Read on Twitter
Cool study, & #39;Inferences about drug safety in phase 3 trials in oncology: Examples from advanced prostate cancer& #39; in @JNCI_Now from @JoshDragoMD @gonen_mithat et al 1/ https://academic.oup.com/jnci/advance-article/doi/10.1093/jnci/djaa134/5898636#.X0-1lcZ0WyJ.twitter">https://academic.oup.com/jnci/adva...
Authors took 3 prostate cancer #RCTs where patients had similar baseline characteristics, and they compared adverse event rates in placebo arm 2a/
An aside: LOTSA great meta-research on clinical trials to be done by throwing away the experimental arm and just analyzing placebo data... but I digress 2b/
What they find is a dog& #39;s breakfast of heterogeneity. Like this: & #39;risks for grade 3–4 AEs showed high variability [in placebo arms], with differences up to 3.6-fold between trials.& #39; 3/
& #39;No trial reported time-to-event analyses, or any other inferential analyses that would have allowed for a valid determination of whether elevated relative risks were causally related to the drug, length or frequency of follow-up between drug and placebo, or chance& #39; 4/
Authors conclude: & #39;No substantiated, quantitative conclusions about the relative toxicity of these drugs could be drawn based on published data& #39;
Given that decisions to use one drug or another are often based on safety, how is it these problems persist in 2020? 5/
Given that decisions to use one drug or another are often based on safety, how is it these problems persist in 2020? 5/
One hint is suggested in historian David Jones& #39; outstanding book about cardiology, & #39;Broken Hearts.& #39; Medical innovation systematically discounts and overlooks safety, unless of course... 6/
A new rival technology comes along, in which case differences in safety become a focal concern- esp for the motivated party. So why might this not be operative in prostate cancer, where there are many rival products? Intellectual collusion? Who knows... 7/
Authors point to factors that might explain heterogeneity: a) differences in appointment frequency (more frequent visits = > AEs); b) inter-rater variability for AEs; c) geographic variation in populations- almost always not reported 8/
I submit that the variability and opacity of AE event reporting is one of the most important, easily addressable ethical failings in current trial practice. Why do a trial if we can& #39;t evaluate a Rx& #39;s risk/benefit balance?
Does not get moral attention it deserves. Why? 9/
Does not get moral attention it deserves. Why? 9/
Thanks to @JNCI_Now for publishing work that asks how we can better serve patients by improving the design and reporting of trials END/