A Novel User-Friendly Model to Predict Corticosteroid Utilization in Newly Diagnosed Patients with Ulcerative Colitis

Area of Science:

• Gastroenterology
• Clinical Epidemiology
• Predictive Modeling

Background:

• Corticosteroid (CS) use is a significant indicator of poor prognosis in ulcerative colitis (UC).
• Predicting CS utilization is crucial for managing UC progression and patient outcomes.

Purpose of the Study:

• To develop and validate a prognostic model for predicting corticosteroid (CS) use in newly diagnosed ulcerative colitis (UC) patients.
• To identify key predictors of CS utilization using routinely available clinical data.

Main Methods:

• A nationwide VA cohort of newly diagnosed UC patients was retrospectively analyzed.
• Multivariate logistic regression was employed for model development and internal validation.
• Candidate predictors included demographics, laboratory results, and index colonoscopy findings.

Main Results:

• The study followed 699 UC patients for a median of 8 years, with 41.2% requiring CS.
• Key predictors identified were age, non-African American ethnicity, hypoalbuminemia, iron-deficiency anemia, and baseline disease extent/severity.
• The validated model demonstrated good discrimination (AUC 0.71) and calibration.

Conclusions:

• A novel, internally validated prognostic model effectively predicts CS use in new-onset UC.
• The model utilizes readily available baseline data for risk stratification.
• This tool can aid clinicians in anticipating and managing CS needs in UC patients.