Structural Sensitivity in HIV Modeling: A Case Study of Vaccination

Area of Science:

• Epidemiology
• Mathematical Modeling
• Public Health

Background:

• Structural assumptions in infectious disease models, including network or compartmental types and individual heterogeneity, can heavily influence predictions.
• Understanding the impact of these assumptions is crucial for accurate modeling of disease dynamics and intervention effectiveness.

Purpose of the Study:

• To explore the implications of structural assumptions on human immunodeficiency virus (HIV) model predictions and policy conclusions.
• To assess the robustness of model inference by examining variations in parameter and simulation complexity.

Main Methods:

• Utilized eight related HIV transmission models, systematically varying parameter complexity (e.g., age, HCV comorbidity) and simulation complexity (aggregated compartmental, individual compartmental, network models).
• Conducted a case study on the effects of a hypothetical HIV vaccine across multiple population subgroups.

Main Results:

• Differences in HIV incidence reduction estimates between network and individual compartmental models were less significant than those between these and aggregated compartmental models.
• Aggregated models showed substantial overestimation of HIV incidence reductions compared to more disaggregated models.
• Age structure complexity buffered aggregation effects and influenced the vaccine effectiveness threshold for overestimation.

Conclusions:

• Structural assumptions, particularly the level of aggregation in compartmental models, profoundly affect HIV model predictions and vaccine impact assessments.
• Parameter complexity also influences cost-effectiveness estimates, but its role in differentiating network model projections is inconsistent.
• Robust assessment of structural assumptions is essential for reliable HIV modeling and informed public health policy.