OBJECTIVES: Identifying ulcerative colitis (UC) patients with increased risk of colectomy is essential for appropriate treatment. We aimed to develop a prediction model assessing the risk of having colectomy within the first 10 years after diagnosis. MATERIAL AND METHODS: A population-based inception cohort of UC patients diagnosed in south-eastern Norway between 1990 and 1994 has been followed for 10 years. Altogether 519 patients were recruited including 49 patients who were colectomized. Based on the best-fitted multivariate model, the probabilities of colectomy were computed for selected levels of baseline covariates, and the results arranged in a prediction matrix. The following risk factors at diagnosis were analyzed: age, smoking, sex, disease extent, weight loss and fever and need for systemic steroids. Biochemical markers included C-reactive protein (CRP, <30 or ≥30 mg/l); erythrocyte sedimentation rate (ESR, <30 or ≥30 mm/h) and hemoglobin (Hgb, <10.5 or ≥ 10.5 g/dL). RESULTS: Extent of disease, age (<40 years, ≥40 years), need for systemic steroids and CRP or ESR (<30 or ≥30) at diagnosis were independently associated with colectomy and were combined in a prediction matrix. The probabilities of colectomy during the follow-up period ranged from 2.6% to 40.1% depending on the combination of predictors at diagnosis. CONCLUSIONS: Our prediction model revealed significant differences in the probability of undergoing colectomy during a 10-years course of disease, which supports an early individualized treatment approach in UC.
OBJECTIVES: Identifying ulcerative colitis (UC) patients with increased risk of colectomy is essential for appropriate treatment. We aimed to develop a prediction model assessing the risk of having colectomy within the first 10 years after diagnosis. MATERIAL AND METHODS: A population-based inception cohort of UC patients diagnosed in south-eastern Norway between 1990 and 1994 has been followed for 10 years. Altogether 519 patients were recruited including 49 patients who were colectomized. Based on the best-fitted multivariate model, the probabilities of colectomy were computed for selected levels of baseline covariates, and the results arranged in a prediction matrix. The following risk factors at diagnosis were analyzed: age, smoking, sex, disease extent, weight loss and fever and need for systemic steroids. Biochemical markers included C-reactive protein (CRP, <30 or ≥30 mg/l); erythrocyte sedimentation rate (ESR, <30 or ≥30 mm/h) and hemoglobin (Hgb, <10.5 or ≥ 10.5 g/dL). RESULTS: Extent of disease, age (<40 years, ≥40 years), need for systemic steroids and CRP or ESR (<30 or ≥30) at diagnosis were independently associated with colectomy and were combined in a prediction matrix. The probabilities of colectomy during the follow-up period ranged from 2.6% to 40.1% depending on the combination of predictors at diagnosis. CONCLUSIONS: Our prediction model revealed significant differences in the probability of undergoing colectomy during a 10-years course of disease, which supports an early individualized treatment approach in UC.