Obstructive sleep apnea among gravidas with chronic hypertension compared to matched controls: a prospective cohort study

Abstract

Background:

Prior studies of obstructive sleep apnea (OSA) risk in gravidas with chronic hypertension (cHTN) did not control for obesity as a risk factor for OSA. We therefore performed this study to evaluate whether OSA is more prevalent among gravidas with cHTN compared to normotensive gravidas matched for body mass index (BMI) and gestational age (10–20 weeks) (primary outcome). We also assessed if OSA is more severe when co-morbid with cHTN in pregnancy (secondary outcome).

Methods:

This was a single-center, prospective cohort study. Adult gravidas, 10–20 weeks gestation, with and without cHTN were enrolled and BMI-matched. All subjects answered OSA screening questionnaires and underwent a home sleep test once between 10–20 weeks gestation. Pregnancy outcomes were followed for all subjects. We performed univariable and multivariable logistic regression to model the relationship between cHTN status and OSA.

Results:

100 pregnant subjects (50 with cHTN and 50 normotensive) completed a home sleep test of 2 hours or more. There were no differences in demographic variables between the two groups, except gravidas with cHTN were significantly older than normotensive subjects (mean±SD 34±4 vs. 30±6 years; p< 0.001). OSA was more prevalent [64% vs 38%; p=0.009; OR (95% CI) 2.90 (1.30, 6.65), p = 0.01] and more severe in gravidas with cHTN (moderate or severe OSA 59% vs. 21%; p=0.009). After controlling for age, we found no overall association between cHTN on OSA risk [aOR (95% CI) 2.22 (0.92, 5.40), p=0.076]. However, among gravidas older than 25 years, cHTN was associated with higher odds of OSA [aOR (95% CI), 2.64 (1.06, 6.71), p = 0.038].

Conclusions:

CHTN and age are important risk factors for OSA in gravidas. Gravidas with cHTN should be screened for OSA in early pregnancy. Future studies may validate screening tools that include cHTN and age, and investigate the role of OSA therapy in blood pressure control.

INTRODUCTION:

Cardiovascular disease is now the leading cause of maternal death in the United States.¹ Chronic hypertension (cHTN) complicates 3 to 5% of all pregnancies and is associated with severe maternal and neonatal morbidity.^2,3 Obstructive sleep apnea (OSA) is a known risk factor for cHTN, preeclampsia and other cardiovascular diseases in pregnant and non-pregnant adults, but no studies have yet shown that OSA treatment can mitigate these adverse peripartum outcomes.^4–8 However, OSA is associated with increased risk of requiring cesarean delivery and has been implicated as a risk factor for adverse peripartum outcomes.^5,9–11 Recent guidelines regarding safe administration practices for neuraxial morphine at the time of cesarean delivery identify OSA as a risk factor for requiring increased postpartum monitoring.¹² OSA as a co-morbidity in gravidas is likely under-appreciated and under-diagnosed due to a number of factors: a lack of validated screening tools; limited longitudinal studies on OSA across trimesters and postpartum; and practical challenges that limit the availability of sleep testing.¹³ In addition, OSA may be under-recognized in reproductive age women as the clinical presentation can differ compared to the presentation in postmenopausal women and men.¹⁴

Prior studies of OSA risk in gravidas with cHTN did not control for obesity as a risk factor for OSA.^{15 16} Although the associations between OSA, cHTN and adverse pregnancy outcomes such as preeclampsia have raised considerable concern,⁸ proposed pregnancy specific OSA screening tools have not been successfully validated in unrelated datasets.^17–20

METHODS:

This study was approved by the Duke University’s Institutional Review Board (IRB # Pro00081272) in March 2017 and written informed consent was obtained from all subjects participating in the trial. The trial was registered prior to patient enrollment at clinicaltrials.gov (NCT03230110, Principal investigator: Dr. Jennifer Dominguez, Date of registration: June 2017).

Aims:

The primary aim of this prospective observational study was to evaluate whether undiagnosed OSA is more prevalent among early gestation gravidas with cHTN compared to normotensive gravidas matched for body mass index (BMI) and gestational age. The secondary aim was to evaluate if OSA is more severe in gravidas with cHTN compared to normotensive gravidas. We hypothesized that OSA would be more prevalent and more severe in gravidas with cHTN compared to normotensive controls. Our tertiary aims were to identify significant risk factors for OSA in early gestation by comparing characteristics of OSA positive and negative subjects in this cohort, as well as to evaluate the utility of several OSA-screening tools to identify OSA in this cohort enriched with subjects with cHTN: the Berlin questionnaire (BQ); the Epworth sleepiness scale (ESS); the American Society of Anesthesiologists (ASA) checklist; and an OSA-risk score developed in a cohort of gravidas by Facco et al.^21–23 We also collected pregnancy outcomes for all subjects, as well as in-laboratory polysomnography data on a subset of subjects by chart review.

Participants:

From June 2017 through March 2019, we enrolled pregnant subjects at 10–20 weeks gestation from obstetric clinics in Durham, North Carolina. Subjects were English-speaking and 18 years or older, and were assigned to one of 2 groups: 1) cHTN (on anti-hypertensive medication and/or hypertensive blood pressures [as defined by the American College of Obstetricians and Gynecologists⁷] documented on 2 clinic visits), or 2) normal blood pressure with no history of, or treatment for, cHTN, and matched for BMI (+/− 3 kg/m²) with the cHTN group. We enrolled women with cHTN first, and then the BMI-matched controls. Subjects were excluded if they were being treated for OSA, were on chronic opioids or alpha-blockers (which can interfere with the home sleep test device used in the study), had secondary hypertension, or did not speak English. Causes of secondary hypertension include diabetes prior to pregnancy > 10 years or type 1 diabetes mellitus, chronic renal disease, collagen vascular diseases, hyperthyroidism, pheochromocytoma, renal artery hyperplasia, Cushing’s syndrome, hyperaldosteronism, and maternal coarctation of the aorta.²⁴ Demographic information including race and ethnicity was collected from all subjects by self-report per institutional guidelines with the following options given: Hispanic, Latina or Spanish; Not Hispanic, Latina or Spanish; Not reported or declined; Black or African-American; White; Asian; Native Hawaiian or Pacific Islander; American Indian or Alaskan Native; and Other. No additional details were collected from subjects that reported “Other” for race, but subjects could self-report “Other” and designate a “Hispanic, Latina or Spanish” ethnicity status.

OSA risk assessments:

Following recruitment and written informed consent, subjects answered a set of questions via a secure, web-based survey created and stored in REDCap (Research Electronic Data Capture).²⁵ These questions regarded sleep quality, daytime sleepiness, self-report of snoring volume and frequency. Sleep questions included the BQ, ESS, and ASA checklist which were scored according to published methods.^21–23 The BQ contains 10 items that are classified into 3 categories: 1) snoring, 2) daytime sleepiness, and 3) BMI > 30 kg/m² and chronic hypertension. The BQ considers a person high risk for OSA if 2 of 3 categories are scored as positive. The ESS is considered concerning for excessive daytime sleepiness if scores range between 11 and 24. The ASA checklist is considered high risk for OSA if score is 2 or greater. An OSA risk score, proposed by Facco et al,¹⁸ was also calculated for all subjects. This score relies on the sum of age and BMI, in addition to 15 points for cHTN and 15 points for frequent snoring. Subjects with 75 or more points are considered high-risk for OSA.

OSA status assessment:

Subjects received an FDA-approved WP200U (Itamar Medical Ltd., Caesarea, Israel) home sleep test device and instructions for using this device during one night of sleep. This Type III, wrist-worn device uses finger plethysmography (peripheral arterial tone, oxyhemoglobin saturation, and heart rate), actigraphy (movement), acoustic decibel detection (snoring volume), and accelerometry (body position) to help diagnose sleep-related breathing disorders (including snoring), and to give information about sleep stages and position during actual sleep time. The WP200U measures changes in arterial blood volume in the fingertip as a result of cyclic sympathetic activation and subsequent peripheral vasoconstriction in people with respiratory disturbances during sleep, but does not directly measure breathing or respiratory impedance. A proprietary algorithm combines this information with heart rate acceleration, changes in oxygen saturation and actigraphy to generate a report with the following parameters: 1) sleep times and stages, 2) respiratory disturbance index (RDI), 3) apnea-hypopnea index (AHI), 4) oxygen desaturation index (ODI), 5) snoring (expressed as decibels and percent of sleep time spent snoring), 6) sleep position, and 7) average oxygen saturation and nadirs. WP200 was validated against full, in-home ambulatory polysomnography (PSG) in gravidas; the sensitivity and specificity of Watch-PAT to identify AHI > 5 events/hr on in-home, full PSG were 0.88 and 0.87 respectively.²⁶ WP200U and its predecessors (WP100 and WP200) have been used in other studies of gravidas,^27–29 and validated in many other populations with simultaneous in-laboratory PSG as detailed in a recent meta-analysis.³⁰ The meta-analysis of 14 studies that conducted simultaneous WP100 or 200 and in-lab or home PSG on 909 non-pregnant subjects showed a high correlation between WATCH-PAT and PSG for RDI and AHI across a range of ages (r = 0.893 [95% CI, 0.857–0.920; p< 0.001]).³⁰ There were hypertensive patients included in all of the studies. The WP200 was also validated in adolescents and children.^31,32

Subjects wore the device in their own home over one night of sleep. Those who were unable to sleep on the proposed night were allowed to repeat the test on a second night of sleep, and only the second night of sleep was analyzed. The home sleep test device was retrieved by research staff. An automatic computerized algorithm scored the sleep data, and generated a report that was reviewed by the principal investigator (JD). Studies were considered valid if the subjects slept for at least 2 hours. Subjects received a follow-up phone call or e-mail to give them the results of their home sleep test. Those who screened positive for OSA (defined as AHI ≥ 5 events/hour) by home sleep test or who reported symptoms of sleep disordered breathing despite AHI < 5 events/hr were referred for further evaluation and overnight in-laboratory PSG if clinically indicated. AHI ≥ 5 events/hour was chosen to define OSA because it represents the threshold for referral for in-laboratory PSG, and an OSA screening measure must be able to indicate sensitively and specifically when a referral for PSG is needed. The subject’s primary obstetric provider was notified of the finding and the need for referral. Subjects with AHI ≥ 5 events/hour were followed up by e-mail questionnaire at 5 and 20 weeks after enrollment to ascertain if sleep referrals were pursued.

Pregnancy outcomes assessment:

Pregnancy outcomes and in-laboratory PSG results, if available, were collected postpartum by chart review of discharge summaries, operative notes, and delivery notes. In-laboratory PSG reports were reviewed by the principal investigator. Due to the size of our study and rarity of many outcomes of interest, we constructed a maternal composite outcome comprised of peri-partum diagnoses (preeclampsia, eclampsia, gestational diabetes, cardiomyopathy, congestive heart failure, cerebrovascular accident) and peri-partum complications [postpartum hemorrhage (estimated blood loss ≥ 500 ml for a vaginal delivery or ≥ 1000 ml for a cesarean delivery], maternal intensive care unit (ICU) admission, post-operative wound infection, and maternal death). Peripartum diagnoses and complications were extracted postpartum from discharge summaries, operative notes and delivery notes. We similarly constructed a fetal composite outcome comprised of fetal growth restriction, preterm delivery (<37 weeks), oligohydramnios, neonatal ICU admission and fetal demise. Study data were collected and managed using REDCap electronic data capture tools hosted at Duke University.²⁵ Subjects were compensated for their participation.

Power and Sample Size:

To address the primary hypothesis we planned a sample size of 100 subjects: 50 with cHTN; and 50 normotensive controls in order to detect a 30% difference in OSA prevalence with 92% power in a conservative Fisher’s exact test based on an expected 40% prevalence of second trimester OSA (defined as AHI ≥ 5 events/hr) among subjects with cHTN from preliminary data from the work of others.^16,33 As BMI is a known confounder for OSA and cHTN, we performed a matched enrollment and interim data inspection to ensure that the cHTN and normotensive control groups were well-balanced on BMI.

Statistical Analysis:

Patient characteristics, questionnaire responses, and sleep study results were summarized according to cHTN or normotensive control status via mean (SD) or median [Q1, Q3] for continuous variables and count (%) for categorical variables. The groups were compared with t-test, Wilcoxon Rank Sum tests, Chi-Square or Fisher exact tests as appropriate. Given the higher than expected prevalence of OSA in our study, the primary hypothesis that OSA prevalence differs between gravidas with cHTN and normotensive controls, was assessed with the more powerful chi-square test rather than a Fisher’s Exact test. We pursued multivariable logistic regression to control for patient demographic factors found to be out of balance between the two groups in the univariable comparison (p<0.05). While our study design ensured BMI was balanced between groups, age was not strictly controlled for, and after study completion we found an imbalance between groups for this potential confounder. To address the impact of age in our study, in addition to multivariable regression adjusting for age, we also pursued a post-hoc sensitivity analysis among those 25 years and older based on previously accepted definition of “young adults”, there being non-overlapping age range between groups, and a statistical measure of influence (diagonal of Hat matrix).³⁴

The secondary and tertiary aims were evaluated by comparing AHI, ODI, and other sleep study characteristics between cHTN and control subjects and differences in baseline factors and screening questionnaire responses between those with and without OSA via t-test, Wilcoxon Rank Sum tests, Chi-Square or Fisher exact tests as appropriate. Statistical analysis was performed using SAS v9.4 (SAS Institute INC, Cary, NC), all tests were two-sided, and statistical significance was assessed at the 0.05 alpha level. For the secondary and tertiary analyses, we performed multiple testing correction via the family-wise error rate control Hochberg method.

RESULTS:

Of the 102 subjects enrolled, 100 subjects completed baseline OSA screening questionnaires and a valid home sleep test of 2 hours or more: 50 subjects with cHTN and 50 normotensive subjects. Two subjects did not complete a home sleep test within the gestational window, and were thus excluded from the final analysis (Figure 1). Demographic data and initial blood pressure measurements are presented in Table 1. Sixty-two percent of subjects in the cHTN group were on anti-hypertensive medication. Subjects in the cHTN group were significantly older than subjects in the normotensive group (mean±SD 34±4 vs. 30±6 years; p< 0.001). Further we found that the normotensive subjects had a disproportionate number of young subjects (those <25 years of age)³⁴; there were 13 normotensive subjects in this age group compared to only 1 cHTN subject younger than 25 years. There were no significant differences between the groups in BMI, neck circumference, gestational age, primiparous status, smoking or race/ethnicity. Home sleep test results are presented in Table 2 and Figure 2 by enrollment group. OSA was more prevalent [64% vs. 38%; p=0.009; OR (95% CI) 2.90 (1.30, 6.65), p = 0.01] (Tables 2 & 3) and more severe in the cHTN group (59.4% moderate/severe vs. 21%; p=0.009).

Figure 1. Subject flow diagram.

Two groups of gravidas were recruited: those with chronic hypertension, and a group of pregnant controls matched for body mass index and gestational age (10–20 weeks gestation).

Table 1.

Subject demographics by enrollment group

Demographics	Chronic Hypertension (N=50)	Control (N=50)	Total (N=100)	P-Value
Age (Yrs)	34.32±4.41	29.62±6.05	31.97±5.77	<0.001*
Pre-pregnancy BMI (kg/m2)	37.64±9.60	37.30±9.02	37.47±9.27	0.856*
BMI at enrollment (kg/m2)	38.22±9.31	38.00±9.17	38.11±9.19	0.906*
Neck circumference (cm) †	37.18±4.27	36.45±3.91	36.82±4.09	0.377*
Gestational Age (Wks)	15.43±2.78	15.55±2.59	15.49±2.67	0.828*
Primiparous	16 (32%)	20 (40%)	36 (36%)	0.405‡
Smoker	1 (2%)	0 (0%)	1 (1%)	>0.999§
Race/Ethnicity				0.928§
White	18 (36%)	20 (40%)	38 (38%)
Black	28 (56%)	26 (52%)	54 (54%)
Asian	0 (0%)	1 (2%)	1 (1%)
Hispanic	2 (4%)	1 (2%)	3 (3.0%)
Other	2 (4%)	2 (4%)	4 (4%)
Anti-hypertensive medications taken	31 (62%)	--	--	--
Systolic blood pressure ∥	129.82±10.73	113.34±10.08	121.58±13.26	<0.001*
Diastolic blood pressure ∥	83.62±7.02	74.32±5.72	78.97±7.90	<0.001*

Numerical data are summarized by mean±SD and categorical data are summarized by count (%).

BMI (body mass index)

^*T-test

^†Missing for 1 control subject

^‡Chi-Square test

^§Fisher test

^∥Non-invasive blood pressure was measured once during the enrollment visit.

Table 2.

Home sleep test results by enrollment group

Home sleep test results	Chronic Hypertension (N=50)	Control (N=50)	Total (N=100)	P-Value
Sleep duration (hrs)	5.90±1.33	6.32±1.43	6.11±1.39	0.128*
Total home sleep test time (hrs)	8.02±1.50	8.50±1.65	8.26±1.59	0.128*
OSA positive (AHI ≥ 5 events/hr)	32 (64%)	19 (38%)	51 (51%)	0.009‡
OSA severity ¶				0.009‡
Mild (5 ≤ AHI < 15 events/hr)	13 (41%)	15 (79%)	28 (55%)
Moderate (15 ≤ AHI < 30 events/hr)	16 (50%)	2 (11%)	18 (35%)
Severe (30 events/hr ≤ AHI)	3 (9%)	2 (11%)	5 (10%)
AHI events/hr	9.2 [2.3, 17.4]	3.3 [2.2, 6.3]	5.1 [2.3, 13.0]	0.010#
ODI events/hr	2.4 [0.5, 9.1]	0.8 [0.3, 2.0]	1.3 [0.3, 4.4]	0.007#
RDI events/hr	16.4 [9.0, 20.9]	9.6 [6.2, 13.5]	11.8 [7.2, 18.1]	0.002#
Mean oxygen saturation (%)	96 [95, 97]	96 [95, 96]	96 [95, 97]	0.625#
Mean of desaturation nadirs	93 [92, 94]	93 [92, 94]	93 [92, 94]	0.177#
Mean snoring decibels (dB)	41 [40, 42]	40 [40, 41]	40 [40, 41]	0.080#

Numerical data are summarized by mean±SD or median [Q1, Q3], and categorical data are summarized by count (%).

OSA (Obstructive sleep apnea); AHI (Apnea-hypopnea Index); ODI (oxygen desaturation index); RDI (respiratory disturbance index)

^*T-test

^†Missing for 1 control subject

^‡Chi-Square test

^§Fisher test

^¶Only relevant for OSA positive sleep tests (AHI > 5 events/hr)

^#Wilcoxon rank sum test

Figure 2. Distribution of Apnea-hypopnea Indices and Oxygen Desaturation Indices sleep study metrics between study groups.

Gravidas with chronic hypertension had significantly higher values of Apnea-Hypopnea Index (AHI) (p=0.010) and Oxygen Desaturation Index (ODI) (p=0.007) than control subjects.

Table 3.

Obstructive sleep apnea risk by chronic hypertension status and age

	Univariable		Multivariable		Multivariable Age 25+ years
Effect	OR (95% CI)	P-Value	aOR (95% CI)	P-Value	aOR (95% CI)	P-Value
Chronic hypertension vs. normotensive controls	2.90 (1.30, 6.65)	0.010	2.22 (0.92, 5.40)	0.076	2.64 (1.06, 6.71)	0.038
Age	1.09 (1.02, 1.18)	0.017	1.07 (0.99, 1.16)	0.119	1.08 (0.98, 1.20)	0.142

A multivariable analysis was conducted to study the association between age and cHTN in relation to OSA status since age differed significantly between groups (Table 3). In the multivariable analysis that controlled for the effect of age we did not observe a significant association between cHTN and OSA in the overall cohort [aOR (95% CI) 2.22 (0.92, 5.40), p = 0.076]. However, we did a post-hoc multivariable analysis in which we excluded all young subjects (<25 years) (n = 14) and found that in this subgroup, cHTN was associated with higher odds of positive OSA status after controlling for age [aOR (95% CI) 2.64 (1.06, 6.71), p = 0.038].

Results of OSA screening questionnaires and risk scores according to OSA status are presented in Table 4. OSA positive gravidas had higher pre-pregnancy and pregnancy BMI, larger neck circumference, as well as higher scores on the Facco et al. risk score.¹⁸ The Facco et al. score was ≥ 75 for 86% of OSA positive subjects, and 45% of OSA negative subjects (PPV = 0.66, NPV = 0.79). There was no difference in gestational age, race/ethnicity, BQ, ESS or ASA checklist scores between OSA positive vs. negative subjects.

Table 4.

Subject characteristics and questionnaire scores according to OSA status

Characteristics and questionnaires	OSA Positive (N=51)	OSA Negative (N=49)	P-Value	Corrected p-value
Age (Yrs)	33.35±5.19	30.53±6.04	0.012*	0.156
Pre-pregnancy BMI (kg/m2)	41.18±8.74	33.61±8.24	<0.001†	0.002
BMI at enrollment (kg/m2)	41.71±8.64	34.37±8.28	<0.001†	0.002
Neck circumference (cm) ‡	38.15±4.19	35.41±3.50	0.001†	0.002
Gestational Age (Wks)	15.00 [13.29, 17.71]	15.43 [13.29, 18.00]	0.514*	0.999
Primiparous	17 (33%)	19 (39%)	0.571§	0.999
Smoker	1 (2%)	0 (0%)	>0.999∥	>0.999
Race/Ethnicity			0.479§	0.999
White	16 (31%)	22 (45%)
Black	31 (61%)	23 (47%)
Asian	1 (2%)	0 (0%)
Hispanic	1 (2%)	2 (4%)
Other	2 (4%)	2 (4%)
Chronic Hypertension	32 (63%)	18 (37%)	0.009§	0.126
Anti-hypertensive medications taken	21 (41%)	10 (20%)	0.025§	0.275
Systolic blood pressure ¶	123.39±13.33	119.69±13.06	0.165†	0.999
Diastolic blood pressure ¶	80.73±6.80	77.14±8.60	0.023†	0.275
Berlin total score	2 [1, 2]	2 [1, 2]	0.092*	0.828
Berlin high risk #	36 (71%)	31 (63%)	0.436§	0.999
Epworth total score	2 [0, 3]	2 [1, 4]	0.200*	0.999
Facco et al. score	96 [82, 108]	74 [66, 83]	<0.001*	0.002
Facco et al. high risk (score > 75)	44 (86%)	22 (45%)	<0.001§	0.002
ASA checklist score	2 [1, 2]	2 [1, 2]	0.078*	0.780
ASA high risk **	37 (73%)	29 (59%)	0.158§	0.999

Numerical data are summarized by mean±SD or median [Q1, Q3], and categorical data are summarized by count (%).

OSA (Obstructive Sleep Apnea); BMI (body mass index); ASA (American Society of Anesthesiologists)

^*Wilcoxon test

^†Equal Variance T-Test

^‡Missing for 1 OSA negative subject.

^§Chi-Square

^∥Fisher Exact

^¶Non-invasive blood pressure was measured once during the enrollment visit.

^#Berlin questionnaire high risk for OSA if ≥ 2 of 3 categories scored positive.

^**ASA checklist high risk for OSA if score ≥ 2.

Pregnancy outcomes for 97 subjects are reported by OSA status (Supplementary Table 1); a composite maternal and neonatal outcome is reported. The composite maternal outcome was not significant between OSA positive and negative subjects (48% vs. 36%, p = 0.239). There were significantly more adverse composite neonatal outcomes in the OSA positive group compared to the OSA negative group (46% vs. 21%, p = 0.01), although composite neonatal outcome did not remain statistically significant when the p-value was corrected for multiple comparisons (corrected p = 0.103).

Follow-up with a sleep medicine specialist was recommended to all OSA-positive subjects (n=51, AHI ≥ 5 events/hr on home sleep test). Fifteen of the 51 subjects were evaluated by a sleep medicine specialist at our institution; in-laboratory PSG was recommended for 14/15 subjects. Further testing was not recommended for one woman with mild OSA (AHI = 7.2 events/hr on home sleep test) and no symptoms. Twelve of 14 subjects for whom in-laboratory PSG was recommended completed this testing. Results of the home sleep test and the in-laboratory PSG and the time interval between the two studies are presented in (Supplementary Table 2). Sleep medicine referral was recommended to one woman who was symptomatic despite a normal home sleep test (AHI = 0.3 events/hr), and she was found to have severe OSA by in-laboratory PSG at 32 weeks gestation.

DISCUSSION:

In this prospective cohort study, we found that early gestation gravidas with cHTN have a higher rate of undiagnosed OSA compared to BMI- and gestational age-matched normotensive gravidas. We also found that OSA is more severe among gravidas with cHTN in early gestation. Our results suggest that among gravidas 25 years and older cHTN is independently associated with a higher odds of OSA. The BQ, ESS and ASA checklist did not differentiate between subjects with and without OSA in this cohort. This is consistent with previous studies.^17–19 We evaluated the OSA scoring system proposed by Facco et al, but it was limited by a 45% false positive rate.

While previous studies have reported a higher prevalence of OSA in gravidas with cHTN, none have controlled for the effect of obesity on this relationship.^15–18,35 The NuMom2b sleep-disordered breathing substudy, which conducted prospective home sleep tests on over 3700 nulliparous gravidas, found that cHTN was strongly associated with OSA in early and mid-pregnancy.⁸ However, cHTN was not included in the final prediction model generated from this cohort because it did not improve the model that included age, BMI and endorsement of frequent snoring.³⁶ This cohort was also younger and of lower mean BMI than our cohort. We also found age to be associated with OSA in pregnancy. However, after we controlled for cHTN, the effect of age on OSA risk in pregnancy was not significant in either the overall cohort or in the subset of patients 25 years and older.

OSA has also been associated with cHTN in non-pregnant adults, and treatment of OSA with positive airway pressure has been shown to reduce blood pressure.³⁷ Some have hypothesized that intermittent hypoxemia may trigger a signaling cascade that stimulates the sympathetic nervous system, causes endothelial dysfunction and ultimately cHTN.³⁸ These pathways may also connect OSA and preeclampsia, but the pathogenesis of these conditions remains largely unknown.

None of the OSA screening tools showed evidence of clinical utility. The OSA scoring system proposed by Facco et al. that includes age, BMI, cHTN and frequent snoring had a PPV = 0.66, and a NPV = 0.79. Our previous study in gravidas with class III obesity suggested that the Facco et al. risk score is highly sensitive, but not specific for OSA in this population because of the large contribution of BMI to the score.²⁰ A reliable screening tool for OSA in pregnancy has not yet been developed and validated in a non-related cohort and has challenged routine OSA screening in prenatal clinics. However, based on the findings of our study, we recommend assessing any patient with cHTN in pregnancy for symptoms of OSA, particularly if co-morbid with advanced maternal age and/or class 3 obesity and referring patients that are symptomatic for further sleep evaluation.

The study was not powered to detect differences in maternal and neonatal outcomes between gravidas with and without cHTN and/or OSA. When we analyzed a composite neonatal outcome, there were significantly more adverse outcomes among the neonates of OSA positive subjects compared to those of OSA negative subjects although composite neonatal outcome did not remain statistically significant when the p-value was corrected for multiple comparisons. This agrees with data from a large retrospective database study, and a meta-analysis.^11,39,40 The mechanisms that may connect maternal OSA with adverse neonatal outcomes are largely unknown, but may relate to the greater burden of co-morbidities in this patient population, namely preeclampsia, which is a risk factor for preterm birth and gestational diabetes. This is an area that warrants further investigation.

While overnight in-laboratory PSG is the gold standard for diagnosing OSA, there can be practical limitations to obtaining PSG during pregnancy. In particular, the feasibility of in-laboratory PSG can be challenging for gravidas and those with caregiving responsibilities, and may discourage them from pursuing diagnostic testing. In this study, only 25% of subjects who were referred for sleep medicine consultation underwent in-laboratory PSG and evaluation for treatment. Those who did waited an average of two months after referral. PSG studies supported the diagnosis of OSA suggested by the home sleep test results in the majority of subjects (11/12 subjects).

A strength of this study is the enrollment of a BMI- and gestational age-matched control group to isolate the impact of cHTN on OSA risk, and a very low drop-out rate. The device used in this study (WP200U™; Itamar Medical Ltd, Caesarea, Israel) was well-tolerated by subjects; this is significant as nighttime discomfort and frequent awakening can be issues for gravidas and have been limitations of previous studies.²⁰ By using actigraphy to estimate actual sleep time, this device overcomes one of the limitations of other unattended home sleep test devices, which is that they may overestimate sleep time (denominator of AHI measurement) and underestimate the AHI. This is particularly crucial in gravidas who may awaken several times during the night.

Our study had the following limitations: a higher decline rate among subjects approached for the cHTN group compared to the control group; insufficient sample size to study interactions between cHTN and baseline factors; not using the gold-standard PSG to diagnose OSA; not repeating OSA tests later in pregnancy; and limiting enrollment to English speaking women, which may limit the generalizability of our findings. We approached 67 gravidas to enroll 51 subjects with cHTN (23.8% decline rate) compared to a 10.5% decline rate among the potential controls. The reasons for the higher decline rate among gravidas with cHTN are unclear, and may have biased our results towards including more subjects that suspected they had sleep disordered breathing and thus were more motivated to participate. Gravidas with cHTN may have had more complicated pregnancies that required more medical appointments, and they may have been less inclined to take on the additional burden of participating in research. The sample size of 100 provided sufficient power for addressing the primary and secondary hypotheses but did not provide adequate power to control for the potential confounding effect of age on the relationship between cHTN and OSA. Further our enrollment irrespective of age resulted in a severely imbalanced rate of young subjects (< 25 years of age) in the cHTN group (1 vs. 13), which limited our ability to analyze the full age range.³⁴ We did conduct a secondary multivariable analysis in which we excluded those subjects younger than 25 years old to better understand the role of cHTN and age given the outsized influence on the model estimation of the youngest subjects. While the WP200U does not use direct measures of breathing to calculate apneas and hypopneas, we were able to demonstrate consistency between WP200U and in-laboratory PSG results in a subset of subjects. Study subjects underwent home sleep testing in early pregnancy (10–20 weeks gestation); many of the women who tested positive in this cohort may have had OSA prior to pregnancy. However, evidence from other studies suggests that, at later gestation, we would have found a higher prevalence of OSA.^8,33 Further studies are needed to assess the progression of OSA during and after pregnancy.

In summary, we found that the rate of OSA among gravidas with cHTN in early gestation is nearly twice that of normotensive gravidas matched for BMI and gestational age. Our results suggest that older gravidas with cHTN are at greater risk of OSA, compared to older normotensive gravidas, but future studies are needed to further study this relationship in larger cohorts of high-risk gravidas.

Key Points:

Question:

Is obstructive sleep apnea (OSA) more prevalent among gravidas with chronic hypertension (cHTN) compared to normotensive controls matched for body mass index (BMI) and gestational age, and is OSA more severe in gravidas with chronic hypertension compared to controls?

Findings:

OSA is nearly twice as prevalent and more severe among gravidas with chronic hypertension in early gestation compared to BMI-matched controls; age appears to be an important variable in this relationship.

Meaning:

Chronic hypertension and age are important risk factors for OSA in gravidas after controlling for BMI.

Glossary of Terms:

ASA

American Society of Anesthesiologists

AHI

apnea-hypopnea index

BQ

Berlin questionnaire

BMI

body mass index

cHTN

chronic hypertension

ESS

Epworth sleepiness scale

ICU

intensive care unit

ODI

oxygen desaturation index

OSA

obstructive sleep apnea

PSG

polysomnography

RDI

respiratory disturbance index