Externally validated risk prediction models for gestational diabetes mellitus: A systematic review and meta-analysis.
Beminate Lemma Seifu, Yitayeh Belsti, Nigus Bililign Yimer et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review and meta-analysis evaluated the external validation of 33 gestational diabetes mellitus (GDM) risk prediction models across 26 studies. The Teede et al. model demonstrated the most extensive validation with a pooled AUC of 0.72, while the Nanda et al. model achieved the highest pooled discrimination at 0.77. Methodological weaknesses were prevalent in the validations.
Key findings
- The Teede et al. model had 11 external validations and a pooled AUC of 0.72 (95% CI: 0.67-0.76).
- The Nanda et al. model achieved the highest pooled AUC of 0.77 (95% CI: 0.74-0.80) from 5 validations.
- 69.23% of studies assessing GDM prediction models had a high risk of bias.
- Meta-analysis was feasible for only four models due to limited data.
Why it matters
Understanding the performance of GDM risk prediction models is essential for early identification and prevention strategies. Clinicians can benefit from insights into which models are more reliable and applicable in diverse populations.
What to keep in mind
Many validations were methodologically weak, and the majority of studies had a high risk of bias.