(109) PREDICT-GDM: Prospective Evaluation of a Risk-based model for Diabetes Intervention & Care Tailoring

We conducted a single center, parallel-group, proof-of-concept RCT (Aug 2022–Jul 2024), enrolling individuals with singleton GDM pregnancies achieving glycemic targets with LM alone at baseline. Participants were randomized to routine care (RC) in the Diabetes in pregnancy clinic or to a modified care (MC) pathway using a previously developed and validated prediction model. Individuals identified at low risk of LM failure were discharged to routine obstetric care with care provider blinded remote glucose monitoring. Those identified at high risk of LM failure remained in the DIP clinic (Fig.1). The primary outcome was validation of model performance. Exploratory outcomes included glycemic control, maternal/neonatal outcomes, patient satisfaction and resource utilization.

Results:

Of 582 individuals with GDM, 211 were eligible; 150 consented and were randomized. In the intervention arm, 18 (24%) were identified as low-risk and 57 as high-risk. All low-risk participants maintained glycemic control on LM, yielding NPV of 100%, specificity of 30.2%, and AUC of 0.732 (95% CI: 0.649–0.814). Patient satisfaction (n=67) was high (mean 3.86/5), with no significant group differences. There were lower admission days (2.8 ± 0.86 vs 3.6 ± 1.3, p=0.006) in the low-risk group and a trend toward lower caesarean rates (6.7% vs 35.8%, p=0.051). Compared with low-risk controls, low-risk individuals in the MC pathway differed in the rate of postprandial dinner glucose above target ,post-partum hemorrhage and mean specialist visits (Table 1).

Conclusion:
This easily implementable tool accurately identifies nearly a quarter of low-risk GDM patients who can safely be managed in a lower-intensity care model thus reducing resource use without compromising outcomes.