Mathematical modelling to guide colorectal cancer screening and surveillance policies

This thesis consists of three main parts, each addressing different topics regarding the evaluation and improvement of CRC screening and surveillance. In the first part of this thesis, we describe possibilities to improve the CRC screening programme. In Chapter 2, we study the benefit-harm balance of participating in CRC screening for many subgroups aiming to help individuals making an informed decision about screening participation. To do so, we combine risk estimates of the benefits and harms of CRC screening, derived with the ASCCA model, with the relative importance of CRC screening outcomes obtained with a preference eliciting survey in order to obtain the benefit-harm balance of screening. In Chapter 3, we evaluate the clinical utility of a new stool test in a large-scale paired-intervention study conducted in the Dutch CRC screening programme. In Chapter 4, we assess the accuracy of various summarising measures commonly used to report adherence over multiple rounds of stool-based CRC screening. In addition, we assess the impact of using these summarising measures, rather than using detailed longitudinal adherence data, on model-predicted CRC screening effectiveness using the ASCCA model. The second part of this thesis investigates surveillance in two populations who are at increased risk of CRC. In Chapter 5, we evaluate whether stool-based surveillance could serve as an alternative to colonoscopy surveillance in a post-polypectomy surveillance population. In Chapter 6, we study the optimal surveillance strategy for individuals with a family history of CRC, considering colonoscopy surveillance, FIT-based surveillance and surveillance including both colonoscopy and FIT. The third part of this thesis focuses on the impact of the COVID-19 pandemic on CRC screening programmes. For this purpose, multiple independent models are used to answer the same research question. In addition to the ASCCA model, the MISCAN-Colon model, the Policy1-Bowel model, and the OncoSim model are used, which are developed for the Netherlands, Australia, and Canada, respectively. We study the short-term and long-term impact of hypothetical disruptions to CRC screening programmes in three countries, e.g. the Netherlands, Australia and Canada, in Chapter 7. In Chapter 8, we investigate two approaches for managing the screening backlog that results from a three-month screening disruption in the same three countries. The objective is to provide guidance on how to manage catch-up screening within available colonoscopy capacity. In Chapter 9, we estimate the global impact on CRC burden due to COVID-19 related decreases to organised CRC screening based on real-world data. To do so, all four models are used to draw conclusions for countries other than those for which the model was originally developed. As this provides an additional level of uncertainty, the results are aggregated across the multiple models to take this uncertainty into account. Chapter 10 summarizes the main findings presented in this thesis. Moreover, we discuss methodological issues and provide recommendations for future research.