Background
In 2019, a BMJ Rapid Recommendation panel judged that for adults with a predicted 15-year risk of CRC below 3%, the benefits of colorectal cancer (CRC) screening would not outweigh the burdens and potential harms. The panel recommended against CRC screening for these individuals. However, as the number of individuals with a risk below 3% is unknown, the population-level impact of this recommendation is uncertain. For example, a majority of CRC cases might be missed if a majority of the population is classified as low-risk and is discouraged from screening (Geoffrey Rose’s prevention paradox). Our study estimated this population-level impact for Switzerland as a case study.

Methods
First, using the recommended QCancer risk calculator, we predicted the CRC risk of all individuals in the population-based Swiss Health Survey. To derive the distribution of risk-based screening start age, we assumed that predicted risk was calculated every 5 years between ages 25-70 and that screening started when this risk exceeded 3%. Second, the MISCAN-Colon microsimulation model evaluated biennial Fecal Immunochemical Test (FIT) screening with this risk-based start age. As a comparison, we simulated screening initiation based on age and sex only, and risk-based screening initiation using a hypothetical, better risk prediction tool with an Area Under the Curve (AUC) of 0.84 instead of QCancer.

Results
Starting screening only when QCancer-predicted CRC risk exceeded 3% meant 82% of females and 90% of males would not start screening before age 65 and 60, respectively (Figure 1). Table 1 shows that this would require 57% fewer FITs for females and 43% fewer FITs for males, compared to the current Swiss recommendation to start CRC screening at age 50. However, it would also yield an 8-14% smaller mortality reduction and 19-32% fewer lifeyears gained by screening. Screening all females from age 65 and all males from age 60 had a similar impact as screening only when QCancer-predicted risk exceeded 3%. Using a hypothetical, better risk prediction tool, a 3% threshold would still decrease mortality reduction and lifeyears gained with 17-34%, although reducing the number of FITs by 78-87% compared to current recommendations in Switzerland.

Conclusions
With the currently recommended QCancer risk calculator, the BMJ Rapid Recommendation would lead to a 10-15 years average delay in screening initiation. Although screening burdens would be halved, lifeyears gained by screening would be reduced substantially compared to current screening recommendations, even with a hypothetical risk prediction tool with increased AUC. This suggests that the 3% 15-year risk threshold to start screening might be too high, and that one could consider this to be a case of Geoffrey Rose’s prevention paradox. We believe these findings generalizable to other Western countries.

Figure 1 – Distribution of the age of screening initiation if screening starts when 15-year risk of CRC as predicted by the QCancer tool exceeds 3%. Predicted risk is recalculated every 5 years.

Table 1 – Population-level impact of 3 CRC screening start criteria compared to the current Swiss recommendation to start screening at age 50: 1) The BMJ Rapid Recommendation using the QCancer risk calculator for risk prediction CRC; 2) screening from age 60 (males) or 65 (females); 3) the BMJ Rapid Recommendation using a hypothetical, better tool for risk prediction.
CRC = Colorectal Cancer, FIT = Fecal Immunochemical Test

Florian Froehlich

Appearances