April 3, 2020

COVID-19 science: Aerosol transmission

Underneath the current debate surrounding whether people should be advised to wear masks in public are rapid developments in understanding the dangers of aerosol transmission of SARS-CoV-2. New evidence shows that the virus can travel more than two meters and remain infectious in aerosol-sized particles (those under 5 micrometers) expelled through regular breathing (i.e., without sneezing or coughing). The widely cited March 17 Lancet article “Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1” highlighted the virus’s viability on different surfaces—such as copper, steel, and cardboard—but the data on aerosol viability were just as significant and less-noticed at the time.

A subsequent article in the New York Times notes that “The finding on aerosol in particular is inconsistent with the World Health Organization’s position that the virus is not transported by air.” Quoting the study’s lead, Dr. Vincent Munster, a virologist at the National Institute of Allergy and Infectious Diseases, the article goes on to say:

To assess the ability of the virus to survive in the air, the researchers created what Dr. Munster described as “bizarre experiments done under very ideal controllable experimental conditions.” They used a rotating drum to suspend the aerosols, and provided temperature and humidity levels that closely mimic hospital conditions.

In this setup, the virus survived and stayed infectious for up to three hours, but its ability to infect drops sharply over this time, he said.

He said the aerosols might stay aloft only for about 10 minutes, but Dr. Linsey Marr [an expert at Virginia Tech on aerosol transmission of viruses who was not a member of the research team] disagreed with that assessment, and said they could stay in the air for three times longer. She also said that the experimental setup might be less comfortable for the virus than a real-life setting.

For example, she said, the researchers used a relative humidity of 65 percent. “Many, but not all viruses, have shown that they survive worst at this level of humidity,” she said. They do best at lower or much higher humidity. The humidity in a heated house is less than 40 percent, “at which the virus might survive even longer,” she said.

Mucus and respiratory fluids might also allow the virus to survive longer than the laboratory fluids the researchers used for their experiments.

Other experts said the paper’s findings illustrate the urgent need for more information about the virus’s ability to survive in aerosols, and under different conditions.

Strong anecdotal evidence of aerosol transmission comes from an incident in Mount Vernon (Washington State), in which a choir rehearsal on March 10 resulted in a COVID-19 outbreak that has so far sickened dozens and killed two people. A March 29 article in the Los Angeles Times noted:

Sixty singers showed up. A greeter offered hand sanitizer at the door, and members refrained from the usual hugs and handshakes.

“It seemed like a normal rehearsal, except that choirs are huggy places,” [the choir’s conductor Adam] Burdick recalled. “We were making music and trying to keep a certain distance between each other.”

After 2½ hours, the singers parted ways at 9 p.m.

Nearly three weeks later, 45 have been diagnosed with COVID-19 or are ill with the symptoms, at least three have been hospitalized, and two are dead.

The outbreak has stunned county health officials, who have concluded that the virus was almost certainly transmitted through the air from one or more people without symptoms

Estimate of current number of actual infections on Vancouver Island

On March 22, I said we’d likely see a COVID-19 death on Vancouver Island in the next two weeks. Yesterday, two deaths were announced. Using Michael Lin’s method, these deaths can anchor an estimate of the total number of infections to date (including those of people who have recovered) on the Island.

If we assume (as Lin does) an infection fatality rate of 1 percent and three to four weeks between infection and death, then the two deaths announced today suggest the Island had about 200 infections three to four weeks ago—say, on Sunday March 8. International evidence indicates that infections double every week in the absence of aggressive closure of businesses and social distancing. Those measures were widely adopted in BC around March 18. So, applying this growth rate, we had about one and a half doublings (to 600 infections) before the 18th. At that point, using my March 22 assumption, the rate of doubling dropped to once every two weeks. This gives us an estimate of about 1,200 infections on the Island to date, or about one for every 725 people.

This number is over twice the rough estimate provided by BC’s health authorities, a week ago (on March 28). Bonnie Henry estimated that BC had a total of 2,000 to 3,000 infections, or on a proportional basis about 350 to 500 infections on the Island.

If around 1,200 people on the Island are (or have been) infected, some 75 have tested positive, and about a fifth of the remainder have likely recovered. That leaves about 900 people with current infections, about half of whom are likely asymptomatic. The rest, some 450 people, think (or hope) they simply have a cold or a mild case of the flu; some of them are certainly self-isolating.

It’s worth noting that experts are vigorously debating SARS-CoV-2’s infection fatality rate, with some suggesting the IFR is closer to 0.5 percent. If that’s indeed the case, then the number of current (and recovered) infections on the Island is about 600.

March 30, 2020

For a moment’s light diversion, this video indicates the kind of reaction online learning may provoke in some communities in coming weeks and months.

The video—only three minutes long—has been translated into twenty languages and already watched millions of times.

Assessment of BC hospitalization modeling:

British Columbia’s public health authorities released results of their hospitalization models on Friday, March 27th.  I’ve reviewed the models’ assumptions and results against other models and data. A key thing to keep in mind is that these are not epidemiological models; they don’t generate estimates of the incidence of COVID-19 and the associated mortality and morbidity rates in BC from underlying parameters such as the virus’s assumed reproduction rate and infection fatality rate. Rather they use data on how the COVID-19 pandemic has developed in other jurisdictions—in particular, South Korea, Hubei, and northern Italy—to estimate the province’s coming need for acute and critical care hospital beds and ventilators.

The models suggest that BC should be able to cope—just—with a worst-case COVID-19 trajectory that resembles the northern Italian experience. I think they’re likely correct, with a couple of caveats.

First, this chart, which appeared early in Friday’s technical presentation, is misleading.

Looking at the green line that represents Italy, one would assume that the number of “Confirmed cases per 1 million population” was about 600 on March 23. In fact, the figure was over 1,000; and as of today, the figure is over 1,600. In the chart, the green line simply ends at the top boundary and doesn’t represent the actual figure for Italy on March 23.

The Lombardy region—the worst-case comparison case for the BC models—had 3,480 cases per million on March 26, as shown in the table below (published in the New York Times). While the BC models may have used a figure of this magnitude in estimating the worst-case hospitalization load for the province, nothing of the kind appears anywhere in the slide deck I reviewed.

We can, however, use this figure to estimate the total number of critical care hospitalizations the province will experience if it follows a north-Italian trajectory. Taking BC’s population as 5.1 million, a confirmed case rate of 3,480 cases per million produces a total of about 17,750 cases during the entire course of the pandemic. If we use the BC models’ assumption that “4.7% of all COVID-19 cases will be admitted to critical care,” then the province can anticipate a total of 834 critical care patients through the course of the pandemic.

According to the BC models, under the worst-case northern Italian scenario, COVID-19 induced demand for critical care in the province will peak at about 215 beds, 33 days into the province’s epidemic. If the total load turns out to be something in the vicinity of 834 critical care patients, however, this estimated peak requirement for beds may be too low. Much depends on the figure used for “average length of stay” (ALOS) in critical care, which the BC models put at 10 days. The shorter the ALOS, the lower the peak demand. Michael Lin, in his slides on COVID-19 that I cited last week (updated to yesterday), notes that the “average hospital stay is 3 weeks.” The ALOS in critical care will be shorter than this period, of course, because the patients there will either die or be moved to acute care as their condition improves. Nonetheless, the 10-day ALOS in the BC models seems arbitrarily short.

Also, notice that the fine print on the key slide (see below) says that the Italian epidemic had not, at the time the models were run, reached its peak. (The peak may have happened in the last couple of days.) Given that the full severity of the north-Italian epidemic was not known at the time, I find it odd that the models would be used to conclude that BC will be able to cope with an epidemic of that magnitude.

These caveats aside, the BC models’ approach and numbers appear sound on balance. The advanced preparations (especially cancelation of elective surgeries) to create surge capacity in the province’s health care system will likely be enough to cope with the predicted wave of hospitalizations in coming weeks.

March 26, 2020

The Tradeoff: “Let it rip” vs. “Lock down”

The debate of the hour concerns whether the cure of shutting down the economy is worse than the damage the disease would do if allowed to run (more or less) its course. Although Trump has voiced this concern most vocally, it has turned up in a variety of ideological spaces, including in this piece last week by Thomas Friedman, who cites the opinion of Yale’s David Katz. Gary Mason once again hits the mark with his review of the issue here. It’s also now the subject of academic discussion and analyses, well-summarized in this just published New York Times article. The first substantial economic model of the tradeoffs by Martin Eichenbaum of Northwestern and colleagues appeared the day before yesterday, and can be found here.

The “let it rip” vs. “lock down” dichotomy is a misleading and false one, I believe. The best approach is clearly something in between. Tomas Pueyo argues for one such approach in his piece in Medium five days ago, “The Hammer and the Dance” (which addresses some flaws in the highly influential Imperial College paper that suggests any relaxation of initial efforts at suppression will allow the virus to reemerge vigorously); and the Eichenbaum paper also identifies a middle road.

However, all such analyses are enormously sensitive to the modeler’s initial choice of parameter values—the pathogen’s attack rate, reproduction number (“R0”), infection fatality rate, and estimated “value” of life. The Eichenbaum paper, for instance, uses a relatively low value for the reproduction number. It also assumes that those who survive infection are substantially immune; therefore, the authors say, the optimal policy response to the pandemic involves a managed increase in a society’s herd immunity.

“Absent vaccines, the only way to prevent a recurrence of the epidemic is for enough of the population to acquire immunity by becoming infected and recovering. The optimal way to reach this critical level of immunity is gradually increase containment measures as infections rise and slowly relaxing them as new infections wane.”

I question this approach, not just because of the extraordinary number of deaths it implies, but also because we still don’t know the degree and duration of immunity of those who have recovered from the disease. Also, as Pueyo argues: “RNA-based viruses like the coronavirus or the flu tend to mutate around 100 times faster than DNA-based ones—although the coronavirus mutates more slowly than influenza viruses.”

In situations of extreme uncertainty like this, I’m a firm believer in the value of quick-and-dirty estimates that provide ballpark figures for comparison. In this case, I think the relative balance of costs of the two approaches—let it rip vs. lockdown—can be clarified by some back-of-the-envelope estimates.

The following are some numbers for Canada:

Let it rip:

If we assume, under this policy approach, that SARS CoV-2 eventually affects two-thirds of the Canadian population—about 25 million of 38 million people—which is a standard “attack rate” figure, and if we assume that the infection fertility rate (IFR) is 1 percent, as Michael Lind argued in the deck I discussed two days ago, then without any efforts to limit infections, Canada would experience nearly 250,000 deaths and about 2.5 million severe illnesses, probably mostly occurring within one year. It’s worth noting that such an outcome would nearly double the total annual deaths in Canada (about 290,000 a year).

But the IFR is almost certainly too low in such a scenario, because the health system would be overwhelmed and ill people (with both COVID-19 and other illnesses) who would otherwise recover would die. Pueyo suggests that, in such circumstances, the case fatality rate from COVID-19 should be doubled from 2 percent to 4 percent. If we similarly assume a doubling of the IFR from 1 to 2 percent, then Canada would experience 500,000 deaths from the 2.5 million cases of severe COVID-19 illness (in other words, the ratio of deaths to severe illness would double).

To estimate how much we’d be willing to pay to avoid such an outcome, we can use a standard “value of statistical life” (VSL) figure of US $6.1 million in 1999 dollars (similar to that used by the US EPA, as outlined in this report here), which converts to US $9.3 million in 2020 dollars. If we discount this VSL by one-third to take into consideration higher average age of those who die from COVID-19, then the imputed value of avoiding the projected COVID-19 deaths alone to Canada is about US $3 trillion. I’d argue that we should escalate this figure by at least a third, to account for the imputed value of avoiding the 2 million cases of severe COVID-19 illness, as well as the imputed value of the collateral deaths resulting from an overwhelmed health system. This suggests we should be willing to pay about US $4 trillion to avoid the deaths that would arise from a “let it rip” policy. Note that this amount is in excess of two years of Canada’s GDP (US $1.74 trillion), and that it does not incorporate the economic costs arising from lost demand and productivity as a result of such a shock, which could themselves be enormous.

Such calculations, of course, are fraught with problems, not least the ethical challenge of putting a value on human life. But the US $4 trillion figure should put an end to the idea that COVID-19 is “just another flu.” The costs to Canada—economically, socially, and morally—of treating this pathogen as “just another flu” are almost beyond imagining.

Lock down:

Most estimates of the consequences of a “suppression” approach to corralling COVID-19, as in the chart below (extracted from the recent Imperial College study), indicate that the approach would achieve mortality and morbidity reductions in the neighborhood of 75 percent.

This suggests that “lock down” could reduce the above imputed cost to Canada of illness and lost lives to US $1trillion. But then we should add the costs to the Canadian economy of such an approach because of lost economic activity directly arising from reduced travel, closed commercial facilities, social distancing and the like.

Here even back-of-the-envelope estimates become very difficult, in part because the Canadian economy isn’t isolated from the global economy; the pathogen’s ravages beyond our borders—and policies implemented in response by other countries—are now gravely affecting our economy.  So the question then becomes: What is the likely additional cost of “lock down” policies in Canada, given the most probable mix of economic costs inevitably spilling into our country from outside our borders? Put simply, how much extra economic damage will “lock down” cause to our economy beyond what we’re already going to suffer from things happening around the world?

Since the Canadian economy is deeply dependent on trade, we can’t escape the coming global downturn. Various analyses, including an updated assessment by JP Morgan, suggest an immediate global contraction of between 10 and 20 percent of GDP. (The JP Morgan study estimates that the Chinese economy itself contracted over 40 percent this past quarter.) Any contraction exceeding 10 percent, if it extends beyond two quarters, qualifies as a depression. However, most of these analyses also argue that the downturn will be followed by a fast recovery (so the recession or depression will be “V-shaped”). I’ll argue in future posts that there are strong reasons to believe the recovery won’t be so fast.

A reasonable first-order estimate, by my view, is that the global effects of COVID-19 will cause the Canadian economy to contract by 10 percent immediately; then the Canadian economy will recover one third of that amount in each of the three subsequent quarters (i.e., out to the end of the first quarter in 2021). Let’s assume that domestic “lock-down” policies of eight-weeks’ duration (in line with Pueyo’s “hammer” approach) double that economic impact, so the immediate total effect is a 20 percent contraction, with a one third recovery for the three subsequent quarters. In this case, the additional lost national income due to “lock down” is about US $110 billion ($43.5 billion in the first quarter—or 10 percent of the quarter’s national income—plus $32.6 billion, $21.7 billion, and $10.9 billion in the subsequent three quarters).

On balance, then, it appears that the incremental cost to Canada of the “lock down” policy would be less than half of the imputed cost of death and sickness under the “let it rip” policy.

In sum, the idea that the cure (defined as lock down) is worse than the disease (defined as let it rip) is almost certainly wrong.

March 24, 2020

COVID-19 Science

Three items strike me as important.

  • The first is a research paper, “High Temperature and High Humidity Reduce the Transmission of COVID-19,” by Chinese researchers Wang et al.; it can be found here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3551767The paper correlates temperature and humidity with the coronavirus reproductive number, R, across 100 Chinese cities from January 21 to 23. The abstract states: “We find, under a linear regression framework for 100 Chinese cities, high temperature and high relative humidity significantly reduce the transmission of COVID-19, respectively, even after controlling for population density and GDP per capita of cities. One degree Celsius increase in temperature and one percent increase in relative humidity lower R by 0.0383 and 0.0224, respectively. This result is consistent with the fact that the high temperature and high humidity significantly reduce the transmission of influenza. It indicates that the arrival of summer and rainy season in the northern hemisphere can effectively reduce the transmission of the COVID-19.”The paper has not been peer reviewed yet, and some epidemiologists have been skeptical of the results. But the paper’s key graphs (four reproduced below) are notable:

The following two maps extrapolate the results to show the possible seasonal difference globally from this March to July.

The paper’s findings could help explain the observed gradient of the disease in the United States, where northern states are seeing more severe outbreaks than those in the south. On this matter, President Trump, alas, may be right.

  • The second item of interest is this report in Buzz Feed on the results of testing in Iceland, where a far larger portion of a population has been tested that anywhere else. The research indicates a very high rate of asymptomatic transmission.The article says: “The number of individuals tested by the country’s health authorities and the biotechnology firm deCode Genetics — 3,787 — roughly translates to 10,405 per million, which compares to about 5,203 in South Korea, 2478 in Italy, and 764 in the UK.”Thorolfur Guðnason, Iceland’s chief epidemiologist, said: “Early results from deCode Genetics indicate that a low proportion of the general population has contracted the virus and that about half of those who tested positive are non-symptomatic. The other half displays very moderate cold-like symptoms.”The article concludes: “The testing has provided evidence revealing that a significant portion of those who catch the disease do so with no or mild symptoms — and confirmed multiple pieces of research that have shown that asymptomatic individuals contribute to the transmission of the disease in great numbers.”
  • The final item is this paper in The New England Journal of Medicine published yesterday on the problem of triage, titled “Fair Allocation of Scarce Medical Resources in the Time of COVID-19.” The article discusses the ethics of triage, and considers six specific recommendations for allocating medical resources in the crisis: “maximize benefits; prioritize health workers; do not allocate on a first-come, first-served basis; be responsive to evidence; recognize research participation; and apply the same principles to all Covid-19 and non–Covid-19 patients.” This will likely become the go-to article for triage committees to guide their decision making.

The US situation: While the daily growth rate shows Italy has turned the corner:

The growth rate in the US remains stunning:

According to the New York Times earlier today, Andrew Cuomo, the governor of New York says that “the state now projects that it may need as many as 140,000 hospital beds to house virus patients, up from the 110,000 projected a few days ago. As of now, only 53,000 are available. Up to 40,000 intensive-care beds could be needed.”

We’re seeing a health catastrophe unfold in New York that will probably exceed that of Wuhan itself. More generally, the coronavirus emergency seems to dramatically exacerbated by a host of systemic failures in the United States economy, culture, social structure, and political system: these include enormous inequalities in wealth and availability of health care at the individual and community levels, a pathological culture of individualism, severe regional and racial socio-economic divisions, deep political polarization, a fractured and discordant federal system, episodes of near paralysis in the federal legislature, and thoroughly incompetent presidential leadership. It’s hard to image a mixture of social properties that could more effectively hinder an effective response to this crisis in an otherwise wealthy country.

March 22, 2020

COVID-19 Science

The slide deck at the following link is from a talk by Michael Lin, an Associate Professor of biochemistry and neurobiology at Stanford. It’s by far the best technical summary I’ve seen of all the relevant science:


It’s a densely packed short course in 37 slides, updated to yesterday. If you can spare 45 minutes, it’s well-worth the time. It covers definitional confusion surrounding the pandemic; health impacts on the individual; infection rates and likely trajectories; what kills the virus; promising existing treatments; prospects for vaccines; and policy failures by the US CDC and FDA. The “Recommendation” slides at the end detailing personal hygiene measures (some of which differ from conventional advice) are themselves invaluable.

Italy turns the corner; but the US?

Italy extended its “red zone” restrictions on commerce, gatherings, and movement to the entire nation on March 9. The policies are now having the desired effect. The daily growth of cases has dropped from around 25 percent in the first third of March to 10 percent yesterday. This is the lowest daily growth rate since the beginning of the crisis in the country. The absolute number of additional cases per day is still rising (5,476 yesterday, up from 4,825 the day before), but that figure should start to fall soon too.

Meanwhile the growth in the US is astonishing (see chart below), especially in New York City, which now has almost 16,000 cases. The total number of cases in the US (now over 33,000, up 9,340 since yesterday) will probably exceed Italy’s this week.

Projection of infections on Vancouver Island:

Michael Lin, in his slides (9 through 15), walks through a calculation of current and projected infections for California; I’ve used his approach to estimate the same numbers for Vancouver Island.

I’m going to “nerd out” for a moment (my MIT training is kicking in), but it’s important to lay out how I ran this calculation. Feel free to skip the following paragraphs, to reach the conclusion below.

The number of diagnosed cases in a region is not terribly useful for tracking the course of an epidemic, because the figure is influenced by a host of extraneous factors, including rates of (and criteria for) testing. Instead, one wants to get at the underlying number of infections by the pathogen—in this case the SARS-CoV-2 virus. The total number of infections will be the total of documented cases of COVID-19 (from positive tests), of undocumented cases of COVID-19, and of SARS-CoV-2 infections that are asymptomatic and haven’t resulted in the disease COVID-19.

The best way to calculate the total infections is to work back from the number of deaths. Lin starts from the 10 deaths in California in the week before March 18; he assumes a lag time of 4 weeks before death; and he assumes, based on evidence to date from around the world (slide 10), that the SARS-CoV-2 Infection Fatality Rate (IFR) is 1 percent. Based on these assumptions, he estimates that 4 weeks before March 18th California had 1,000 infections.

Island Health hasn’t had a death yet, which means we have to interpolate some numbers. Thirty cases had been diagnosed on the Island as of March 20, and the figure will probably pass 50 in the next week. Given that the case fatality rate is around 2 percent, we can expect the first death on the Island in the next two weeks. If it happens a week from now, then using an IFR of 1 percent, and walking back four weeks, there were about 100 infections on the Island three weeks ago.

Lin estimates, based on an average of data from countries outside China (slide 12), that new infections double every week. So we likely had somewhere around 400 new infections on the Island this past week (starting with 100 three weeks ago, then 200 new infections the following week, and 400 last week), for a total of about 700 infections currently on the Island as a whole. That’s about 1 in every 1,200 people on the Island (total population about 800,000).

Many people on the Island now seem to have adopted social distancing practices, so let’s assume the rate of new infections drops sharply going forward, and the doubling time for new infections increases from one week to two (a big and important assumption). So, in two weeks the Island will have 800 new infections, and two weeks after that (April 20), another 1,600. At that point the total number of infections (some of which will have resolved by then) will be 3,100 (700 today, plus 800 two weeks from now, plus 1,600 in four weeks).

On slide 15, Lin writes that an “estimated 5% of total infected (not diagnosed) cases require hospitalization and 2.5% require ICU. Average hospital stay is 3 weeks, and starts 2 weeks after infection (1 week after diagnosis).“ So the total of 3,100 people on the Island who are infected in four weeks will produce (over the course of their illnesses) about 150 hospitalizations and 75 people requiring intensive care and ventilation.

Island Health has 96 ICU beds and 140 ventilators. Given that many of those ICU beds will be occupied by people with other health crises, it would seem likely that at least some of the extra 75 people requiring ICU care in four weeks will not receive that care.

The bottom line is that the health care crisis on the Island will probably become acute about the end of April.

March 20, 2020

COVID-19 Health Care Response

The most notable development on this topic is the now-open conversation about the necessity of triage. The discussion seems to be most advanced in Washington state, where medical leaders are quietly developing triage policy. The NY Times says: “The triage document, still under consideration, will assess factors such as age, health and likelihood of survival in determining who will get access to full care and who will merely be provided comfort care, with the expectation that they will die.” Gary Mason discusses the ethical challenges in his column in the Globe today.

March 19, 2020

COVID-19 Science

The world will pass ten thousand COVID-19 deaths today.

However, with news yesterday that China had its first day of no new local infections (although 34 came from outside the country’s borders), the debate now seems to be focusing on “where are we on the curve?”

This article, just published in the NY Times, includes curves for a large number of countries (at the bottom) that can be skimmed quickly:


This screen grab gives you a sense for the divergence in national trajectories:

In looking through the data, I’m struck by the sharp uptick in recent days in Singapore, Hong Kong and Taiwan, all of which seemed to be flattening the curve successfully a few days ago. The situation in Italy remains critical, with just over 400 new cases yesterday. But there’s a hint that the curve may have shifted from exponential to linear in the last four days.

China and South Korea have shown that it’s possible to corral the infection, but they’re now engaged in an experiment that the whole world will be watching: Can they relax the social and economic controls without a resurgence of new cases? The Imperial College study I cited yesterday suggests not, and certainly these countries will have to work hard to prevent infections from the rest of the world spilling back across their borders, and this in turn will continue to have implications for critical international supply chains.

An assessment in the NY Times by Ezekiel J. Emanuel (chairman of the department of medical ethics and health policy at the University of Pennsylvania), Susan Ellenberg (professor of biostatistics there) and Michael Levy (professor of epidemiology) gives us a realistic sense for what we can expect. They say we should be ready for a minimum of an eight week of aggressive social distancing. Here are key excerpts:

“A likely scenario is that there will be subsequent waves of the disease. That’s what happened in Denver in the 1918-19 influenza pandemic and in Toronto during the 2003 SARS outbreak. Over the next few months, South Korea, China and other countries will generate some relevant evidence to show how this might play out.


The next round of social distancing will be activated more rapidly, because officials — and the public — will be more prepared. It should also be shorter, because we can assume that most of the people who were initially infected are likely to be immune next time around. But it will still disrupt people’s lives and the economy. We will still have canceled conferences and sporting events. People will not frequent restaurants and will not travel. The service industry will be severely curtailed. And it’s going to happen again and again.


Maybe the best analogy is pumping a car’s brakes on an icy road. Either doing nothing or slamming on the brakes leads to an accident. So we pump the brakes — pushing on the brakes, then easing up, and then applying them again — and after three or four times we slow down enough to stop.


The alternative to this roller coaster would be even more drastic. It would require sustaining social distancing until there are no more cases whatsoever and then closing borders to all travelers — no contact with the outside world — for 18 months or more. While the United States and many other countries, like Denmark and Germany, have instituted travel bans, sealing the country off for over a year until a coronavirus vaccine is discovered seems implausible. But who knows. If the situation becomes dire enough, the previously impossible could become inevitable.”

Note that the “alternative” mentioned in the last paragraph is the “suppression” scenario the Imperial College modelers think will be necessary.

From a complexity science point of view, there seems to be very high “path dependency” of national trajectories, depending on how fast they move at the beginning of the pathogen’s spread in their communities. It appears that a little as a day or two of delay in implementation of these measures can make a world of difference. There will be many doctoral dissertations exploring these data in years to come.

March 18, 2020

COVID-19 Science

  • The most significant study in the last two days was released by a team of epidemiological modelers under the direction of Neil Ferguson at Imperial College in London: “Impact of non-pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and healthcare demand.”The researchers make two key points: First, “mitigation” efforts to slow the pandemic will not be enough to keep health systems from being overwhelmed “many times over.” Second, more aggressive efforts at “suppression” to actually reverse the progress of the pandemic will have to be sustained (with only the possibility of intermittent, local relaxations) until a vaccine is developed, because the infection will rebound when these measures are substantially relaxed. The researchers say:“Suppression will minimally require a combination of social distancing of the entire population, home isolation of cases and household quarantine of their family members. This may need to be supplemented by school and university closures, though it should be recognised that such closures may have negative impacts on health systems due to increased absenteeism. The major challenge of suppression is that this type of intensive intervention package – or something equivalently effective at reducing transmission – will need to be maintained until a vaccine becomes available (potentially 18 months or more) – given that we predict that transmission will quickly rebound if interventions are relaxed.”Implications: We’re in for a much longer haul than public officials are acknowledging. Markets, I think, are trying to internalize this information, which explains their dramatic decline. But we’re well outside the boundaries of conventional economic models: it’s not at all clear what the economic “floor” will be when significant parts of the world economy are shut down for a year or more—an enormous and sustained contraction of demand.
  • A paper by Li et al. published two days ago in Science (“Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus, SARS-CoV2”) fills a key gap in knowledge about transmission. Using a statistical analysis of Chinese data, the authors say:“We estimate 86% of all infections were undocumented (95% CI: [82%–90%]) prior to 23 January 2020 travel restrictions. Per person, the transmission rate of undocumented infections was 55% of documented infections ([46%–62%]), yet, due to their greater numbers, undocumented infections were the infection source for 79% of documented cases.”So, to translate to the Canadian situation, before we implemented rigorous travel and social distancing policies (in just the last week), the majority of infections in our country have likely been “undocumented” (exhibiting either no symptoms or symptoms resembling a cold or mild flu). Although these cases were substantially likely less likely to cause other infections than “documented” (test-positive) infections, their higher overall prevalence is likely emerging now as a major source of documented infections. This may explain the jump in the rate of increase in daily infections a week ago, from a percentage figure in the mid-teens to one in the mid-twenties to low thirties. (NB: A daily growth of 25 percent on yesterday’s base of 569 cases gives a total of 8,300 cases in 12 days, and 460,000 cases in 30 days.)Implications: If one were to try to design a pathogen to shut down highly networked economic systems, it would be hard to do a better job than this coronavirus. Its deadly enough to overwhelm our health-care systems if it gets loose, yet stealthy enough that keeping it from getting loose means keeping just about everyone at home, especially anyone with the sniffles, a slight cough, or a headache. Boris Johnson’s pivot in the last two days, apparently influenced by the Imperial College study, shows his government recognized that letting the virus rip through a population (to build herd immunity) isn’t a viable public policy response. But the alternative laid out in the same study—18 months or more of “suppression”—could crush our economies, so it isn’t necessarily better.