Material and Methods 
This prospective single-center study took place at the Lille University Hospital Center, at the Jeanne de Flandre maternity ward, and included a population of obese pregnant women who were offered overnight in-hospital polysomnography for OSA screening. The Lille Hospital Ethics Committee approved this study (CPP 09/65 N° 2009-A01018-49).
Its principal objective was to study the prevalence of OSA in the population of obese women receiving prenatal care and giving birth at our hospital, specifically those with a prepregnancy BMI >35 kg/m². The first of our secondary objectives was to compare the course of pregnancy in women with and without OSA, especially for the onset of pregnancy-related vascular disorders (pregnancy-related hypertension, preeclampsia, eclampsia, HELLP syndrome, or fetal growth restriction, FGR), but also for their pregnancy outcome (type of delivery and newborn’s characteristics). The last secondary objective was to examine whether various criteria, including maternal age, parity, BMI, history of chronic hypertension, history of diabetes, family history of OSA, weight gain during pregnancy, and gestational diabetes might be risk factors for developing OSA in this population.
Participation in the study was offered to all pregnant women aged at least 18 years receiving prenatal care at the Jeanne de Flandre maternity ward with a prepregnancy BMI >35 kg/m². The polysomnography had to take place after 24 and before 32 weeks of gestation. Women were excluded if they refused to participate, did not sign the informed consent, had a twin or higher-order multiple pregnancy, had a guardian or conservator, or took medication likely to modify OSA. Obstetricians explained the study to women meeting the inclusion criteria in a special consultation and gave them written information about it. After time to consider participation, the women who agreed provided written informed consent before inclusion.
These women were then followed up monthly from the fourth month of pregnancy, during prenatal consultations, when obstetric data were collected. Fetal monitoring took place by monthly ultrasound to evaluate fetal weight and by fetal and uterine artery Doppler scans. Polysomnography took place between 24 and 32 weeks of gestation. Mode of delivery and neonatal status were recorded at birth.
The following obstetric data were collected during pregnancy: maternal age, parity, prepregnancy BMI, type 1 or 2 diabetes, chronic hypertension, history of phlebitis and/or pulmonary embolism, family member with OSA, gestational diabetes, weight gain during pregnancy, hospitalization during pregnancy, and presence of any pregnancy-related vascular disease or complication (pregnancy-related hypertension, preeclampsia, eclampsia, HELLP syndrome, or FGR). These pregnancy-related vascular diseases were grouped together as a composite criterion. The criteria for their diagnosis were those defined by the French national guidelines.21
Polysomnography was performed during one night of hospitalization in the hospital’s sleep laboratory, recording an electroencephalogram, electro-oculograms, submental and bilateral anterior tibialis electromyography, an electrocardiogram, nasal and oral air-flow, oxygen saturation and thoracic and abdominal movement. Various clinical indicators were also collected to assess the existence of OSA (ronchopathy or snoring, respiratory pauses, nocturia, night sweats, morning headaches, perception of nonrestorative sleep, and excessive daytime somnolence). Analysis of these clinical and polysomnographic data enabled us to define two groups of women: one group with OSA and one without it. AHI ≥5 measured by polysomnography defined OSA. Among the women with OSA, we distinguished those with mild or moderate sleep apnea (AHI <30) and those with severe OSA (AHI ≥30). Severity was also assessed by simultaneous measurement of arterial oxygen desaturation and consideration of the complete clinical picture (hypersomnolence, neuropsychological disorders, and hypertension). Ventilation by continuous positive airway pressure (CPAP) was proposed to the women with severe OSA, with continuing medical follow-up at the sleep center.
Data about delivery and the baby were collected postpartum: mode of delivery, spontaneous or induced labor, term at delivery, birth weight, acid-base status, and NICU transfer.
The sample size was calculated on the basis of the estimated 95% confidence interval (95% CI) of the theoretical frequency of women with a BMI >35 and OSA (that is, the principal study objective). This 95% CI was calculated with Sachs’ method, and the sample size was calculated to obtain a given level of precision for this CI, defined as half of it. By using the standard formula, which can determine positive predictive value from sensitivity, specificity, and prevalence, we were able to estimate the frequency of OSA in women with a BMI >35 kg/m2 at 56%. We set the precision at 12.5% (length of CI: 25%). In these conditions, we calculated that 68 women were necessary for the study (theoretical frequency estimated at 43.3–68.2%).22The frequency of OSA was estimated by its 95% CI. For the descriptive data analysis, the qualitative data were presented as numbers and percentages, and the quantitative data as means and their standard deviations. The normality of the numeric parameters was verified graphically and by the Shapiro-Wilk test. The groups with and without OSA were compared by Chi-square or Fisher’s exact tests for the qualitative variables (e.g., age, BMI, weight gain, birth weight, Apgar score) and by the Kruskal Wallis or Mann-Whitney tests for the quantitative variables (e.g., chronic hypertension, diabetes, the vascular disease composite criterion). Significance was set at 5%. The analysis was performed with SAS software (version 9.4, SAS Institute, Cary, NC).