Interpretation
We evaluated a range of potential cut-points from our predictive model that could be used to prioritize groups of women to target for different interventions. For example, the Diabetes Prevention Program demonstrated that lifestyle modification in women with a history of GDM reduces the risk of type 2 diabetes by 50%, but is costly to implement at $9,000/year.35 For our model, a cut-point of 0.24 would identify 56% of women for additional intervention and had reasonable sensitivity (80%) and low specificity (58%), suggesting it could be a useful cut-point to identify a high risk subgroup of women for the most intensive and costly postpartum lifestyle modification interventions. Conversely, a cut-point of 0.13 would identify 76% of the population for further intervention, had higher sensitivity (96%) and lower specificity (35%) and may be more appropriate for low-cost interventions, such as metformin or other pharmacological interventions. Future cost-effectiveness analyses are required to investigate the implications of offering interventions to women with GDM based on different cut-points from the proposed model.
Our results indicate a clinical prediction model may be useful around the time of delivery to identify women with recent GDM at high risk for impaired glucose tolerance. However, a larger study with longer follow-up is needed to assess the model’s ability to predict type 2 diabetes. In addition, future work should explore whether covariates such as socioeconomic status, healthcare access, and diet help to predict type 2 diabetes risk.