I think things are turning around with COVID-19

[gbwillner]

So I have access to all the data for COVID-19 and have been tracking it daily since 3/11. I think that we are hitting the start of the flattening of at least the death curve.

I focused on following deaths because we don't know the prevalence of the disease and it seemed to be the only reliable metric. Now, one issue with deaths is that it is reflecting transmissions that occurred ~12-16 days ago, since the average time before symptoms post contraction of the disease is ~4-5 days and the few studies published so far show that approx. patient course from admission to ICU to death is about 10-14 days. So looking at deaths today means understanding what happened in people contracting COVID about 2 weeks ago (on average).

From 3/11 to 3/26 there was a general trend where the doubling of deaths was speeding up from 4 to every 3 days. This started to slow on 3/27, and between then and 4/1 it was 4 days. for 4/2 it was 5 days. Assuming that the death rates for the disease are stable (and there would be no reason to doubt that as there are no specific new therapies to treat), this means that transmissions are decreasing (and subsequent deaths in the minority of patients who get sick enough to go to the ICU).

Following a similar trend, the number of daily deaths grew at a very alarming rate between 3/11 and 4/4. While there were days that deaths were less than the day before, this was likely due to stochastic events and dealing with small numbers. This trend never repeated and was rare. For example, between 3/13 and 4/4 there were two days where the deaths were lower the the preceding day, and in both cases these death numbers were preceded or followed by a large jump in the total deaths- these were on 3/21 (53 deaths, with 59 deaths the prior day, followed by 126 deaths), and 3/23 (73 deaths, following the large jump on 3/22 of 126 deaths and followed on 3/24 with 204 deaths). However, the new deaths total for 4/4 was 1352 and has been decreasing for 2 continuous days now (1184 in 4/5 and 1182 4/6). It's still early, but given a large step downward and the start of a possible trend is hopeful.

While it may be desirable to track hospitalizations, these data have been hard to track since they are not very complete and have only been available since 3/21. Although flawed, these numbers also follow this trend, as COVID+ hospitalizations peaked on 4/1 with 4482 new hospitalizations and has been variable since between 1488 and 3796 admissions.

Lastly, although even more flawed due to a lack of tests or strategy for testing, the new positive cases also follows this trend, with 33,840 cases on 4/4 and a decrease since. Like deaths, this number has steadily increased daily with rare exception except for a single day (3/24). However, I don't think this turnaround can be attributed to testing availability. One metric that has only trended up every day is the proportion of cases that test positive (positive test rate). This started at 10.0% on 3/16 and slowly crept up incrementally every day until 4/3 when it hit 19.3%. Since then that number has also come down slightly and stabilized at 18.8% for the last 3 days.

My take from this is that this might be early signs that whatever we were doing as a country starting about 2 weeks ago is likely starting to put a dent in spread of COVID-19.

---

Members don't see this ad.

 

[BoneWizard]

Couldn't this just be due to it being the weekend?

 

[gbwillner]

Couldn't this just be due to it being the weekend?

I suppose it is possible that states had incomplete death records because of the weekend and this could certainly have contributed or accounted for what I found in the data. I have state-specific data and didn't look carefully at this to make sure all were updated. This may be validated if we have a huge jump upwards today. On the contrary, previous weekends did not show this trend.

 

[Napoleon1801]

So I have access to all the data for COVID-19 and have been tracking it daily since 3/11. I think that we are hitting the start of the flattening of at least the death curve.

I focused on following deaths because we don't know the prevalence of the disease and it seemed to be the only reliable metric. Now, one issue with deaths is that it is reflecting transmissions that occurred ~12-16 days ago, since the average time before symptoms post contraction of the disease is ~4-5 days and the few studies published so far show that approx. patient course from admission to ICU to death is about 10-14 days. So looking at deaths today means understanding what happened in people contracting COVID about 2 weeks ago (on average).

From 3/11 to 3/26 there was a general trend where the doubling of deaths was speeding up from 4 to every 3 days. This started to slow on 3/27, and between then and 4/1 it was 4 days. for 4/2 it was 5 days. Assuming that the death rates for the disease are stable (and there would be no reason to doubt that as there are no specific new therapies to treat), this means that transmissions are decreasing (and subsequent deaths in the minority of patients who get sick enough to go to the ICU).

Following a similar trend, the number of daily deaths grew at a very alarming rate between 3/11 and 4/4. While there were days that deaths were less than the day before, this was likely due to stochastic events and dealing with small numbers. This trend never repeated and was rare. For example, between 3/13 and 4/4 there were two days where the deaths were lower the the preceding day, and in both cases these death numbers were preceded or followed by a large jump in the total deaths- these were on 3/21 (53 deaths, with 59 deaths the prior day, followed by 126 deaths), and 3/23 (73 deaths, following the large jump on 3/22 of 126 deaths and followed on 3/24 with 204 deaths). However, the new deaths total for 4/4 was 1352 and has been decreasing for 2 continuous days now (1184 in 4/5 and 1182 4/6). It's still early, but given a large step downward and the start of a possible trend is hopeful.

While it may be desirable to track hospitalizations, these data have been hard to track since they are not very complete and have only been available since 3/21. Although flawed, these numbers also follow this trend, as COVID+ hospitalizations peaked on 4/1 with 4482 new hospitalizations and has been variable since between 1488 and 3796 admissions.

Lastly, although even more flawed due to a lack of tests or strategy for testing, the new positive cases also follows this trend, with 33,840 cases on 4/4 and a decrease since. Like deaths, this number has steadily increased daily with rare exception except for a single day (3/24). However, I don't think this turnaround can be attributed to testing availability. One metric that has only trended up every day is the proportion of cases that test positive (positive test rate). This started at 10.0% on 3/16 and slowly crept up incrementally every day until 4/3 when it hit 19.3%. Since then that number has also come down slightly and stabilized at 18.8% for the last 3 days.

My take from this is that this might be early signs that whatever we were doing as a country starting about 2 weeks ago is likely starting to put a dent in spread of COVID-19.

Good news for once.

 

[gbwillner]

So maybe as BoneWizard predicted, there was a spike today in deaths and the start of a trend downward could itself have been a stochastic event (there were 1941 deaths today- an all time high by almost 600 deaths). HOWEVER, there were a few other interesting observations today: although % positive rate came back up to 19.1%, this is still near/below the ATH of 4/3 and is the 4th day of at least being relatively stable. Total positive cases is again stable/slightly lower than ATH of 4/4. Even though there was a large gap day-to-day for deaths today compared to yesterday, we are still showing a slowing down of death doubling times.

 

[PhillyMed777]

I hope you are right about everything.
Please keep us updated with more information.

 

[yaah]

It's difficult - at what point do increasing deaths become difficult to measure? Are we including people dying at home? How long does measurement of deaths lag actual occurrence? Does seem to be slowing, or at very least undershooting the anticipated curve. But you need several days or a week worth of data I think to see anything real.

 

[Med Director New England]

It's difficult - at what point do increasing deaths become difficult to measure? Are we including people dying at home? How long does measurement of deaths lag actual occurrence? Does seem to be slowing, or at very least undershooting the anticipated curve. But you need several days or a week worth of data I think to see anything real.

??
Death is the easiest metric to follow. Death dates are known, death causes are documented; prompt reporting is required in every state and the information is easy to obtain, death / unit time easy to measure, there is high awareness of Covid and relatively easy to diagnose now. Maybe an occasional case of Covid related death mis-classified but that would have been happening all along and more likely to happen earlier on, thus would not impact trending.

what is hard about this ?

 

[LADoc00]

Infections in Wuhan now kicking up because people are finally moving around. As much as I hate to say it we are in some dystopian nightmare scenario. I dont even want to look at the occasional slide anymore, I just want to get drunk when I wake up.

Dear Lord, this must be what people trapped in ghetto feel like every day of their lives..I have never had more sympathy for the downtrodden than right now.

 

[CDX-2]

Infections in Wuhan now kicking up because people are finally moving around. As much as I hate to say it we are in some dystopian nightmare scenario. I dont even want to look at the occasional slide anymore, I just want to get drunk when I wake up.

Dear Lord, this must be what people trapped in ghetto feel like every day of their lives..I have never had more sympathy for the downtrodden than right now.

But the new infections are apparently all related to foreigners and overseas travel (as per the ever so truthful Chinese Government)!

 

[yaah]

??
Death is the easiest metric to follow. Death dates are known, death causes are documented; prompt reporting is required in every state and the information is easy to obtain, death / unit time easy to measure, there is high awareness of Covid and relatively easy to diagnose now. Maybe an occasional case of Covid related death mis-classified but that would have been happening all along and more likely to happen earlier on, thus would not impact trending.

what is hard about this ?

Because a lot of people are dying at home or in nursing homes without any testing. It may or may not be more than an "occasional" misclassification of death. In fact, it may be more likely to be happening now in places with high hospital loads where there is limited capacity.

There are data from various locales that the death rate is higher than normal even excluding confirmed COVID cases.

I mean, I'm not crazy. This is in the media too (article which states "massive underreporting") New Yorkers dying at home not counted in COVID-19 death statistics

I know it's a current talking point on the right that deaths are being overcounted as COVID, but the truth is probably the opposite. The degree to which this is true is uncertain though. Is it a lot or a little? What percent? I tend to think it's a smaller percentage, but that's just guessing.

 

[gbwillner]

DATA UPDATE
Ok, I now have had an additional week of data to work with. Beyond tracking deaths and death doubling times (which admittedly was a very limited stat), I started tracking: daily deaths and rate of change of daily death rates (to better approximate acceleration of deaths), and new positive cases and the rate of change (acceleration) of new positive cases. rate of change I think is a good indicator of when things are slowing down or speeding up. I also tracked net patient flow (new admissions vs. discharges) and positive test rate of overall tests performed. I have this all plotted out but for brevity I will post my findings below.

Given the current restrictions of movements:

• New positive cases grew exponentially (linear on a log(2) plot) until about 3/27, then started to flatten. Positive tests were initially a poor metric since testing was itself an issue and not reliable until at least 3/17 when ~25k daily tests were being performed.
• As mentioned before, the positive test rate grew steadily from 3/17 at 10% until 4/3 where it got to ~19%. It has hovered there ever since.
• Daily tests performed grew quickly from less than 10k before 3/16 to 100K by 3/26. This is likely why data such as positive test rate is all over the place prior to 3/17 but very consistent thereafter.
• Daily deaths grew exponentially until about 4/4, then started to flatten. Looking at the delta of the daily deaths by assigning a polynomial treadline to the data shows that the change grew in a positive direction steadily until a peak on 4/4-4/5, then started to quickly decrease. It crossed zero (decreased daily deaths) on 4/10
• tracking rate of change for new cases and deaths was difficult because there IS an apparent lag in some reporting, particularly over weekends which can move the data. However, assigning polynomial treadlines to the data helps us assess the findings (as used above). They are:
  • rate of change of daily deaths (delta death/daily deaths) grew rapidly in a positive direction until 3/25. This signifies the peak of the growth phase of new deaths. This is the best and more specific correlate to death doubling times. There was a second "bump" positively around 4/9, but this may be due to the significant increase in cases on a single day. Regardless, it crosses 0 on 4/12 and begins to accelerate in a negative direction. This may still change with new data.
  • rate of change of daily new cases is tracked similarly. the caveat is that mew case information as stated is not reliable prior to 3/17. However, it is clear that the acceleration of new cases is already decreasing as these data become reliable. Initially there is a very rapid deceleration from 3/13-3/21. then this deceleration becomes flat/ slowly decreasing from 3/22 to about 4/1. then decelerates again to crossing into negative acceleration (decreasing cases) on 4/6. this treadline seems to then stabilize and then approach zero again.

My takeaways from this (given current restrictions)

• Social distancing had a significant affect on the spread of this disease, showing rapid deceleration of new cases from the time these data had been available and were likely an accurate reflection. This resulted in subsequent deceleration of the daily death rate about 10 days later, and showed a similar effect in the negative acceleration wherein the death data lagged the positive case data as expected. It is possible these phenomena are unrelated to social distancing, but since this was the only method used and not any specific treatment, it seems quite likely that this lead to a deceleration of the spread of disease and subsequent deaths.
• The "peak" for new cases was likely 4/3-4/5 with about 32.7K new daily cases. New cases then flattened and the started trend negative.
• The "peak" for daily deaths was likely 4/7- 4/11 with about 2050 new deaths. These similarly flattened until 4/12 when they apparently begain a downward trend (though this is very recent and can still change).
• COVID-19 patient flow grew at a higher than exponential rate until about 3/26, likely due to a lack of adequate detection in existing patients. It then stabilized at about 2000 new patients per day and took a significant downturn on 4/12. I believe this is due to the weekend effect where patients were either not discharged until Monday or a delay in reporting weekend discharges. This will become clearer with the next few days' worth of data.
• The positive test rate, given the population that is tested- those with respiratory symptoms coming into the hospital- is likely to truly be about 20% of all patients currently. Given that this number has not really decreased may mean that we are not really treading downwards with new cases and that is a recent data bias (and we are only STABLE), or that clinicians are simply better at detecting COVID-19 specific symptoms and selectively testing patients more likely to be positive, or that total test numbers are coming down plus the preceding account, meaning we are being more selective about whom to test but the trend is downwards. The total tests performed plateaued at about 150K daily tests but has trended slightly down since 4/10.

That's all I got for now.

 

[Med Director New England]

Thanks for the update and gathering the info.

so now the million dollar question - when will it be safe to relax social distancing and how best to do it.

my doctor friends are beginning to clamor for re-opening offices for primary care and preventative carewith more stringent standards for hygiene, patient screening & continuing telemedicine for patients that are in the high risk groups based on age / co-morbidities. At some point folks with chronic or undiagnosed conditions worsening will be worse than risk of possible Covid exposure.
Also a bit ironic that one can go to the liquor store currently but not their primary care doctor !?

 

[WEBB PINKERTON]

The liquor store exception is bizarre. Last time I checked you can buy liquor at CVS and walgreens.

#FireFauci

 

[gbwillner]

The liquor store exception is bizarre. Last time I checked you can buy liquor at CVS and walgreens.

#FireFauci

This is heavily state dependent. In some states grocery stores and drug stores can sell liquor, in others they require a specific license and you can only get liquor at liquor stores.

 

[Allegri]

This is all US data, not worldwide?

 

[gbwillner]

This is all US data, not worldwide?

US only

 

[deleted314957]

This is heavily state dependent. In some states grocery stores and drug stores can sell liquor, in others they require a specific license and you can only get liquor at liquor stores.

Hey, y’all don’t mess with our corn likker in Florida!


Sent from my iPad using Tapatalk

 

[gbwillner]

Hey, y’all don’t mess with our corn likker in Florida!


Sent from my iPad using Tapatalk

in TX you can't buy alcohol on Sundays before noon. I'm sure it's to keep people from public intoxication at church.

 

[Thr0mbus]

The last thing we need right now is for all the alcoholics to dry up.

 

[yaah]

Thanks for the update and gathering the info.

so now the million dollar question - when will it be safe to relax social distancing and how best to do it.

my doctor friends are beginning to clamor for re-opening offices for primary care and preventative carewith more stringent standards for hygiene, patient screening & continuing telemedicine for patients that are in the high risk groups based on age / co-morbidities. At some point folks with chronic or undiagnosed conditions worsening will be worse than risk of possible Covid exposure.
Also a bit ironic that one can go to the liquor store currently but not their primary care doctor !?

Seconded - great post, GBWillner!

I think part of the issue around relaxing social distancing is so many people have a different definition of what it means, from going out a bit more to full-on open all the restaurants and clubs and disneyworld. A lot of the less affected population centers in states with significant social distancing measures (like Ohio and parts of michigan and new york) are not overwhelmed.

As soon as hospitals and clinics are satisfied they are not at risk for getting to excessive capacity they should probably start to get back to treating non-covid patients - there has to be a huge backlog and the longer it waits the more difficult decisions everyone has to make. We can still do this while maintaining precautions. You can also still maintain a decent telemedicine presence for a lot of visits or follow ups.

The other question is at what point do you have to resinstitute social distancing and will people actually comply given that they may not see the consequences.

 

[ScubaV]

I work in a state that's part of the 7 NE states that agreed to coordinate their economic response. I'm currently working every other week, having to use PTO or be unpaid the other half. Our system has not been overwhelmed and to my knowledge has not even approached 100% capacity. I'm hoping we can restart elective procedures soon since there's a lot of ways to manage social distancing in a controlled setting, but some of that depends on the state and is not just a system decision.

In a broader context I hope we can relax some of the strict lockdown measures soon, but there's no way things can just flip back to normal until there's a vaccine. In the months to come, I know I'll still be avoiding people as much as possible and maintaining more thorough hand washing and cleaning of potential fomites like grocery store carts.

 

[deleted915203]

Yesterday daily deaths in the US hit a record high. People seem to act like things have peaked... but not in terms of deaths.

Daily deaths is the stat that I look at. Is there a better stat?

 

[BoneWizard]

I think it'd be reasonable to think that peak deaths would trail peak number of cases by a few days.

 

[Dro133]

Yesterday daily deaths in the US hit a record high. People seem to act like things have peaked... but not in terms of deaths.

Daily deaths is the stat that I look at. Is there a better stat?

That was largely a result of an additional 3,700 deaths that occurred since March 11th of people who were presumed to have died from Covid but had not officially tested positive.

 

[gbwillner]

I noticed that everytime I post a conclusion that day data come out to contradict it. Good to see there may have been an alternative reason for those deaths. Even with the ATH yesterday, it was not significantly more deaths than a few days ago at the peak.

 

[deleted915203]

 

[Med Director New England]

I noticed that everytime I post a conclusion that day data come out to contradict it. Good to see there may have been an alternative reason for those deaths. Even with the ATH yesterday, it was not significantly more deaths than a few days ago at the peak.

I think the spike in deaths is related to now including probable Covid deaths In the count, as others have mentioned.

Too bad this was changed mid crisis - The trend now is more difficult to follow.

one thing is certain the need for ICU beds and vents seems to have been over estimated (thank goodness). Tons of surge capacity in and around Boston barely or not being used at all.

 

[gbwillner]

I think the spike in deaths is related to now including probable Covid deaths In the count, as others have mentioned.

Too bad this was changed mid crisis - The trend now is more difficult to follow.

one thing is certain the need for ICU beds and vents seems to have been over estimated (thank goodness). Tons of surge capacity in and around Boston barely or not being used at all.

This may simply be due to cities like Boston and Houston, as major hubs for tertiary care, having huge excess capacity. I'm more worried for places like New Orleans, Atlanta, Washington, etc..

 

[LADoc00]

Just marathoned through 12 hours of teleconferences with public health, ID experts, pathologists, primary care docs...ouch
BUT one thing that has been hammered home is that there is a now a real fear is that patients stay away completely from healthcare for YEARS.

Entire specialties in medicine maybe in the short term at least totally done for...

 

[ScubaV]

Just marathoned through 12 hours of teleconferences with public health, ID experts, pathologists, primary care docs...ouch
BUT one thing that has been hammered home is that there is a now a real fear is that patients stay away completely from healthcare for YEARS.

Entire specialties in medicine maybe in the short term at least totally done for...

Did they give any data to support this fear? Certainly the economic aftershocks will mean less people can AFFORD healthcare in the short to medium term, but I'm not sure how there would be a fundamental change in people's desire for care.

 

[Med Director New England]

Just marathoned through 12 hours of teleconferences with public health, ID experts, pathologists, primary care docs...ouch
BUT one thing that has been hammered home is that there is a now a real fear is that patients stay away completely from healthcare for YEARS.

Entire specialties in medicine maybe in the short term at least totally done for...

Imagine the epiphany many ER docs are having now. They are always the first to tell you that 80% of what walks thru their door never should have been an ER visit, rather a PCP visit or urgent care if after hours. But they are happily seen in the ED, billed at a very high rate.

Now many of these patients that are in that 80% group have actually stopped coming b/c in general they are scared to visit the ED. if this habit persists for months or years EDs will be over staffed for some time.

public health wise and from a stewardship of medical resources perspective this is a good thing - but EDs have been built and staffed for years based on these visits. These ED docs / staff will need to pivot quickly. Maybe more migrate to urgent care setting ???

 

[Spikebd]

Imagine the epiphany many ER docs are having now. They are always the first to tell you that 80% of what walks thru their door never should have been an ER visit, rather a PCP visit or urgent care if after hours. But they are happily seen in the ED, billed at a very high rate.

Now many of these patients that are in that 80% group have actually stopped coming b/c in general they are scared to visit the ED. if this habit persists for months or years EDs will be over staffed for some time.

public health wise and from a stewardship of medical resources perspective this is a good thing - but EDs have been built and staffed for years based on these visits. These ED docs / staff will need to pivot quickly. Maybe more migrate to urgent care setting ???

Which specialties besides EM will be most devastated in the post-pandemic new world order? Won't people still want their plastics procedures, knee replacements, etc?
Major cancer resections can't really wait. Not sure how the GI cancer/Gyn Onc/Breast/ENT/Neurosurgery/GU resection volume can really drop.. those cancers will continue to happen. Same thing with hemepath. I'd imagine screening GI stuff picks back up. Maybe less derm biopsies? IDK.
I just don't know which specialties could be devastated as LADoc is referring to, aside from EM.

 

[LADoc00]

EVERYONE will be bad touched by this. Everyone. Even Psych.

If I had to predict who is getting the biggest beat stick I would guess Ortho, sadly. I have some solid buddies in that field but I think their volumes and income will be hit the hardest.

After that likely Plastics. ENT, GI, Urology, PATH, Radiology, Anesth....even Peds.

Good time to be at VA folks.

 

[WEBB PINKERTON]

I guess the surgeon owned surgery center in town will be out of business post covid? Will the few independents out there now run to hospital employment? My niece the psych said things are booming for her. I don't see psych being hurt too bad.

I agree it will be years till things get back to normal. Everyone that thinks a switch will be flipped and things will be hopping again are in for a rude awakening.

 

[octopusprime]

Regionally our ortho, GI, ENT, GU, Plastics and anesthesia groups are down 85-95%. And the ones that have big outpatient surgery centers where they just scope people all day and have loads of staff are getting completely and utterly destroyed. Plenty of these large are already bankrupt, their accounts are entirely empty, the only thing keeping their doors open is the PPP and CMS grants which--combined with the 5-15% of normal volume--is generating just enough to prevent them from laying off all their staff/RNs/midlevels...it's going to get ugly.

 

[LADoc00]

Molecular testing maybe, maybe 70% sensitive with NP swabs.

Insufficient test kits.

Seroconversion rate of maybe 60%.

NO TESTING OPTION LOOKS GOOD FOLKS, yet politicians and financial types keep saying "testing testing testing" like that is the magic bullet;.

Is this like the Death of Stalin scene where the doctors just stand around staring at each because no one wants to give the authorities the bad news that indeed Stalin is dead or what??

Did anyone do the ASCP town hall on a "universal testing protocol" which basically amounted to this: "this sucks, this sucks and this sucks and all in all no one has any test kits anyway."

 

[octopusprime]

I find this interesting:

Sweden resisted a lockdown, and its capital Stockholm is expected to reach 'herd immunity' in weeks

Sweden went against the grain by keeping public life as unrestricted as possible when the coronavirus hit. Now, it says its strategy appears to be working.

www.cnbc.com

Unrelated note, found this emerging Swedish company that looks promising: Sojaludd Gron.

 

[ccous]

I find this interesting:

Sweden resisted a lockdown, and its capital Stockholm is expected to reach 'herd immunity' in weeks

Sweden went against the grain by keeping public life as unrestricted as possible when the coronavirus hit. Now, it says its strategy appears to be working.

www.cnbc.com

Unrelated note, found this emerging Swedish company that looks promising: Sojaludd Gron.

Our major cities were also probably well on their way to this considering the virus had to have been spreading across the country throughout Jan, Feb, and before the first stay home mandates in mid March. Community spread slowly ramped up as it does during the lag phase of the exponential curve. By the time we started testing surely we were already a ways up the log phase. We had to slow it down obviously since hospitals finally started to get slammed. But as is being discussed we need to start uncovering the real prevalence.

 

[WEBB PINKERTON]

Apparently we learned NOTHING from the Theranos debacle. Where are CAP and ASCP right now? It is turning into the wild west and our leaders are silent.

 

[LADoc00]

The future of successful coronavirus response: Mass testing at work and in church and self-administered tests

Has anyone clued them onto the fact THIS IS IMPOSSIBLE:
researchers are calling for up to 20 million tests a day in the U.S. to safely reopen the economy


Just think about this at a practical level: you would have to have folks standing by to take specimens, process them and turn a result around almost instantly at every worksite, mall, theater and church.

You would stand a better chance of just taking everyone who is negative now into a massive orbiting space station to live forever.

This is all Sci Fi, are these idiots imagining we will have some type of wearable device constantly monitoring for the virus?? That IS Theranos. Literally the initial idea was that insane.

I give up.

 

[gbwillner]

UPDATE: May the 4th be with you....

It's been a few weeks since my last update here, because it is a lot of work to keep up with the data and because I didn't think I had much more to add. However, after a few weeks and almost 30k deaths later I thought it was time for a reality check.

• Death doubling times. Initially I used this metric to measure how quickly the virus was spreading, given our testing and tracking were very incomplete and the only reliable metric was deaths. As long as the death doubling times were constant, then the spread (about 8-14 days prior to the measurement) were growing exponentially. As these numbers increased, the spread was slowing down. The problem with this metric was that you are limited to the reported deaths in a day. Anyway, As I mentioned from the beginning, death doubling times actually increased from doubling every 4 days (3/11 to 3/15) to 3 (3/16 to 3/26). Then there was an initial slight slow back to 4 days from 3/27-4/1. This suggested that the peak of spread and growth of the pandemic was about 10 days before these times, or really early to mid-March. At that point social distancing kicked and and really took the sails out of the spread. On 4/3-4 it took 5 days to double current deaths; 4/5-4/6 it was 6 days. After that there was a daily slow-down with the exception of 4/11 (around the time they changed how deaths were counted) and 4/15 (wherein it took 14 days to double deaths, as the day before). We still have not doubled the deaths of 4/17, but it will probably be tomorrow, where it would have been 17 days.
• Log plot new positive cases. Plotting daily deaths and new cases since 3/11 supports the above. There is a linear growth of new cases until about 3/23, then the curve bends and grows slowly. The peak of new daily cases flattens and remains steady (the top of the flattened peak) on about 4/4. Several times the numbers trend down only to shoot up with a a single or few data points. Using a 4-day moving average (the approx. time to symptoms) shows that there has been a downtrend in new cases since 4/25. It is possible the cycling seen (downtrend followed by positive spike) is due to under-reporting on the weekends, but it is not super clear.
• Log plot new daily deaths. Similar to new cases and as expected, new daily deaths appear to grow at an exponential rate until about 3/31 and then start to slow. This is about a week or so after the new cases, which makes sense. The daily deaths continue to grow at a slower and slower pace until about 4/12, then they reach a steady state. Like new cases, downturns start and then invalidated by single or few points that seem to oscillate wider than with new cases. Over the past 5 days, daily deaths have dropped 50%, but using the same 4 day moving average, this does not signify a reliable and significant downtrend (yet). However, other factors lead one to believe we are at the back of the flattened curve here as well.
• Daily net Patient flow. Tracking patients admitted for COVID-19 and those discharged can create a picture if the problem is getting worse or getting better (i.e., more patients are being discharged). Since we are assuming that the problem will get worse, then get better, then maybe worse again, I plotted net patient daily flow (numbers + or- from the previous day) and used an order 3 polynomial trendline. These data are fairly stochastic and difficult to review without the trendline. The peak net growth of new admissions was 3/31-4/4. Growth slowed after that and turned negative 4/24. The negative patient flow may be slowing as well since 5/2.
• Positive test rate. As previously mentioned, positive tests grow rapidly once testing was reliably established after 3/17, approaching but never quite reaching 20% (19.67%). Between 4/3 and 4/21 we were testing more than 150K patients per day. Since then we are testing about 250K per day, and since 4/21 the positive test rate has started to drop fairly dramatically and consistently, now standing about about 16.1%. These data seem small, but remember we are testing 250K samples per day so the changes are uniform and likely substantial.
• Rate of change of new cases. A better way to assess if things are getting better or worse for new cases and deaths was to measure the daily changes in cases and deaths as a proportion of the cases of the prior day. Think of this as acceleration. Speed is the number of cases every day- if things are accelerating they are getting worse, and if they are decelerating they are getting better, even if there are more new cases every day. These data were again very stochastic, so I used a 3 order polynomial trendline to track changes. These show deceleration of new cases that crosses zero and turns negative on about 4/8. There is then stability and the new cases accelerate negatively starting about 4/20.
• Rate of change of deaths. As above, the same plot with deaths shows a steady deceleration until ~4/17 when it crosses zero. There is then steady negative acceleration until current time.

It's not clear what will happen with restrictions being lifted, but I assume we will see a reversal of these general trends.

Maybe I'll post the plots this time if there is demand. Enjoy.

 

[deleted314957]

Thank you for all the thoughtful time and effort!


Sent from my iPad using Tapatalk

 

[gbwillner]

Some good news.....

Today saw the highest total tests performed AND the fewest positive cases recorded since 3/26. The positive test rate continues to decrease and is now 14.3% of all tests.

 

[gbwillner]

We also saw the fewest deaths recorded nationally since 3/31, and had less than 1,000 deaths for the second consecutive day.