Looking through the @pfizer data submitted to #vrbpac now. Fascinating stuff here. Thread. 1/
Baseline characteristics. Not bad. ~10% Black. Would have liked to see more >=75 years old. 2/
Efficacy overall (that's the 95% you keep hearing about) and stratified by age group.
Looks similar whether > or < 55 years of age. 3/
More subgroups. Looks efficacious across multiple slices of the data (though smaller numbers make the precision of these estimates low). 4/
DAMN. Red line is cumulative cases in placebo. Blue is cumulative in vaccine group. Separation at 14 days which, interestingly, is BEFORE dose 2. But don't get carried away, the overall benefit occurs here after dose 2. Still - might be partial protection with 1 dose. 5/
What about prevention of SEVERE COVID-19. Only 3 cases in placebo group, 1 in vaccine group - so overall efficacy 66% with a gigantic confidence interval. Don't be disheartened - this is lack of data not lack of efficacy. 6/
This looks at efficacy from the point of dose 1, not 7 days after dose 2, so this as expected doesn't look as good. You can get infected a day after the first shot - not enough time to build immunity. Critical to continue to be careful until AFTER dose 2. 7/
(Actually you should stay careful no matter what since this is not a 100% effective vaccine - no vaccine is).8/
Adverse events.
Lots to parse here.
126 serious events in vax group, 101 in placebo. Not bad.
But LOTS of injection site reactions - this shoudl be expect4ed. Systemic AEs (fever, etc) also seem common.
Overall good news. 9/
More granular AE data - these are unsolicited - ie stuff the participants brought up on their own.
Take home is to expect injection site reactions, fatigue, fever, chills, muscle pain - but most patients not bad enough to bother telling study coordinators about it. 10/
6 deaths (4 placebo, 2 vax).
Vax group was one heart attack long after vax and one death from "arteriosclerosis" 3 days after dose 1. Not sure what that latter one means? Would like a bit more detail here. 11/
In terms of serious adverse events - rates are nice and low.
Eyebrows raised at a higher number of appendicitis cases in vax vs. placebo group, but this could just be a statistical fluke. I guess something to watch for during roll-out. 12/
All in all, this is REALLY encouraging data. I'll be very surprised if the EUA isn't granted on 12/10. And given this - I would 100% get this vaccine myself.
13/
Thread:
Even a mediocre vaccine can end the pandemic. But there are some caveats. I wrote about this on vox.com last week, vox.com/21528373/vacci…
but here are the highlights: (1/n)
Let's assume that, on average, every person with COVID-19 can infect 2 additional people (a bit lower than the R0 of 2.5 but makes math easier). (2/n)
To stop the pandemic, we need to prevent disease in 1 out of every 2 people.
So if the vaccine is 100% effective, we'd need to vaccinate 50% of the population.
(Technically vaccinate or infect 50% of the population but trying to stay simple.)
(3/n)
I have no idea which #vaccine @realDonaldTrump was talking about today. But if we are going to have a vaccine before 2021, it will be one of these seven.
There will be no "antibody passports" for a while. Even if an antibody test has a low (say 5%) false positive rate, if YOU get a positive test, it may only be 50/50 (or less) that you actually have antibodies. WTF? (1/n)
It comes down to the false positive rate versus the positive predictive value. The FPR is how often a test comes back positive in a group of people WITHOUT antibodies. For this example, let's say that's 5%. (2/n)
OK - but as an individual, that number doesn't mean a lot. After all, you don't know if you truly have antibodies or not. That's why you're getting the test. (3/n)
We're testing the wrong people for #Covid_19, let me explain (a thread). (1/12)
OK - tests are limited. That's a given. If they were unlimited, we'd test everyone. That's not an option. We need to triage. (2/12)
But what health systems are doing is selecting those who get tested. And they are picking a very specific group to test:
They focus on those with "classic" symptoms - like fever.
(3/12)
1/ A student just asked me why, if the p-value for a study is 0.04, we can't say the study has a 4% chance of being a false positive. First off, we definitely can't, even under idealized conditions - here's a brief thread as to why.
2/ (Apologies to the great John Ioannidis who does this better than I ever could), and to @VinayPrasadMD whose "Tweetorials" are an amazing epiphenomenon in and of themselves.
3/ Imagine a world of scientific hypotheses - all those hypotheses out there, floating in the ether. "Atorvastatin reduces nose bleeds" is out there. So is "marijuana use increases the chance of graduating college". Some of these are true hypotheses, some are not.