Estimating dengue transmission intensity from serological data: A comparative analysis using mixture and catalytic...

Area of Science:

• Epidemiology
• Infectious Diseases
• Biostatistics

Background:

• Dengue virus (DENV) infection is a significant global health issue.
• Accurate estimation of the force of infection (FOI) is crucial for effective DENV intervention strategies.
• Traditional catalytic models for DENV FOI estimation rely on antibody thresholds, which can lead to misclassification and biased results.

Purpose of the Study:

• To compare the performance of mixture models against traditional catalytic models (time-constant and time-varying) for estimating DENV FOI.
• To evaluate the application of mixture models using both simulated and real-world serological data.

Main Methods:

• Utilized simulated data to compare bias and coverage of FOI estimates from mixture models and catalytic models.
• Applied mixture models and catalytic models to serological data from Vietnam (2004-2009) and Indonesia (2014).
• Assessed the impact of antibody titre distributions on model performance.

Main Results:

• Simulation studies indicated that mixture models produced less biased FOI estimates compared to catalytic models.
• Both time-varying catalytic and mixture models achieved >95% coverage of the true FOI, with mixture models showing reduced uncertainty.
• Application to Vietnamese data revealed that mixture models often yielded higher FOI and seroprevalence estimates than catalytic models.

Conclusions:

• Mixture models are a valid and potentially less biased alternative to catalytic models for DENV FOI estimation.
• These models are especially beneficial when dealing with serological datasets exhibiting largely overlapping antibody titre distributions.
• The findings support the broader adoption of mixture models in DENV epidemiological research.