Overcoming time-varying confounding in self-controlled case series with active comparators: application and...

Area of Science:

• Pharmacoepidemiology
• Biostatistics

Background:

• Confounding by indication is a significant challenge in pharmacoepidemiology.
• Self-controlled study designs effectively handle time-invariant confounding.
• Indications for medication use can change over time, posing a challenge for traditional self-controlled designs.

Purpose of the Study:

• To formalize and describe methods for incorporating active comparators into self-controlled studies.
• To address time-varying confounding by indication in pharmacoepidemiologic research.
• To enhance the robustness of self-controlled case series studies.

Main Methods:

• Formalization of methods using active comparators in self-controlled case series.
• Derivation of the active comparator rate ratio via explicit regression coefficient comparison or nested regression models.
• Application and comparison of methods in two case studies using the UK's Clinical Practice Research Datalink.

Main Results:

• Demonstrated approaches for deriving active comparator rate ratios in self-controlled case series.
• Evaluated methods in studies on thiazolidinedione and fractures, and fluoroquinolone and uveitis.
• Provided a framework for applying these advanced statistical techniques.

Conclusions:

• The formalized methods offer a robust approach to mitigate time-varying confounding by indication.
• Active comparators enhance the validity of self-controlled case series in pharmacoepidemiology.
• Recommendations are provided to support the design, execution, and interpretation of these studies.