Endocrine Abstracts

Introduction: Type 1 diabetes (T1D) in pregnancy is known to be associated with increased risks of birth defects and miscarriages. Despite the technological improvement and availability of continuous glucose monitoring (CGM), women with T1D who are adequately controlled according to CGM average and HBA1C still face these complications. Glucose variability was suggested as an explanatory factor of the difference between actual obstetric outcome and the expected outcome according to currently available monitoring tools.

Aim: To present an algorithm and to verify that it has a practical potential to notify early enough of increased risk of miscarriage or birth defect.

Methods: Change-point detection methods were applied on CGM data of eight pregnant women with T1D during the entire first trimester of pregnancy.

Results: Seven pregnancies ended with good outcome and one pregnancy ended at a late miscarriage (patient H). The average first trimester HbA1C of the pregnancies with good outcome was 6.58±1.2 (range 5.1–8.8), which is similar to the HbA1C of the pregnancy that ended with a late miscarriage (6.6). The daily averages of CGM data were not higher for patient H. Nevertheless, the statistical method (aimed at detection of an increase in the hourly StDev of glucose levels) did detect an exceptional elevation of glucose variability of patient H, at 8 weeks of pregnancy, which was 7 weeks before the actual miscarriage. Namely, we could have predicted this poor outcome early enough to lessen glucose variability.

Conclusions: Analysis of eight first-trimester follow-ups (of more than 1600 hours ≈ 20,000 measurements for each patient) showed the change-point detection method to fully discriminate between proper pregnancies and pregnancies with adverse outcome. This approach might serve as an additional tool to detect pregnancies at risk in real-time and to plan better glucose control in order to improve pregnancy outcome.