Good morning! It has been 388 days since the first documented human case of COVID-19.
Today we will have an in-depth on the data supporting the Pfizer vaccine, which were released yesterday by the US FDA. There’s a lot in there.
As for COVID-19, we will cover some reader comments and headlines in today’s issue.
As usual, bolded terms are linked to the running newsletter glossary.
Keep the newsletter growing by sharing it! I love talking about science and explaining important concepts in human health, but I rely on all of you to grow the audience for this:
Now, let’s talk COVID.
Pfizer vaccine briefing book
Yesterday, the FDA made the briefing book for the Pfizer vaccine, with detailed safety and efficacy data, available to the public. This is dealt with in detail in today’s In-depth piece.
Potential attack rate of 70% for SARS-CoV-2
Early in the COVID-19 pandemic, epidemiologist Dr. Marc Lipsitch, predicted that COVID-19 would become a pandemic and that 40 to 70% of people could potentially become infected. This was in mid-February, and at that time, I still thought there was a chance to contain this thing. Turns out Dr. Lipsitch was very right, and I was wrong.
Here’s a collected thread of his tweets on the topic: https://threadreaderapp.com/thread/1228373884027592704.html?refreshed=1582681665
I want to highlight one thing that he said, though:


His statement on a 40 to 70% “attack rate” really shocked me at the time. I didn’t think it was possible. The “attack rate” is the percentage of the population that ultimately becomes infected.
Well, it turns out he was probably right about that estimate too. A paper published yesterday in Science compared two Brazilian cities, Manaus and Sao Paolo. In Sao Paolo, eventually, substantial protective measures were put in place. Not so in Manaus.
The attack rate of COVID-19 in Sao Paolo appears to have been about 29% by their analysis. In Manaus, it was 76%. You can read the study here: https://science.sciencemag.org/content/early/2020/12/07/science.abe9728
This study gives us a sense of what might have happened if we did nothing to control COVID-19. 76% is a big number. If that were the US population, that would be 249 million people. If 0.5% to 1% of those people died, as we have observed roughly during the pandemic, that’s 1 to 2 million dead in the US.
That’s what can happen without mitigation, because the natural attack rate of COVID-19 appears to be very, very high if you do nothing to stop it.
What am I doing to cope with the pandemic? This:
Dr. Atomic
We watched the stream of this last night, which I discussed in yesterday’s newsletter. It’s still fantastic, and I wanted to mention that my mother found a performance of it on YouTube in case you missed it.
I’ll admit it was a little more spectacular to see the simulated nuclear explosion live in the theatre, where it filled the entire space with blinding light, rather than on a screen, where it…did not do that.
Pfizer vaccine data
You’ve all been patient as I’ve railed about the absence of specific data about the vaccines from Pfizer, Moderna, and AstraZeneca. For the Pfizer vaccine, that absence is now filled by the FDA’s released of a detailed briefing book for their upcoming meeting regarding this vaccine. You can read it in its entirety here (PDF at link): https://www.fda.gov/media/144245/download
There’s a lot of great information in this report. Where I want to start is with this figure, which is the key figure in the whole report, assessing whether the vaccine prevents disease or not:

This is what science looks like at its best, right here. The intervention in this trial was designed thoroughly and thoughtfully based on 200 years of vaccination experience and understanding. The result is that after the first dose of vaccine is given, the red placebo line and the blue vaccinated line diverge.
The placebo line continues to increase in cumulative COVID-19 cases, while the blue line goes essentially flat. There are some scattered cases, but they do not seem to accelerate over time nor do they show nearly the same slope of increase as the placebo group. the experiment was designed to ask whether the vaccine worked, and it answered that question elegantly. The vaccine works to prevent disease. I’m sold on that.
The next big question is whether or not the vaccine prevents severe disease. This question can be looked at in a few ways; we can look over the entire study period, or we can look after the entire vaccine course is given. The trial’s intended design was to look at severe cases acquired after the second dose of vaccine was given. This was also true for the regular disease case endpoint as well; presumably when the study was designed, the investigators thought that the first dose would not be very protective and there would be a higher rate of cases between the first and second doses than were observed. Boy, were they wrong—and I’m sure delightedly so.
Anyway, because of this design, Pfizer first reported the number of severe disease cases in both arms starting 7 days after the second dose was given, in patients with no history of COVID-19. There were 4 cases total by this measure; 3 in the placebo arm and 1 in the vaccine arm, for approximately a 66% efficacy in prevention of severe disease. I think this is a little misleading, though. First, this is quite a small number. The “95% confidence interval (CI)” (the range of values in which the analysis is 95% sure contains the “true” value) is from -124.8% to 96.3%. So ranging from the possibility that the vaccine is actually worse than the placebo substantially to the possibility that it’s extremely effective. This is not a sign of true uncertainty, but a reflection of very small numbers. It stands to reason, to my mind, that given the large number of cases prevented overall, the vaccine will also substantially reduce the number of severe cases.
This isn’t guaranteed, but I think the remainder of the severity data strongly suggest it. Looking at severe cases in all patients after dose 1 was given, we see there were 9 cases in the placebo group and 1 in the vaccinated group. This gives an efficacy against severe disease of 88.7% (95% CI: 20.1—99.7). That result suggests a clear effect that may be very strong or may be somewhat weak, but it’s there either way.
In light of this, we can be relatively confident that the vaccine will reduce cases of disease but also reduce the likelihood that those cases will be severe. That will take some burden off of the hospitals, which is great.
Having determined that the vaccine is effective in preventing both general COVID-19 as well as severe disease, the next question I want to ask is how the vaccine performs in various special demographics. We know that demographics and health conditions matter in COVID-19, so we certainly want the vaccine work well in those with the highest-risk factors.
Patients with “any comorbidity” (that is, any disease other than COVID-19), had a 95.3% vaccine efficacy in this trial, compared with 94.7% vaccine efficacy for patients without any comorbidities. This is certainly not an actual difference; the true efficacy is probably the same for both groups.
Interestingly, the picture is different for one specific type of comorbidity: patients with any malignant cancer, in whom the vaccine was only 75% effective. This is not surprising; chemotherapy substantially interferes with immune responses. However, it is something that patients with cancer will probably need to keep in mind when seeking a vaccine. It may be preferable to be vaccinated during a “drug holiday,” a break from treatment that is a common part of many chemotherapy regimens. I’m sure that the National Comprehensive Cancer Network, NCCN, will have some guidelines on this eventually—they provide recommendations regarding other vaccines as well.
For all other specific comorbidities examined, vaccine efficacy was at least 93%.
The vaccine was highly effective across age categories as well, and was 100% effective in patients over 75, but there were only about 1600 patients over 75 in the study, so that may be an overestimate.
It was also highly effective in a variety of racial groups, though there were some groups that seemed to show lower efficacy. However, I’m not sure just how seriously to take that, because there was wide variation in racial representation, and it may be that patients of different races have differential exposure to the virus in the places where the study was conducted. I wouldn’t make too much of those variations, especially considering that they don’t seem to follow any particular pattern and are all the result of analyses based on less than 5 total cases.
Additional categories where the vaccine was highly effective were gender and obesity status, and Hispanic or Latino ethnic status. Most efficacy numbers were consistently in the mid-90s. It looks like this vaccine works well in a diverse group of people.
One group had very poor vaccine efficacy though—people who tested positive for SARS-CoV-2 infection at the start of the study. There were about a thousand such patients, and there was 1 case of COVID-19 in each of the placebo arm and in the vaccinated arm among this group. This is not very surprising, since this group contained people who were either actively infected or recovering from infection at the time they enrolled in the study, and a vaccine generally cannot protect you if you are already sick.
So far we have a picture of a vaccine intervention that is highly effective across diverse patients, that appears to be capable of preventing both generalized disease cases and severe disease cases, and that seems to work in patients who have existing disease conditions that may increase risk of severe COVID-19. All good news on the efficacy front.
Efficacy isn’t the whole picture, though. We also need to understand safety.
Safety data are really difficult to talk about in detail because they are generally just big tables of numbers. What’s more, these numbers are collected in exhaustive detail. If you’re in a clinical trial and you get shot with a gun, your adverse event of gunshot wound will be recorded as a safety event in the trial. An independent commission that monitors safety in the trial will discuss your gunshot wound during a meeting, and they will discuss it seriously to assess whether it might have been caused by the treatment.
You may think this example is silly, and in most cases it is silly. But imagine that the product being tested was alcohol. It’s very possible that alcohol use could increase your risk of being shot, because it has behavioral effects. A lot of medications can have behavioral effects. When safety is monitored in a clinical trial—and safety is always monitored in a clinical trial—we look at everything in detail.
That said, I’m going to summarize. The safety events that are observed in the Pfizer vaccine data are what you would expect to see from an injected vaccine. There are a large number of injection site reactions. These injection site reactions—redness around the injection site, for example—were much more common with vaccine than placebo. This suggests that they are genuinely reactions to the vaccine rather than reactions to being jabbed with a needle. In a way that is good, as injection site reaction is generally a result of immune response.
The results categorize the severity of these reactions as “mild,” “moderate,” or “severe.” In vaccinated patients, most reactions were mild or moderate. “Moderate” is probably a sign of substantial discomfort, though; “severe” usually means the patient required medical attention, while moderate is a step down from that. I’ve heard anecdotally that this vaccine causes a bit more discomfort than the typical seasonal flu vaccine, and that may well be the case based on these data.
However, reactions in the “severe” condition were rare, as were “serious adverse events,” or SAEs. I always look straight for the SAEs, because these are the things that really can be scary. Keep in mind that people in clinical trials lead full lives that may cause SAEs that will be picked up by the trial, so there are expected to be some in any trial. The big question is whether they occur more frequently, or with a defined pattern, in the patients receiving the treatment of interest. They are usually the rarest events, but they are the ones that you really don’t want to see. Fewer than 1% of patients in this trial had an SAE. The rate of SAEs did not appear to be substantially different in placebo vs vaccinated conditions. That is all good news.
Finally, death is a safety event. There were more deaths in the placebo group than in the vaccinated group. While I’d prefer there not be any deaths in a trial at all, if there have to be, that’s the pattern one would prefer.
Two specific events that looked elevated from my trip through the data were headache and fatigue; be prepared for those if you do eventually receive this vaccine. However, none of the “safety signals” measured in this trial suggest risk of anything serious.
We need to be aware, though, that this trial had a short follow-up period. We are discussing safety results here that cover only a median follow-up time of two months. Some patients were followed for longer, but two months was the median time. That is not very long. It is possible that there will be more safety events observed as time goes on; the trial continues to monitor patient safety.
Thankfully, the staged rollout of the vaccine and the limitations on its manufacturing may be a help to us here. If there is some sort of silent safety event that has not yet been detected, it may be found before the vaccine is widely available, during the protracted months of its roll-out. I do not expect there to be such a surprise safety event, but since it has happened before with medicines, I’m never going to say never.
Another thing that may change with added follow-up is the efficacy. While the efficacy monitoring period is formally over, since the trial is continuing I am sure that Pfizer will continue to monitor the patients in the trial for COVID-19 infection. We may learn that immunity induced by the vaccine wanes slightly over time. Perhaps a third dose of the vaccine will be warranted after a year or two; it’s impossible to tell at this point.
It is also impossible to tell if the vaccine has any effect on transmission of the virus. Specifically, the report has this to say:
Data are limited to assess the effect of the vaccine against transmission of SARS-CoV-2 from individuals who are infected despite vaccination. Demonstrated high efficacy against symptomatic COVID-19 may translate to overall prevention of transmission in populations with high enough vaccine uptake, though it is possible that if efficacy against asymptomatic infection were lower than efficacy against symptomatic infection, asymptomatic cases in combination with reduced mask-wearing and social distancing could result in significant continued transmission. Additional evaluations including data from clinical trials and from vaccine use post-authorization will be needed to assess the effect of the vaccine in preventing virus shedding and transmission, in particular in individuals with asymptomatic infection.
Translated from the complex nature of medical writing, this says, “we’re not sure yet, we’ll get back to you.” I hope they do soon.
All in all, I’m very confident about the power of this vaccine to substantially reduce the threat posed by COVID-19 to most people, for a meaningful period of time. Since the duration of immunity to the virus appears to be at least a year, if not longer, I expect the vaccine will be similar given the strong immune response that it generated in Phase 2 trials. Having seen the data, I am confident that we have won a major battle in the war against COVID-19. Now we just have to keep winning.
Lots of comments in response to yesterday’s newsletter, once again!
Reader Lisa Hertel left the following:
You're right about hospital workers needed to get tested. When I began working in a hospital, I was tested for both TB and rubella. I knew my rubella titer was good, because it's also routinely tested in women when you get married. (I am older than the MMR vaccine, and was of the era when it's predecessor had questionable effectiveness. I actually had mumps and chicken pox.)
In the future, I suspect if the COVID-19 vaccine works, it will be one of those things that colleges require, much as they do the meningitis vaccine today. The question is, how long will immunity last? One problem with hurrying the data is that we only know for as long as we've been giving the vaccine, which is a few months.
Excellent question. Here’s what I said:
It's a great question. We do know that immunity induced by the virus appears to have a relatively long duration, but that's from a handful of small studies. We also know from earlier results that the vaccine from Moderna, at least, induces more consistently high immune reactions than are seen in patients who had natural infection. I'm hopeful that we'll get at least a year of immunity out of this crop of vaccines, if not longer. But ask me again in a year.
Reader EH asked:
There’s a raging pandemic and it’s not well-controlled in places across the country. How do you administer vaccines to people who may already be infected? Could inoculating someone who is currently-infected make the vaccine more dangerous than for someone with no detectable trace of the virus? Would there be a need to get tested for the virus first, wait forever the result (which could be minutes, hours or days) and only then get the shot? And what about the booster - is there some likelihood of becoming infected between first and second shots?
Thanks, as always, for an excellent newsletter, Dr. Skylar!
Here’s my response:
Thank you for reading!
Generally speaking, you don't give vaccines to people who are sick with the thing the vaccine is against. It is often recommended not to give vaccines to sick people at all, because the existing illness may leave the immune response somewhat weakened in response to the vaccine. This is not true for all vaccines, however, and I suspect that advice will be provided in the prescribing information for COVID-19 vaccines about how to evaluate sick patients for potential vaccination.
I don't think that vaccinating a currently-infected person could add risk of negative outcomes for them, to answer your second question. The dose of vaccine is smaller than the amount of virus that is running around being produced, and it also is not competent to cause cell damage or death. I might be hesitant to give the vaccine to a patient with severe COVID-19, because they already have a pretty damaging immune response, but I don't expect that this will ever be a consideration. When someone has severe disease, you know.
Anyway, I doubt that a negative test for COVID-19 will be a prerequisite for vaccination. As will be described in tomorrow's newsletter, the vaccine was given to about 500 COVID-19 positive patients in the Pfizer trial. I don't believe they had any meaningful problems from receiving the vaccine, though unsurprisingly it was less effective at preventing COVID-19 in the group of patients who were actively sick with COVID-19. However, it may well have prevented future disease, a year from now. We're not quite there with the follow-up yet.
There IS a chance of becoming infected between the shots. However, at least for the Pfizer vaccine, it looks like the first dose provides protection starting about 7 days after it is administered. Don't go burning your mask and running around in the streets kissing people once you're vaccinated. Give it some time. Maybe a lot of time, for that specific activity.
For the record, I’m being glib with that thing about burning your mask and kissing people. We don’t know if the vaccine will prevent transmission, so even if you’re protected, you may still pose a risk to others.
You might have some questions or comments of your own! Send them in. As several folks have figured out, you can also email me if you have a comment that you don’t want to share with the whole group.
Join the conversation, and what you say will impact what I talk about in the next issue.
Also, let me know any other thoughts you might have about the newsletter. I’d like to make sure you’re getting what you want out of this.
This newsletter will contain mistakes. When you find them, tell me about them so that I can fix them. I would rather this newsletter be correct than protect my ego.
Though I can’t correct the emailed version after it has been sent, I do update the online post of the newsletter every time a mistake is brought to my attention.
No corrections since last issue.
See you all next time.
Always,
JS