Inference About Separable Causal Effects With Longitudinal Bivariate Ordinal Responses With Missingness and Censoring

Area of Science:

• Statistics
• Biostatistics
• Epidemiology

Background:

• Causal inference methods primarily focus on univariate response variables.
• Longitudinal studies present unique challenges for causal inference, including missing data and censoring.
• Handling bivariate response variables in causal inference requires specialized approaches.

Purpose of the Study:

• To develop a novel framework for causal inference with bivariate responses in longitudinal studies.
• To address the complexities of missingness and censoring in bivariate causal inference.
• To provide a transparent interpretation of treatment effects on multiple outcomes.

Main Methods:

• Decomposed treatment framework to separate overall treatment effects into effects on individual responses.
• Identification of separable treatment effects using observed data under specified conditions.
• Likelihood-based estimation and hypothesis testing for separable treatment effects.

Main Results:

• The proposed decomposed treatment framework allows for the identification of separable treatment effects.
• The sum of separable treatment effects is shown to equal twice the overall treatment effects.
• The methods were validated through real data analysis and simulation studies.

Conclusions:

• The novel framework effectively handles causal inference with bivariate responses in longitudinal data, even with missingness and censoring.
• The decomposed treatment approach offers enhanced interpretability compared to traditional methods.
• The proposed methods demonstrate practical utility and effectiveness in real-world and simulated scenarios.