Semiparametric outcome regression-based estimator of Mann-Whitney-type causal effect

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Observational studies often face challenges with confounding variables.
• Estimating causal effects requires robust statistical methods that account for complex data structures.
• Rank-based methods offer advantages in handling outcome transformations and confounding.

Purpose of the Study:

• Introduce a novel semiparametric estimator for Mann-Whitney-type causal effects.
• Develop and validate the cumulative probability model (CPM) for causal inference.
• Assess the causal effect of HIV status on albuminuria in a Nigerian cohort.

Main Methods:

• Developed a semiparametric estimator based on the cumulative probability model (CPM).
• Formalized estimation under standard causal assumptions: consistency, no interference, ignorability, and positivity.
• Designed accompanying inference procedures for the proposed estimator.
• Conducted simulations to evaluate performance against misspecified parametric models.
• Applied the method to a cohort of people with HIV (PWH) in Northern Nigeria.

Main Results:

• The CPM estimator demonstrated reduced variability and improved predictive accuracy in simulations.
• The method outperformed mis-specified parametric transformations under varying conditions.
• Successfully applied to assess the causal effect of HIV status on albuminuria levels.
• Highlighted the value of semiparametric methods for causal inference beyond average treatment effects.

Conclusions:

• The cumulative probability model (CPM) provides a valuable semiparametric approach for causal inference in observational data.
• This method offers flexibility and robustness, particularly in the presence of confounding.
• Findings underscore the utility of advanced statistical techniques for understanding health disparities, such as the impact of HIV on kidney function.