Simulation models in population breast cancer screening: A systematic review

Area of Science:

• Oncology
• Health Services Research
• Biostatistics

Background:

• Simulation models are crucial for evaluating breast cancer screening programs.
• Existing models have been widely used to inform policy and decision-making.
• Critical evaluation is needed to assess the reliability and validity of these models.

Purpose of the Study:

• To critically evaluate published simulation models for general population breast cancer screening.
• To identify limitations and provide direction for future breast cancer modeling.
• To assess the accuracy of predicted mortality reduction and cost-effectiveness.

Main Methods:

• Systematic literature search to identify relevant simulation models.
• Qualitative assessment framework including model type, parameters, approach, validation, and outcomes.
• Comparison of model-predicted mortality reduction and cost-effectiveness against meta-analyses of randomized controlled trials.

Main Results:

• Seven simulation models were identified, primarily based on tumor progression.
• Models showed internal and cross-validation but lacked external validation.
• Predicted mortality reduction was overestimated (11-24%) compared to randomized controlled trials (10%).
• Potential harms of screening are only recently reported.
• Most scenarios demonstrated acceptable cost-effectiveness.

Conclusions:

• Published breast cancer screening simulation models exhibit a high risk of bias.
• There is a critical need for externally validated models using systematic evidence.
• Improved model validation is essential for accurate evaluation of breast cancer screening effectiveness and harms.