Search in the Abstract Database

Search Abstracts 2016

* = Presenting author

OP016 Development and validation of diagnostic criteria for IBD-unclassified (IBDU) in children: a multicentre longitudinal study from the paediatric IBD Porto Group of ESPGHAN

L. Birimberg Schwartz1, D. Zucker2, A. Akriv2, C. Salvatore3, F. Cameron4, I. Lazowska5, L. Yianni6, P. Siba7, S. Kolacek8, C. Romano9, S. Buderus10, A. Pærregaard11, J. C. Escher12, D. Turner*13

1Shaare Zedek Medical Centre, Paediatrics, Jerusalem, Israel, 2The Hebrew University, Jerusalem, Israel, 3Sapienza University of Rome, Rome, Italy, 4Yorkhill Children’s Hospital, Glasgow, United Kingdom, 5Medical University of Warsaw, Warsaw, Poland, 6University Hospital Southampton, Hampshire, United Kingdom, 7University of Dundee, Scotland, United Kingdom, 8Zagreb University Medical School, Zagreb, Croatia, 9University of Messina, Messina, Italy, 10St Marien Hospital, Bonn, Germany, 11Hvidovre University Hospital, Copenhagen, Denmark, 12Erasmus Medical Centre, Holland, Netherlands, 13Shaare Zedek Medical Centre, Jerusalem, Israel

Background

There are no available criteria by which to classify and diagnose inflammatory bowel disease unclassified (IBDU). We aimed to derive and validate diagnostic criteria of IBDU in children on the largest IBDU cohort ever constructed.

Methods

This was a multicentre retrospective longitudinal study from 23 centres affiliated with the paediatric IBD Porto Group of ESPGHAN. Both a hypothesis-driven judgemental approach and mathematical modelling (including classification and regression tree analysis) were used for creating the classification algorithm.

Results

Enrolled were 749 children with IBD: 236 (32%) with Crohn’s colitis (CC), 272 (36%) with ulcerative colitis (UC), and 241 (32%) with IBDU (mean age 10.9 ± 3.6 years, 53% males). Median follow-up was 2.8 years (IQR 1.7–4.3). A set of 23 features were clustered in 3 classes according to how frequently they may be seen in UC: 6 features in class 1 (0% prevalence in UC), 12 in class 2 (more than 0% but < 5% prevalence), and 5 in class 3 (5% to 10%) (Table 1). According to the algorithm (Figure 1),CC should be diagnosed by ≥ 1 class-1 feature, ≥4 class-2 features, or 1–3 class-2 features with ≥ 3 class-3 features. When having no class-1 and 2 features with ≤ 2 class 3 features, patients are diagnosed as UC. All other combinations should prompt the diagnosis of IBDU. This algorithm differentiated UC from CC and IBDU with 80% sensitivity (95% CI [71–88]) and 84% specificity (95% CI [77–89]). The algorithm also differentiated between CC from IBDU and UC with 78% sensitivity (95% CI [67–87]) and 94% specificity (95% CI [89–97]).

Conclusion

The validated algorithm can adequately classify children with colonic IBD into CC, UC, and IBDU, thus enabling the standardisation of the diagnostic criteria of IBDU.




The chosen algorithm based on the hypothesis-driven analysis for the differential diagnosis of IBD subgroups.

Final division of 23 features by classes