Optimally capturing latency dynamics in models of tuberculosis transmission.

Area of Science:

• Epidemiology
• Mathematical Modeling
• Infectious Disease Dynamics

Background:

• Modern tuberculosis (TB) models utilize various structures to simulate TB latency.
• The ability of these structures to accurately reproduce empirical TB activation dynamics is not well-understood.

Purpose of the Study:

• To identify which TB model structures accurately replicate observed TB activation dynamics.
• To compare parameter estimates from validated models with those used in prior studies.

Main Methods:

• A systematic review of 88 TB modeling articles was conducted to classify latency structures.
• Six identified model structures were fitted to activation data from 1352 infected contacts in Australia and the Netherlands.
• Parameter estimation was performed to assess model fit and identify significant differences.

Main Results:

• Only TB models incorporating two distinct latency compartments accurately reproduced observed activation dynamics.
• Significant age-related differences in parameter estimates were observed.
• Parameter estimates differed markedly from those in previous models, particularly the duration of the initial latency phase.

Conclusions:

• TB models simulating latency require two compartments and age-stratification for accurate dynamic replication.
• The initial latency phase duration in two-phase models should be shorter than previously assumed.
• The study provides a catalog of parameter values and an estimation approach for future TB model calibration.