What 95% vaccine efficacy means

Continuing the discussion from I've had my first shot; can I act like I'm fully vaxxed?:

Bolding mine.

This is such an important misunderstanding, one which an enormous amount of people make, that I wanted to address it in its own thread. (Not meaning to call you out – no one explains this, and lots of people make the mistake, including me, until I looked it up a while back.)

The efficacy percentages you see are about the reduction in risk. A vaccinated person doesn’t have a 5% or 10% chance of getting Covid (over what time frame?). No, they have a 90 or 95% reduction in risk of getting Covid.

5% of the vaccinated people in the trials didn’t get Covid. 0.04% did.

The mRNA vaccines are some of the most effective vaccines produced, period. MMR has 97% efficacy against measles, and 88% against mumps. Flu vaccine varies between 40% and 60% efficacy.

Here’s a source for the numbers.

It also means that you are far less likely to die from Covid-19 than a non-vaccinated person and if you do get the disease it will far more likely to be a mild case.

Also getting the vaccine makes you less likely to become a petri dish for the virus to mutate into another strain that may be resistant to the present vaccine. I wish that message would get put out to people who are “waiting” to see how it all goes. Now is not the time to wait.

I’m the guy you quoted. I take no offense at you picking on my words to make your point. And I do agree 100% with your math and your epidemiology.

But I will say I think we’re saying the exact same thing, albeit with different (mostly) unstated assumptions. And IMO the key to successful messaging is to understand which set of unstated assumptions our audience is unwittingly using as they interpret your words or mine. Metaphorically, if I’m using inches and you’re using yards but neither of us is identifying our units, and the audience isn’t actually thinking about units as they read our competing words, there’s a lot of room for the audience to take away the wrong idea.

So …

If I am unvaccinated and wander around in public unmasked among a crowd that has some unknown number of spreaders in it for a day, I will be exposed to X number of opportunities for infection. And between luck and the skill of my natural immune system, my genetic susceptibility, etc., I will actually contract COVID not from every one of those X encounters, but from some small fraction of them, maybe 1%. Let’s just call it 1% for simplicity of explanation; I’m not claiming that number is scientifically accurate; in fact the real number is much, much smaller. All this is before we consider anyone being vaccinated. If I’m masked instead, the percentage chance is less, but the same logic applies. I’m exposing myself to some specific, if not strictly known, risk.

So my basic risk for my day’s activities out in public is arguendo 1% of X. As you say, the vaccine takes that risk and cuts it by 90 or 95%. Said another way, the risk become 5% or 10% of what it was. Let’s stick with 90/10 for now because that’s what I’d used in my first post. You’re right that the real number from the trials was better.

My point was/is the vaccine stops 9 of 10 of the assaults that would have been successful in infecting you had you not been vaccinated. IOW 10% (due to your vax) of 1% of X (due to luck, etc.) would get through. Your point is that the total risk is the same exact number, 0.04% or whatever. You’re just looking at the end number, not the components of it.

And now here’s where/why that distinction matters.

For Joe/Jane Average, the actual risk of contracting COVID 6 months ago when nobody was vaccinated wasn’t really all that large. A tiny fraction of 1% per day, and ballpark 10%-15% per year of typical distanced/masked public behavior. And yet that risk was enough to make total hermits of WAG 50 million Americans, and radically alter the public behavior of another WAG 200 million, myself certainly included.

The OP of that thread was asking how much to alter his behavior. And my point was/is from his POV as if he was the the only vaccinated person on Earth, his risk is now 10% of what it was. If he was willing to upend his life before, perhaps he still needs to keep it partly upended now, not throw caution to the winds. Conversely, if he was incautious before, he has (some) more justification for (some) further incaution now.

As any given individual considers how to behave post-vax, they need to be clear that they have reduced the risk by 90% (or 95%, or 97%, or whatever the data shows). As you say, their new risk isn’t 10%. It’s 10% of their old risk.

Which in turn means that if they drop 90% of their other precautions, they’re restoring the exact same risk they just spent 14 months hiding at home from. Just like drivers with anti-lock brakes tend to speed more and brake later, folks need to think carefully about their total risk and how each component changes that. If you add a vaccine and take off a mask or start hanging out in crowded noisy bars, you may well be having the exact same risk, but more fun. Which is a fine decision if made consciously. But a real mistake if made in the wrong-headed belief that your vax drove the total risk to nil.

Though the math above is correct (given the semi-made-up numbers) and the logic is sound, there are two significant faults with my argument above.

One fault is that this exposition assumes everyone else is unvaccinated. As more people are vaccinated the (1% of X) term declines and your personal risk declines in lockstep because out in that bunch of people you encounter that day you’ll encounter fewer infectious people. But not zero. And we each have no way of knowing how vaccinated or not any particular public group is. If we did not have such a large crop of refusers in our midst, we could say that soon enough (a couple months?) with enough people vaccinated, COVID would be largely over. That’s the famous herd immunity. You don’t not catch it because you’re immune; you do not catch it because it’s just not out there to be caught. But those refusers (plus the much smaller group of people legitimately unable to vax) will ensure the herd risk doesn’t decline all the way to truly negligible.

The other major fault (omission actually) is what @Si_Amigo says just above. The vaccine has some effectiveness, call it Y%, on preventing you from contracting COVID given exposure to it sufficient to cause infection were you unvaccinated. And we agree Y is somewhere in the 90/95/97% bracket which is pretty darn spectacular as vaccines go. Yaay Science!

Even better yet, it has 10x Y% effectiveness on preventing serious COVID and 30x Y% effectiveness on preventing death from COVID.

So in addition to reducing your odds of infection it greatly reduces the severity of an infection that does leak through to you. If COVID truly were no worse than the common cold from the git-go we’d maybe have never even noticed it. It’s the severity of unprotected COVID infection that made this such a big deal planet-wide. It’s not true that post-vaxxed COVID is totally harmless; I sure as heck hope I never catch it. But as between catching COVID two unvaccinated months ago and catching it now that I’m vaxxed, I’d much rather choose door #2.

Bottom line on this mega-essay:

So when somebody is considering how to alter their behavior because now they’ve just now been vaxxed, they ideally need to consider that their risk in public has been declining for the last couple of months due to increasing vaccination among the herd out there. And will continue to decrease as long as more and more people are getting vaxxed each day. But will eventually plateau at a number north of zero due to non-compliance, importation from other raggedier regions or countries, new variant mutations, and all the rest. And they need to consider that, due to their own vaccination, their own risk of death has receded A LOT. But not to zero. And their risk of severe COVID has also receded a lot. But not to zero.

I made a comment a month or so ago in the breaking news thread, that the public at large was dropping precautions at or slightly faster than they were getting vaccinated. Which is why (IMO) we were seeing the infection numbers steady or going up while the death numbers were going down. As a group we’re holding the risk of infection roughly constant, just changing which mitigations we’re using to get there. Any one of us can choose how much to join that constant-risk bandwagon or not. But better if we do that knowingly, not unwittingly.

We don’t know how many people are vaccinated, but the number of positive cases informed by an evaluation of the positivity rate can give you a pretty good handle on how much COVID is out there to be caught. Dallas County, where I live, has about 10% as many cases as we had in January, and the positivity rate isn’t spiking.

If I understand the OP correctly (using my layman-y brain)…

95% vaccine efficacy would be kind of like saying, “Getting shot by a bullet that has only 5 percent as much kinetic energy as a bullet normally would,” right?

In other words, it would be a case of “almost everyone who gets shot by such a weak bullet would survive?”

I don’t think that analogy works but will leave it to others to explain why. We do know, as has been reported by various reputable sources including the one in the OP, that the Pfizer, Moderna, Johnson & Johsnon and Astra-Zeneca vaccines are 100% effective in preventing COVID hospitalizations and deaths.

I think the way to think of it in concrete terms is, if there’s currently a base risk of 100 per 100,000 of catching Covid in your community, once you’re vaccinated, you’d have only a 5 in 100,000 risk. You can also affect that risk in other ways, like if you always adhere to masking and distancing, and are only unmasked with your household, etc., your base risk was already lower than 100 in 100,000, and you’re lowering that risk by 90 or 95%. So you could have a miniscule risk.

Or, if you get vaccinated and then start doing all the things you’ve been waiting to do, and you stop taking precautions, and hang out unmasked with unvaccinated people, you might be raising your base risk, but then it still gets reduced because of the vaccine. You’ll still come out ahead, I think, because 90-95% is a huge reduction in risk, and also because if you do get Covid, its severity will also be reduced. But this is why vaccinated people should not just jump back to life as normal, at least until many more people are vaccinated and community spread is way down.

5% DID NOT get Covid
.04% DID get COvid

What about the other ~95%?

You misunderstood what he meant.

What he was saying was that the number of people who got COVID in the trial was not 5%, but rather was 0.04%.

He was knocking down a sorta-strawman he’d developed in response to something I had written. Which he had interpreted as me asserting that 5% got COVID. He was debunking his (mis-)understanding of my argument.


Primary efficacy analysis demonstrates BNT162b2 to be 95% effective against COVID-19 beginning 28 days after the first dose;170 confirmed cases of COVID-19 were evaluated, with 162 observed in the placebo group versus 8 in the vaccine group

And, these numbers are how “95% efficacy” is determined. The vaccine group only had 5% of the number of COVID cases that the control group did, and, thus, the vaccine reduced incidence of the disease by 95%.

I understand. But with only 8 cases, we can say there were no deaths and hospitalizations, but thats out of 8 total cases. After 200 million people have been vaccinated with moderna and pfizer, who knows what % of those who do get infected will need to be hospitalized or die.

Yes. That’s right.

Out of the 6,000 or so breakthrough infections, 396 people were hospitalized and 74 people died, according to CDC data released last week.

6,000 is an underestimate – it’s only 0.007% of fully vaccinated. The CDC knows that they aren’t getting all of the cases reported.

I wonder what the efficacy means on a personal level. If a vaccine reduces risk by 95% for a population, it doesn’t necessarily mean my personal risk is reduced 95%.

I’d assume some people don’t produce a good immune response to the vaccine, so their personal risk might only be reduced by some small amount. Which would imply that people with a “typical” immune response would have a reduced risk greater than 95%.

Your second sentence doesn’t follwo.

If you assume the trail members were no more or less immune capable than the population at large, then if the trial group got 95%, so should the population at large.

What we don’t know is the shape of the “how immune is each individual” curve.

I am talking about the “how immune is each individual” curve. If the population as a whole is 95% and some individuals are below that, then other individuals must be greater than 95%.

Thanks for the enlightening thread! So a minuscule .00047% of those vaccinated got COVID severely enough to be hospitalized? Do I have that right? (And 30% of those vaccinated who did get COVID were asymptomatic.) Even bumping that up because only 86% of states reported, it doesn’t seem like something I should worry about. Please correct me if I’m wrong. Trying to fight my own ignorance here.

If a case occurs after both doses of Moderna or Pfizer (or the single J&J) but before the two-week period following that is up, does it count as a breakthrough case? I assume not, but the article doesn’t say.

I’m not sure , but I suspect it would not be counted as a breakthrough case.

I don’t know that we can really calculate the rates yet, because even in states that are reporting, there has been some trouble getting complete data. And infections will necessarily lag the vaccination numbers. So it will wind up being higher than it looks now. But I think it is looking very good.

And, yes, I think one of the most important things is the 30% being asymptomatic, and low rates of hospitalization and death.

A follow-up question regarding an individual’s status - is there some clinical sign that would inform us after we’re vaccinated as to whether we would fall into the higher risk or lower risk category? Is the presence of Covid-19 antibodies what we’re looking for? Or do we have to get sick and die for us to know that we were among those who were not sufficiently protected?

95% means jackshit. If and when I become omnipotent, I shall go back in time and ban these reports from being made public.It’s confused the issues.
It means that 95% of the Covid cases in the trial were in the placebo group not the vaccinated group. That’s all.
Doesn’t mean the data was useless, far from it but it has limitation both Pfizer and Mordena were tested in the US only and during a lull in the pandemic in the country, ie late summer and autumn of 2020, if they had been tested elsewhere at different time, might have seen different outcomes. J&J famously has a lower efficacy, but it was tested during the start of the winter surge and in multiple places.
SinoVac has seen efficacy between 50% in Brazil and 85% in Turkey.

Which brings up another problem, variants can make the efficacy numbers go haywire, SinoVac was facing the very nasty P1 variant in Brazil while it faced the original wild virus in Turkey. Both Pfizer and and Morderna’s number are against the original wild virus. And we know variants can cause a substantial drop in efficacy, AZ had to be discontinued in S Africa since it was functionally useless against the South African variant.
(If you want to be scared, whole virus vaccines like SinoVac have less significant drops in efficacy versus variants as opposed to subunit vaccines like all the ones approved in the US, so the fact SinoVac saw such a reduction should concern you).

That said, all the vaccine greatly reduce your chance of getting it but to use @LSLGuy favourite expression, the risk remains non-zero and the more time goes by the more chances of you getting it eventually.