Developing a score chart to improve risk stratification of patients with colorectal adenoma.

ABSTRACT

BACKGROUND AND STUDY AIMS: Current surveillance guidelines risk stratify patients with adenoma by using only one or two factors adenoma multiplicity or presence of an advanced adenoma characteristic. Combinations of adenoma characteristics are not considered, which limits the predictive value of these guidelines. The aim of the study was to develop a scoring system for more refined risk stratification of patients with adenoma. PATIENTS AND METHODS: The Dutch Pathology Registry (PALGA) was used to identify newly diagnosed patients with adenoma in 10 Dutch hospitals between 1988 and 2002. Medical records were reviewed until 1 December 2008 for follow-up. Logistic regression analysis was used to assess patient- and adenoma-related predictors of metachronous advanced neoplasia. The prediction model was validated by bootstrapping and cross-validation. A score chart was developed based on identified adenoma-related predictors. The discriminative ability of the prediction model was compared with currently used risk stratifications in surveillance guidelines. RESULTS: A total of 2914 patients with adenoma were included (mean age 61 years; 55 % male). The score chart consisted of characteristics that contributed 1 point (size ≥ 10 mm, villous histology, proximal location, having 2 - 4 adenomas) or 2 points (having ≥ 5 adenomas). A patient's adenoma risk score could range from 0 to 5 points. A score of 5 for a 75-year-old man implied a 5-year risk of advanced neoplasia of 46 %. The discriminative ability of the model was moderate (c-statistic 0.712) but better than risk stratifications in current international guidelines, which had c-statistics of 0.642 - 0.674. CONCLUSION: A score chart that combines adenoma-related predictors of advanced colorectal neoplasia optimized the risk stratification of patients with adenoma for appropriate surveillance colonoscopy intervals.