COVID19: Social Distancing might last to 2022

[AggieMike88]

Unknown if it has to be as drastic as we are doing know, but following the cautionary basics might make sense.

 

[PeterNSteinmetz]

I am not aware of any current academic studies which show that the current coercive government imposed measures have been responsible even for the existing decrease in the exponential growth rate.

What Kissler et al showed and was just published in Science is that a single period of social distancing is unlikely to be sufficient to avoid over-whelming critical care beds in the US and that if SARS Cov-2 is moderately seasonal, this may make things worse in the fall.

 

[PeterNSteinmetz]

I have an essential job which provides the food you're putting on your table. If you would like those food providers to stay home, good luck eating. As for stress relief, you're allowed to relieve stress all you want so long as it does not put others at risk. We tried that plan and the people weren't able to manage the not putting others at risk part.

Really not a lot of good evidence that the risk was particularly high or that the coercive interventions have reduced that risk in any substantial way.

I know the party line is that it was such a terrible risk that something draconian had to be done - but the numbers just aren’t bearing that out.

 

[PeterNSteinmetz]

So what do you think we should do about it, and on what timeline?

In brief, continue to make the best recommendations based on available evidence and immediately remove all coercive measures.

Focus available resources on better studies and immediately remove barriers to provision of services and products.

Do not provide any more financial handouts, aka “stimulus”, and reduce taxes to stimulate the economy.

In other words, let the free market and people’s own judgements decide.

 

[PeterNSteinmetz]

What impact do you think your proposal would have on the number of COVID-19 hospitalizations and deaths?

I think that total mortality and morbidity, including all causes, would likely be decreased by this when averaged over a year or two.

In terms of Covid-19, I would also expect a decrease in hospitalizations and deaths in the course of a month or two as obstacles to better testing, treatments, etc took effect.

However, I think any detailed predictions are quite tenuous at this time given the poor data we have on this particular illness.

What I am more confident of is that not using coercive measures, when one is not certain the person or persons they are being used against are an imminent threat, will work out best for everyone in the long run.

 

[PeterNSteinmetz]

I disagree.

OK. I'm always happy to look at papers and serious studies that indicate otherwise. My view on whether the numbers go one way or another tends to be very empirically driven.

 

[PeterNSteinmetz]

It sounds like the bottom line of what you're saying is that there's not good enough evidence to show which approach is more effective, yours, or the one being pursued currently. I'm not sure why "coercive" matters, other than to add emotional impact to the argument, since all laws are by nature coercive.

Please note that my suggested policy involves removing the coercions. So that would be no laws requiring people to do things regarding Covid-19 they do not otherwise wish to do.

So big difference in coercion level between my proposal and what is presently being done.

Practically, in the long run, I think there is very good evidence that lack of coercion in the absence of imminent threat results in better outcomes for everyone, even if that might be difficult to predict in a particular case. So we should default to lack of coercion absent clear evidence of an imminent threat.

But at that level, we are drifting into politics, so we best discuss that in another forum if you want.

 

[PeterNSteinmetz]

You seem to be assuming that there isn't an imminent threat.

No I am not assuming that, but that clearly is the primary question with an epidemic like this.

I would submit that the present data available do not strongly support the idea that a person of unknown SARS Cov-2 status presents an imminent threat of causing serious injury or death to the average person in their casual, possibly unchosen, interactions on the street or other public places.

I would also say that the imminent threat should be very clear to justify the use of coercion.

The present data fail to meet that standard.

Chosen interactions in non-public places don’t matter for this sort of evaluation, so bars and restaurants don’t matter. There is also the fact that the purported victim can take much more peaceful measures to protect themselves on the street, such as traveling only in cars or wearing a mask, rather than coercing others to not be in a public place.

 

[PeterNSteinmetz]

I am surprised the prevalence estimate is only 2.8%. Will have to read that study.

Didn’t the Gangelt Germany study find 15%? They were involved very early on. I bet there is a lot of heterogeneity in this.

 

[PeterNSteinmetz]

I do understand that a lot of the people posting here fall into a higher risk group for Covid-19 and so I definitely advocate for them to make there own judgements and take appropriate precautions.

You can reduce your risk of being infected to almost 0 by staying at home and not letting anyone else in.

I have some friends in the very high risk category who choose to do just that right now.

 

[PeterNSteinmetz]

Unfortunately it’s a really terrible study. The test they use has a 1.5% CI. And they detected a 1.5% positive rate...

So it means precisely nothing. Unfortunately.

You can use this results to justify both a CFR of 0% and one of 100%.

I read the whole paper and will have to disagree. Firstly I think this was a rather well done study given the evidence available regarding the sensitivity and specificity of the test at the time.

It is important to note that estimates based on sampling like this will always have a a statistical distribution and a certain range in the confidence intervals. Nonetheless, the mean estimate of prevalence, as adjusted, is likely the best estimator from a mean squared error point of view.

What a 95% confidence interval means is not that “precisely nothing” is known, but rather that there is a 95% chance that the estimated value lies within the specified interval. In this case, the estimated intervals based on the available data are that the prevalence lies between 2.5 and 5.7% using a number of different assumptions. Certainly if new data becomes available indicating the sensitivity and specificity of the test are actually different, then that would have to change, as the authors note. But that is not the case now and there is no existing data to suggest the 95% CI should include 0.

So I think the authors are on rather good ground when they assert that the number of infected in this population was likely 50-85 times higher than the reported cases. We’ll have to see how it does in peer-review of course, but this group is well known and competent.

 

[PeterNSteinmetz]

The issue was when they calibrated the test, they only performed 30 negative tests. They need to go back and re-calibrate the test.

As noted in the paper, they also used a much larger calibration sample which was obtained by the manufacturer.

The idea that somehow the true prevalence could be as low as 0 is based on speculation. Note the “if” in the statement by the authors quoted above,

The numbers quoted by the authors for the mean prevalence are the best estimates based on the current data.

 

[PeterNSteinmetz]

I ran across this “peer review”. I’m not this sort of numbers guy, so I don’t know the validity of this critique, and I don’t know anything about the author. But he echoes some of your comments.

https://medium.com/@balajis/peer-re...in-santa-clara-county-california-1f6382258c25

I read that “review”. His primary concern seems to be that given possible values of the false positive rate for the test, it is conceivable that with a higher FPR, a fair number of the observed positives could be false positives.

While that is true, it does not imply that such a result is very probable (he could compute that using his models).

The best estimate based on this data of adjusted prevalence is as given by the authors of the paper. There is some finite probability that the actual prevalence is different than the mean as stated by the given confidence intervals.

He does raise an interesting point regarding the implication that Covid-19 would have to be spreading much more quickly than viruses in prior pandemics for the stated results to be true. I think that is a good observation. And is in accordance with an extrapolation by a Reason editor that a 50-85X ratio would imply that nearly all of New York City is presently infected.

I agree completely with his argument that more serological surveys are badly needed.

Another point of reference is that the 0.15 - 0.2% infection fatality rate is not so far from the 0.3% estimate from the Gangelt study, at least in absolute terms (clearly a 50% difference relatively speaking).

The Stanford study adds to a growing body of evidence that the true infection fatality rate (vs case fatality rate) is on the order of 0.1-0.3%, a far cry from the feared 8% based on initial reports, and perhaps 2-3X more deadly than the seasonal flu.