Abstract Background Approximately 30% of hospitalized patients with acute severe ulcerative colitis (ASUC) may be refractory to systemic glucocorticoid (GC) therapy, administered orally or intravenously necessitating salvage therapy 1. Clinical tools predicting response to GC could allow for early initiation of salvage therapy thus reducing GC exposure and improving overall outcomes. However, comparative studies for available prognostic scores are limited and yield conflicting results 2,3. We aimed to determine the best performing score on a cohort of patients with ASUC from a large tertiary hospital. Methods A retrospective cohort study was conducted between 2017-2022. Hospitalized patients with ASUC, as defined by Truelove and Witts criteria and treated with GC, were included. Patients were classified as GC-refractory (GC-Ref) if they had no clinical and biochemical response, requiring salvage therapy. Assessments for GC response were performed on the third and seventh day of hospitalization. The predictive value of the Mayo endoscopic and Adams scores was evaluated on admission, while the Gibson, Ho, Lindgren, and Oxford scores were assessed on the third day. We compared scores using a multivariate logistic regression and ROC analysis. Results Ninety-six patients met the inclusion criteria and were included, 56 (58%) were GC-responsive (GC-Res) and 40 (42%) were GC-Ref. The mean age of admission was similar between the groups (GC-Res: 42.2 ± 16.7 years; GC-Ref: 37.15 ± 15.3; P=0.1). The GC-Ref group had more extensive disease (65% Vs 38.2%, P=0.01), (Table 1). On admission, the GC-Ref group had significantly more bowel movements per day (14.6 Vs 11.3, P=0.05), elevated CRP levels (91.5 mg/L Vs 58.4 mg/L, P=0.005), higher MAYO score (2.8 Vs 2.3, P=0.028) and lower albumin levels (32 g/L Vs 35.4 g/L, P=0.008) and body mass index (20.3 kg/m2 Vs 23.8 kg/m2, P=0.02). The Gibson score (based on three or more stools per day and a CRP-to-albumin ratio (CAR) above 0.85) and Mayo Endoscopic Score demonstrated appropriate sensitivity and significant predictive value for identifying refractory patients (aOR: 11.3 (95% CI 1.87-68), P=0.008 and aOR: 7.9 (95%CI 1.8-34), P=0.006); respectively. The Gibson score, showed the best overall performance with a sensitivity of 0.7, specificity of 0.75 and positive predictive value of 0.71 (Table 2). Of the individual elements of the various scores, CAR measured on the third day of hospitalization had the highest predictive value (aOR: 12.9 (95% CI 2.5-67), P=0.002) with sensitivity of 0.72, and specificity of 0.65. Conclusion In our cohort, the Gibson score proved to be the most effective predictive tool for identifying patients with steroid refractory ASUC. This need to be confirmed in larger prospective cohorts.
Read full abstract