Balancing scores and causal diagrams.

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Propensity scores (PS) estimate the probability of exposure given confounders.
• Disease risk scores (DRS) estimate disease probability given confounders and fixed exposure.
• Both PS and DRS are balancing scores crucial for confounding adjustment, particularly with high-dimensional confounders.

Purpose of the Study:

• To elucidate the function of PS and DRS in confounding adjustment.
• To utilize causal diagrams for a graphical representation of these concepts.

Main Methods:

• Employing causal diagrams to illustrate the theoretical properties of PS and DRS.
• Demonstrating the application of PS and DRS in statistical analysis for confounding adjustment.

Main Results:

• Graphically depicting the balancing properties: conditional on PS, exposure is independent of confounders.
• Showing that conditional on DRS, potential outcomes under no exposure are independent of confounders.
• Illustrating practical methods for using PS and DRS in analyses.

Conclusions:

• Causal diagrams offer a clear visualization for understanding confounding adjustment.
• Researchers can enhance their comprehension of PS and DRS utility through causal diagrams.