,
5 tweets,
1 min read
Read on Twitter
Good passage from the guidelines of reporting for clinical research in urology (ncbi.nlm.nih.gov/pubmed/30537407): 
This is a common fallacy: within a group of patients that got a treatment, showing that a specific risk factor is associated with poorer outcome, and concluding that patients with that risk factor are therefore poor candidates for that treatment.
...in doing so, failing to appreciate that the treatment still may have improved the outcome for those patients (vs. them not getting the treatment).
...which is why assessing whether a treatment "works" or "doesn't work" for selected patient groups based only on data for patients that received the treatment is a bit of a fool's errand. It's forgetting basic experimental design.
e.g. "Patients with pulmonary hypertension have poorer outcomes in TAVR" to conclude "patients with pulmonary hypertension are poor candidates for TAVR" rather than comparing outcomes for AS patients with PH if they got TAVR versus an alternative (or nothing)
