Poster Presentation International Association of the Diabetes and Pregnancy Study Groups 2022 - Hosted by ADIPS

StUdy of Gestational diabetes And Risk using Electronic Data (SUGARED) (#139)

Tessa Weir 1 2 , Travis Stenborg 1 3 , Sarah Glastras 1 2
  1. University of Sydney, St Leonards, NSW, Australia
  2. Endocrinology, Royal North Shore Hospital, St Leonards, New South Wales, Australia
  3. ARC Centre in Data Analytics for Resources and Environments (DARE), St Leonards, New South Wales, Australia

Background: Gestational diabetes (GDM) affects 1 in 6 pregnancies within Australia [1]. The incidence of GDM increased due to changed diagnostic criteria, implemented as a result of the Hyperglycaemia and Adverse Pregnancy Outcome (HAPO) study [2]. Bayesian methods are better at modelling complex data relationships compared to traditional methods of analysis. Hence, we reanalysed the HAPO data using Bayesian methods to identify subgroups of women with GDM with low or high risk of adverse outcomes.

Method: Bayesian regression analysis was applied to the HAPO dataset (> 23,000 women unclouded by treatment interference). This involved modelling birth outcome (birth weight >90th percentile) as a function of glucose levels on the 75g oral glucose tolerance test and selected maternal variables (field centre, ethnicity, age and BMI), assuming a linear combination of fixed and random effects. Random intercepts and random slopes were associated with clusters in the maternal variables of interest.

Results: The model, including fasting, 1-hour and 2-hour glucose levels and selected covariates, predicted birth weight >90th percentile. Change in each glucose measure’s response as a function of the other two measures, with a single covariate, BMI, are seen in Figure 1. The model was then extended to include covariates field centre, ethnicity, age and BMI.

62e1e0b44177f-Figure+1+Mixed+Effects+Model.png

Figure 1: Mixed effects model. Estimated probability of birth weight >P90 by plasma glucose measures and BMI class

Conclusion: Our sophisticated statistical methods allow us to understand the complex inter-relationships between factors affecting adverse birth outcomes, and have established subgroups of women based on BMI and glucose levels at varying risk of birth weight >90th percentile. This model will be expanded to develop a personalised risk prediction tool, assisting clinicians and women to make more individualised decisions regarding treatment, thereby minimising unnecessary obstetric intervention and allowing for a focusing of resources on women at high risk of adverse outcomes.

  1. O'Sullivan EP, Avalos G, O'Reilly M, et al. Atlantic Diabetes in Pregnancy (DIP): The prevalence and outcomes of gestational diabetes mellitus using new diagnostic criteria. Diabetologia. 2011;54(7):1670-1675.
  2. HAPO Study Cooperative Research Group, Metzger BE, Lowe LP, et al. Hyperglycemia and adverse pregnancy outcomes. N Engl J Med. 2008;358(19):1991-2002. doi:10.1056/NEJMoa0707943