I was first wondering why, if we keep hearing about teams rapidly generating vaccines for COVID-19, the common wisdom is that it will take 18 months to start vaccinating at a large scale.

Turns out that the scaling up takes a few months, but the real blocker is the Phase 1 trial, which requires monitoring patient health for 14 months after vaccination.

Doesn't it seem like the cost-benefit analysis changes a bit if we're in the midst of a pandemic? Wouldn't it be worth cutting it down to e.g. 3 months before at least vaccinating the highest-risk populations? Is anyone official even thinking about this?

New to LessWrong?

New Answer
New Comment

6 Answers sorted by


Apr 24, 2020


The article I summarize here goes over some of the specific bad vaccine reactions for SARS-1. I expect similar challenges for SARS-2.


In situations where someone took the vaccine, then gets infected by the contagion, they can have a bad reaction where the course of the disease is more severe than if they had not been vaccinated at all.*

Here's some of what we know about those bad reactions**

  • Th-2 type immune activation definitely happens
    • This is an allergy-like immune response
    • Th-2 reactions happen in severe cases of COVID-19 generally
  • There might also be a bad complement system related reaction
    • Complement system means protein complexes that kill cells by poking holes in membranes
  • Antibody Dependent Enhancement (ADE) might be possible, but is not likely
    • This is when imperfect antibodies are used as an anchor for the virus to infect white blood cells.
    • UPDATE: A related thing now strikes me as somewhat likely. It might be fusing with some white blood cells (at least T-cells) and ordering them to apoptose (activate cell-death). Article, h/t CellBioGuy.

* Going off of some other bits of research on this, these individuals probably have lower virus-titer, but higher severity and lethality. A damaging immune response, basically.

** Which vaccine types cause this bad reaction? For SARS-1, any whole-S-protein vaccines were more prone to this bad reaction. Some smaller S-protein fragments didn't have this issue, hopefully the same fix works for SARS-2. I heard of at least one case where an N-protein-only vaccine attempt also resulted in the Th-2 reaction, though. It's not totally clear how to avoid triggering it.

So maybe the speed-up you really want is to vaccinate, then deliberately expose to the live virus, and monitor what happens?

This is the type of test I'd rather we do on animal models than humans, to be frank. It seemed that you could test this phenomenon just fine with SARS-1 in animal models.

Thank you, this is the sort of answer I was looking for- I'd naively had the prior that "no effect" was the only non-negligible possibility besides "positive effect".

I'm normally on team "the FDA is making everything too slow," but in this case I actually think there's good reason to be really careful with those Phase-1 vaccine development trials. You don't need to be using a live vaccine to actively make things worse!

How long does testing for this problem actually take? I'm not sure.

If we were sure this was the only thing we needed to worry about (but I don't think that it is)... getting a line of evidence where some vaccine takers are exposed deliberately some time after vaccination could maybe speed up weeding out v

... (read more)
Informed volunteers would be heroes, but I think there are enough heroes to make vaccines available months sooner and to save millions of lives. At least it should be in the Overton window to ask for a voluntary trial with the understanding that there's substantial risk.
While we should at least ask and assess that question... Yes, it would be good to rule out those things that we do know to expect. And I think animal results* could check this one somewhat. But corners are already being cut, and I still expect some degree of surprises. I do feel like there's a lot we don't know with this virus. I don't know that the problems will be limited to the things we currently know to look for, and I'd be a little surprised if timing was not at least a bit of an influencing factor. * Apparently they haven't found/developed an easy animal that catches the virus, but they are doing animal testing in parallel to check the type of immune response? And the vaccine test result with the 4 macaques at least seemed promising, now up by another 10 macaques tested with that same inactivated-virus vaccine. ---------------------------------------- While it definitely helps that we have some experience with SARS-1, we can't totally rely on what we know about SARS-1 and trust that it'll apply to SARS-2. (I think SARS-1 and SARS-2's genetic similarity was said to be only ~80%? This is about as much as we share in common with cows. There can be meaningful differences between the two.) Here's just one example. Did you see the "UPDATE" I added to my answer above? Says something like "Oh hey, I guess it probably does make immune cells apoptose?" SARS-1 doesn't do that thing. As in, the article specifically mentions that they tried with SARS-1 and SARS-2, and only SARS-2 gets into T-cells like this. And they weren't sure which receptor was responsible. It's great to see a lot of science happening on this, but it's also something of a marker of our vast uncertainties paired with its high priority.

Adele Lopez

Mar 18, 2020


From what I can tell, it looks like the main danger is with a live vaccine, where the vaccine can give the disease to a large number of people (biggest actual disaster seems to have been the Cutter incident, which infected 40,000 people with polio).

I assume that the trial is also there to catch potential black swan issues.

IIRC the Covid-19 vaccines on trial are not live, so the case for doing the 14 month watch was not as strong as I expected. Certainly worth considering more carefully at least.

Whether or not a person gets a disease when you give them a live vaccine is information you learn very soon. It's seems implossible to me that a live vaccine causes the virus 2 months afterwards let alone 12.

2Adele Lopez4y
That sounds right to me.

Thomas Kwa

Mar 19, 2020


I don't think the timeline for Phase 1 trials looks anything like a 14 month delay before Phase 2 trials start.

  • [https://www.lesswrong.com/users/adele-lopez-1] already mentioned the live vs inactivated vaccine distinction.

  • Metaculus (admittedly not the best source of predictions) gives 45% that a vaccine is distributed starting in 2020. One user gives only 70%, taking into account the high urgency and high risk-tolerance of countries like China.

  • The American NIH says "If the clinical trial enrolls participants as planned, researchers hope to have initial data from the clinical trial within three months." This means either (a) they're being slightly misleading, or (b) that further trials will start immediately after that point.

Here's a news article reporting a 14-month Phase 1 trial for the COVID-19 vaccine, and I've seen the "12-18 months until vaccine deployment" timeline from Dr. Fauci and the NIH in several sources.

I think our main confusion is whether Phase 1 trials have to be complete before Phase 2-3 trials start. Surely if Phase 1 trials took 14 months and Phase 2 and 3 trials take additional serial time, there's no way to get the vaccine in mass production within 12-18 months? I'm not 100% sure of this, though.

You're right; the current plan condenses and overlaps the three phases in order to save a lot of time.

You're right, you can overlap Phases if the FDA cooperates.


Apr 22, 2020


There's a very tiny percentage chance that there's a completely unexpected long-term complication. Widespread distribution and vaccination with such a complication could be extinction-level.

Uhuh, and waiting a year will help us avoid this how? I mean maybe not giving it to the people in the underground bunker that we've stuck a breeding population in that can repopulate the earth after the non killer robot apocalypse makes sense on these grounds, but they aren't going to catch Covid-19 anyways.

If it starts to kill everyone within 14 months, we'll almost certainly know before everyone in poor countries gets it distributed to them, and then the underdeveloped inherit the world.

The odds that there's some serious side effect that isn't extinction-level are many orders of magnitude higher, and the approval system was made in advance with the full knowledge and careful consideration of the potential of epidemics.
3Timothy Underwood4y
Based on my priors about how groups like the FDA and CDC work that seems unlikely to be true. My strong impression is that the system is designed to minimize the odds of things going wrong in a way that will generate headlines based on errors of commission. Is there any source showing how the decided on this trial duration as something that would balance the risks involved with a deadly epidemic?
I believe so, but I lack the requisite domain-specific knowledge to extract it, or even to evaluate those reasons once they have been extracted. The thing about the size of the federal government is that there was a team of people with domain-specific knowledge integrating and responsive to public comments and suggestions from people with and without domain-specific knowledge. Their records *ARE* available, if you can figure out what to ask for. The general summary is probably fairly accurate, but it would be a major error to think that the actual policy was strictly to highly optimize for the fairly accurate summary.


Mar 18, 2020


I can't remember the term it's antibody specific amplification or something, but some promising vaccines will actually make you more susceptible.

I think we will likely get some acceleration out of heroic volunteers, if not in the US then elsewhere.

The SARS-1 virus has been known for 15+ years now, and vaccine trials resulted in vaccine-induced infection aggravation (see also here). There is some reasonable concern that vaccination for COVID could trigger the cytokine storm effect. It doesn't help that the novel coronavirus is already mutating, which " raises the alarm that the ongoing vaccine development may become futile in future epidemic if more mutations were identified." Coronaviruses are hard to fight.

William Walker

May 04, 2020


In 1957 they put together a flu vaccine in four months.

Of course, in 1969 we had flight-ready NERVA nuclear rocket engines. In 1962, Orion was ready for nuclear flight... progress doesn't always go in a straight line.

Least of all in biotech, where our incentives are so twisted that we use the most dangerous treatments first.

Moderna says they'll have their RNA vaccine in production by July, BTW. And there are safer treatments under study:


7 comments, sorted by Click to highlight new comments since: Today at 11:09 AM

The main piece of data that would help answer this question is case-studies of past vaccines, whether they had safety problems and what those problems were, and when the problems manifested. Given that there's a new influenza vaccine every year, and I've never heard of any year's influenza vaccine being rejected on safety grounds, my guess is that 18 months is much too conservative.

I think each year's flu vaccine is a slight modification on an existing vaccine. This may well (read: I have no idea, but it sounds plausible) make it faster to safety test the flu vaccine than a vaccine for a novel disease.

This is correct. We have lots of infrastructure and expertise for making new flu vaccines every year. It's not a good model for how long we should expect safety testing to take for a vaccine for a new virus. We don't have any licensed vaccines for any coronavirus, for example.

I hardly think it answers the question, but this might be of interest: https://www.cdc.gov/vaccinesafety/concerns/concerns-history.html

Are there any sources that describe why 14 months is the trial period (or provide justification for picking trial periods of various lengths)?

This would be nice to have for 'magic numbers' in general – as is common in well-documented source code.

Seems like it ought to be more of a continuous variable, rather than this discrete 14 month trial: at time t, we've observed x people for y months to see if they have wierd long-term side effects, so we should be willing to vaccinate z more people.