South Korea and Italy have been dealing with the novel coronavirus and its associated disease, COVID-19, for about the same amount of time. South Korea got started a few days earlier, but data from the WHO show clearly that each country began to find a significant number of cases somewhere around mid-to-late February of this year.1 From that point until the end of the month, the two nations had epidemic curves that tracked each other closely, but then something changed: new cases in South Korea began to fall off precipitously while the opposite happened in Italy. Even more alarmingly, deaths in South Korea stayed at a slow trickle while mortality in Italy took off on a sharp upward curve.
As of today, here is each country’s “naive” overall COVID-19 death rate, which is calculated by dividing the number of reported deaths by the number of reported cases in each country:
- South Korea: 0.8%
- Italy: 6.2%
Now, these death rates are called “naive” because the numbers that go into them are prone to bias and flaws, but that is still a fairly extreme difference. The South Korean death rate suggests that COVID-19 is something like 4–8x as deadly as a regular seasonal influenza (“flu”) virus, which is obviously awful but at least sounds like it might be within the range of manageable.2 A death rate of 6.2%, however, as Italy is currently experiencing, is 30–60x as mortal as the flu. With epidemiological experts predicting that 40–70% of the world’s population might eventually become infected with this new virus, that translates to approximately 8–14 MILLION deaths in the United States alone, or about 2 to 4% of our population.3 That would obviously be a disaster beyond reckoning.
So which is it, which is the correct death rate? Is this virus more akin to a terrible flu, or is this something that could redefine American life, or even humanity in general?
Digging into the Death Rates by Age
Ten days ago, the best COVID-19 data I could find came from China, and the resulting analysis strongly suggested that this new virus has a strong tendency to cause serious disease and death in two groups of people: those who are older and those who have baseline medical conditions, such as heart or lung disease.
With the pandemic spreading, more countries now have data to report, and South Korea and Italy are both providing accessible daily updates that are granular enough to allow for the extraction of findings deeper than simple naive death rates. Again, I thank the public health teams and medical workers in these countries for their invaluable service and generous contributions that have made this possible.
To cut to the chase, the multi-country international data continue to suggest that COVID-19 is most deadly for people over the age of 50, with the really serious risk starting past the age of 70. I have not been able to find case or mortality breakdowns by underlying health condition, so we’ll have to limit the present analysis to age only, but the results are still informative.
In South Korea, the Korean CDC has recorded 7,979 confirmed COVID-19 cases and 67 deaths, with the specific counts distributing thusly by age:4
Keep an eye on that weirdly huge and aberrant spike in the 20–29 age bracket, because we’ll come back to that soon. In the meantime, let’s translate this age-based distribution of disease burden into a death-rate-by-age chart, which should look familiar by now:
The South Korean data corroborate many other reports, all of which support the emerging theory that COVID-19 causes death almost solely in the second half of life.
Italy has also found the virus to have a predilection for older age groups, but the underlying case details have some important differences. According to data from the Istituto Superiore di Sanità (ISS), Italy has so far had 16,295 confirmed COVID-19 cases with 1,017 associated deaths.5 Here is the distribution of those cases and deaths, by age:
Converting those data to a death-rate-by-age chart yields a predictable narrative, in which older people are disproportionately impacted:
Unless you’ve been reading with a hawk eye, the charts so far for South Korea and Italy have probably looked quite similar. Without doubt, COVID-19 has shown a reliable propensity to cause increasing morbidity and mortality with advancing age, and this has been true in both of the countries we’re currently examining. However, what you may not have noticed is that, while the overarching trends are consistent, the exact distributions of cases and deaths by age really haven’t been that similar.
Digging into Case Rate Differences
If we divide the data into two broad age categories, the dissimilarities between South Korea and Italy become apparent. Take, for example, case distribution by age:
The above chart immediately tells us that Italy has recorded three-fourths of its COVID-19 cases in people aged 50 or older, whereas South Korea has found fewer than half of its cases in this age group. This finding could be the result of at least three possible factors, which might also be acting in combination with each other:
- Possibility 1: The virus is behaving differently between the two countries, attacking older people at a far higher rate in Italy
- Possibility 2: The two countries are testing their age groups at different rates, with Italy testing relatively more seniors
- Possibility 3: The underlying age demographics of the two countries are different, with Italy having more older people
I’m going to ignore the first possibility simply because we have no controlled data with which to investigate it. On a basic-science level, I think it’s unlikely that the virus is acting differently between countries. There can certainly be biologic predispositions to pathogen infectivity and morbidity (e.g., human genetic variations), and these can track with nationality and geography, but it would be malpractice to assume the existence of such differences before looking for explanatory cultural and societal factors, which are typically both more powerful and more likely.
Basic elements of societal structure, such as where older people live or how younger people congregate, can easily affect the propensity of a virus to spread within certain age groups, and differences in such factors could explain the age-based case rate discrepancies between South Korea and Italy. However, trying to analyze such cultural factors at this point would be nothing more than conjecture.
Possibility two, that South Korea and Italy are testing their populations at different rates, is much more amenable to analysis. I have not been able to find age-specific testing data for either country, but we do have reliable numbers for overall testing and for population size.4,6,7 From these, we can see right off the bat that South Korea is testing a relatively large amount of its population:
The really interesting finding, though, is that South Korea and Italy are clearly testing different people. We don’t have testing data by age, but we do know the proportions of positive tests, and this time, the story for each country is reversed:
South Korea is testing more of its population, but when Italy tests people, they’re much more likely to actually have the virus. If we combine this chart with the prior one, we are led to the general assessment that Italy is likely to be testing only the people who are sicker or more obviously infected. If you revisit the country-specific case and death distribution charts above, you’ll see that this is likely correct. The most informative age band is the 20–29-year-old one, which includes 28.5% of South Korea’s known cases but only about 5% of Italy’s.
In fact, the 20–29-year-old bracket in South Korea is a complete aberration for positive case volume; it’s 50% higher than the next closest age band in the country. I’m not sure what accounts for this, other than that South Korea has put a huge emphasis on accessible testing, and relatively healthy young people might be more likely and able to take advantage of this. This could be especially true with projects such as drive-through testing that the country has implemented.
Regardless of the exact reason, it seems very likely that Italy is testing disproportionately more of its older citizens, possibly because they are more symptomatic, and this is certainly a large contributory factor to its high rate of cases in elderly people.
Possibility three, that the underlying population demographics differ between South Korea and Italy, is also likely to be contributing to age-based differences in case rates between the countries.
Recent data from the UN support this idea fairly clearly.7 First let’s look granularly, by 10-year age bracket:
And then let’s assess the same demographic differences via our binary age buckets from before:
Based on these charts, it is fairly uncontroversial to say that Italy has an older population than South Korea, and that this difference is most stark first in the 20–40-year-old age demographic and then, with the opposite polarity, in the 70–80+ age band. This discrepancy is certainly contributing to Italy’s higher density of cases in the older age bands.
Digging into Death Rate Differences
We have so far found that a larger proportion of Italy’s known COVID-19 cases have occurred in an older population than have South Korea’s reported cases. However, there is a second issue that we must tackle, which is that, even accounting for age, more Italians are dying of the virus than South Koreans:
The death rate among people aged 0–49 in either country is reasonably low, echoing our repetitive age-dependent theme, and it is also based on a fairly small number of deaths (eight in total, actually) that makes it susceptible to small-sample-size variability and fluctuation. While the rate in this young bracket is still higher in Italy, this isn’t what is making this virus a terrifying pandemic. These low youth death rates also reinforce the fact that nobody is likely to die of a cold, regardless of what society or medical system they find themselves in, and young people overwhelmingly get benign, self-limited colds from this new coronavirus.
The death rates in the age-50+ demographic, however, are more concerning. And in this bracket, Italy has a rate that is more than 4x that of South Korea. Just as with case volume, there are several possible factors that could account for this, and they are again not mutually exclusive:
- Possibility 1: Older Italians are less healthy or otherwise more susceptible to dying from the virus
- Possibility 2: When Italy identifies infected older people, they are disproportionately more likely to be quite sick
- Possibility 3: The Italian medical system is less able to care for its severely ill cases
In the first possibility, Italy’s 50+ population could be less healthy at baseline than the same age group in South Korea, and from what we know from the Chinese data, this would indeed predict a higher mortality rate for the Italian group. Public health data to interrogate this thesis likely exist, but I have not yet found them. Biologic differences, such as genetic variances, between older Italians and South Koreans could also be present, and these might account for some difference in viral mortality, but this premise is neither likely nor easily investigable at this time.
Possibility two, that Italy is finding older cases only when those people are more sick, is on the contrary quite likely, in my opinion. We have already observed that South Korea is testing its population at a higher rate than is Italy, including those at low risk of mortality. It seems likely that Italy’s current approach to testing is therefore identifying a disproportionately high number of seriously ill people, who are also likely to be older, and people who are older and more severely ill are also more likely to die.
This theory fits with news reports of Italy’s testing approach, such as detailed by Reuters: “In Italy at first, regional authorities tested widely and counted all positive results in the published total, even if people did not have symptoms. Then […] Italy changed tack, only testing and announcing cases of people with symptoms.”8
With this approach to testing, Italy is also likely to be missing many mild coronavirus cases from the denominator of its death rate calculation, a type of bias which would tend to artificially inflate the rate and which I previously considered in more detail when analyzing the data from China.
The third possibility, that Italy may be having a hard time caring for its sickest cases, is also an unfortunate possibility. Before getting into this topic, I want to be clear that the capabilities and efficacy of any complex system, such as a national health system, are never a direct reflection of the people in that system or even its available resources. I found multiple reports stating that Italy funds its healthcare system adequately and similarly to South Korea, and that the Italian medical system is “solid” or even “top-notch.”8,9,10
Despite this, there are also convincing reports that Italian medical resources have been overwhelmed by sick COVID-19 cases. Stories of makeshift ICU beds, not enough mechanical ventilators, exhausted doctors and nurses, and terrible life-altering choices have come to light.11 It’s impossible to think that these individual cases are not showing up in the overall data, that these anecdotal tales of an overstretched system are not telling the truth about what happens to otherwise-curable sick people when resources become thin. That is, regardless of country, higher rates of sick people die when there are more of them, and this is likely happening to at least some extent in Italy.
For one comparison of medical resources between South Korea and Italy, which also includes a cautionary note about the current situation of the United States, we can look at the availability of acute-care hospital beds per thousand people in each country:12
Digging into Absolute Case and Death Differences
We have so far investigated the differences in relative COVID-19 case and death rates between South Korea and Italy. This has mostly amounted to trying to determine why Italy has been reporting a larger proportion of infected older people, and why they have been dying at a higher rate. While rate differences are important, I don’t want to gloss over the fact that differences between absolute numbers can also be important, or even critical.
Perhaps most germane, South Korea and Italy have similar absolute population sizes: about 51 million people for the former and 61 million for the latter. These figures provide the generic denominator for all absolute-number comparisons, so it’s convenient that they’re not too far apart.
In the very first chart in this article, I showed you the daily novel coronavirus case volume for South Korea versus Italy. It made the point that South Korea has somehow gotten control of new cases, whereas Italy has continued to see a surge in them. However, looking at the same data grouped by cumulative case volume is even more illustrative of this point:
South Korea began to have COVID-19 cases before Italy did, but fortunes flipped on March 10th. Since then, Italy has been on a rapid upward trajectory that has both a present value and a rate of change that are greater than those for South Korea. This is all despite the fact that South Korea is clearly testing more and therefore should be more likely to have larger absolute numbers. So, while relative rate differences do exist between the countries, there are also clearly differences in absolute case and death volumes. These figures, especially case volume, are relevant because they interact in absolute terms with the resources that any given system, or any given country, has at its disposal.
As a word of warning and fallibility, I consider this section to be my weakest of this whole piece. The data here are more suggestive than conclusive, and the conclusions are therefore more like conjecture. I’m somewhat at the mercy of the prevailing journalism winds because the forces we’re dealing with exist primarily (or only) at the cultural or country-wide level, and I only know what I can read or calculate. This could easily all be wrong.
Still, Italians are getting sick and dying in larger numbers than are South Koreans, so let’s try to think about why that is and how we might prevent it for the United States.
Let’s start with testing for infection by the virus. Testing alone, by definition, doesn’t help anybody. It doesn’t cure you or change the course of your illness. But it’s critical on a population level because it allows us to understand the state of a pandemic and to plan our society’s response. Based on disease spread, every society has to choose basically one of two strategies: containment or mitigation. Containment means identifying and quarantining every positive case, as well as tracking all the possible contacts of those cases and then testing and quarantining them, as need be. Containment is only effective in the “early” stages of an epidemic, where “early” is defined by the amount of spread in a population as well as the resources available to monitor that population. South Korea has seemingly done an amazing job of staying true to a “containment” strategy, as they are testing at a high rate in a very regimented and highly accessible way and then following up with textbook quarantine procedures.13 This has likely contributed significantly to the falloff in their daily new-case rate, but it takes a serious national investment to uphold.
Italy, on the other hand, has passed the point of being able to track each case and adhere to the other demands of containment. In one report, an Italian doctor related that he “had spent a 12-hour day tracing people who had been in contact with just one positive patient, to ensure those who next need testing are found.” He added, “you can do that if the number of cases remains two to three, but if they grow, something has to give. The system will implode if we continue to test everyone actively and then have to do all this.”8
Italy is instead pursuing mitigation as its strategy for COVID-19. With mitigation, you accept that community spread of the disease is inevitable, and you forgo attempting to identify each individual case of the virus in favor of implementing strategies that slow its transmission on the population level.
Most important among these control strategies for Italy was its decision on March 9th to institute a national quarantine, or “lockdown.”14 Under the Italian lockdown, schools, theaters, bars, and nightclubs are closed; funerals, weddings, and sporting events are suspended; and restaurants, bars, malls, and markets are allowed to operate only within limited hours and with strict spacing between attendees.
This aggressive set of Italian measures falls under the mitigation strategy of “social distancing,” and it is likely to save lives.
The United States Is More Like Italy Than South Korea
Before we talk about social distancing, let’s just get this out of the way: As far as the novel coronavirus and COVID-19 go, the United States is much more like Italy than South Korea.
We already covered the hospital bed data, where we’re not even on par with Italy.12 We also don’t have a national health system, and this virus does not pick and choose between those who have good health insurance and those who don’t; the COVID-19 virus doesn’t even spare the rich and famous, just ask the Utah Jazz or Tom Hanks.15 Experts are also advising that measures like paid sick leave might be crucial mitigating strategies, so that sick people have less of an incentive to go to work and spread the illness, and we also don’t have that.9
Perhaps most damningly, however, we’re just not testing.
As of today, the CDC reports that we have run 16,542 tests for this novel coronavirus.16 Note that this does not mean that 16,542 U.S. patients have been tested, as many of these tests have been for the same patients, for various reasons.17 In any case, even if this number were reflective of actual individual patients, it would roughly equate to a testing rate of 0.005% for the United States population thus far. Another source, based on individually collated state data, thinks that the testing number might be more like 19,066, with some results still pending; no matter what, it is small.18
As for our other numbers, we have roughly 2,182 identified positive test results out of 17,892 completed tests, with 49 reported deaths. Those figures correspond to a 12.2% positive test rate and a 2.25% naive death rate. Those numbers are closer to Italy than they are to South Korea, and whether that’s due to testing practices or actual spread and morbidity isn’t that material, since the two are clearly intertwined.
Regardless, in the United States, we have been testing at a very low rate while we have been undeniably experiencing community-based spread of COVID-19. That means that we have missed our opportunity for a containment-based approach and that we are necessarily committed to a mitigation-based approach, just as Italy currently is.
In a “moderate” disease outbreak, such as for pandemic influenza or the new novel coronavirus, the U.S. Department of Health and Human Services (HHS) estimated, in 2017, that we’d have one million hospitalizations and 200,000 people who need intensive care in an ICU. For a “very severe” scenario, such as that which occurred with the 1918 flu, HHS thinks we’re likely to have 9.6 million hospitalizations and 2.9 million people who need intensive care.19
As a comparison, tallies indicate that we have about 93,000 ICU beds in our country, at the most, with possibly 160,000 mechanical ventilation machines.9,19 If our healthcare capacity is to keep up with a moderate to severe pandemic, we have to spread the disease out over months. Many months.
In fact, since we’re doing “naive” math today, the naive division of 2.9 million ICU patients by 93,000 ICU beds comes out to 31.2, meaning that we conceivably could have to find a way to spread 2.9 million patients across 31.2 divisions of time, whether those divisions are days, weeks, or months. From what I’ve personally seen of ICU-level patients with viral respiratory disease, that division will not be “days.”
Okay Here It Is: Social Distancing
The truth is that I’m writing this part of this article at 4 A.M., west-coast time in California, while I’m supposed to be asleep in Maryland. I’m supposed to be asleep in Maryland because I’m supposed to be going to the wedding of one of my best childhood friends, Jeremy, at 5 P.M. east-coast time today. But I’m obviously not going to do that, despite my love for my friend and despite our 20-year history, and that’s because I decided, almost exactly this time two days ago, that this stupid piece-of-crap virus was the real deal. And I decided that I had to put my money where my mouth was and start doing “social distancing.”
Social distancing is all about doing things that keep you out of large gatherings, out of public, away from other people, whenever possible. It’s wonderful sometimes, but it is also the worst when it’s mandated and necessary. This concept of social distancing also forms the basis of the “flatten the curve” graphic that we’ve all probably seen a hundred times by now:
Basically, the thinking behind “flatten the curve” comes from CDC planning and posits that any given healthcare system has a certain capacity, which can be overloaded by pandemics if they spread too quickly.20 This is especially true for contagious viruses, like the current novel coronavirus. If you can keep people apart enough, however, and if you can keep them practicing basic hygiene measures like hand-washing and regular environmental cleaning, “flatten the curve” suggests that you might be able to slow the course of the disease in the population to the point where you don’t overwhelm the available healthcare resources at any given moment. Seems to make sense, just watch the gif.
If you’re older than 50, or if you have a significant chronic medical condition, social distancing for COVID-19 is about protecting yourself. You can’t catch a virus if you’re not exposed to it. Maybe you can’t run forever, but maybe you can outlast things until a vaccine or a disease-specific treatment is available or at least until the pandemic spike decreases and hospital beds open up again.
If you’re younger than 50, however, social distancing for this virus is primarily about protecting other people. If you’ve read this far, you already know that you, personally, are unlikely to die from COVID-19. But you should also realize that what for you is a slightly annoying cold might turn out to be a deadly pneumonia when you accidentally give it to your mother or father, and that’s the truth.21 Sorry. You should further consider that young people are always going to need hospitalization for non-COVID-19 reasons, such as appendicitis or car accident trauma, so having every hospital bed taken up by older patients with viral pneumonia means that this virus could end up killing young people indirectly if it deprives the system of enough resources.
And if you want to know the data behind social distancing, I have personally reviewed the original research behind the two most prominent studies that most recent news articles have been citing.22,23 Unfortunately, and it brings me no pleasure to report this, social distancing seems to work. That is, there is very strong evidence that social distancing dramatically lowers the immediate positive case velocity, and there is also decent evidence that social distancing additionally lowers the overall “excess death” amount, when all is said and done.
Both of these reference articles on social distancing studied data from the 1918 flu pandemic, and here is an amalgam of my favorite figure from each:
In the first half of this figure, the black bar represents excess pneumonia and influenza deaths in cities that implemented social interventions early and aggressively, versus the gray bar for cities that intervened late or not at all. The little asterisk means that the difference between the two is statistically significant, p<0.05.22
In the second half of the figure, the top scatter plot shows American cities distributed horizontally by their social response time to pandemic flu and vertically by their time to peak excess death rate. You can see that faster responses corresponded strongly to longer times to peak death rates (“flattening the curve”), p<0.001. The bottom scatter plot shows American cities again distributed horizontally by social response time, but this time distributed vertically by eventual excess mortality. Here you can see that faster responses correlated with less death, p=0.008.23 Blue city-name coloration just means that the original authors considered those cities to be outliers for one reason or another.
We might quibble about which specific social distancing interventions are the most effective, but it is pretty clear that social distancing works. Unfortunately. And it’s time to do it, everywhere.
And as for which death rate is “right”…well, I’m not entirely sure. It seems like it will end up depending on how effectively we’re able to follow the data that we’re shown and how well we’re able to care about each other and do the work that is required. My best guess and biggest hope is that the final true death rate will be less than 1%.
So I’m doing my best to do the work and to do what the science and the history say will help take care of all of us. But man am I depressed about it in this exact moment. I’m sorry, Jeremy. And congratulations. I miss you. And I’m there in spirit. I’ll see you soon.
- Novel Coronavirus (2019-nCoV) situation reports. https://www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports/
- CDC. Estimated Influenza Illnesses, Medical visits, Hospitalizations, and Deaths in the United States — 2018–2019 influenza season. Centers for Disease Control and Prevention https://www.cdc.gov/flu/about/burden/2018-2019.html (2020)
- Axelrod, J. Coronavirus may infect up to 70% of world’s population, expert warns. https://www.cbsnews.com/news/coronavirus-infection-outbreak-worldwide-virus-expert-warning-today-2020-03-02/ (2020)
- Korea Centers for Disease Control and Prevention (KCDC). Press Release: Updates on COVID-19 in Republic of Korea (13 March 2020). http://www.cdc.go.kr (2020)
- COVID-19 Task force of the Department of Infectious Diseases and the IT Service Istituto Superiore di Sanità (ISS). Daily infographic: Integrated surveillance of COVID-19 in Italy – 3-13-2020 | Sorveglianza integrata COVID-19: i principali dati nazionali (Integrated surveillance COVID-19: the main national data) | EpiCentro L’epidemiologia per la sanità pubblica. https://www.epicentro.iss.it/coronavirus/bollettino/covid-19-infografica_eng.pdf (2020)
- Rosini, U. COVID-19 Italia – Monitoraggio situazione. GitHub https://github.com/pcm-dpc/COVID-19 (2020)
- United Nations, Department of Economic and Social Affairs, Population Division. World Population Prospects 2019, Online Edition Rev. 1. https://population.un.org/wpp/Download/Standard/Population/ (2019)
- Parodi, E., Jewkes, S., Cha, S. & Park, J.-M. Special Report: Italy and South Korea virus outbreaks reveal disparity in deaths and tactics. Reuters (2020)
- Carroll, A. E. Here’s the Biggest Thing to Worry About With Coronavirus. The New York Times (2020)
- Jolicoeur, L. & Mullins, L. Harvard Global Health Expert: Mass. Hospitals Face Capacity Problem If Coronavirus Cases Spike Quickly. WBUR https://www.wbur.org/news/2020/03/10/coronavirus-covid-19-massachusetts-hospital-capacity-ashish-jha (2020)
- Parker, C. ‘Every ventilator becomes like gold’ – doctors give emotional warnings from Italy’s Coronavirus outbreak. World Economic Forum https://www.weforum.org/agenda/2020/03/suddenly-the-er-is-collapsing-a-doctors-stark-warning-from-italys-coronavirus-epicentre/ (2020)
- Health equipment – Hospital beds – OECD Data. OECD https://data.oecd.org/healtheqt/hospital-beds.htm (2020) doi:10.1787/0191328e-en
- Bicker, L. Coronavirus in South Korea: How ‘trace, test and treat’ may be saving lives. BBC (2020)
- Di Donato, V., Reynolds, E. & Picheta, R. All of Italy is in lockdown as coronavirus cases rise. CNN (2020)
- Europe travel ban, Tom Hanks, NBA suspends games. MSNBC.com https://www.msnbc.com/msnbc/watch/europe-travel-ban-tom-hanks-nba-suspends-games-80545349586 (2020)
- CDC. Coronavirus Disease 2019 (COVID-19) | Testing in U.S. Centers for Disease Control and Prevention https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/testing-in-us.html (2020)
- Chuck, E. The U.S. has tested more than 8,500 specimens for coronavirus. That doesn’t equal 8,500 patients. NBC News https://www.nbcnews.com/health/health-news/u-s-has-tested-more-8-500-specimens-coronavirus-doesn-n1154156 (2020)
- The COVID19 Tracking Project. COVID19 Tracker. COVID19 Tracker https://covidtracking.com (2020)
- Toner, E. & Waldhorn, R. What US Hospitals Should Do Now to Prepare for a COVID-19 Pandemic. Clinicians’ Biosecurity News | Johns Hopkins Center for Health Security http://www.centerforhealthsecurity.org/cbn/2020/cbnreport-02272020.html (2020)
- Wilson, M. The story behind ‘flatten the curve,’ the defining chart of the coronavirus. Fast Company https://www.fastcompany.com/90476143/the-story-behind-flatten-the-curve-the-defining-chart-of-the-coronavirus (2020)
- Piper, K. If you can stay home now, you make things safer for the people who can’t. Vox https://www.vox.com/future-perfect/2020/3/11/21171903/coronavirus-social-distancing-pandemic-covid19 (2020)
- Hatchett, R. J., Mecher, C. E. & Lipsitch, M. Public health interventions and epidemic intensity during the 1918 influenza pandemic. Proc. Natl. Acad. Sci. U. S. A. 104, 7582–7587 (2007)
- Markel, H. et al. Nonpharmaceutical interventions implemented by US cities during the 1918-1919 influenza pandemic. JAMA 298, 644–654 (2007)
- Cover image: “Novel Coronavirus SARS-CoV-2“; transmission electron micrograph of SARS-CoV-2 virus particles, isolated from a patient. Image captured and color-enhanced at the NIAID Integrated Research Facility (IRF) in Fort Detrick, Maryland. Credit: NIAID. March 9, 2020.