Brendon Stiles, Professor and Chief, Thoracic Surgery and Surgical Oncology at Montefiore Health System, shared a post on X:
“Have to admit that I am a bit disappointed watching many RadOnc like, retweet, and embrace this study without pointing out critical flaws – particularly after the whole hullabaloo with the IRRADIaTE radiation complications paper. Pretty classic example of confirmation bias…
- Cohort used non-parallel data sources (prospective vs retrospective; UK-heavy vs international registry)
- Spanned different eras of care and center and geographic effects not obviously controlled
- Matching variables may not include the “real drivers” of treatment selection
- The “sRP” arm is also heterogeneous and skewed toward open surgery -which has known higher comps
- And most relevant for the most “significant” finding – the complication window is asymmetrical for the clinical question. Silly to compare short term comps between these treatments
“Both effective” could be overreach from observational matched data. Conclusion that both are effective and sFT has “favorable therapeutic ratio” may be reasonable as hypothesis, but causal language is overstated given residual confounding and the periop only toxicity focus.
ut sadly I don’t think this paper will get that same type of in depth analysis by those who have already made up their mind. Admittedly, surgeons have confirmation bias too. We should all try to recognize it.”
Other articles featuring Brendon Stiles on OncoDaily.