of the night spent in each sleep position, and REI and ODI.
Sleep position scoring.
Going-to-bed position, sleep-onset position, and waking-up position were manually recorded from each Nox study. The first non-upright position was recorded as going-to-bed position. The first non-upright position that lasted for more than 5 minutes was recorded as sleep-onset position. The 5-minute minimum epoch to define sleep-onset position was selected based on prior studies’ use of intervals of 3- [29] to 10-minute [30] epochs to establish similar sleep position variables. Waking-up position was recorded as the last non-upright position before going upright at the end of the recording.
Total number of sleep position changes throughout the night was also recorded. The number of sleep position changes was calculated using 30-second epochs; only positions that lasted 30 seconds or longer were included in this calculation. This standard is used to exclude insignificant periods of brief movement [31, 32].
Fetal growth.
The fetal growth cohort is a subset of the longitudinal cohort with (1) available fetal biometric data and (2) a known single maternal race, due to the growth percentile calculator requiring one race as input.
We reviewed prenatal and delivery medical records. Estimated due date was based on the last menstrual period and confirmed by first-trimester ultrasounds. We extracted fetal biometric data from clinical ultrasounds performed in the second trimester (between 17–24 weeks), including estimated fetal weight (EFW), biparietal diameter, occipitofrontal circumference, head circumference, abdominal circumference, and femur length. Clinical ultrasounds were performed at two different centers, both of which use similar methods to derive EFW [33].
We derived EFW and birthweight percentile by using the FetalGPSR software and the normal reference range of the fetal biometry created by the National Institute of Child Health and Human Development (NICHD) [34]. The NICHD reference criteria were selected for this study due to similarities between participants in this study and the study population used to develop NICHD standards. Growth percentiles are a method of quantitatively comparing a fetus’s growth to standard measures for their gestational age, developed using data from a large sample of pregnancies, by accounting for potential confounding variables such as maternal demographics [34]. Fetal growth velocity was expressed as change in percentile per week, by comparing EFW and birthweight. Abnormal fetal growth was categorized into small-for-gestational-age as less than the 10th percentile, and large for gestational age as greater than the 90th percentile.
Analysis
Among participants included in all cohorts, descriptive statistics were used for demographic variables and data were reported as mean/standard deviation for continuous variables and n/ % for categorical variables. In the early pregnancy cohort, an analysis of variance (ANOVA) was used to examine differences in REI and ODI by sleep position to determine the validity of the REI and ODI measures, as there is a known correlation between time spent supine and these respiratory parameters during sleep [4]. Linear regression analyses examined associations among supine sleep position and REI and ODI, adjusting for BMI. Similar analyses were performed using third-trimester data from the longitudinal cohort. In the longitudinal cohort, a multinomial logistic regression was performed to predict predominant sleep position as a function of going-to-bed position and waking-up position. In assessing the changes in sleep position patterns over time between early and late pregnancy, longitudinal regression models were used to determine if the number of position changes throughout the night or the overall distribution of sleep position changes over time, using predominant sleep position as the outcome of interest.
Differences in key potential confounders (BMI, age, GDM, tobacco use, and IVF experience) were compared between predominantly supine sleepers versus predominantly non-supine sleepers in early and late pregnancy using t-tests for continuous variables and chi-squared tests for categorical variables. Additional chi-squared tests were performed to test the association between supine sleep in early and late pregnancy and adverse outcomes including gestational hypertension, preeclampsia, gestational diabetes, and preterm birth. Predominant sleep position was defined as the position with the relative majority in the breakdown of percentages of one sleep study.
Linear regression models were used to examine associations between sleep position (percent of sleep spent supine and number of sleep position changes, assessed individually in separate models) and growth velocity. Both unadjusted models (separately at early and late pregnancy) and multivariable models controlling for maternal age, race, BMI, GDM, tobacco use in early pregnancy, HDP including chronic hypertension and preeclampsia, and REI were explored. Preexisting DM and use of IVF for the current pregnancy, potential confounders due to their reported effects on risk of growth acceleration [35] and restriction [36], respectively, were not included in multivariable models because of their low prevalence in this cohort. We performed a sensitivity analysis to determine if results changed after excluding for preexisting DM. An independent samples t-test was completed to assess any associations between first-trimester sleep position and neonatal size at birth. Significance level was set at 0.05 a priori.
Results
A total of 446 participants were included in the primary analysis. Demographic data for the three cohorts are presented in Table 1. Average maternal BMI at screening, in early pregnancy, was elevated in all cohorts. Average gestational age at which sleep studies were completed was 11 weeks + 3 days (range [6, 20] weeks) in early pregnancy, and 32 weeks + 3 days (range [29, 37] weeks) in late pregnancy.
All Cohort Demographics
| Early pregnancy cohort (N = 446) | Longitudinal cohort (N = 126) | Fetal growth cohort (N = 83) | |
|---|---|---|---|
| Maternal age (years)—mean (SD) | 30.22(5.72) | 30.46(5.40) | 31.28(5.29) |
| Maternal BMI (kg/m2)—mean (SD) | 34.60(7.23) | 34.00(5.14) | 34.46(5.27) |
| Parity—mean (SD) | 1.22(1.37) | 1.14(1.09) | 1.12(1.09) |
| Nullipara—n (%) | 160(35.9) | 40(31.7) | 29(34.9) |
| Race/ethnicity—n (%) | |||
| Non-Hispanic white | 222(50.4) | 71(56.3) | 54(65.1) |
| Non-Hispanic black | 71(15.9) | 11(8.7) | 8(9.6) |
| Hispanic or Latina American | 122(27.4) | 34(27.0) | 18(21.7) |
| Indian/Alaska Native | 16(3.6) | 2(1.6) | 0(0.0) |
| Asian | 15(3.3) | 4(3.2) | 3(3.6) |
| Multiple | 48(10.8) | 11(2.5) | — |
| Other/Unknown | 6(1.3) | 1(0.8) | — |
| Pre-gestational DM—n (%) | 18(4.0) | 6(4.8) | 5(6.0) |
| GDM—n (%) | 43(9.6) | 18(14.3) | 12(14.46) |
| Maternal HDP—n (%) | — | 39(31.0) | 29(34.9) |
| Tobacco in early pregnancy—n (%) | 52(11.7) | 10(7.9) | 6(7.2) |
| IVF—n (%) | 13(2.9) | 1(0.8) | 1(1.2) |
| Early pregnancy cohort (N = 446) | Longitudinal cohort (N = 126) | Fetal growth cohort (N = 83) | |
|---|---|---|---|
| Maternal age (years)—mean (SD) | 30.22(5.72) | 30.46(5.40) | 31.28(5.29) |
| Maternal BMI (kg/m2)—mean (SD) | 34.60(7.23) | 34.00(5.14) | 34.46(5.27) |
| Parity—mean (SD) | 1.22(1.37) | 1.14(1.09) | 1.12(1.09) |
| Nullipara—n (%) | 160(35.9) | 40(31.7) | 29(34.9) |
| Race/ethnicity—n (%) | |||
| Non-Hispanic white | 222(50.4) | 71(56.3) | 54(65.1) |
| Non-Hispanic black | 71(15.9) | 11(8.7) | 8(9.6) |
| Hispanic or Latina American | 122(27.4) | 34(27.0) | 18(21.7) |
| Indian/Alaska Native | 16(3.6) | 2(1.6) | 0(0.0) |
| Asian | 15(3.3) | 4(3.2) | 3(3.6) |
| Multiple | 48(10.8) | 11(2.5) | — |
| Other/Unknown | 6(1.3) | 1(0.8) | — |
| Pre-gestational DM—n (%) | 18(4.0) | 6(4.8) | 5(6.0) |
| GDM—n (%) | 43(9.6) | 18(14.3) | 12(14.46) |
| Maternal HDP—n (%) | — | 39(31.0) | 29(34.9) |
| Tobacco in early pregnancy—n (%) | 52(11.7) | 10(7.9) | 6(7.2) |
| IVF—n (%) | 13(2.9) | 1(0.8) | 1(1.2) |
n(%) reported for categorical variables; mean(SD) for continuous variables.
Race/ethnicity totals do not equal cohort N due to multiple races.
BMI, body mass index; DM, diabetes mellitus; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; IVF, in vitro fertilization.
All Cohort Demographics
| Early pregnancy cohort (N = 446) | Longitudinal cohort (N = 126) | Fetal growth cohort (N = 83) | |
|---|---|---|---|
| Maternal age (years)—mean (SD) | 30.22(5.72) | 30.46(5.40) | 31.28(5.29) |
| Maternal BMI (kg/m2)—mean (SD) | 34.60(7.23) | 34.00(5.14) | 34.46(5.27) |
| Parity—mean (SD) | 1.22(1.37) | 1.14(1.09) | 1.12(1.09) |
| Nullipara—n (%) | 160(35.9) | 40(31.7) | 29(34.9) |
| Race/ethnicity—n (%) | |||
| Non-Hispanic white | 222(50.4) | 71(56.3) | 54(65.1) |
| Non-Hispanic black | 71(15.9) | 11(8.7) | 8(9.6) |
| Hispanic or Latina American | 122(27.4) | 34(27.0) | 18(21.7) |
| Indian/Alaska Native | 16(3.6) | 2(1.6) | 0(0.0) |
| Asian | 15(3.3) | 4(3.2) | 3(3.6) |
| Multiple | 48(10.8) | 11(2.5) | — |
| Other/Unknown | 6(1.3) | 1(0.8) | — |
| Pre-gestational DM—n (%) | 18(4.0) | 6(4.8) | 5(6.0) |
| GDM—n (%) | 43(9.6) | 18(14.3) | 12(14.46) |
| Maternal HDP—n (%) | — | 39(31.0) | 29(34.9) |
| Tobacco in early pregnancy—n (%) | 52(11.7) | 10(7.9) | 6(7.2) |
| IVF—n (%) | 13(2.9) | 1(0.8) | 1(1.2) |
| Early pregnancy cohort (N = 446) | Longitudinal cohort (N = 126) | Fetal growth cohort (N = 83) | |
|---|---|---|---|
| Maternal age (years)—mean (SD) | 30.22(5.72) | 30.46(5.40) | 31.28(5.29) |
| Maternal BMI (kg/m2)—mean (SD) | 34.60(7.23) | 34.00(5.14) | 34.46(5.27) |
| Parity—mean (SD) | 1.22(1.37) | 1.14(1.09) | 1.12(1.09) |
| Nullipara—n (%) | 160(35.9) | 40(31.7) | 29(34.9) |
| Race/ethnicity—n (%) | |||
| Non-Hispanic white | 222(50.4) | 71(56.3) | 54(65.1) |
| Non-Hispanic black | 71(15.9) | 11(8.7) | 8(9.6) |
| Hispanic or Latina American | 122(27.4) | 34(27.0) | 18(21.7) |
| Indian/Alaska Native | 16(3.6) | 2(1.6) | 0(0.0) |
| Asian | 15(3.3) | 4(3.2) | 3(3.6) |
| Multiple | 48(10.8) | 11(2.5) | — |
| Other/Unknown | 6(1.3) | 1(0.8) | — |
| Pre-gestational DM—n (%) | 18(4.0) | 6(4.8) | 5(6.0) |
| GDM—n (%) | 43(9.6) | 18(14.3) | 12(14.46) |
| Maternal HDP—n (%) | — | 39(31.0) | 29(34.9) |
| Tobacco in early pregnancy—n (%) | 52(11.7) | 10(7.9) | 6(7.2) |
| IVF—n (%) | 13(2.9) | 1(0.8) | 1(1.2) |
n(%) reported for categorical variables; mean(SD) for continuous variables.
Race/ethnicity totals do not equal cohort N due to multiple races.
BMI, body mass index; DM, diabetes mellitus; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; IVF, in vitro fertilization.
When comparing participants who completed the third-trimester visit to those who were offered the third-trimester visit but did not complete it, those who completed the study were slightly older 30.6 ± 5.5 versus 28.9 ± 5.6 years, but did not differ on any other demographic or baseline risk.
Sleep position in early and late pregnancy
Mean number of position changes was similar in early (14.19 ± 7.82, range [0,53]) versus late pregnancy (14.58 ± 8.25, range [1,46]) (p = 0.65). There was a significant correlation between sleep-onset position and predominant sleep position in both early (p = 0.001) and late (p < 0.01) pregnancy. However, supine going-to-bed position predicted predominant supine sleep in only 47% of women. Most participants (76.2% in first trimester and 77.2% in third trimester) spent at least part of the night in each of the main sleep positions: supine, left, and right. There was a significant change in sleep position between early and late pregnancy (p = 0.04) with a reduction in the percentage of participants sleeping predominantly supine (51.6% to 30.2%) and an increase in predominant left lateral sleepers (24.6% to 37.3%) (Figure 2), along with a reduction in the average proportion of time spent in the supine position overnight (early pregnancy = 40.7% vs. late pregnancy = 32.9%, p = 0.015). Thirty (23.8%) women maintained their predominant supine sleep from early to late pregnancy.
Change in predominant sleep position over time, N = 126. Percentage of total participants in each predominant sleep position within the Longitudinal Cohort changed significantly between early and late pregnancy (p = 0.04), notably with a decrease in predominantly supine sleepers and an increase in predominantly left lateral sleepers.
In the early pregnancy cohort, REI were higher in predominantly supine sleepers compared to predominantly non-supine sleepers (3.93 vs. 2.45, t = −2.24, p = 0.026). Furthermore, rates of GDM were higher in predominantly supine sleepers compared to predominantly non-supine sleepers (13% vs. 6%, χ2 = 5.58, p = 0.018). Supine position was not associated with higher incidence for HDP (χ2 = 0.625, p = 0.425), but was associated with lower incidence of preterm birth (χ2 = 6.252, p = 0.012).
In the fetal growth cohort, rates of GDM were again significantly higher in predominantly supine sleepers than predominantly non-supine sleepers in early pregnancy (22.5% vs. 6.8%, χ2 = 4.10, p = 0.04).
Sleep position and breathing parameters during sleep
Mean REI was 3.3 (SD = 6.9, range [0,84]) among all first-trimester sleep studies and 5.5 (SD = 7.9, range [0.2,69.1]) in the third trimester, representing a significant increase in REI over pregnancy (p < 0.01). In the fetal growth cohort, first-trimester mean REI was 2.5 (SD = 2.3) among predominantly supine sleepers and 3.4 (SD = 3.4) events per hour among predominantly non-supine sleepers, and third-trimester mean REI was 7.6 (SD = 9.1) for predominantly supine sleepers and 5.3 (SD = 9.3) events per hour for predominantly non-supine sleepers. These differences within trimesters were not statistically significant. Mean ODI did not change significantly between first (6.0, SD = 7.6) and third (6.9, SD = 7.1) trimester. In both early and late pregnancy, percent of sleep spent supine was significantly associated with both REI and ODI (p values < 0.05, r values ranged from 0.24 to 0.35). After adjusting for BMI, percent of time spent in the supine sleep position was significantly associated with ODI in the third trimester (ß = 0.18, p = 0.036) (Figure 3).
Respiratory event indices as a function of percent of the night spent supine REI: respiratory event index; ODI: oxygen desaturation index. *Only ODI remained significantly associated with percent of sleep spent supine, in the third trimester, after adjusting for BMI (Figure 3D).
Maternal sleep patterns and fetal growth velocity
Average gestational age at second-trimester anatomy scan and at birth are reported for the 83 participants included in the fetal growth cohort in Table 2. There was a large distribution of growth percentiles at both time points. Of these 83 participants, 47 in the second trimester and 11 at birth had babies exhibiting abnormal growth (growth percentile <10th or >90th), with five displaying abnormal growth at both timepoints. In the second trimester, 46 of these 47 babies (97.9%) with abnormal growth were above the 90th percentile, and at birth, 8 of the 11 (72.7%) were above the 90th percentile.
Overview of Gestational Age and Growth Percentile Distributions
| Second-trimester anatomy scan | Birth | |
|---|---|---|
| GA (weeks + days) | 19 + 4 (1 + 3) | 38 + 5 (1 + 4) |
| Growth percentile | 81.4 (22.1) | 54.4 (24.2) |
| Second-trimester anatomy scan | Birth | |
|---|---|---|
| GA (weeks + days) | 19 + 4 (1 + 3) | 38 + 5 (1 + 4) |
| Growth percentile | 81.4 (22.1) | 54.4 (24.2) |
N = 83 participants in fetal growth cohort; mean (SD) reported. GA, gestational age.
Overview of Gestational Age and Growth Percentile Distributions
| Second-trimester anatomy scan | Birth | |
|---|---|---|
| GA (weeks + days) | 19 + 4 (1 + 3) | 38 + 5 (1 + 4) |
| Growth percentile | 81.4 (22.1) | 54.4 (24.2) |
| Second-trimester anatomy scan | Birth | |
|---|---|---|
| GA (weeks + days) | 19 + 4 (1 + 3) | 38 + 5 (1 + 4) |
| Growth percentile | 81.4 (22.1) | 54.4 (24.2) |
N = 83 participants in fetal growth cohort; mean (SD) reported. GA, gestational age.
Overall, average fetal growth percentile decreased significantly between the second trimester and birth (p < 0.0001), resulting in an average negative growth velocity, measured in growth percentile change/week (mean = −1.37, SD = 1.57). Models did not suggest significant effects of percentage of time spent in supine sleep in early or late pregnancy on growth velocity (b = −0.033, p = 0.75 early pregnancy and b = 0.014, p = 0.90 late pregnancy). There were no statistically significant differences in neonatal size among women who predominantly slept in the supine versus non-supine position in early pregnancy (p-values > 0.56).
In the aggregate sample of the 83 participants included in the fetal growth cohort, there were no significant differences in mean growth velocity between predominantly supine sleepers and predominantly non-supine sleepers in early or late pregnancy. Neither unadjusted nor multivariable models, while simultaneously adjusting for maternal age, race, BMI, GDM, tobacco use in early pregnancy, HDP including chronic hypertension and preeclampsia, and REI, indicated that predominant supine sleep position predicted growth velocity. The adjusted multivariable model is presented in Table 3. The number of changes in sleep position was not significantly associated with growth rates in early or late pregnancy. Of note, this sample was underpowered to detect differences in fetal growth.
Adjusted Multivariable Model of Growth Velocity as a Function of Percent of the Night Spent Supine in Late Pregnancy
| ß | Std. Error | 95% CI | |
|---|---|---|---|
| Percent supine (third trimester) | 0.014 | 0.009 | −0.017, 0.019 |
| Maternal age | 0.057 | 0.040 | −0.058, 0.102 |
| Maternal BMI | 0.052 | 0.043 | −0.065, 0.106 |
| Tobacco use | −0.052 | 0.779 | −1.932, 1.357 |
| GDM | 0.182 | 0.574 | −0.172, 2.108 |
| Any HDP | 0.057 | 0.619 | −0.899, 1.558 |
| REI (third trimester) | −0.150 | 0.024 | −0.083, 0.014 |
| Race (non-Hispanic black) | −0.072 | 1.219 | −3.271, 1.570 |
| ß | Std. Error | 95% CI | |
|---|---|---|---|
| Percent supine (third trimester) | 0.014 | 0.009 | −0.017, 0.019 |
| Maternal age | 0.057 | 0.040 | −0.058, 0.102 |
| Maternal BMI | 0.052 | 0.043 | −0.065, 0.106 |
| Tobacco use | −0.052 | 0.779 | −1.932, 1.357 |
| GDM | 0.182 | 0.574 | −0.172, 2.108 |
| Any HDP | 0.057 | 0.619 | −0.899, 1.558 |
| REI (third trimester) | −0.150 | 0.024 | −0.083, 0.014 |
| Race (non-Hispanic black) | −0.072 | 1.219 | −3.271, 1.570 |
BMI, body mass index; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; REI, respiratory event index.
Adjusted Multivariable Model of Growth Velocity as a Function of Percent of the Night Spent Supine in Late Pregnancy
| ß | Std. Error | 95% CI | |
|---|---|---|---|
| Percent supine (third trimester) | 0.014 | 0.009 | −0.017, 0.019 |
| Maternal age | 0.057 | 0.040 | −0.058, 0.102 |
| Maternal BMI | 0.052 | 0.043 | −0.065, 0.106 |
| Tobacco use | −0.052 | 0.779 | −1.932, 1.357 |
| GDM | 0.182 | 0.574 | −0.172, 2.108 |
| Any HDP | 0.057 | 0.619 | −0.899, 1.558 |
| REI (third trimester) | −0.150 | 0.024 | −0.083, 0.014 |
| Race (non-Hispanic black) | −0.072 | 1.219 | −3.271, 1.570 |
| ß | Std. Error | 95% CI | |
|---|---|---|---|
| Percent supine (third trimester) | 0.014 | 0.009 | −0.017, 0.019 |
| Maternal age | 0.057 | 0.040 | −0.058, 0.102 |
| Maternal BMI | 0.052 | 0.043 | −0.065, 0.106 |
| Tobacco use | −0.052 | 0.779 | −1.932, 1.357 |
| GDM | 0.182 | 0.574 | −0.172, 2.108 |
| Any HDP | 0.057 | 0.619 | −0.899, 1.558 |
| REI (third trimester) | −0.150 | 0.024 | −0.083, 0.014 |
| Race (non-Hispanic black) | −0.072 | 1.219 | −3.271, 1.570 |
BMI, body mass index; GDM, gestational diabetes mellitus; HDP, hypertensive disorders of pregnancy; REI, respiratory event index.
Discussion
This study has demonstrated several important findings. (1) When utilizing objective measures of sleep position, we demonstrated that going-to-bed position predicts predominant supine sleep in less than half of all pregnant women and constitutes a poor surrogate for the predominant sleep position overnight. (2) Longitudinal data demonstrated that pregnant women changed position on average 14 times during the night, both early and late in gestation and, (3) significantly fewer women were predominantly supine sleepers in late pregnancy. Though we did not investigate reasons for changes in sleep position, this may possibly be due to advice about better placental perfusion in a lateral sleeping position, or to physical discomfort while supine with pregnancy progression. (4) Similarly to the non-pregnant population, the supine position was associated with measures of SDB, both in early and late pregnancy. However, only the association between the percent of sleep spent supine and ODI in the third trimester remained significant after adjusting for BMI. Finally,(5) Being a predominantly supine sleeper, spending a larger percentage of the night supine, and changing sleep position more frequently were not associated with differences in fetal growth velocity expressed as change in growth percentile/week.
Supine sleep position has gained significant attention in the obstetric population in recent years due to a described association with stillbirth [9–11]. Though stillbirth is a rare outcome, its frequency has not decreased in high-income countries in recent years [16]. Understandably, many studies utilized a case–control design. However, inherent limitations of this design include potential recall bias [15]. Furthermore, many of the published studies employed going to bed position as a surrogate for the predominant sleep position. This study demonstrates that going to sleep position predicts the predominant sleep position in less than half of the women. In a previous study that examined the typical sleep positions in late pregnancy in three cohorts (healthy pregnancy, HDP, and small for gestational age babies), we similarly demonstrated that sleep-onset position was the dominant sleep position in just 54% of the sample [38]. This calls into question the validity of going to bed position as a surrogate to predominant sleep position, a measure used in the bulk of the research that demonstrated a positive association of supine sleep with adverse fetal outcomes such as reduced birthweight and stillbirth [37].
There is a small but statistically significant association between ODI and percent of sleep spent supine during the third trimester of pregnancy. This suggests a potential role for the gravid uterus in the pathophysiology of SDB. This third-trimester association was not present when using REI as the dependent variable and adjusting for BMI. These third-trimester observations are similar to the results of a prior study, which found no significant association between REI and time spent supine despite a significant negative correlation between percent of sleep spent supine and SpO2 nadir, in a cohort of low-risk pregnancies [4]. This discrepancy may support the ongoing debate about whether REI, or AHI, is the best and only measure of SDB, as alternative measures may be more sensitive or more predictive of adverse outcomes.
It is possible that the lack of association between supine sleep and fetal growth rates observed in the multivariable analyses is due to multiple opposing associations. Specifically, it has been suggested through cross-sectional analyses that supine sleep is associated with fetal growth restriction [38], but incidence of gestational diabetes—a condition associated with larger babies [39]—was significantly higher in our predominantly supine cohort versus predominantly non-supine sleepers. This could help explain the trends observed in our analyses by masking overall fetal growth restriction, though the regressions controlled for GDM. Similarly to GDM, SDB has been associated with growth acceleration in some studies of lean pregnant women [7]. As supine sleep was positively associated with measures of SDB, it is possible that SDB may have contributed to the opposing directions in fetal growth, though this was not apparent based on our model. Hence, despite potential hemodynamic changes expected in the supine position in a cohort with obesity, supine sleep was not negatively associated with fetal growth in late pregnancy, suggesting potential protective factors, though sample size was a limitation as mentioned below.
Strengths and limitations
This study has several strengths. This study utilizes objective data to determine sleep position in a protocol that is not aimed at assessing sleep position, minimizing potential bias from participants being more aware of their position and altering their natural sleep patterns. Participants were tested in the comfort of their own beds and were not given any instructions on sleep position. Furthermore, the population studied is racially and ethnically diverse, and with overweight or obesity, risk factors for adverse fetal and neonatal outcomes. Thus, the results of this study may be generalized to a larger, similarly diverse population of pregnant people with BMI ≥27 kg/m2. This study also has the advantage of utilizing longitudinal data on the same participants to examine changes in sleep position over time in pregnancy and longitudinal, rather than cross-sectional, fetal growth.
Limitations of this study include those inherent to level-III recording devices, lack of video recording option for position confirmation and electroencephalogram to determine sleep onset. It is currently unclear if the potential impact of supine position varies depending on whether the participant is asleep or awake. Furthermore, the sleep studies were conducted over only one night in early pregnancy and one night in late pregnancy; this analysis assumes that sleep position patterns are relatively consistent within trimesters. The Noxturnal devices may also underestimate REI compared to in-laboratory polysomnography, though the device has been validated for the detection of obstructive sleep apnea. In addition, this study is limited by the small sample size for the growth analysis and is underpowered to detect differences in fetal growth velocity or other perinatal outcomes, partially due to inconsistent timing of second-trimester anatomy scans. Although the use of longitudinal fetal growth data was innovative, EFW calculated from ultrasound and birth weight are inherently different measures, and future studies should utilize multiple ultrasounds to better compare growth percentiles over time. Finally, although these analyses include a large number of covariates, there may exist some residual confounding. The participant pool allows for wide generalizability of the results among pregnant populations with overweight and obesity, these results cannot be applied to a more general population, as the implications of sleep habits on fetal growth may be different in lean individuals or those without the same comorbidities.
Conclusion
Going-to-bed position predicts predominant supine sleep position in less than half of women with overweight and obesity. Time spent supine in both early and late pregnancy correlates with measures of sleep-disordered breathing. Sleeping supine in early or late pregnancy, and number of sleep position changes, do not appear to be associated with longitudinal fetal growth velocity. More prospective studies are needed to evaluate the potential for sleep position changes over time as a potentially modifiable risk factor for maternal and neonatal health outcomes, as well as other factors that influence fetal growth rate. Further examinations of fetal growth and the mechanisms behind the associations observed in this study will help elucidate the clinical significance of these findings in regard to maternal and neonatal health.
Future studies should aim to examine sleep measures to investigate interactions between position, actual sleep (vs. awake but supine), and fetal measures, as these measures may also have implications for pregnant women on prolonged bed rest. Additional research should also focus on mechanistic pathways behind the observed associations and on obtaining repeated measures.
Funding
This work was funded by National Heart Lung and Blood Institute R01HL130702 (GB) and T35 HL094308 (JK), and the National Institute for Child Health and Human Development R01HD078515 (GB).
Acknowledgments
The authors thank the participants of this study.
Disclosure Statement
Financial disclosure: The authors have no financial conflicts of interest to report. Nonfinancial disclosure: None.
Data Availability
De-identified data will be made available upon reasonable request.
Comments