Obstructive Sleep Apnea in Gestational Diabetes: Prevalence, Predictive Factors and the Development of a Screening Tool

Presentation Number: SAT 594
Date of Presentation: April 1st, 2017

Ekasitt Wanitcharoenkul*1, Boonsong Ongphiphadhanakul2, Naricha Chirakalwasan3, Somvang Amnakkittikul4, Suranut Charoensri5, Sunee Saetung6, Punyu Panburana4, Sommart Bumrungphuet5 and Sirimon Reutrakul4
1Faculty of Medicine Ramathibodi hospital, Bangkok, Thailand, 2Faculty of Medicine, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand, 3Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand, 4Faculty of Medicine Ramahibodi Hospital, Bangkok, Thailand, 5Faculty of Medicine Ramathibodi Hospital, Bangkok, Thailand, 6Department of Medicine, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand


Obstructive sleep apnea (OSA) has emerged as a risk factor for gestational diabetes (GDM). The exact prevalence of OSA in pregnancy is unknown. In addition, studies revealed that questionnaires typically used to screen for OSA in non-pregnant population were not accurate in predicting OSA in pregnancy. Factors predicting OSA in GDM women have not been studied to date. The objectives of this study were to investigate the characteristics of GDM women with OSA, and to develop a simple and practical screening tool for OSA in GDM using classification tree analysis.

Methods: Diet-controlled, obese GDM women were enrolled at gestational age (GA) 24-36 weeks. Baseline characteristics, glycemic parameters and neck circumference were obtained. The participants completed the Berlin questionnaire which assessed three categories (snoring, daytime fatigue and hypertension/BMI). High risk for OSA was considered when two of three categories were positive. The participants underwent a diagnostic test for OSA using an overnight home monitoring device (WatchPAT200). OSA was diagnosed when an apnea hypopnea index (AHI) ≥5. Parameters were compared between OSA and non-OSA groups. A classification tree method was applied to develop an algorithm to predict OSA.

Results: Of the 82 women who completed the study, 43 (52.4%) had OSA. The prevalence of OSA was 52.4%. There were no significant differences in age, pre-pregnancy BMI, current BMI, GA at sleep assessment, fasting glucose or HbA1c at sleep assessment between OSA and non-OSA women. Neck circumference was significantly larger in the OSA than in the non-OSA group (median 35.0 vs. 34.5 cm, p = 0.018). High risk for OSA as assessed by Berlin questionnaire (2 of 3 positive categories) did not differ between OSA and non-OSA groups (16% vs. 10%, p=0.424). However, significantly more women with OSA scored positive in at least 1 category of Berlin questionnaire than the non-OSA women (74%. vs. 49%, p=0.017).

We developed a screening tool using a decision tree by including the two variables, neck circumference and Berlin questionnaire. The result was validated using statistical bootstrap. Using this model, the number of positive categories from Berlin should be considered first. Those with ≥1 positive category were considered high risk for OSA (63% by WatchPAT200 testing). For those with negative results in all categories of Berlin, neck circumference should be considered next. Those with neck circumference >35.5 cm were considered as high risk for OSA (100% by WatchPAT200). Those with neck circumference ≤35.5 cm were considered as low risk for OSA (23% by WatchPAT200). The sensitivity and specificity using this classification tree were 86% and 51% respectively. The overall accuracy was 70%.

Conclusion: OSA is highly prevalent in diet-controlled obese GDM women. An algorithm using a neck circumference and Berlin questionnaire could help in the screening for OSA in GDM.


Nothing to Disclose: EW, BO, NC, SA, SC, SS, PP, SB, SR