Oral

Introduction:
Several studies have proposed models to predict disease outcomes in pediatric ulcerative colitis (UC), notably PROTECT, Schechter, and PIBD-ahead, but none have been validated by exteal cohorts.

Objectives: 
We aimed to explore these predictive models in a prospective, multicenter inception pediatric UC cohort.

Methods:
We included UC-diagnosed children at 17 centers and followed up on them at three and 12 months thereafter, as well as at their last follow-up. Outcomes included steroid-free remission (SFR), sustained SFR (SSFR), and acute severe colitis (ASC).

Results:
A total of 223 children were included, of whom 74 (34%), 97 (43%), and 52 (23%) presented with mild, moderate, and severe disease, respectively, by the pediatric UC activity index (PUCAI). The SFR rate was 35% at three months and 47% at 12 months (62% of those with mild disease at diagnosis, 40% with moderate disease, and 42% with severe disease; p=0.01). Fifty-three children (24%) developed ASC during the first year. The sensitivity/specificity/PPV/NPV of the PROTECT model for predicting SFR at three and 12 months were 55%/59%/42%/71% and 80%/29%/50%/62%, respectively. The sensitivity/specificity/PPV/NPV of Schechter's criteria to predict SSFR at 12 months were 50%/60%/35%/74%. Only PUCAI at baseline and three months predicted ASC.

Conclusion:
None of the predictive models of SFR in pediatric UC achieved sufficient accuracy, as all were far from that which was reported in their original cohorts. PUCAI was the only predictor of ASC. This highlights the necessity for exteal validation of any prediction model before its implementation in clinical practice.