Matt_W
1668
So that means an infection rate for placebo receivers of 90/15000 or 0.6%, which is about 6x lower than the national (EDIT: all time) infection rate. (EDIT: I guess you’d have to look at the study duration to see what the actual infection rate is expected to be. Not sure what the duration was/is.)
But given that 0.6% is the expected infection rate, the chance of there being 5 or less positive cases in a population of 15,000 is very close to zero. (Far outside any standard significance interval.) Strongly suggests the vaccine is very effective, unless there’s some other correlating factor.
Well, phase 3 didn’t go on too long; median follow up of a little over 2 months. Add to that some selection bias (those in the study would likely have been more aware of COVID factors including their behaviors to have chosen to participate in the first place).
Spot on.
Ephraim
1670
I also don’t know where they are running the clinical trials. The picture in the WaPo article showed a volunteer getting the vaccine in Rockville, Maryland (where my sister lives, coincidentally). It’s a DC suburb that borders Bethesda, where the NIH is located, so it wouldn’t surprise me if the Moderna study was done there. And Maryland has been in pretty good shape up until lately, at least according to my sister. I think it’s their spike in cases that’s leading Moderna to be so optimistic that the next week to 10 days should let them get to their desired number of infected participants (hopefully all still in the placebo group!).
My understanding was the vaccine signups in the US were mostly recruiting in hotspot areas (and many geographic areas which is part of the expense of the trial), so the trials could get the exposures needed. When the trials were being set up this summer, my state had a 0.2% positive rate, so I didn’t bother to volunteer, I’d never be chosen.
Joke’s on me. Positivity rate has gone up 10-15x where I live [~3%]. Of course, we’re still not a hotspot, relative to many other places in the country where there isn’t enough testing, and positivity rates are upwards of 40-60%.
Matt_W
1672
I kind of wish they’d report results this way. I totally get that you want to answer people’s question: “If I get this vaccine, how likely is it to prevent infection?” And that’s what the 95% number provides. But to a layperson eyeballing numbers, both 5 and 90 look low compared to 30,000 study participants. You might suspect that random chance could produce those numbers even if the vaccine was totally ineffective. But no, that’s not the case at all. Even if you look at the average infection rate across the whole group (0.317%), the chance of there being a random <=5 cases in a group of 15,000 is less than a trillionth of a percent.
I think you can use binomial statistics to show that 95 and 5 isn’t really going to happen [very often] by chance. [P=3e-21 is infrequent enough for me.]
Simply sit the person down with a coin and say “flip this coin 95 times. If you get 90 heads, you can stop. If not, then start another round of 95 flips. You’re welcome to continue as long as you want.”
Matt_W
1674
Yeah, that might work, though I suspect most people’s intuition will fail even then. You might expect if you do it 300 times you’ll get at least one success. (You definitely won’t.)
Strollen
1675
How do we know what the national infection rate is? Not arguing, I’ve been really struggling to figure it out.
If we use active case 4.2 million/ 330= 1.3% total number of cases 11.5%/330 = 3.5%. But we know there are a lot of unidentified cases. I assume that participants are tested way more than the average American. But I assume there is also a time function and other factors.
The other thing that seems pretty obvious to me is that participants in the trials are different than the average American, and especially different than Tumpist and Anti-vax folks.
If you think the Covid is hoax, exaggerated etc. you don’t sign up for trial of a vaccine with potential for numerous complications. Likewise, I bet the trial participants are lot more likely to use masks, practice social distance than the average American.
Now none of this detracts from the dramatic difference between the vaccine group and the control group. But I do think its worth studying all the participants just to show how much behavior effects your changes of getting the virus.
Are you positing that cases and controls change their behavior in different ways after they are vaccinated? I suppose in a really poorly designed study, you would “know” that you were not vaccinated [1], and thus you might be more cautious than a vaccinated person. This would lead to an observed decrease in effectiveness - people who received the vaccine would engage in more risky behavior because they thought they were “safe”, and thus get infected more often.
It seems unlikely that the opposite occurs - you know you didn’t receive the vaccine, so you engage in more risky behavior.
[1] Some trials in the past have given contols a completely different vaccine (or similar) that produced an immune response such that their arm might be sore after the procedure. It’s entirely possible for people to “cheat” by getting an antibody test.
Strollen
1677
No I’m not saying that at all. I am saying the mere fact that they were volunteers participants in a trial at all makes it likely their behavior is different than the average Americans. For instance, I suspect, that mask usage among all trial participants was higher than the average American, even before they got the injections.
It looks like even the placebo group has a below average infection rates. But since I’m not really sure how we measure infection rate, I’m not positive about that.
That characteristic should be constant across both groups, however, since they both think they got the trial vaccine.
Alstein
1679
They knew they had a 50% shot of getting the placebo, and you don’t want to get a disease that has a 1% shot of killing you. These folks were probably well aware of the virus, but you can do everything right and still get it.
Ironically, the researchers think the US has been so bad at COVID that it caused results sufficient to deternine vaccine effectiveness a lot quicker than expected.
I don’t think rate of mask usage really impacts anything other than viral load (which probably impacts severity, so that could be an issue there, though I think it’s unlikely- we’ll find out eventually when vaccinations hit red areas).
Yes, I do think they had below-average rates in the placebo group. If behavior was different, you’d see both groups have higher rates, and socioeconomic status probably plays a role (guessing most test subjects were more affluent, and being affluent makes you less likely to get it due to less involuntary opportunities for exposure)
2020 OECD Report for the state of health in the Europe is out. It’s entire first half is dedicated to looking at the numbers and facts of the first wave of the Covid-19 surge.
https://www.oecd-ilibrary.org/social-issues-migration-health/health-at-a-glance-europe-2020_82129230-en
Direct link:
No revolutionary new conclusions in it, but it contains a lot of numbers that pretty much confirm the rightness of the recommendations that most of the sensible epidemiologists have been screaming all year (i.e., close early, close hard, etc).
MikeJ
1682
I’d heard of this, and it is just observational, and not a big set, but a bigger difference than I’d thought:
https://www.nature.com/articles/s41598-020-77093-z
Study included either asymptomatic COVID-19 patients (Group A) or severely ill patients requiring ICU admission (Group B).
The prevalence of vitamin D deficiency was 32.96% and 96.82% respectively in Group A and Group B.
The fatality rate was high in vitamin D deficient (21% vs 3.1%). Vitamin D level is markedly low in severe COVID-19 patients. Inflammatory response is high in vitamin D deficient COVID-19 patients. This all translates into increased mortality in vitamin D deficient COVID-19 patients.
JD
1683
Not sure if it was already discussed elsewhere, but… huh.
https://www.who.int/news-room/feature-stories/detail/who-recommends-against-the-use-of-remdesivir-in-covid-19-patients
WHO has issued a conditional recommendation against the use of remdesivir in hospitalized patients, regardless of disease severity, as there is currently no evidence that remdesivir improves survival and other outcomes in these patients.
This recommendation, released on 20 November, is part of a living guideline on clinical care for COVID-19. It was developed by an international guideline development group, which includes 28 clinical care experts, 4 patient-partners and one ethicist.
The guidelines were developed in collaboration with the non-profit Magic Evidence Ecosystem Foundation (MAGIC), which provided methodologic support. The guidelines are an innovation, matching scientific standards with the speed required to respond to an ongoing pandemic.
Work on this began on 15 October when the WHO Solidarity Trial published its interim results. Data reviewed by the panel included results from this trial, as well as 3 other randomized controlled trials. In all, data from over 7000 patients across the 4 trials were considered.
The evidence suggested no important effect on mortality, need for mechanical ventilation, time to clinical improvement, and other patient-important outcomes.
The guideline development group recognized that more research is needed, especially to provide higher certainty of evidence for specific groups of patients. They supported continued enrollment in trials evaluating remdesivir.
Updated 20 November 2020
- A conditional recommendation is issued when the evidence around the benefits and risks of an intervention are less certain. In this case, there is a conditional recommendation against the use of remdesivir. This means that there isn’t enough evidence to support its use.
strategy
1684
Uh oh. Glad the wife got me chucking Vitamin Ds a few months ago… I’m perenially low on that.
MikeJ
1685
I remember a while back they were testing vitamin D levels in people in Toronto and prevalence of vitamin D deficiency scaled with how dark a persons skin was. From like 25% for white skin to something like 90% of people with dark skin. Vitamin D deficiency is also worse in winter.
Covid seems to hit Black people in the US harder and so far it seems worse in the darker months. There’s no shortage of possible explanations for those but it’s interesting that vitamin D deficiency lines up with both. Maybe a lot of lives could be saved with supplements.
I know coming across this paper has highly motivated me to keep up the supplements!
Well worth a watch from June…
JD
1687
Latest one in:
Oxford University said interim analysis from its phase 3 vaccine trial showed that the 70% effectiveness came from combining two dose regimes. One was 90% effective, the other 62%.
In a statement, Prof Andrew Pollard, the director of the Oxford Vaccine Group and chief investigator of the Oxford vaccine trial, said: “These findings show that we have an effective vaccine that will save many lives. Excitingly, we’ve found that one of our dosing regimens may be around 90% effective and if this dosing regime is used, more people could be vaccinated with planned vaccine supply.
“Today’s announcement is only possible thanks to the many volunteers in our trial, and the hard working and talented team of researchers based around the world.”
His colleague, Prof Sarah Gilbert, professor of Vaccinology at the University of Oxford, said: “The announcement today takes us another step closer to the time when we can use vaccines to bring an end to the devastation caused by Sars-CoV-2. We will continue to work to provide the detailed information to regulators. It has been a privilege to be part of this multi-national effort which will reap benefits for the whole world.”
(Note: Pfizer/Biontec & Moderna are at 95%.)
