Simulation-based validation of a method to detect changes in SARS-CoV-2 reinfection risk

Area of Science:

• Epidemiology
• Mathematical Modeling
• Infectious Disease Dynamics

Background:

• High global seroprevalence of SARS-CoV-2 necessitates understanding reinfection risks.
• Models tracking reinfection trends must be robust against data biases.

Purpose of the Study:

• To perform simulation-based validation of a catalytic model for detecting changes in SARS-CoV-2 reinfection risk.
• To assess model robustness against biases from imperfect data observation and mortality.

Main Methods:

• Simulated primary and reinfection datasets based on South African SARS-CoV-2 epidemic data.
• Incorporated biases like imperfect observation and mortality into simulations.
• Employed a Bayesian approach with a negative binomial distribution to fit the catalytic model.

Main Results:

• The catalytic model successfully detected changes in reinfection risk when simulated.
• Model parameters converged in most scenarios, aligning with anticipated outcomes.
• Low observation probabilities (10%) led to poor convergence and low observed cases.

Conclusions:

• The catalytic model is robust to imperfect observation and mortality in most scenarios.
• Model performance may differ in settings with smaller epidemics; further validation is recommended.
• Ensuring model parameter convergence is crucial to avoid false positives in detecting reinfection risk shifts.