Developing the Covid-19 vaccine

[shpalman]

Good news. Another cheaper vaccine, and one that will probably be available to countries that have trouble accessing medication because of US sanctions, like Iran and Venezuela, which have both been badly hit. And more for the rest of us, I guess.

... and Hungary too, which might be an issue if the EU doesn't certify it.

Apparently it's more effective than the Oxford one because it uses two different adenoviral vectors for its two doses.

They've only had 39 cases of covid in the trial so the result isn't as robust as from the trials which have performed their analysis after >90 cases.

[Bird on a Fire]

Good news. Another cheaper vaccine, and one that will probably be available to countries that have trouble accessing medication because of US sanctions, like Iran and Venezuela, which have both been badly hit. And more for the rest of us, I guess.

... and Hungary too, which might be an issue if the EU doesn't certify it.

Apparently it's more effective than the Oxford one because it uses two different adenoviral vectors for its two doses.

They've only had 39 cases of covid in the trial so the result isn't as robust as from the trials which have performed their analysis after >90 cases.

Hungary wouldn't have any trouble accessing other vaccines, and isn't under sanctions (yet, lol). That's just Orban and his cronies looting public money.

[sTeamTraen]

With those numbers and proportionate numbers of cases, there would be 31 cases with the half dose first and 100 with the full dose first. 90% (62%) efficiency suggests to me that the treatment group had 10% (38%) of the number of cases of the control group. That would mean (3 Tx, 28 control) cases in the half-dose trial and (27 Tx, 73 control) in the full-dose trial, and that would also give the expected (30 Tx, 101 control) in the overall score, for an average efficacy of 70% (i.e., the vaccine prevented 71 out of 101 cases in the treatment group). Assuming equal sample sizes in each group:

My previous post on this (re the AstraZeneca vaccine) contained an error (for the full-dose cohort I had calculated on the basis of the total sample for the study, not the sample for that cohort) and a faulty assumption (i.e., equal numbers of people in each group). It was pointed out to me that in many of the other trials, there have been more people in the treatment group than the control group, so I revised my numbers (and spotted the first error myself while doing so).

Without knowing the exact number in each group we can't determine how many cases there were, but if we assume it's about 2:1 (treatment:control), we could have numbers like this:

Half-dose: Treatment N=1804, 5 cases; Control N=937, 26 cases
1 - ((5/1804)/(26/937)) = 0.9001151 = 90.01151%

Full-dose: Treatment N=6074, 45 cases; Control N=2821, 55 cases
1 - ((45/6074)/(55/2821)) = 0.6200048 = 62.00048%

(Of course, the percentages are unlikely to be anything like as precisely close to an integer as those.)

Still, though, 5 cases is a small number. One case more or fewer would change the effectiveness of the half-dose regime by 2 percentage points. But of course, if the vaccine was 100% effective, we would have 0 cases, and be jumping up and down, even though that's a *very* small number. Maybe a better indication of the sample size problem is having only 26 cases in the control group.

[jdc]

That seems reasonable put like that. I'm just anticipating unexpected stumbling blocks to emerge.

I'll take your bet - either we get a vaccine, or I get some money. 2021 is looking up.

Consolation betting - same thinking my dad used when he went to the bookies to put £200 on the UK voting for Brexit.

I think stumbling blocks have already been a factor in delaying the progress so far (not sure how unexpected they've been), in terms of things like Novavax having trouble scaling up or J&J/AZ pausing their trials. I suppose we could see more things cropping up but I'd have thought most of the opportunity for stumbling blocks would be past by now?

They've got through phase 1 and 2 trials, and are well on with phase 3 so various hurdles like problems finding enough recruits or not getting the hoped-for phase 1/2 results have already not cropped up. We've already been through some of the others (as above). What's left? Disappointing phase 3 results comes to mind (please no), or more of the adverse events that have briefly paused trials. After that isn't it mostly about passing regulatory checks? I don't know much about this element of vaccines, I'm usually reading about alleged problems with existing vaccines rather than the approval process for new jabs. Shame tomp isn't here, I bet he'd know.

I asked tomp, he has come back with the following:

Normally it takes months to approve a new medicine cos the company sends all the data from pre-clinical to the latest phase 3 results, via manufacturing and packaging and everything to us and then the assessors go through it all slowly and methodically and then things like risk might plans are discussed and agreed.

For a seasonal vaccine, it's often done as a variation to the license, which is a far shorter process and they pretty much just need to demonstrate human safety and efficacy equivalent to previous vaccines.

For the covid vaccines, we have been doing two things to speed stuff up. 1st we assessed, agreed and pre-approved packaging and labelling as much as possible. Then we have been doing a rolling review of data so the pre-clinical, phase 1 and 2 studies will already have been assessed and GMP compliance at proposed sites is being done too.

We also agreed to checkpoints throughout the studies when certain numbers of patients had been vaccinated/placeboed and certain numbers had caught COVID. So, depending on the apparent efficacy, we will know whether we can, in theory grant conditional approval with the studies still ongoing.

So when they send us the data, the review process will be far shorter. And, of course, since the whole world is desperate for this and there's silly national pride at stake, every regulator will be dedicating the max useful number of resources to this to speed up the decision-making process further still to beat the others.

Normally we have a CHMP meeting at routine intervals (the next is 7-10 December) and documents need submitting well in advance. This time, I expect an ad-hoc extraordinary CHMP for no more than 1 week after the rapporteur has finished their assessment.

The agency closes for Christmas on 23 December this year so maybe in the 2 weeks between then or perhaps twixt Xmas and New year. Since nobody will be doing much at Christmas and we are now used to virtual meetings and we'd all like to know we were doing our bit to help the world, it is very possible.

That's great. Thanks to tomp and his proxy poster.

[jdc]

Was the UK given the half first dose and Brazil the full first dose ? If so could the different % success be due to something to do with the countries ?

When university researchers were distributing the vaccine at the end of April, around the start of Oxford and AstraZeneca’s partnership, they noticed expected side effects such as fatigue, headaches or arm aches were milder than expected.

“So we went back and checked … and we found out that they had underpredicted the dose of the vaccine by half,” said Pangalos.

Instead of restarting the trial, he said researchers decided to continue with the half dose and administer the full dose booster shot at the scheduled time ...

About 3,000 people were given the half dose and then a full dose four weeks later, with data showing 90% were protected. In the larger group, who were given two full doses also four weeks apart, efficacy was 62%.

Volunteers in Brazil have begun receiving a trial vaccine against COVID-19, in Latin America’s first phase 3 COVID-19 clinical trial.

The trial officially began on Saturday 20th June and will enrol 5,000 volunteers across the country. Vaccinations will take place in Sao Paulo, Rio de Janeiro and a site in the Northeast of Brazil.

The pooled analysis included data from the COV002 Phase II/III trial in the UK and COV003 Phase III trial in Brazil. Over 23,000 participants are being assessed ... COV002 is a single-blinded, multi-centre, randomised, controlled Phase II/III trial assessing the safety, efficacy and immunogenicity of AZD1222 in 12,390 participants in the UK. ... COV003 is a single-blinded, multi-centre, randomised, controlled Phase III trial assessing the safety, efficacy, and immunogenicity of AZD1222 in 10,300 participants in Brazil.

So... 3k UK volunteers got the half dose, 9k UK vols got the full dose, and 10k Brazil vols all got the full dose. I think. The full dose figures are therefore for Brazil + UK and the half dose figures are UK only. Does that look right?

[sTeamTraen]

Was the UK given the half first dose and Brazil the full first dose ? If so could the different % success be due to something to do with the countries ?

See this excellent article from the always-fabulous Hilda Bastian. There were a lot of differences between the UK cohort and the Brazilian cohort, and there are a lot of questions about the trials of the Oxford vaccine.

[PeteB]

See this excellent article from the always-fabulous Hilda Bastian. There were a lot of differences between the UK cohort and the Brazilian cohort, and there are a lot of questions about the trials of the Oxford vaccine.

Thanks - looks like the trials were a bit of a mess, don't know if it is incompetence or covering something up

[MartinDurkin]

Today's update from the Zoe team on the Vaccines

https://www.youtube.com/watch?v=rD6t5OHbDN0

ETA: for some reason the link jumps to the end of the video for me, but hopefully can be reset to the start once you are in Youtube.

[jdc]

Was the UK given the half first dose and Brazil the full first dose ? If so could the different % success be due to something to do with the countries ?

See this excellent article from the always-fabulous Hilda Bastian. There were a lot of differences between the UK cohort and the Brazilian cohort, and there are a lot of questions about the trials of the Oxford vaccine.

Blimey. So many problems highlighted there it's hard to know where to start.

How about: "In fact, it’s impossible to know, at this point, just how many analyses these researchers have run, and on which data."

[sTeamTraen]

How about: "In fact, it’s impossible to know, at this point, just how many analyses these researchers have run, and on which data."

It's very sad that in 2020 we have claims like the ones we saw on Monday being made in a press release, and not even one that's accompanying a preprint. I know quite a few proponents of open science at Oxford and the couple I've spoken to are more than a little disappointed.

[shpalman]

With those numbers and proportionate numbers of cases, there would be 31 cases with the half dose first and 100 with the full dose first. 90% (62%) efficiency suggests to me that the treatment group had 10% (38%) of the number of cases of the control group. That would mean (3 Tx, 28 control) cases in the half-dose trial and (27 Tx, 73 control) in the full-dose trial, and that would also give the expected (30 Tx, 101 control) in the overall score, for an average efficacy of 70% (i.e., the vaccine prevented 71 out of 101 cases in the treatment group). Assuming equal sample sizes in each group:

My previous post on this (re the AstraZeneca vaccine) contained an error (for the full-dose cohort I had calculated on the basis of the total sample for the study, not the sample for that cohort) and a faulty assumption (i.e., equal numbers of people in each group). It was pointed out to me that in many of the other trials, there have been more people in the treatment group than the control group, so I revised my numbers (and spotted the first error myself while doing so).

Without knowing the exact number in each group we can't determine how many cases there were, but if we assume it's about 2:1 (treatment:control), we could have numbers like this:

Half-dose: Treatment N=1804, 5 cases; Control N=937, 26 cases
1 - ((5/1804)/(26/937)) = 0.9001151 = 90.01151%

Full-dose: Treatment N=6074, 45 cases; Control N=2821, 55 cases
1 - ((45/6074)/(55/2821)) = 0.6200048 = 62.00048%

(Of course, the percentages are unlikely to be anything like as precisely close to an integer as those.)

Still, though, 5 cases is a small number. One case more or fewer would change the effectiveness of the half-dose regime by 2 percentage points. But of course, if the vaccine was 100% effective, we would have 0 cases, and be jumping up and down, even though that's a *very* small number. Maybe a better indication of the sample size problem is having only 26 cases in the control group.

Efficacy is (cases among placebo subjects - cases among vaccinated subjects)/(cases among placebo subjects) isn't it?

https://phastar.com/blog/250-statistici ... ccine-data

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

[sTeamTraen]

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Ha, yes, you're right. I did those numbers yesterday evening (not fovkerf-influenced, AFAIK) and was probably getting confused about what the 100 meant. Damn round numbers.

When I stop making that mistake, I have difficulty in getting the numbers to work unless the size of the treatment and control groups are close to identical.

[sTeamTraen]

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Ha, yes, you're right. I did those numbers yesterday evening (not fovkerf-influenced, AFAIK) and was probably getting confused about what the 100 meant. Damn round numbers.

When I stop making that mistake, I have difficulty in getting the numbers to work unless the size of the treatment and control groups are close to identical.

Just missed the edit window... it is completely absurd that we are having to attempt to construct the contingency tables from the frequencies.

[Bird on a Fire]

Look at all these people doubting the word of Big Pharma press releases! I bet they're all big fans of Chairman Mao

[EACLucifer]

Look at all these people doubting the word of Big Pharma press releases! I bet they're all big fans of Chairman Mao

In China, discussion of this sort of thing even in a far more private forum than this could land you in court charged with "rumour mongering", or getting roughed up by the local police - both happened to Li Wenliang before his death.

Just for context, you know.

It's vital proper evidence is submitted prior to approval, but this kind of online discussion simply isn't possible to do safely in China.

[jimbob]

With those numbers and proportionate numbers of cases, there would be 31 cases with the half dose first and 100 with the full dose first. 90% (62%) efficiency suggests to me that the treatment group had 10% (38%) of the number of cases of the control group. That would mean (3 Tx, 28 control) cases in the half-dose trial and (27 Tx, 73 control) in the full-dose trial, and that would also give the expected (30 Tx, 101 control) in the overall score, for an average efficacy of 70% (i.e., the vaccine prevented 71 out of 101 cases in the treatment group). Assuming equal sample sizes in each group:

My previous post on this (re the AstraZeneca vaccine) contained an error (for the full-dose cohort I had calculated on the basis of the total sample for the study, not the sample for that cohort) and a faulty assumption (i.e., equal numbers of people in each group). It was pointed out to me that in many of the other trials, there have been more people in the treatment group than the control group, so I revised my numbers (and spotted the first error myself while doing so).

Without knowing the exact number in each group we can't determine how many cases there were, but if we assume it's about 2:1 (treatment:control), we could have numbers like this:

Half-dose: Treatment N=1804, 5 cases; Control N=937, 26 cases
1 - ((5/1804)/(26/937)) = 0.9001151 = 90.01151%

Full-dose: Treatment N=6074, 45 cases; Control N=2821, 55 cases
1 - ((45/6074)/(55/2821)) = 0.6200048 = 62.00048%

(Of course, the percentages are unlikely to be anything like as precisely close to an integer as those.)

Still, though, 5 cases is a small number. One case more or fewer would change the effectiveness of the half-dose regime by 2 percentage points. But of course, if the vaccine was 100% effective, we would have 0 cases, and be jumping up and down, even though that's a *very* small number. Maybe a better indication of the sample size problem is having only 26 cases in the control group.

Efficacy is (cases among placebo subjects - cases among vaccinated subjects)/(cases among placebo subjects) isn't it?

https://phastar.com/blog/250-statistici ... ccine-data

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Inside Science had a discussion about vaccine effectiveness, and the interviewee obviously got it wrong, as she was implying that it was worth vaccination with a 40% effectiveness on occasion.

I assume it's proportions rather than cases but otherwise, Yes, that's what I understand it too.

So 40% effective is 100/x placebo and 60/x vaccinated.

[shpalman]

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Ha, yes, you're right. I did those numbers yesterday evening (not fovkerf-influenced, AFAIK) and was probably getting confused about what the 100 meant. Damn round numbers.

When I stop making that mistake, I have difficulty in getting the numbers to work unless the size of the treatment and control groups are close to identical.

I'm sure I read somewhere that one subset of the study had a 2:1 ratio between Tx and control but I can't find it now.

Inside Science had a discussion about vaccine effectiveness, and the interviewee obviously got it wrong, as she was implying that it was worth vaccination with a 40% effectiveness on occasion.

I assume it's proportions rather than cases but otherwise, Yes, that's what I understand it too.

So 40% effective is 100/x placebo and 60/x vaccinated.

Yes it should be proportions, but I get the published percentages from the published numbers of cases if x is the same between Tx and placebo groups (for all the different vaccines reported so far).

When the 70% result came out for the Oxford vaccine, the news was basically "ah but it might be 90%" but also "well the 'flu one is only 40-60% and anything over 50% counts as working". This latter point was obviously not made when the Moderna and Pfizer results were reported. I can't imagine that the results would have been reported in quite the same way by the UK media if it hadn't been developed by a UK team.

But the Oxford trial featured weekly swabs, so would pick up asymptomatic cases in a way which the other trials won't. Then it depends if you assume the vaccine reduces case severity (a severe case becomes a mild case and a mild case becomes an asymptomatic case) or stops severe cases but leaves the asymptomatic ones. But it makes sense to me that a trial which is regularly swabbing participants will pick up more cases in the vaccinated arm than one which relies on participants presenting with some sorts of symptoms.

[jimbob]

Inside Science had a discussion about vaccine effectiveness, and the interviewee obviously got it wrong, as she was implying that it was worth vaccination with a 40% effectiveness on occasion.

She was saying it was the ratio of control:trial so a 90% effectiveness meant that 90% of the infections were in the control leg.

It was clearly a brain fart on her part.

[sTeamTraen]

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Do we have an external source for that? I only see a total of 131 infections in the AZ press release. With the 30-101 constraint it's hard to make anything work except with a 50-50 Tx:control split (for which I'd also quite like to see a source!)

[shpalman]

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Do we have an external source for that? I only see a total of 131 infections in the AZ press release. With the 30-101 constraint it's hard to make anything work except with a 50-50 Tx:control split (for which I'd also quite like to see a source!)

Overall, there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection. The researchers said it worked out at 70% protection, which is better than the seasonal flu jab.

See also ... about 24,000 trial participants (treatment and control groups). In the treatment group, 8,895 participants received two full doses of the vaccine, spaced one month apart, and 2,741 patients got a half dose at first, followed by a full dose a month later.

[sTeamTraen]

We know that in total there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection so I've no idea what you are doing with your numbers, you've got a total of 50 cases in vaccinated subjects.

Do we have an external source for that? I only see a total of 131 infections in the AZ press release. With the 30-101 constraint it's hard to make anything work except with a 50-50 Tx:control split (for which I'd also quite like to see a source!)

Overall, there were 30 cases of Covid in people who had two doses of the vaccine and 101 cases in people who received a dummy injection. The researchers said it worked out at 70% protection, which is better than the seasonal flu jab.

Huh. It would be nice to have something that might not be a journalist writing down numbers presented orally. This is a f.cking mess.

See also ... about 24,000 trial participants (treatment and control groups). In the treatment group, 8,895 participants received two full doses of the vaccine, spaced one month apart, and 2,741 patients got a half dose at first, followed by a full dose a month later.

Huh. So the reported Ns in the press release were for the treatment groups only. Well f.cking done, Astra f.cking Zeneca's f.cking PR flacks.

So that makes 11,636 people in the treatment group(s) and possibly even slightly more in the control group(s). If there were only 1,400 in the half-dose control group to make 2:1, you'd need 11,000 in the full-dose control group, making 9:11. FFS.

[shpalman]

Where are you getting this 2:1 thing from?

[jdc]

https://www.bloomberg.com/news/articles ... r-business

AstraZeneca Plc is likely to conduct an additional global trial to assess the efficacy of its Covid-19 vaccine, according to the company’s chief executive officer, after current studies raised questions over its level of protection.

“Now that we’ve found what looks like a better efficacy we have to validate this, so we need to do an additional study,” CEO Pascal Soriot said in his first interview since the data were released. It will probably be another “international study, but this one could be faster because we know the efficacy is high so we need a smaller number of patients.”

Soriot said he didn’t expect the additional trial to hold up regulatory approvals in the U.K. and European Union. Clearance from the U.S. Food and Drug Administration may take longer because the regulator is unlikely to approve the vaccine on the basis of studies conducted elsewhere, especially given the questions over the results, he said. Authorization in some countries is still expected before the end of the year, he said.

[jdc]

A day after the data were unveiled the head of the U.S. vaccine program known as Operation Warp Speed said that the regimen showing the higher level of effectiveness was tested in a younger population. He also said that the half-dose was given to some people because of an error in the quantity of vaccine put into some vials. None of those details were disclosed in Astra or Oxford’s original statements.

Astra and researchers have declined to provide more data ahead of a peer-reviewed analysis that is expected to be published in the coming weeks. Results have been submitted to an undisclosed journal, Astra said in a statement.

[shpalman]

“... this one could be faster because we know the efficacy is high so we need a smaller number of patients.”

Does it actually work like that?

If you want the trial to go faster and/or need fewer subjects you need to do it in a hotter covid spot.

By the way, the Guardian have mentioned that data will be published in the Lancet over the weekend.