Did you catch it ?

[PeterNSteinmetz]

It is "much lower" because the R0 is an exponential factor, as in e^(r0*t). If you double the R0, you halve the doubling time. That's a very big deal for exponential growth. (5 doublings is 32x initial; 10 doublings in the same time period is 1024x initial. Huge difference in growth of raw numbers.) Measles is one of the most contagious diseases known, undoubtedly. Covid-19 has an R0 about double that of seasonal flu, based on current knowledge.

While the point is well taken taken that the doubling rate depends on R0, the relationship depends on the latency assumed in a model and has a different form than the equation given. (https://en.wikipedia.org/wiki/Basic_reproduction_number).

For a very simply model, the relationship between the doubling time, Td, and R0, would be as follows: Td = ln(2) tau / ln(R0). Agreed that a 2X change if R0 results in changing the timescale of the exponential growth by a factor of 2.

Since exponential growth always looks nearly vertical on an appropriate timescale, the practical question in the present case is whether the 2X difference in R0 versus the seasonal flu really impacts what should be done.

As modeled by Kissler et al. with these types of numbers, a single period of social distancing with relatively severe measures may be insufficient to avoid over-running existing healthcare resources in the US. If SARS Cov-2 is moderately seasonal, it may make things worse in the fall.

Are there any academic studies using the current data which suggest that coercive social distancing measures ordered by the government have increased the effective R0? While there have been a lot of reports in the lay press, there are a lot of other confounding factors that one would have to account for to reasonably demonstrate this has worked in the US. I have not seen any so far but this is rapidly developing.

 

[Bell206]

I want to see the numbers for NON-COVID-19 deaths.

Those deaths will be culled out in the NVSS Provisional Death counts which are determined from submitted death certificates. They are currently tracking several different COVID categories which at the end will be quantified at years ends. The same method is used with the flu, overdoses, etc.
https://www.cdc.gov/nchs/nvss/vsrr/COVID19/index.htm

 

[Juliet Hotel]

Those deaths will be culled out in the NVSS Provisional Death counts which are determined from submitted death certificates.

What examiners are able to put on death certificates varies by location. In some locations they can't list Covid-19 on the cert unless they have a positive test. But examiners can't perform testing because they don't have tests available.

 

[Kenny Phillips]

What examiners are able to put on death certificates varies by location. In some locations they can't list Covid-19 on the cert unless they have a positive test. But examiners can't perform testing because they don't have tests available.

NYC is now counting deaths that are likely to have been caused by C19, hence a spike in the numbers. If others did that, I'd expect the total of deaths to be 40K+ already.

 

[Bell206]

What examiners are able to put on death certificates varies by location.

FWIW: there's specific guidance from CDC/NVSS on COVID death certificates and if I recall there are provisions if the COVID can not be validated by test or autopsy. However, this data lags behind the current public data but will end up in the final reports. The deputy coroner who I asked said on days where there are elevated deaths that preclude subsequent testing/investigation they include additional under-lying information on the certificate to aid in future classification per NVSS guidance. It's my understanding that while there may be some initial skewed data on COVID deaths, in the end analysis the numbers will be more accurate as they are required for future policy and procedures.

 

[Juliet Hotel]

FWIW: there's specific guidance from CDC/NVSS on COVID death certificates and if I recall there are provisions if the COVID can not be validated by test or autopsy.

There are. But from what I've read, there are places where local regulations contradict or are otherwise not in alignment with that guidance. If the CDC says list it even if you can't test for it but the local regulations say its illegal for the coroner to do that, who is the coroner supposed to listen to?

 

[Palmpilot]

New York Gov. Cuomo issues executive order requiring face coverings in public

Excerpt:

ALBANY, N.Y. (WIVB) — The healthcare system has stabilized, Gov. Andrew Cuomo says, but caution still must be taken.


On Wednesday morning, Cuomo announced he was issuing an executive order that New Yorkers must wear coverings on their nose and mouth while near others in public.


People must wear coverings if they cannot be six feet or more from someone. State residents have three days to comply with this order.​

 

[Kenny Phillips]

There are. But from what I've read, there are places where local regulations contradict or are otherwise not in alignment with that guidance. If the CDC says list it even if you can't test for it but the local regulations say its illegal for the coroner to do that, who is the coroner supposed to listen to?

What the coroner reports to the CDC and what he puts on the death certificate aren't the same thing.

 

[Juliet Hotel]

What the coroner reports to the CDC and what he puts on the death certificate aren't the same thing.

A article I read indicated death certificate is what gets reported to CDC. If that's not the case then I stand corrected.

 

[Bell206]

What the coroner reports to the CDC and what he puts on the death certificate aren't the same thing.

It's my understanding the CDC oversees the NVSS and NCHS which manages and collects all vital statistics to include the standard death certificates issued by all states. The CDC then uses this mortality information to directly update all their current data and guidance. Perhaps there are additional CDC information requirements but the CDC has been providing direct guidance to those who certify deaths and how to complete the death certificates in reference to COVID. There's another CDC webinar tomorrow on this subject if you wish to watch.
https://emergency.cdc.gov/coca/calls/2020/callinfo_041620.asp

 

[Juliet Hotel]

https://www.cnn.com/2020/04/06/health/coronavirus-coroners-uncounted-deaths-invs/index.html

 

[IK04]

"If you die in my county, I will not know if you died of Covid-19," Romann said. "I will, however, be able to tell if you legally smoked pot."

What a totally screwed up world we have allowed to devolve to this level....

 

[chemgeek]

While the point is well taken taken that the doubling rate depends on R0, the relationship depends on the latency assumed in a model and has a different form than the equation given. (https://en.wikipedia.org/wiki/Basic_reproduction_number).

For a very simply model, the relationship between the doubling time, Td, and R0, would be as follows: Td = ln(2) tau / ln(R0). Agreed that a 2X change if R0 results in changing the timescale of the exponential growth by a factor of 2.

Since exponential growth always looks nearly vertical on an appropriate timescale, the practical question in the present case is whether the 2X difference in R0 versus the seasonal flu really impacts what should be done.

As modeled by Kissler et al. with these types of numbers, a single period of social distancing with relatively severe measures may be insufficient to avoid over-running existing healthcare resources in the US. If SARS Cov-2 is moderately seasonal, it may make things worse in the fall.

Are there any academic studies using the current data which suggest that coercive social distancing measures ordered by the government have increased the effective R0? While there have been a lot of reports in the lay press, there are a lot of other confounding factors that one would have to account for to reasonably demonstrate this has worked in the US. I have not seen any so far but this is rapidly developing.

Well, in NY, the effective R0 was reduced from 1.8 to 0.9 in about 4 weeks. The reduction can't be attributed to herd immunity, because only a small fraction of the population has been infected. Similar story in every other region that has implemented physical distancing. There is lots of raw data in the US that shows dramatic changes in exponential growth rates about 10-14 days after implementation of physical distancing. The NY data is particularly dramatic, with caseload growth going from a doubling time of less than 2 days to over 15 days in a 4 week period. Similar, but less dramatic trends can be seen in other regions.

These policies are primarily intended to staunch exponential growth and get to a point where control is possible through contact tracing of isolated outbreaks. However, to keep effective R0 values below 1 some vestiges of physical distancing are likely to be required to remain in place until herd immunity can be achieved through vaccination. Had doubling times in NY remained uncontrolled at 2-3 days doubling times for even a few weeks, the impact would have been devastating beyond imagination. It is bad enough as it is.

There is more interesting data at the county level that shows the advantages of intervening early, instead of waiting for high caseloads to react. Some rural counties in NY state that implemented controls before things really took off, because of statewide stay at home orders, are seeing declines to pre-emergent caseload levels after a brief peak, and may be in the enviable position to control future outbreaks through testing and tracing. If we can get enough tests. Our local rural county had an initial outbreak of cases that persisted for 2 weeks, peaked, is now declined to an average or 2-3 cases per day.

 

[chemgeek]

NYC is now counting deaths that are likely to have been caused by C19, hence a spike in the numbers. If others did that, I'd expect the total of deaths to be 40K+ already.

Those categories of deaths (confirmed and probable) are being kept separately, but different data aggregators are treating them differently. So if you are tracking statistics, you need to figure out who is reporting what. So today there were 600 additional confirmed deaths in NY, but some aggregators are reporting a spike of over 4000 because of the inclusion of a backlog of probable deaths. It's a pain to when you are doing data analysis and projections.

 

[deonb]

I want to see the numbers for NON-COVID-19 deaths. Probably some shocking numbers for everyone except the vultures who are hyping the relative comparison to normal flu and miscellaneous virus deaths. Those numbers are a whole lot bigger than the average dumass knows..

Here you go. Comparing deaths from all causes.

It's 2 weeks old though - take that red line on the right and double it.

 

[PeterNSteinmetz]

Well, in NY, the effective R0 was reduced from 1.8 to 0.9 in about 4 weeks. The reduction can't be attributed to herd immunity, because only a small fraction of the population has been infected.

I would very much like to see a publication and careful study of this sort with appropriate control for population density etc. Haven’t seen that yet.

As I understand it, there are a number of possible issues with this analysis.

1. There are a number of cases where the data don’t line up so well.

2. The timing is often off, with the beginnings of the downturn being relatively early compared to the implementation of coercive lockdown measures.

3. Measured decreases in social mobility do not agree with the levels of coercive lockdowns enacted.

4. Since we don’t know the actual fraction infected yet, it is hard to know if this is some type of effect of saturating easily infected remaining population.

I agree that it is possible the observed increases in the doubling time may be due to social distancing. But there are other possible explanations. Would love to see a properly designed analysis. So far, none published that I know of.

And the important question, in terms of governments imposing coercive measures, is whether those actions in particular helped or hurt in terms of total mortality and morbidity.

 

[PeterNSteinmetz]

Here you go. Comparing deaths from all causes.

It's 2 weeks old though - take that red line on the right and double it.


View attachment 84838

What happens in the last two weeks? Is that supposed to be a data point at the end, or just an artifact? Or a point for the whole two weeks? And did you add that?

 

[deonb]

What happens in the last two weeks? Is that supposed to be a data point at the end, or just an artifact? Or a point for the whole two weeks? And did you add that?

Since the graph was released 2 weeks ago COVID-19 deaths went from 9780 in New York to 16700 - those are still April. The datapoint at the end of the graph is just basically the deaths from March 4 to April 4.

If you shift it forward by 2 weeks the last datapoint becomes March 17 to April 17, which will shift the March 4 to 17 deaths into the previous month, but of any significant count (only 110 deaths US-wide on March-17).

Ignore my 2-week thing, it's only a complication. There will be new data at the end of April. The point was just that even for the half we have right now, the count is much larger than any previous total deaths-from-all-causes count, including 9/11.

 

[PeterNSteinmetz]

Ignore my 2-week thing, it's only a complication. There will be new data at the end of April. The point was just that even for the half we have right now, the count is much larger than any previous total deaths-from-all-causes count, including 9/11.

I will be very curious to see when the numbers come out. In Europe as of about a week ago, total mortality was actually down somewhat, excepting Italy.

 

[Kenny Phillips]

Here you go. Comparing deaths from all causes.

It's 2 weeks old though - take that red line on the right and double it.


View attachment 84838

As of now, the death toll in NYC from COVID-19 is more than four times the number that died when the twin towers were attacked.
China is now 'fessing up to a few thousand more deaths.

 

[Half Fast]

As of now, the death toll in NYC from COVID-19 is more than four times the number that died when the twin towers were attacked.
China is now 'fessing up to a few thousand more deaths.

Are you suggesting we respond as we did after 9/11, and start bombing China? At least that would drive up my company stock.

"All we are saying,
Is give war a chance."​

 

[Kenny Phillips]

Are you suggesting we respond as we did after 9/11, and start bombing China? At least that would drive up my company stock.

"All we are saying,
Is give war a chance."​

Actually, I'd say we grossly overreacted the last time, when we could have just used a couple of old nukes that we had in stock to stop any threat.
I'm just pointing out to some on here that said this would be nothing, that it is indeed something (and it may be much worse, if the South Korean reports of reinfection turn out to be the case; then we'll need a Manhattan Project vaccine drive.)

 

[deonb]

Actually, I'd say we grossly overreacted the last time, when we could have just used a couple of old nukes that we had in stock to stop any threat.
I'm just pointing out to some on here that said this would be nothing, that it is indeed something (and it may be much worse, if the South Korean reports of reinfection turn out to be the case; then we'll need a Manhattan Project vaccine drive.)

If it really turns out to be the case, and the secondary infection CFR is as high as the primary, we'd have to do a drive-to-zero - a vaccine won't actually help.

Thankfully that's fairly unlikely.

 

[chemgeek]

I would very much like to see a publication and careful study of this sort with appropriate control for population density etc. Haven’t seen that yet.

As I understand it, there are a number of possible issues with this analysis.

1. There are a number of cases where the data don’t line up so well.

2. The timing is often off, with the beginnings of the downturn being relatively early compared to the implementation of coercive lockdown measures.

3. Measured decreases in social mobility do not agree with the levels of coercive lockdowns enacted.

4. Since we don’t know the actual fraction infected yet, it is hard to know if this is some type of effect of saturating easily infected remaining population.

I agree that it is possible the observed increases in the doubling time may be due to social distancing. But there are other possible explanations. Would love to see a properly designed analysis. So far, none published that I know of.

And the important question, in terms of governments imposing coercive measures, is whether those actions in particular helped or hurt in terms of total mortality and morbidity.

I'm not sure I understand points 1 and 2. There is already good data, publicly available so you can look at it yourself, where you can examine the hypothesis that physical distancing has effectively reduced viral transmission. The NY state data is an excellent case in point. You can (and I have) compared that data to other states and also to the US as a whole, excluding NY state, which was one of the first and earliest to implement stay-at-home policies. The difference in that data is stark, and if conforms very nicely to the expected outcome. You can also look at other state data and compare implementation dates of stay-at-home policies and caseload growth, and there are generally similar results. Physical distancing is not only well correlated by the data, but it also has a known causative mechanism. It is probably too early to have peer-reviewed publications on this topic, because we are still in the middle of the outbreak. There will be plenty of peer-reviewed publications when this thing has run its course. Bottom line: current data strongly supports the contention that physical distancing has had a significant impact on viral transmission, and public health and infections health experts, who are also monitoring the data in real time, agree. There is no existing supporting data that other factors (Sunspots? Eating broccoli?) are responsible for the decrease in case growth rates we have seen in many locales.

Regarding point 4. There is no good evidence, even if one considers extremely optimistic estimates of exposure rates, that herd immunity is a factor in the current progress of the epidemic in the US. With the known R0 range for COVID-19, herd immunity would require somewhere between 70-90% exposure of the population. The most GENEROUS antibody study in the US so far (for which I have not seen supporting experimental protocols and data) shows about 15% exposure of ONE local population. As interesting as that result is, that is not anywhere close enough to have a significant impact on viral transmission. There are also some important experimental concerns about antibody testing that should be satisfied before we accept them at face value. Many of the currently available antibody tests are of questionable accuracy. A proper study would include validation data (something analytical chemists are quite familiar with) to ensure it is not picking up false positives from non-COVID coronaviruses, which are very common.

Science is based on hypothesis (reasonable cause-effect mechanism) and supporting data, not rationalization. The overwhelming scientific consensus points to the importance of physical distancing in controlling the current outbreak. It's not really very controversial, as this is a field of science that is relatively well understood. This isn't the first rodeo for viral outbreaks and their behavior. There is virtually no disagreement (I mean, is there really ANY disagreement on this point?) that physical distancing has reduced both morbidity and mortality due to COVID-19, and publicly available data supports that claim. If people are not exposed, they can't get sick and they can't die from the virus. Balancing economic and public health issues is another thing, and that is a question of ethics and sociology, not science.

We are about to get some uncontrolled experiments in epidemiology by selected regions of the country that will likely relax their physical distancing policies while still in the initial exponential growth phase of their outbreaks. We shall see how well that turns out. Texas may be a good laboratory.

I understand the need to rationalize some sort of better, happy ending for this pandemic. Science says there is no pain-free happy ending right now. Disney's Law ("Wishing will make it so") does not apply, unfortunately. We have to try to do the best with the hand we are dealt. The virus is nobody's "fault." We are wasting our time fault-finding. We can only move forward. The sooner we prioritize building nationwide testing, restoring health care capacity, and development of a deployable and effective vaccine (and maybe some therapeutics along the way as a stopgap), the faster we emerge from this mess. The only tool in our toolbox at the moment, in the absence of these things, is avoidance, unfortunately.

 

[PeterNSteinmetz]

Thanks for the continued discussion as I am very curious about this issue.

I'm not sure I understand points 1 and 2.

For #1. While the contrast between Kentucky and Tennessee is often held out as proof that social distancing works, epidemiologists in Tennessee note that actual measures of distancing obtained from cell phone records indicate more compliance in Tennessee, yet that is the state with the higher number of cases, so they don’t buy the explanation that differences in social distancing policies are the explanation.

For #2. Nationally the measured exponential growth rate began to drop about March 25. My understanding is that the predictions of the delay between implementation of coercive social distancing measures and the drop in propagation were for a greater delay than what was observed.

I would be quite happy to see a dataset with the timing of implementation of the policies and the confirmed case rates on a state by state basis. I can always be persuaded by the data.

The NY state data is an excellent case in point. You can (and I have) compared that data to other states and also to the US as a whole, excluding NY state, which was one of the first and earliest to implement stay-at-home policies. The difference in that data is stark, and if conforms very nicely to the expected outcome.

Can you provide the data and your analysis? I have not had the time to try and pull this together myself yet but would be very happy to see it.

It is probably too early to have peer-reviewed publications on this topic, because we are still in the middle of the outbreak.

Well we have preprints on all sorts of other aspects of this. You say you’ve already performed an analysis. I imagine those who work in this area specifically have had plenty of time to do something similar. Where are those preprints? Perhaps they will be forthcoming, but until they do so, I think it is a bit premature to predict what they will say.

My suspicion is that the relationship here is more nuanced and not so obviously clear. Therefore it takes some time and effort to tease apart the various factors. This is not a physics or chemistry experiment, but rather biomedical and social science where the evidence is often much softer and difficult to interpret than in the hard sciences.

Regarding point 4. There is no good evidence, even if one considers extremely optimistic estimates of exposure rates, that herd immunity is a factor in the current progress of the epidemic in the US.

I agree that the results from the recent studies in the US strongly suggest herd immunity is likely not a factor, at least not in the normal sense.

A proper study would include validation data (something analytical chemists are quite familiar with) to ensure it is not picking up false positives from non-COVID coronaviruses, which are very common.

What did you think of the validation procedures in the recent Stanford study? They manufacturer reportedly had something like 390 negatives against sera from prior to the epidemic.

There is overwhelming scientific consensus points to the importance of physical distancing in controlling the current outbreak.

Argument from authority. I’ll be convinced by the data, analysis, and publications. Where are they?

There is virtually no disagreement (I mean, is there really ANY disagreement on this point?) that physical distancing has reduced both morbidity and mortality due to COVID-19, and publicly available data supports that claim.

Again, please provide citations. I have seen no preprints addressing this. Without citations to at least preprints, it seems unjustified to assert this.

If people are not exposed, they can't get sick and they can't die from the virus. Balancing economic and public health issues is another thing, and that is a question of ethics and sociology, not science.

Agreed the cost trade offs are not a scientific question and that in the limit distancing has to work to stop the spread. The question is whether in this case the interventions used have worked, and if so, how well.

One of my primary interests is whether the coercive orders made a difference. It is possible that people were distancing based on recommendations prior to the coercive orders, yet it may be that the coercive orders have been the primary thing causing other damage, such as deaths due to other causes and economic damage. That is a very important question from a policy perspective and measuring the relative contributions will require the type of data and analyses I have been discussing.

 

[MuseChaser]

I "liked" the last two posts because, although they disagree on interpretation of the data, they are both very thoughtful, measured, and logical interpretations. I wish I knew what the reality TRULY was.

A few, much less scientific personal thoughts..

1. Massachusetts and the SF Bay area, independently, have both recently released random testing samplings that indicate that 50 to 85 times MORE people have been infected and tested positive for antibodies than expected. Not 50 to 85 more people... not 50 to 85 percent more people... 50 to 85 TIMES more people. Therefore, the mortality rate would seem to also be 50 to 85 times lower than previously reported.

2. We, as a globe, country, society, or however you'd like to sectorize the population, have been through terrible diseases in the past. Polio. The Spanish Flu epidemic of 1918. It's a long list. Tragic. Many, many people died and families suffered. We did the best we could, but couldn't save millions of people. Unfortunately, terrible tragedies are part of life.

3. Right now, there is an endless, and I mean ENDLESS barrage of media attention to COVID. Not just the news, but CEOs from virtually every company sending out missives, television shows showing how much they care, constant compassionate talk about all of the suffering caused by COVID... and yet, COVID itself is causing true suffering in an EXTREMELY small percentage of the population. The virtually universal suffering, at least in the US, is caused not directly by COVID, but by our reaction to it. I understand, COMPLETELY, the dangers of this highly infectious virus in terms of overwhelming our health care system if everyone who would be severely impacted by it got impacted simultaneously.. that is THE concern. We want to be able to care for those who truly need the care. Is the best way to care for those people crippling many of our hospitals, now on the verge of bankruptcy due to lack of patients, or is the best way to care for those people remaining active, taking intelligent precautions, and keeping a strong economy and nation capable of helping those who need help?

This whole thing reminds me of the health-care equivalent of an argument used, in my opinion validly, to show the illogicalness of socialism. If you took all of the world's wealth and distributed it equally among the world's population, everyone would have approximately $7000. In other words, EVERYONE would be powerless to provide any real financial or powerful assistance to ANYONE. Right now, we are ALL suffering in some way, and that does NO one any good.

What is the point in living if your only goal is subsistence, rather than productivity, benevolence, and fulfillness? What is the point in being safe... if there is no OTHER point?

 

[PeterNSteinmetz]

Thanks for the kind remarks. While I am largely in agreement with you on the political aspects, I will just comment on the science that you mention under #1.

The Chelsea, Massachusetts data says 1/3 of randomly selected are seropositive for antibodies. Wow. https://www.bostonglobe.com/2020/04...ples-taken-chelsea-show-exposure-coronavirus/

That is about twice the fraction from the Gangelt study in Germany and 10 times the 2.5-4% estimated in Santa Clara County. The fraction seropositive may reflect how far along a community is in the growth of the infection and these different numbers may reflect that.

50 to 85 TIMES more people. Therefore, the mortality rate would seem to also be 50 to 85 times lower than previously reported.

Interesting, the numbers from that article about Chelsea would suggest an infection fatality rate of 0.33%. This is in rough agreement with the results in Gangelt and about 50% more, relatively speaking, than the estimates from the Santa Clara study of 0.12-0.2%. All are much lower than the feared 8% from initial reports and in an absolute sense on the same order as the seasonal flu for infection fatality rate.

As I imagine others here will be quick to point out (hat tip @chemgeek), SARS Cov-2 does appear to have an effective R0 of about twice the seasonal influenza from estimates early this month, so that it's doubling time is about half that of the seasonal flu. (But I really don't know yet what to make of the linear growth in confirmed cases in the last 2 weeks, that seems incompatible with exponential growth.)

 

[chemgeek]

Can you provide the data and your analysis? I have not had the time to try and pull this together myself yet but would be very happy to see it.

You can find the data on GitHub and explore to your heart's content. JHU has the largest dataset, although it can be harder to navigate. The NY Times has simplified and re-aggregated that dataset for confirmed cases and deaths, aggregated by states and counties in separate files. That dataset has the same data in an easier to use format.

If you look at the growth of cases in many states, you will find that they are still growing exponentially, although rates are slowing in many locations, some more than others. For example, in Texas, new cases are doubling about every 5 days or so. In NY it has slowed to a t(2) of 15 days. NY has passed the inflection point of the logistic curve and is probably going to wind up with about 16,000 confirmed deaths or so. Nationwide, the inflection point is a moving target (hasn't been reached yet) and logistic curve analysis points to somewhere north of 60,000 deaths. That's if nothing changes from the current situation.

 

[Kenny Phillips]

I "liked" the last two posts because, although they disagree on interpretation of the data, they are both very thoughtful, measured, and logical interpretations. I wish I knew what the reality TRULY was.

A few, much less scientific personal thoughts..

1. Massachusetts and the SF Bay area, independently, have both recently released random testing samplings that indicate that 50 to 85 times MORE people have been infected and tested positive for antibodies than expected. Not 50 to 85 more people... not 50 to 85 percent more people... 50 to 85 TIMES more people. Therefore, the mortality rate would seem to also be 50 to 85 times lower than previously reported.

I hope that we don't find out that much of that antibody testing is pickup up on coronaviruses that cause the common cold; that's been mentioned before, and it would effectively trash any results.
But even if this has a kill rate of the typical flu, it spreads much more easily. And it seems to have singled out a lot of notable people, killing several (RIP John Prine).
So I look at the high indicated infection rates with quite a bit of skepticism, at least for now.

 

[PeterNSteinmetz]

You can find the data on GitHub and explore to your heart's content. JHU has the largest dataset, although it can be harder to navigate. The NY Times has simplified and re-aggregated that dataset for confirmed cases and deaths, aggregated by states and counties in separate files. That dataset has the same data in an easier to use format.

What I am particularly curious about is the comparison which you said you made that shows clearly that the times of implementation of coercive and voluntary policies for distancing correlate with a change in the in the growth rate of the infections.

I am aware of these different datasets regarding the reported cases, thanks, but where is the data about the policy implementation times?

And what methods of comparison did you use to conclude there was a relationship in the time series?

As I have noted repeatedly, I am not aware of any present preprints or studies that do that in a serious manner.

Nationwide, the inflection point is a moving target (hasn't been reached yet) and logistic curve analysis points to somewhere north of 60,000 deaths. That's if nothing changes from the current situation.

The US national rate of confirmed cases appears to be a linear growth mode for the last two weeks, quite strongly so actually. That strikes me as quite odd and indicating some other type of limited process. Other people have speculated it may reflect a limit on the number of tests which can be processed per day, but I have been unable to find a serious reference on that.

(https://www.worldometers.info/coronavirus/country/us/ shows the US cases on both a linear and log scales.)

In terms of US confirmed cases, the overall trajectory looks consistent with an initial exponential growth phase but then striking some sort of limit that produces linear growth, rather than something having changed the exponential growth rate, which was my initial thought about it.

I do hope that whatever is going here limits deaths to that many. If one looks at the US national death rates, the curve strikes me as still fitting an exponential, with a weird bump on April 14. A davies test for changes in slope of the log of deaths as a function of time between March 16 and April 18 yields a highly significant result (p<2x10^-16) with 5 inflection points on March 30, April 3, April 8, April 12, and April 17.

 

[Palmpilot]

I'll be interested in seeing what the brain trust thinks about this:

https://rt.live/

"These are up-to-date values for Rt, a key measure of how fast the virus is growing. It’s the average number of people who become infected by an infectious person. If Rt is above 1.0, the virus will spread quickly. When Rt is below 1.0, the virus will stop spreading."​

 

[chemgeek]

You can find implementation dates for distancing policies here. If you examine data on a short time scale it will always look linear. It's easier to see if growth is exponential by looking at a semilog plot. If that is linear, growth is exponential. It is also easy to clearly see on semilog plots when growth rates change. Both the US overall or NY state data has not grown linearly. You can see that by simply examining a full plot of the raw data. It's not a straight line. Both caseload and deaths data has closely followed a canonical logistic curve, which is a self-limiting exponential growth function typical of phenomena like epidemics.

On the other hand, my local mostly rural county, which intervened very early in the epidemic, never saw exponential growth, but rather a languid linear growth for about two weeks, followed by a decline in case growth over a 3 week period to early epidemic levels. We are small enough that our limited testing capacity can keep up with the necessary tracing and isolation required to keep things under control.

 

[Palmpilot]

There has been a lot of discussion of the importance of testing and contact-tracing in order to reopen the economy in a safe manner. I ran across a fifteen-minute portion of Friday's White House briefing, in which some interesting details of the logistics of those subjects were discussed. (Scroll to about the 4:17:00 mark.)

(For those who are interested in the entire discussion of testing, scroll to about the 3:45:00 mark.)

 

[chemgeek]

I'll be interested in seeing what the brain trust thinks about this:

https://rt.live/

"These are up-to-date values for Rt, a key measure of how fast the virus is growing. It’s the average number of people who become infected by an infectious person. If Rt is above 1.0, the virus will spread quickly. When Rt is below 1.0, the virus will stop spreading."​

What you are referring to is the exponential growth factor, also known as R0, or "R-nought". R0 is intrinsic to the virus, but the "effective" R0 can be lowered by interventions. If R0>1, infections grow exponentially. The larger R0 the faster it grows. If R0=1 the infections are stable, neither increasing or decreasing. If R0<1 infections will exponentially decline. The lower R0 the faster the decline. Covid-19 has an intrinsic R0 estimated to be at least 2.3-2.6, which is very high compared to seasonal flu. Interventions like distancing, vaccination, hygiene, can decrease the effective R0. The goal of public health efforts is to drive R0 under 1.0. If it gets much above 1 an epidemic is difficult to control. Consider that seasonal flu is R0 of 1.3-1.6.

 

[chemgeek]

BTW, don't get too excited yet about antibody testing, as it is fraught with problems. These tests are based on ELISA assays, which require very specific antibody-antigen interactions to work as intended. Many if not all,of the current test kits are using antigens that have significant cross-reactivity with other common coronavirus strains that cause the common cold. False positives are going to be very prevalent, especially considering that the true negatives are expected to be 85-95% of the test pool. ELISA assays are notorious in the research lab for their ability to be insufficiently specific.

 

[Palmpilot]

What you are referring to is the exponential growth factor, also known as R0, or "R-nought". R0 is intrinsic to the virus, but the "effective" R0 can be lowered by interventions. If R0>1, infections grow exponentially. The larger R0 the faster it grows. If R0=1 the infections are stable, neither increasing or decreasing. If R0<1 infections will exponentially decline. The lower R0 the faster the decline. Covid-19 has an intrinsic R0 estimated to be at least 2.3-2.6, which is very high compared to seasonal flu. Interventions like distancing, vaccination, hygiene, can decrease the effective R0. The goal of public health efforts is to drive R0 under 1.0. If it gets much above 1 an epidemic is difficult to control. Consider that seasonal flu is R0 of 1.3-1.6.

OK, sounds like they are just using a different designation for what has been discussed here all along.

 

[Palmpilot]

BTW, don't get too excited yet about antibody testing, as it is fraught with problems. These tests are based on ELISA assays, which require very specific antibody-antigen interactions to work as intended. Many if not all,of the current test kits are using antigens that have significant cross-reactivity with other common coronavirus strains that cause the common cold. False positives are going to be very prevalent, especially considering that the true negatives are expected to be 85-95% of the test pool. ELISA assays are notorious in the research lab for their ability to be insufficiently specific.

Yeah, Dr. Birx talked about the importance of not giving people false reassurance due to false antibody positives in Friday's briefing.

 

[IK04]

BTW, don't get too excited yet about antibody testing, as it is fraught with problems. These tests are based on ELISA assays, which require very specific antibody-antigen interactions to work as intended. Many if not all,of the current test kits are using antigens that have significant cross-reactivity with other common coronavirus strains that cause the common cold. False positives are going to be very prevalent, especially considering that the true negatives are expected to be 85-95% of the test pool. ELISA assays are notorious in the research lab for their ability to be insufficiently specific.

Yep. Just because I had "it" in January, doesn't mean it was exactly the same flavor of Corona in COVID-19. I'm beginning to believe most everything we are being told is based on bad guesses and politically influenced worse actions...

 

[PeterNSteinmetz]

Many if not all,of the current test kits are using antigens that have significant cross-reactivity with other common coronavirus strains that cause the common cold. False positives are going to be very prevalent, especially considering that the true negatives are expected to be 85-95% of the test pool. ELISA assays are notorious in the research lab for their ability to be insufficiently specific.

Curious, what did you think of the controls which were run by the manufacturer of the kits used in the Santa Clara study? They tested against 371 pre-Covid samples with 369 negative results. It strikes me that argues that that test is not reacting with other coronaviruses which were around at the time, though I suppose depending on where the pre-Covid samples were obtained, they might not have contained the other coronaviruses.

Article available here: https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v1

 

[Palmpilot]

Yep. Just because I had "it" in January, doesn't mean it was exactly the same flavor of Corona in COVID-19. I'm beginning to believe most everything we are being told is based on bad guesses and politically influenced worse actions...

The idea that the medical folks are basing their statements on politics is pure fantasy.