Clinical Variables as Prognostic Tools in Pediatric-onset Ulcerative Colitis: A Retrospective Cohort Study

Background: Clinical variables may identify a subset of patients with pediatric-onset ulcerative colitis (UC) (<18 years at diagnosis) at risk for adverse outcomes. We postulated that routinely measured clinical variables measured at diagnosis would predict colectomy in patients with pediatric-onset UC. Methods: We conducted a chart review of patients with pediatric-onset UC at a single center over a 10-year period. We compared patients with and without colectomy across several variables, used proportional hazards regression to adjust for potential confounders, and assessed the ability of a UC risk score to predict colectomy. Results: Among 470 patients with inflammatory bowel disease ICD9-coded encounters, 155 patients had UC and 135 were eligible for analysis. The 1- and 3-year colectomy rates were 16.7% (95% confidence interval [CI]: 11.0%-24.8%) and 35.6% (26.7%-45.4%). White blood cell (WBC) count and hematocrit measured at diagnosis were associated with colectomy at 3 years, even after correcting for potential confounding variables. A UC Risk Score derived from the WBC count and hematocrit was strongly associated with colectomy risk, with a high negative predictive value (NPV) for colectomy at 1 and 3 years (NPV = 0.95 and 0.89, respectively), but low positive predictive value (PPV = 0.22 and 0.38, respectively). Conclusions: A risk score calculated from WBC and hematocrit measured at diagnosis was associated with, but incompletely predictive of, colectomy in pediatric-onset UC. These data suggest 1) routinely measured clinical variables may have a prognostic role in risk stratification, and 2) multicenter prospective studies are needed to optimize risk stratification in pediatric UC. Our findings have impact on the design of such studies.