I think all of this government data needs to be viewed with the knowledge that no government would admit to complete vaccine failure. Apparent vaccine effectiveness is therefore a given and has little probative value.
Very very interesting and would most definitely be interested in seeing (and sharing) your weekly combination of useful charts. This chart, whilst requiring some explanation initially is indeed quite useful. Based on how vaccine efficacy has been trending downwards I would suggest that the key test would be between weeks 44 and 50 (generally around week 45). If they haven't crossed over into negative territory in that timeframe then it seems likely that they will be approaching unity or zero but never quite get there as you noted.
Given how small the previously uninfected unvaccinated population has become, and given how much it is decreasing with time, what will you do in terms of analysis when that segment of the population eventually reaches zero? How will vaccine efficacy continue to be properly measured if the only two groups left are the vaccinated and previously infected unvaccinated?
Well for the 60+ year olds we are pretty much at their limits of detection as it is. For 80+ there are somewhere between 15-20k non-immune people left by my calculations. We are likely to have to infer numbers by attrition and predictions.
It is also possible we WILL see a sudden drop in VE implying a turn over to negative efficiency (I suspect right now, without viral changes, what we will see will be the limit approaching 1 (equal efficacy between vaccination and nothing)). The sudden VE cliff is really what we want to look out for. I suspect any such cliff will be mostly in those with the lowest antibody numbers first - so a clear and significant downward move in a short period of time, would indicate a change in the nature of the virus itself more than a steady decline in the population's protections.
It is a fair question what to do when we effectively believe the entire population has had COVID-19. I'm not 100% sure honestly. I need to think a bit about that. I've been working this weekend on projecting forward and think I have something almost ready to go for a first attempt. Hopefully with some predictions being made, we can start testing more fully. The most interesting outcome will be a much greater than expected shift in vaccinated case rates and a matching prediction. Right now though, the Vaccinated Rate is moving relatively slowly.
A sudden plunge in VE could imply a turn over to negative efficacy as a change in the nature of the virus, however a decline towards 1 and then towards negativity might be possible also should it not? I'm thinking such an outcome might be because of the nature of the immune response (as generated by the vaccines) rather than the change in the virus (via mutation).
the vaccines (and natural infection) produce various different antibodies to different regions of the spike protein. Most of these are, initially, neutralizing antibodies, mainly to the receptor-binding domain (RBD) and to the non-terminal domain (NTD). However some are facilitative or non-neutralizing antibodies targeted at the NTD. The balance of the antibodies shortly after vaccination favours the neutralizing antibodies, however this balance can shift (i) as antibody levels decline; and (ii) as viral mutants arise such as the delta strain whereby the NTD and RBD are changed in such as a way as to have decreased affinity with the neutralizing antibodies, but increased affinity with the facilitative antibodies.
So might it not be possible in the case of someone who was vaccinated but did not receive boosters, to reach a point where the balance of antibody types gradually shifts in favour of non-neutralizing antibodies just due to natural decline of the overall antibody levels, resulting in perhaps a weak negative efficacy effect whereby the vaccinated person is now more likely to become infected than a naive unvaccinated person simply because they now have non-neutralizing antibodies in sufficient proportion to the neutralizing antibodies in their systems, even at overall low antibody levels generally?
Obviously this is complicated and I'm Just a Guy on the Internet telling people they are wrong.
I suspect, that if we see ADE which is what you are proposing - and I think is a very real possibility moving forward - there will be a sudden shift on the charts. More importantly, I really think we would be looking at a lagging shift in admissions and mortality as well. If this is true ADE (as opposed to a Marek's type situation where the virus never evades the antibody response, but can select for more virulent strains because the vaccinated population has protection against those outcomes, but not infection) then the enhanced viral replication and cellular infiltration indicated by ADE would generate significantly worse outcomes in the elderly (those most at risk to start with and currently seeing the least benefit vs the non-immune at this time).
So, while I agree that the gradual drop in antibody levels could (as indicated by the letter you reference and the paper out of Japan in August ) lead to an ADE response over time - I think that we would see this as a drop not just a gradual slide into less than 1 (right now it cannot go below zero, so if/when we start sliding below 1 I'll need to adjust that chart)
I think all of this government data needs to be viewed with the knowledge that no government would admit to complete vaccine failure. Apparent vaccine effectiveness is therefore a given and has little probative value.
Very very interesting and would most definitely be interested in seeing (and sharing) your weekly combination of useful charts. This chart, whilst requiring some explanation initially is indeed quite useful. Based on how vaccine efficacy has been trending downwards I would suggest that the key test would be between weeks 44 and 50 (generally around week 45). If they haven't crossed over into negative territory in that timeframe then it seems likely that they will be approaching unity or zero but never quite get there as you noted.
Given how small the previously uninfected unvaccinated population has become, and given how much it is decreasing with time, what will you do in terms of analysis when that segment of the population eventually reaches zero? How will vaccine efficacy continue to be properly measured if the only two groups left are the vaccinated and previously infected unvaccinated?
Well for the 60+ year olds we are pretty much at their limits of detection as it is. For 80+ there are somewhere between 15-20k non-immune people left by my calculations. We are likely to have to infer numbers by attrition and predictions.
It is also possible we WILL see a sudden drop in VE implying a turn over to negative efficiency (I suspect right now, without viral changes, what we will see will be the limit approaching 1 (equal efficacy between vaccination and nothing)). The sudden VE cliff is really what we want to look out for. I suspect any such cliff will be mostly in those with the lowest antibody numbers first - so a clear and significant downward move in a short period of time, would indicate a change in the nature of the virus itself more than a steady decline in the population's protections.
It is a fair question what to do when we effectively believe the entire population has had COVID-19. I'm not 100% sure honestly. I need to think a bit about that. I've been working this weekend on projecting forward and think I have something almost ready to go for a first attempt. Hopefully with some predictions being made, we can start testing more fully. The most interesting outcome will be a much greater than expected shift in vaccinated case rates and a matching prediction. Right now though, the Vaccinated Rate is moving relatively slowly.
A sudden plunge in VE could imply a turn over to negative efficacy as a change in the nature of the virus, however a decline towards 1 and then towards negativity might be possible also should it not? I'm thinking such an outcome might be because of the nature of the immune response (as generated by the vaccines) rather than the change in the virus (via mutation).
I suggest this because as this paper notes: https://www.journalofinfection.com/article/S0163-4453(21)00392-3/fulltext
the vaccines (and natural infection) produce various different antibodies to different regions of the spike protein. Most of these are, initially, neutralizing antibodies, mainly to the receptor-binding domain (RBD) and to the non-terminal domain (NTD). However some are facilitative or non-neutralizing antibodies targeted at the NTD. The balance of the antibodies shortly after vaccination favours the neutralizing antibodies, however this balance can shift (i) as antibody levels decline; and (ii) as viral mutants arise such as the delta strain whereby the NTD and RBD are changed in such as a way as to have decreased affinity with the neutralizing antibodies, but increased affinity with the facilitative antibodies.
So might it not be possible in the case of someone who was vaccinated but did not receive boosters, to reach a point where the balance of antibody types gradually shifts in favour of non-neutralizing antibodies just due to natural decline of the overall antibody levels, resulting in perhaps a weak negative efficacy effect whereby the vaccinated person is now more likely to become infected than a naive unvaccinated person simply because they now have non-neutralizing antibodies in sufficient proportion to the neutralizing antibodies in their systems, even at overall low antibody levels generally?
Obviously this is complicated and I'm Just a Guy on the Internet telling people they are wrong.
I suspect, that if we see ADE which is what you are proposing - and I think is a very real possibility moving forward - there will be a sudden shift on the charts. More importantly, I really think we would be looking at a lagging shift in admissions and mortality as well. If this is true ADE (as opposed to a Marek's type situation where the virus never evades the antibody response, but can select for more virulent strains because the vaccinated population has protection against those outcomes, but not infection) then the enhanced viral replication and cellular infiltration indicated by ADE would generate significantly worse outcomes in the elderly (those most at risk to start with and currently seeing the least benefit vs the non-immune at this time).
So, while I agree that the gradual drop in antibody levels could (as indicated by the letter you reference and the paper out of Japan in August ) lead to an ADE response over time - I think that we would see this as a drop not just a gradual slide into less than 1 (right now it cannot go below zero, so if/when we start sliding below 1 I'll need to adjust that chart)