Quantitative analyses on the global coronavirus pandemic

Month: March 2020

Austrian and German Mortality Rates Appear Low

Austria and Germany are both suffering exponential growth of confirmed COVID-19 cases.  They are being infected as fast or faster than any other country in the world.  In fact, Austria now has a hotness index of 183 meaning that 183 per million of its population have now been confirmed with COVID-19.  Germany is close behind with 143 per million.  But their mortality rates have remained remarkably low.  The coincident mortality rates for Austria and Germany are 0.24% (with 95% confidence range of 0.09% – .65%) and 0.23% (0.16% – .35%).  But these estimates are clearly too low because as we have explained before in the exponential growth phase of a pandemic, the coincident mortality rate is always too low.  Few cases go from diagnosis to death in just one day.  Even though some cases are diagnosed after death, most take 3–10 days to resolve themselves either as cures or deaths.  The full range of days from diagnosis to death may be -2 to 21 days.  We have modeled dozens of countries’ death rates and find that the death count compared to the number of confirmed cases 3–7 days before provides the most stable forecast.  We call this the lagged5 mortality rate.   The lagged5 mortality rate for Austria is 0.8% (0.3–2.1%) and 0.8% for Germany (0.5–1.1%).  These are significantly lower than the current WHO estimate of 3.4% and rules that out as statistically improbable for Austria and Germany.  Soon we can test whether it is lower than the optimistic rate of 1.0% estimated by many scientists as the true underlying mortality rate.  We have also ruled the theory that COVID-19 is just like bad common flu.  For a long time the Germans (and Americans and Brits and others) have been willing to let the virus run amok in their country because they thought it was just like a severe common flu.  In fact, Angela Merkel did not come out and make a strong warning until March 11.  Now that the Germans know better they are finally taking drastic measures to try to curb the spread in their country and the EU.

The question to ask is why do they seem to tolerate the coronavirus better any other country.  Italy’s high 16% lagged5 mortality rate may be partially understood in terms of its high median age.  Poorer healthcare systems in Italy, Iran, and Spain could account for the high lagged mortality rate seen in those countries.  For example, Germans have 8900 hospital beds per million citizens compared to 4400 per million for Italy.  One other major factor that can affect our estimated mortality rate is testing.  The faster and more thoroughly a country test for COVID-19, the longer the time from diagnosis to case resolution.  For example, soon after South Korea identified that the coronavirus was inside their country they rolled out extensive testing countrywide.  Thus, both their coincident and lagged5 mortality rate is a low 1.0% and 1.1%, respectively.  Being able to identify the true number of infections allows them to have a good fix on the denominator of the mortality rate and requires that we use a longer lag time between diagnosis and resolution for our estimates.  If we assume that the time lag was longer than 5 days in Austria and Germany, the mortality rates would have to be adjusted higher.  We think it is too early to make a definitive call.  But we think we can safely conclude that the true COVID-19 mortality rate is not 3.4% or 0.3%, but rather something closer to 1% or lower.  Extenuating factors such as an older population and overwhelmed medical facilities can drive this rate higher than 1%. 

NYC Is Now Worse Than China Two Days After Wuhan Quarantine

China announced plans to quarantine Wuhan on January 22 when the number of confirmed COVID-19 cases was near 500 and threatened to go exponential or out of control.  The number of confirmed cases in NY state has now passed that threshold by three days and it is growing exponentially.

Most of the confirmed cases in NY, NJ, and CT are located in the metro-NYC area.  If the US wants to adopt China’s successful, albeit painful, quarantine program they must do so NOW.  Many people with the benefit of hindsight criticized China for such draconic measures while others criticized them for doing it too late.  WHO seemed to be the only major organization that had praised China for its decisive action at that time.  Now is the time for Governor Cuomo, or really President Trump to adopt a strategy similar to China’s before the virus escapes the hot zones such as metro-NYC, Seattle, and San Franciso and the pandemic gets out of control everywhere in the USA.  The entire metro-NYC area should be quarantined because so many NJ and CT residents work in NYC.  They need to be allowed to travel freely within the quarantine zone for critical business or else too many people will be tempted to violate the boundaries.  Currently, Mayor De Blasio favors a shelter-in-place policy for NYC but Governor Cuomo is downplaying it.  They, along with the governors of NJ and CT, need to adopt it now for the entire metro-NYC area before they lose control of the situation and forever regret that they acted too late.  Yes, it would be painful but we are talking about the lives of 56M citizens in the Northeast.

Domestic Travel Restrictions Could Help Fight COVID-19

Some of the current forecasts for infection rate in the US is very scary but way too pessimistic.  The dire worst-case forecasts of 200,000,000 infections in the US this year is certainly possible if nothing is done at all.  Fortunately the US has taken some steps already that will help to mitigate these disastrous scenarios:

  1. Imposed stringent international travel restrictions first on China then Iran, Italy, and South Korea and then extending it further to most of Europe all helped to stem the inflow of infection.  The US should have imposed restrictions on countries quickly and scientifically, based on their hotness, rather than politics. Moreover, now that the infection is inside the US most of these travel restrictions will have minimal impact.
  2. Declared a national emergency to deploy all necessary resources to fight this internal plague and imposed restrictions on congregation, but fell short by not deploying the National Guard to expand hospital facilities.
  3. Reducing structural handicaps such as improving the speed and efficiency of testing, tracking and treatment.  This requires paid family leave, free testing and free treatment (to cover the high percentage of under-insured patients) and tons of liquidity to ease financial strains on the system.  Should be done soon.
  4. Encouraging best practice individual behavior such as social distancing and institutional practices such as work from home and temporary business closure.
  5. Begun imposing internal traffic restrictions.

Step 5 we believe is very important and must be done as soon as possible.  The infection has spread inside the US but could be still be isolated into a handful of metro areas that must be quarantined.  If these areas cannot be effectively isolated, the total population of the US at 330M could be exposed, and the worst case could become reality.  If US citizens are allowed to move freely within the country 330M is the total possible exposure.  Cities that have been relatively unscathed such as Houston and St. Louis will become infected soon without these travel restrictions.  If the US adopts the inhumane UK philosophy of “herd immunity” at this point and allow the virus to spread widely inside, 200M US citizen could get infected.  However if the hot zones could be segregated to pockets of contagion then the total possible exposure number could be reduced to possibly below 100M, and the total infections could be reduced to 60M rather than 200M. This proposal would be similar to the Wuhan model that has worked in China. 

In China, the coronavirus infection has basically ended with just 158 infected per million citizens.  If this were possible in the US it would have meant only 52,000 total infections.  This may not be possible now because there are already 6–10 metro areas in the US that are Wuhan-hot that need to be quarantined.  Seattle, San Francisco, and New York City are already past the 400 or so confirmed case level when China isolated Wuhan.  But 90% of the cases in the US are still concentrated in major metropolitan statistical areas (MSAs) — see this very useful NY Times map.  Each of the top 50 US MSAs should view themselves as a Singapore for this fight and adopt Singapore’s best practices. One objection is that no city in the US is like Singapore, an autonomous city-state.  Imposing internal travel bans could allow each of them to act like Singapore temporarily.  Trump needs to authorize each MSA real powers, or manage this process at the national level.  Singapore saw their first case of COVID-19 on Jan 23, just a couple of days after the US.  They have managed to maintain their zero death record for 8 weeks and has limited their total infections to just 42 per million citizens.  This would translate to a total of just 14,000 infections for the US.  Singapore’s infection has not ended so the number of total infections will grow, but it seems to be under better control without the exponential growth seen in much of the US now.  If we have the will to restrict domestic travel so that each MSA can manage itself like a Singapore with full national support we can limit our losses.  Not doing this could make the dire forecasts of 100,000,000 infections real.

What Makes Scandinavia So Special?

Scandinavia, generally considered to comprise the countries of Norway, Sweden, and Denmark, contain beautiful scenery and very nice people but those are not reasons why I would want to visit them this time of the year.  After all they are all countries on Trump’s banned list.  They are all massively infected just like the rest of Europe.  However, they all have remarkably low COVID-19 death rates.  What makes them so special?
First of all let me caution that the outbreak is still early in these countries and estimating COVID-19 mortality rates is very tricky in the early phase due to exponential growth and small number statistics leading to large error uncertainties.  But it is extremely important to be able to identify early best practice countries so that others may learn from them.  For exponential growth every day that passes could have an enormous impact on final death rate.  The table below summarizes the data for these three countries and for Scandinavia as a whole.

Norway       2051,10830.3%0.9%
Denmark       143  82710.1%0.5%
Sweden       9596120.2%0.7%
Scandinavia       1362,89650.2%1.0%
95% confidence low0.1%0.5%
95% confidence high0.4%2.3%

I would say that the data is extremely intriguing but borderline statistically significant.  If we combine all three countries and analyze them as one region with larger statistics we find the coincident mortality rate is a very low 0.2% with a 95% confidence interval of 0.1% to 0.4% — consistent with common flu and way below the 3.4% estimate of the WHO.  As we noted earlier the coincident mortality rate is very likely low because the denominator overstates the actual number of cases evolving to death or recovery.  A better estimate is to assume that it takes on average 3–7 days for the disease to resolve itself from day of diagnosis.  This lagged5 rate is 1.0% with a 95% confidence range of 0.5% to 2.3% , still lower than the WHO estimate. 

If this is a real phenomenon, what might be the cause?  One possibility is that the median age in Scandinavia is about 41, substantially younger than the median age of Italy at 47.  Age might account for the difference since it is well known that age has a big impact on COVID-19 mortality.  It is also possible that the better average health of Scandinavians in terms of fewer diabetes and lower heart disease incidence also plays a strong role in keeping mortality low.  As more data is collected it might be possible to detect other differences that could help other countries manage the COVID-19 outbreak better.

Germany vs Spain: Coronavirus

Most of Europe has been experiencing a huge COVID-19 outbreak in recent weeks.  If the EU was considered as a single country it would have a population 40% more than the US.  If we take just the top 9 most infected EU countries that we track it would closely match the population of the US.  These countries already have over 30,000 confirmed cases and the US is headed there before the end of this month.  These countries also have over 1,700 reported deaths.  Hopefully the US will not match this death count soon since the aggregated number include the tragedy unfolding in Italy. 

There are no winners in this global tragedy but there are definitely countries that are suffering more than others.  Clearly Italy is a big loser with 1266 deaths to date, increasing rapidly.  There also appears to be other countries such those in Scandinavia that appear to have very low mortality rates but the count is too low to be statistically significant yet.  But there are two countries, neither bordering  Italy, whose confirmed cases are both rising rapidly but who are experiencing very different mortality rates: Spain and Germany.  Spain and Germany are both headed on parallel exponential paths toward full on contagion with Germany expecting the majority of its citizens will contract the disease before it’s over without quick, concerted intervention.  

Spain  1376.3911953.1%15.5%
Germany554,599    90.2%0.4%
95% confidence low0.1%0.2%
95% confidence high0.4%0.8%

Germany has a coincident mortality rate of 0.2% that is statistically very significantly different than Spain’s 3.1%.  Out best guess of the real underlying mortality rate (comparing to the confirmed cases an average of 5 days earlier) is 0.4% for Germany versus 15.5% for Spain, with Spain suffering 40 times worse.  The table only gives the 95% confidence levels for Germany for clarity but the confidence interval for Spain is tighter than for Germany due its higher number count.  Our best guess is that Spain will be another Italy if they don’t take drastic actions to stop the spread of the virus.

What makes them so different?  It is not median population age which is a significant contributor to the fortunes of the Scandinavian countries since Spain’s median age at 44.9 yrs is actually lower than that of Germany’s at 45.7 yrs.  It may be the speed with which each country has responded to the pandemic.  For Germany the difference may be Angela Merkel who has been taking this virus very seriously.  If politics and egos were not factors, Trump could do well to examine Germany’s approach carefully and see if anything could be learned from them.  Will the fate of the US follow Italy and Spain or will it follow Germany?  The death rate could be a difference between thousands versus tens of thousands of dead Americans.

Trump Broadens International Travel Ban

Last night, Trump finally acknowledged the seriousness of the COVID-19 outbreak in the US and put in place a necessary first step to reduce the number of deaths in the US: restricting travel from 26 countries in Europe to the US for the next 30 days.  He bungled this announcement in a number of ways, but exempting the UK appears to be purely political.  He says that the UK is doing a great job of handling the coronavirus while all the evidence points to gross mismanagement of the outbreak by Boris Johnson.  The following graph shows that the UK is still in the uncontrolled exponential phase of infection and there is no sign that it will not grow as bad as Italy or Iran.  

The current mortality rate is 1.7% and the best guess lagged5 mortality rate is 4.8% — near the global average.  The hotness measure for the UK is 8.7 per million citizens so they still have a chance to stop the carnage if they act quickly before the medical system gets overwhelmed.  For Trump to cordon off much of hot Europe but allow the entry of potentially infected UK citizens makes little sense.  He should base his decision to buffer the US on facts — not politics.  The virus does not discriminate between Brits, Americans, and Italians.

In any case, this first step will not stop the coronavirus from spreading in the US since it is already inside.  Many observers were looking for Trump to declare a national emergency and put plans into place for imposing internal quarantines like one for New Rochelle, NY.  This along with leadership regarding social distancing could do more to reduce the eventual fatality count in the US.  Instead he continues to flaunt his fabled immunity and hold massive rallies counter to the advice of doctors and scientists and nearly every other government official.  Further there were no announcements of substantial policy changes such as paid sick leave and universal free virus testing and treatment that would help to ensure that every infected patient is identified and treated.

US Hot Zones That Need to Be Contained

The number of confirmed COVID-19 cases in the US has grown to 1,698 (Mar 12, midnight GMT).  This number is much lower than the tens of thousands of cases that are probably already active in the US due to substantial under testing.  A number of areas in the US are now experiencing concentrated exponential growth of COVID-19.  These areas need to isolated immediately.  New Rochelle, NY has been “contained” but this is not enough.  Too many infections have already leaked out or begun in the rest of metro-NYC as people have traveled and congregated freely, and allowed the virus to proliferate.  NY state has 328 cases identified in just the last 12 days — much of it in metro-NYC.  All of metro-NYC should be contained, if not quarantined.  NYC, however, is not the only hot spot in the US. 

New York3280
New Jersey301

Washington state has more than 420 cases and a hot zone around Seattle that must be contained.  California with 252 cases has two hot zones around San Francisco-Silicon Valley and Los Angeles that should be contained.  Finally, Massachusetts with 108 cases has a hot zone around Boston that needs to be contained.  These five metro areas should cordoned off and controlled as soon as possible.  In addition, other infected areas around DC, Miami, Atlanta, Denver, Philadelphia, etc. should be preparing to implement congregation and travel restrictions now. 

COVID-19 Is Now Officially a Pandemic

The World Health Organization (WHO), after many weeks of dithering, has finally declared COVID-19 a pandemic.  Most scientists and healthcare professionals already had been treating it as such anyway.  Scientific advisers to Chancellor Merkel warned her that as many as 2/3 of all Germans may contract the coronavirus and she needed to attack it seriously and immediately as a major pandemic.  The US, as many had urged for weeks, is finally ramping up testing and found 313 new cases today – more than the cumulative total for weeks up to March 5th.  The stock market took another major tumble today with the Dow now down more than 20% — forecasting a recession is coming to the US.  (These forecasts are sometimes scoffed at since the Dow has forecast 9 of the past 5 recessions.)  Is the world ending?  Is there any good news?

I think the risks are many and they are far more likely than optimistic scenarios painted by Trump, but there are some signs that the world may avoid some of the near apocalyptic scenarios if governments act forcefully and quickly.

The first sign is that the Chinese contagion finally broke its fever with just 18 new cases and 11 new deaths — record low case number for China last seen Jan 15th — eight weeks ago.  These numbers and the number of active case have been trending down for more than three weeks now.  If China doesn’t suffer reinfection, it means that the US might only suffer similarly until the end of April.  Of course, this assumes that the US adopts similarly draconian measures as the Chinese — not a slam dunk given today’s political environment.  Following the Chinese containment model as the Italians appear to be doing now will set back any country’s economy severely for at least 1–2 quarters.  

The second sign is that some countries appear to have escaped any major fatalities.  For example Norway has 629 confirmed cases and hotness of 116 per million citizens and yet has suffered no deaths.  The apparent mortality of zero, although statistically significant, is very misleading since case count has exploded in recent days.  The lagged5 mortality of 0/113, that is 0 out of 113 confirmed cases 5 days ago in Norway, is also misleading because Norway has only had one resolved case so far so the lag time between disease diagnosis and resolution may be longer than 5 days especially since we suspect that they may be doing a better than average job of testing potential cases early like South Korea.  So even though Norway and its neighbor Denmark (514 case with no deaths) are hot and show zero mortality on all 3 of our measures, it is still too early to say whether they provide an optimistic model for testing, tracking, and treatment of COVID-19 that may be emulated by other countries.  There is another country, though, where optimism may be justified and that is Singapore.  It has experienced the coronavirus outbreak for more than two months already.  And yet it still has 0 fatalities out of 178 confirmed cases (coincident).  It also has 0/130 mortality on a 5-day lagged basis.  In fact it has zero mortality on a resolved basis as well with 96 recovered patients and no deaths.  This is beginning to be statistically significant and rules out some of the more apocalyptic scenarios envisioned by some.  It may make sense for researchers to look at how Singapore test, track, and treat potential victims and how they restrict travel, activity, and quarantine to protect their citizens.

COVID-19 Hotspots Around the World

When people look at COVID-19 hotspots around the world, they usually look at a list of total infections and deaths by country. These are not the best statistics to figure out where the most dangerous areas for infections are located — the deadliest manifestations of the current epidemic.

 Infections Deaths  Mortality
Italy           167101496316.2%16.4%38.6%
S. Korea          1457513580.8%0.9%19.0%
Iran           9880422913.6%8.3%9.6%
Switzerland          5849730.6%2.5%50.0%
China  578076131363.9%3.9%5.0%
Spain 361690352.1%12.4%20.6%
Singapore        2916600.0%0.0%0.0%
France           271784331.8%7.8%71.4%
U.S.            3971303.1%13.6%66.7%

On a list of most infections, China clearly tops the list with 80,761.  On a list of most deaths, China again tops the list with 3,136. But China has a huge population of citizens who can potentially become infected.  A normalized ratio of infections per million of population is a more accurate measure of how widespread the infection is in each country, how deeply the infection affects each person,  and how well each country’s government is managing the crisis within its own borders.  On such a “hotness” measure, China has already been exceeded by Italy, South Korea, Iran, and Switzerland.  In these countries, COVID-19 is generally less under control.  Travelers would do well to avoid these hotspots.  Indeed this figure could have given the Italian government government an earlier warning as to how serious their problem was and led them to close their borders sooner rather than later.  Since this outbreak is still in its dynamic stage, the hotness figure by itself has a major issue: countries like South Korea that are doing a great job testing their citizens will inevitably come up with higher figures, while countries like the U.S. that are dragging their feet will show lower figures.   As testing ramps up, the U.S. hotness will surely rise and begin to more accurately reflect how thoroughly they are caring for their citizens.   (A table of number of tests done per million population could be used to correct this problem but some countries such as the U.S. have stopped reporting this number.)

One other statistic that people are obsessed with, and rightly so, is the mortality rate.  This is the ultimate measure of how serious a disease is.  Many people have downplayed the seriousness of COVID-19 by claiming it’s just like the seasonal flu that kills hundreds of thousands each year without causing anywhere near the panic that COVID-19 seems to evoke.  The main reasons seasonal flu is more accepted is that there is a fairly effective vaccine each year for the flu and the mortality rate is near 0.1%. It affects one in a thousand, so most people are not directly affected (and some people like President Trump don’t even realize that a close relative, in this case his grandfather, has died from the seasonal flu). COVID-19’s impact could potentially be much higher.  In a pandemic, nearly 10% of a country’s population could be infected.  That 167 per million in Italy could be hundreds of times larger.  Imagine if the number of deaths increased a hundred-fold.  The global mortality rate according to WHO is 3.4% with a wide range of uncertainty.  The apparent coincident mortality rate in Italy is 6.2% — nearly twice as high (possibly due to its relatively older population).  But even this high figure may be understated because it assumes that every diagnosed case could immediately result in death.  In reality COVID-19 can take about 5 days to manifest itself in symptoms, and then 9 to 16 more days to resolve itself.

We suggest that a rate be calculated by looking at the current deaths divided by the number of confirmed infections 5 days earlier would provide a more accurate estimate of the true mortality.  Both of these are imperfect while the epidemic is ongoing and the true mortality rate can only be calculated when the outbreak has ended and all the confirmed cases have been resolved into deaths or recoveries.  In China, the current outbreak is near its peak and the coincident rate (3.9%) is now nearly the same as the lagged5 rate (3.9%) and fairly close to the resolved rate (5.0%).  All these rates may need to be adjusted down because of the COVID-19 cases that were too mild to require a test.  However, in some countries like South Korea, we suspect that they are doing a much more thorough testing process than other countries and they therefore have a very low coincident and lagged rate and we suspect eventually a very low resolved mortality rate near 1%.  This would still make COVID-19 10 times more deadly than the seasonal flu, a serious problem worldwide, and a grave problem for certain countries.

The US Could Stop the Coronavirus If the Government Acts Promptly and Forcefully

The US has had a two month head start on the Chinese to prepare for COVID-19 caused by the new Coronavirus, but it is poorly prepared at this moment.  Ever since the Chinese government warned the world about the new virus and published its genetic code in early January (1/11), the world knew something big was coming that could severely test the healthcare and economic systems of countries worldwide.  The US has chosen to downplay the threat and now that it is on US soil, the US is not well prepared.

It did do one thing right at the beginning on February 2nd, imposing a travel ban on foreign nationals who had traveled through China recently, and requiring 14-day quarantines for returning American citizens.   In the interim it should have build up a robust testing, tracking, and quarantine system to accommodate future patients.  It should have gotten additional funding to make sure that the 70 million or so under-insured Americans are covered to meet this crisis.  Without this safety net there will be big incentives for citizens to avoid testing.  Finally, it should have put into place a set of policies to deal with an epidemic that might require the kind of draconian quarantine set up by the Chinese for Wuhan but that few Americans would accept voluntarily.  Now we see the CDC is not prepared to test the number of potential patients that have hit our shores.  Having lost precious days and weeks, hopefully the government will be able to recover to meet this historic challenge.

In the meantime this is how the number of infections in the US has grown over the last 2 months.  From the first positively identified case on Jan 21st, it stayed below 15 until Feb 16th, grew slowly (including 49 repatriated cases from Wuhan and Diamond Princess) until Mar 1st when wider availability of test kits from the CDC exploded the number of confirmed cases nearly exponentially.

Feb 1st marked the death of the first American due to COVID-19.  Since then the number of deaths has grown to 22 (on 3/8).  The apparent fatality rate in the US appears to be 4% (=22/538) as of March 8th, but the real rate could be as high as 10% or as low as 1%.  To find the fatality rate one has to adjust the apparent rate lower due to under testing of the real population of infected patients in the US that could be as high as 2,200 (on 3/8) or 220 (on 3/2).  Dividing the number of deaths (22) by 2200 yield 1%, while dividing 22 by 220 yield 10%.  The real number of infections 6 days prior makes more sense to use as a denominator than the coincident case number because COVID-19 appears to take up to 14 days to resolve itself after diagnosis.   Another way to estimate the real fatality rate is to compare deaths to total number of resolved cases which on Mar 8th amounted to 30 (=22+8 recovered).  We believe this percentage, ~70%, is unusually high currently due to CDC’s prior policy to limit testing to very serious cases – sometimes after the patient has already died.

No matter how you slice this, it is clear that the COVID-19 situation is very serious and requires a very serious, immediate, coordinated response by the government to prevent huge numbers of unnecessary deaths in the US.

Page 2 of 2

Powered by WordPress & Theme by Anders Norén