Covid-19 Vaccine Efficacy Explained | It's Not What You Think

What does the Covid-19 Vaccine Efficacy actually mean? It's not what you think. Unless you’ve been living under a rock these past two weeks, you probably heard the great news that two of the leading vaccine contenders for Covid-19 have had some amazing initial results. The joint effort by BioNTech and Pfizer announced on November 9th that their vaccine candidate has a greater than 90% efficacy and just this week on Nov 16th Moderna announced that their vaccine candidate has a 94.5% efficacy. I’m not an expert on vaccines, but based on everything I’ve been reading, this seems like an absolutely major success that we should all be celebrating.

And it’s because I’m not an expert on vaccines that I’m not going to discuss the differences between these two vaccines or how they’ll be distributed or even how likely they are to have side effects. There are a bunch of amazing YouTubers putting out really great content on just this. Instead, since my expertise is in data, I’m going to help you understand what those efficacy numbers actually mean. And I’ll fully admit that when I first read the news stories about these vaccine candidates, I did not interpret those values correctly at all...and I’m guessing neither did you.

Welcome to Data Demystified, I’m Jeff Galak and if you stick around, I’m going to debunk what I think is the most common incorrect interpretation of those efficacy numbers and then explain to you what they actually mean.

Follow me at:

Equipment Used for Filming:

[DataDemystified]

UPDATE - There is an error in this video and I thank all the viewers who pointed it out. To be clear, the intuition of efficacy is correct in that it reflects the relative impact of the vaccine compared to the control group, rather than your personal likelihood of contracting Covid after you get the vaccine. That said, in my haste to make this video in a timely manner, I muddled the math. Efficacy should be calculated as 1 - (% Infected in Vaccine condition / % Infected in Control Condition). Sorry for the error. I hope that the intuition was the key takeaway rather than the calculation.

[atributetoantifafallenhero]

I got jabbed 4 times and got covid 4 times. Every time its getting worse and worse. God help me. What is going on?

[DrRickKellyCrossroadHealthDPC]

@DataDemystified nice presentation. There is one factor that is not being mentioned or even alluded to in most discussions of Covid vaccine efficacy. That is duration. I think the lay public believes efficacy is an either/or, or yes/no proposition. But in actuality, the reported efficacy might be at 28 days after the second vaccine dose, or 60 days, or something else.

Consider a major league pitcher who throws 90 mph fastballs into the strike zone. The ball leaves his hand, and after traveling 60 1/2 feet, it is traveling 90 mph. What if the catcher, instead, was 70 feet or 80 or 120 feet away? Will the ball still be traveling that fast? What if the strike zone was 12 inches wide instead of 17 inches?

Despite these amazing early efficacy rates, we now understand that there is a significant drop off in efficacy, both due to time since previous vaccination, as well as immune evasion by new variants. So, as often is the case, perception is not the same as reality.

[Waltuh]

What’s the efficacy rate of the vaccine?

[sndsofscnce]

This was great and clearly explained. I'd been wondering what these numbers meant since they were reported.

[pollysey]

Only a small handful are admitting this is what the numbers mean.

[sankalp]

Great Video! Definitely learned something very useful. However, I have two questions:

1 - Do the initial volunteers for both the vaccine and placebo groups, have no history of infection before being administered dozes?
2 - Once the vaccine and placebo are administered to the groups, what's the process like to evaluate their conditions later? Do we wait a certain time before testing both of these groups for covid-19 infection? If yes, then isn't it allowing a lot more uncertainty in reporting the efficacy numbers?

[Brenstar]

My first interpretation of the efficacy numbers were the percentage of people who were "cured" of Corona. But that tells you how misleading the whole thing was that everyone had a different interpretation.

[samdog]

August shots, more and more vaxed getting very ill.

[thebrocialist]

I’m not a data expert, but I didn’t assume what you did when I saw the headline claims about the vaccines’ efficacy.

[dimitrioskalentakis]

Shouldn’t the Pfizer efficacy be calculated as follows?
(85-9)/85= 89.4% (i.e. 1- Relative risk)
This is a different number from 90.4% you mention.
Thanks

[andrewmgomel]

Confused, as I thought an efficacy rate of 90% definitely does mean that your chances of infection decrease at a rate of 90% - as in, if the rate before infection were x, the new rate would be x - 0.9x. Are you just stating that it isn't a 90% nominal decrease starting from 100%?

[geoffphillips]

The unwary victim blindly believes the propaganda of others.

[jaster]

Hi the future here, didn't work well. Cheers.

[mindfuleats]

End the horror of the global pandemic by all lining up to take something that doesn't stop transmission or contagion? Lol

[thethinkingman-]

efficacy is about COMPARING vaccinated to unvaccinated (during clinical trials)
94 got covid.
9 were vaccinated and 85 were not.
the number of vaccinated people (who got covid) was 10.58% of the number of UNvaccinated people who got covid (ie. 9/85)
so the vaccinated are said to be 89.42% better off when COMPARED to those not vacinated.
so efficacy is said to be 89.42% ( bigger % means more sales)

[wills]

In what year do you think safety will be verified? 202? Or 2030+?

[chasecolin]

How do Type I and II errors come into play here?

[bubbarand]

A bookmaker would call it "odd" - like 2 to 1 for the flip of a coin, or 50 percent. Also meaning that the efficacy of the flip of a coin is "fifty". Right? If I'm wrong, then tell me how I'm wrong. The efficacy of a straight answer is zero.

[roshx]

The calculation of efficacy is not right here, it's 1-vaccine/placebo cases not placebo/total. If the arms had equal numbers of cases that means the vaccine is 0% effective not 50%