Validation of models using basic parameters to differentiate intestinal tuberculosis from Crohn's disease: A...

Area of Science:

• Gastroenterology
• Internal Medicine
• Diagnostic Accuracy

Background:

• Limited external validation data exists for models differentiating Crohn's disease (CD) from intestinal tuberculosis (ITB).
• Accurate differentiation between CD and ITB is crucial for appropriate patient management.

Purpose of the Study:

• To externally validate and compare existing models for distinguishing CD from ITB.
• To assess the diagnostic performance of models utilizing clinical, endoscopic, and pathology data.

Main Methods:

• Retrospective collection of data from newly diagnosed ITB and CD patients across 5 centers in Thailand and Hong Kong.
• Application of data to established models: Lee et al., Makharia et al., Jung et al., and Limsrivilai et al.
• Comparison of model performance using Area Under the Receiver Operating Characteristic Curve (AUROC) and misdiagnosis rates.

Main Results:

• Limsrivilai's clinical-endoscopy-pathology (CEP) model achieved an AUROC of 0.887, outperforming other models.
• The CEP model demonstrated significantly better performance than the clinical-endoscopy (CE) model (AUROC: 0.824) and Jung's model (AUROC: 0.798).
• Misdiagnosis of ITB as CD was lowest with Limsrivilai's CEP model (9.6%) compared to Jung's (15.7%) and Makharia's (66.3%) models.

Conclusions:

• Models incorporating more diagnostic modalities and parameters exhibit superior performance.
• Despite improved accuracy, clinical application remains challenging due to persistent misdiagnosis rates, particularly of ITB as CD.
• Future models should integrate additional modalities like imaging and serological tests for enhanced diagnostic accuracy.