Finally a voice of reason

There's an economic impact for sure, and there's a philosophical argument in there somewhere. But objectively the shelter in place rules work... and the truth still stands that if you locked everyone at home for 14 days this would disappear and we could be coming out the other side of this that much sooner

That would only be true if we could test everyone before they re-emerge out of their burrow. We know that a substantial percentage of people infected are asymptomatic and while they are less likely to transmit the disease, with prolonged contact as it happens within a househould, they can. The median incubation period is 5 days and the 97% range is 11.5 days. So with a 2 week quarantine, unless you lock everyone in a separate room, you may have an infection chain within a household unit and come out with a carrier at the tail end.

As mentioned above, we don't need zero new infections. We need a number of new cases and a R0 that is low enough that the methods of tracing and contact isolation become sufficient to eventually eliminate the bug from a population.

As for it 'becoming seasonal', as these epizoonotic viruses go, its not all that likely. This isn't the first one. They tend to come, run their course and disappear. There may be another round of this 5 years from now when there is another jump of a virus from bat or pangolin to humans, but that wouldn't be the same SARS-cov2 but rather SARS-covX etc.
 
I spray everything with denatured alcohol. But I've done that for years. An awesome time to be an introverted germophobe.
The only problem with denatured alcohol is that you usually don't know what you are buying, if it came form a DIY store, or a hardware store. They often denature the ethanol with methanol. The problem is, there is sometimes little ethanol, and the can is mostly ethanol. While that doesn't matter for most tasks, for this purpose, methanol doesn't do much as a disinfectant.

[1] https://www.cdc.gov/infectioncontrol/guidelines/disinfection/disinfection-methods/chemical.html , one sentence mentions it in the link
[2] https://medcraveonline.com/JBMOA/JBMOA-07-00247.pdf
 
But it appears the swift measures taken weeks ago at the county level had a positive effect

...But objectively the shelter in place rules work...

I would use some caution in ascribing causation here at least at the national level (where I have looked at the numbers). I can think off hand of 4 reasons why the exponential growth rate of the number of total cases might decrease (which it does appear to have done in a statistically significant way around March 25.

1. The measures overall taken across the US, such as shelter in place, lockdowns, and just the advice for social distancing, may have worked.
2. It is getting warmer and perhaps the virus is sensitive to temperature as many other seasonal flus are.
3. This could be some weird function of testing availability. Perhaps an exponential growth in cases exhausted extant testing supplies and now a more linear increase in the number of available test kits is dominating the overall effect.
4. It may be that number of asymptomatic cases really dominates the symptomatic ones, by some estimates a factor of 5, and as per the Oxford modeling study a large fraction of the population has already been exposed and we are simply seeing a slowdown in the rate of the number of cases that get noticed and tested.

Overall likely good news, but I think it will take some time before people are able to parse out what may have helped here in the US.
 
Yes, that is what is meant by exponential growth. The rate of growth appears to have changed in a statistically significant way since March 1st around March 25th.

What you see as inflection point is the time period when the large commercial labs and a number of approved hospital labs came online and the states started to report those results along with those obtained in their own labs.
 
I missed this earlier that the Apple News item is the item based on the Kinsa temperature readings. This is very weak evidence because of large potential confounds. It is entirely possible that if you tell people there is a terrible pandemic and that we having to go to lockdowns, that one of the things people do is take their temperature because they are afraid they have this terrible pandemic. And if a large number of them are not in fact infected, the average temperature measured will drop.
 
What you see as inflection point is the time period when the large commercial labs and a number of approved hospital labs came online and the states started to report those results along with those obtained in their own labs.

Could be, as in explanation #3 in my post. But it does seem odd. Do you know of any datasets describing the availability of tests as a function of time? That would be quite interesting.
 
Could be, as in explanation #3 in my post. But it does seem odd. Do you know of any datasets describing the availability of tests as a function of time? That would be quite interesting.

I am closely watching the number of tests and the sources of test results at my hospital. We are testing like crazy but have yet to find a single confirmed case. There is also a substantial backlog of cases which are stratified by priorities. The highest priority is on acutely ill inpatients and the lowest priority are healthy contacts and low symptomatic outpatients. I told one of the management folks today that once the positives start showing up, we should not mistake the fact that the results on the sickest patients show up first for an indication that our community is more severely affected. This nonlinearity in the reporting is systemic. First the state and CDC labs were capped out, then the commercials became available, as they don't restrict samples they got swamped and positive results got delayed among the flood of tests from the worried well who had the sniffles and talked their doc into sending out the test.
 
This morning an article or two mentions April 15th as projected virus peak? I realize some spitballing going on.

Either way, a ho-hum week and more ahead.

I’m of the mindset it’s widespread, the more testing, large uptick in positives. It seems symptoms vary widely too. Time to just eat ‘regular’ critters, common livestock.

It’s amazing the documentaries out in recent years, which prophesied the rise of such a virus from the Chinese ‘wet markets’? If so, maybe time to get ahead of it?
 
This morning an article or two mentions April 15th as projected virus peak? I realize some spitballing going on.

Either way, a ho-hum week and more ahead.

While I loathe any cartographer who represents raw case numbers (rather than per/100k) with 'red circles', the JHU resource website seems to be one of the better collections .

https://coronavirus.jhu.edu/map.html

If you look at the data for the different countries, you can look at log plots and 'daily increase' numbers. You can compare the pattern in the different countries. I am somewhat hopeful that the 'daily increase' numbers in a number of countries (Italy, Germany, Switzerland) seem to 'break'. If what happened in South Korea is any guide, there is probably going to be a bit of a 'rats tail' to the initial peak as some of the infection chains within families take a few cycles of isolation to burn themselves out.

The problem in the US is that the epidemic is focused on a few hot-spots and the strategy seems to be to apply the isolation measures to everyone until things are at least getting better. Given the particular political, logistical and attitude considerations in the US, that's probably the only way to get it done, but the economic damage is without doubt larger than what it could be otherwise. We are basically telling people in Maine to hunker in the basement because there is a tornado in Kansas.
 
The problem in the US is that the epidemic is focused on a few hot-spots and the strategy seems to be to apply the isolation measures to everyone until things are at least getting better. Given the particular political, logistical and attitude considerations in the US, that's probably the only way to get it done, but the economic damage is without doubt larger than what it could be otherwise. We are basically telling people in Maine to hunker in the basement because there is a tornado in Kansas.
Well, not exactly. I'm not familiar with the situation in Maine, but in Vermont, while there may not be a tornado of COVID-19 cases, there have been 256 confirmed cases, and clear evidence of community transmission. If the number of cases explodes, our rather limited hospital resources here would likely be quickly overwhelmed. I think the lockdown order is mainly intended to prevent that from happening.

I also think the only way to make these measures unnecessary would be to make testing available on a widespread basis so that we could identify and isolate the people that are carrying the virus... but that's still clearly weeks or even months off.
 
We're not making the progress we should be. The hammer should have come down sooner.
 

Attachments

  • 31leonhardt-newsletter-chart-jumbo.png
    31leonhardt-newsletter-chart-jumbo.png
    97.1 KB · Views: 40
Well, not exactly. I'm not familiar with the situation in Maine, but in Vermont, while there may not be a tornado of COVID-19 cases, there have been 256 confirmed cases, and clear evidence of community transmission. If the number of cases explodes, our rather limited hospital resources here would likely be quickly overwhelmed. I think the lockdown order is mainly intended to prevent that from happening.

It was an analogy and I used ME as a state where tornados (the actual meteorological phenomenon) are uncommon.

At a lower level of spread in a community, 'social distancing' and cancellation of high risk gatherings (apparently singing is a high risk activity) may be all that is needed for it to burn itself out. But that only works if you dont bring in new cases from the the high-incidence areas. Absent a national border that you can control (like South Korea did), your local control effort would be swamped by the new imports.
 
From an interview from the Director of The Centers for Disease Control.


“At the end of the day, most of us who get this infection will recover. The majority of people do — probably 98%, almost 98.5%, 99% recover. The challenge is the older, the vulnerable, the elderly, those with significant medical conditions where this virus has shown a propensity to have a significant mortality.”


https://apple.news/AZWeq6JfQTsuAUyB5BoY4HQ


Cheers
 
We're not making the progress we should be. The hammer should have come down sooner.

Those raw numbers are great to motivate people, they dont really allow you to draw conclusions.

Below is the log plot and the daily increases for italy (off the JHU site). The log plot is no longer a straight line and their daily increases are receding. And its not because 'everyone in italy now has it', its probably a function of the control effort:

Italy_3_31_log.jpgItaly_3_31_daily_inc.jpg
 
Looking at the data for the US, the last several days has shown a decrease in the rate of new cases day over day. Also I bet you could correlate the days that showed a large increase occurred on days more testing was available.

Day. New. Delta
March 26. 17,200.
March 27. 18,700. 1500
March 28. 19,500. 800
March 29. 19,900. 400
March 30. 20,400. 500

Ref: https://www.worldometers.info/coronavirus/country/us/
 
Last edited:
It was an analogy and I used ME as a state where tornados (the actual meteorological phenomenon) are uncommon.
As was I using Vermont (even though I do live there) as an example of a state where the incidence is currently at a low level, but which is taking the problem very seriously, and for good reason.

At a lower level of spread in a community, 'social distancing' and cancellation of high risk gatherings (apparently singing is a high risk activity) may be all that is needed for it to burn itself out. But that only works if you dont bring in new cases from the the high-incidence areas. Absent a national border that you can control (like South Korea did), your local control effort would be swamped by the new imports.
Exactly, and unlike tornadoes (to strain the analogy), people can travel from pretty much anywhere to anywhere, and in much less time than it takes weather systems to cross the country. Vermont's Governor Scott has asked all incoming visitors or returning snowbirds to self-quarantine for 14 days. Works only if everyone complies, no way to enforce it without taking everyone crossing the state line into custody (won't happen).

My point is that it's now a national problem and has to be attacked on a national basis.
 
The same log plot from the same site Screenshot_20200331-112141_Firefox.jpg for the US. We're not making the progress we should be.
 
I will only observe that the data on confirmed cases (with deaths lagging about 14 days behind) clearly shows that NY state has flattened the curve significantly by the imposition of stay-at-home policies, painful as they are. We are reaching what appears to be peak caseload as of this writing. The reduction in the exponential growth of the epidemic followed right on cue about 7-10 days after the imposition of the isolation policies, and was dramatic. (The delay is due to those who were already infected, but not yet diagnosed at the time of the policy implementation.) Meahwhile, the national case growth (with NY excluded) continues largely unabated.

Other states should take note, instead of being smug that exponential growth can't happen in their region. Isolation policies have more impact earlier in an epidemic that later. So if you wait until the exponential curve is well under way, the 7-10 day delay to see effects has a much larger impact than if you start early when caseload is low. I feel badly for those regions where leaders are ignoring math and science. People, not necessarily the leaders, will suffer.
 
Yes. By my estimates, Vermont's doubling rate has increased from 2.4 days a week ago to 5.2 days as of yesterday. One week after lockdown seems a little soon for it to be entirely due to that, but people may have been isolating effectively even before the order came. Only time will tell for sure, but we seem to be flattening the curve.
 
It is getting warmer and perhaps the virus is sensitive to temperature as many other seasonal flus are
Is this true? Unless warm air breaks it down I believe 70*F is a pretty "perfect" spot for this thing to survive for days outside of a human host. Florida is plenty warm, and there was virtually no measures there for a while, and even with measures a generally flip attitude to this (my brother has basically ignored this and gone about his life as usual, as many there have), and they're now 6th most infected state in the US and in the top half for infections per million. Louisiana, very warm, has the third highest infection rate per million in the whole country, behind New York and New Jersey. You also had Iran (warm), Italy (warm), Spain (warm), etc.

The measures overall taken across the US, such as shelter in place, lockdowns, and just the advice for social distancing, may have worked
That was basically my point.. from what we know so far, looking at China's aggressive lockdown and other places that took action early it seems that the infection isn't spreading as rapdily as it has elsewhere that are basically letting this run its course

That would only be true if we could test everyone before they re-emerge out of their burrow.
Good point
 
Is this true? Unless warm air breaks it down I believe 70*F is a pretty "perfect" spot for this thing to survive

Yes, most seasonal flus die down in the summer. There are several hypotheses for why that is the case, amongst them warmer temperatures, though there are others, such as people spending less time indoors and kids being out of school.

OTOH, it is unclear that SARS and MERS followed that pattern, so since those are more closely related to Covid-19, Covid-19 may not follow a seasonal pattern.

I suspect it will take quite a bit of research after the fact to really pin down which factors were causative and by how much. At lot we don't know about this one at this point.
 
Below is the log plot and the daily increases for italy (off the JHU site).

Would you care to speculate on what process causes that approach to the limit of cases in Italy? That intrigues me. That is a log plot and it looks like the log of the difference between the number of cases and the limit is decreasing exponentially. If it was a log decay to the limit of the number of cases, that would seem to indicate an exhaustion of the available non-infected people. But in this case, it is an approach on the plot of the logarithm.
 
I will only observe that the data on confirmed cases (with deaths lagging about 14 days behind) clearly shows that NY state has flattened the curve significantly by the imposition of stay-at-home policies, painful as they are. ... Other states should take note,...

It is a good temporal observation, but as in my post above, I would be a bit cautious about assuming causation. There are other potential explanations and the lack of knowing generally how many people have been exposed and when is really hurting the ability to make good inferences at this point. In the UK for example, a modeling study at Oxford suggested that in fact over 50% of the population has already been exposed and what we are seeing in their numbers is some type of function of testing bias. The estimates of the number of asymptomatic cases in the literature have varied from 20-86% and clearly that has a big impact on interpretation here.

Whether governors should institute lockdown policies based on the current data is actually a separate political question of how certain one has to be that a person infected with Covid-19 is a danger to others, but that is likely to lead to thread lock here. (I suspect the moderators are just tolerating Covid-19 since it is of great interest to the community, though has little to do with aviation.)
 
Would you care to speculate on what process causes that approach to the limit of cases in Italy? That intrigues me. That is a log plot and it looks like the log of the difference between the number of cases and the limit is decreasing exponentially. If it was a log decay to the limit of the number of cases, that would seem to indicate an exhaustion of the available non-infected people. But in this case, it is an approach on the plot of the logarithm.

I only see the log plot as a tool to visually check for the second order derivative of a exponential process. Is the 'getting worse getting worse' or is the 'getting worse getting a little bit not as bad'. Given that this only affects a small proportion of the population in all of the countries involved, I dont anticipate any approach to any limit.


Here is a link to a Arcgis site that feeds the data of the italian civil defense.

https://www.arcgis.com/apps/opsdashboard/index.html#/b0c68bce2cce478eaac82fe38d4138b1

You can select 'Nuovi positivi' for the graph in the right lower corner and remove the 'totale positivi' from the graph above it. and get this:


Italy_3_31_nuovi_guariti.jpg


As expected for a infection that as far as we know only has two outcomes of recovery and death, the number of 'dimessi guariti' (discharged recovered) and 'deceduti' (deceased) increases with a 2-3 week lag to the 'currently infected' number. At some point, the daily number of 'newly recovered' and 'newly dead' will exceed the 'new cases' and the 'number currently infected' (prevalence) will peak and eventually recede.
 
Regarding that "Exponential" claim:

He should stick to shop tools. Either that or explain the death curve. Are we now testing more people for death? Do we have enough death-test kits?
 
I am closely watching the number of tests and the sources of test results at my hospital. We are testing like crazy but have yet to find a single confirmed case.

Good that you are testing as much as possible. Do have a question; You have yet to find a single confirmed case? Is there an issue with the test? Not sure how many serious illnesses you have (and have been tested), but if there are a lot of very sick people, either their illness is not Covid-19 or the tests are giving you false negatives - that's disturbing.

There is also a substantial backlog of cases which are stratified by priorities. The highest priority is on acutely ill inpatients and the lowest priority are healthy contacts and low symptomatic outpatients.

From this, I assume (perhaps wrongly) that you do have many patients showing the symptoms associated by Covid-19 and requiring hospitalization.

I told one of the management folks today that once the positives start showing up, we should not mistake the fact that the results on the sickest patients show up first for an indication that our community is more severely affected. This nonlinearity in the reporting is systemic. First the state and CDC labs were capped out, then the commercials became available, as they don't restrict samples they got swamped and positive results got delayed among the flood of tests from the worried well who had the sniffles and talked their doc into sending out the test.

What turnaround time are you seeing from the labs?
 
It is a good temporal observation, but as in my post above, I would be a bit cautious about assuming causation. There are other potential explanations and the lack of knowing generally how many people have been exposed and when is really hurting the ability to make good inferences at this point. In the UK for example, a modeling study at Oxford suggested that in fact over 50% of the population has already been exposed and what we are seeing in their numbers is some type of function of testing bias. The estimates of the number of asymptomatic cases in the literature have varied from 20-86% and clearly that has a big impact on interpretation here.

Whether governors should institute lockdown policies based on the current data is actually a separate political question of how certain one has to be that a person infected with Covid-19 is a danger to others, but that is likely to lead to thread lock here. (I suspect the moderators are just tolerating Covid-19 since it is of great interest to the community, though has little to do with aviation.)

To confirm you look at the death growth curve. Deaths are not significantly affected by testing rates. (They don't get missed and are rarely misattributed.) It will lag 14 days or so behind the caseload curve. The NY data is just starting to show a decrease in the growth rate of COVID deaths. We'll know for sure in about a week to 10 days from now. The derivative curve for caseload is also reaching close to peak. These are all promising, if at the same time ghastly signs.
 
To confirm you look at the death growth curve. Deaths are not significantly affected by testing rates. (They don't get missed and are rarely misattributed.) It will lag 14 days or so behind the caseload curve. The NY data is just starting to show a decrease in the growth rate of COVID deaths. We'll know for sure in about a week to 10 days from now. The derivative curve for caseload is also reaching close to peak. These are all promising, if at the same time ghastly signs.

It is a better more certain measure for sure. But my point is one can’t assume causation from either measure and a temporal correlation. A lot of variables in play here and very poor measurements of fundamentals like rate of showing symptoms or whether there is even excess mortality due to Covid-19 (some data from Europe that Covid-19 displaces other deaths but has not contributed to an increase in overall deaths).

Isolation measures contributed to decreased Covid-19 deaths seems like a good first hypothesis, I agree, but I suspect it will be a study for some time to try and figure out what the causal factors were and how effective each was.
 
Last edited:
Good that you are testing as much as possible. Do have a question; You have yet to find a single confirmed case? Is there an issue with the test?

I'll send you a pm. Nothing proprietary but some of this is from internal memos.
 
... and the truth still stands that if you locked everyone at home for 14 days this would disappear and we could be coming out the other side of this that much sooner

Was thinking about it more and this is in general an important observation. If you lock everyone in their house for a month, the propagation of this will stop. The problem is, everyone will have starved to death. Similar to how if you ban all GA flying, there will be no more deaths due to GA accidents.

There is a tradeoff involved at less absurd levels of intervention, and a fundamental question, though a political one so I will stop after noting it, is what is the appropriate tradeoff between other deaths and other costs and Covid-19 deaths and illnesses? Not a question we have much good data to answer presently, but there is a tradeoff there in reality whether we choose to acknowledge it or not.
 
Another item I will share from something I have been tracking -- which is total US confirmed cases and deaths. It appears from some basic statistical testing that the rate of exponential growth started decreasing in a significant manner about March 25. While deaths in the US has accelerated a bit today, 14 days from then will be April 8th. Let's hope the deaths decelerate then for the US as a whole also.
 
Yes, most seasonal flus die down in the summer. There are several hypotheses for why that is the case, amongst them warmer temperatures, though there are others, such as people spending less time indoors and kids being out of school.

OTOH, it is unclear that SARS and MERS followed that pattern, so since those are more closely related to Covid-19, Covid-19 may not follow a seasonal pattern.

I suspect it will take quite a bit of research after the fact to really pin down which factors were causative and by how much. At lot we don't know about this one at this point.
Flus (influenza) != coronaviruses
I wouldn't try to compare one to the other. Some "common colds" are caused by coronaviruses, some of which do follow the seasons as you describe.
 
Flus (influenza) != coronaviruses
I wouldn't try to compare one to the other. Some "common colds" are caused by coronaviruses, some of which do follow the seasons as you describe.

You are correct, the normal seasonal flu is not a coronavirus. Nonetheless, many viral infections follow a seasonal pattern, with infections lower in the summer. As you note, some of these are colds, which are coronaviruses. Pretty active debate amongst experts right now if COVID-19 will follow a seasonal pattern. We can hope but remain vigilant.
 
You are correct, the normal seasonal flu is not a coronavirus. Nonetheless, many viral infections follow a seasonal pattern, with infections lower in the summer. As you note, some of these are colds, which are coronaviruses. Pretty active debate amongst experts right now if COVID-19 will follow a seasonal pattern. We can hope but remain vigilant.

Just for clarification. What the CDC tracks routinely are 'flu like illnesses and pneumonia'. That contains a basket of bugs including influenza A, B, parainfluenza, various bacteria, adenovirus, respiratory synctial virus and yes non-SARS coronaviridae. Anyone who wants to make comparisons to 'the flu' needs to be aware that they are comparing one specific virus to a basket of pathogens, not one specific virus. While there are 'respiratory panel' labtests, many patients who present to a hospital with 'the flu' don't end up getting diagnosed with a specific virus. For the epidemiologists and vaccine designers to figure out what bugs are circulating that year isnt trivial.
 
Back
Top