This unfounded claim utterly scorches the reputation of the FDA at a critical time — what a complete embarrassment. https://twitter.com/us_fda/status/1297662384060981248
The 35% figure is apparently drawn, per @EricTopol and others, from the Mayo Clinic’s single-arm / non-randomized / non-placebo controlled observational study, which hence tells us little. But I’ve read the paper and I think there are additional concerns

https://www.medrxiv.org/content/10.1101/2020.08.12.20169359v1.full.pdf
(The 35% relative reduction appears to reflect the relative difference in mortality for those who received plasma with a low dose of antibodies (the putative active ingredient) compared to a high dose: 13.7% (low dose) —> 8.9% (high dose) = 35% relative reduction in mortality).
Regardless, in the paper, the authors’ argue that the antibody levels of the plasma are pseudo-randomized, because nobody knew them at time of transfusion. But the groups were not balanced: the low-antibody group appears to be sicker.
Here’s the relevant table.
1. % with P/F ratio < 300 was 35.9% in low-antibody recipients vs 27.3% in high-antibody recipients

2. % with 5+ severe risk factors was 25.5% in low-dose group versus 19.0% in high-dose group.

They were also older (p>0.05).
The significant difference in % with a P/F ratio < 300 — the threshold for a diagnosis of ARDS — concerns me. Controlling for that post-facto is inadequate. That’s why we - and the FDA historically - has to require randomized trials to determine drug efficacy.
Clinician observations and patient testimonials also add little. If a patient improves after receipt of plasma, it is impossible to know if it was because of the plasma or in spite of it, or if it made no difference at all — because many patients improve.
Plasma may or may not work, it could be helpful or harmful. But we don’t know yet, even though we could have if we had done things right: if we had carried out a large, national randomized trial instead of an expanded access program that sheds little light on efficacy.
You can follow @awgaffney.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: