CAUSAL INFERENCE FROM OBSERVATIONAL STUDIES WITH CLUSTERED INTERFERENCE, WITH APPLICATION TO A CHOLERA VACCINE STUDY

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Observational studies are crucial for vaccine effectiveness research but face challenges like non-randomization and interference.
• Interference occurs when vaccinating one person affects another's risk of disease, complicating effect estimation.
• Existing methods often assume independent treatment choices, which may not hold in clustered populations.

Purpose of the Study:

• To propose novel causal estimands for vaccine effects in observational studies with interference.
• To develop methods that account for within-cluster dependence in vaccination decisions.
• To provide tools for policymakers to quantify population-level vaccine impacts.

Main Methods:

• Developed new causal estimands addressing within-cluster dependence in treatment selection.
• Proposed inverse probability-weighted estimators for the new causal estimands.
• Derived large-sample properties of the proposed estimators.
• Conducted a simulation study to evaluate finite-sample performance.

Main Results:

• The proposed causal estimands allow for more relevant quantification of vaccine effects in the presence of interference.
• Inverse probability-weighted estimators demonstrated good performance in simulations.
• The methods were successfully applied to a large cholera vaccination study in Bangladesh.

Conclusions:

• New causal estimands and estimation methods are effective for analyzing vaccine effects in observational studies with interference.
• These methods offer valuable insights for public health policy and decision-making.
• The approach is applicable to various infectious diseases where interference is a concern.