Easing of Restrictions from 10th May

[Sunny]

So if you understand that 95% efficacy means that 95% of a cohort will be protected, surely you understand that this means 5%, or 1 in 20 will not be protected?

If it wasn't so, and that 5% had some level of protection, then the efficacy would be higher.

It doesn't mean that. A 95% efficacy means that the chance of a vaccinated person getting Covid is 95% less than someone who is not vaccinated. This does not mean that the other 5% of the population is 100% guaranteed to get Covid or a severe case of covid. So to simply say that 5% of 500k people or 25,000 vaccinated people are at risk of Covid which Philip Nolan claimed and seemed to suggest is in his model is simply incorrect. There are other public health measures and other reasons that will reduce the risk further so only a % of the 5% will actually catch Covid and even then might not be severe.

Going by what Nolan said, they have modelled 25,000 (5% of population) vaccinated elderly people catching covid because of the 95% efficacy and calculated deaths and hospitalisations on this number.

 

[Leo]

It doesn't mean that. A 95% efficacy means that the chance of a vaccinated person getting Covid is 95% less than someone who is not vaccinated.

You can't apply an efficacy rating to a single person, that's not what it is intended to measure or convey. It is a community measure assuming ideal conditions. A vaccine will provide a wide range of protection. Some will be fully, or 100% protected, many somewhat less protected, some not at all.

here are other public health measures and other reasons that will reduce the risk further so only a % of the 5% will actually catch Covid and even then might not be severe.

Yes, once you start to consider the effects of public health measures on outcomes, then you need to speak in terms of vaccine effectiveness, and not efficacy.

 

[joe sod]

It's not really nphet that is the issue but the power that they have been given by the government and media, why ? Why have we given this body so much power in comparison to our European peers, they maybe correct about the delta wave sweeping Europe but nobody else is panicking like us.
In other countries economic and social consequences have a much higher weighting in decision making than here. Yet we have the highest per capita debt load in the eu with the pandemic set to cost 50 billion but nobody sounding alarms about this. The 50 billion spent is spending that will not go on "green infrastructure" building houses and could put the whole welfare system in jeopardy if interest rates begin rising faster

 

[Sunny]

You can't apply an efficacy rating to a single person, that's not what it is intended to measure or convey. It is a community measure assuming ideal conditions. A vaccine will provide a wide range of protection. Some will be fully, or 100% protected, many somewhat less protected, some not at all.


Yes, once you start to consider the effects of public health measures on outcomes, then you need to speak in terms of vaccine effectiveness, and not efficacy.

95% efficacy does not mean that 95% of the population are protected from getting covid and that 5% have zero protection. It means the chance of getting Covid or severe Covid is reduced by 95% compared to an unvaccinated population. 100% of the unvaccinated population will not get Covid. It will only be small % of that population

Assume an attack rate of about 10% (No idea what it is for Delta). If we have a population of 1,000,000 people that are vaccinated and a population of 1,000,000 that are not vaccinated. We can assume we will see 100,000 cases in the non vaccinated population. In the vaccinated population, what Philip Nolan is saying that 5% of the vaccinated population could get covid so we would expect to see 50,000 cases and has implied that this is in the model. That is simply not true. A 95% efficacy means that we would expect to see 5000 cases based on the attack rate and the efficacy.

Apologies if my maths are wrong. I am hungover!....

 

[Purple]

I am hungover!....

Well it is Friday.
That's my understanding of it as well but there's a high likelihood I'm wrong.

 

[Paul O Mahoney]

Efficacy = what is the desired effect usually in a controlled environment

Effectiveness = what happens in the real world

So, a drug might have 95% efficacy but be 90% effective in the real world.

I might be wrong too.

 

[Sunny]

Efficacy = what is the desired effect usually in a controlled environment

Effectiveness = what happens in the real world

So, a drug might have 95% efficacy but be 90% effective in the real world.

I might be wrong too.

Thats my understanding as well. I think Philip Nolan is just guilty of phrasing something carelessly as I doubt they have modelled the scenario he described. It is just worrying that the guy in charge of modelling made a statement that because of 95% vaccine efficacy, 5% of vaccinated elderly people are at risk of catching covid. He then backed it up with actual numbers of 25,000 out of 500,000 vaccinated people could get covid. This is simply wrong as he is assuming 100% attack rate.

It would be good to see the complete model released rather than tweets. At the very least they should be able to provide the basic assumptions made under each scenario.

 

[Paul O Mahoney]

Thats my understanding as well. I think Philip Nolan is just guilty of phrasing something carelessly as I doubt they have modelled the scenario he described. It is just worrying that the guy in charge of modelling made a statement that because of 95% vaccine efficacy, 5% of vaccinated elderly people are at risk of catching covid. He then backed it up with actual numbers of 25,000 out of 500,000 vaccinated people could get covid. This is simply wrong as he is assuming 100% attack rate.

It would be good to see the complete model released rather than tweets. At the very least they should be able to provide the basic assumptions made under each scenario.

If you click on the links on the Twitter above you can see all the reports to date, and a plethora of references to back up the assumptions made.

Its heavy reading btw, and a lot of cross referencing and dead boring.

They use "homogeneous mixing" as the basis of everything eventhough he said it's not real world but it has other advantages such as simplicity as fewer variables are needed to forecast. Its internationally used.

 

[Sunny]

If you click on the links on the Twitter above you can see all the reports to date, and a plethora of references to back up the assumptions made.

Its heavy reading btw, and a lot of cross referencing and dead boring.

So they didn't provide a nice one page summary???

Thanks. I might have a look over the weekend but I don't think I have the attention span to be honest! I will continue to make unsupported accusations of flawed modelling as it is the lazier option!

By the way, I should state that I am in awe of what these guys produce from a modelling point of view. But I strongly believe that models work best when they are completely open to scrutiny and are not black boxes that only the people who designed them know how they work. We had a whole financial crisis built on black box models. It's good if they are publishing everything. I hope the academic and public health world step up and voice their opinions.

 

[Paul O Mahoney]

So they didn't provide a nice one page summary???

Thanks. I might have a look over the weekend but I don't think I have the attention span to be honest! I will continue to make unsupported accusations of flawed modelling as it is the lazier option!

By the way, I should state that I am in awe of what these guys produce from a modelling point of view. But I strongly believe that models work best when they are completely open to scrutiny and are not black boxes that only the people who designed them know how they work. We had a whole financial crisis built on black box models. It's good if they are publishing everything. I hope the academic and public health world step up and voice their opinions.

It's all new given the speed of transmission since it began they have achieved quite a bit, if I knew reports were in the public domain I would have read them last and this year, but you would expect them to run more scenarios and show different results by changing key variables, a sensitivity analysis I think they call it.

 

[Sunny]

It's quotes like this that really bother me

At the NPHET briefing on Thursday, Prime Time asked Professor Philip Nolan, Chair of the Irish Epidemiological Modelling Advisory Group, if he'd run the models without the step change in social mixing included, and what the outcome would be.

"I did, and the answer is: it’s modest," he said. "Frankly, it is anywhere between Alpha-only and the optimistic scenario."

So we are only seeing a 'modest' change in the forecasts by not opening indoor hospitality on the 19th. Given that they are now seeing a risk (but won't tell us the probability) of anything from Alpha only outbreak with hospitality closed to a more severe outbreak than Alpha if opened, why aren't we locking down again? The answer has to be the probability of an outbreak like January to be so unlikely that NPHET are prepared to take the risk.

So the question still remains. How much extra risk would we have taken on by opening up indoor hospitality on the 19th? They have obviously modelled it.

 

[Sophrosyne]

Efficacy = what is the desired effect usually in a controlled environment

Effectiveness = what happens in the real world

So, a drug might have 95% efficacy but be 90% effective in the real world.

I might be wrong too.

No you are right.

Efficacy means that under the same conditions as the trials, vaccine reduces infection by the same percentage as that in the trials.

The Pfizer trial found an infection rate of 0.74% in the placebo group and 0.04% in the vaccinated group.

Therefore, the vaccine reduced the infection rate by 0.7%.

Scaled up this means an efficacy rate of 95.05% (0.70 / 0.74)

However, trials do not show how the percentage might vary on populations with different exposures and different transmission and attack rates.

The modelling is attempting to achieve this with best and worst case scenarios in the Irish context.

Other countries model on their own prevalent conditions.

 

[joe sod]

The Pfizer trial found an infection rate of 0.74% in the placebo group and 0.4% in the vaccinated group.

Therefore, the vaccine reduced the infection rate by 0.7%.

That doesn't make sense, 0.74 % to 0.4% is actually a 50% reduction not 0.7%, !! Are you sure you have your figures correct. These don't sound like Pfizer figures anyway ?

 

[Purple]

The Pfizer trial found an infection rate of 0.74% in the placebo group and 0.4% in the vaccinated group.

Therefore, the vaccine reduced the infection rate by 0.7%.

Scaled up this means an efficacy rate of 95.05% (0.7 / 0.74)

Should that be 0.04%?

 

[24601]

So if you understand that 95% efficacy means that 95% of a cohort will be protected, surely you understand that this means 5%, or 1 in 20 will not be protected?

If it wasn't so, and that 5% had some level of protection, then the efficacy would be higher.

That's not what it means and he has been widely ridiculed for presenting this impression. For what it's worth I think he was being a bit careless with his language rather than actually suggesting that 1/20 fully vaccinated people have no protection.

In any event, NPHET don't believe any of these pessimistic scenarios are at all probable or we'd all be back doing laps of our 2km radius.

 

[Sophrosyne]

Should that be 0.04%?

Yes it should. I have amended post.

 

[Sophrosyne]

These don't sound like Pfizer figures anyway ?

Phase 111 of the Pfizer trial enrolled 43,661 people half of whom received a placebo and the other half were vaccinated.

There were 170 confirmed cases of which 162 (0.74%) were in the placebo group and 8 (0.04) in the vaccinated group.

 

[Purple]

Yes it should. I have amended post.

Lucky guess on my part.

 

[Paul O Mahoney]

In any event, NPHET don't believe any of these pessimistic scenarios are at all probable or we'd all be back doing laps of our 2km radius.

I wouldn't be so sure of that, looking back to February they over egged the forecasts too and the reality, while still terrible were a good 30% lower.

Providing good analytics and forecasting is essential for good decision making in any business and if anything is flawed it leads to flawed decisions with a scale of irrelevant to serious , and this is a pandemic.

 

[Paul O Mahoney]

Phase 111 of the Pfizer trial enrolled 43,661 people half of whom received a placebo and the other half were vaccinated.

There were 170 confirmed cases of which 162 (0.74%) were in the placebo group and 8 (0.04) in the vaccinated group.

Yep and the initial trial size was 35,000 the additional numbers were for other minorities, but I also know that production was still started based on the original and in September 2020.

I can say that now as it's in the public domain.