Your Coronavirus Test Is Positive. Maybe It Shouldn’t Be.


Some of the nation’s leading public health experts are raising a new concern in the endless debate over coronavirus testing in the United States: The standard tests are diagnosing huge numbers of people who may be carrying relatively insignificant amounts of the virus.

Most of these people are not likely to be contagious, and identifying them may contribute to bottlenecks that prevent those who are contagious from being found in time. But researchers say the solution is not to test less, or to skip testing people without symptoms, as recently suggested by the Centers for Disease Control and Prevention.

Instead, new data underscore the need for more widespread use of rapid tests, even if they are less sensitive.

“The decision not to test asymptomatic people is just really backward,” said Dr. Michael Mina, an epidemiologist at the Harvard T.H. Chan School of Public Health, referring to the C.D.C. recommendation.

“In fact, we should be ramping up testing of all different people,” he said, “but we have to do it through whole different mechanisms.”

In what may be a step in this direction, the Trump administration announced on Thursday that it would purchase 150 million rapid tests.

The most widely used diagnostic test for the new coronavirus, called a PCR test, provides a simple yes-no answer to the question of whether a patient is infected.

But similar PCR tests for other viruses do offer some sense of how contagious an infected patient may be: The results may include a rough estimate of the amount of virus in the patient’s body.

“We’ve been using one type of data for everything, and that is just plus or minus — that’s all,” Dr. Mina said. “We’re using that for clinical diagnostics, for public health, for policy decision-making.”

But yes-no isn’t good enough, he added. It’s the amount of virus that should dictate the infected patient’s next steps. “It’s really irresponsible, I think, to forgo the recognition that this is a quantitative issue,” Dr. Mina said.

The PCR test amplifies genetic matter from the virus in cycles; the fewer cycles required, the greater the amount of virus, or viral load, in the sample. The greater the viral load, the more likely the patient is to be contagious.

This number of amplification cycles needed to find the virus, called the cycle threshold, is never included in the results sent to doctors and coronavirus patients, although it could tell them how infectious the patients are.

In three sets of testing data that include cycle thresholds, compiled by officials in Massachusetts, New York and Nevada, up to 90 percent of people testing positive carried barely any virus, a review by The Times found.

On Thursday, the United States recorded 45,604 new coronavirus cases, according to a database maintained by The Times. If the rates of contagiousness in Massachusetts and New York were to apply nationwide, then perhaps only 4,500 of those people may actually need to isolate and submit to contact tracing.

One solution would be to adjust the cycle threshold used now to decide that a patient is infected. Most tests set the limit at 40, a few at 37. This means that you are positive for the coronavirus if the test process required up to 40 cycles, or 37, to detect the virus.

Tests with thresholds so high may detect not just live virus but also genetic fragments, leftovers from infection that pose no particular risk — akin to finding a hair in a room long after a person has left, Dr. Mina said.

Any test with a cycle threshold above 35 is too sensitive, agreed Juliet Morrison, a virologist at the University of California, Riverside. “I’m shocked that people would think that 40 could represent a positive,” she said.CORONAVIRUS SCHOOLS BRIEFING: The pandemic is upending education. Get the latest news and tips as students go back to school.Sign Up

A more reasonable cutoff would be 30 to 35, she added. Dr. Mina said he would set the figure at 30, or even less. Those changes would mean the amount of genetic material in a patient’s sample would have to be 100-fold to 1,000-fold that of the current standard for the test to return a positive result — at least, one worth acting on.

The Food and Drug Administration said in an emailed statement that it does not specify the cycle threshold ranges used to determine who is positive, and that “commercial manufacturers and laboratories set their own.”

The Centers for Disease Control and Prevention said it is examining the use of cycle threshold measures “for policy decisions.” The agency said it would need to collaborate with the F.D.A. and with device manufacturers to ensure the measures “can be used properly and with assurance that we know what they mean.”

The C.D.C.’s own calculations suggest that it is extremely difficult to detect any live virus in a sample above a threshold of 33 cycles. Officials at some state labs said the C.D.C. had not asked them to note threshold values or to share them with contact-tracing organizations.

For example, North Carolina’s state lab uses the Thermo Fisher coronavirus test, which automatically classifies results based on a cutoff of 37 cycles. A spokeswoman for the lab said testers did not have access to the precise numbers.

This amounts to an enormous missed opportunity to learn more about the disease, some experts said.

“It’s just kind of mind-blowing to me that people are not recording the C.T. values from all these tests — that they’re just returning a positive or a negative,” said Angela Rasmussen, a virologist at Columbia University in New York.

“It would be useful information to know if somebody’s positive, whether they have a high viral load or a low viral load,” she added.The Coronavirus Outbreak ›

Officials at the Wadsworth Center, New York’s state lab, have access to C.T. values from tests they have processed, and analyzed their numbers at The Times’s request. In July, the lab identified 794 positive tests, based on a threshold of 40 cycles.

With a cutoff of 35, about half of those tests would no longer qualify as positive. About 70 percent would no longer be judged positive if the cycles were limited to 30.

In Massachusetts, from 85 to 90 percent of people who tested positive in July with a cycle threshold of 40 would have been deemed negative if the threshold were 30 cycles, Dr. Mina said. “I would say that none of those people should be contact-traced, not one,” he said.

Other experts informed of these numbers were stunned.

“I’m really shocked that it could be that high — the proportion of people with high C.T. value results,” said Dr. Ashish Jha, director of the Harvard Global Health Institute. “Boy, does it really change the way we need to be thinking about testing.”

Dr. Jha said he had thought of the PCR test as a problem because it cannot scale to the volume, frequency or speed of tests needed. “But what I am realizing is that a really substantial part of the problem is that we’re not even testing the people who we need to be testing,” he said.

The number of people with positive results who aren’t infectious is particularly concerning, said Scott Becker, executive director of the Association of Public Health Laboratories. “That worries me a lot, just because it’s so high,” he said, adding that the organization intended to meet with Dr. Mina to discuss the issue.

The F.D.A. noted that people may have a low viral load when they are newly infected. A test with less sensitivity would miss these infections.

But that problem is easily solved, Dr. Mina said: “Test them again, six hours later or 15 hours later or whatever,” he said. A rapid test would find these patients quickly, even if it were less sensitive, because their viral loads would quickly rise.

PCR tests still have a role, he and other experts said. For example, their sensitivity is an asset when identifying newly infected people to enroll in clinical trials of drugs.

But with 20 percent or more of people testing positive for the virus in some parts of the country, Dr. Mina and other researchers are questioning the use of PCR tests as a frontline diagnostic tool.

People infected with the virus are most infectious from a day or two before symptoms appear till about five days after. But at the current testing rates, “you’re not going to be doing it frequently enough to have any chance of really capturing somebody in that window,” Dr. Mina added.

Highly sensitive PCR tests seemed like the best option for tracking the coronavirus at the start of the pandemic. But for the outbreaks raging now, he said, what’s needed are coronavirus tests that are fast, cheap and abundant enough to frequently test everyone who needs it — even if the tests are less sensitive.

ADVERTISEMENTContinue reading the main story

“It might not catch every last one of the transmitting people, but it sure will catch the most transmissible people, including the superspreaders,” Dr. Mina said. “That alone would drive epidemics practically to zero.”

One Meeting in Boston Seeded Tens of Thousands of Infections, Study Finds


On Feb. 26, 175 executives at the biotech company Biogen gathered at a Boston hotel for the first night of a conference. At the time, the coronavirus seemed a faraway problem, limited mostly to China.

But the virus was right there at the conference, spreading from person to person. A new study suggests that the meeting turned into a superspreading event, seeding infections that would affect tens of thousands of people across the United States and in countries as far as Singapore and Australia.

The study, which the authors posted online on Tuesday and has not yet been published in a scientific journal, gives an unprecedented look at how far the coronavirus can spread given the right opportunities.

“It’s a really valuable study,” said Dr. Joshua Schiffer, a physician and mathematical modeling expert who studies infectious diseases at the Fred Hutchinson Cancer Research Center in Seattle and was not involved in the research.

Dr. Schiffer said that the new genetic evidence fit well with what epidemiologists and disease modelers have been learning about the coronavirus. The Biogen conference, he said, was just one of many similar events that amplified and spread the virus in its early months. “I don’t think it’s a fluke at all,” he said.

The results came out of a project that began in early March at the Broad Institute of Harvard and M.I.T., a research center specializing in large-scale genome sequencing. As a wave of Covid-19 patients crashed into Massachusetts General Hospital, the Broad researchers analyzed the genetic material of the viruses infecting the patients’ cells. The scientists also looked at samples from the Massachusetts Department of Public Health, which ran tests around Boston at homeless shelters and nursing homes. All told, the scientists analyzed the viral genomes of 772 people with Covid-19 between January and May.

The researchers then compared all of these genomes to trace where each virus came from. When a virus replicates, its descendants inherit its genetic material. If a random mutation pops up in one of its genes, it will also get passed down to later descendants. The vast majority of such mutations don’t change how the virus behaves. But researchers can use them to track the spread of an epidemic.

“It’s kind of like a fingerprint we can use to follow viruses around,” said Bronwyn MacInnis, a genomic epidemiologist at the Broad Institute.

The first confirmed case of the coronavirus in Boston turned up on Jan. 29. The patient had traveled from Wuhan, China, and his virus carried distinctive mutations found in Wuhan. But Dr. MacInnis and her colleagues didn’t find any other viruses in Boston from later months with the same genetic fingerprint. It’s likely that the patient’s isolation prevented the virus from spreading.

But as February rolled on, the researchers determined, at least 80 other people arrived in Boston with the virus. Undiagnosed, they spread it to others.

Most of the viral lineages in Boston have a genetic fingerprint linking them to earlier cases in Europe, the study found. Some travelers brought the virus directly from Europe in February and March, whereas others may have picked up the European lineage elsewhere in the northeastern United States.

Dr. MacInnis and her colleagues took a detailed look at a few key places to see how the virus swept through the city. At Massachusetts General Hospital, for example, they found that coronaviruses in patients did not share many of the same mutations. That was a relief, because it meant that the hospital was not a breeding ground where a single virus could spread quickly from patient to patient.CORONAVIRUS SCHOOLS BRIEFING: The pandemic is upending education. Get the latest news and tips as students go back to school.Sign Up

But that’s exactly what happened in a skilled nursing home where 85 percent of patients and 37 percent of the staff were infected. The researchers identified three different virus lineages in the home, but one of them accounted for 90 percent of the infections.

Such superspreading events are a hallmark of the coronavirus. When an infected person shows up in the right place — generally inside, with poor ventilation and close contact with other people — the virus can infect a lot of people in very little time. These unfortunate events don’t happen often, and so most people who get infected with the coronavirus don’t pass it on to anyone else.

The virus that raged through the nursing home didn’t spread beyond its walls, as far as Dr. MacInnis and her colleagues could tell. But when the virus showed up at the Biogen conference, the story turned out very differently.

The researchers were able to sequence 28 viral genomes from people at the meeting. All of them shared the same mutation, called C2416T. The only known samples with that mutation from before the Biogen event came from two people in France on Feb. 29.The Coronavirus Outbreak ›

It’s possible that a single person came to the meeting from Europe carrying the C2416T mutation. It’s also possible that the virus carrying this mutation had already been in Boston for a week or two, and someone brought it into the meeting.

As the attendees spent hours together in close quarters, in poorly ventilated rooms, without wearing masks, the virus thrived. While replicating inside the cells of one meeting attendee, the virus gained a second mutation, called G26233T. Everyone who was subsequently infected by that person carried the double-mutant virus.

From the meeting, the researchers concluded, this lineage spread into the surrounding community. In a Boston homeless shelter, for example, researchers found 51 viral samples with the C2416T mutation, and 54 with both mutations.

“We had no idea it would be associated with the conference,” Dr. MacInnis said. “It came as a complete surprise.”

The researchers estimated that roughly 20,000 people in the Boston area could have acquired the conference virus.

New York saw a similar pattern, according to Matthew Maurano, a computational biologist at N.Y.U. Langone Health. After many viral strains arrived from Europe in February, a few came to dominate the city. “A lot of lineages die off, and some spread enormously,” Dr. Maurano said.

The Boston double-mutant spread particularly far. Researchers identified this lineage in samples collected later in Virginia, North Carolina and Michigan. Overseas, it turned up in Europe, Asia and Australia.

Dr. Jacob Lemieux, a co-author of the new study and an infectious disease physician at Massachusetts General Hospital, said it was impossible at the moment to determine how many people acquired the virus in the months after the Biogen conference. But it would be in the tens of thousands.

Six months after the conference, Dr. MacInnis said that it should serve as a warning to anyone who thinks life can return to an unmasked version of normal before the virus is brought under control.

“One bad decision can affect a lot of people,” she said. “And the ones who suffer the most from that reality are the most vulnerable among us.”

When covid-19 becomes a chronic illness


The symptoms began in March, says Laura, a British woman in her mid-20s. At first covid-19 felt like a bad case of flu: a dry cough, fever, shortness of breath, loss of smell, “horrendous nausea” and general fatigue. After three weeks of rest, things started to improve. Five months later, she has still not recovered. Sometimes her symptoms ease for a week or two, but they inevitably return. “When it’s bad I can’t even go on work calls, because if I talk too much I can’t breathe.”

In March, as covid-19 cases began their exponential rise in country after country, doctors focused on saving patients’ lives. Speedy sharing of knowledge, clinical trials and hands-on experience have made the illness less deadly. In Britain about half the patients treated in intensive-care units (icus) in the weeks to the middle of April died. By the end of June mortality was below 30%. Reductions were seen across all age groups, which means the fall cannot have been caused by fewer frail old people arriving in hospital (see chart 1). In places where the epidemic has subsided, calmer wards have meant better care. But improved knowledge about treatment probably accounts for much of the improvement.

Doctors have learned a lot. They have stopped rushing covid-19 patients onto ventilators, which can cause lung damage. Oxygen supplied through small prongs in the nostrils is much less invasive and often does the job. In British icus the share of covid-19 patients on ventilation fell from 90% in the early days to 30% in June. Treatment protocols have improved further with the addition of dexamethasone, an immune-dampening drug that increases survival rates in patients who need oxygen.

Now, though, doctors and scientists are shifting their focus to those who survive the infection—including the subset of people like Laura, who have never been ill enough to be hospitalised, but who have also never recovered sufficiently to return to normal life.

In most people, covid-19 is a brief, mild illness. Between a third and a half of those infected do not notice any symptoms. In those who do become unwell symptoms usually clear within two to three weeks with just home rest. In Europe only around 3-4% of those who become infected are admitted to hospital.

Yet at the same time it is becoming clear that some small but significant proportion of those infected have symptoms that persist for months. Prolonged recovery is not unusual for patients hospitalised for pneumonia, a frequent complication of covid-19. It is also common for people who have been admitted to an icu, who are by definition seriously ill. But many clinicians say that the share of covid-19 patients with lingering problems is far higher than is seen with other viral illnesses such as influenza. The problems are also more varied, often including lung, heart and psychological symptoms, says Sally Singh of the University of Leicester, who leads the development of a covid-19 rehabilitation programme for Britain’s health service.

The walking wounded

Anecdotal reports of long-lasting illness have been around since the early days of the pandemic. But with more than 22m cases confirmed worldwide, and with infection rates having peaked several months ago in most rich countries, statistical patterns about the virus’s lingering effects are starting to emerge. A paper in the British Medical Journal on August 11th concluded that as many as 60,000 people in Britain have long-term symptoms. Yet only about 6% of Britain’s population—around 4m people—seems to have been infected with the virus so far.

Severity of the illness is one predictor of lasting problems. Ian Hall, the director of the Biomedical Research Centre at the University of Nottingham, reckons that 30-50% of patients hospitalised with covid-19 have significant symptoms six to eight weeks after they have been discharged. That number rises further for patients who were admitted to an icu. But even those who escaped with a mild illness, like Laura, are at risk. More than 10% of them remain unwell for more than three weeks, according to a patient-tracking study that follows mostly American and British patients. They struggle with fatigue, breathlessness, body aches and cognitive problems which many describe as “brain fog” (see chart 2).

Some long-term covid-19 patients may be suffering from undiagnosed conditions such as diabetes or thyroid dysfunction, which are “unmasked” by the infection, says Avindra Nath of the National Institutes of Health in America. For others, the collection of symptoms is suggestively similar to those seen in chronic fatigue syndrome (cfs). The biological causes of cfs are still poorly understood, but data from America indicate that three-quarters of cases follow viral or bacterial infections. One hypothesis is that the syndrome is caused by the immune system failing to properly stand down after being called on to battle an infection. It may be that sars-cov2, the virus that causes covid-19, is unusually likely to provoke such a lingering over-reaction.

If the root cause is not known, then the growing understanding of just what covid-19 can do to the body can at least suggest what sorts of care long-term sufferers may need. The hallmark of many covid-19 cases is damage to the lungs. Aggressive inflammation leads to the destruction of lung tissue and the formation of scars. The scarring, in turn, impedes the flow of oxygen from the lungs into the blood. That can cause breathlessness, even with light exercise. Small studies of covid-19 patients discharged from hospitals have found that 25-30% have impaired oxygen flow.

The prognosis is unclear. People treated in icus for other viral infections usually recover about 80% of their previous lung function fairly quickly, but the final 20% can take three to six months, says Dr Hall. And in some cases lung scarring can worsen over time, especially if combined with new health problems later in life.

Breathing problems can also arise from another effect of covid-19—its tendency to cause blood clots, which is unusual for a respiratory virus. When they form in the lungs, clots can choke off blood flow, making it even harder to absorb oxygen. And the virus may cause breathlessness in a more subtle way, too, by damaging the lining of blood vessels, which limits how much blood can flow through them.

Covid-19 can also damage the heart. It can inflame the tissues that surround the organ, as well as the blood vessels that ferry nutrients to it. That can weaken the heart muscle, and eventually lead to heart failure. Blood clots cause problems here, as well, since the heart must pump harder to push blood through partly blocked vessels. Over time, that can weaken the muscle.

Nobody knows exactly how often such cardiac complications occur. But news from Germany is worrying, says Clyde Yancy, a cardiologist at Northwestern University, in Illinois. Using mri scans, one study found evidence that covid-19 causes inflammation and other heart changes—including in people who had tested positive for the virus more than two months earlier and were, by the time of their scans, free from symptoms. The changes were small, and not enough by themselves to cause clinical symptoms. But even a minor injury to the heart may eventually lead to heart failure if it lingers for long enough, says Dr Yancy.

Least understood are the long-term effects of covid-19 on the nervous system and the brain. Patients with lingering post-covid symptoms complain of headaches, tingling and numbness in the feet, and other neurological problems. Problems that suggest a dysfunction of the autonomic nervous system, such as irregular heart beat, dry mouth and gastrointestinal problems are also common, says Dr Nath. But the exact cause of such symptoms remains unclear, as does the reason why more than half of those infected suffer a temporary loss of their sense of smell. (Some, rather than losing it, have it altered instead, so that things smell different after the infection than they did beforehand.)

Confronted with a baffling array of symptoms and few detailed explanations about exactly what is going wrong, doctors are desperate for guidance. A third of general practitioners in Britain already have patients with lasting post-covid symptoms. For now, the best they can offer is referral to lung or cardiac rehabilitation. Such therapy may improve a patient’s quality of life with something as simple as a breathing exercise. Britain, Belgium and other countries are setting up specialised covid-19 rehabilitation programmes for those recovering from the disease. Waiting lists are already long.

In the absence of understanding, doctors must fall back on lessons from other illnesses. Lingering symptoms are not the exclusive preserve of covid-19. Full recovery from other viral diseases such as influenza can occasionally take months. Data on cfs suggests chances of recovery are best in the first three months.

More specific data are on the way. Studies in America, Britain, China and Europe have enrolled thousands of patients, and should begin reporting initial results in the next few months. But for now, those suffering the lingering effects of the disease must deal not only with the physical symptoms, but with uncertainty about just how long it will take them to get better. ■

What Happens When COVID-19 Collides With Flu Season?


Many questions remain about how flu season might affect the pandemic, and vice versa. For example, would coinfection with influenza worsen the course of COVID-19? Experts also aren’t certain whether influenza vaccination could help protect against COVID-19 or whether steps taken to mitigate COVID-19 will reduce the burden of the coming flu season.

Some hints have come from preliminary research conducted in China, where influenza was still widely circulating when the first novel coronavirus infections emerged, and in the southern hemisphere, which is currently in the midst of its flu season.

At least 2 things are clear: Quicker and more widely available testing is needed to distinguish between COVID-19 and influenza, which have similar symptoms, at least at first, but require different treatments. On top of that, a severe influenza season—the result of more virulent strains, inadequate vaccination rates, or a combination of both—coupled with a COVID-19 pandemic that shows no signs of abating, could overwhelm already taxed emergency departments and intensive care units.

As pulmonary and critical care specialist Benjamin Singer, MD, wrote in a recent editorial, influenza and other causes of pneumonia represent the eighth leading cause of US deaths in nonpandemic years.

“We can expect that the new reality of COVID-19 will only complicate the next influenza season,” Singer, of the Northwestern University Feinberg School of Medicine, concluded in his editorial.

Flu-Like but Not Alike

Distinguishing between influenza and COVID-19 “has important prognostic implications,” Singer said in an interview. “In many ways it matters that you find out quickly.”

While the course of influenza is rapid, COVID-19 “kind of limps along a little bit,” he said. Knowing the reason for a patient’s respiratory symptoms “helps inform what you can expect.”

Identifying the cause, of course, helps determine how best to treat respiratory symptoms, Singer noted. Although supportive care for influenza and COVID-19 is similar, drug treatments don’t overlap, he said.

“We have things that we can do for COVID if we know someone’s infected,” he said. “If they have influenza, we can give antivirals targeted against influenza.”

But mistakenly treating patients with influenza as though they have COVID-19 is wasteful and potentially harmful, Singer said.

For example, he noted, randomized controlled trials have found that intravenous remdesivir, a broad-spectrum antiviral that is not approved anywhere in the world for any use, was more effective than a placebo in treating severe and moderate COVID-19. Remdesivir has received Emergency Use Authorization to treat COVID-19 from the US Food and Drug Administration and regulatory agencies in a few other countries, but, as an unapproved drug, it has been in short supply. Meanwhile, although earlier studies found that remdesivir had antiviral activity against influenza A, the drug has not been tested in patients with the flu, so there’s no evidence it’s effective in treating that disease.

Another drug, the corticosteroid dexamethasone, appears to be effective in some patients hospitalized with COVID-19, but it could harm those who instead have influenza. A recent preliminary report found that dexamethasone resulted in a lower 28-day mortality rate among patients hospitalized with COVID-19 who were receiving respiratory support. However, in 2019 clinical practice guidelines, the Infectious Diseases Society of America (IDSA) specifically advised against using corticosteroids to treat seasonal influenza unless clinically indicated for other reasons, such as asthma. Data from randomized controlled trials of corticosteroid treatment of influenza aren’t available, but 2 meta-analyses of observational studies suggested that corticosteroid treatment of patients hospitalized with influenza was associated with increased mortality, according to IDSA.

A retrospective study from Wuhan, China, suggested that lopinavir-ritonavir combination therapy led to faster resolution of pneumonia than standard care alone among patients with both COVID-19 and influenza. However, the World Health Organization (WHO) on July 4 discontinued the lopinavir-ritonavir arm of its Solidarity trial because interim results found the treatment, which is approved for HIV, produced little or no mortality reduction in patients hospitalized with COVID-19.

“Although we need more data to confirm the conclusion, we prefer to use the lopinavir-ritonavir to treat all COVID-19 patients with influenza,” coauthor Rui Zeng, MD, PhD, a kidney specialist on the faculty of Wuhan’s Tongji Medical College at Huazhong University of Science and Technology, said in an email.

Another reason it’s important to determine whether respiratory symptoms are due to influenza or to COVID-19 (or both) is that mitigation efforts for the former aren’t as strict as for the latter. “We’ve never told people with influenza to isolate themselves from everyone else,” Osterholm, founder and director of the Center for Infectious Disease Research and Policy at the University of Minnesota, said in an interview.

Without quickly learning which virus they have, some people with COVID-19 during flu season might mistakenly attribute their symptoms to influenza and not take the necessary precautions to prevent spreading severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which is more easily transmitted, he said.

In addition, distinguishing between COVID-19 and influenza will be vital for disease surveillance, the authors of a recently published letter noted. Given the overlap of symptoms, systematic testing for SARS-CoV-2 and influenza will be needed during the upcoming flu season to determine the contributions of each viral illness to the burden of respiratory disease, the authors wrote.

Considering Coinfection

Physicians in several countries have reported patients who tested positive for both COVID-19 and seasonal influenza. “We have seen patients with both viruses,” Singer said. “It was early on, in March.”

But such patients have represented a small minority.

“The chances are more likely that they have one or the other,” Osterholm said, noting that only 3% or 4% of the population have SARS-CoV-2 infection, while 10% to 20% might become infected with influenza virus, so the odds of being infected with both are small.

Early reports from China suggested that coinfection with other respiratory diseases was extremely rare in patients with COVID-19. For example, in a study of 99 patients with COVID-19 admitted to Wuhan Jinyintan Hospital from January 1 to January 20, none tested positive for any of 9 other respiratory pathogens, including influenza A and influenza B.

However, Zeng’s study, conducted at Wuhan’s Tongji Hospital, which the government had designated for treating patients with severe COVID-19, produced a much different finding. Of 544 patients with polymerase chain reaction–confirmed COVID-19 who were admitted from January 28 to February 18, 11.8% were coinfected with influenza A or influenza B. Zeng noted that the influenza infection rate in his study was similar to that reported in the US during the 2018-2019 influenza season.

“Coinfection was a significant risk factor for prolonged hospital stay,” Zeng said. In addition, his study found that COVID-19 patients who were coinfected with influenza shed SARS-CoV-2 longer than other COVID-19 patients (17 days vs 12 days on average). “We don’t know the reason.”

Studies about coinfection in the US have found rates more in line with those at Jinyintan Hospital than at Tongji Hospital.

A recent study in JAMA found that out of 1996 patients hospitalized with COVID-19 in metropolitan New York City who were tested for other respiratory viruses, only 42 (2.1%) were coinfected, and only 1 was coinfected with influenza. The patients were hospitalized between March 1 and April 4.

In Northern California, laboratories that simultaneously tested for SARS-CoV-2 and other respiratory pathogens found a 10-fold higher coinfection rate (20.7%) than the New York study, but only 0.9% of specimens were coinfected with influenza. The authors, who reported their findings in a JAMA research letter, studied 1217 specimens, 116 of which had tested positive for SARS-CoV-2 and 318 for other respiratory pathogens. Of the 116 that were positive for SARS-CoV-2, 24 were positive for at least 1 other respiratory pathogen. However, only 1 of the 116 was positive for influenza.

During the pandemic, “the possibility of COVID-19 should be considered regardless of positive findings for other pathogens,” Japanese researchers recommended in a recent case report about a 57-year-old restaurant worker.

COVID-19 Protection From Flu Shots?

study conducted at Ohio’s Wright-Patterson Air Force Base during the 2017-2018 flu season recently caught the attention of Luigi Marchionni, MD, PhD, an oncologist and computational biologist at Johns Hopkins University. The study compared the influenza vaccination status of approximately 6000 Department of Defense personnel with their respiratory virus status.

“That paper didn’t find vaccination was making people more likely or less likely to get another infection from another virus,” Marchionni explained. However, it did find that influenza vaccination was associated with a higher risk of non-SARS coronavirus infection, offset by a lower risk of influenza, parainfluenza, respiratory syncytial virus, and some other respiratory infections.

Marchionni wondered whether the finding of an association between flu shots and coronavirus infections might bode ill for influenza vaccination in the middle of a coronavirus pandemic. So he and his coauthors explored a possible county-level association between influenza vaccination coverage in people aged 65 years or older and the number of COVID-19 deaths.

Their findings, which have not yet been peer-reviewed, suggest that influenza vaccination in that age group is negatively associated with COVID-19 mortality. Marchionni said he and his coauthors have submitted an expanded version of their paper to a peer-reviewed journal.

“I’m quite confident in the fact that influenza vaccination in the population is associated with less [COVID-19] mortality,” Marchionni said. “There are many plausible biological explanations.”

Another study that has not yet undergone peer review also found that patients with COVID-19 who were immunized against influenza fared better than those who had not. The authors analyzed data from 92 664 confirmed COVID-19 cases in Brazil and found that recently vaccinated patients had, on average, an 8% lower chance of needing intensive care, an 18% lower chance of requiring invasive respiratory support, and a 17% lower chance of dying.

Can We Curb Flu Along With COVID-19?

Intuitively, it makes sense that wearing masks, social distancing, working from home, closing schools, and other strategies to minimize the spread of COVID-19 would lessen transmission of other respiratory infectious diseases as well.

That appeared to be the case in Taiwan, researchers concluded in a recent brief report. They compared 25 weeks of case data for severe influenza, invasive Streptococcus pneumoniae disease, and pneumonia deaths from 2016 to 2020. All 3 trended downward in 2020 compared with the same weeks in previous years, especially after Taiwan implemented COVID-19 prevention strategies. The downward trend does not appear to be a result of negligence in reporting cases, the authors noted, because there were still substantial cases of severe influenza-like illness reported. However, they tested negative for influenza.

Japanese researchers also observed less influenza activity week by week in 2020 compared with the previous 5 seasons. They speculated that high awareness among the Japanese public of measures to reduce COVID-19 transmission early in the year might explain their finding, according to a recent research letter in JAMA.

And researchers in Qatar recently reported a “dramatic decrease” of laboratory-confirmed influenza A after the state closed schools on March 10, although laboratory-confirmed cases of other respiratory pathogens, including influenza B, barely budged. Seasonal variations likely do not explain the 30-fold drop in laboratory-confirmed influenza A cases between February 13 to March 14 and March 15 to April 11, because a similar decline was not seen between the same periods in 2019, the authors wrote.

The situation in the southern hemisphere might provide more clues as to what the northern hemisphere can expect in the upcoming flu season. Or, as Osterholm cautioned, it might not.

“We’re seeing an incredibly mild flu season in the southern hemisphere,” he said. “To date, we’ve seen virtually little, little activity…We don’t know what’s going on right now.” And that’s throughout the southern hemisphere, including COVID-19 hotspots such as Brazil, Osterholm noted. “We have to be careful not to assume that’s what’s going to happen in the northern hemisphere.”

The best-case explanation for the southern hemisphere’s mild flu season is that COVID-19 mitigation strategies are tamping down the spread of other respiratory viruses, said Brendan Flannery, PhD, coauthor of the letter calling for systematic testing for both influenza and COVID-19. But the worst-case scenario is that COVID-19 has overwhelmed health care systems, so people with the flu are staying home and not being counted or seeking care but getting lost in the crowd of COVID-19 patients, said Flannery, lead investigator from the US Centers for Disease Control and for the US Flu Vaccine Effectiveness Network.

“We’re all going to learn a lot,” Osterholm said of the upcoming flu season. “We can speculate until we’re blue in the face, and I don’t think we know yet what’s going to happen.”

What if ‘Herd Immunity’ Is Closer Than Scientists Thought?


We’ve known from the beginning how the end will arrive. Eventually, the coronavirus will be unable to find enough susceptible hosts to survive, fading out wherever it briefly emerges.

To achieve so-called herd immunity — the point at which the virus can no longer spread widely because there are not enough vulnerable humans — scientists have suggested that perhaps 70 percent of a given population must be immune, through vaccination or because they survived the infection.

Now some researchers are wrestling with a hopeful possibility. In interviews with The New York Times, more than a dozen scientists said that the threshold is likely to be much lower: just 50 percent, perhaps even less. If that’s true, then it may be possible to turn back the coronavirus more quickly than once thought.

The new estimates result from complicated statistical modeling of the pandemic, and the models have all taken divergent approaches, yielding inconsistent estimates. It is not certain that any community in the world has enough residents now immune to the virus to resist a second wave.

But in parts of New York, London and Mumbai, for example, it is not inconceivable that there is already substantial immunity to the coronavirus, scientists said.

“I’m quite prepared to believe that there are pockets in New York City and London which have substantial immunity,” said Bill Hanage, an epidemiologist at the Harvard T.H. Chan School of Public Health. “What happens this winter will reflect that.”

“The question of what it means for the population as a whole, however, is much more fraught,” he added.

Herd immunity is calculated from the epidemic’s so-called reproductive number, R0, an indicator of how many people each infected person spreads the virus to.

The initial calculations for the herd immunity threshold assumed that each community member had the same susceptibility to the virus and mixed randomly with everyone else in the community.

“That doesn’t happen in real life,” said Dr. Saad Omer, director of the Yale Institute for Global Health. “Herd immunity could vary from group to group, and subpopulation to subpopulation,” and even by postal codes, he said.

For example, a neighborhood of older people may have little contact with others but succumb to the virus quickly when they encounter it, whereas teenagers may bequeath the virus to dozens of contacts and yet stay healthy themselves. The virus moves slowly in suburban and rural areas, where people live far apart, but zips through cities and households thick with people.

Once such real-world variations in density and demographics are accounted for, the estimates for herd immunity fall. Some researchers even suggested the figure may be in the range of 10 to 20 percent, but they were in the minority.

Assuming the virus ferrets out the most outgoing and most susceptible in the first wave, immunity following a wave of infection is distributed more efficiently than with a vaccination campaign that seeks to protect everyone, said Tom Britton, a mathematician at Stockholm University.

His model puts the threshold for herd immunity at 43 percent — that is, the virus cannot hang on in a community after that percentage of residents has been infected and recovered.

Still, that means many residents of the community will have been sickened or have died, a high price to pay for herd immunity. And experts like Dr. Hanage cautioned that even a community that may have reached herd immunity cannot afford to be complacent.

The virus may still flare up here and there, even if its overall spread is stymied. It’s also unclear how long someone who has recovered may be immune, and for how long.

The coronavirus crashed this year’s Purim celebrations in the Orthodox Jewish neighborhoods of New York City, tearing through the parades and masquerades in Brooklyn on March 9 and 10.

Schools and synagogues soon shut down to quell the spread, but it was too late. By April, thousands in the Brooklyn communities were infected, and hundreds had died.

“It’s like a black hole in my memory because of how traumatic it was,” said Blimi Marcus, a nurse practitioner who lives in Borough Park, which was hit hard by the virus.

But all that has changed now, Ms. Marcus added: “The general feeling is one of complacency, that somehow we’ve all had it and we’re safe.”CORONAVIRUS SCHOOLS BRIEFING: The pandemic is upending education. Get the latest news and tips as students go back to school.Sign Up

Is it possible that some of these communities have herd immunity? In some clinics, up to 80 percent of people tested had antibodies to the virus. The highest prevalence was found among teenage boys.

But people at clinics are more likely to be showing symptoms and therefore more likely to be infected, said Wan Yang, an epidemiologist at Columbia University’s Mailman School of Public Health in New York. Random household surveys would probably find lower rates — but still well above the 21 percent average reported for New York City, she said.

Researchers in Mumbai conducted just such a random household survey, knocking on every fourth door — or, if it was locked, the fifth — and took blood for antibody testing. They found a startling disparity between the city’s poorest neighborhoods and its more affluent enclaves. Between 51 and 58 percent of residents in poor areas had antibodies, versus 11 to 17 percent elsewhere in the city.

The lowest-income residents are packed tightly together, share toilets, and have little access to masks. “These factors contributed to a silent infection spread,” said Dr. Jayanthi Shastri, a microbiologist at Kasturba Hospital in Mumbai who led the work.

Most researchers are wary of concluding that the hardest-hit neighborhoods of Brooklyn, or even those in blighted areas of Mumbai, have reached herd immunity or will be spared future outbreaks.

But models like Dr. Britton’s hint that it’s not impossible. Other researchers have suggested, controversially, that herd immunity can be achieved at rates of immunity as low as 10 or 20 percent — and that entire countries may already have achieved that goal.

Criticism trailed Sunetra Gupta, a theoretical epidemiologist at Oxford University, after a widely circulated interview in which she said that London and New York may already have reached herd immunity because of variability among people, combined with a theoretical immunity to common cold coronaviruses that may protect against the new one.

“That could be the explanation for why you don’t see a resurgence in places like New York,” she said.

Most experts reject that notion. Several studies have shown that certain immune cells produced following infection with seasonal coronaviruses may also recognize the new coronavirus.

But “where is the evidence that it’s protective?” asked Natalie Dean, a biostatistician at the University of Florida.The Coronavirus Outbreak ›

These cities have not returned to pre-pandemic levels of activity, other experts noted.

“We are still nowhere near back to normal in our daily behavior,” said Virginia Pitzer, a mathematical epidemiologist at the Yale School of Public Health. “To think that we can just stop doing all that and go back to normal and not see a rise in cases I think is wrong, is incorrect.”

A second wave might also hit groups or neighborhoods that were spared by the first, and still wreak havoc, she said. Immunity is a patchwork quilt in New York, for instance: Antibodies were present in 68 percent of people visiting a clinic in the Corona neighborhood of Queens, for instance, but in just 13 percent of those tested at a clinic in the Cobble Hill section of Brooklyn.

But another group, led by the mathematician Gabriela Gomes of the University of Strathclyde in Britain, accounted for variations within a society in its model and found that Belgium, England, Portugal and Spain have herd immunity thresholds in the range of 10 to 20 percent.

“At least in countries we applied it to, we could never get any signal that herd immunity thresholds are higher,” Dr. Gomes said. “I think it’s good to have this horizon that it may be just a few more months of pandemic.”

Other experts urged caution, saying these models are flawed, as all models are, and that they oversimplify conditions on the ground.

Jeffrey Shaman, an epidemiologist at Columbia University, said it wasn’t clear to him that Dr. Gomes’s model offered only one possible solution. And he was suspicious of the big ranges among the four countries.

“I think we’d be playing with fire if we pretended we’re done with this,” Dr. Shaman said.

The new models offer food for thought, he and other experts said, but should not be used to set policy.

“Mathematically, it’s certainly possible to have herd immunity at these very, very low levels,” said Carl Bergstrom, an infectious disease expert at the University of Washington in Seattle. “Those are just our best guesses for what the numbers should look like.”

“But,” he added, “they’re just exactly that, guesses.”

But what about immunity at levels lower than those needed for herd immunity?

“Definitely the disease would not spread as well if it gets back into New York,” said Joel Miller, a mathematical modeler at La Trobe University in Australia. “The same level of behavior change will have more effect on the disease now than it did four months ago.”

Thinking of a city or country as composed of subgroups, demarcated by age, race and level of social activity, might also help governments protect those with the least immunity.

That perspective also might help put a renewed focus on groups who require the higher levels of immunity, because of greater exposure levels and other inequities, including Black and Latino residents, said Dr. Manoj Jain, an infectious disease expert at Emory University. “That’s where this info is very useful,” he said.

The models also suggest a vaccination strategy: Rather than uniformly vaccinate all groups, governments could identify and immunize those most likely to be exposed in “superspreader” events.

“Getting those people vaccinated first can lead to the greatest benefit,” said Dr. Michael Mina, an immunologist at Harvard University. “That alone could lead to herd immunity.”

Vaccination schemes for other pathogens have successfully exploited this approach. For example, when children were given the pneumococcal vaccine in the early 2000s, rates of bacterial pneumonia in the elderly rapidly dropped because of a “herd effect.”

Vaccines that offer just 50 percent protection are considered to be moderately effective, but at that efficiency, even a low herd immunity target would require that a large proportion of the population be immunized, Dr. Bergstrom noted.

If there are early reports of side effects that may scare away some people, he said, “we’d do well to start thinking about all that now.”

Back in Brooklyn, fewer than 1 percent of people tested at neighborhood clinics over the past eight weeks were infected with the virus. But there are still handfuls of cases, Ms. Marcus said, adding that her 10-year-old niece was in quarantine because a counselor at her day camp had tested positive.

“Sometimes that’s all you need, right?” she said. “I’m still hoping we don’t see what we had in March and April, but I’m not so sure that we’ve seen the end of it.”

The Plan That Could Give Us Our Lives Back


mina is a professor of epidemiology at Harvard, where he studies the diagnostic testing of infectious diseases. He has watched, with disgust and disbelief, as the United States has struggled for months to obtain enough tests to fight the coronavirus. In January, he assured a newspaper reporter that he had “absolute faith” in the ability of the Centers for Disease Control and Prevention to contain the virus. By early March, that conviction was in crisis. “The incompetence has really exceeded what anyone would expect,” he told The New York Times. His astonishment has only intensified since.

Many Americans may understand that testing has failed in this country—that it has been inadequate, in one form or another, since February. What they may not understand is that it is failing, now. In each of the past two weeks, and for the first time since the pandemic began, the country performed fewer COVID-19 tests than it did in the week prior. The system is deteriorating.

Testing is a non-optional problem. Tests permit us to do the most basic task in disease control: Identify the sick, and separate them from the well. When tests are abundant, they can dispel the fear of contagion that has quieted public life. “The only thing that makes a difference in the economy is public health, and the only thing that makes a difference in public health is testing,” Simon Johnson, the former chief economist of the International Monetary Fund, told us. Optimistic timelines suggest that vaccines won’t be widely available, in the hundreds of millions of doses, until May or June. There will be a transition period in which doctors and health-care workers are vaccinated, but teachers, letter carriers, and police officers are not. We will need better testing then. But we need it now, too.

Why has testing failed so completely? By the end of March, Mina had identified a culprit: “There’s little ability for a central command unit to pool all the resources from around the country,” he said at a Harvard event. “We have no way to centralize things in this country short of declaring martial law.” It took several more months for him to find a solution to this problem, which is to circumvent it altogether. In the past several weeks, he has become an evangelist for a total revolution in how the U.S. controls the pandemic. Instead of restructuring daily life around the American way of testing, he argues, the country should build testing into the American way of life.

The wand that will accomplish this feat is a thin paper strip, no longer than a finger. It is a coronavirus test. Mina says that the U.S. should mass-produce these inexpensive and relatively insensitive tests—unlike other methods, they require only a saliva sample—in quantities of tens of millions a day. These tests, which can deliver a result in 15 minutes or less, should then become a ubiquitous part of daily life. Before anyone enters a school or an office, a movie theater or a Walmart, they must take one of these tests. Test negative, and you may enter the public space. Test positive, and you are sent home. In other words: Mina wants to test nearly everyone, nearly every day.

The tests Mina describes already exist: They are sitting in the office of e25 Bio, a small start-up in Cambridge, Massachusetts; half a dozen other companies are working on similar products. But implementing his vision will require changing how we think about tests. These new tests are much less sensitive than the ones we run today, which means that regulations must be relaxed before they can be sold or used. Their closest analogue is rapid dengue-virus tests, used in India, which are manufactured in a quantity of 100 million a year. Mina envisions nearly as many rapid COVID-19 tests being manufactured a day. Only the federal government, acting as customer and controller, can accomplish such a feat.

If it is an audacious plan, it has an audacious payoff. Mina claims that his plan could bring the virus to heel in the U.S. within three weeks. (Other epidemiologists aren’t as sure it would work—at least without serious downsides.) His plan, while costly, is one of the few commensurate in scale to the pandemic: Even if it costs billions of dollars to realize, the U.S. is already losing billions of dollars to the virus every day. More Americans are dying of the coronavirus every month, on average, than died in the deadliest month of World War II. Donald Trump has said that the U.S. is fighting a “war” against an “invisible enemy”; Mina simply asks that the country adopt a wartime economy.

George Packer: We are living in a failed state

We have been covering coronavirus testing since March. For most of that time, the story has been one of failure after failure. But in the past few weeks, something has changed. After months without federal leadership, a loose confederation of scientists, economists, doctors, financiers, philanthropists, and public-health officials has assembled to fill in that gap. They have reexamined every piece of the testing system and developed a new set of tactics to address the months-long testing shortage. Mina’s plan is the most aggressive of these ideas; other groups—such as the new nonprofit Testing for America, founded by private-sector experts who helped the White House in the spring—have advanced their own plans. Taken together, they compose a box of tools that could allow the country to fix its ramshackle house.

The government has also done more in the past month to stimulate the creation of new kinds of tests than it has done in any period of the pandemic so far. The National Institutes of Health has awarded $248 million in grants to companies so that they can scale up alternate forms of COVID-19 testing as quickly as possible. The Centers for Medicare and Medicaid has begun to support the nascent testing market as well. This investment is belated and too meager—by comparison, the government has spent more than $8 billion on vaccine development—but it is significant.

If the new proposals make anything clear, it’s that it is in our power to have an abundance of tests within months—and to return life to normal, or something close to it, even before a vaccine is found. There is a way out of the pandemic.

Today, if you go to the doctor with a dry cough and fever, and get swabbed for COVID-19, you will probably receive a test that was not designed for an out-of-control pandemic. It’s called a “reverse-transcription polymerase chain reaction” test, or PCR, test, and it is one of the miracles of medicine. The PCR technique has allowed us to probe the genomes of the Earth: Its invention, in 1983, cleared the way for the Human Genome Project, the early diagnosis of certain cancers, and the study of ancient DNA. It works, in essence, like a zoom-and-enhance feature on a computer: Using a specific mix of chemicals, called “reagents,” and a special machine, called a “thermal cycler,” the PCR process duplicates a certain strand of genetic material hundreds of millions of times.

When used to test for COVID-19, the PCR technique looks for a specific sequence of nucleotides that is unique to the coronavirus, a snippet of RNA that exists nowhere else. Whenever the PCR machine—as designed and sold, for instance, by the multinational firm Roche—encounters that strand, it makes a copy of both that sequence and a fluorescent dye. If, after multiplying both the strand and the dye hundreds of millions of times, the Roche machine detects a certain amount of the dye, its software interprets the specimen as a positive. To have a “confirmed case of COVID-19” is to have a PCR machine detect the dye in a sample and report it to a technician. Tested time and time again, the PCR technique performs stunningly well: The best-in-class PCR tests can reliably detect, in just a few hours, as few as 100 copies of viral RNA in a milliliter of spit or snot.

The PCR test has anchored the American response to the pandemic. In CDC guidelines written by a council of state epidemiologists, a positive PCR result is the only way to confirm a case of COVID-19. And the Food and Drug Administration, which regulates all COVID-19 tests used in the U.S., judges every other type of test against PCR. Of the more than 62 million COVID-19 tests conducted in the U.S. since March, the overwhelming majority have been PCR.

However, a small but growing pile of clinical evidence—and a sky-scraping stack of real-world accounts—has revealed glaring issues with PCR tests. From a public-health perspective, the most important questions that a test can answer are: Is this person infected and contagious now? and If he’s not contagious, might he be soon? But these are not questions that even a positive PCR result can address. And especially as they’re conducted in the U.S. today, PCR tests do not tell us what we need to know to stop the virus.

Imagine that, at this instant, you are exposed to and infected with the coronavirus. You now have COVID-19—it is day zero—but it is impossible for you or anyone else to know it. In the following days, the virus will silently propagate in your body, hijacking your cells and making millions of copies of itself. Around day three of your infection, there might be enough of the virus in your nasal passages and saliva that a sample of either would test positive via PCR. Soon, your respiratory system will be so crowded with the virus that you will become contagious, spraying the virus into the air whenever you talk or yell. But you likely will not think yourself sick until around day five, when you start to develop symptoms, such as a fever, dry cough, or lost sense of smell. For the next few days, you will be at your most infectious.

And here is the first problem with PCR. To cut off a chain of transmission, public-health workers have to move faster than the virus. If they can test you early—around day three of your infection, for instance—and get a result back in a day or two, they may be able to isolate you before you infect too many people.

But right now, the U.S. is not delivering PCR results anywhere close to that fast. Brett Giroir, the federal coronavirus-testing czar, admitted to Congress last month that even a three-day turnaround time is “not a benchmark we can achieve today.” As an outbreak raged in Arizona this summer, some PCR results took 14 days or more to come back. That’s worse than useless—“I would not call that a test,” Johnson, the economist, told us—because most bouts of COVID-19 last 14 days or fewer. “The majority of all U.S. tests are completely garbage, wasted,” Bill Gates, who has helped fund COVID-19 testing, recently said.

After your symptoms start around day five, you might remain symptomatic for several days to several months. But some recent studies suggest that by day 14 or so—nine days after your symptoms began—you are no longer infectious, even if you are still symptomatic. By then, there is no longer live virus in your upper respiratory system. But because millions of dead virus particles line your mouth and nasal cavity, and because they contain strands of intact RNA, and because the PCR technique is very sensitive, you will still test positive on a PCR test. For weeks, in fact, you may test positive via PCR, even after your symptoms abate.

And here is PCR’s second problem: By this point in your illness, a positive PCR test does not mean what you might expect. It does not mean that you are infectious, nor does it necessarily mean that there is live SARS-CoV-2 virus in your body. It does not make sense to trace any contacts you’ve had in the past five days, because you did not infect them. Nor does it make sense for you to stay home from work. But our country’s public-health infrastructure cannot easily distinguish between a day-two positive and a day-35 positive.

The final issue with PCR tests is simple: There aren’t enough of them. The U.S. now runs more than 700,000 COVID-19 tests a day. On its own terms, this is a stupendous leap, a nearly 800-fold increase since early March. But we may be maxing out the world’s PCR capacity; supply chains are straining and snapping. For months, it has been difficult for labs to get the expensive chemical reagents that allow for RNA duplication. Earlier this summer, there was a global run on the tips of pipettes—the disposable plastic basters used to move liquid between vials. Sometimes the bottleneck is PCR machines themselves: As infections surged in Arizona last month, and people lined up to be tested, the number of tests far exceeded the machines’ capacity to run them.

When tests dwindle, the entire medical system suffers. In Arizona, many doctors’ offices were short-staffed at the peak of the outbreak, because any doctor exposed to the virus needed to test negative before returning to work, and the system simply couldn’t handle the volume of tests. “We’ve had people out seven to 10 days” waiting for a negative result, Catherine Gioannetti, the medical director of health and safety for Arizona Community Physicians, told us. “It’s essentially a broken system, because we don’t have results in a timely fashion.”

If labs don’t have the capacity to turn around doctors’ tests, which are often fast-tracked, they definitely do not have the capacity to test contagious people who are wholly asymptomatic. These silent spreaders may remain infectious for weeks but never develop any symptoms. They are the virus’s “secret power,” one testing executive told us, and they account for 20 to 40 percent of all infections. Some evidence suggests that they may be more infectious than symptomatic people, carrying higher viral loads for longer.

The challenge is clear: We need an enormous number of tests. As some have argued since the spring, the American population at large—and not just feverish, coughing people—has to be screened. Let’s say, for instance, that you wanted to test everyone in the U.S. once a week. That’s 45 million tests a day. How can we get there?

In the immediate future, the only way to increase testing is to squeeze more tests out of the existing PCR system. Our best bet to do so fast is through a technique called “pooling,” which could get a few hundred thousand more tests out of the system every day.

Pooling is straightforward: Instead of testing each sample individually, laboratories combine some samples, then test that “pooled” sample as one. The technique was invented by Robert Dorfman, a Harvard statistician, to test American soldiers for syphilis during World War II. Today it is commonly used by public-health labs to test for HIV. It works as follows: A lab technician mixes 50 HIV samples together, then tests this pool. If the result is negative, then none of the patients has HIV—and the researcher has evaluated 50 samples with the same materials it takes to run one test.

But if the pooled sample is positive, a new phase starts. The technician pools the same specimens again, this time into smaller groups of 10, and retests them. When one of these smaller pools is positive, she tests each individual sample in it. By the end of the process, she has tested 50 people for HIV, but used only a dozen or so tests. This approach saves her hundreds of tests over the course of a day.

Pooling is a great first step to maximizing our test supply, Jon Kolstad, an economist at UC Berkeley, told us. This is in part because regulators and public-health officials are already familiar with it. The FDA has told Quest Diagnostics, LabCorp, and BioReference, three major commercial laboratories, that they can start pooling a handful of coronavirus samples at a time. In some parts of New England that haven’t seen much of the virus, pooling could triple or even quadruple the number of tests available, a team at the University of Nebraska has found.

But pooling is only a stopgap. It works best for diseases that are relatively rare, such as HIV and syphilis. If a disease is too common, then the work of pooling—the laborious mixing and remixing of samples—is more work than it’s worth. (About twice as many Americans have been infected with the coronavirus as have contracted  HIV since 1981.) In Arizona and some southern states burning with COVID-19, traditional pooling would not be worth the effort, the same Nebraska team found.

Kolstad and Johnson, the MIT economist, are experimenting with ways to increase the efficiency of pooling. By grouping samples more deliberately, they can create larger pools of people with similar risksA group of office workers might be at lower risk than a group of meatpackers who work close together, and even within a meatpacking plant, workers on one side of the plant might be at greater risk than those on the other. And because pooling saves money, companies and colleges and schools could run more tests. This would create a virtuous cycle. Each day, a person has a certain chance of being infected that varies with the prevalence of the disease in a community. Test every day, and there is simply less time between tests in which a person could have been infected. This makes it possible to build larger pools of people who are likely negative.

Starting up these systems would require clearing logistical and regulatory hurdles—a positive coronavirus sample is a low-level biohazard, and the FDA regulates it as such. Dina Greene, who directs lab testing for Kaiser Permanente in Washington State, says that contamination problems are already difficult for labs to manage, and would be more so if labs have to manually mix together samples.

Kolstad has been thinking through this problem. His team is experimenting with a different technique, which one might call “intermediate pooling.” Instead of having labs make pools on the back end, Kolstad proposes deploying trained nurses in mobile pooling labs in retrofitted vans. It would work well for nursing homes, he says: The nurses might arrive at a certain time every day, test every employee, pool the samples in the van, and then drop them off at a nearby clinical lab. (Because the FDA regulates pooling in clinical labs more strictly than in this type of “surveillance testing,” it may also be easier to obtain FDA approval for this plan.) Kolstad and his team are trying out this technique with a network of nursing homes in the Boston area, and delivering the pooled tests to a nearly complete, fully automated COVID-19 testing facility run by Gingko Bioworks, a $4 billion start-up in Cambridge that is pitching another method to scale up U.S. testing, one that could vastly increase the pace of processing.

Since its founding in 2009, Ginkgo Bioworks has specialized in synthesizing new kinds of bacteria for use in industrial processes. Its engineers spin new forms of DNA using genetic-sequencing machines made by Illumina, a large and publicly traded biotechnology company. But in the spring, as viral testing buckled, Ginkgo’s engineers realized that their Illumina machines could be put to another use: Instead of creating genes, they could identify existing ones—and do so much faster than a PCR machine can.

Unlike PCR machines, which can analyze at most hundreds of individual samples per run, sequencing machines can read thousands of samples simultaneously. A high-end PCR machine, operated by a round-the-clock staff, can run up to 1,000 samples a day; a single Illumina machine can read more than 3,000 samples in half that time. Ginkgo has sharpened that advantage by building its fully automated factory in Boston, centered on Illumina machines, which it says could test about 250,000 samples a day. It aims to open the facility by mid-October; in two months more, another three could go up and Ginkgo could be testing 1 million samples each day.

The company has designed its supply chain to withstand high demand. It has rejected some reagents, for instance, because it doesn’t trust that there will be enough of them; it uses saliva samples, not swabbed nose or throat samples, because it does not think there are enough swabs in the world to meet demand. The genetic-sequencing supply chain is already built for such scale because other automated factories—doing noninvasive neonatal testing, for example—already use Illumina machines.

Both Ginkgo and an Illumina-backed start-up, Helix, have received NIH grants to rapidly scale up their testing. If the technique receives FDA approval, as many expect, the two companies could as much as triple the country’s testing capacity. “In three months, I think we could be at between 1 and 3 million additional tests per day in this country, without any problem at all,” John Stuelpnagel, a bioscience entrepreneur and one of the founders of Illumina, told us.

The approach has its challenges. Any samples must be shipped to one of Ginkgo’s or Helix’s centralized testing locations, which imposes a huge logistical obstacle to scaling up. The incumbent testing companies—Quest and LabCorp—have achieved dominance because of their ability to collect samples from places where they’re tested. But in Ginkgo’s full vision, 1 million tests will cover far more than 1 million people.

The key to this approach is “front-end pooling.” Imagine that every day, when kids arrive in their classroom, they briefly remove their mask and spit into a bag. (It is a perfect plan for second graders.) The bag would then be shipped to the nearest Ginkgo factory, which could test the pooled sample and deliver a single result for the classroom by the next morning. “If you pool together one classroom, and test that classroom together, then if you get a positive, you can send the whole classroom home,” Blythe Adamson, an economist and epidemiologist at the nonprofit Testing for America, told us. “For children, it protects their privacy—we don’t know which student” tested positive.

Front-end pooling could also drive costs down, partly by saving on materials. “Do 10 people spit in one bag? That’s one-tenth the cost,” Jason Kelly, Ginkgo’s chief executive, told us. “It’s logistically simpler, because one bag shows up, not 10, so there’s 10 times less unboxing, 10 times less robotic movement.” The challenge, he said, is chiefly one of industrial design, not molecular biology: There is no FDA-approved device, at present, that will let 10 kids safely spit into one vial. We should have federally backed development and fast regulatory approval for that kind of device, Kelly said.

The Ginkgo sequencing approach and front-end pooling have never been tried before, because they make sense only in a pandemic. Only at the scale of tens of thousands of tests do Ginkgo tests start to cost less than PCR, Adamson said. But at that scale, their cost drops quickly in comparison—possibly down to $20, Stuelpnagel said, if not $10, compared with more than $100 for a PCR test.

“You’d never do [any of this] for HIV,” Kelly said. “It’s only in a pandemic you go, ‘Oh my God, we’re undertesting by a factor of 10.’”

But what if testing needs to scale up not 10 times, but 20, or 50, or 100 times? That’s where another type of test—an antigen test—comes in.

At the same time that Ginkgo and other next-gen sequencing tests should come online, antigen tests will be scaling up. Unlike a PCR or a Ginkgo-style test, an antigen test does not identify any of the virus’s genetic material. Instead, it looks for an antigen, a slightly redundant name for any chemical that’s recognized by the test. Antigen tests aren’t as sensitive as genetic tests, but what they sacrifice in accuracy, they make up in speed, cost, and convenience. Most important, an antigen test can be conducted quickly at a “point of care” location, such as a doctor’s office, nursing home, or hospital.

Two of the most anticipated such tests are already on the market. Manufactured by two companies, Quidel and Becton, Dickinson and Company, they look for an antigen called “nucleocapsid,” which is plentiful in the SARS-CoV-2 virus. The companies say they will be making a combined 14 million tests a month by the end of September; for comparison, the U.S. completed 23 million total tests in July. This scale alone will make this type of test an important factor in fall testing. Hospitals and doctors told us they are eager to get their hands on antigen tests, in part because they’re worried about dealing with COVID-19 during the coming flu season. In years past, if a patient had a cough and a runny nose in December, she would likely be diagnosed with the flu, even if she tested negative on a rapid flu test. “But now we can’t presume [patients] have the flu,” because they might have COVID-19, says Natasha Bhuyan, the West Coast medical director for One Medical, a chain of primary-care clinics. An antigen test seems to offer a way out of this dilemma.

The tests cost less than half as much as standard PCR tests, and they don’t need to be sent away to a lab. They can deliver a result in 15 minutes. But this approach has downsides. While the tests work well enough, successfully identifying most people with high viral loads, they have sometimes delivered false positives. Last week, Ohio Governor Mike DeWine tested positive on the Quidel test, leading him to cancel a meeting with President Trump. But later that day, he tested negative, three times, when analyzed by PCR.

And while these tests will be useful, they have their own supply-chain drawbacks. Both companies’ tests can be interpreted only with a proprietary reader, and while many clinics and offices already have these readers on hand, neither company is prepared to mass-produce them at the same scale as the tests. (Quidel now makes 2,000 of its readers a month, but is aiming to scale to 7,000 a month by September, a spokesperson told us.) Because both tests look for nucleocapsid, which exists only inside the coronavirus, they need a way to sever the virus’s outer membrane. This requires more reagents. For many technicians, these drawbacks aren’t worth the benefits. “Most people who are real lab experts are steering away from all that stuff because they can’t justify it,” Greene, the Kaiser lab director, said.

The readers are a particular sticking point for Michael Mina, the Harvard epidemiologist. He calls the BD and Quidel systems “Nespresso tests,” because, just as a Nespresso pod can transform into coffee only through a Nespresso brewer, they can deliver results only when their readers are at hand. “What I want is the instant coffee of tests,” he told us. What if there was an antigen test that could be made in huge numbers and didn’t require a specialized reader? What if it worked more like a pregnancy test—a procedure you can do at home, and not only at a doctor’s office?

Such tests exist—and have existed since April—and they are made by e25 Bio, a 12-person company in Cambridge. An e25 test is a paper strip, a few inches long and less than an inch wide. It needs only some spit, a saline solution, and a small cup—and it can deliver a result in 15 minutes. Like a pregnancy test, the strip has a faint line across its lower third. If you expose the strip to a sample and it fills in with color, then the test is positive. It does not require a machine, a reagent, or a doctor to work.

Its unusual quality is that it does not look for the same antigen as other tests. Instead of identifying nucleocapsid, the e25 test is keyed to something on the outside of the virus. It reacts to the presence of the coronavirus’s distinctive spike protein, the structure on the virus’s “skin” that allows it to hook onto and enter human cells. “I think we’re the only company in North America that has developed a spike antigen test,” Bobby Brooke Herrera, e25’s co-founder and chief executive, told us.

This has several advantages. It means, first, that the e25 test does not have to rupture the virus, which is why it doesn’t need reagents. And it means, second, that the e25 test is actually looking for something more relevant than the virus’s genetic material. The spike protein is the coronavirus’s most important structure—it plays a large part in determining the virus’s infectiousness, and it’s what both antibodies and many vaccine prototypes target—and its presence is a good proxy for the health of the virus generally. “We’ve developed our test to detect live viruses, or, in other words, spike protein,” Herrera said.

Working with two manufacturers, e25 thinks that it could make 4 million tests a month as soon as it receives FDA approval. Within six weeks of approval, it could make 20 million to 40 million tests a month. In short, e25 could single-handedly add as many as 1.2 million tests a day to the national total.

But FDA approval has not yet arrived, because the FDA compares every test to PCR, and no antigen test, however advanced, can stand up to the accuracy and sensitivity of the PCR technique. “The FDA, early on in the outbreak, said we had to follow a rubric of 80 percent sensitivity compared to PCR. How they got that number, I’m uncertain, but my best guess is it came from influenza epidemics in the past,” Herrera said.

This requirement has made antigen tests worse, Herrera argues, because it causes manufacturers to prioritize sensitivity at the cost of speed or convenience. It’s why other antigen tests use readers, or centrifuges, or look for nucleocapsid, he contends. By slightly weakening those guidelines, to 60 or 70 percent sensitivity, the FDA could let cheaper at-home tests come to market. The models that e25 uses show that even an at-home test that caught 50 percent of positives and 90 percent of negatives could detect outbreaks and reduce COVID-19 transmission.

Recall the coronavirus’s infection clock—how, from day zero of an infection to day five, the amount of the virus in your system exponentially increases; how it begins to ebb with the onset of symptoms; how, by day 14 or so, the PCR test is likely detecting only the refuse RNA of dead virus. While antigen tests need the equivalent of 100,000 viral strands per milliliter, a typical PCR test can detect a positive from as little as 1,000 strands per milliliter. There is only about a day at the beginning of an infection when the two tests would give different results—when there are more than 1,000 viral strands per milliliter of your saliva or snot but fewer than 100,000, according to Dan Larremore, a mathematician at the University of Colorado at Boulder. During that period—approximately day two or day three of an infection—antigen tests are truly inferior to PCR tests.

Yet the opposite is true as COVID-19 fades: There are potentially weeks at the end of an infection when there is enough viral RNA to clear the threshold for a positive PCR test but not enough to set off an antigen test. During that period, antigen tests, such as e25’s, outperform PCR tests, Mina argues, because they identify only people who are still contagious. So why, he asks, are they judged against PCR tests—and kept off the market—for failing to find the virus when there is no intact virus to find?

Antigen tests are not better than PCR tests in every instance. When someone at a hospital presents with severe COVID-19-like symptoms, for example, health-care workers cannot risk a false negative: They will need a PCR test. Some experts worry that at-home tests will have a much lower accuracy rate than advertised. Laboratory tests are conducted by professionals on machines they are familiar with, but amateurs will conduct at-home tests, which risks introducing errors not captured by official ratings or even imagined by regulators. At a national scale, this could mean that someone might have COVID-19, fail to realize it, and infect other people. “What’s concerning is the salami slicing of sensitivity. A percent here, a percent there, and pretty soon you’re talking real people,” Alex Greninger, a laboratory-medicine professor at the University of Washington, told us. Jennifer Nuzzo, an epidemiologist at the Johns Hopkins Center for Health Security, told us that it’s not yet clear whether people who receive a positive result on an at-home test will report that information to health authorities and choose to self-isolate.

But given that they are cheaper than PCR tests, have a faster turnaround time, and can be conducted at home, these paper tests do seem different, in a useful way. In some cases, they answer a more helpful question than PCR tests. There is good evidence to infer that a high viral load, which is what antigen tests detect, is correlated with infectiousness. The more virus in your body, the more contagious you are.

In that light, paper antigen tests aren’t SARS-CoV-2 tests at all, not like PCR tests are. They are rapid, cheap COVID-19 contagiousness tests. That shift in thinking, Mina argues, should undergird a shift in our national strategy.

Mina wants to coat the country in COVID-19 contagiousness tests. To understand the scale of his vision, start with the closest American analogue, the ubiquitous, paper-based, inexpensive at-home pregnancy test. Americans use 20 million of those each year. This is not sufficient for Mina’s plan. “Ideally, we’re making way more than 20 million [paper tests] a day,” Mina said. Entering a grocery store? Take a test first. Getting on a flight? There’s a test station at the gate. Going to work? Free coffee is provided with your mandatory test. He began pitching the idea as a moonshot in July, but it quickly took hold. By the end of the month, Howard Bauchner, the editor in chief of The Journal of the American Medical Associationgushed on a podcast that ubiquitous tests were “the best way we can get back to a semblance of working society.”

The idea has gained other advocates. Last month, a panel of experts convened by the Rockefeller Foundation called for the U.S. to do 3.5 million rapid antigen tests a day, or 25 million a week—five times more than the number of PCR tests they recommended. The researchers compiled a list of 12 rapid tests in development, including e25’s, and called for an aggressive government-led effort to support them. (The Rockefeller Foundation has also provided funding to the COVID Tracking Project at The Atlantic.) “These sort of tests are on the horizon, but getting them into the hands of everyone who needs them—schools, employers, health providers, public essential workers, vulnerable communities—will require the muscle that only the federal government can provide,” the experts wrote.

The muscle, specifically, of a wartime economy. The experts called for the White House to invoke the Defense Production Act, a Truman-era law that allows the federal government to compel companies to mass-produce goods in moments of national crisis. (Manufacturers are compensated for their effort at a fair price.) Only naked federal authority could push production fast enough to make enough tests in time to curb the virus, they wrote.

Herrera, the e25 executive, has been waiting for months for the government to invoke such power. There is essentially no resource constraint on the raw materials that make up antigen tests, but there is a profound limit to available productive capacity. “Being able to manufacture these products,” Herrera said, “is where the bottleneck lies.” And after it has the tests, Herrera believes, the company will need help sending them where they’re most needed. If testing companies are to save the world, they need federal support to do it.

And here is the tragedy—and the promise—of Mina’s moonshot: To fix testing, the federal government must do exactly what it has declined to do so far. Why is testing still a problem? Partly because the CDC and the FDA bickered in February and delayed by weeks the initial rollout of COVID-19 tests. Partly because infections continued to grow in the spring and summer, further boosting the number of tests needed to track the virus. But those reasons alone still do not explain the fundamental issue: Why has the U.S. never, not since the pandemic began, had enough tests?

The answer is because the Trump administration has addressed the lack of testing as if it is a nuisance, not a national-security threat. In March and April, the White House encouraged as many different PCR companies to sell COVID-19 tests as possible, declining to endorse any one option. While this idea allowed for competition in theory, it was a nightmare in practice. It effectively forced major labs to invest in several different types of PCR machines at the same time, and to be ready to switch among them as needed, lest a reagent run short. Today, the government cannot use the Defense Production Act to remedy the shortage of PCR machines or reagents—because the private labs running the tests are too invested in too many different machines.

Because of its trust in PCR, and its assumption that the pandemic would quickly abate, the administration also failed to encourage companies with alternative testing technologies to develop their products. Many companies that could have started work in April waited on the sidelines, because it wasn’t clear whether investing in COVID-19 testing would make sense, Sri Kosaraju, a member of the Testing for America governing council and a former director at JP Morgan, told us.

The Trump administration hoped that the free market would right this imbalance. But firms had no incentive to invest in testing, or assurance that their investments would pay off. Consider the high costs of building an automated testing factory, as Ginkgo is doing, said Stuelpnagel, the Illumina co-founder. A company would typically amortize the costs of that investment over three to five years. But that calculation breaks down in the pandemic. “There’s no way that we’re doing high-throughput COVID testing five years from now. And I hope there’s not COVID testing being done three years from now that would require this scale of lab,” he said. Companies aren’t built to deal with that level of uncertainty, or to serve a market that would dramatically shrink, or disappear altogether, if their product did its job. Even if the experimentation would benefit the public, it doesn’t make sense for individual businesses to take on those risks.

So nothing happened—for months. Only in the past few weeks has the federal government begun to address these concerns. The NIH grants awarded to Ginkgo, Helix, Quidel, and others were aimed, in part, at providing capital that would let businesses scale up quickly. And the Centers for Medicare and Medicaid has started to ensure that demand will exist for an experimental test: It has promised to buy Quidel or BD tests for every nursing home in the country.

But even if those companies succeed in delivering what they’ve promised, life will not go back to normal. An extra 1 million tests a day will allow us to ramp up contact-tracing operations and slow down the virus, but they will not change the texture of daily life in the pandemic, especially if there is another resurgence of the virus in the winter. For that, Mina’s moonshot is required. It will require much more than the $200 million the federal government has invested in testing technology so far, and it will require the full might of the federal government, with its unique ability to coerce manufacturing capacity. But its costs are not astronomical. If every paper test costs $1, as Mina hopes, and every American takes a test once a week, then his plan will cost about $1.5 billion a month. Congress has already authorized at least $7 billion to fix testing that the Trump administration had declined, for months, to spend. And even if Mina’s plan cost $300 million each day, the annual expense would amount to a fraction—about 3 percent—of the more than $3 trillion Congress has already spent dealing with the economic fallout of the pandemic. Yet the plan wouldn’t merely mitigate the harm of the pandemic. It could end it. To escape the pandemic in this way, the U.S. must make hundreds of millions of contagiousness tests—tests that are not perfect, but just good enough.

Mass-producing a cheap thing fast is, as it happens, something the United States is very good at, and something this country has done before. During the Second World War, the U.S. realized that the most effective way of shipping goods to Europe was not to use the fastest ship, but to use cheap “Liberty ships,” which were easy to mass-produce. The Allies “created this model of a ship that was kind of cheap, not as fast as they could make it, and not as good as they could make it,” Mark Wilson, a historian at the University of North Carolina at Charlotte, told us. “They were building cheap—one might say disposable—ships. They weren’t very good. But they just wanted to out-volume their opponents.”

We must out-volume the virus, and what will matter is not the strength of any one individual ship, but the strength of the system it is part of. When the FDA regulates tests, though, it looks at the sensitivity and specificity of a single test—how well the test identifies illness in an individual—not at how the test is part of a testing regimen meant to protect society. For this reason, Mina proposes that the FDA make room for the CDC or the NIH to oversee the use of contagiousness tests. “I think the CDC could potentially create a certification process really simply. They are the public-health agency, and could say, ‘We will evaluate different manufacturers. None of these will be fully regulated by law, but here are the ones you should or should not choose.’”

Paper tests do have downsides. Testing tens of millions of people each day would be an unprecedented biotechnical intervention in the country, and it might have unpredictable, nasty side effects. Mina’s plan is “being pushed without really thinking through the operational consequences,” Nuzzo said on a recent press call. Brett Giroir, the federal testing czar, has worried that a deluge of positive paper tests could lead asymptomatic people to swamp the rest of the medical system. “You do not beat the virus by shotgun testing everybody, all the time,” he said on the same call. Paper tests are based on an inference about human behavior. For example, if people knew that every paper test would catch only seven or eight infections out of every 10 (compared with PCR, which would catch all 10), would they keep taking them? Would the country’s testing system split in two, delivering PCR tests for the rich and cheap paper tests for the poor? Each way of testing for the virus is not only a technology or a medical device. Each is its own hypothesis about public health, human behavior, and market forces.

So here is what May 2021 could look like: Vaccines are rolling out. You haven’t gotten your dose yet, but you are no longer social distancing. When your daughter walks into her classroom, she briefly removes her mask and spits into a plastic bag; so do all the other children and the teacher. The bag is then driven across three states and delivered to the nearest Ginkgo processing facility. When you arrive at work, you spit into a plastic cup, then step outside to drink coffee. In 15 minutes, you get a text: You passed your daily screen and may proceed into the office. You still wear your mask at your desk, and you try to avoid common areas, but local infection levels are down in the single digits. That night, you and your family meet your parents at a restaurant, and before you proceed inside, you all take another contagiousness test. It’s normal, now, to see the little cups of saliva and saline solution, each holding a strip of color-changing paper, sitting on tables near the entrance of every public place. And before you fall asleep, you get a text message from the school district. Nobody in your daughter’s class tested positive this morning—instruction can happen in person tomorrow.

There is no technical obstacle to that vision. There is only a dearth of political will. “The lack of testing is a motivation problem,” Stuelpnagel said. “It’s going to take a lot of effort, but it should take a lot of effort, and we should be willing to take that effort.” Mina is frustrated that the answer is so close, and so doable, but not yet something the government is considering. “Let’s make the all-star team of people in this field, pay them whatever they need to be paid, put billions of dollars in, and get a working test in a month that could be truly scalable. Take it out of the free-market, capitalistic world and say: ‘This is a national emergency’—which,” he said, “it is.”

Covid-19 Drug Research Is a Big Huge Mess

THE PANDEMIC DISEASE Covid-19 does much more to the human body than a typical respiratory virus. In addition to neurological problems ranging from a loss of sense of smell to outright seizures, surprising gastrointestinal symptoms and kidney damage, and a potentially fatal haywire immune response, the disease also messes with a person’s blood. The sickest people start forming clots, potentially leading to stroke, heart attack, lung damage … it’s a mess. Physicians started noticing all this early in the pandemic, of course. The question was—and remains—what to do about it all.

“So, someone comes into the hospital and needs a blood-thinning medication to keep them from clotting,” says Linda Wang, a cardiologist who specializes in that problem—it’s called anticoagulation—at Duke Clinical Research Institute. But which patients would benefit the most? Which drug should they get? How much? When? Figuring out that kind of thing is the foundational behind-the-scenes work of medicine, where clinical trials of protocols and medications connect with on-the-ground clinical work. Except, when it came to anticoagulants and Covid-19, the research hasn’t happened yet. “Each hospital tried to develop their own protocol,” Wang says. “Could we have joined up hospital networks and developed a coordinated anticoagulant regimen? Or, if we can’t agree, develop two or three regimens and compare those?”

Yes. Well, they could have. Researchers across multiple medical centers could have done just what Wang proposes here, recruited thousands of patient-volunteers and then randomly assigned them to get either a treatment or not (that’d be a control group), with neither scientists nor participants knowing who was getting what until they look at the final data. That’s called a double-blind, randomized, controlled clinical trial—an RCT. Not to be too blunt about it, but instead of dithering about the evidence, they could have just counted how many people recovered and how many people died. That’s using mortality as an endpoint, in the lingo of the field. And then they’d know what works. “But we didn’t do that. Instead, every hospital just launched its own variation on things. And at the end of the day, we couldn’t study this in a multi-center, large-scale fashion,” Wang says. “We couldn’t even agree on how to measure the outcome.”

Things might get better; better studies of anticoagulants are in the works. But for now, eight months into a global pandemic, physicians and scientists still don’t know a hell of a lot about how to fight it. The story of anticoagulants is also the story of hydroxychloroquine, which took months to knock down as a preventative or a treatment, and convalescent plasma taken from the blood of recovered patients, which—while promising—has been the subject of a fragmented, delayed research effort. Doctors know that the steroid dexamethasone gets people out of the hospital quicker. They know that keeping really sick people on their fronts instead of their backs helps. They know that a couple other familiar antiviral drugs don’t work. And despite hundreds of millions of dollars in research money, tens of thousands of volunteer subjects, and the diligent spadework of thousands of researchers, that’s basically it. With a couple of important exceptions, we’re all still kind of clueless.Exclusive Offer.Don’t miss the future. Get 1 year for $5.Subscribe Now

Which leads to an important question: What the hell? It turns out that while scientists have been working crazy hard to figure all this out—to evaluate old drugs, test protocols, find new approaches—from a statistical and evidentiary perspective, they’ve mostly been spinning their wheels. The large-scale trial that has arguably done the most important work, the United Kingdom-based Randomised Evaluation of Covid-19 Therapy (Recovery) trial, spun up in part via the authority and infrastructure of the UK’s National Health Service. In the US, a lack of central planning, methodological obstacles, and professional pressures meant that since the pandemic began, everyone raced off at top speed, but in different directions, producing incompatible, unusable, or incoherent results—if they got results at all.Get WIRED AccessSUBSCRIBEMost Popular

An examination in July in the journal JAMA Internal Medicine proves it. Of 1,551 Covid-19 studies entered into the US registry between March and May, the study says, just over half were randomized clinical trials, and only 10 percent of those were double-blinded and had more than 100 participants. A fifth of the RCTs were on hydroxychloroquine or its cousin chloroquine, a dead end. The vast majority of the trials in the registry—76 percent—covered just one hospital. Only a third used mortality as an endpoint. Only 13 percent of the observational studies were prospective—that is, they looked forward instead of merely reviewing what happened, a significant weakness.

These results extend an analysis by Stat in July which showed that of studies begun or planned in since January, one in six were on the chloroquine family. That meant that of the 685,000 patient volunteers who anyone planned to enroll in any study of Covid-19, 237,000—more than one in three—were captured by studies of those drugs and therefore nothing else. Overall, less than 40 percent of all those trials would have the statistical power to conclude anything meaningful.

None of this was bad science, exactly. It’s all in bounds, nothing unethical or methodologically unsound. The observational studies, even the retrospective ones, were critical in drawing the outlines of the pandemic and its manifestations. The problem is, taken in aggregate, most of that science didn’t change policy. It doesn’t save any lives. “We had this massive activation of clinical research worldwide,” says Mintu Turakhia, a cardiologist at Stanford University and the lead author of the JAMA Internal Medicine paper. “That’s all great. But the problem is, we didn’t really have a strong sense of what kind of evidence to expect and where we will be after the first wave of research.”

The result of the non-results? “There aren’t that many studies that are going to move the needle in terms of generating evidence,” Turakhia says. “You can comment on the public health response, but the scientific response lagged also. Especially in the US, we just haven’t activated the machinery.”

Part of what went wrong seems to have been a strategic misstep. The science funding agencies of the US, like the National Institutes of Health, have announced efforts to set up the kind of large-scale, multi-arm “adaptive” trials that most researchers agree are the way to get big, world-denting results. But so far the US iteration, the Adaptive Covid-19 Treatment trial, has only conclusively shown that the (expensive) drug remdesivir, made by the pharmaceutical company Gilead, can reduce the time Covid-19 patients spend in the hospital. That’s a perfectly fine endpoint—no shenanigans involved—but it’s not mortality. Meanwhile the UK’s Recovery trial has shown that the steroid drug dexamethasone saves lives, and that the AIDS drugs lopinavir and ritonavir and the antimalarial and autoimmune drug hydroxychloroquine do not. Those drugs all looked like they might help from the early days of the pandemic, they’re all relatively cheap, and those findings changed the global standard of care.

US agencies and research centers may yet mount more coordinated efforts, but, like, tick-tock, y’all. In the US, political actors like the president talked up hydroxychloroquine before a central effort could take off, bolluxing research already in progress. The same thing seems to be happening with convalescent plasma—already another arm of the Recovery trial, not for nothin’. “I think the clinical trial enterprise, certainly in the US, was not designed for speed and for a pandemic,” says Walid Gellad, director of the Center for Pharmaceutical Policy and Prescribing at the University of Pittsburgh. “We’re seeing the results of that, basically, in what we see now.”Most Popular

Another problem has been the kind of internecine push-and-pull among hospitals and individual researchers. They’re all frenemies, all chasing the goal of helping people, but also getting published in big journals in pursuit of tenure and grants. That’s not necessarily bad—if the energy gets directed. “The lowest-lift study you can do as a clinician scientist is to write up the cases that come through your center. It’s not that hard to do, and it’s a low lift. But if you want impact, you’ve got to get over that,” Turakhia says. “We have to get away from academic opportunism, just so you have a paper, and figure out how to get together and work collaboratively.”

That opportunism isn’t just ambition. It actually risks disrespecting (if not outright harming) patients. “When we do clinical research, it isn’t just a researcher saying, ‘Here’s a good idea, let’s go do it.’ Research is a high-stakes endeavor for all of us. Our patients are volunteering, in most cases, to be parts of these studies, contributing data and their bodies to help us advance knowledge. There’s a cost to doing research,” says Wang, who wrote a commentary that ran alongside the JAMA Internal Medicine article. “Wouldn’t it seem possible, especially in this age of communication and technology, to be more efficient early on?”

Gellad takes an even harder line. “Every little group was doing its own trial rather than having an organized, central effort to say, ‘These are the most important central efforts. These are the trials we’re going to do,’” he says.

Blame the system, if you want. Big therapeutic trials are expensive, so only pharmaceutical companies and governments tend to have the bank accounts to pull them off. A whole grab bag of potential funders, from the NIH to the Gates Foundation and on and on, pulls researchers in many directions. A lack of central patient data means that even when hospital systems and researchers want to collaborate, it’s hard for them to talk to each other, digitally speaking. The mechanisms for protecting patients’ rights and keeping them safe during research trials are scattered and independent; no one is suggesting eliminating the institutional review boards at individual hospitals and research centers, but a big study protocol might have to deal with dozens of them, each one with veto power. And in the end, as the reporter Susan Dominus shows in a recent article in The New York Times Magazine, hospitalists and clinicians might feel that their duty to patients means they should try anything and everything to save their lives, rather than enroll them in studies that might randomize them to the control group (even though the study might eventually save more lives overall).

These problems have always challenged drug trials and the people who mount them. As with so many system failures, the pandemic has only made the issue worse. “There is no doubt we lack any sort of organized and systematic approach to testing therapeutic ideas,” says Peter Bach, director of the Center for Health Policy and Outcomes and the Drug Pricing Lab at Memorial Sloan Kettering Cancer Center. Bach says that small trials that risk false positive results, studies that use squishy outcomes instead of mortality, and all the other weaknesses that lead to biased results and lack of generalizability are obviously bad, “but I don’t know what to say other than it is always like this, really.”

Exposing these problems might provide the incentive and ideas to fix them. Turakhia thinks a solution—maybe for the next pandemic—would be a whole network of centers ready to mount clinical trials at a moment’s notice. Just fill in the nouns on the paperwork. “We need a bunch of sites that are a priori ready to go. ‘We’ve signed off, the IRBs have a fast-track mechanism,’” he says. “You just need the right infrastructure and the buy-in and commitment to the vision. The operational aspects, the approvals, and all that—you can get all that up and running.”

It’s the kind of system that could actually make the world better, if someone builds it. “All of us agree there is an imperative to do this, and time is of the essence,” Wang says. “Now we just need to make the machine that makes this run a little faster. And I’m certain this machine will persist after the pandemic.” The shift from spinning wheels to gears turning in sync won’t be an easy lift, but it’s clearly a necessary one.

Scientists May Be Using the Wrong Cells to Study Covid-19

BY NOW THERE’S little doubt about hydroxychloroquine: It doesn’t work for treating Covid-19. But there’s a bigger, more important lesson hidden in the story of its failure—a rarely mentioned, but altogether crucial, error baked into the early research. The scientists who ran the first, promising laboratory experiments on the drug had used the wrong kind of cells. Instead of testing its effects on human lung cells, they relied on a supply of mass-produced, standardized cells made from a monkey’s kidney. In the end, that poor decision made their findings more or less irrelevant to human health. Worse, it’s possible that further research into novel Covid-19 cures will end up being compromised by the same mistake.SUBSCRIBE

The problem began in early February, when the scientific journal Cell Research published data from the Wuhan Institute of Virology suggesting that a pharmaceutical cousin of hydroxychloroquine was “highly effective” at controlling infections with SARS-CoV-2, the virus that causes Covid-19. (In 2005, lab tests of the same drug found it could inhibit the coronavirus that caused the original SARS outbreak.) A separate, full-fledged study from a different Chinese group, which appeared in the journal Clinical Infectious Diseases on March 9, found hydroxychloroquine to be more potent, and has since been cited hundreds of times. About a week after that, a third journal, Cell Discovery, put out the results from another study by the Wuhan group, which concluded that hydroxychloroquine, in particular, “has a good potential to combat the disease.”

The experiments described in those three papers—and also the one on SARS from 15 years ago—all suffered from the same problem: They started with a set of kidney cells that traces back to an African green monkey dissected in the spring of 1962.

It’s not at all uncommon for individual scientists—or even entire subfields of research—to waste their time in just this way, by choosing the most familiar animal or “model system” as the basis for their work, even when it’s not well suited to the question at hand. Rodent research findings, for example, have been notoriously misleading on a number of important topics, including potential treatments for amyotrophic lateral sclerosis (ALS) and tuberculosis. Cell lines, too, can be misapplied out of habit or convenience. The ones derived from an African green monkey kidney, known as Vero cells, are especially popular among virologists, in part because they contain fewer antiviral proteins known as interferons than other cells, and thus provide a fertile breeding ground for certain viruses that are otherwise quite fickle and difficult to grow in the lab. (Lab mice have long been used to study cancer for the same reason: They happen to be startlingly adept at getting tumors.)

The experiments all suffered from the same problem: They started with a set of kidney cells that traces back to an African green monkey dissected in the spring of 1962.

Vero cells have at times been indispensable to the study of coronaviruses. One of the earliest papers to identify the pathogen behind the 2003 SARS outbreak, for instance, used this kind of cell to grow the virus so that scientists could study it in detail. But the cells’ more recent use provides a cautionary tale. Whereas hydroxychloroquine does appear to stop SARS-CoV-2 from infecting Vero cells, it fails to do the same for human lung cells in a dish. According to research from Stefan Pöhlmann, head of the Infection Biology Unit at the German Primate Center in Göttingen, and his collaborators, the devil was in the details of how the cells interact with the SARS-CoV-2’s dreaded ‘spike’ protein. Human lung cells contain at least two different enzymes that can help the virus sneak through their membranes. With Vero cells, however, only one of those modes of entry is available—and it turns out to be the one that hydroxychloroquine will block. Pöhlmann and his team published the results in the journal Nature on July 22. For him, it’s a clear example of why using human lung cells is really important in studying this pandemic virus. Vero cells should be “handled with caution,” Pöhlmann says. “It’s true that the Vero cells are very popular. But unfortunately for this particular aspect of Covid-19 research, they are absolutely not useful. I think this is now clear to the field.”Get WIRED AccessSUBSCRIBEMost Popular

The notion that doctors might once more forge ahead with ad hoc treatments based on nothing more than Vero-cell results worries Vincent Racaniello, a microbiologist at Columbia University in New York and host of the popular TWiV virology podcast. Racaniello says that he received a message from someone who was excited about a study posted to a preprint server that suggested the antidepressant Prozac inhibits the SARS-CoV-2 virus—but this work, too, was done in Vero cells. He was unimpressed.

Both Racaniello and Pöhlmann note that Vero cells might be fine for investigating drugs that work by slowing or stopping the virus from making more copies once it’s already inside a cell. But they say that not all scientists are aware of those nuances. There are plenty who have abandoned their main areas of expertise to explore Covid-19, including virologists who have never worked with coronaviruses before. Madhu Pai, epidemiologist and director of the Global Health Program and TB Centre at McGill University in Montreal, has described this as the “Covidization of research,” where “well-intentioned scientists with real expertise in one field intrude into another, passing judgment where they lack expert-level training and insight.” The result, he says, is to “open the door to big mistakes with bad consequences.”

Concerns about the overreliance on Vero cells in this pandemic are not limited to the study of potential therapies. Experts have also cautioned against Vero-based research into how efficiently the new coronavirus infects its hosts, and whether patients with positive Covid test results are necessarily contagious after the first week or so of illness

Pöhlmann expects that academic journals will start paying more attention to the flaws of Vero cells—and demanding evidence from human lung cells—when reviewing studies of potential therapies for Covid-19. But this vetting is even less likely to occur for preprint papers, so journalists and the public should be vigilant as well.

All this hand-wringing over Vero cells isn’t just a minor scientific quibble. Sure, more than six months into this pandemic, scientists finally figured out why hydroxychloroquine was a bust, but that only happened after massive amounts of time and money were spent on clinical trials of a drug with potentially fatal side effects. The US scrambled to secure a stockpile of hydroxychloroquine pills; by mid-June it had distributed 31 million of them to state and local health departments. Meanwhile, at least one-third of lupus patients in the US who take hydroxychloroquine had trouble accessing the drug—which is known and approved to treat their condition—because of shortages resulting from the Covid-19 hype. We would have avoided all of this if scientists had double-checked their Vero findings in human lung cells before racing to publish their results. The cells that scientists select are tiny, but their repercussions are anything but small.

How to Think Like an Epidemiologist

Don’t worry, a little Bayesian analysis won’t hurt you.

There is a statistician’s rejoinder — sometimes offered as wry criticism, sometimes as honest advice — that could hardly be a better motto for our times: “Update your priors!”

In stats lingo, “priors” are your prior knowledge and beliefs, inevitably fuzzy and uncertain, before seeing evidence. Evidence prompts an updating; and then more evidence prompts further updating, so forth and so on. This iterative process hones greater certainty and generates a coherent accumulation of knowledge.

In the early pandemic era, for instance, airborne transmission of Covid-19 was not considered likely, but in early July the World Health Organization, with mounting scientific evidence, conceded that it is a factor, especially indoors. The W.H.O. updated its priors, and changed its advice.

This is the heart of Bayesian analysis, named after Thomas Bayes, an 18th-century Presbyterian minister who did math on the side. It captures uncertainty in terms of probability: Bayes’s theorem, or rule, is a device for rationally updating your prior beliefs and uncertainties based on observed evidence.

Reverend Bayes set out his ideas in “An Essay Toward Solving a Problem in the Doctrine of Chances,” published posthumously in 1763; it was refined by the preacher and mathematician Richard Price and included Bayes’s theorem. A couple of centuries later, Bayesian frameworks and methods, powered by computation, are at the heart of various models in epidemiology and other scientific fields

As Marc Lipsitch, an infectious disease epidemiologist at Harvard, noted on Twitter, Bayesian reasoning comes awfully close to his working definition of rationality. “As we learn more, our beliefs should change,” Dr. Lipsitch said in an interview. “One extreme is to decide what you think and be impervious to new information. Another extreme is to over-privilege the last thing you learned. In rough terms, Bayesian reasoning is a principled way to integrate what you previously thought with what you have learned and come to a conclusion that incorporates them both, giving them appropriate weights.”

With a new disease like Covid-19 and all the uncertainties it brings, there is intense interest in nailing down the parameters for models: What is the basic reproduction number, the rate at which new cases arise? How deadly is it? What is the infection fatality rate, the proportion of people with the virus that it kills?

But there is little point in trying to establish fixed numbers, said Natalie Dean, an assistant professor of biostatistics at the University of Florida.

“We should be less focused on finding the single ‘truth’ and more focused on establishing a reasonable range, recognizing that the true value may vary across populations,” Dr. Dean said. “Bayesian analyses allow us to include this variability in a clear way, and then propagate this uncertainty through the model.”

A textbook application of Bayes theorem is serology testing for Covid-19, which looks for the presence of antibodies to the virus. All tests are imperfect, and the accuracy of an antibody test turns on many factors including, critically, the rarity or prevalence of the disease.

The first SARS-CoV-2 antibody test approved by the F.D.A., in April, seemed to be wrong as often as it was right. With Bayes theorem, you can calculate what you really want to know: the probability that the test result is correct. As one commenter on Twitter put it: “Understanding Bayes’ theorem is a matter of life and death right now.”

Joseph Blitzstein, a statistician at Harvard, delves into the utility of Bayesian analysis in his popular course “Statistics 110: Probability.” For a primer, in lecture one, he says: “Math is the logic of certainty, and statistics is the logic of uncertainty. Everyone has uncertainty. If you have 100 percent certainty about everything, there is something wrong with you.”

By the end of lecture four, he arrives at Bayes’s theorem — his favorite theorem because it is mathematically simple yet conceptually powerful.

“Literally, the proof is just one line of algebra,” Dr. Blitzstein said. The theorem essentially reduces to a fraction; it expresses the probability P of some event A happening given the occurrence of another event B.

“Naïvely, you would think, How much could you get from that?” Dr. Blitzstein said. “It turns out to have incredibly deep consequences and to be applicable to just about every field of inquiry” — from finance and genetics to political science and historical studies. The Bayesian approach is applied in analyzing racial disparities in policing (in the assessment of officer decisions to search drivers during a traffic stop) and search-and-rescue operations (the search area narrows as new data is added). Cognitive scientists ask, ‘Is the brain Bayesian?’ Philosophers of science posit that science as a whole is a Bayesian process — as is common sense.

Take diagnostic testing. In this scenario, the setup of Bayes’s theorem might use events labeled “T” for a positive test result — and “C” for the presence of Covid-19 antibodies:

Now suppose the prevalence of cases is 10 percent (that was so in New York City in the spring), and you have a positive result from a test with accuracy of 87.5 percent sensitivity and 97.5 percent specificity. Running numbers through the Bayesian gears, the probability that the result is correct, and that you do indeed have antibodies is 79.5%. Decent odds, all things considered. If you want more certainty, get a second opinion. And continue to be cautious.

An international collaboration of researchers, doctors and developers created another Bayesian strategy, pairing the test result with a questionnaire to produce a better estimate of whether the result might be a false negative or a false positive. The tool, which has won two hackathons, collects contextual information: Did you go to work during lockdown? What did you do to avoid catching Covid-19? Has anyone in your household had Covid-19?

“It’s a little akin to having two ‘medical experts,’” said Claire Donnat, who recently finished her Ph.D. in statistics at Stanford and was part of the team. One expert has access to the patient’s symptoms and background, the other to the test; the two diagnoses are combined to produce a more precise score, and more reliable immunity estimates. The priors are updated with an aggregation of information.

“As new information comes in, we update our priors all the time,” said Susan Holmes, a Stanford statistician, via unstable internet from rural Portugal, where she unexpectedly pandemicked for 105 days, while visiting her mother.The Coronavirus Outbreak ›

Updated August 4, 2020

That was the base from which Dr. Holmes refined a preprint paper, co-authored with Dr. Donnat, that provides another example of Bayesian analysis, broadly speaking. Observing early research in March about how the pandemic might evolve, they noticed that classic epidemiological models tend to use fixed parameters, or constants, for the reproduction number — for instance, with an R0 of 2.0.

But in reality, the reproduction number depends on random, uncertain factors: viral loads and susceptibility, behavior and social networks, culture and socioeconomic class, weather, air conditioning and unknowns.

With a Bayesian perspective, the uncertainty is encoded into randomness. The researchers began by supposing that the reproductive number had various distributions (the priors). Then they modeled the uncertainty using a random variable that fluctuates, taking on a range of values as small as 0.6 and as large as 2.2 or 3.5. In something of a nesting process, the random variable itself has parameters that fluctuate randomly; and those parameters, too, have random parameters (hyper-parameters), etcetera. The effects accumulate into a “Bayesian hierarchy” — “turtles all the way down,” Dr. Holmes said.

The effects of all these up-and-down random fluctuations multiply, like compound interest. As a result, the study found that using random variables for reproductive numbers more realistically predicts the risky tail events, the rarer but more significant superspreader events.

Humans on their own, however, without a Bayesian model for a compass, are notoriously bad at fathoming individual risk.

“People, including very young children, can and do use Bayesian inference unconsciously,” said Alison Gopnik, a psychologist at the University of California, Berkeley. “But they need direct evidence about the frequency of events to do so.”

Much of the information that guides our behavior in the context of Covid-19 is probabilistic. For example, by some estimates, if you get infected with the coronavirus, there is a 1 percent chance you will die; but in reality an individual’s odds can vary by a thousandfold or more, depending on age and other factors. “For something like an illness, most of the evidence is usually indirect, and people are very bad at dealing with explicit probabilistic information,” Dr. Gopnik said.

Even with evidence, revising beliefs isn’t easy. The scientific community struggled to update its priors about the asymptomatic transmission of Covid-19, even when evidence emerged that it is a factor and that masks are a helpful preventive measure. This arguably contributed to the world’s sluggish response to the virus.

“The problems come when we don’t update,” said David Spiegelhalter, a statistician and chair of the Winton Centre for Risk and Evidence Communication at the University of Cambridge. “You can interpret confirmation bias, and so many of the ways in which we react badly, by being too slow to revise our beliefs.”

There are techniques that compensate for Bayesian shortcomings. Dr. Spiegelhalter is fond of an approach called Cromwell’s law. “It’s heaven,” he said. In 1650, Oliver Cromwell, Lord Protector of the Commonwealth of England, wrote in a letter to the Church of Scotland: “I beseech you, in the bowels of Christ, think it possible you may be mistaken.”

In the Bayesian world, Cromwell’s law means you should always “keep a bit back — with a little bit of probability, a little tiny bit — for the fact that you may be wrong, ”Dr. Spiegelhalter said. “Then if new evidence comes along that totally contradicts your main prior belief, you can quickly ditch what you thought before and lurch over to that new way of thinking.”

“In other words, keep an open mind,” said Dr. Spiegelhalter. “That’s a very powerful idea. And it doesn’t necessarily have to be done technically or formally; it can just be in the back of your mind as an idea. Call it ‘modeling humility.’ You may be wrong.”

I’d Need Evidence Before I Got a Covid-19 Vaccine. It Doesn’t Exist Yet.

Coronavirus vaccines are rapidly advancing through the development pipeline. The University of Oxford’s vaccine is in large trials in BritainBrazil and South Africa. In the United States, researchers just began enrolling around 30,000 volunteers to test Moderna’s vaccine, and more trials are starting every day. Operation Warp Speed has set an ambitious goal of delivering 300 million doses of a safe, effective vaccine by January.

But the concept of developing a vaccine at “warp speed” makes many people uncomfortable. In a May survey, 49 percent of the Americans polled said they plan to get a coronavirus vaccine when one is available, 20 percent do not, and 31 percent indicated that they were not sure. The World Health Organization considers “vaccine hesitancy” a major threat to global health, and poor uptake would jeopardize the impact of a coronavirus vaccine.RelatedScientists Worry About Political Influence Over Coronavirus Vaccine ProjectAug. 2, 2020

This hesitancy isn’t surprising. Why should we expect Americans to agree to a vaccine before one is even available? “I think it’s reasonable to be skeptical about a vaccine that doesn’t exist yet,” Dr. Paul Offit, the director of the Vaccine Education Center at Children’s Hospital of Philadelphia, told Today.

I’m a vaccine researcher, and even I would place myself in the “not sure” bucket. What we have right now is a collection of animal data, immune response data and safety data based on early trials and from similar vaccines for other diseases. The evidence that would convince me to get a Covid-19 vaccine, or to recommend that my loved ones get vaccinated, does not yet exist.

That data can be generated by the large trials that are just beginning, known as Phase III or efficacy trials. Some have argued that we already have enough safety and immune response data to start vaccinating people now. But this would be a big mistake.

This is how Phase III trials work: Thousands of healthy adult volunteers are randomized to receive either a new Covid-19 vaccine or a control — a placebo or an already licensed vaccine for another disease. Then they go about their normal lives. They do not know what they have received (known as “blinding”) so the two groups behave similarly in terms of risk taking.

Participants are monitored for side effects and contacted regularly to ask about symptoms and to be tested for infection. The goal is to compare the rates of disease or infection across the two groups to measure how well the vaccine prevents Covid-19 “in the field.”

It is possible that some Covid-19 vaccines may not prevent infection entirely, but they could still prepare a person’s immune system so that, if infected, they would experience milder symptoms, or even none at all. That’s similar to the flu vaccine: It’s not perfect, but we advise people to get it because it reduces intensive care admissions and deaths.

How many people need to be protected by a vaccine before it’s recommended for widespread use? Ideally, rates of disease will be 70 percent lower in vaccinated people than in unvaccinated people. The World Health Organization says a vaccine should be at minimum 50 percent effective, averaged across age groups. (We know from influenza that vaccines don’t always work as well on older adults whose immune systems have declined.)

This benchmark is crucial because a weak vaccine might be worse than no vaccine at all. We do not want people who are only slightly protected to behave as if they are invulnerable, which could exacerbate transmission. It is also costly to roll out a vaccine, diverting attention away from other efforts that we know work, like mask-wearing, and from testing better vaccines.

The last thing Phase III trials do is examine safety. Earlier trials do this, too, but larger trials allow us to detect rarer side effects. One of those rare effects researchers are paying attention to is a paradoxical phenomenon known as immune enhancement, in which a vaccinated person’s immune system overreacts to infection. Researchers can test for this by comparing the rates of disease severe enough to require hospitalization across the two groups. A clear signal that hospitalization is higher among vaccinated participants would mark the end of a vaccine.

The speed of the trials depends on how quickly we can detect a difference between the two groups. If two vaccinated people became sick versus 10 who got a placebo, it could be because of chance. But if it were 20 compared to 100, we would feel much more confident that the vaccine was working.

Key to getting a quick result is placing the trial in outbreak hot spots where people are most likely to be infected. We can even target the highest-risk people within those areas, using mobile teams to travel to neighborhoods, bringing the trial directly to the people. Some trials explicitly prioritize essential workers like health care workers or grocery employees. Others are simply focused on enrolling large numbers of participants as fast as possible.

Combining those efforts, it could take as little as three to six months to generate enough convincing safety and efficacy data for companies to apply for expedited review by the Food and Drug Administration.

There are ways for vaccines to be approved without definitive efficacy data, based on animal or immune response data instead, but the bar is extremely high, and for good reason. A precondition is that efficacy trials are not possible, typically because the disease is so rare or sporadic that it would require hundreds of thousands of participants to be followed for many years to tell if the vaccine is effective (rabies, for example). That is not the situation here.

While there is promising data from smaller trials that measured the antibody response in people who got a vaccine, it’s not enough to approve a vaccine. We don’t know the level of antibodies needed to prevent infection from this virus. There is a history of vaccines with promising immune response data that did not pan out in the field.

With this in mind, the F.D.A. has committed to the need for traditional efficacy trial data to approve Covid-19 vaccines. And it follows the W.H.O.’s recommendation, stating that vaccines must be at least 50 percent effective to be approved.

I worry nonetheless that public pressure may mount to approve a product that doesn’t meet our standards. Other countries may decide to approve vaccines based on weaker evidence. Russia, for example, claims to be on track to approve a vaccine in just a few weeks.

We must resist the desire to rush out a product. Creating vaccines is hard, and we should be prepared for the reality that some promising ones will not meet the F.D.A.’s criteria. Researchers and the government should also commit to transparency so that people can see the results for themselves to understand the regulatory decisions.

Waiting for a better vaccine to come along may feel like torture, but it is the right move. With so many potential shots on goal, scientists are optimistic that a safe and effective vaccine is out there. We can’t afford to jeopardize the public’s health and hard-earned trust by approving anything short of that.