SW-3C - How should we interpret self-controlled study designs in the era of causal inference?

Background: Self-controlled designs, such as self-controlled case series (SCCS) and case-crossover (COO), inherently control time-invariant factors, making them a powerful yet underutilized tool in causal inference. Properly integrating these methods into broader causal frameworks can significantly strengthen real-world evidence generation.

Objectives: To explore:
1. The principles of self-controlled designs, and their applications.
2. The unique advantages these study designs.
3. How these designs align—or diverge—from the target trial emulation framework and their role in estimating causal effects.
4. Methodological and communication challenges, including issues of time-varying confounding and other complexities, when adopting these designs for causal inference.
5. The implications of self-controlled designs for generating RWE, including recent innovations and approaches.

Description: 1. Overview of Self-Controlled Designs
Speaker: Andrew Yuen, UCL
This presentation will introduce SCCS and CCO designs, emphasising their methodological foundations, strengths, and limitations. Examples of their applications will be highlighted. (10 min)

2. Unique advantages and applications of self-controlled approaches
Speaker: Professor Edward Lai, NCKU
In this session, we will delve deeper into the strengths of self-controlled designs. This session will cover published studies underscore the versatility of self-controlled approaches and how they can tailor to different research questions. (15 min)

3. Self-Controlled Designs in Causal Inference
Speaker: Professor Ian Douglas, LSHTM
This session will consider the interpretation of self-controlled designs in causal inference frameworks, particularly their compatibility with the target trial emulation framework. Differences in research questions versus other causal methods will be discussed.(15 min)

4. Challenges and Limitations in understanding and communication
Speaker: Dr Ju-Young Shin, SKKU
An exploration of practical and methodological challenges, including time-varying confounding, reverse causality, and the nuances of adapting self-controlled designs to causal inference. Limitations of the study designs will also be addressed. (15 min)

5. Future Directions and Innovations
Speaker: Dr Kenneth Man, UCL
This presentation will highlight emerging advancements, such as hybrid designs that combine SCCS with other methods, novel analytical approaches, and innovations in real-world settings. Insights into how these advancements may overcome current limitations will be provided. (15 min)

Panel Discussion and Audience Q&A (20 min)
Moderator: Dr Kenneth Man, Mr Andrew Yuen
Panel members: Dr Cindy Zhou, Dr Sudha Raman, and all speakers