Predictive model may assist patients, providers in decision making for advanced neoplasia
Click Here to Manage Email Alerts
A predictive model for advanced colorectal neoplasia may help increase colorectal cancer screening adherence, according to a study published in Gut.
“The study examined whether and how well phenotypic factors (socio-demographic factors, lifestyle factors, and personal and family medical history) can separate risk for advanced colorectal neoplasia among average-risk persons due for colorectal cancer screening,” Thomas Imperiale, MD, from the Regenstrief Institute and Indiana University School of Medicine, told Healio Gastroenterology. “The results suggest that the risk prediction model derived and split-sample validated model achieves very good discrimination in risk for advanced neoplasia, information that may be useful to patients and their providers when discussing which test or strategy to choose for first-time screening.”
Imperiale and colleagues identified patients who underwent first-time screening colonoscopy. They measured sociodemographic and physical features, medical and family history and lifestyle factors. Researchers stratified scores with comparable risks into risk categories.
Results showed advanced colorectal neoplasia prevalence was 9.4% in 3,025 patients from the derivation set. Investigators reported the 13-variable model produced risk advanced colorectal neoplasia groups comprising 1.5% (95% CI, 0.72% to 2.74%) in the low-risk group, 7.06% (95% CI, 5.89% to 8.38%) in the intermediate-risk group and 27.26% (95% CI, 23.47% to 31.30%) in the high-risk group (P < .001).
The validation set included 1,475 patients with an advanced colorectal neoplasia prevalence of 8.4%.
“In the validation set of 1,475 subjects (AN prevalence of 8.4%), model performance was comparable (c-statistic = 0.78), with AN risks of 2.73% ([95%] CI, 1.25% to 5.11%), 5.57% ([95%] CI, 4.12% to 7.34%) and 25.79% ([95%] CI, 20.51% to 31.66%) in low-risk, intermediate-risk and high-risk subgroups, respectively (P < .001), containing proportions of 23%, 59% and 18%,” Imperiale and colleagues wrote.
Collapse
Read more about
Click Here to Manage Email Alerts