Apollo's Arrow: The Profound and Enduring Impact of Coronavirus on the Way We Live

by Nicholas A. Christakis MD PhD

In its approach, China had essentially detonated a social nuclear weapon. And so it was able to stop the spread of the virus. By late March, the number of new reported cases in the nation dropped from thousands per day to less than fifty per day.35 By April, the daily case count hit zero, and this in a country of 1.4 billion people.

The United States had placed broad travel restrictions on China on January 31, on Iran on February 29, and on Europe on March 11. But it became clear through these genetic analyses that the greatest risk to Americans was from domestic importations from other U.S. states rather than from foreign arrivals. The SARS-2 variants found in Connecticut at that time came from several other locations, mostly Washington, and none of the patients examined at that point had been abroad. Since there are numerically so many more domestic than international travelers, it should not be surprising that the domestic risk is greater. These scientists concluded that the imposition of restrictions on international travel had a limited effect on the virus’s spread.

By May 1, COVID-19 had become the leading daily killer in the United States, far eclipsing the deaths caused by the seasonal flu and even surpassing cancer and heart disease.

Like the fishmonger before him, Liu proved to be a super-spreader—twenty-three guests of the Metropole also developed SARS, including seven from the ninth floor, where he had stayed. These guests went on to seed the epidemic throughout the world. Later, the World Health Organization reported that nearly half of all the cases seen worldwide from this pandemic could be traced back to Liu’s twenty-four-hour stay at the Metropole Hotel.

So, early in an epidemic, it is hard to know the CFR. Still, most authorities, using a broad array of data, samples, and methods from countries around the world, concluded that the overall CFR for COVID-19 was in the range of 0.5 to 1.2 percent. Since roughly half (or more) of patients with SARS-2 are asymptomatic, this meant that the IFR was half as much as the CFR, so in the range of 0.25 to 0.6 percent. A CFR of about 0.5 to 1.2 percent is at least ten times less deadly than SARS-1. Compare that also to the usual seasonal flu, which has an overall CFR of about 0.1 percent. To summarize, SARS-1 was ten times deadlier than SARS-2, which in turn is ten times deadlier than the ordinary flu. But there is a further wrinkle. While SARS-2 is less deadly than SARS-1 in a given single case, that does not mean that it’s less dangerous overall. Imagine a population of one thousand people and a pathogen that infects twenty of them; this pathogen makes those people seriously ill and kills two of them. That yields a CFR of 10 percent. Now imagine another population of one thousand people and another pathogen that does the same thing—it makes twenty people seriously ill and kills two of them—but this pathogen also infects another one hundred eighty people, making them mildly or moderately ill but not killing them. The CFR for the latter case is two deaths out of two hundred patients, or 1.0 percent, making the second disease appear much milder. But in reality, it’s actually a worse dis...

Another crucial feature that made SARS-1 easier to control than SARS-2 is that it was generally not transmissible before a patient was symptomatic. That was why a large percentage of SARS-1 infections showed up in medical professionals—they were exposed to SARS-1 patients who were already quite sick. It was precisely when these patients went to the hospital, often near death, that they were most infectious. But SARS-2 is transmissible before symptoms show up.

The time between when a person is infected and when he or she shows signs or symptoms is called the incubation period. This ranges from two to fourteen days in SARS-2 (hence the recommended fourteen days of isolation) and is typically about six to seven days. For SARS-1, the incubation period was shorter, ranging from two to seven days. But there is another important interval, the latent period, which is the time between when a person is infected and when he or she becomes infectious—that is, able to spread the disease to others. The incubation period and the latent period are not always the same. The latent period is often shorter than the incubation period in SARS-2, but that was generally not the case in SARS-1. The difference between the latent period and the incubation period is sometimes known as the mismatch period; it’s calculated by measuring the incubation period and subtracting the latent period.

While COVID-19 patients on average take about seven days from exposure to show symptoms, a meaningful percentage of carriers can spread the disease for two to four days before they are symptomatic.

R0 is the average expected number of secondary infections for each primary case in a naive and wholly susceptible population with no prior history of the disease. The R0 captures the capacity of a pathogen to start an outbreak, and it reflects the degree to which it is infectious in the absence of any measures to control it. Re, however, reflects the real-time spread of the epidemic later in its course, when the population is no longer “naive.” The Re is susceptible to human responses.

Consider the following simplified illustration of this idea. Say there is a group of one hundred people who are infected with a virus. One is a super-spreader who can spread disease to three hundred people, and ninety-nine are not infectious at all. The average R0 in this hypothetical population of one hundred is 3.0, but there is a very large variation in infectiousness. If just one person from such a population is chosen at random to travel to another place, that means that, ninety-nine out of a hundred times, there will be no epidemic as a result. By comparison, if there is a group of one hundred people who can each spread the disease to three people, the average R0 is still 3.0, but now there is no variation in infectiousness. Picking one member of this second group to travel to another place means that surely an epidemic will start there. Although in both groups the pathogen has the same average R0, the fact that the variation of the R0 is smaller for the latter group means that the germ is much more likely to seed new infections in other communities. An epidemic with large variation in individual R0 manifests itself with many super-spreaders and super-spreading events. This is what happened with SARS-1.42 For SARS-1, it was estimated that four importations were necessary for one transmission chain to be initiated (the other three importations would fail to start epidemics and die out). However, when an outbreak did occur, it was likely to be explosive. For SARS-2, it...

Most estimates place the R0 of SARS-2 as roughly the same as that of SARS-1, about 3.0. This is actually rather worrisomely high for a pathogen; compare it to the R0 of ordinary influenza, which is 0.9 to 2.1.55 But SARS-2 has a smaller dispersion in Re, meaning that transmission chains are somewhat less likely to be dead ends, which makes it easier to reliably spread SARS-2 than SARS-1. SARS-2 is also less deadly than SARS-1, with a CFR of less than 1 percent (compared to around 10 percent for SARS-2). As we saw, this makes SARS-2 paradoxically more concerning, because larger numbers of people survive, and move about for longer, to transmit it. Finally, and perhaps most important, unlike SARS-1, SARS-2 has a positive mismatch period; this allows for asymptomatic transmission and makes traditional public health responses based on detection and quarantine very difficult.

For SARS-2, however, probably at least 40 percent of the human population worldwide will be infected in the end, and perhaps as much as 60 percent.

The 1957 pandemic ended after three years, when enough people had become immune that the population acquired herd immunity. In part, this was achieved through widespread vaccination (flu vaccines were invented in 1945).68 Possibly the virus also became less virulent over time, which is another typical feature of infectious diseases.

The 1918 flu pandemic (incorrectly labeled as originating in Spain) affected and killed many more people than the 1957 pandemic. Perhaps thirty-nine million people died worldwide, which was 2.1 percent of the world population, and some experts put the worldwide toll as high as one hundred million (given possible misidentification of deaths and poor reporting).

Given how deadly this pandemic was, it’s surprising that it’s not more salient in our collective memory outside of public health circles. Everyone learns facts about World War I, but few learn much about the pandemic, which was much deadlier.

By tracking excess deaths overall or for a specific category of causes of death, such as “pneumonia and influenza,” it is possible to detect a spike in cases in the United States in late February and early March, even before COVID-19 was widely recognized as a problem; this helps quantify possible underreporting due to misdiagnosed or overlooked cases. For instance, an analysis of national data from early January 2020 through March 28, 2020, by scientists at Yale found that California reported 101 deaths due to COVID-19 during this period, but there were actually 399 excess deaths due to pneumonia and influenza. Many or most of those deaths were surely from COVID-19. Looking at all causes of death combined in New York and New Jersey in this same period, the investigators found that there was as much as a threefold increase in total deaths as the pandemic swept through those two states during its first wave, regardless of how many cases of COVID-19 were actually diagnosed by health-care personnel.85 And looking at the entire United States through the end of May 2020, researchers concluded that the reported number of COVID-19 deaths was likely 22 percent lower than the true number of deaths.

The virus kills some people directly, by infecting them, and others indirectly, by, for example, prompting people to delay going to the hospital for other conditions and thus needlessly dying, or by increasing suicides as a result of depression due to job loss or social isolation. But the pandemic also saved some lives too. For instance, motor-vehicle fatalities fell during the winter and spring of 2020, as fewer people were on the road; there were fewer deaths due to complications from noncritical medical procedures, as hospitals had canceled elective procedures; fewer babies were born premature (possibly because their homebound mothers were under less physical stress or were less exposed to all pathogens); and fewer people lost their lives to respiratory conditions, as air pollution was reduced due to the cessation of manufacturing activity.

The second plague, one of the most catastrophic pandemics ever, began in Central Asia and reached Europe in 1347.93 It lasted on and off for nearly five hundred years, then disappeared in the 1830s.

The first wave of the second plague, which lasted from 1347 to 1353, is what we think of as the Black Death, although that term was not used at the time. During this first wave, driven by a favorable environment of densely packed towns and cities and substantial poverty, as much as half of the population of Europe was wiped out. This force was so powerful that it even acted as a selection pressure, changing the course of human evolution. As we will see in chapter 8, many people may have genetic features today that reflect the fact that their ancestors were the ones who survived.

For instance, in the fifteenth century, the Office of Health in Venice constructed facilities on outlying islands where all arriving ships had to sequester for forty days—which is the origin of the term quarantine (based on the Italian word for “forty,” quaranta). This duration had its foundation in the Bible, which frequently refers to the number forty in the context of purification, such as the forty days and nights that the flood in Genesis lasted, the forty days that Moses spent on Mount Sinai before receiving the Ten Commandments, the forty days of Christ’s temptation, and the forty days of Lent.

The scale of mortality of the plague is truly difficult to comprehend. As noted above, it is estimated that 30 to 50 percent of the entire European population died in a five-year period during the first wave, from 1347 to 1351.101 In some locations, whole villages and populations were wiped out completely. It is estimated that 60 percent of the population of both Florence and Siena died.

In any epidemic, a basic educational task of leaders is to help people understand what is actually happening.21 In fact, maintaining public trust can be seen as its own nonpharmaceutical intervention, not just a way to boost the efficacy of others.

In 1918, collective will was harnessed with greater ease than it was in 2020 because the Spanish flu had erupted in the midst of World War I. Members of the public supported ordinances that limited their freedoms because these rules were seen as a way to protect American troops abroad. One announcement from the Red Cross explicitly stated, “The man or woman or child who will not wear a mask now is a dangerous slacker.” The governor of California described mask-wearing as the “patriotic duty” of every American.

Sweden, alone among its Scandinavian neighbors, adopted a tactic like this, aiming to isolate the vulnerable and aged while allowing the young and healthy to go about their business as sensibly as possible in order to achieve sufficient levels of population immunity. By May 2020, Sweden had at least four times the rate of COVID-19 deaths as its Nordic neighbors with similar demographics and economies, and it began to gingerly correct its course.26 Ultimately, the architect behind the strategy admitted that the policy had not worked out as planned.27 Crucially, the Swedish economy was just as hard-hit as the economies of its neighbors who had pursued full lockdowns. Physical and economic health are inextricably linked.

Early on, many authorities, including the CDC and the surgeon general, recommended against universal mask adoption because the United States had such limited supplies of personal protective equipment. It was feared that recommending universal mask-wearing would deprive health-care workers of precious supplies. The World Health Organization also discouraged masks as late as April 2020.30 If conserving masks for health-care workers was the true policy motivation, it made no sense to mislead the public, who were understandably confused by the mixed messages in the statement “This mask is so valuable, it has to be given to a health-care provider, but in any case it would not help you and you do not need it.” Needed credibility was lost over this shifting story.

If 70 percent or more of the population in a typical urban situation used masks of decent quality (that is, about 70 percent effective), it would prevent a large-scale outbreak of a respiratory pathogen of a moderately contagious disease such as COVID-19.35

But as we saw in chapter 1, in most circumstances, closing borders alone does not work. It may delay the arrival of a pathogen, but, with very rare exceptions, it does not stop it, even in island nations where it’s easier to implement.

There is not really any such thing as totally sealing off borders—citizens abroad can still return home, and people can move illegally.

The health-care system strained under the load, and very worrisome reports reached me from my medical friends in the city. They described the situation as “gruesome” and “unreal” with the incessant arrival of very sick patients, and the ICUs bursting. It was like “a refugee camp in a war zone,” one nurse observed.117 Working conditions were extremely difficult, requiring providers to be in layers of protective equipment. They were uncomfortable, but they also feared the imminent unavailability of PPE. Doctors and nurses improvised equipment from office supplies or brought PPE in from home. Hospital morgues filled up. On March 25, some eighty-five refrigerated trailers were sent by FEMA to provide places for the bodies.118 Rules were relaxed so that local crematories could “work around the clock.”

To the extent that the risk of death seemed distant and abstract—a problem for them rather than us—the economic sacrifice and disruption to our lives seemed unwarranted.

One study evaluated how public health messaging could be more successful. What was more effective, telling people, “Follow these steps to avoid getting coronavirus” or telling them, “Follow these steps to avoid spreading coronavirus”? It turns out that the emphasis on the public threat of coronavirus is at least as effective as, and sometimes more so than, the emphasis on the personal threat. This is in keeping with other work that shows that people are motivated to get vaccinated not only out of self-interest but also out of concern for the common good.

However, to put the challenge in perspective, one of the fastest vaccines ever developed, the vaccine for the Ebola virus, approved in 2019, took five years.60 The usual time frame is closer to ten years.

Vaccine and drug development is usually extraordinarily laborious. The process costs around one billion dollars from start to finish, and it typically takes ten years.

Testing drugs always requires volunteers willing to assume some risk by taking a new and incompletely understood pharmaceutical agent. Ultimately, there must be some “first in human” or “first human dose” studies, known as phase 1 studies. These studies are then followed by phase 2 studies, which involve slightly larger numbers of subjects and attempt to further explore the safety of the drug and, more important, get an initial sense of its efficacy. If this phase is promising, then phase 3 studies, involving a much larger number of people in a randomized controlled trial, are initiated with an eye to quantifying the actual efficacy of the drug. In such pivotal phase 3 studies, one group of people is randomly assigned to get the active agent and another, the control group, gets either a placebo or whatever constitutes the usual standard of care. During phase 3, scientists also assess the possible emergence of rare toxicities that could not have been observed in the small phase 1 and phase 2 studies. Finally, even after a drug is released, ongoing monitoring is necessary to make sure the pharmaceutical works in a broader, representative sample of people who get it and to check for even rarer toxicities (this is known as phase 4).

As we saw before, given the R0 of SARS-2, roughly 67 percent of the population needs to be immunized.

In March 2020, as the lockdowns were coming into force in Europe, seismologist Thomas Lecocq of the Royal Observatory of Belgium noticed that the Earth was suddenly still.1 Every day, as we humans operate our factories, drive our cars, even simply walk on our sidewalks, we rattle the planet. Incredibly, these rattles can be detected as if they were infinitesimal earthquakes. And they had stopped. After Lecocq’s initial observation, seismologists around the world began to share data. With the anthropogenic shaking of our planet subdued, they could even detect the rushing of rivers far away. The unexpected calm allowed them to use naturally occurring background vibrations, such as the crashing of distant ocean waves, to better understand the deformation of the Earth’s stony crust. The coronavirus had changed the way the Earth moved.

That is, since we are likely to reach an attack rate of roughly 40 to 50 percent by 2022 no matter what we do, unless the vaccine becomes widely available in early 2021 (which would be the fastest a vaccine had ever been developed by far), it will not make much of a difference in the overall course of the pandemic (though even then, the vaccine would still be enormously valuable to protect uninfected people).

Either way, until 2022, Americans will live in an acutely changed world—they will be wearing masks, for example, and avoiding crowded places. I’ll call this the immediate pandemic period. For a few years after we either reach herd immunity or have a widely distributed vaccine, people will still be recovering from the overall clinical, psychological, social, and economic shock of the pandemic and the adjustments it required, perhaps through 2024. I’ll call this the intermediate pandemic period. Then, gradually, things will return to “normal”—albeit in a world with some persistent changes. Around 2024, the post-pandemic period will likely begin.

In California, several churches took Governor Newsom to court over his stay-at-home orders, arguing that they violated religious freedom, but a federal appeals court allowed his order to stand. The Supreme Court eventually agreed with the appeals court, ruling that as long as government public health policies did not single out churches for special restrictions or benefits, the restrictions did not violate the First Amendment.

Incredibly, many hospitals actually lost money or faced bankruptcy during the pandemic, despite being packed with patients and despite providing a critical lifesaving service to our nation, because reimbursement for caring for seriously ill people with infection is less than reimbursement for elective procedures for trivial problems.

The pandemic also highlighted the long-standing issue of iatrogenic (doctor-caused) illnesses or injuries arising in the course of medical care. By some metrics, this problem is a leading killer in our society, resulting in as many as fifty thousand to a hundred thousand deaths every year, in and out of hospitals. Medical errors run the gamut from surgical mistakes (leaving a sponge in somebody’s abdomen) to medication mistakes (prescribing Lasix instead of Losec, for instance—one is a diuretic and the other reduces stomach acid). People associate medical error with a surgeon accidentally removing the wrong kidney, but hospital-acquired (also known as nosocomial) infections happen vastly more often—things like urinary tract infections, surgical site infections, lung infections, bloodstream infections. The ugly truth is that most of these stem from preventable breakdowns in sterile procedures—that is, poor hygiene. Overall, possibly as many as 1 percent of patients admitted to American hospitals die from a medical mistake!

It can be difficult to differentiate the adverse economic impact of the virus itself from the adverse economic effects of the nonpharmaceutical interventions people implemented in response. Viruses can sicken and kill people and compromise the economy directly. And the precautions that people take in response, such as not spending their money or avoiding social interactions, can have adverse economic impacts on their own. A careful analysis of the 1918 pandemic in the United States that took advantage of variation in the timing of the arrival of the virus from place to place as well as variation in the timing of implementation of the nonpharmaceutical interventions as a kind of natural experiment concluded that it was the pandemic itself that depressed the economy, not the public health responses.

One day in 1902, an American second lieutenant named George Marshall crossed a small river on the island of Mindoro, Philippines, to pay a visit to a local leader and his three young daughters. The attractive girls were apparently the main draw; they laughed and made chitchat and sang “delightfully.” Social visits were conducted in the mornings in those days to avoid the oppressive heat. Marshall left. But later that same day, he had cause to return to the village, where he joined in the funeral for the same four family members who had welcomed him only hours before. They were dead from a cholera outbreak that struck precipitously, eventually killing five hundred of the twelve hundred residents of the hamlet.

A review in 2008 documented that, contrary to the hopes from forty years earlier, there had been 335 new infectious diseases in the period between 1940 and 2004 and that their threat to global health was actually increasing.

It’s worth noting that all of our modern plagues are zoonoses coming to us from wild animals. Our other major contagious diseases (from smallpox to TB to measles) mostly came to us from animals that we domesticated beginning ten thousand years ago, which has allowed us time to coevolve with them, to some extent, and develop at least some genetic resistance. These were pathogens that once afflicted the wild ancestors of our cows, pigs, sheep, chickens, and camels. For example, measles appears to have entered our species quite recently, in the sixth century BCE, from a progenitor in cattle that caused a disease called rinderpest. This coincided not only with humans living so close to cattle (domesticated millennia earlier) but also with the establishment of large cities.

Out of the U.S. population of 330 million, about 3 million people die each year, for a crude death rate of 9.1 people per thousand. If, for the sake of argument, we assume that over a year, the coronavirus pandemic causes one million deaths in the United States, the crude death rate would rise to 12.1 per thousand. The average person’s absolute risk of dying from the virus would remain small—roughly a three-out-of-a-thousand chance (1,000,000 extra deaths from COVID-19 divided by 330,000,000 people). That seems low, but this level of mortality would still far surpass all of the threats to life an average person faced that year, making COVID-19 the number-one cause of death.

One careful analysis of weekly mortality data from Sweden compared deaths per day in 2020 to prior years, quantifying the excess deaths and assessing the impact on mortality across all ages (using the method invented by William Farr in the nineteenth century, discussed in chapter 2). It estimated that SARS-2 was a kind of shock that, if sustained, would lop off three years of life expectancy from men and two years from women.

Another important subtlety here is that, while the great majority of deaths from COVID-19 occur in the elderly, this is true of virtually all other causes of death too.

Being young and not fearing COVID-19 because fatal disease seems rare is a reasonable posture (though not caring about infecting others is not!). But it is crucial to recognize that, at baseline, young people do not face quantitatively large risks of death of any kind. Nevertheless, COVID-19 increases the baseline risk of death at every age.

By using a standard benchmark of five hundred thousand dollars as the economic value of a year of life (or ten million dollars per life, regardless of age), we can estimate that one million coronavirus deaths (at the rough age distribution at which they occur) would be worth about six trillion dollars. Even at the highest end of a range of estimates of the consequences to our economy, including the expenditures by our government, we do not reach that sum. Strictly from an economic perspective, our response was commensurate to the threat posed by the pathogen. It’s a bad virus.