Prediction Models for Spontaneous Preterm Birth: A Systematic Review

Three prospective cohort studies, with sample sizes from 135 to 1,107, were identified in a systematic review of prediction models for spontaneous preterm birth in asymptomatic low-risk singleton pregnancies. All models used logistic regression and showed moderate to strong apparent discrimination (AUC 0.781–0.94), with cervical length as a consistent predictor. The review found marked heterogeneity in predictors and modeling strategies, and all included studies were assessed as having a high risk of bias with no validation performed.

Use Cases

• Evaluating methodological quality of existing preterm birth prediction models using the PROBAST tool.
• Identifying consistent clinical predictors, such as cervical length, for future model development.
• Informing the design of robust multicenter studies by highlighting gaps in validation and reporting standards.
• Conducting a narrative synthesis of heterogeneous medical literature where meta-analysis is not feasible.

Strengths

• The review screened 16,475 records, providing a systematic and broad search base.
• Data extraction and quality assessment followed established checklists (CHARMS and PROBAST).
• Provides specific performance metrics for identified models, including AUC ranges and odds ratios.

Limitations

• The dataset is a review article (PDF), not a primary data table, limiting direct computational analysis.
• Substantial heterogeneity in predictors and outcomes across the three included studies limits comparability.
• All included studies were assessed as having a high risk of bias and lacked internal or external validation.

Provenance

Source

Gagana Hanumaiah

Collection Method

Systematic literature review following PRISMA guidelines, searching PubMed, Scopus, Embase, and Cochrane Library until 20 August 2025.

Time Range

Literature search conducted until 20 August 2025; review published/updated in 2026.

Freshness

Last updated 2026-03-23