Poster Presentation Australasian Diabetes in Pregnancy Society Annual Scientific Meeting 2019

Can risk prediction calculators be used to inform models of care aligned to risk stratification in women with GDM? (#81)

Tang Wong 1 2 3 , N Wah Cheung 3 4 , Glynis Ross 1 3 , Robyn Barnes 1 5 , Jeff Flack 1 2 6
  1. Department of Diabetes and Endocrinology, Bankstown-Lidcombe Hospital, Sydney, NSW, Australia
  2. University of NSW, Sydney, Australia
  3. University of Sydney, Sydney, Australia
  4. Department of Diabetes & Endocrinology, Westmead Hospital, Westmead, NSW, Australia
  5. University of Newcastle, Newcastle, NSW, Australia
  6. Western Sydney Univerisity, Sydney, NSW, Australia

Background:
Increasing workload pressures in the context of finite resources and infrastructure has important implications for developing appropriate models of care in GDM management

Aim:
To propose a framework using risk calculators to triage women with GDM into clinics with resources commensurate to their risk profile.

Methods:
We developed risk calculators and nomograms from predictive models of women diagnosed with GDM(ADIPS1998 criteria1) at Bankstown-Lidcombe hospital (1992-2013). The risk calculators assessed the probability of insulin therapy; large for gestational-age infant(LGA) and a composite neonatal outcome (≥1 of the following; needing insulin therapy, pre-term labour, caesarean section, LGA, neonatal hypoglycaemia/jaundice) according to antenatal maternal parameters. Boxplots were created for predicted probabilities to determine the quartiles of risk for each outcome assessed. Only the risk calculators derived from models using continuous variables were used due to higher performance and a more Gaussian distribution.

Results:
There were a total of 3095 singleton births to GDM women used to generate the predictive models and derive the risk calculators (accessible on www.gdmriskcalculator.com). A total of 2541 women had complete data to calculate predicted probabilities for all three endpoints.  Predicted probabilities of >43.5%, >15.5% and >72.5% represented the highest quartile of risk for insulin therapy, LGA and the composite outcome, respectively. In contrast, predicted probabilities of <17.0%, <8.5% and <50.5% represented the lowest quartiles of risk for the three endpoints.

In our proposed model of care, those at highest obstetric risk (those at the highest quartile for risk of LGA and the composite outcome) could be managed in a high resourced setting. Women not in the highest quartile of obstetric risk, and having the lowest quartile of risk for requiring insulin therapy, could be managed in a low resourced setting (see table1).

5cf0d02de6133-Model+of+care.jpg

Conclusion:
In our proposed model of care, GDM risk calculators could assist with the allocation of resources to women according to risk strata of both adverse obstetric outcome and the likelihood of requiring insulin therapy.

  1. Hoffman, L., et al., Gestational diabetes mellitus--management guidelines. The Australasian Diabetes in Pregnancy Society. Med J Aust, 1998. 169(2): p. 93-7.