Optimizing Initial Screening for Colorectal Cancer Detection with Adherence Behavior

Abstract

Background: Two-stage screening programs for colorectal cancer (CRC) detection typically involve a first-stage test that evaluates fecal-hemoglobin (f-Hb) concentration in stool samples, with positive results leading to a recommended second-stage diagnostic test (colonoscopy).

Methods: We explore the design of the first-stage test—specifically the selection of f-Hb cutoffs to report test outcomes—to balance screening effectiveness (CRC and polyp detection) and efficiency (colonoscopy costs), considering that not all individuals follow up with a colonoscopy. We propose an information design model that integrates Bayesian persuasion with information avoidance to address this issue. The model is applied to the design of Singapore’s CRC screening program and calibrated using data from multiple sources, including a nationwide survey of 3,920 respondents in Singapore.

Results: Our findings indicate that under certain conditions, using a single cutoff maximizes follow-up adherence, while showing exact f-Hb readings optimizes detection effectiveness. Raising the cutoff to 39 µg/g, as compared to current practices, could detect 21% more CRC and polyp cases, reduce colonoscopies by 27%, and lower lifetime CRC risk by 11%. This adjustment would lead to a reduction of public healthcare expenditure by S$20 million and individual spending by S$12 million on average in screening costs.

Conclusions: Choosing appropriate cutoffs for the first-stage test can significantly enhance the screening effectiveness while efficiently managing colonoscopy demands. The prevalent practice of using lower cutoffs to achieve high sensitivity may lead to excessive unnecessary colonoscopies and diminished follow-up adherence.

Note:
Funding Information: None to declare.

Declaration of Interests: None to declare.

Keywords: Cancer Screening, Cutoff Selection, Adherence, Bayesian Persuasion, Information Avoidance