Skip to content

When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS Absolutely Maybe

Phase 3 Results in Sight for Some, While One Vaccine is Put on Hold: Covid-19 Vaccine Race, Month 9

Cartoon of vial of vaccine looming up on coronavirus

We’re now into the ninth month since sequencing of the genome of SARS-Cov-2, the virus that causes Covid-19. And this is the fourth of my month-by-month posts about the vaccine race. This month, there’s a lot of good news: We are on the brink of the first phase 3 trial results. But it’s a bit of a wild ride, and there’s a lot to sort through. Let’s start, though, with the vaccine trials put on hold on September 9 to investigate an unexplained serious illness in a participant.

The question marks around the Oxford/AstraZeneca vaccine

One of the original frontrunners, this vaccine has been hitting bumps and sliding behind schedule. Back in mid-May, their spokesperson said “Oxford is likely to have the first efficacy data in the world”, perhaps in August, and suggested the vaccine could be in use in the fall.

Instead, in August, despite having a few months’ head start – its phase 2/3 trial in the UK began in the last week of May – the Oxford/AstraZeneca vaccine was no longer among the first 5 vaccines likely to report phase 3 trial results. With the news this week that the trials for the vaccine are on hold, more ground will be lost – and concerns in the wake of the announcement of a serious adverse event could slow down recruitment, even if independent review gives them an all-clear quickly.

Here’s the publicly available information, in a timeline with some context:

  • July ?: A serious adverse event for a participant in the phase 2/3 trial in the UK was reported, and that triggered a pause. This was not discussed publicly at the time.
  • July 12: An amended version of the information leaflet for prospective volunteers included this: “one volunteer in the trials of ChAdOx1 nCoV-19 developed symptoms of transverse myelitis (inflammation in the spinal cord), which has not required medical treatment and is being investigated, though the cause is uncertain”. (Nature reported the patient information contents this week.)
  • August 5: The information leaflet was amended again to say that the participant had experienced an “unrelated neurological illness”. A spokesperson has now reported it was multiple sclerosis.
  • August 14: After about 10 weeks, the UK trial had only recruited 7,573 out of an original target of 10,260 (now 12,330). (For perspective, another Covid-19 vaccine trial in a much smaller country managed up to 7,000 a week.) There were also 3,032 participants recruited to a 5,000-person phase 3 trial in Brazil, and 914 in a 2,000-person phase 2 trial in South Africa. Counting those in their phase 1/2 trial, about 12,500 people had had the vaccine injected at that point.
  • August 31: Recruitment began for a 30,000-person phase 3 trial in the US.
  • September 6: Trials put on hold for safety review.
  • September 8: Journalists at STAT News reported that the US trial had been put on hold because of an unexplained serious adverse event in the UK trial, which was to be investigated by an independent committee. The illness has yet to be explained, but AstraZeneca said the volunteer was in hospital, with “symptoms consistent with a condition called transverse myelitis, or inflammation of the spinal cord”. The person was recovering, they said, and expected to leave hospital that day. (There’s been no further word that I’ve been able to find.)
  • September 9: The UK and South African trials were confirmed to also be on hold. However, the Serum Institute of India, which recently began a 1,700-person phase 3 trial of the vaccine, tweeted they were unaffected and their trial was not on hold. India’s regulator didn’t see it that way, issuing the Institute with a notice demanding they show cause why their license to run clinical trials shouldn’t be suspended for not notifying them of the event.
  • September 10: The Indian trial was put on hold.
  • [Update] September 12: AstraZeneca announced UK regulator cleared the UK trial to re-start based on recommendation of the independent review committee. Regulators reportedly cleared the Indian trial to proceed, and the Brazilian one.
  • [Update] September 18: Regulators reportedly cleared the South African trial to continue. US trial still on hold.

We haven’t been told if this second event is transverse myelitis, or enough details around the July event. Transverse myelitis is a rare condition, but you can see warnings about it in product information leaflets and informed consent forms for several vaccines. Although viruses can cause it, there isn’t enough evidence to be certain that vaccines do. There isn’t enough evidence to give vaccines the all-clear on it either. That’s according to both the National Academy of Medicine (formerly IOM) and the US Agency for Healthcare Research and Quality (AHRQ) (in 2011 and 2014 respectively).

The NIH Director has said that if it’s determined that the Oxford/AstraZeneca vaccine caused serious neurological damage, that’s the end of it, and all the doses that have been produced “will be thrown away”. But if you’ve been following the events of the last few days, you’ll likely have seen others saying a single serious adverse event in tens of thousands of people shouldn’t doom a vaccine.

I think that’s a hard case to make if it were found likely to cause an extremely serious condition, among the first 10- or 20,000 people vaccinated. If it were to happen at that rate among a billion people, tens of thousands of people would be severely harmed. It’s doubtful that most people would agree to a vaccine after seeing that happening anyway, given a choice of vaccines.

Even if review determines this second event wasn’t caused by the vaccine, the vaccine doesn’t have much buffer left. If lightning struck a third time, the concern would be intense.


The participant information for the trial was amended, and now says there have been an unspecified number of “unexplained neurological symptoms” – could there be more than 2? – and not given the total all-clear:

After independent review, these illnesses were either considered unlikely to be associated with the vaccine or there was insufficient evidence to say for certain that the illnesses were or were not related to the vaccine.

Safety and acceptability were already issues to keep a keen eye on for this vaccine. After its trials got started, the team added an arm with acetaminophen to try to reduce the adverse events, but it didn’t help all that much. As I discussed previously, the rate of some severe adverse events in the week after vaccination almost reached the level designated “common” by the FDA’s guidance, and would be classified as common by the EU equivalent, the European Medicines Agency.

Here’s a table I put together in a previous post, with data on headaches from the vaccine’s phase 1/2 trial, to show what that looks like. Note the comparison group is another vaccine: There was no placebo group. And the numbers are small, too, so a lot could change. What’s more, the dose in the US trial is different to the doses in the UK trials.

Covid-19 (n = 487)32%34%28%6%
Meningococcal (n = 477)59%35%5%1%
Headaches for people without preventive paracetamol

Other vaccines, including several of the Covid-19 vaccines that have reported clinical trial results, have reported lower rates of adverse events so far. The bar for demonstrating adequate enough safety should be high for vaccines that would be the first of their kind to ever be approved for general use. No vaccine that’s based on the adenovirus that Oxford/AstraZeneca’s one is, has ever been approved for use in humans.

In August, further changes were made to the phase 2/3 trial in the UK:

Taken together, the new outcomes I’ve highlighted above raise more questions. Those outcomes – signs of immunity to the adenovirus in this vaccine – touch on a reported concern with adenovirus-based vaccines. These vaccines’ effectiveness could be weakened by pre-existing immunity to the vaccine carrying the antigen in, or gained from the first shot in the 2-injection course.

It’s all about balance. If effectiveness is high, then problematic tolerability is one thing. But if effectiveness is lower than you would like and adverse effects are higher, that could tip the balance even for a vaccine that “works”. A vaccine with low effectiveness and low uptake wouldn’t provide a community with anywhere near enough protection. Fingers crossed, though – we need all the good options we can get. Next let’s turn to the vaccines that do seem to have results coming soon.

The first fully-recruited phase 3 trial plus likely first results on Covid-19 protection: 3 inactivated virus vaccines

The first phase 3 trial to reach full enrolment was in the middle of August. It was a 15,000-participant trial in the United Arab Emirates (UAE), of the 2 Sinopharm inactivated virus vaccines – an “old school” type of vaccine. It was 5,000 for each of the vaccines, plus a 5,000-person placebo group. They got there in less than a month, with full enrolment in the middle of August. This is the trial that got up to 7,000 new participants a week. They’ll be looking at outcomes soon.

This trial is called #4Humanity, and now an international extension has started, apparently an additional 30,000 people. And by the end of August, another 16,000 people from the UAE had joined – so they’re past halfway to the 45,000 participant target for the combined trials. (You’ll see later how many more trials are in the works for these vaccines.)

#4Humanity has involved a lot of community mobilization and organization. The principal investigator is Nawal Al Kaabi, a chief medical officer, pediatrician, and vaccine researcher who trained in the UAE and Canada. She’s interviewed in this story about the trial. (Thanks to Dr Dalia Dawoud for pointing me to it.) In a conference presentation on August 26, Al Kaabi said that adverse effects in the trial were “within the expected range”, presumably meaning similar to the earlier stage trials. No data released yet to judge that though.

Meanwhile, the same message is coming from Brazil about the third inactivated virus vaccine from China, which is in a phase 3 trial there: Sinovac’s Coronavac. According to the president of the Butantã Institute running the trial, there have been no serious adverse events, with 4,000 having had their first vaccine. The trial is for 8,870 healthcare workers, and they are expecting to have the trial full by the end of September. Again, no data released yet to judge.

These vaccines, along with CanSino’s (see below), have emergency use authorization in China. Coronavac, for example, is reported to have been used for 24,000 people in China. They are reporting adverse events are very uncommon. The 2 from Sinopharm have reportedly been used by hundreds of thousands of people. As I said, this is a wild ride! And again – fingers crossed.

Overview of phase 3 trials with earlier phase results

The first 5 trials likely to have phase 3 results are rounded out by Germany’s BNT/Pfizer vaccine, and the NIH/Moderna vaccine from the US. They both started phase 3 trials in the US on the same day, at the end of July, although the BNT/Pfizer one is a compressed phase 1/2/3 trial. Both are for 30,000 participants.

However, the BNT/Pfizer trial is recruiting a bit more quickly, and in other countries as well as the US. As the 2-injections for that vaccine are 3 weeks apart – a week less than for NIH/Moderna’s vaccine – presumably they will have results sooner. Recruitment for the NIH/Moderna trial – called the Cove trial – has apparently slowed down to increase the proportion of people of color, who have been under-represented. The BNT/Pfizer trial has more diverse participation, through international sites.

Both trials now have websites, with recruitment numbers updated once a week. On September 7, BNT/Pfizer had 25,189 participants, and at the rate they’ve been going, expect it to be fully recruited any day now. On September 4, the Cove trial (NIH/Moderna) had 21,411 participants. They, too, must be getting very close. [Updates] On September 11, the number had increased to 23,497 – so it is slowing down. On September 12, BNT/Pfizer announced they were seeking FDA approval to expand the trial to 44,000, to include ages down to 16, and people with HIV or hepatitis (B and C), “as well as provide additional safety & efficacy data”. They “continue to expect that a conclusive readout on efficacy is likely by the end of October”.

You can see all the trials known to be in phase 3 in the table that follows. There are quite a few groups saying that they will be in phase 3 trials soon. But in the past few months, a few of these kinds of claims have turned out be more aspirational than reliable. So I’ve narrowed the overview a bit this month, to vaccines with phase 3 trials already underway, or at least with the trial registered along with particularly convincing claims that they will start this month.

For the trials included here, all the result reports (journal or preprint) and trial register entries associated with them are in my tagged public Zotero collection: now with 102 records, for vaccines from 44 groups, with 51 preclinical reports and 11 clinical trial reports. (More on that, with a link, at the bottom of this post.)


PreclinicalPhase 1

Phase 1/2

Phase 2

Phase 3


Johnson & Johnson
Primate, non-primaten.a.Unpublishedn.a.60,000
(Planning: Brazil, Chile, Colombia, Mexico, Peru, Philippines, South Africa, Ukraine, USA)
Non-primate only108n.a.50840,000 planned but not enough sites publicly confirmed
(Pakistan, planning – Saudi Arabia, Mexico)

Primate, non-primate (b2 only)b1:




b1 & b2:

(USA – incl 45 above)

30,000, adding 14,000*
(Argentina, Brazil, Turkey, USA – cleared to start in Germany)
ChAdOx1 nCov-19/AZD1222
UK (Oxford Uni)
Primate, non-primaten.a.1,077n.a.12,330**




Primate, non-primate45n.a.Unpublished30,000
Sputnik V (Gam-COVID-Vac)
Inactivated vaccines:
BBIBP-CorV x 2 (1 Wuhan, 1 Beijing)

Primate, non-primate
(Beijing only?)
n.a.320 (Wuhan)

n.a.15,000 (both vaccines)

30,000 (both vaccines)
(Bahrain, Jordan, UAE)

3,000 (Beijing only)

Apparently additional:

6,000 (both vaccines)

? (both vaccines)
(Morocco, Serbia)

? (Beijing only)
PicoVacc (CoronaVac)
Primate, non-primateUnpublished



(Reportedly approved to also run in another unnamed country)

Vaccines with published or preprint results (and/or Phase 3 trial register entry)
* a combined phase 1/2/3 trialwith 2,000 phase 1/2
** a combined phase 2/3 trial
*** an unrandomized phase 3 trial

n.a. = not applicable (where only combined phase 1/2 trial so far reported)
Sources: unless otherwise linked, the sources from my tagged public Zotero collection

Clinical trial results since the last round-up

Summaries of all the previously released clinical trials are in my previous posts. There have only been 2 new reports this month. Waiting in particular on the 600-person randomized phase 2 trial for the NIH/Moderna vaccine. That reached full recruitment on July 8, so it’s presumably over a month since the last person got their second injection. Last time, they didn’t do a preprint, so we might be waiting for a while as it wends its way through a journal’s editorial process.

BNT/Pfizer vaccine

There have been 3 of these in clinical trials: BNT162b2 is the one in the 30,000 trial, but there are still trials for b1 and b3 as well. A preprint of results for phase 1/2 of b2 was posted on August 28. It’s a small number of people: 13 groups of 15 people each in 2 age groups and 3 dose levels (195 people in total). In each of those groups, 3 people got placebo injections. Altogether, 72 people had the b2 vaccine, and only 24 had the dose that’s gone to phase 3 (30µg). So like the NIH/Moderna vaccine, there is still very little data in humans for this vaccine.

This preprint reports on antibodies and adverse events: a future preprint will report on T-cells. In this preprint, people 65 and older did better with the b2 vaccine than b1. There were no serious adverse events, 7/24 non-severe adverse events (29%) (of which less than half were considered related to the vaccine), and 2/24 had severe adverse events (8%). The rate of fever was 17% for 18-55 year-olds, and 8% for 65-85 year-olds.

A preclinical report in primates and non-primates for b2 was also posted as a preprint – and a phase 1/2 trial for b2 in China was registered.

Sputnik V (Gam-COVID-Vac)

Last month’s roundup post ended up with the strange story of Russia registering this vaccine – which is actually a pair of vaccines – after just 2 phase 1/2 trials with 38 people each. Although it’s a world’s worst case of jumping ahead of the evidence, it wasn’t the world’s first case of emergency use authorization – that was China, authorizing the CanSino vaccine for use in the military back on June 29.

If you’re interested in this vaccine, the best coverage I’ve found are the articles at Meduza (a news service by journalists based in Riga, because independent investigative journalism within Russia isn’t possible). Their August 12 article is a good jumping off point. (They also have an interesting Twitter account.)

But back to the clinical trial results. The report of the phase 1/2 trials was published in The Lancet on September 4. The trials tested 2 different forms of the vaccine – a frozen version (ideal) called Gam-COVID-Vac, and a version that doesn’t need to be frozen and could be used in parts of the world where maintaining a cold enough distribution chain isn’t possible. (That one’s called Gam-COVID-Vac-Lyo.)

There were different regimes, too, because they had 2 versions of the vaccine, each based on a different recombinant adenovirus vector. Participants got either 2 injections of a single vector, or an injection of each. And it’s that “1 of each” regime for Gam-COVID-Vac that’s going into phase 3. (They concluded the “Lyo” version worked adequately, although not as well as frozen, but it’s more complex to manufacture.)

There was no placebo group, which isn’t that unusual for phase 1/2 trials. What is odd, though, is that they didn’t randomly allocate people to the groups. The 38 people getting Gam-COVID-Vac regimes were mostly men – only 6 were women, and they were in the group getting 1 of each, leaving 2 of the groups 100% male. And they were a very young group of healthy people – for the 1 of each vaccine vector group, the average age was 26.4. They clocked up the highest rate of fever of any Covid-19 vaccine trial so far – everyone: 95% had mild fever, and the 1 other person had a moderate fever. Around half had headaches. There were no severe or serious adverse events.

The data on signs of immune response were heavily criticized in an open letter to the authors and Lancet from a group of scientists, suggesting “apparent duplication” of results for participants, which was a big leap. Some others have said it could be problematic data visualization, not data that were fudged in some way. Meduza reports that the authors have sent a rebuttal to the journal, and the deputy research director at the Gamaleya Institute (the home of this vaccine – hence “Gam”-COVID-Vac) said there weren’t mistakes. If only they had published the full data instead of just visualizations, this all could have been avoided, if he’s right. Sigh.

Meanwhile, the first batch of Gam-COVID-Vac has been released for general use, and Moscow’s mayor says he hopes the city’s population will be vaccinated within months.

Again, wishing them all the very best – but this is hair-raising.

In other recent news…

  • China’s vaccine regulator, the National Medical Products Administration (NMPA), issued its standards for Covid-19 vaccine approval. They will accept 50% effectiveness, but prefer at least 70%. (Equivalents from the FDA and the European Medicine Agency.)
  • The Canadian 696-participant phase 1/2 trial for the CanSino vaccine is now definitely off – a victim of China-Canada political tensions.
  • More in the ongoing saga of Moderna executives’ stock sales: NPR reports that they have now scooped up $90 million personally. Especially given how much public money has flowed into Moderna, they ask if there is more than just “bad optics” here.
  • A rare interview with the leader of Operation Warp Speed.
  • The latest Coalition for Epidemic Preparedness Innovations (CEPI) survey of capacity to produce Covid-19 vaccines concludes the globe now has enough for their target of 20% of people in all countries. That would need 2 billion doses, and they now estimate capacity to produce 2-4 billion globally. (But that doesn’t mean it will be distributed fairly.) They reported that India has the largest capacity for non-RNA vaccine substances, Europe for RNA-based substances, and China the greatest capacity for vaccines.
  • CEPI is banking on the Brighton Collaboration to produce standards for data on Covid-19 vaccine adverse events and harm/benefit assessment.
  • A Covid-19 vaccine nasal spray developed by the University of Hong Kong has been approved for clinical trials.
  • And the FDA Commissioner announced that the October 22 meeting of the FDA vaccine advisory committee will be virtual, and made public.

What a dramatic month! I think we can expect many more of those.


Cartoon of facing off coronavirus

All my Covid-10 vaccine race posts

All previous Covid-19 posts

My op-ed about Covid-19 vaccine adverse effects at WIRED

Disclosures: My only interest in Covid-19 trials is as a person worried about the virus, as one of my sons is immunocompromised. I have worked for an institute of the NIH in the past, but not the one working on the vaccine (NIAID). More about me.

The cartoons are my own (CC BY-NC-ND license). (More cartoons at Statistically Funny and on Tumblr.)

Background note: The timing for these roundup posts is based on the January 2020 sequencing of the genome for SARS-Cov-2. The sequencing was submitted for release on January 5, and published on January 12.

Sources for the table are included among the records in my public Zotero collection of Covid-19 vaccines with preclinical or clinical trial results either as publications or preprints, and associated trial registry entries. They are tagged with names and type of record (e.g. Phase 1). Please let me know if I’m missing any! On September 10, it included:

  • 44 vaccine groups with published or posted results;
  • 51 preclinical reports;
  • 2 phase 1 trial reports;
  • 7 phase 1/2 reports;
  • 2 phase 2 reports;
  • 40 trial registry entries associated with these vaccines (some with multiple, some with none).

Updates September 12: Added Oxford/AstraZeneca’s small unrandomized phase 3 trial in Russia; the report about Sinopharm’s vaccines being used for hundreds of thousands of people in China; and the updated recruitment data for the Cove trial.

Updates September 13: Re-start of the UK, Brazilian, and Indian trials for the Oxford/AstraZeneca vaccine; added section on the amended participant information; expansion to 44,000 planned for BNT/Pfizer. And a correction: the original described the adenovirus vector for the Oxford/AstraZeneca as adenovirus 5 – thanks to Snow Leopard for alerting me to the error.

Update September 16: Increase to 10,000 for Brazilian trial of Oxford/AstraZeneca vaccine.

Update September 20: Regulators clearing Oxford/AstraZeneca’s South African trial.

Update September 23: Sinopharm’s phase 3 trial in Argentina was registered: table corrected – had been based on a media report that it would be 4,000 people & both vaccines.

Update September 24: Added September 6 as the date Oxford/AstraZeneca trials put on hold, after media report.

Update October 16: Corrected table – originally had attributed preclinical results for Sinopharm vaccines to the Wuhan vaccine instead of the Beijing vaccine.

Update on 10 December: Correction to table – BNT/Pfizer results formerly listed as phase 1/2 moved to phase 1. Although titled phase 1/2, no phase 2 results were included. (Grateful to Julia Belluz from Vox, who pointed out this.)

  1. Just a note, you wrote “No vaccine based on an adenovirus 5, as the Oxford/AstraZeneca one is, has ever been approved for use in humans.”

    But the Ox/AZ vaccine is based on a chimpanzee adenovirus, not (Human) adenovirus 5 or adenovirus 26 like the other adenovirus based COVID19 vaccines.

    The rest of the article was great, keep up the good work.

    1. Yikes – thanks! Corrected, with acknowledgment. As far as I know, the only adenovirus-vector-based vaccine approved at all is the J&J ebola one (which is adenovirus 26).

Leave a Reply

Your email address will not be published. Required fields are marked *

Add your ORCID here. (e.g. 0000-0002-7299-680X)

Related Posts
Back to top