Harvard Professor Robert Yeh discusses the challenges and importance of determining the credibility of observational studies in cardiology. They explore the use of causal language in research and the need to differentiate between association and causality. The chapter also discusses the difficulties of analyzing observational data, avoiding biases like immortal time bias, and advancing observational research in medicine by borrowing techniques from adjacent fields.
Observational comparison studies in cardiology should aim to simulate a randomized clinical trial to ensure credibility.
Collaboration between clinical experts and causal inference experts is crucial for improving the design and analysis of observational studies in cardiology.
Deep dives
The Importance of Credibility in Observational Research in Cardiology
In this podcast episode, Bobby Yay, a professor of medicine at Harvard and a cardiologist, discusses the importance of credibility in observational research in cardiology. Yay emphasizes the need to raise the bar for the quality of observational studies and improve their credibility. He acknowledges that while observational research is necessary due to the inability to conduct randomized trials for every question, there are challenges in determining which studies are credible. Yay advocates for a closer collaboration between clinical experts and causal inference experts to improve the design and analysis of observational studies.
Understanding Observational Studies and Their Limitations
Yay provides an overview of observational studies and explains how they differ from randomized trials. He highlights that observational studies analyze real-world practice, while randomized trials involve carefully designed experiments. Yay acknowledges that most clinical decisions rely on observational studies since conducting randomized trials for every question is not feasible. However, he points out the challenge of confounding, where various factors can influence treatment decisions and outcomes, affecting the reliability of observational studies.
The Problem of Confounding in Cardiovascular Observational Research
Yay discusses the crux of the problem in cardiovascular observational research, which is confounding. Confounding occurs when factors influencing treatment decisions are also connected to the outcome of interest. Yay highlights that these factors can make it challenging to determine whether treatment or patient differences are responsible for observed outcomes. He provides examples such as the selection bias in a heart attack study and immortal time bias. Yay emphasizes the importance of addressing confounding issues and striving for credible observational research.
Emulating Target Trials and Improving Observational Study Quality
Yay introduces the concept of target trial emulation, which involves designing observational studies to mirror theoretical randomized trials. This approach helps researchers define time zero, treatment assignment, and other essential parameters clearly. He highlights the need for clinical experts and causal inference experts to collaborate in designing and analyzing observational studies. Yay also mentions the RCT duplicate project, which attempted to emulate randomized trials using non-randomized data, showing promise but leaving room for improvement. He calls for embracing advances in adjacent fields and adopting more rigorous techniques to enhance the quality of observational studies.
Sensible Medicine is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
I learned a lot from this conversation. One of the main lessons is that no matter how well the authors avoid causal language, the intent of an observational comparison study is causal. And if that is so, the main thrust of these efforts ought to be simulate, as close as possible, a randomized clinical trial.
One of my favorite parts of our chat was Bobby’s now famous explanation of immortal time bias using Cheetos.
Let us know what you think. I hope to do more of these types of conversations.
JMM
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.sensible-med.com/subscribe
Get the Snipd podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share & Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered podcast player
Listen to all your favourite podcasts with AI-powered features
Discover highlights
Listen to the best highlights from the podcasts you love and dive into the full episode