Maternal obstructive sleep apnea (OSA) is associated with increased rates of chronic hypertension, preeclampsia, diabetes, depression, asthma, preterm birth,1 reduced fetal growth,2 markers of fetal distress,3 and cesarean delivery.4 Obstructive sleep apnea is characterized by inflammation, sympathetic activation, and oxidative stress.5,6 These factors in conjunction with the physiologic demands of pregnancy are thought to be responsible for the increased perinatal and maternal morbidity seen in affected pregnancies. Few pregnant women are referred for formal sleep evaluation, although snoring, a prominent symptom of OSA, is more common in pregnant than in nonpregnant young women.7 It is likely that OSA and other types of sleep-disordered breathing are underdiagnosed during pregnancy. Given that OSA is a risk factor for maternal and perinatal morbidity, the identification of patients at risk who require further evaluation and possibly intervention is an important consideration.
There are many screening tools that are highly sensitive and specific for the identification of nonpregnant patients at high risk for OSA, including snoring, tiredness, observed apnea and blood pressure (STOP),8 STOP-BANG,8,9 the American Society of Anesthesiologists checklist,10 Flemons Index,11 Berlin,10,12 and the Epworth Sleepiness Scale.13 However, none have been validated in the obstetric population. The goal of this project was to estimate the predictive ability of current OSA screening tools and individual questions in these tools for identifying pregnant patients with OSA as detected by ResMed ApneaLink (sleep monitor). A secondary aim was to assess the sensitivity and specificity of particular comorbidities at predicting patients who have OSA.
MATERIALS AND METHODS
This prospective study targeted women 27 weeks of gestation and greater between 2010 and 2012. We recruited volunteer participants either from outpatient clinics or the inpatient antepartum obstetric service at Barnes-Jewish Hospital in St. Louis.
Patients with chronic pulmonary disease, on opioids, or in whom delivery was imminent or indicated were excluded. Based on a sensitivity of 0.88, a precision of 0.09, an OSA prevalence rate of 20% in our pilot data, a type I error of 0.05, and a power of 0.8, the sample size of 208 was calculated. A sample size of 250 patients was chosen to account for potential associations with severity of OSA.
Study participants completed a questionnaire consisting of six OSA screening tools (Berlin, American Society of Anesthesiologists checklist, STOP, STOP-BANG [STOP plus body mass index (BMI), age, neck circumference, and gender], Flemons Index, and Epworth Sleepiness Scale). Body mass index was calculated as weight (kg)/[height (m)]2. The 39-question survey included demographic information, medical comorbidities, and evaluations by the examiner (BMI, neck circumference, tonsillar examination, and presence of craniofacial abnormalities). The data were collected by trained research assistants who were part of the research team in the Department of Anesthesiology. The collected data were then reviewed for quality by the project coordinator. Before analyzing the data, 5% of the cases were randomly selected in addition to cases with outlier information for validation. Less than 1% of the data needed to be corrected or converted into missing values after medical record review. No values were imputed. The study was approved by the Human Research Protection Office of Washington University. All appropriate safety and medical privacy measures as described by federal law under the Health Insurance Portability and Accountability Act were taken. Information was entered into REDCap (Research Electronic Data Capture)14 software and exported into Microsoft Excel. Additionally, patients were asked to wear a sleep monitor overnight. This device records respiratory pressure, pulse rate, and oxygen saturation using a finger pulse oximetry probe and a nasal cannula with no oxygen flow. At least 2 hours of assessment time is required for accurate measurement. The sleep monitor software uses standard measurements and definitions of the American Academy of Sleep Medicine. The sleep monitor results were first used to categorize the study sample into two groups based on each patient's apnea–hypopnea index. Apnea–hypopnea index is defined as total number of apneas (reduction of airflow to 0–20% lasting 10 seconds or longer) and hypopneas (reduction of airflow to 50% lasting 10 seconds or longer) occurring per hour of sleep. These pauses must be associated with a decrease in oxygen saturation.
Per the American Academy of Sleep Medicine, we defined OSA-negative as apnea–hypopnea index less than 5 and OSA-positive as apnea–hypopnea index 5 or greater. Differences between the two groups for each of the evaluation measures and for patient demographics as well as baseline health characteristics were then assessed using χ² test for all categorical variables and Student's and Fisher's t test for the parametric variables and the Mann-Whitney U test for the nonparametric variables. Receiver operating characteristic (ROC) curves were constructed for each OSA screening tool, and the overall predictive ability for OSA was estimated using the area under ROC curves.
Further predictive statistics including sensitivity, specificity, positive and negative predictive values, and κ for each screening tool as compared with the patient's apnea–hypopnea index score were calculated and compared across instruments. Kappa values from 0.61 to 1.00 are considered substantial to excellent agreement; values between 0.41 and 0.60 represent moderate agreement; values between 0.21 and 0.40 indicate fair agreement; and 0 and 0.20 as slight or poor agreement.15 Pairwise comparisons of ROC curves were performed using z-statistics. A deconstruction analysis was performed to estimate the contribution of the various elements in each screening tool. Association between those significant elements and patient apnea–hypopnea index score was estimated using logistic regression. Additionally, all selected OSA screening tool items were then tested using a multivariate prediction model. A two-sided P value of ≤.05 was considered statistically significant for all statistical tests performed. The screening tool deconstruction was assessed using a simple test of association and logistic regression. All analyses were conducted using SAS 9.2.
A total of 293 patients were enrolled in the study, and 248 patients had sleep monitor results. Forty-five of the patients either chose not to wear the monitor or did not wear it long enough. Two hundred eighteen (88%) of the patients were OSA-negative and 30 patients (12%) were OSA-positive. The average age of the cohort was 28 years, and 76% were multiparous. There were 127 (54%) African Americans and 122 (43%) Caucasians. The mean gestational age at recruitment was 32 weeks. The average prepregnancy weight was 70 kg and 81 kg at the time of sleep study with an average pregnancy weight gain of 12 kg. The median BMI was 31, and 31% had BMIs greater than 35. The median neck circumference was 36 cm.
Table 1 shows results of the sleep evaluation. Two hundred eighteen (88%) of the patients were OSA-negative and 30 patients (12%) were OSA-positive. Obstructive sleep apnea–positive patients had significantly higher numbers of desaturations, apneas, hypopneas, and snoring events. Of those patients who were OSA-positive, 57% had mild OSA, 23% had moderate OSA, and 20% had severe OSA based on the apnea–hypopnea index (apnea–hypopnea index 30 or greater).
Baseline patient characteristics are shown in Table 2. Obstructive sleep apnea–positive patients had greater BMIs both before pregnancy and as measured at the time of sleep study. Neck circumference was significantly higher in the OSA-positive group. Obstructive sleep apnea–positive patients also had significantly higher rates of preeclampsia, hypertension (chronic and gestational), chronic diabetes mellitus, and asthma. There were no differences is tobacco use, alcohol, or drug use.
The OSA screening tools results were compared with the apnea–hypopnea index scores (Tables 3 and 4). Kappa values range from fair (0.3) to poor (0.1) demonstrating poor agreement between the screening tools and with the exception of the STOP-BANG where the negative predictive values were much higher than the positive values, indicating that this tool is good at identifying those without OSA.
The predictive ability for OSA was compared using ROC curves (Fig. 1). The overall predictive abilities of the screening tools for OSA were modest and not significantly different between the various tools. Table 5 deconstructs the screening tools into individual elements to examine their performance at detecting patients who are OSA-positive. In general, items aimed at assessing fatigue or tiredness were unsuccessful in differentiating OSA-negative and OSA-positive patients with the exception being “falling asleep while talking to someone.” There were multiple questions about snoring; all were significantly associated with OSA, particularly loud and frequent snoring. Other questions that were able to differentiate between OSA-negative and OSA-positive patients include cessation of breathing during sleep, BMI, neck circumference greater than 40 cm, awakening from sleep with choking sensation, and frequent arousals from sleep. These items were validated in a multivariate prediction model as shown in Table 6. Body mass index greater than 35, history of dozing off while talking to someone, and history of treatment for hypertension remained significant with an overall model fit of 0.83.
Obstructive sleep apnea is associated with maternal cardiovascular morbidity and in-hospital death,1 low birth rate, and preterm delivery.4 Treatment of OSA in the nonobstetric population has clear benefits including improvements in systemic hypertension, ventricular function, and management of noninsulin-dependent diabetes mellitus.16 In the obstetric population, the treatment of OSA positively affects hemodynamic and biophysical profiles of patients with OSA and chronic hypertension.17–20 Therefore, identification of at-risk patients is important. Given the potential burden of testing, an accurate and easy-to-administer screening tool for pregnant patients has a role in the care of obstetric patients.
None of the popular OSA screening tools was successful at detecting OSA in the third trimester. This finding is consistent with recent reports. The Berlin and Epworth Sleepiness Scale were poorly predictive of OSA in pregnant women and was associated with a high false referral rate for polysomnography.21 Facco et al22 found that the Berlin and Epworth Sleepiness Scale did not accurately predict OSA in high-risk pregnant women. In contrast to the modest performance of the screening tools, individual components of the questionnaire were strongly associated with OSA. This suggests that a new tool consisting of these components may have improved accuracy. Based on our univariate analysis, the following elements were significant: objective measurements (BMI, neck circumference), medical comorbidities (treatment for high blood pressure, diabetes, asthma), and symptomatic components (snoring, stop breathing during sleep, fall asleep while sitting and talking with someone, awaken from sleep with choking sensation, frequent arousals from sleep). Further multivariate analysis found that only BMI greater than 35, falling asleep while talking with someone, and history of treatment for hypertension remained significant.
None of the investigated screening tools have been validated in the obstetric population and are problematic in several aspects. Male gender is clearly not relevant. The inclusion of age as a risk factor may well be appropriate because maternal age is a risk factor for third-trimester OSA among women without baseline sleep-disordered breathing23; however, age older than 50 years is clearly not applicable. Tiredness and daytime sleepiness are common complaints of pregnancy, yet these questions are heavily represented on some of the screening tools. Our results demonstrated that only “falling asleep while sitting and talking with someone” was a predictor of OSA. Snoring, a principle symptom of OSA, affects up to one third of women by the third trimester24,25 and was not significant in our multivariate analysis.
Objective measurements and comorbidities are also part of many screening tools. Neck circumference greater than 40 cm and BMI greater than 35 were able to distinguish between OSA-positive and OSA-negative patients but only BMI remained in our final model. High BMI is clearly a risk factor, but the cutoff range for this population remains to be determined.
This study has several limitations. We only studied patients in the third trimester of pregnancy (greater than 27 weeks of gestation). It is likely that the incidence and magnitude of OSA change during pregnancy, so studying a single trimester allows us to draw stronger conclusions. However, we were unable to study several patients with symptoms strongly suggestive of severe OSA or who had a diagnosis of OSA because they were delivered before 27 weeks of gestation. Another limitation is that our pool of OSA-positive patients was smaller than expected; only 30 patients (12%) were OSA-positive. A third limitation is that we utilized a portable monitor and not polysomnography. Portable monitors are unable to monitor electroencephalography, electrocardiography, or observe sleep. However, studies in nonpregnant patients have demonstrated strong correlations (r=0.89 and r=0.978)26,27 with high sensitivity and specificity for the apnea–hypopnea index derived from this particular device and that obtained by polysomnography.28 This is the means by which we identify patients as OSA-positive or OSA-negative. This study is unable to address when or how often OSA screening should be performed during pregnancy. In a study of pregnant women who underwent first- and third-trimester polysomnography, 10.5% of women in their first-trimester and 26.7% of women in their third-trimester had OSA.23 In addition, appropriate cutoff values may differ in the first and third trimesters. Early testing seems preferable if the goal is to evaluate and treat to improve outcomes. However, testing too early may miss patients who develop OSA during pregnancy.
In conclusion, our results demonstrate that none of the established OSA screening tools are able to identify OSA-positive patients in their third trimester. Future directions include establishing reliability and validity of this tool and understanding how this screening tool functions throughout pregnancy.
1. Louis JM, Mogos MF, Salemi JL, Redline S, Salihu HM. Obstructive sleep apnea and severe maternal-infant morbidity/mortality in the United States, 1998–2009. Sleep 2014;37:843–9.
2. Fung AM, Wilson DL, Lappas M, Howard M, Barnes M, O'Donoghue F, et al.. Effects of maternal obstructive sleep apnoea on fetal growth: a prospective cohort study. PLoS One 2013;8:e68057.
3. Bourjeily G, Butterfield K, Curran P, Messerlian G. Obstructive sleep apnea is associated with alterations in markers of fetoplacental wellbeing. J Matern Fetal Neonatal Med 2015;28:262–6.
4. Chen YH, Kang JH, Lin CC, Wang IT, Keller JJ, Lin HC. Obstructive sleep apnea and the risk of adverse pregnancy outcomes. Am J Obstet Gynecol 2012;206:136.e1–5.
5. Iturriaga R, Moya EA, Del Rio R. Inflammation and oxidative stress during intermittent hypoxia: the impact on chemoreception. Exp Physiol 2015;100:149–55.
6. Mills PJ, Kennedy BP, Loredo JS, Dimsdale JE, Ziegler MG. Effects of nasal continuous positive airway pressure and oxygen supplementation on norepinephrine kinetics and cardiovascular responses in obstructive sleep apnea. J Appl Physiol (1985) 2006;100:343–8.
7. Venkata C, Venkateshiah SB. Sleep-disordered breathing during pregnancy. J Am Board Fam Med 2009;22:158–68.
8. Chung F, Yegneswaran B, Liao P, Chung SA, Vairavanathan S, Islam S, et al.. STOP questionnaire: a tool to screen patients for obstructive sleep apnea. Anesthesiology 2008;108:812–21.
9. Chung F, Subramanyam R, Liao P, Sasaki E, Shapiro C, Sun Y. High STOP-Bang score indicates a high probability of obstructive sleep apnoea. Br J Anaesth 2012;108:768–75.
10. Chung F, Yegneswaran B, Liao P, Chung SA, Vairavanathan S, Islam S, et al.. Validation of the Berlin questionnaire and American Society of Anesthesiologists checklist as screening tools for obstructive sleep apnea in surgical patients. Anesthesiology 2008;108:822–30.
11. Finkel KJ, Searleman AC, Tymkew H, Tanaka CY, Saager L, Safer-Zadeh E, et al.. Prevalence of undiagnosed obstructive sleep apnea among adult surgical patients in an academic medical center. Sleep Med 2009;10:753–8.
12. Ahmadi N, Chung SA, Gibbs A, Shapiro CM. The Berlin questionnaire for sleep apnea in a sleep clinic population: relationship to polysomnographic measurement of respiratory disturbance. Sleep Breath 2008;12:39–45.
13. Johns MW. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep 1991;14:540–5.
14. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2008;42:377–81.
15. Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assess 1994;6:284–6.
16. Zhong W, Tang YG, Zhao X, Go FY, Harper RM, Hui H. Treating obstructive sleep apnea with continuous positive airway pressure benefits type 2 diabetes management. Pancreas 2007;43:325–30.
17. Poyares D, Guilleminault C, Hachul H, Fujita L, Takaoka S, Tufik S, et al.. Pre-eclampsia and nasal CPAP: part 2. Hypertension during pregnancy, chronic snoring, and early nasal CPAP intervention. Sleep Med 2007;9:15–21.
18. Edwards N, Blyton DM, Kirjavainen T, Kesby GJ, Sullivan CE. Nasal continuous positive airway pressure reduces sleep-induced blood pressure increments in preeclampsia. Am J Respir Crit Care Med 2000;162:252–7.
19. Blyton DM, Sullivan CE, Edwards N. Reduced nocturnal cardiac output associated with preeclampsia is minimized with the use of nocturnal nasal CPAP. Sleep 2004;27:79–84.
20. Blyton DM, Skilton MR, Edwards N, Hennessy A, Celermajer DS, Sullivan CE. Treatment of sleep disordered breathing reverses low fetal activity levels in preeclampsia. Sleep 2014;36:15–21.
21. Antony KM, Agrawal A, Arndt ME, Murphy AM, Alapat PM, Guntupalli KK, et al.. Obstructive sleep apnea in pregnancy: reliability of prevalence and prediction estimates. J Perinatol 2014;34:587–93.
22. Facco FL, Ouyang DW, Zee PC, Grobman WA. Development of a pregnancy-specific screening tool for sleep apnea. J Clin Sleep Med 2012;8:389–94.
23. Pien GW, Pack AI, Jackson N, Maislin G, Macones GA, Schwab RJ. Risk factors for sleep-disordered breathing in pregnancy. Thorax 2014;69:371–7.
24. O'Brien LM, Bullough AS, Owusu JT, Tremblay KA, Brincat CA, Chames MC, et al.. Snoring during pregnancy and delivery outcomes: a cohort study. Sleep 2013;36:1625–32.
25. Bourjeily G, Raker CA, Chalhoub M, Miller MA. Pregnancy and fetal outcomes of symptoms of sleep-disordered breathing. Eur Respir J 2010;36:849–55.
26. Erman MK, Stewart D, Einhorn D, Gordon N, Casal E. Validation of the ApneaLink for the screening of sleep apnea: a novel and simple single-channel recording device. J Clin Sleep Med 2007;3:387–92.
27. Ng SS, Chan TO, To KW, Ngai J, Tung A, Ko FW, et al.. Validation of a portable recording device (ApneaLink) for identifying patients with suspected obstructive sleep apnoea syndrome. Intern Med J 2009;39:757–62.
© 2015 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
28. Oktay B, Rice TB, Atwood CW Jr, Passero M Jr, Gupta N, Givelber R, et al.. Evaluation of a single-channel portable monitor for the diagnosis of obstructive sleep apnea. J Clin Sleep Med 2011;7:384–90.