Core Concepts: Self-Controlled Designs in Pharmacoepidemiology.

Area of Science:

• Pharmacoepidemiology
• Biostatistics
• Observational Study Design

Background:

• Uncontrolled confounding is a major challenge in pharmacoepidemiological studies, arising from poorly measured, unmeasured, or unknown confounders.
• Traditional study designs often rely on external control groups, which can be susceptible to confounding.
• Self-controlled designs offer an alternative by comparing outcomes within the same individual over different time periods.

Purpose of the Study:

• To introduce readers to various self-controlled study designs used in pharmacoepidemiology.
• To provide an overview of the terminology, key publications, and practical implementation of self-controlled designs.
• To discuss recent developments and serve as a starting point for researchers interested in applying these methods.

Main Methods:

• Review of existing literature on self-controlled study designs.
• Annotated reference list highlighting key publications.
• Practical description of implementation and visualization with recent examples.

Main Results:

• Self-controlled designs control for all time-stable confounders and negate the need for external controls when time-varying confounding is absent.
• These designs rely on strong, not always verifiable, assumptions.
• The review provides a practical guide and discusses recent advancements in the field.

Conclusions:

• Self-controlled designs are valuable tools in pharmacoepidemiology for addressing confounding.
• Understanding their assumptions and implementation is crucial for accurate application.
• This review serves as a foundational resource for researchers exploring self-controlled methodologies.