Search in the Abstract Database

Search Abstracts 2011

* = Presenting author

P321. Paediatric IBD – How reliably does the Montreal classification perform?

M.E. Sherlock1, G. Tomlinson2, T. Walters1, D. Urbach2, A.M. Griffiths1

1The Hospital for Sick Children, Toronto, ON, Canada; 2Toronto General Research Institute, Toronto, ON, Canada

Background: The Montreal Classification [1] was developed to provide a uniform system of designating subgroups of patients with inflammatory bowel disease (IBD), with the aim of facilitating multi-center genotype-phenotype correlation studies. Although the classification is frequently used for paediatric IBD patients, its reliability in this population has never been evaluated.

Aim: To determine the reliability with which experienced IBD clinical researchers apply the Montreal Classification in paediatric settings.

Methods: Case scenarios were constructed using de-identified data from the medical records of 50, randomly-selected, paediatric IBD patients. Data pertaining to clinical presentation, radiologic and endoscopic investigations were recorded. International IBD experts classified these cases using the Montreal Classification on two separate occasions. Inter-rater and intra-rater reliability in the assignment of an overall diagnosis of Crohn's disease (CD), ulcerative colitis (UC) or inflammatory bowel disease – type undefined (IBDU) were determined using the kappa (κ) statistic. The single kappa statistic for inter-rater reliability, which summarizes the level of agreement amongst all raters, was calculated using a SAS® Macro [2]. The kappa statistic for intra-rater reliability was determined for each rater individually using standard SAS® coding. Kappa values of ≤0.2, 0.2–0.4, 0.4–0.6, 0.6–0.8 and >0.8 are considered to indicate poor, fair, moderate, good and excellent reliability respectively.

Results: Twelve international paediatric IBD experts participated. The inter-rater reliability for the overall diagnosis of CD, UC or IBDU, using all 50 cases was good (κ = 0.60, 95% CI 0.58–0.63). The reliability improved (κ = 0.64, 95% CI 0.64–0.71) when the analysis was limited to cases with an optimal diagnostic work-up (n = 39). The inter-rater reliability was moderate for assigning disease location (L1-L3) and perianal disease (κ = 0.58, 95% CI 0.54–0.62), but only fair for assigning the category L4 (upper GI tract disease). Agreement was good when assigning disease behavior (κ = 0.76 (95% CI 0.73–0081) and excellent for disease extent in UC patients. The median kappa statistic for intra-rater reliability for overall diagnosis of CD, UC or IBDU was excellent (κ = 0.81) (range 0.52 to 1.00).

Incomplete diagnostic workup, the presence or absence of perianal disease and upper GI tract findings led to variation in classification.

Conclusion: The Montreal classification performs moderately well for classifying paediatric IBD patients; however the level of agreement was not perfect and this potential for misclassification should be considered when interpreting the findings of phenotypic correlation studies.

1. Silverberg MS, Satsangi J, Ahmad T, Arnott ID, Bernstein CN, Brant SR, et al. Toward an integrated clinical, molecular and serological classification of inflammatory bowel disease: Report of a Working Party of the 2005 Montreal World Congress of Gastroenterology. Can J Gastroenterol. 2005; Suppl A:5–36.

2. Chen B, Zaebst D, Seel L. A macro to calculate kappa statistics for categorizations by multiple raters. SUGI 30 Proceedings 2005. Paper 155–30.