Developing and Calibrating a Colorectal Cancer Microsimulation Model for Northern Ireland

Area of Science:

• Epidemiology
• Biostatistics
• Health Economics

Background:

• Individual-level microsimulation models are crucial for colorectal cancer (CRC) screening evaluation.
• Regional relevance necessitates detailed natural history and population-specific calibration.
• This study introduces the first CRC natural history microsimulation model for Northern Ireland (NI).

Purpose of the Study:

• To develop a CRC natural history microsimulation model tailored to Northern Ireland.
• To evaluate the Northern Ireland Bowel Cancer Screening Programme (NI BCSP).
• To provide a regionally specific platform for screening strategy assessment and policy support.

Main Methods:

• Simulated individual trajectories from adenoma onset to CRC diagnosis.
• Calibrated eight natural history parameters using empirical and Approximate Bayesian Computation (ABC) methods, including ABC-Markov Chain Monte Carlo (ABC-MCMC).
• Informed other parameters using NI-specific data.

Main Results:

• The frequentist and ABC rejection calibration informed the ABC-MCMC prior distribution.
• ABC-MCMC yielded 55 parameter sets but was constrained by limited calibration targets and parameter identifiability.
• Calibration was demonstrated as feasible within a data-limited context.

Conclusions:

• The developed model is the first NI-specific CRC microsimulation tool.
• It offers a tailored platform for evaluating screening strategies and informing policy.
• Further refinement and additional calibration targets are recommended to enhance parameter estimation.