A deep dive into principal stratification and causal inference

In this episode, I talk with 2 experts from Novartis and Roche. We cover the following questions:

• What is Principal Stratification?
• How would you describe principal stratification to a non-statistician?
• Where do you see the benefits of this estimand compared to the other typical strategies?
• Which critique points do are usually raised against this approach?
• How do you implement/calculate corresponding estimates for this estimand?
• What references would you recommend for further reading?

Björn and Kaspar recommend the following very useful references:

References:

• Books:
  • Introduction into potential outcomes and causal inference: https://www.cambridge.org/core/books/causal-inference-for-statistics-social-and-biomedical-sciences/71126BE90C58F1A431FE9B2DD07938AB
  • Hernan and Robins: https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/ 
  • Pearl, the book of why: https://www.amazon.de/Book-Why-Science-Cause-Effect/dp/046509760X
• Papers:
  • Paper draft: https://arxiv.org/abs/2008.05406 with markdown: https://oncoestimand.github.io/princ_strat_drug_dev/princ_strat_example.html and github: https://github.com/oncoestimand/princ_strat_drug_dev
  • Magnusson et al (Siponimod Beispiel): https://onlinelibrary.wiley.com/doi/full/10.1002/sim.8333 and the corresponding EPAR: https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-evaluation-anticancer-medicinal-products-man-revision-5_en.pdf
• Oncology estimand working group: http://www.oncoestimand.org/