Early Predictors of Steroid Failure in Acute Severe Ulcerative Colitis: A Retrospective Study With the Evaluation of the Admission Model for Intensification of Therapy in Acute Severe Colitis (ADMIT-ASC) Score.

BACKGROUND AND

aimsAcute severe ulcerative colitis (ASUC) is a life-threatening gastroenterological emergency in which early identification of patients at risk of corticosteroid failure is critical for timely therapeutic escalation. This study aimed to identify early predictive factors of corticosteroid failure and to evaluate the performance of the Admission Model for Intensification of Therapy in Acute Severe Colitis (ADMIT-ASC) score in predicting therapeutic failure in a North African cohort.

methodsThis was a retrospective, single-center study. Patients hospitalized for ASUC between September 2021 and January 2026 with complete data for ADMIT-ASC calculation were included (n = 87). The primary outcome was corticosteroid failure, defined as the requirement for rescue anti-TNF therapy or emergency colectomy during the same index hospitalization. Disease severity was assessed using the Truelove and Witts (TW) criteria, the Lichtiger score, the Ulcerative Colitis Endoscopic Index of Severity (UCEIS), the Oxford criteria (Travis score), and the ADMIT-ASC score. Fisher's exact test was used when expected cell counts were below 5; the chi-square (χ²) test otherwise.

resultsOf 87 patients included, 53 (60.9%) achieved clinical remission under intravenous corticosteroids; 27 patients (31.0%) required rescue therapy during the index hospitalization: 21 received anti-TNF rescue therapy (primarily infliximab 5 mg/kg), and 6 (6.9%) underwent emergency surgical colectomy; and the remaining 7 (8.1%) patients had incomplete outcome documentation in the hospital records, were retained in the denominator, and were excluded from the primary outcome analysis. The presence of more than two additional TW criteria (including fever, tachycardia, anemia, and elevated CRP beyond stool frequency) was the strongest admission predictor (χ² = 11.44, OR = 7.27 (95% CI: 2.37-22.31), p = 0.001). The ADMIT-ASC score ≥ 3 was significantly associated with failure (χ² = 4.34, OR = 3.92 (95% CI: 1.20-12.80), p = 0.037; sensitivity of 33.3% (95% CI: 18.6-52.2%), specificity of 88.7% (95% CI: 77.4-94.7%)). In a Firth penalized logistic regression, including both admission predictors, both remained independently associated with failure: adjusted OR 7.37 (95% CI: 2.35-23.12) for additional TW criteria and 5.10 (95% CI: 1.44-18.07) for ADMIT-ASC ≥ 3 (p = 0.012). The Lichtiger score ≥ 11 showed a trend toward significance without reaching it (p = 0.114). The UCEIS ≥ 7 was not significantly associated with failure (p = 0.631). The Travis score at day three remained the strongest predictor overall (OR = 153.1; p < 0.001).

conclusionsTraditional clinical markers at admission - particularly the number of additional TW criteria - remain highly effective predictors of corticosteroid failure. The ADMIT-ASC score is a promising adjunct for structured early risk stratification, with a high specificity (88.7%), complementing rather than replacing traditional clinical assessment and the Travis score at day three.