Identification of causal intervention effects under contagion

Area of Science:

• Epidemiology
• Biostatistics
• Infectious Disease Modeling

Background:

• Intervention effects for transmissible diseases are complex due to indirect impacts on others.
• Existing two-person partnership models often oversimplify transmission dynamics.

Purpose of the Study:

• To propose novel causal intervention effects for two-person partnerships under general transmission dynamics.
• To provide nonparametric identification results for estimating these effects in empirical trials.

Main Methods:

• Developing causal estimands for infectious disease transmission within partnerships.
• Utilizing time-to-infection or binary outcome data for estimation.
• Leveraging conditional independencies induced by contagion for identification.

Main Results:

• Proposed causal estimands are applicable under arbitrary infectious disease transmission dynamics.
• Nonparametric identification strategies are demonstrated for empirical data.
• The approach is illustrated with a realistic HIV vaccine trial simulation.

Conclusions:

• Contagion is a causal phenomenon that can be exploited to identify meaningful causal estimands.
• The proposed methods enhance the conceptual usefulness of partnership models for real-world intervention trials.
• This work provides a framework for evaluating interventions in the presence of disease transmission.