Effect of colorectal cancer screening on colorectal cancer surgery outcomes: nationwide cohort study.

Abstract

National colorectal cancer screening commenced in Denmark in 2014. Little is known about the effects of organized colorectal cancer screening on intraoperative and postoperative events. The aim of this nationwide cohort study was to evaluate the difference in intraoperative and postoperative outcomes between patients with screen-detected colorectal cancer and non-screen-detected colorectal cancer within the first 90 days after surgery. National register data were collected for Danish residents diagnosed with colorectal cancer between January 2014 and March 2018. Outcomes for the two cohorts were reported as relative risk or weighted mean difference. Intraoperative outcomes were blood loss, blood transfusion, tumour perforation, and organ lesion. Postoperative outcomes were complications (surgical and non-surgical) and 90-day mortality. Discrete data estimates were calculated from a general linear model. Analyses were adjusted for potential healthy user bias with respect to sex, age, location of the cancer (colon/rectum), and Charlson co-morbidity index. In total, 10 606 patients were included. Compared with patients in the non-screen-detected colorectal cancer group (4497 patients), patients in the screen-detected colorectal cancer group (6109 patients) had reduced intraoperative blood loss (-52 mL, 95% c.i. -67 to -37, P < 0.001), a shorter duration of hospitalization (-2.3 days, 95% c.i. -2.8 to -1.8, P < 0.001), and reduced rates of intraoperative organ lesion (0.76, 95% c.i. 0.59 to 0.99, P = 0.042), surgical complications (0.79, 95% c.i. 0.73 to 0.87, P < 0.001), non-surgical complications (0.68, 95% c.i. 0.60 to 0.78, P < 0.001), and 90-day mortality (0.29, 95% c.i. 0.21 to 0.39, P < 0.001). In comparison with non-screen-detected colorectal cancer, surgery for screen-detected colorectal cancer remains associated with improvement in several intraoperative and early postoperative outcomes after considering healthy user bias.