The purpose of this study was to evaluate the effectiveness of standard laboratory tests in detecting severe ulcerative colitis in children at the time of diagnosis. A total of 427 children aged 4 to 17 years old who had just been diagnosed with ulcerative colitis (UC) were recruited in the study. Boosted classification trees were used to characterize the predictive ability of disease attributes based on clinical disease severity as measured by the Pediatric Ulcerative Colitis Activity Index (PUCAI), severe versus not severe, and total Mayo score, severe versus not severe; mucosal disease as measured by the Mayo endoscopic subscore, severe versus not severe; and extensive disease versus not extensive. The average age was 12.7 years, with 49.6 percent being female and 83 percent being Caucasian. Severe overall Mayo score was observed in 28%, mean PUCAI score was 49.8 20.1, and severe mucosal disease was present in 33% with widespread involvement in 82%. White blood cell count, erythrocyte sedimentation rate, and platelet count (PLT) were found as the top three blood laboratory tests for predicting illness extent and severity using classification and regression trees. Albumin was used instead of PLT to assess mucosal severity. Using conventional clinical cut-points, classification models for PUCAI and total Mayo achieved sensitivity of at least 0.65, with misclassification rates of around 30%.
In respect to objective clinical scores, a combination of white blood cell count, erythrocyte sedimentation rate, and either PLT or albumin is the most predictive subset of routine laboratory tests for distinguishing severe from nonsevere clinical or mucosal illness at diagnosis.
