Objective
Obstructive sleep apnea (OSA) involves episodic nocturnal apneas. Using polysomnography, we examined the predictive capacity of screening questionnaires (Berlin) in pregnancy. Incorporating simultaneous fetal heart rate monitoring (FHM), we examined the association of maternal apnea with FHM abnormalities.
Study Design
We enrolled 100 pregnant women at 26-39 weeks of gestation with OSA screening and baseline data ascertainment who underwent polysomnography and FHM for ≥3 hours. The relationship between maternal characteristics, OSA, and FHM was explored with multivariate analyses that were controlled for potential confounders.
Results
When compared with polysomnography, sensitivity and specificity by Berlin screening was 35% and 63.8%, respectively; the snoring component of the Berlin correlated better with oxygen desaturation <95% ( P = .003). Body mass index was a significant confounder (r s = 0.44; P < .0001). No association was observed between FHM abnormalities and OSA parameters.
Conclusion
In pregnancy, the Berlin questionnaire poorly predicts OSA. It is unclear whether fetal compromise during maternal apnea is a mechanism in OSA that is related to pregnancy outcome.
Obstructive sleep apnea (OSA) is characterized by episodes of airflow limitation that cause intermittent hypoxia. Studies have observed that in nonpregnant patients, it is an independent risk factor for hypertension, coronary artery disease, and atherosclerosis. Although the true prevalence rate in pregnancy is unknown, many physiologic changes contribute to increased risk for OSA. To date, few studies have investigated OSA in pregnancy, and most studies have failed to adjust for potential maternal confounders.
The diagnosis of OSA is established by polysomnography, but time and expense limitations have lead to the development of several validated screening tools, which includes the Berlin questionnaire. Although the Berlin questionnaire has been shown to have a positive predictive value as high as 89%, recent analyses suggest the predictive performance of the questionnaire may be quite variable; the sensitivity and specificity range from 57-86% and 43-97%, respectively.
Among pregnant women, snoring, which is a risk factor for OSA, increases through latter gestation; although not all “snorers” have OSA, it has been associated with adverse pregnancy outcomes, intrauterine growth restriction, and preeclampsia. Furthermore, maternal apnea episodes have been associated with fetal heart rate decelerations that may be a contributing factor to documented adverse pregnancy outcomes.
We hypothesized that the Berlin questionnaire is a valid tool for the screening of OSA in pregnancy, when compared with the gold standard, polysomnography. We therefore sought to investigate the performance of the Berlin questionnaire among pregnant women in a large, prospectively acquired cohort. We also sought to investigate the suggested association between sleep-related maternal apnea events and changes in fetal oxygenation status, as measured by fetal heart rate monitoring (FHM).
Methods
Basic study design
Institutional review board approval was obtained from both Baylor College of Medicine and the Harris County Hospital District. All pregnant women during the 9-month study interval who were admitted to the antepartum service at the Ben Taub General Hospital were approached for participation in the study. Enrollment was halted when we reached an initial cohort of 100 women. Inclusion criteria included singleton pregnancy with unrelated condition for antepartum admission and gestational age ≥26 weeks by best obstetric estimate (with at least 1 confirmatory sonogram). Exclusion criteria consisted of hospital stay <4 hours or immediate delivery within 4 hours, multifetal gestation, known or suspected fetal growth restriction <10%, multiple fetal anomalies or death, known severe cardiopulmonary disease, and known OSA.
Once consented, the patient completed the Berlin questionnaire in their native language. The Berlin is a well-validated screening tool for OSA among nonpregnant subjects that stratifies patients into low-risk and high-risk categories for OSA. The Berlin questionnaire uses 10 self-administered questions at an 8th-grade equivalent regarding risk factors that focus on characteristics of snoring, sleepiness throughout the day, and presence of hypertension and increased body mass index (BMI). A questionnaire for a high-risk patient will have 2 of 3 symptom categories positive. In this study, we considered a high-risk Berlin questionnaire score to be evidence of symptom-diagnosed OSA. After completion of the questionnaire, patients were monitored with simultaneous and time-synched FHM and polysomnography, as described further later ( Figure ).
Maternal data collection
At time of enrollment, extracted maternal baseline data included maternal age, ethnicity, height, prepregnancy weight, and pregnancy weight. Of note, prepregnancy weight was asked of the patient and then confirmed with the medical record. Maternal data also included gestational age, smoking history, gravidity, parity, abortion history, pregestational diabetes mellitus, history of thyroid disease, chronic hypertension, previously diagnosed sleep disorder, gestational diabetes mellitus, gestational hypertension, preeclampsia, preterm labor, preterm premature rupture of membranes, urinary tract infection, abruption, hydramnios, and intrauterine growth restriction. In all instances, standard clinical definitions based on the American College of Obstetrics and Gynecology practice guide lines were used for diagnosis.
BMI classification
Each woman was stratified into BMI category according to the International Obesity Task Force classification: underweight, <19 kg/m 2 ; normal weight, 19-24.9 kg/m 2 ; overweight, 25-29.9 kg/m 2 ; class I obesity, 30-34.9 kg/m 2 ; class II obesity, 35-39.9 kg/m 2 ; and class III obesity, >40 kg/m 2 . BMI was calculated using height and weight data (kilograms/square meter) that were collected during initial assessment and during postpartum hospitalization. For analyses of prepregnancy BMI, we defined BMI as 19-24.9 kg/m 2 as normal, when compared with those women with a BMI >25 kg/m 2 . However, the reference “normal” range for analyses of BMI at delivery was 19–29.9 kg/m 2 (normal and overweight). This acknowledges that a normal-weight woman who gains up to the recommended (Institute of Medicine) amount during pregnancy will have a body mass index of ≤29.9.
Simultaneous polysomnography and FHM monitoring
The polysomnography device that was used in the study was the ResMed ApneaLink (ResMed Corp, San Diego, CA), which records continuous pulse oxymetry and nasal airflow. Standard external FHM that was in use at Ben Taub General Hospital was used along with the WatchChild system (Hill-Rom, Batesville, IN) to record fetal heart rate.
Once placed on FHM, the polysomnography monitor was prepared and placed on the patient. The polysomnography monitor was then started, and the start time was electronically marked on the fetal heart rate tracing strip and recorded by the WatchChild system. Patients were monitored at night for 3-6 hours (dependent on maternal sleeping behavior), and the study was interrupted only if the patient experienced intense discomfort with the monitors or if the patient spontaneously awoke. Occasionally (<3 occurrences/subject), the patient had to be awakened for repositioning of the devices. At the conclusion of the study, the devices were removed, and the fetal heart rate strip was printed from the WatchChild system. In addition, the polysomnography data were downloaded onto the study computer that generated a comprehensive sleep study report.
FHM data
After the data were collected, the FHM tracings were reviewed independently by 3 maternal-fetal medicine faculty members at Baylor College of Medicine who analyzed the tracings and noted the following information as previously described: severe variables (<70 beats/min; >60 seconds), late decelerations (≥1 deceleration/30 minutes), bradycardia (<110 beats/min), tachycardia (>180 beats/min), poor long-term variability (variable decelerations with poor beat to beat), variable decelerations with tachycardia, recurrent prolonged variable decelerations (>2 decelerations of <70 beats/min for >90 seconds in 15 minutes), sinusoidal pattern (long-term variability frequency, 2-5 cycles/minute), ≥1 variable decelerations in 2 consecutive 30-minute windows, increased variability (>25 beats/min for at least 30 minutes), baseline fetal heart rate <100 beats/min with accelerations, baseline fetal heart rate between 100 and 200 beats/min without accelerations, and suspected fetal arrhythmia. Consensus was reached with a fourth independent reviewer when 2 of 3 maternal-fetal medicine physicians noted the same abnormality during the same time interval. There were complete tracings on the 100 participants and no missing data, because the monitors were adjusted to maintain fetal tracing throughout the sleep study interval.
Polysomnography data
Two pulmonologists independently interpreted the polysomnography data and verified the sleep study report. Polysomnography data were validated by the number of apnea events and hypopnea events per hour (Apnea-Hypopnea Index [AHI]) and used an a priori determination of the AHI of ≥5 as diagnostic of OSA. Similarly, apnea-hypopnea episodes (AHE; defined as total number of apneas and hypopneas throughout the monitored period) and oxygen desaturation events (ODE; defined as total number of oxygen desaturation events throughout the monitored period), time when oxygen saturation was <90%, time when oxygen saturation was <85%, nadir of oxygen desaturation, oxygen desaturation index, percentage of flow-limited breaths with snoring, and the percentage of flow-limited breaths without snoring were also verified.
Associating FHM and polysomnography data
The FHM and polysomnography data sheets were then reviewed; the precise recorded time of apnea event, presence of an associated fetal heart rate tracing abnormality, degree of oxygen desaturation, and nadir of oxygen desaturation were noted. At the onset of each subject’s recorded monitoring, time-syncing for each FHM and polysomnography tracing occurred and was validated to be within <10 seconds of precision.
Statistical analysis
Women with and without OSA as diagnosed by accepted polysomnography were compared. For calculation of power, the primary objective was to test the null hypothesis pertaining to the use and validation of the Berlin Questionnaire in pregnant women. We assumed a prevalence of 40% by questionnaire. With an alpha of .05, we anticipated that to detect a difference in the questionnaire vs polysomnography at <20% would require a minimum of 80 subjects. Given the concomitant fetal heart rate tracing as a pilot study, we arbitrarily (but a priori) increased the total number of enrollees to 100. Nominal data were analyzed with χ 2 or Fisher’s exact test with Yates correction for continuity, Student t test for continuous data, and Wilcoxon rank sums for ordinal or nonnormal data. The relationship between standard Berlin measures and each of its components (eg, snoring questions, sleepiness questions) and the gold-standard polysomnography test (along with various components measured by polysomnography, including AHI, AHE, and ODE) were analyzed by linear and logistic regression analyses. R 2 and corresponding adjusted odds ratios with 95% confidence intervals (CIs) and probability values were determined with polysomnography as the outcome variable and Berlin as the predictor of interest. The relationship between OSA and measures was examined overall and at various strata of BMI per aforementioned Institute of Medicine Guidelines (≤25, 25-29, ≥30 kg/m 2 ); statistical hypotheses were tested with the use of 2-tailed 95% CIs. Specifically, our analysis was adjusted for the potential confounders of maternal age, parity, and race and then stratified by maternal BMI to examine for consistency in finding, which was performed by adding this variable to each model on the second step. Data from logistic regression modeling are reported as adjusted odds ratios with 95% CIs and probability values for estimation of statistical significance. All statistical analysis was performed with SAS System statistical software (SAS Institute Inc, Cary, NC).
Results
Study population
A total of 176 patients met the criteria over a 9-month period and were approached for participation in the study; 76 patients declined to participate, and 100 patients consented, administered the questionnaire, and underwent at least 3 hours of nocturnal polysomnography with FHM. All 100 patients had complete FHM tracings. Most of our patients were Hispanic and were admitted for unrelated conditions, preterm labor, gestational diabetes mellitus, preeclampsia, preterm premature rupture of membranes, chronic hypertension, trauma, gestational hypertension, pregestational diabetes mellitus, and urinary tract infection (in order of frequency), none of which was significant enough to warrant immediate delivery ( Table 1 ). Using the “gold standard” polysomnography diagnostic criteria, 20 patients (20%) were diagnosed with OSA; by Berlin questionnaire measures, a significantly higher number (36 patients; 36%) would have been diagnosed with OSA. As described in Table 1 , subjects were similar by virtue of maternal characteristics and indication for antepartum admission ( Table 1 ).
| Characteristic | Overall (n = 100) | OSA + (n = 20) | OSA – (n = 80) | P value |
|---|---|---|---|---|
| Maternal age, y a | 26.6 ± 7.1 | 28.7 ± 6.8 | 28.6 ± 7.2 | .85 |
| Body mass index, kg/m 2a | 27.5 ± 7.2 | 28.6 ± 8.6 | 27.2 ± 6.9 | .73 |
| Gestational age, wk a | 32.3 ± 3.5 | 32.5 ± 2.9 | 32.2 ± 3.6 | .75 |
| Gravidity, n a | 3.3 ± 2.1 | 3.5 ± 2.9 | 3.2 ± 1.8 | .91 |
| Parity, n a | 1.72 ± 1.5 | 1.6 ± 1.4 | 1.8 ± 1.5 | .63 |
| Smoker, n (%) | 4 (4) | 1 (5) | 3 (3.8) | .99 |
| Amniotic fluid index a | 13.6 ± 4.6 | 12.9 ± 4.6 | 13.8 ± 4.6 | .55 |
| Pregestational diabetes mellitus, n (%) | 7 (7) | 1 (5) | 6 (7.5) | .99 |
| Chronic hypertension, n (%) | 9 (9) | 2 (10) | 7 (8.8) | .99 |
| Gestational diabetes mellitus, n (%) | 21 (21) | 3 (15) | 18 (22.5) | .55 |
| Gestational hypertension, n (%) | 7 (7) | 1 (5) | 6 (7.5) | .99 |
| Mild preeclampsia, n (%) | 9 (9) | 1 (5) | 8 (10) | .68 |
| Severe preeclampsia, n (%) | 4 (4) | 0 | 4 (5) | .58 |
| Preterm labor, n (%) | 24 (24) | 8 (40) | 16 (20) | .06 |
| Preterm premature rupture of membranes, n (%) | 11 (11) | 3 (15) | 8 (10) | .68 |
| Urinary tract infection, n (%) | 4 (4) | 0 | 4 (5) | .58 |
| Trauma, n (%) | 9 (9) | 4 (20) | 5 (6) | .076 |
| Polyhydramnios, n (%) | 2 (2) | 0 | 2 (2.5) | .99 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
