Covid-19: The US Lags Others in the Recovery Thus Far

In those countries where the spread of Covid-19 was not addressed early, all that policy-makers could then do to break its exponential growth was to lockdown the economy.  Schools were closed; non-essential businesses such as theaters, retail establishments, barbershops and hair salons, and similar were also all closed; workers were told to work from home whenever possible; and travel by other than private means was sharply curtailed.

This did succeed in reversing what had been an exponential rise in the spread of the disease, although at a tremendous cost to the economy.  While figures are not yet available on the extent of the downturn, it is clear that this will be the sharpest fall in the US economy since at least the Great Depression.  And the suddenness of the fall is unprecedented.

But as noted, the lockdowns did stabilize the number each day of new cases and of deaths, and started to bring those numbers down.  The disease was still spreading, but not at the pace of before.  There is now pressure from some quarters to lift or even fully end those lockdowns, and that process has indeed already started in much of the US as well as in other countries.  It is still too early, however, to say whether such easing will lead to a resurgence of the disease.  As was seen in January through March, several weeks will go by before one observes whether the number of daily cases will have been affected, and a further two or three weeks before one will see an impact on the daily number of deaths.

One can, however, at this point examine what the impact was of the lockdowns on the spread of the disease.  For the US, those lockdown measures were introduced starting in mid-March and lasted through end-April before they started to be partially lifted in certain jurisdictions.  And one can compare the US record to that of a number of Western European nations who also failed to stop the spread of the virus early, who then had to impose lockdown measures to break the exponential growth and start to bring it down.

The chart at the top of this post shows how the US record compares to that of a number of Western European countries in terms of the number of daily deaths from Covid-19.  It is not good.  The US is an outlier, with significantly less of a decline in the daily number of deaths than what all of these comparator countries have been able to achieve.

The chart tracks, by days from the peak day in the country, the daily deaths from the disease (using 7-day moving averages to even out the day to day fluctuations in the statistics), with the figures for each country indexed to 100 for the number of deaths on its peak day.  The data cover the period through May 17, and were calculated from the cross-country data assembled by Johns Hopkins.

Thus, for the US there were 29 days (as of May 17) since the peak day in the US of April 18, and by that point the number of daily deaths (using 7-day moving averages) was about 65% of what it was on the peak day.  In terms of absolute numbers, the US had 2,202 deaths (in terms of the 7-day average ending on that day) on April 18, and by May 17 (29 days later) the number of deaths had fallen to 1,434 (or 65% of 2,202).

European countries all did better.  By 29 days after their respective peaks, the number of deaths had fallen to 47% of what it had been in the UK, and to just 13% of what it had been in Austria (with Ireland tracking even lower, but only on its day 22).  The other European countries are all in between.  More could have been added.  I had originally included five other Western European countries in the chart, but it was then hopelessly cluttered.  So I removed those five as they were generally smaller countries (Belgium, the Netherlands, Denmark, Finland, and Portugal), plus their curves all fell in between those of Ireland on the low side and the UK on the high side.

Would the record be different if one drew a similar chart for the number of confirmed cases rather than the number of deaths?  Not really:

Here the UK curve tracks more closely to the US curve until day 28 from the respective peaks, but then fell below.  While this is speculation, one wonders if those in the UK started to take the social distancing measures more seriously once their prime minister, Boris Johnson, ended up in the intensive care unit of a hospital due to the disease (where one should keep in mind that the number of cases will then be affected only several weeks later).

Sweden may also be of interest.  In contrast to other countries, Sweden never issued legally binding lockdown orders, but rather just guidelines.  The result, however, was that the number of cases has not come down much from its peak (see the chart).  While still early, the number of daily new cases is close to 90% of what it was at its peak.  This is similar to what it was for the US at the same point in terms of the number of days from the respective peaks.  The UK path was also broadly similar at that point.

There is a difference, however, in terms of how far deaths had come down (the chart at the top of this post).  The path for Sweden has been below that of the US and in the range of other European countries.  From these observations alone one cannot say why Sweden has seen a greater reduction in its daily number of deaths (relative to their respective peaks) despite a similar number of cases as the US (relative to their peaks).  It might be because Sweden enjoys a much better health care system than the US (despite the US spending 60% more than Sweden as a share of GDP).  The age composition of those coming down with the disease might also be a factor, if younger people are, on average, a higher share of those being infected in Sweden than in the US.

But overall, the key question is why has the US performed more poorly than all the others in bringing down the number of deaths?  There are a number of possible reasons, and these reasons are not mutually exclusive – they could all be contributory.  They include:

a)  There was no national lockdown order given, but rather different states issued their orders at different times, mostly between mid-March and the beginning of April.  Indeed, a few, generally less populous, states never even issued formal lockdown orders, but simply guidelines.  This would spread out the impact, leading to less of a fall in the number of deaths relative to the national peak for any given day.

b)  Those lockdown orders varied greatly in terms of their degree of strictness.  Some were strong, and some notably lax.  Furthermore, enforcement was typically lax.  The lockdown orders were usually more serious (and much more seriously enforced) in Europe (but with Sweden as an exception).

c)  Cultural factors undoubtedly also entered.  Some Americans took social distancing measures seriously – others did not.  Indeed, some have been especially loud and insistent on not obeying such orders, in a childish display of contrariness.  They assert they have a constitutional right to do as they please (even if this may infect others with a deadly disease).

But perhaps the most important reason for the poor record of the US has been the failure of responsible presidential leadership.  There has been no coherent, and scientifically informed, national policy.  Trump spent two months denying that the virus was a concern, and the US failed to take the critical early actions which could have stemmed the spread (as the developed countries of East Asia and the Pacific were all able to do, and successfully so).  Then, when he was finally forced to admit the obvious (spurred more by a crashing stock market than by the disease itself), he has only reluctantly backed the measures needed to address the crisis.  And he has personally not modeled the behavior that the federal government’s own guidelines call for:

a)  He refused, and continues to refuse, to wear a mask in public.

b)  He continued to shake hands with those close by (leading to awkward, and amusing, moments when the other party had begun some other action of greeting).

c)  Rather than follow the social distancing guidelines at his highly publicized daily press briefings, for several weeks he had for the cameras a large number of officials and assistants all standing shoulder to shoulder around him.

d)  Most recently, Trump confirmed that he has started to take the controversial drug hydroxychloroquine, despite FDA warnings that to do so was dangerous.  Indeed, a recent study found that a higher share of Covid-19 patients who took the drug ended up dying than did those not given the drug.  Along with some of his other suggestions (such as to examine ingesting bleach or some other disinfectant to kill the virus – which health officials hastened to tell everyone not to do as it could kill them), Trump has conveyed to the public a disrespect for science and instead to do what he believes “in his gut”.

Coupled with Trump’s twitter outbursts (including the early encouragement of small, but well-organized, groups of gun-brandishing demonstrators in several states calling for an immediate lifting of the lockdown measures), it should not be a surprise that the US has been a laggard compared to what other nations have been able to accomplish.

Politicizing this public health crisis, as Trump has, will now also make it more difficult to emerge from it.  Guidelines that had been prepared by the CDC on how to safely reopen the economy, and which would have been issued on May 1, were instead suppressed by the White House.  Trump instead announced (following intensive lobbying by affected industries), that he did not want cautions to continue, but rather that everything should be quickly reopened back to “where it was” three months ago.

With such political pressures superseding the recommendation of health professionals, many will approach any opening even more cautiously than they otherwise would have.  With uncertainty as to whether restaurants, say, were re-opened because it was truly safe or because of political pressures, many will hold off on patronizing them for an extended time.  I certainly will.

Covid-19 by State: The Impact of Urbanization on the Spread

A.  Urban Concentration and Covid-19 Cumulative Deaths as of May 3

The virus that causes Covid-19, like other such viruses, spreads person to person.  Thus one should expect that there will be a more rapid pace of spread in urban areas, where people are in closer day-to-day contact.  This is not an indication of what the ultimate spread might be, as catching an infectious disease is a one-time event and contacts with others still add up over time.  It is just that instead of encountering a certain number of people in one day, it might instead take several days or even weeks.  But greater person-to-person contact increases the likelihood that one will catch the disease earlier.

Thus one should expect that at this point in the middle of the spread of Covid-19, those states that are more highly urbanized will have seen a greater number of deaths from the disease (per unit of population) than states that are more rural.  And that is indeed what one finds, although with some interesting exceptions.

The chart above shows the number of deaths in each US state per million of population, plotted against the percentage share of the urban population in the state.  The share of the state’s population that is defined as residing in an “urban” area comes from the US Census Bureau, which applies a very specific (and uniform) definition of what it labels as urban.  The calculations are based on what the Census Bureau defines as “urbanized areas”.  Under this definition, the urban population is the total population in the state living in an area with a dense urban core, including in the surrounding (suburban) areas meeting certain population density requirements, and with a total population within that area of 50,000 or more.  (Note that the Census Bureau also has a broader concept of what it considers “urban” that includes communities down to a population of 2,500.  Statements on urban populations in states are often based on this broader definition.)

While this is the best one can do in defining what it means to be living in an urban area, note that it is still highly imperfect for the purposes here.  Urban areas differ greatly.  The day-to-day contact one would experience in New York is quite different from what would normally find in a city of 50,000.  Even comparing similarly large cities, it will be quite different between New York and, say, Los Angeles.  Still, it is of interest to see whether states with a higher share of their population living in urbanized areas, as defined by the Census Bureau, have at this point in the spread of Covid-19 experienced a higher fatality rate from the disease.

The chart indicates that in general they have.  The data on the number of deaths from Covid-19 comes from the data set maintained by the New York Times, with the figures as of May 3, 2020 (and downloaded in the afternoon of May 4).  The Census Bureau figures on state total populations and on those living within urbanized areas (of 50,000 or more) are all from the 2010 census.  While these are now ten years old and will be updated once the 2020 census is completed, for the purposes of this exercise they more than suffice.  The relative populations across states will not have changed all that much.

At this point in the pandemic, states with urban population shares of up to almost 60% have uniformly relatively low (as compared to other states) death rates from Covid-19 per million of population, with all at about 100 or less (Mississippi is at 102).  Half the states (25 of the 51 including Washington, DC, as a 51st) fall into this category, with their names on the chart crowded and overlapping.  For those interested, the figures for individual states can be found in a table at the bottom of this post.

The states with urban population shares of just below 60% (Indiana) up to 80% then show more variety.  The fatality rates are very low for some (e.g. Hawaii, at 12.5 per million with an urban share of 71.5%) and substantially higher for others (e.g. Louisiana, at 434 per million and an urban share of 61%).

But the most substantial variation is seen in those states with an urban share of 80% or more.  The fatality rate at this point in the pandemic is just 18 per million in Utah despite an urban share of 81%, while it is close to 1,000 per million in the state of New York with an urban share of 83%.  Several other states in this group also have relatively low fatality rates, including California, Arizona, Nevada, and Florida.  Thus while there is a clear association seen between a higher share of a state’s population living in an urbanized area and the deaths per million from Covid-19, that relationship is not fate.  There are important exceptions.

The broad range in cumulative death rates among the states with the higher urban population shares is a consequence of several factors.  While it is not surprising that a higher urban share appears to make a location more vulnerable to a rapid spread of the virus, it is also clear that it is not inevitable.  There are a number of exceptions.  California, while vulnerable, imposed state-wide lockdown orders relatively early, for example.  The Utah public health system has also functioned particularly well.  And the state totals may be consistent with some very limited evidence (but disputed, and far from certain) that the virus that causes Covid-19 might spread less in warmer and moderately humid environments.  This might in part explain the low rates seen, despite high urbanized shares, in Arizona, California, Florida, and Nevada, as well as in Texas and Hawaii.

At the other end, the areas around New York City (in the states of New York, New Jersey, and Connecticut) saw an early and rapid spread of the virus before many were aware of it.  Based on analysis of the genome, researchers have found that the virus found there had in most cases arrived from Europe rather than directly from China.  Furthermore, they found that it was introduced to the New York area from multiple independent sources (i.e. not from just one traveler) and that it may well have arrived already in January.  There has also been a recent report that the virus had already been introduced into Europe as early as late December.  A recent analysis of a sample of bodily fluids taken from a French man living in the Paris region, who went to a local hospital on December 27 with a case of suspected pneumonia, indicated that he in fact had the virus that causes Covid-19.  He had not traveled abroad.

Thus bad luck can also play a role.  A region with a high degree of urban concentration (such as New York), with frequent travelers to and from a region where the disease had spread but where this was not known at the time (Europe), would be particularly susceptible to a highly infectious viral disease such as Covid-19.

Florida may be a surprising case.  It is a state with a relatively high share (87%) of its population residing in urbanized areas (as defined by the Census Bureau measure).  But its cumulative death rate (as of May 3) is also relatively low.  Florida has been criticized for not shutting down the spring break holidays of mid-March when numerous college students from around the country fly to Florida for parties and more.  But while the impact on cases leading to deaths in Florida itself appears to have been limited, outbreaks of the virus in other parts of the US have been traced to the spring break vacationers in Florida then returning to their homes across the US.

B.  Urban Concentration and the Recent Daily Path of Covid-19 Deaths

The picture outlined above is a static one, as it focused on the rate of fatalities from the disease at a particular point in time (May 3).  It is also of interest to review what the path has been in daily deaths from the disease, particularly over the past several weeks.  The social distancing measures that the states imposed in mid to late-March (with a good deal of variation in both when they were imposed and how strong the measures were) would be expected to have an impact on reducing the pace of the spread, with a lag of a few weeks.  They would then hopefully reduce the number of deaths from the disease a further week or so later.

In this, it is clear that the social distancing measures did succeed in flattening and then bringing down the curve, but with an important difference between the more highly urbanized states and the less urbanized ones:

The fatality rate for the US as a whole has come down since reaching a peak of about 2,000 deaths per day in mid-April (using 7-day moving averages to smooth out day-to-day fluctuations, where the dates shown are for the end of each 7-day period).  The number of deaths then fell to just below 1,800 by May 4, a reduction of 10%.  Based in part on this, the Trump administration is now encouraging states to lift their social distancing measures so that economic activity would, they hope, then recover.

But while the number of fatalities from this disease have begun to fall in the US as a whole, this has been entirely in the more urbanized states.  Between the 7-day periods ending on April 17 and on May 4, the number of fatalities in the highly urbanized states fell by 25%.  During that same period, they rose by 15% in the less urbanized states.

While the daily number of deaths remains at this point higher in the more urbanized states than in the less urbanized ones, this might soon change:

The daily number of new confirmed cases of Covid-19 is now higher in the less urbanized states.  While the measurement of confirmed cases has been suspect (it depends on how broadly one is testing), it is better now than it was in March and even early April, when testing supplies were still limited and constrained the availability of testing.  And the chart suggests that with the number of new confirmed cases now higher in the less urbanized states than in the more urbanized ones, and still heading upwards, the number of deaths from the disease in the less urbanized states may soon be higher in absolute number.

C.  What is the Plan? 

The Trump administration, and especially Trump himself, are now encouraging states to lift their social distancing measures.  The stated aim is for the economy then to recover.  However, with all the disruption that has resulted from the failure of the Trump administration to take this pandemic seriously early on, it is far from clear that this will suffice.  The economy has been severely affected, where an astounding 30 million Americans (18% of the labor force) have already applied for unemployment insurance as of the week of April 25.  Such a sharp and rapid collapse is unprecedented.  It did not happen even in the Great Depression.

The Trump administration has argued that with the daily number of deaths from Covid-19 now falling in the US, the time has come to reopen businesses.  And a number of governors, primarily Republicans in the more rural states, have started to do this, arguing that with their more rural spaces there is no longer a need for such social distancing.  But as seen in the charts above, while the accumulated number of deaths per million from Covid-19 has often (but not always) been less in the less urbanized states, the absolute number of deaths in these states has continued to grow over the last several weeks even while they have gone down significantly in the more urbanized states.  And the number of deaths each day may indeed soon be higher in the less urbanized states than in the more urbanized ones.

But what is the plan to address this?  From all I can see, there is no plan.  The Trump administration has not set out any coherent plan to safely reopen the economy.  Rather, it has simply called for the lifting of social distancing measures while hoping for the best.

Could there be a plan?  Certainly.  As public health experts have called for from the start, and as the developed market economies of East Asia and the Pacific have demonstrated is possible, management of a pandemic requires wide testing of those who appear they may have the disease, isolation if the test proves that they do, tracing the contacts of all those found to have the disease, and then testing and quarantining for about two weeks those contacts who might have been exposed to the virus.

This can be most easily done early in the course of a pandemic, when the number of cases is relatively small.  However, in January (and still through February) Trump insisted that all was fine and under control, and little was done.  Now, with over 27,000 new confirmed cases each day (as of the week ending May 4), this will be far more difficult.  The social distancing measures were implemented to stabilize the situation and then bring this number down to more manageable levels.  But while they succeeded in bringing the total number down from its peak (the daily number of new cases had been over 31,000), it is still far too high.

In addition to bringing down the daily number of new cases to more manageable levels, the social distancing measures were also put in place to give the government time to develop the capacity then to carry out the standard public health measures of testing, isolating, contact tracing, and quarantining.  But while some states appear to be building up that capacity to the extent they can, the evidence for others is scant, and for few, if any, does the capacity appear to be anywhere close to adequate.

And what is certainly missing is any leadership at the top – from Trump and his administration.  States have rather been left largely on their own, with some assistance perhaps at the working levels but without a clear nationally-led program to build the necessary capacity.

The economy of course certainly needs to be reopened, with the social distancing measures loosened and eventually lifted.  The issue is not whether this should be done but instead under what conditions.  Rather than lead a national effort to bring down the number of daily new cases through a coherent and consistent program of social distancing measures (which may well differ between urban and rural areas, but not based on political boundaries), and using the time thus gained to ramp up the public health capacity that is required, the Trump administration has floundered, with a response that has been limited, ineffective, and rudderless.

 

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

The data underlying the chart at the top of this post:

Urban population %

Deaths per million

Vermont

17.4%

84.7

Wyoming

24.5%

12.4

Maine

26.2%

42.9

Montana

26.5%

16.2

Mississippi

27.6%

102.1

South Dakota

29.9%

25.8

West Virginia

33.2%

27.0

Arkansas

39.5%

26.1

North Dakota

40.0%

37.2

Kentucky

41.0%

58.8

Iowa

41.7%

60.4

Alaska

44.5%

9.9

Oklahoma

45.8%

63.4

New Hampshire

47.3%

65.3

Alabama

48.7%

60.7

Kansas

50.2%

49.8

Idaho

50.5%

40.8

New Mexico

53.8%

73.3

Nebraska

53.8%

42.7

Tennessee

54.4%

34.7

North Carolina

54.9%

45.7

South Carolina

55.8%

59.5

Wisconsin

55.8%

59.6

Missouri

56.6%

63.1

Minnesota

58.0%

79.0

Indiana

59.2%

174.6

Louisiana

61.3%

434.3

Oregon

62.5%

28.5

Ohio

65.3%

90.0

Georgia

65.4%

120.6

Michigan

66.4%

409.7

Delaware

68.7%

197.1

Virginia

69.8%

82.5

Pennsylvania

70.7%

223.8

Hawaii

71.5%

12.5

Washington

75.0%

124.9

Texas

75.4%

35.4

Colorado

76.9%

167.0

Illinois

80.0%

205.1

Arizona

80.1%

56.6

Utah

81.2%

18.1

New York

82.7%

990.2

Maryland

83.5%

204.7

Connecticut

84.8%

681.6

Nevada

86.5%

97.0

Florida

87.4%

73.3

California

89.7%

60.0

Massachusetts

90.3%

611.5

Rhode Island

90.5%

304.0

New Jersey

92.2%

895.3

District of Columbia

100.0%

417.1

 

Trump’s Incompetent Management of the Covid-19 Pandemic: The Consequences for Health

Trump and his administration have utterly mismanaged the Covid-19 pandemic.  The direct result of this ineptitude has been tens of thousands more Americans dying (already) than would have been the case had the US managed the pandemic as well as any of the developed countries of the Asia and Pacific region.  The difference is not small.  This is also not a calculation comparing what the US did to what theoretically might have been possible.  Rather, it is a comparison to what these seven countries actually achieved, facing the same virus as the US.

The chart above shows the cumulative number of deaths from Covid-19 for the US and for seven Asian/Pacific countries between March 1 and April 25, with the numbers all scaled, for comparability, to what they would have been at the US population level.  The data comes from that assembled on an on-going basis by Johns Hopkins University, where a description is available here, and the country data itself available here.  The country population figures to work out the rates per million are those reported by the UN.

US deaths from the virus totaled 53,755 as of April 25.  The figures for each of the seven Asian/Pacific countries are, at this scale, all scrunched up at the bottom of the chart and are almost indistinguishable.  Each and every one has done far better than the US.

Focusing on the figures for April 25 only, one has:

The highest number of deaths among the seven Asian/Pacific countries (when scaled to the US population) would have been South Korea at 1,562.  That is 97% less than the US had.  The levels at all the other countries would have been even lower.  The least would have been Taiwan, with just 83 deaths from the virus (99.8% less than the US), despite far closer links to China.  Not re-scaled for population, the number of deaths so far in South Korea have totaled 242.  In Taiwan, there have been only 6.

US fatalities from Covid-19 have also just passed another significant milestone.  As of April 25 they now exceed the total number of US deaths in combat during World War I:

The figures for combat deaths are all from the Department of Veteran Affairs.  The total number of deaths from Covid-19 surpassed the number of combat deaths during the entire period of the Vietnam War on April 22, and passed the number of combat deaths during the Korean War on April 16.  And Covid-19 deaths of course continue to rise.  Over the most recent week, the daily increase averaged 2,156.

Yet Trump continues to assert that he and his administration have done a superb job of managing this crisis.  If there are any issues, he has at various times asserted the blame is with China, the Chinese President Xi Jinping (whom he, at other times, lauded), the WHO, Obama, Hillary Clinton, Democratic mayors, Democratic governors, Democrats in Congress and the Senate, Democrats more generally, the news media, government civil servants (the “deep state”), and others.

But not himself.  What can he be faulted for?  David Frum (a longtime self-described “conservative Republican” who served as a speechwriter for George W. Bush, but who is now very much a critic of Trump) puts it well in an April 7 article in The Atlantic.  I will quote it at length:

“That the pandemic occurred is not Trump’s fault. The utter unpreparedness of the United States for a pandemic is Trump’s fault. The loss of stockpiled respirators to breakage because the federal government let maintenance contracts lapse in 2018 is Trump’s fault. The failure to store sufficient protective medical gear in the national arsenal is Trump’s fault. That states are bidding against other states for equipment, paying many multiples of the precrisis price for ventilators, is Trump’s fault. Air travelers summoned home and forced to stand for hours in dense airport crowds alongside infected people? That was Trump’s fault too. Ten weeks of insisting that the coronavirus is a harmless flu that would miraculously go away on its own? Trump’s fault again. The refusal of red-state governors to act promptly, the failure to close Florida and Gulf Coast beaches until late March? That fault is more widely shared, but again, responsibility rests with Trump: He could have stopped it, and he did not.

The lying about the coronavirus by hosts on Fox News and conservative talk radio is Trump’s fault: They did it to protect him. The false hope of instant cures and nonexistent vaccines is Trump’s fault, because he told those lies to cover up his failure to act in time. The severity of the economic crisis is Trump’s fault; things would have been less bad if he had acted faster instead of sending out his chief economic adviser and his son Eric to assure Americans that the first stock-market dips were buying opportunities. The firing of a Navy captain for speaking truthfully about the virus’s threat to his crew? Trump’s fault. The fact that so many key government jobs were either empty or filled by mediocrities? Trump’s fault. The insertion of Trump’s arrogant and incompetent son-in-law as commander in chief of the national medical supply chain? Trump’s fault.

For three years, Trump has blathered and bluffed and bullied his way through an office for which he is utterly inadequate. But sooner or later, every president must face a supreme test, a test that cannot be evaded by blather and bluff and bullying. That test has overwhelmed Trump.”

Since April 7 one could add much more to this list, most recently Trump’s comments at his April 23 press briefing on the possible curative effects of injecting disinfectants into the body.  Health professionals around the country scrambled to warn the public not to do this – it could be deadly – and Trump is responsible for this confusion.  Or the ousting in the last week of Dr. Rick Bright, a highly respected medical researcher with a career on vaccine development, who as the head of BARDA (the Biomedical Advanced Research and Development Authority) had a lead role in the urgent development of a vaccine for Covid-19.  The ousting of such an official in the middle of a pandemic is astonishing for any reason.  Dr. Bright, in a statement issued through his lawyers, said it was because he had resisted administration pressure to promote the anti-malarial drugs chloroquine and hydroxychloroquine that Trump had repeatedly touted, even though there was no good evidence of their efficacy.  Indeed, a recent study found that hydroxychloroquine significantly increased the number of deaths of Covid-19 patients compared to those given the usual care.

Trump, however, is unwilling to take responsibility for these repeated failures.  In a March 13 press briefing he famously said “I don’t take responsibility at all” (where this was with specific reference to the lack of adequate testing in the US for a critical two months).  Rather, it was someone else’s fault.

Leadership requires taking responsibility.  Successful presidents do.

Covid-19: Using Fatality Rates to Estimate the Degree of Undercounting and Undertesting

The United States terribly bungled the introduction of testing for the virus that causes Covid-19, with a delay of more than a month during a critical time.  Even now, the testing rate is still less than half that of a number of European countries (per unit of population).  As a result, public health officials as well as regional hospitals have been “flying blind” as they try to manage their way through this crisis.

It is widely acknowledged that with the continued limitations on testing, the US is severely undercounting the number of Covid-19 cases in the country.  But undercounting by how much?  One can arrive at a rough estimate using the measured case fatality rates (which are based on the number of cases that have been confirmed by lab tests), and estimates of what the fatality rate is among all those infected by the virus (which is formally called the “infection fatality rate”).  And the figures suggest that we may be undercounting the number of Covid-19 cases by a factor of perhaps six to nine.  That is, the true number of Covid-19 cases in the US as of the evening of April 15 was not 633,000, but rather between 3.8 and 5.7 million.  That is not a small difference, and provides an indication of how much better the US needs to get at testing if it is to be able to reopen the economy safely.

The method used here is based on aggregate numbers, which limits its precision.  Hence the broad range given of six to nine.  But only aggregate numbers are at this point publicly available.  I would also emphasize that I am an economist (and know a bit of math) and not an epidemiologist.  Estimates provided by epidemiologists using micro data sets that trace samples of individuals, will be far better than what can be provided here, and presumably such work is now underway.  But for the purpose here, which is to arrive at a rough estimate of what the true Covid-19 case count might be in the US, the method should suffice.

The “case fatality rate” is simply the number of deaths from some disease divided by the number of cases of those that had the disease.  Once an epidemic is over, with all cases known and resolved (either by the patients surviving, or not), this rate can be measured easily.  But it is far more difficult to arrive at a good estimate during an epidemic.  The number of cases may be rising exponentially, as are also the number of fatalities from the disease.  But the fatalities occur with a lag.  That is, as the disease runs its course over some period of time, the fatalities today will be linked to the number of those coming down with the disease some number of days earlier, not to the number of those who have been diagnosed with the disease as of today.  Secondly, and importantly, it can be difficult to know during the course of the epidemic whether the number of cases that have been confirmed by some diagnostic test matches well or not the number of actual cases.  No testing is universal, but there are different degrees of how adequate or inadequate it is.

To address the first issue, one can compare the number of fatalities today to what the number of confirmed cases were some number of days before.  While there will be some distribution of the number of days between when symptoms are first observed and the death of those who will die from the disease, if that distribution does not vary over time then one can use a simple average for the number of days for this period.  That distribution might start to lengthen once we have more effective therapies to treat those with Covid-19, but so far we do not.

The chart at the top of this post then shows what the observed ratios were in the aggregate numbers for the US, between deaths from Covid-19 and confirmed cases with time lags of 10 days, 7 days, 6 days, 5 days, and none.  These are all “case fatality rates”, where the no lag case (where the current total number of deaths is divided simply by the current total number of confirmed cases) is called the “naive case fatality rate”.  The data comes from the data set assembled by The New York Times, which it is making available through GitHub.

The figures cover the period from March 30 (when the total number of fatalities first passed 3,000, and the number of confirmed cases totaled 163,000) to April 15 (when fatalities were almost ten times higher at 28.3 thousand, and the number of confirmed cases totaled 630,000).  When there is no lag, with the number of total deaths as per some date divided by the total number of cases as of that same date, the curve (shown in black on the chart above) will be low in a situation where both cases and deaths are rising.  And it will be rising as the growth rate of each slows (as we will discuss below) because then the number of deaths will more closely “catch up” with the number of confirmed cases.

At the opposite extreme, the curve in red in the chart of the ratio with a lag of ten days will be high (as current total deaths are calculated as a ratio to the number of cases of many days before, when the cases were much smaller), and will decline over time as the underlying growth rates (of cases and deaths) slow and there is again a “catch up”.

In the middle, a lag of six days leads to a ratio which is very close to constant, at about 6.0%.  It fits well, but also suggests that since late March there has not been any working off of the backlog of untested cases.  If, as we will discuss below, the infection fatality rate were around 1.0% or less (which appears to be the general consensus among the experts at this point), then that curve should be falling from 6% down towards 1%.  It isn’t.  The only alternative explanation would be that the US is in fact testing most of those who have caught the virus, and that the fatality rate from the disease is therefore in fact 6%.  I do not believe that to be true (and it would be horrendous if it were), which implies not only that there has been substantial under-testing, but also that the margin of under-testing has not improved.

One might also question whether the six days for a lag that leads to a close fit in this data is too short given what is known about the disease.  But it is consistent with how a “confirmed case” is recorded in the statistics.  It is always based on a lab test for the virus that causes Covid-19, and these tests are only given to patients after they show symptoms (and for many, after possible other causes of those symptoms are ruled out).  But when someone comes down with symptoms, there will always be some lag, on the order of perhaps 2 to 4 days between when the symptoms start and when they seek a test.  There may then be a delay of another day or two before the test is taken.  There will then be another delay between when the test is taken and when the results are found.  Due to the bungled roll-out of testing in the US, these delays have been as long as 9 days (at least in the Washington, DC, area).  Finally, the case is only recorded as a confirmed case to be added to the national statistics on the day the test result is made available, not on the day the test was taken.  A death, when it occurs, is then on average some number of days after that.

A six-day lag found in the data between when cases are confirmed and death occurs (in those cases where it is fatal) is therefore consistent with an estimate (from an article published in The Lancet) that the mean time between when symptoms first appear and death was 17.8 days for those who died.

Another way to look at this relationship is in terms of the growth rates:

Here I have only included the curves for the rates (expressed in growth rates per day, but on a seven-day moving average ending on the date shown) for total confirmed cases, for total deaths, and for the confirmed cases lagged by six days.  They have all fallen substantially over this period, which is of course critically important to the nation – exponential growth at such rates would be disastrous if they did not fall and fall quickly.  And one sees that the daily growth in the number of deaths follows very closely to what the daily growth had been in total confirmed cases six days before.

This falling curves also indicate that social distancing has worked.  Social distancing measures were only put into force starting in mid-March, and were then often substantially strengthened over the next week or two as state and local officials saw the number of Covid-19 cases exploding.  By the last week of March the growth in the total number of confirmed cases had then begun to decline (the peak growth rate was 38% a day in the seven-day period ending March 23), with this then falling to 21% a day for the seven-day period ending March 30, and the rest then as shown in the chart.  The growth in the number of deaths then fell similarly, but with a six-day lag.

What does this then imply for the actual number of Covid-19 cases in the US?  As noted above, assuming a six-day lag between when a case is recorded as “confirmed” and when those who will die for the disease do die, leads to a ratio of 6.0% between the number of deaths and number of confirmed cases (six days before).  And also as noted before, the general consensus among epidemiologists, based on what is now known, is that the fatality rate for all those who come down with Covid-19 may be around 1.0% or less.  A good deal of work is underway to try to estimate this better, but it is still early.  One careful estimate, part of the same article published in The Lancet cited before, is that the rate is 0.66%.

Using the 1.0% and 0.66% estimates as a range for what the fatality rate might be for all those who come down with Covid-19, why does the data indicate a ratio of 6.0% in what is observed?  The answer has to be that testing is not capturing a high share of those with the infection.  And while it is not surprising that in the middle of a pandemic the number of cases is being underestimated, what is surprising and significant is the degree of that underestimation for the virus that causes Covid-19.  If the true fatality rate of all those who have the disease is 1.0%, the implication of the 6.0% ratio observed is that the true number of cases in the US is six times higher than what is currently estimated.  And if the true fatality rate is 0.66%, then the true number of cases is nine times higher.  With 633,000 cases as of April 15, that implies the true number of cases is between 3.8 and 5.7 million.

There is both good news and bad news in this.  The good news is that a high share of those infected with the virus may be suffering such mild symptoms (and possibly no symptoms at all) that they do not feel compelled to call a doctor and have a test.  However, with the continued limitations on testing availability there are also many not being tested who clearly should be.  The sum of these two groups together can be easily calculated from the figures.  Simple shares would indicate that those not being tested would be in the range of 83 to 89% of all those infected.  Most of these are probably in the category of those exhibiting mild or even no symptoms, but we will not know how large that share is until the US is testing all those with any symptoms.

The bad news, however, is that those with mild or even no symptoms appear able to pass on the virus to others.  And the person catching the virus from that individual may then suffer severe symptoms.  It all depends on the individual – their genes, age, health condition, and other such factors.

This means that carrying out the testing to identify infected individuals will be far more difficult than would be the case if symptoms were always clear.  But until vaccines are available or effective therapeutics have been developed, there really is no alternative to the traditional public health measures of identification of those infected, isolating those with the disease, tracing all the contacts of those thus identified, and then quarantining all those contacts until it is clear whether or not they have caught the virus.  This will be difficult when a substantial share of those with the disease do not manifest clear symptoms, but is not impossible.  Those who have caught the virus got it from someone, and good public health work can often find them.

This would have been far easier to do if action had been taken immediately, in January, when it was clear that what had happened in Wuhan was serious (one does not impose a full lockdown on a city of 11 million if is it not serious) and the WHO was issuing a series of increasingly urgent warnings (starting from January 5).  Trump, in contrast, famously asserted on January 22, in an interview while at the Davos meetings in Switzerland, that he was not concerned about the virus – that “We have it totally under control”.  He also boasted of having “a great relationship with President Xi” of China, with no problem in cooperation with China on the virus.  “The relationship is very good”, Trump said.

If action had been taken in January or even early February, the US would only have had to deal with on the order of dozens of cases a day.  Identification, isolation, tracing of contacts, and quarantining then would have been far easier than what will now be necessary, when confirmed new cases since April 2 have averaged 30,000 per day.

Social distancing has worked in bringing the growth rate in new cases down to single digits and, with a lag, to a similar slowdown in the growth in fatalities.  But it has been achieved only with an effective shutting down of the economy, with an unemployment rate that is probably now the highest it has been since the Great Depression (we will find out when the April figure is reported at the end of the month).  Trump’s refusal until mid-March to see Covid-19 as the serious threat that it is has come at a tremendous cost.

 

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

Update:  April 18, 2020

After I put up this post yesterday, I came across two reports with figures that are broadly consistent with the findings of the analysis.

There are two components leading to undertesting.  One is that a substantial share of those with the symptoms of Covid-19 are not being tested in the US.  The second is that a substantial share of those carrying the virus are asymptomatic or only mildly symptomatic, and hence do not seek a test (or qualify for a test due to the current limits on testing).  However, those carrying the virus can still spread it to others, even if they are asymptomatic.

The two reports are of analyses where they could test 100% of an isolated population, and then see among all those testing positive what share were asymptomatic.  One report was on testing of the 4,800 sailors on the US Navy aircraft carrier Theodore Roosevelt.  Of the 600 sailors who tested positive, roughly 60% did not show symptoms of Covid-19.  If one assumes that those showing no symptoms would not have otherwise been tested, and that perhaps half of those exhibiting symptoms would not have been tested (in the US testing as it is being carried out now), then 80% of those with the virus would not be recorded in the US as a “confirmed case”.

The figure of one-half not seeking a test is, of course, a pure guess, but is not necessarily unreasonable.  If your symptoms are there but moderate (keep in mind that about 80% of the confirmed cases do not require hospitalization – they stay home instead where they self-monitor), it might well be rational for many individuals not to seek a test unless their condition worsens.  Not only do you have to find a testing location, but you will likely need to line up and wait, and stay relatively close to a large number of individuals who are there precisely because they think they have the disease.  There may also be insurance issues, even though the tests are in principle supposed now to be free to the individual.

The other report was on the testing of all 3,000 individuals in an isolated village in Northern Italy.  It found that 50 to 75% of those who tested positive were asymptomatic (why there is a range on this is not clear to me).  They then isolated those who tested positive (88 individuals), and the number of those sick in the village from Covid-19 fell to seven in less than 10 days.  Hence isolation works well if you can identify all those carrying the virus.  If the share asymptomatic was 50%, and one then assumes that only half of those with symptoms would be tested (under US conditions), then the share of confirmed cases would be just 25% of actual cases.  And if the share asymptomatic was 75%, with half of those with symptoms being tested, then the share of confirmed cases would be just 12.5% of actual cases.

There are limitations, of course, to these analyses.  The samples are not all that large, they are special populations (sailors who are mostly male and mostly aged between 20 and 50, or those living in an isolated village in Italy, where the share elderly may be high), plus the tests were all done at one point in time.  Those asymptomatic at the time of the test may have started to come down with obvious symptoms a few days later.  But while this might apply to some, it is also clear from this and other work that many of those with the disease, but who were tested for some reason (most likely as part of the process of tracing and testing the contacts of those who came down with the disease) never exhibit outward symptoms.

All this suggests that current testing in the US is only capturing a small share of those with the disease.

Trump’s Mismanagement of the Covid-19 Crisis: South Korea Shows What Would Have Been Possible

Source:  David Leonhardt, Newsletter of April 13, 2020, The New York Times

I normally only include charts I have developed myself in this blog, but the chart above, from David Leonhardt of the New York Times, is particularly striking.  It comes from his newsletter of April 13, and shows the daily number of deaths (on a seven-day moving average) per 10 million people, from February 19 to now in the US and in South Korea.

It shows what the US could have achieved had the Trump administration managed this crisis as well as South Korea has.  And one cannot argue that South Korea is a rich country with resources that the US does not have – GDP per capita in the US is double that of South Korea.  Nor is it because of travel bans.  Trump repeatedly asserts that the crisis would have been far greater in the US had he not had the singular wisdom to impose a ban on travel (by non-US citizens) from China on February 2 (and from Europe and other countries later).  But the only travel ban South Korea has imposed has been travel from Hubei Province in China.  And South Korea has far more contact with China, from both business and personal travel and trade in goods, than the US has.  Yet despite this, the deaths from Covid-19 have been far fewer in South Korea than in the US even after scaling for population.

And it is not only South Korea that has demonstrated competence in the management of the Covid-19 virus.  Death rates in other countries of East Asia, all similarly heavily exposed to China, have been even lower than that of South Korea.  In terms of the cumulative number of deaths from Covid-19 since the crisis began (as of April 13), there have been 4 deaths per million of population in South Korea, but just 2 per million in Singapore, 1 per million in Japan, 0.5 per million in Hong Kong, and 0.3 per million in Taiwan.  For the US, in contrast, the total is 71 per million.  (Reminder:  The chart above tracks deaths per day, not the cumulative total, and shows the figures per 10 million of population.)

This also shows that Trump’s repeated assertion that the deaths suffered in the US were inevitable – that nothing more could have been done – is simply nonsense.  Sadly, it is deadly nonsense.  South Korea shows what could have been done.  Travel bans were not important.  Rather, it was the basic public health measures of large-scale testing, identifying those with the virus or who may have been exposed to the virus, quarantining or isolating those exposed (including self-isolating, along with self-monitoring and regular reporting), and then treating in hospitals those who developed severe symptoms.

None of this is new to public health professionals.  And the US has excellent public health professionals.  What was different in the US was Trump, who refused to listen to them and indeed treated many of those in government as enemies to be attacked (as those with expertise were seen as members of the “deep state”).

The US had prepared plans on what to do should an infectious disease such as Covid-19 threaten.  There was, for example, a major effort to develop such plans in 2006/2007, towards the end of the Bush administration.  The work included running exercises similar to war-games of various scenarios (“table-top” exercises), to see how officials would respond and what the likely outcomes then would be.  These plans were further developed during Obama’s two terms in office.  But the Trump administration then ignored this previous preparation, and indeed took pride in dismantling important elements of it.

Dr. James Lawler, now an infectious disease doctor at the University of Nebraska but then serving in the Bush White House, participated in the 2006/2007 task force.  Over the weekend, the New York Times released a trove of over 80 pages of emails (obtained through a Freedom of Information Act request) of late-January to mid-March from Dr. Lawler and other experts, in and out of government, discussing how to address the crisis.  Particularly telling is a March 12 email from Dr. Lawler in which he said:

“We are making every misstep initially made in the table-tops at the outset of pandemic planning in 2006.  We had systematically addressed all of these and had a plan that would work – and has worked in Hong Kong/Singapore.  We have thrown 15 years of institutional learning out the window …”

Throwing those 15 years of institutional learning out the window has had deadly consequences.