April 30 - COVID-19 in Westchester County, NY

This is my 60th day of tracking the daily COVID-19 stats in Westchester County. Much of the day-to-day changes now are just noise.  The data with the least day-to-day noise is the Percent Testing Positive.  This number continues almost a linear drop from its peak 5 weeks ago. 

The Daily Deaths in Westchester increased from yesterday (orange line), but was not statistically different.

The number of New Cases in Bergamo, Italy has been between 35 and 110 since April 6.  This is a very long tail (yellow line).  It is what we should expect in Westchester where the epidemic has been similar.  The virus is very contagious, so it is expected to spread at a finite rate in spite of physical distancing.  As testing becomes more widely available, it is expected that a greater portion of positive cases are mild and do not require hospitalization.  This is true for Westchester as well as Bergamo.

Therapeutics

Preliminary results for the antiviral, remdesivir, show similar efficacy in treating COVID-19 as TAMIFLU does in treating the regular flu.  It reduces the duration of the disease by about 30% and reduces mortality. It is not a cure, but will reduce hospital stays and save lives.

Pfizer and BioNTech started clinical trials of an mRNA vaccine last week in Germany.  The US trial is to start next week. If any company is capable of bringing a complicated vaccine to the mass market, it is Pfizer.  It is one part of their manufacturing they have not moved to China.

April 29 - COVID-19 in Westchester County, NY

We are 43 days into the "Slow the Spread" initiative, 50 days since the containment zone in New Rochelle, 57 days (Mar 3) since the first case in New Rochelle (Westchester) was confirmed, and 67 days since that first case had symptoms (Feb 22). If you are like me, it is starting to get old.  I no longer watch Cuomo at 11:30 or the White House Briefings at dinner.

Just the same, I would like to know what is going on.  Something about virus transmission and lethality does not make sense.  It has become evident that the virus is very widespread in Westchester and less lethal than originally thought.  How and when did it spread so fast? Was it via fomite or aerosol? Were there tens of thousands with the virus circulating prior to the first case being detected? And why was the transmission rate so high even after social distancing?

New Cases in the NYC Metro area (including Westchester) seem to be in more than a linear decay. All of these areas show a similar trend in the last four days.

Neither a Gaussian shape nor a linear decay do a good job of matching the trend of the epidemic.

Deaths in Westchester appear to be declining.  Fitting the data to a Gaussian curve seems appropriate.The early death reporting (spikes) was gathered from press briefings prior to NYS making the data publicly available on a daily basis.  The early data is presumed to be both under-reported (late) and accumulated over several days (instead of a single day).

The IHME model also assumes a Gaussian shape for the Death Curve.  They have found that slowing the spread widens the curve, but does not reduce the area under the curve.  In other words, the same number of people die, they just don't die all at once and flood the hospitals. To account for that recent finding, IHME has changed the models (which are very dependent on recent actual data) so they predict MORE deaths than prior models.  Here is the latest model for NYS.  It predicts that by May 8, deaths will fall below 60 per day (~2% of total deaths). This is the same date as the April 22 model.  On April 1, their model had 95% confidence that the death rate would be 0-50 on May 1!

Reopening Westchester

The Westchester County Executive has not announced any plans for reopening.  He did mention that he plans to go ahead with Bike Sunday on May 3, but with face-covering required.  Bike Sunday is a 6.5 mile shutdown of the Bronx River Parkway for four hours on Sundays in May, June, and September.

His hands are tied because the reopening of Westchester must be done in conjuction with NYC and other NYC metro areas.

Here are the latest NYS requirements for reopening.

April 28 - COVID-19 in Westchester County, NY

The most important data to look at when judging if it is time to loosen restrictions on physical distancing is the hospitalization usage.  This is not available for Westchester County.  The verbal reports state that it is dropping, just like the state utilization shown below. Westchester hospitals (despite being one of the hardest hit areas in the world .... 3% confirmed cases) were never more than 35% full.

All other metrics continue to improve.  Deaths were at their lowest number since April 5 (orange line).

New Cases were at their lowest since March 21 (red line).

All of the NYC Metro area continues its slow burn out of the disease.  Despite over 6 weeks of enforced social distancing, over 20% of tests are positive.  The virus has shown how contagious it is.  The latest study shows that 25% of NYC Metro area have antibodies.

Transmission Risk Scoring of activities by Johns Hopkins - Bloomberg School of Public Health

April 27 - COVID-19 in Westchester County, NY

The COVID-19 epidemic in Westchester County is slowly burning out.

The New Case Count in Westchester does not have the predicted Bell Curve shape.  Instead it has a very long tail.  The decay in New Cases over the last four weeks can be approximated by a straight line (orange line).  Will this be the trajectory for the next few weeks? The New Case Count in Bergamo Italy has leveled at about 50 cases per day for the last 2 weeks. [Bergamo Province is about the same size as Westchester]. This would be the equivalent of Westchester leveling off at about 150 cases per day by next week.

The testing level in Westchester has been rather constant as seen in the blue line below.  The number of positive cases (red line) is definitely correlated to the number of tests.  The more you test, the more positives you detect.  The testing is not a random sample of the population....supposedly it is people with severe symptoms, healthcare professionals, and of course the rich and famous (a few too many of which we have here in Westchester.)  But maybe it is not far from a random sample?  And why has no random sample study been done? If the Percent Testing Positive is constant regardless of the sample size, that would indicate that maybe the sample does represent Westchester at-large? That would mean about 1 in 4 persons in Westchester currently carry the virus?

The trends of 'Percent Testing Positive' for Westchester, NYC, and Nassau County (borders NYC on the east) are very much aligned.  This data has not been filtered or smoothed, but is less 'noisy' than all of the other data. If a growing percentage of the population is resolving from the virus (and exhibiting antibodies to it), that would make the Percent-Testing-Positive drop because they are supposedly immune from the virus. Last's week's report that 20% of the NYC Metro area have antibodies almost perfectly agrees with the ~20% decline in the Percent Testing Positive.  Is that just a coindidence? If we project the trend below into the future, Westchester will be at about 15% in a week which is not too far from the national average.

Deaths in Westchester continue to be flat at around 30 per day. See orange line in the middle chart above. This has been the case for about 3 weeks.  Interestingly, the normal number of deaths per day in Westchester is about 30?  Are almost all deaths in Westchester being categorized as COVID related?

Model your own epidemic

If you have cabin fever, you can create your own model for the epidemic here.  You will probably do just as well as the experts.

April 26 - COVID-19 in Westchester County, NY

The usual charts are further down the page.

Masks
A small study showed that masks did not reduce viral spread from a cough.  Apparently the force of coughed air moves aerosol around the mask instead of depositing it on the mask. The study did not look at whether a mask protects wearers from coughs by others.  It also did not test N95 masks.

Why 'Test & Trace' won't work in NYC Metro area. 
New York Governor Cuomo has stipulated the implementation of 'Test & Trace' as a condition prior to reopening New York.  However, the numbers show that is a fool's errand.

Antibody studies of the NYC Metro show that about 20% of the population have been exposed and recovered from the virus. However, confirmed cases only make up 2% of the population. This data suggests 9 of 10 with the virus have no symptoms or have such mild symptoms that they are not tested.  Of the 18% asymptomatic population (9 of 10 with antibodies), about 80% of them stayed-at-home for the last 5 weeks.  Assuming they were in an average household of three and they took no precautions to protect others in the household because they had no symptoms, those household members are undoubtedly infected.  That computes to another 36% who are infected and who presumably would have the antibodies in another few weeks.

The 20% of asymptomatic people who could not stay-at-home for various reasons, have been interacting with the public for five weeks. With all of the precautions in place, let's assume that the reproduction factor is only 0.5 (only one in two infectious persons transmits the virus to one other person).  That is less than the apparent reproduction rate of 0.9 (see NYS data).  In other words, I am saying that the real people transmitting the virus are not the people working, but people in their homes, transmitting it to other household members.  That matches case data from China.  If the serial interval is five days (see DHS data), then three weeks from now this group will have infected a number equal to themselves.  It is a relatively small number (4% of total population).  If you add up all of these numbers, you have 20% (antibody tested positive) + 36% (infected by asymptomatic persons in the household) + 4% (infected by asymptomatic workers) = 60%!!

This analysis is supported by the virus test results.  No matter what the testing volume, the percent positive in the NYC metro area is about 30% right now.  But we know 20% have the antibody.  That means 50% either have the virus or have the antibody (it is possible, but unlikely to have the IgG antibody and the virus at the same time).  That is very close to the 60% estimate above.

The bad news is that if the virus is that contagious and that wide spread, you will never be able to test and trace all of the people exposed in a population of 20 million.

The good news (if this analysis is correct) is that the virus has just about burned out in the NYC Metro area. Assuming 20% are naturally immune for some reason or another (drugs they are taking, genetics, blood type????), then only another 20% of the people in the NYC Metro area are yet to be infected.  If we continue the social distancing, it will drag out the time needed for the last 20% to be exposed.  For Westchester, I predict that will be another 13 weeks! (Aug 1). Westchester is testing about 1% of the population every week. In 13 weeks, 24% of the population will have been tested.  At the current rate, about 8% of the population will have been confirmed positive (currently about 3% are confirmed positive).  If the real rate is 10 times the number of confirmed, then 80% will have been infected at that point, which is eveyone who could be infected.

It should be obvious by now that if the virus is that wide spread in the NYC metro area, that cutting back on social distancing will not have a drastic effect on hospitalizations. There are not that many people remaining to be infected!

Redoing the antibody test in another 2 weeks would prove or disprove this analysis.  I suppose I am patient enough to wait that long.

==========================

The 'noise' in new cases is so great that I am only showing the 7 day averages.  The noise is caused by daily changes in testing volume.  As stated above, New Cases and Number of Tests is strongly correlated. The more you test, the more you find.

Westchester deaths per day is still about 30 (orange line).

Bergamo, Italy has had a steady stream of about 50 New Cases per day.  Social distancing there is still in effect. My estimates are that this area is also about 'burnt out' with the virus.

April 25 - COVID-19 in Westchester County, NY

The epidemic in Westchester County, New York is in a very slow decline. It is so slow, that it is getting boring to track on a daily basis.  New Cases are plotted below.  Beneath it is the same data with daily-noise filtered out (essentially 7 day averages).

With the noise filtered out it appears that Westchester reached its peak first.  This aligns with suspicions that the metro epidemic started in Westchester.  Westchester had the first containment zone in the metro area.  Nassau peaked later but appears to have a faster decline than Westchester or NYC.

The percent testing positve also continues to show a slow improvement.  The US Average is about 10% testing positive.

Deaths per day in Westchester are essentially unchanged (orange line).

I have expanded my time horizon to May 3.  Originally, all models predicted that the New York epidemic would be over by May 1.  They all assumed a Gaussian shape (a bell curve).  It now appears (especially for Westchester if you look at 7-day averages) to be more like an exponential rise followed by a linear decay.

April 24 - COVID-19 in Westchester County, NY

One puzzling aspect of the epidemic in Westchester is the improvement in total hospitalizations despite only minor reductions in the New Case rate.  Since the peak of the epidemic in Westchester the number of patients hospitalized has dropped from over 1100 to less than 900 as of yesterday. The New Case Rate, which should be the leading indicator of all other metrics, remains stubbornly at about 700 per day after almost five weeks.  See blue line below.

Westchester is also stuck at about 30 deaths per day (see orange line below).  Perhaps the 'steady reduction' in total hospitalizations is only noise in another plateau?

This elongated epidemic is, after all, what we were trying to do with all of the social distancing measures.  The idea was to 'slow' the spread, not to stop the spread.  So we should not be surprised that the epidemic curve has a lower peak but is very broad. The length of the Westchester epidemic (blue line) could be 60 days.

This is different from what was reported in China.  The entire country only had a 30 day event???

Daily New Cases in China.  John Hopkins

But the elongated epidemic is exactly what is being reported out of Italy.

Daily New Cases in Italy.   worldometers

The true infection rate in Westchester is now thought to be about 12% based on recent 'random' antibody testing.  The sample size is still small, but it confirms that there are many people who are asymptomatic or where the symptoms were so mild that they did not qualify for a swab test. I am a bit surprised it was not higher based on the spread of the disease after social distancing.

Yesterday, I heard that a neighbor couple (both in the medical profession and living only two doors away) had tested positive several weeks ago and had quarantined with their two young kids. I do not know if they had symptoms, but during this period I saw him several times working in his yard.  My guess is that they were some of the few asymptomatic persons who got tested because they were exposed to patients. Other than that couple, I know of noone in Westchester testing positive, but it appears one in eight Westchesterians have been infected!

The facts

The White House Task Force Briefing featured the Head of DHS's Science and Technology Group.. Now there is a sleeper cell.  Apparently they do all kinds of R&D to protect the USA from all kinds of threats including bioterrorism. They have a lab that specializes in viral threats and can measure the half-life of the CoronaVirus at various temperatures, humidities, and UV conditions.

They have not published their methods, but the verbal description was compelling enough to make me think they know what they are doing.  If you get off on drilling down to the real scientific questions (what is known and what needs to be known), read their Master Question.

April 23 - COVID-19 in Westchester County, NY

The New Cases in metro NYC (including Westchester) increased in yesterday's reporting.  Why? More tests.  From the beginning of the epidemic the number of new cases correlated with the number of tests.  More testing, more positives.  This is why I have said from the beginning, testing is not the answer.  (If you could test all 350MM US citizens once a week, it might be valuable.)

The testing rate per capita in Westchester is 75,000 per million. It is so large it does not even fit on the chart below.

Chart from Business Insider

In mid-March, 50% of the tests in Westchester were coming back positive. The testing was doubled, then tripled. They still came back about 50% positive.  Finally the percent testing positive is starting to go down.  This could just be because 10-30% of the population has already recovered from the virus when they were tested.  After all, the virus arrived in Westchester in early February.  It has been here almost three months.

Ohio has a large testing program.  Strangely they prioritized prisoners vs. citizens.  In the Marion prison they tested all 2500 inmates and 80% tested POSITIVE!!!  Only 5% of those testing positive had symptoms. There has only been one death at Marion.

New York metro area supposedly has about 10% of the world's cases.  Why?  Testing density.

Since the start of this pandemic there have been so many incidents that point to a virus that is extremely contagious and largely asymptomatic. But nobody seems to be able to put together a random study to show the extent of those claims. That seems like something the WHO could have done in China???

The noise in the chart of New Cases in the NYC metro area is largely due to the fluctuation in number of tests.  Labs must be dumping large batches of results on certain days.  The chart for NYC which is based on the day of the swab is not as noisy.  The weekends on that chart are clearly visible (less swabbing on Saturday and Sunday?).

Compare the red line above (New Cases plotted on day results received) vs below (New Cases plotted on day patient was swabbed.)

Deaths in Westchester are stuck at 30 per day (orange line).

Bergamo, Italy (comparator city) is recovering, but still has about 50 New Cases diagnosed each day.

April 22 - COVID-19 in Westchester County, NY

We are 36 days into the National 'Slow the Spread' initiative, 43 days since the containment zone in New Rochelle, 50 days (Mar 3) since the first case in New Rochelle (Westchester) was confirmed, and 60 days since that first case had symptoms (Feb 22). It is time again to step back for some perspective.

In those 36 days, the virus erupted in New York City and then spread out from there.  There could have been other 'Patient 0s' who started clusters in Queens, Nassau, or Rockland, but it is very possible that all of this started with one person in New Rochelle or Manhattan.

Since then, 800 people have died in Westchester County (about 0.1% of the population) and they are still dying at about 30 per day. The total confirmed cases in Westchester are about 2.5% of the population.  Westchester is quite large -- 30 miles from its border with the Bronx to its northern villages where people like the Clintons, Martha Stewart, and Ralph Lauren live.  The infection is concentrated in areas with denser populations and more Hispanic and African American residents.

Map of Westchester County showing Total Cases by municipality.  Data is about 10 days old.

Although Westchester does not release data by zipcode or percapita (as other entities have been doing), these dense areas appear to have an incidence of disease of well over 3%.  For reference, in Rockland County (across the Hudson River) there are several zipcodes where the incidence is over 3%.  See the red areas below.  These areas in Rockland have a high Hasidic Jewish population. The density of households is one aspect they share with the hardest-hit areas in Westchester.

Rockland County Total Cases.
  Hudson River is shown in blue on the right.  Westchester is to the right of the River.

The good news is that Westchester has had no problem handling the patients needing hospitalization. That is surprizing considering the extent of the infection.  The explanation appears to be in the large number of beds per 10K residents in Westchester compared to the rest of the state. Westchester has 30 hospital beds per 10K compared with 7 per 10K for the rest of NYS. Hospitalizations in Westchester peaked at about 1000, far short of the 3000 beds available.  The $15MM overflow hospital built by the Army Corp of Engineers was not needed.

Of the County's 25,000 cases, over 15,000 are already resolved.

New Cases in Westchester dropped to the lowest number since March 22, almost exactly 4 weeks ago. If the last three weeks is a predictor of the future, Westchester may still be on the plateau.

The gaussian fit suggests that New Cases will not fall below 60 (~2% of total cases) until May1 at the earliest.  The IMHE model was updated and released yesterday for the entire state of New York. It predicts that by May 8, deaths will fall below 60 per day (~2% of total deaths).  Of course, IMHE has almost always been wrong (explain to me how they become more certain the farther they predict into the future? I guess they can be certain that the deaths will not fall below 0, duh.  The 95% certainty range for May 8 is 7 to 175 deaths. We can come back later and check on that.  By the way, on April 1, their model had 95% confidence that the death rate would be 0-50 on May 1!)

Westchester deaths are definitely still on a plateau of about 30 per day (orange line).  The number hospitalized is falling, but is still close to the peak of around 1000.

The NYC metro area does appear to be trending lower in New Cases.

April 21 - COVID-19 in Westchester County, NY

The duration of the Westchester County epidemic is now appearing to be much longer in duration than expected. The duration of the China epidemic was about 30 days, January 23 to February 23 (95% of new cases appeared in this time span, if you believe that. The Wuhan epidemic would be even shorter.)  The duration of the epidemic in Bergamo, Italy appears to be about 45 days. The Westchester epidemic appears to be at least 45 days.

I am not the only one to notice this.  See this Bloomberg article on the flattened but ELONGATED curve.

IMHE has not updated their models since April 17.  Supposedly new ones are coming out today.  They also have increased their estimates of epidemic durations.  Once again, the China curve-shape is not playing out here.  Shown below is IMHE's prediction of the NYS death curve from April 17.  Right below it is the curve from April 1, which predicted the end of the epidemic on about May 1.

New cases in the NYC metro area dropped below their plateau averages, but declaring a trend would be a stretch. NYC and Nassau reported the fewest new cases since March, but Westchester is still struggling to get below 500 new cases per day.

Deaths in Westchester are still on a plateau of 30 per day.

April 20 - COVID-19 in Westchester County, NY

Two critical pieces of data from Westchester's epidemic seem to contradict each other. The rate of new cases has not slowed down, but the hospitalization rate has slowed???  If the infection rate has not slowed down (as evidenced by new cases) then the number of hospitalizations should not be slowing down? (Westchester does not report hospitalization data, but both the state and NYC report decreases in hospitalization.)

New Cases per Day in the NYC Metro Area since start of epidemic

NYS Total Hospitalizations

Possible explanations for this discrepancy are that the test data for new cases is more delayed than the hospitalization data OR that fewer new cases require hospitalization. No one has explained this.

The Percent-Testing-Positive in the NYC metro area continues to drop, but it is nowhere near the national average of about 15%.

There is no change in the comparison of Bergamo, Italy with Westchester County.

The daily death count in Westchester (and in NYS) has not yet started to substantially drop.

How immune is Westchester?

Westchester has more infections per capita than any county in the US (including NYC).  Nassau county is not far behind.  [Why would these less dense places with healthier populations be harder-hit than NYC?] Almost 2.5% of the Westchester population is confirmed infected.  7.5% of Westchester has been tested for the virus.  [Compare that with 1% nationally.] 

The true measure of immunity will come after a random antibody testing study is started this week.  It is rumored to include 3000 households in NYC (which should be comparable to Westchester). In addition, Northwell Health Core Labs (Long Island) is committing to start antibody testing this week at 10,000 tests per day, increasing to over 100,000 per week.

Germany is doing a similar study of 3000 households in the most hard-hit areas.  No results yet.

One sneak peak at what this might look at comes from a study of about 200 birth mothers in NYC in the past month.  Four had symptoms of COVID and tested positive. (Note that is about the same as the 2.5% confirmed in Westchester.)  Testing of the other 200 mothers revealed another 29 positives who were asymptomic, or 15%.  Once again, this does not detect any persons that may have already recovered from the disease. It is very likely that 20% of Westchester have the antibody.  What if there is 50% of the population who is naturally immune for some reason? That would be getting close to herd immunity.

Reopening

As states start to think about reopening places where human contact is more likely, it would be useful to know which types of behaviors are the most effective at preventing transmission. Is it masks?  Is it gloves? Is it not touching your face? Is transmission by air more likely than transmission by touch? Are people using cash more likely to get infected? Shaking hands and opening doors can be avoided. But what about floors and shoes? Should I handle my shoes with care?

Most of the answers to the above question are more opinion than science...and maybe that is the best we are going to do.  UNLESS we try some big experiments.  What if everyone in Kansas went back to work wearing masks but Iowa went back without masks.  Keep everything else the same.  Then collect the data??

Yinon Weiss did an interesting analysis of death outcomes in various states based on when they implemented the stay-at-home order. (This piece of data is the same one used by IHME in their models that predicted the height of the death curve.) Weiss' analysis showed no correlation between when a state implemented the stay-at-home order and the number of deaths per capita 21 days into their epidemic. Early implementation of stay-at-home has been used to explain the low death rate in California, but it doesn't explain many other entities with low death rates that even to this date have not issued stay-at-home orders. The graph is below, but I suggest you go to the article for the full explanation.

Yinon Weiss Analysis