Inter-observer variability for detection of suspected ileocolonic Crohn's disease with magnetic resonance

Jacob Broder Brodersen1,2, Michael Dam Jensen1,2, Mathias Byriel2,3

  • 1Department of Internal Medicine, Section of Gastroenterology, Esbjerg Hospital - University Hospital of Southern Denmark, Esbjerg, Denmark.

Insights

Magnetic resonance enterography (MRE) shows moderate to substantial agreement for evaluating ileocolonic Crohn's disease (CD). While agreement for colonic assessment was lower, it may reflect early disease detection rather than observer variability.

Area of Science:

  • Gastroenterology
  • Radiology
  • Medical Imaging

Background:

  • Magnetic resonance enterography (MRE) is a key imaging modality for visualizing Crohn's disease (CD) and its complications.
  • Accurate assessment of CD, particularly in the ileocolonic region, is crucial for diagnosis and management.

Purpose of the Study:

  • To evaluate the inter-observer agreement among radiologists in detecting and assessing ileocolonic Crohn's disease using MRE.
  • To compare the agreement levels between senior radiologists and junior doctors in training.

Main Methods:

  • A post-hoc analysis of MRE scans from 48 patients with suspected CD was performed.
  • Two senior radiologists and two junior doctors independently assessed MREs for image quality, CD presence, and disease severity, blinded to clinical data.
  • Inter-observer agreement was quantified using Kappa statistics for CD detection and Intraclass Correlation Coefficients (ICCs) for disease severity scoring.

Main Results:

  • Inter-observer agreement for CD detection was moderate to substantial across the ileocolon (κ=0.65), terminal ileum (κ=0.77), and colon (κ=0.59).
  • Agreement was generally higher among senior radiologists compared to junior doctors, although not statistically significant.
  • Inter-observer agreement for disease severity scoring (MaRIA and simplified MaRIA) was poor to moderate (ICC=0.51 and 0.46, respectively), with senior radiologists showing better consistency.

Conclusions:

  • MRE demonstrates moderate to substantial inter-observer agreement for ileocolonic CD evaluation in early disease stages.
  • Lower agreement in colonic assessment may be attributed to the challenges of detecting early-stage disease rather than observer variability.
  • Complementary assessment methods may be necessary to optimize the evaluation of colonic involvement in CD.