The vaccines | vaxxed boosted unvaxxed? New poll

[BD]

I'm realising how much stats have leaked out of my brain.

From (long term) memory significance level is a test of the data where 95% = 5% chance that the null hypothesis is true, confidence level is the probability that if the test was repeated again and again the results would be the same and confidence intervals being a range within which a result would be expected to fall in 95% of the time and constructed from significance level (or perhaps confidence level or both/either - I forget the details). All closely related yet different. I suspect I've been conflating one or more at times.

Kind of, but not quite.

The null hypothesis is either true, or it isn't - there's no probability assigned to it. A given vaccine either reduces cases of Covid, or it doesn't.

For the p-value, we assume that the null hypothesis is correct, and then the p-value denotes the probability of observing our observed data (or data more 'extreme') given the null hypothesis to be true. So we assume that our Covid vaccine does nothing, but then when we see the difference in case rates across the two placebo and the vaccine cohort, the p-value will give an indication of how likely this difference was to happen, under the assumption that there should be no difference.

I don't mean to be a nitpicky bore, but I just thought it might be interesting. And I've heard complaints from a lot of statisticians that p-values are so often misunderstood and misused, especially in medical statistics. To the point where research is being done in such a way that gives the best chance for a 'statistically significant' result, even if the corresponding research is no good (and I've experienced similar to this on a project I was helping on, where p-values were the be-all and end-all, despite us trying to say why they were pointless in our case).

 

[Pogue Mahone]

Kind of, but not quite.

The null hypothesis is either true, or it isn't - there's no probability assigned to it. A given vaccine either reduces cases of Covid, or it doesn't.

For the p-value, we assume that the null hypothesis is correct, and then the p-value denotes the probability of observing our observed data (or data more 'extreme') given the null hypothesis to be true. So we assume that our Covid vaccine does nothing, but then when we see the difference in case rates across the two placebo and the vaccine cohort, the p-value will give an indication of how likely this difference was to happen, under the assumption that there should be no difference.

I don't mean to be a nitpicky bore, but I just thought it might be interesting. And I've heard complaints from a lot of statisticians that p-values are so often misunderstood and misused, especially in medical statistics. To the point where research is being done in such a way that gives the best chance for a 'statistically significant' result, even if the corresponding research is no good (and I've experienced similar to this on a project I was helping on, where p-values were the be-all and end-all, despite us trying to say why they were pointless in our case).

Yeah, people often obsess about p-values too much, to the exclusion of other important factors.

One thing I’ve often wondered is p<0.05 important in fields other than medicine? For us it’s the holy grail. Obviously the smaller the p value the better but that 0.05 figure is the hard cut off between “yes it works” and “no it’s useless “ which always seemed a bit arbitrary to me!

 

[jojojo]

AZ have re-announced their efficacy stats from the USA data. It's now being reported as:

The primary endpoint, vaccine efficacy at preventing symptomatic COVID-19 was 76% (confidence interval (CI): 68% to 82%) occurring 15 days or more after receiving two doses given four weeks apart. In addition, results were comparable across age groups, with vaccine efficacy of 85% (CI: 58% to 95%) in adults 65 years and older. A key secondary endpoint, preventing severe or critical disease and hospitalisation, demonstrated 100% efficacy. There were eight cases of severe COVID-19 observed in the primary analysis with all of those cases in the placebo group.

The vaccine was well tolerated, and no safety concerns related to the vaccine were identified.

https://www.astrazeneca.com/media-c...ry-analysis-confirms-safety-and-efficacy.html

Which makes you wonder why they bothered to do the 79% version a couple of days ago - that just made them look shady.

I'd like to think they were busy doing more detailed analysis on the numbers - like the Novavax decision to announce an overall figure for the UK trial, but then split out a "original covid" and a "UK mutation" stat as well. Maybe the full data/report from AZ will give us a clue.

 

[BD]

Yeah, people often obsess about p-values too much, to the exclusion of other important factors.

One thing I’ve often wondered is p<0.05 important in fields other than medicine? For us it’s the holy grail. Obviously the smaller the p value the better but that 0.05 figure is the hard cut off between “yes it works” and “no it’s useless “ which always seemed a bit arbitrary to me!

I'm not too sure on other fields specifically, but I do know that medicine is always used as the example for 'p-hacking', so within the stats 'community' medicine is seen as the bold kid. Since my little experience in applying stats to real world is all based on medical applications, I have little view on other industries.

Indeed, the hard line of 0.05 is crazy, especially when you consider that if you originally get a p-value of say 0.057, if you just observed more data there's a good chance that p-value would then fall below 0.05. And suddenly your result becomes significant. Not that that's recommended of course! It just shows the non-rigidness (?) of them.

Confidence intervals of effect size should be included more often (or always). If I said that my vaccine reduces infections by 30%, and we know this to be significant because I got a p-value less than 0.05, then you'd be delighted with that. But if I said that my vaccine reduces infections by 30%, with a 95% CI of (2%, 58%), you'd be a lot more suspect. Even though they're from the same test.

 

[lynchie]

AZ have re-announced their efficacy stats from the USA data. It's now being reported as:

The primary endpoint, vaccine efficacy at preventing symptomatic COVID-19 was 76% (confidence interval (CI): 68% to 82%) occurring 15 days or more after receiving two doses given four weeks apart. In addition, results were comparable across age groups, with vaccine efficacy of 85% (CI: 58% to 95%) in adults 65 years and older. A key secondary endpoint, preventing severe or critical disease and hospitalisation, demonstrated 100% efficacy. There were eight cases of severe COVID-19 observed in the primary analysis with all of those cases in the placebo group.

The vaccine was well tolerated, and no safety concerns related to the vaccine were identified.

https://www.astrazeneca.com/media-c...ry-analysis-confirms-safety-and-efficacy.html

Which makes you wonder why they bothered to do the 79% version a couple of days ago - that just made them look shady.

I'd like to think they were busy doing more detailed analysis on the numbers - like the Novavax decision to announce an overall figure for the UK trial, but then split out a "original covid" and a "UK mutation" stat as well. Maybe the full data/report from AZ will give us a clue.

There's 14 cases still to be adjudicated, so I wonder if they went with the interim numbers to avoid having those hanging over them? Clearly didn't go down well though. Whole thing seems utterly ridiculous really.

 

[Pogue Mahone]

I'm not too sure on other fields specifically, but I do know that medicine is always used as the example for 'p-hacking', so within the stats 'community' medicine is seen as the bold kid. Since my little experience in applying stats to real world is all based on medical applications, I have little view on other industries.

Indeed, the hard line of 0.05 is crazy, especially when you consider that if you originally get a p-value of say 0.057, if you just observed more data there's a good chance that p-value would then fall below 0.05. And suddenly your result becomes significant. Not that that's recommended of course! It just shows the non-rigidness (?) of them.

Confidence intervals of effect size should be included more often (or always). If I said that my vaccine reduces infections by 30%, and we know this to be significant because I got a p-value less than 0.05, then you'd be delighted with that. But if I said that my vaccine reduces infections by 30%, with a 95% CI of (2%, 58%), you'd be a lot more suspect. Even though they're from the same test.

To be fair, we would tend to look at 95% CI as well. We’re not savages! Although it’s rare to dig much deeper than that in the stats section.

 

[Pogue Mahone]

AZ have re-announced their efficacy stats from the USA data. It's now being reported as:

The primary endpoint, vaccine efficacy at preventing symptomatic COVID-19 was 76% (confidence interval (CI): 68% to 82%) occurring 15 days or more after receiving two doses given four weeks apart. In addition, results were comparable across age groups, with vaccine efficacy of 85% (CI: 58% to 95%) in adults 65 years and older. A key secondary endpoint, preventing severe or critical disease and hospitalisation, demonstrated 100% efficacy. There were eight cases of severe COVID-19 observed in the primary analysis with all of those cases in the placebo group.

The vaccine was well tolerated, and no safety concerns related to the vaccine were identified.

https://www.astrazeneca.com/media-c...ry-analysis-confirms-safety-and-efficacy.html

Which makes you wonder why they bothered to do the 79% version a couple of days ago - that just made them look shady.

I'd like to think they were busy doing more detailed analysis on the numbers - like the Novavax decision to announce an overall figure for the UK trial, but then split out a "original covid" and a "UK mutation" stat as well. Maybe the full data/report from AZ will give us a clue.

All of this could have been avoided with better communication between AZ and the DSMB. Such a trivial difference to get called out on in public.

 

[Pogue Mahone]

The data coming out of Israel is fantastic. They’re even seeing a similar picture in a town that has 20% SA variant (see replies to tweet) If this keeps up they could essentially wipe the virus out in Israel (assuming no animal reservoir)

 

[BD]

To be fair, we would tend to look at 95% CI as well. We’re not savages! Although it’s rare to dig much deeper than that in the stats section.

Of course, but what about doing away with p-values altogether and having CIs as the main result?

 

[Pogue Mahone]

Of course, but what about doing away with p-values altogether and having CIs as the main result?

Well you don’t need a p value with a CI, right? I always assumed that if the 95% CI excluded no effect then that told you the p value was < 0.05. Using vaccines as the example. If the 95% CI straddled zero then you could be certain that any efficacy claim would not be statistically significant.

 

[christinaa]

My Mrs has, 3 days of feeling very ropey

My wife has had her 1st of 2 shots a couple of weeks back. No side effects.
Her 2nd shot is due on April 7th, 4 weeks after the 1st so I'll know more then.
Her Da, who's 86, was a bit tired after his 2nd Moderna shot - but he's 86.....other than that nothing.

I had the first dose 3 weeks ago. My arm was very sore the second day but got better over the next few days. Other than that, I felt a little achy and slight chills around bedtime the second day. Nothing substantial though. The second dose is typically where the immune response is more noticeable. Most of my family had Moderna and only had headache, aches, chills, or other symptoms for about 12-24 hours (to be clear, each person had one or two of the symptoms, not all).

Thanks for your answers.

 

[Snowjoe]

Woke up this morning feeling aches, sore throat and a bit lethargic but other than that no drama, 8/10 would get stuck again

 

[lynchie]

The data coming out of Israel is fantastic. They’re even seeing a similar picture in a town that has 20% SA variant (see replies to tweet) If this keeps up they could essentially wipe the virus out in Israel (assuming no animal reservoir)

Cases also dropping rapidly in (unvaxed) children as well, as they're not picking it up from adult contacts as often now. Israel's data is basically knocking down the worst-case scenarios for fun.

 

[BD]

Well you don’t need a p value with a CI, right? I always assumed that if the 95% CI excluded no effect then that told you the p value was < 0.05. Using vaccines as the example. If the 95% CI straddled zero then you could be certain that any efficacy claim would not be statistically significant.

Yeah exactly, the 'result' from the p-value is included in the CI results, but with added context.

 

[BarstoolProphet]

Denmark extending the pause in vaccination with AZ to middle of April.

 

[Stanley Road]

Denmark extending the pause in vaccination with AZ to middle of April.

Can they send them to me?

 

[Rajma]

The data coming out of Israel is fantastic. They’re even seeing a similar picture in a town that has 20% SA variant (see replies to tweet) If this keeps up they could essentially wipe the virus out in Israel (assuming no animal reservoir)

Fantastic news, my mum has already been vaccinated with Pfizer (HCW), today’s is my grandparents turn, will be a massive relief that.

 

[Penna]

Woke up this morning feeling aches, sore throat and a bit lethargic but other than that no drama, 8/10 would get stuck again

Woke up this mornin',
I got them vaccine blues ...

 

[Dumbstar]

Woke up this mornin',
I got them vaccine blues ...

 

[Balljy]

I had mine yesterday and was a bit nervous considering I got it last May have been suffering with long covid (I'm now pretty much on first name terms with half the hospital). I was slightly worried what sort of symptoms I'd have but just felt a bit heavy eyed yesterday evening and had a very slight temperature but feel absolutely fine today. Relieved.

The nurses who dealt with my injection couldn't see why I was invited at my age, but I'm assuming I've been put in a category due to the last few months.

 

[zing]

I had mine yesterday and was a bit nervous considering I got it last May have been suffering with long covid (I'm now pretty much on first name terms with half the hospital). I was slightly worried what sort of symptoms I'd have but just felt a bit heavy eyed yesterday evening and had a very slight temperature but feel absolutely fine today. Relieved.

The nurses who dealt with my injection couldn't see why I was invited at my age, but I'm assuming I've been put in a category due to the last few months.

Curious to know if your long covid symptoms improve - have read a few things online indicating it happens for some people.

 

[Balljy]

Curious to know if your long covid symptoms improve - have read a few things online indicating it happens for some people.

That hope had crossed my mind, it is just a hope but I'll be keeping a close eye on it. I'm getting there gradually anyway, and my symptoms are now lung related with my energy pretty much back to how it was before.

 

[11101]

Another tick for the vaccines (Pfizer specifically).

In my province overall cases are just about hitting their peak, but cases in old people's homes have dropped from 568 two months ago to just 36 now, with only 2 symptomatic cases. There were no lockdowns or anything like that until a couple of weeks ago, it's entirely down to the vaccines.

 

[Grinner]

Apparently Krispy Kreme are offering one free donut every day for the year for anyone who shows proof of vaccination. My choice to return to California for the shot seems smart now.

Survive Covid, get diabetes.

 

[RedTiger]

I recieved the Oxford vaccine today.

 

[Bratt]

My mother got her appointment for the vaccine today, so happy. My father who’s the same age haven’t got his appointment yet though.

 

[christinaa]

Woke up this mornin',
I got them vaccine blues ...

 

[langster]

Woke up this morning feeling aches, sore throat and a bit lethargic but other than that no drama, 8/10 would get stuck again

Ahaaaaa!

So you did go for the old smash and sniff option after all

Glad you're ok mate

 

[Pogue Mahone]

 

[McGrathsipan]

Another tick for the vaccines (Pfizer specifically).

In my province overall cases are just about hitting their peak, but cases in old people's homes have dropped from 568 two months ago to just 36 now, with only 2 symptomatic cases. There were no lockdowns or anything like that until a couple of weeks ago, it's entirely down to the vaccines.

That's encouraging

 

[McGrathsipan]

My wife got 1st Astra jab 2.5 weeks ago and she is telling me her arm is actually worse now than when she got the jab. Is this normal?

@Pogue Mahone

 

[11101]

That's quite out of date now. The EU said yesterday they have exported 77m to date, it looks like they are(were) sending about 20m a week.

 

[2cents]

That's quite out of date now. The EU said yesterday they have exported 77m to date, it looks like they are(were) sending about 20m a week.

Do we know which countries the EU is exporting to?

 

[11101]

Do we know which countries the EU is exporting to?

33 different countries, although i don't know who has got what. The only news reports were of 21m to the UK.

 

[Pogue Mahone]

That's quite out of date now. The EU said yesterday they have exported 77m to date, it looks like they are(were) sending about 20m a week.

It’s the zeros that grind my gears. I know it’s not governments exporting, it’s the companies, blah blah blah but the US and the UK have clearly stepped out of line from the rest of the world here and it’s not a good look.

 

[Traub]

It’s the zeros that grind my gears. I know it’s not governments exporting, it’s the companies, blah blah blah but the US and the UK have clearly stepped out of line from the rest of the world here and it’s not a good look.

Yip, I thought Biden was supposed to be different - well according to many on here. Now we are seeing his true colours - we are going through the biggest world crisis for a few generations, and it seems Biden would rather have healthy 20 year old Americans vaccinated than frontline HCW in the most impoverished countries.

https://www.bloomberg.com/news/arti...s-america-first-vaccine-plan-to-corner-market

 

[Pogue Mahone]

Yip, I thought Biden was supposed to be different - well according to many on here. Now we are seeing his true colours - we are going through the biggest world crisis for a few generations, and it seems Biden would rather have healthy 20 year old Americans vaccinated than frontline HCW in the most impoverished countries.

https://www.bloomberg.com/news/arti...s-america-first-vaccine-plan-to-corner-market

Paywalled for me but yeah, it’s annoying. Obviously Trump started this but Biden seems determined to bask in the glow of having way more vaccines than they need.

 

[Classical Mechanic]

It’s the zeros that grind my gears. I know it’s not governments exporting, it’s the companies, blah blah blah but the US and the UK have clearly stepped out of line from the rest of the world here and it’s not a good look.

Thought you said that Airfinity were a corrupted source doing the UK governments bidding?

It's absolutely right to point out that it's not governments exporting. It's private companies. A UK private company is integral to the manufactuer of a highly significant number of vaccines, to the extent that they cannot be made without them and the EU's vaccination program would collapse without them. The idea we're not a part of the global supply of vaccines is false. They are all made cross continents by private companies and then exported by private companies.

The UK doesn't have an export ban. It's the AZ contract and the fact that we only have one manufacturer as to why we're not sending many abroad. Oxford University and AZ did create a non-profit vaccine that's been licensed to the Indian Serum Insitute which has supplied millions of doses around the world, however.

 

[11101]

It’s the zeros that grind my gears. I know it’s not governments exporting, it’s the companies, blah blah blah but the US and the UK have clearly stepped out of line from the rest of the world here and it’s not a good look.

I expected it from the US, and the EU isn't complaining about them as they are exporting vast quantities of materials if not the finished vaccines. It's the UK i am disappointed in, and most of my older family members seem proud to be sticking two fingers up to the world.

 

[Pogue Mahone]

Thought you said that Airfinity were a corrupted source doing the UK governments bidding?

It's absolutely right to point out that it's not governments exporting. It's private companies. A UK private company is integral to the manufactuer of a highly significant number of vaccines, to the extent that they cannot be made without them and the EU's vaccination program would collapse without them. The idea we're not a part of the global supply of vaccines is false. They are all made cross continents by private companies and then exported by private companies.

The UK doesn't have an export ban. It's the AZ contract and the fact that we only have one manufacturer as to why we're not sending many abroad. Oxford University and AZ did create a non-profit vaccine that's been licensed to the Indian Serum Insitute which has supplied millions of doses around the world, however.

This argument is so fecking boring and has been done to death. I just saw that graph and got irritated all over again. There’s some mitigation for the UK in that vaccines wouldn’t be the only export to tank, post-Brexit but the fact that both the UK and US insisted on exclusivity of supply in their contract negotiations with manufacturers, during a global pandemic, is not a good look. And I’ll leave it at that.