Prognostic Prediction Models for Ulcerative Colitis: Systematic Review and Meta-Analysis.

Area of Science:

• Gastroenterology and Hepatology
• Clinical Epidemiology
• Biostatistics

Background:

• Ulcerative colitis (UC) is a chronic inflammatory bowel disease with variable patient outcomes.
• Prognostic models are crucial for personalized medicine in UC but require thorough evaluation.
• Current quality and clinical utility of UC prognostic models are not well-assessed.

Purpose of the Study:

• To systematically review and critically evaluate prognostic prediction models for UC.
• Assess the development, performance, and clinical applicability of existing UC prognostic models.

Main Methods:

• Comprehensive literature search across major databases up to November 2024.
• Data extraction on study characteristics, model development, and validation metrics.
• Risk of bias and applicability assessment using the Prediction Model Risk of Bias Assessment Tool.
• Meta-analysis to evaluate pooled model performance (AUC).

Main Results:

• 30 studies with 7452 UC patients were included; most were retrospective.
• Models predicted treatment response (e.g., to TNF-alpha inhibitors), surgery risk, or relapse.
• Pooled AUC was 0.84, but 97% of studies had high risk of bias.
• Limited external validation (47%) and generalizability concerns (60%) were noted.

Conclusions:

• UC prognostic models demonstrate potential but are limited by high bias and insufficient validation.
• Future research must focus on multicenter validation and refined model development.
• Enhancing model applicability is key for broader clinical implementation in UC management.