Nesting a Target Study within a Target Trial: A Framework for Evaluating Intervention Effects on Disparities

Area of Science:

• Causal inference
• Health equity research
• Epidemiology

Background:

• Evaluating interventions on health disparities requires robust causal methods.
• Existing frameworks may not adequately address ethical considerations or ensure valid disparity measurement.
• There is a need for methods that can generate ethically informed causal evidence to reduce health inequities.

Purpose of the Study:

• To present a novel framework (TS+TT) for nesting a target study (TS) within a target trial (TT) to evaluate intervention effects on disparities.
• To provide an ethically grounded approach for measuring disparities and anchoring causal inference to specific populations and timeframes.
• To demonstrate the framework's application in assessing interventions aimed at reducing disparities in clinical treatment receipt.

Main Methods:

• The TS+TT framework combines a target study for ethical disparity measurement with a target trial for intervention randomization.
• It employs stratified sampling to ensure distributional similarity of social groups on allowable covariates.
• Semiparametric G-computation is extended to handle continuous stochastic interventions and estimate counterfactual disparities.

Main Results:

• The TS+TT framework provides a meaningful causal estimand for evaluating intervention impacts on disparities.
• Demonstrated application shows potential for assessing interventions on pulse oximeter bias and treatment disparities.
• The extended G-computation method allows for estimation of counterfactual disparities in time-to-event outcomes.

Conclusions:

• The TS+TT framework offers a versatile and policy-relevant approach for ethically evaluating interventions on disparities.
• It generates ethically informed causal evidence crucial for reducing health inequities.
• The framework helps avoid exacerbating existing disparities by providing a sound basis for intervention design and evaluation.