Understanding Incremental Propensity Score Approaches: A Guide for Pharmacoepidemiologists

By
Byeong Yeob Choi
September 24, 2026

Pharmacoepidemiology And Drug Safety

Background

Causal inquiries in pharmacoepidemiology often rely on average treatment effects (ATEs) derived from propensity score (PS) methods. However, violations of positivity can lead to significant bias in these estimates, complicating the interpretation of treatment effects. Understanding and addressing these challenges is crucial for accurate causal inference in observational studies.

Data Highlights

No numerical data or trial data provided in the source material.

Key Findings

IPSIs allow for the estimation of treatment effects by adjusting each patient's baseline treatment probability.
Conventional PS methods may produce biased estimates when positivity assumptions are violated.
Parametric bootstrap methods can assess bias due to positivity violations.
Four traditional PS approaches include matching, stratification, IPTW, and covariate adjustment.

Clinical Implications

The application of IPSIs can enhance the validity of causal estimates in pharmacoepidemiology, particularly in settings where treatment assignment is influenced by complex factors. This approach may inform clinical guidelines and prescribing practices more effectively.

Conclusion

IPSIs represent a promising alternative to traditional propensity score methods, addressing the limitations posed by positivity violations in observational studies.