Observer variation of diagnoses based on simple biopsy criteria differentiating among Crohn's disease, ulcerative colitis, and other forms of colitis.

BACKGROUND AND AIM: Simple mucosal biopsy criteria proposed by authors reliably differentiate idiopathic inflammatory bowel disease (IBD) from other forms of colitis (non-IBD) and Crohn's disease involving the colon (CD) from ulcerative colitis (UC). The aim of this study is to investigate the reproducibility of these criteria.

METHODS: Three established pathologists and two medical students blindly examined 20 sets of multiple biopsy slides from patients with CD, 20 from those with UC, and 20 from those with non-IBD. The students had been given instructions on histological definitions using another 15 sets previously. Each observer evaluated 10 histological items required in the criteria and determined categorical diagnoses such as definite IBD and probable UC. Interobserver agreement for the individual histological items was measured by using kappa analysis and Pearson's correlation, while it was measured for categorical diagnoses with the use of Spearman's rank correlation.

RESULTS: All of the individual histological items expressed excellent or fair-to-good agreement among the five observers, although two items associated with the criteria for CD and UC had poor agreements among the students. With regard to categorical diagnoses based on the criteria for IBD and non-IBD, and those for CD and UC, coefficients for Spearman's rank correlation exceeded 0.92 and 0.86 among the pathologists, and 0.76 and 0.74 among the students, respectively; all of the coefficients were statistically significant (P < 0.05).

CONCLUSIONS: The simple criteria were sufficiently reproducible and would help most pathologists to make an automated and objective diagnosis.