COVID-19: What to Expect

Updates are in reverse order, meaning the latest update is just below.

UPDATE: October 20, 2020

Herd Immunity: Doing the Math

In June, we observed that the US had embarked on a herd immunity approach to dealing with the COVID-19 pandemic.

With little public discussion or debate, the US has embarked on a herd immunity approach to the coronavirus. No one’s really describing it as such, but the label is accurate just the same. Over the past month, scores of politicians and business leaders have declared (from the safety of their home offices) that “getting back to work” is more important than saving the maximum number of lives from COVID-19…

Bottom line, the decision to reopen the country was made out of political and economic desperation with suspect motives and dubious scientific backing. While the strategy was purportedly for the benefit of the American worker, it is those very same workers who will end up paying the price.

It was a provocative thing to say at the time, and I received a fair amount of blowback, most of it along the lines of “you’re exaggerating” or “not even Trump would do something like that.” At this point, it is patently obvious that the Trump administration has embraced herd immunity despite the enormous loss of life it would entail. Let’s do the math.

Most advocates admit that herd immunity would require 60-70% of a country’s residents to become infected before protection would be afforded to the general population. In the United States, this would mean between 199-232MM people becoming infected.We’ll go with the average of 215MM.

Compare this to the total known positive cases of about 8.5MM and the likely 35MM who have already been infected but not diagnosed, and the implication is that another 180MM would have to be infected.

The death rate has obviously fallen greatly over the past few months as those infected have tended to be younger and healthier and more effective therapeutics have been developed. But, it still hovers around 2.7% of diagnosed cases and 0.6% of suspected cases.

Applied to 180MM infected, this would mean total deaths exceeding 1 million.




UPDATE:  September 2, 2020

This is an excellent summary of where we are and why. More importantly, it offers simple, clear solutions to America’s coronavirus disaster.  Reprinted from The New Yorker, September 2, 2020.

We Can Solve the Coronavirus-Test Mess Now – If We Want To
by: Atul Gawande

To get out of this pandemic, we need fast, easy coronavirus testing that’s accessible to everyone. From the way people often talk, you might think we need a technological breakthrough to achieve this. In fact, we don’t have a technological problem; we’ve got an implementation problem. We could have the testing capacity we need within weeks. The reason we don’t is not simply that our national leadership is unfit but also that our health-care system is dysfunctional.

Many developed countries have met their testing needs, and ready access to speedy tests has been key to containing outbreaks and resuming social and economic activity. Whether you live in England or South Korea, scheduling is straightforward. No doctor’s order is required. Tests, where indicated, are free. And you typically get results within forty-eight hours.

Appointments can take days, results days more. Most testing in the United States is done by four companies—Quest Diagnostics, LabCorp, BioReference Laboratories, and Sonic Healthcare. Through early August, results routinely took four days or more, making the tests essentially useless. Times improved only when testing volumes declined, because many people gave up on getting tested. The vast majority of infected Americans, including those with symptoms, never get tested. And we have not even reached the fall, when flu season will hit and coronavirus-testing needs and demand are expected to rise substantially. As the saying goes, it’s as messed up as a pile of coat hangers.

Further complicating matters, insurers don’t pay for testing that they don’t consider medically necessary. Yet testing people who don’t have symptoms will be important to getting COVID-19 under control. The Centers for Disease Control and Prevention estimates that around forty per cent of viral transmissions occur before an infected person has any symptoms. Yes, a combination of distancing, hygiene, and mask-wearing when people can’t stay six feet apart can markedly reduce such transmissions. But, in many situations, we can’t count on people being able to maintain those measures. Testing is the only way to know whether a person is potentially contagious and in need of isolation.

Such “assurance testing” has been required by countries such as Iceland, France, and Germany for travellers from abroad in order to avoid a mandatory two-week quarantine; by states from Maine and Massachusetts to Hawaii and Alaska for out-of-state travellers; by many U.S. hospitals for all patients admitted for non-emergent surgery; and by the film and television industry, which plans to do regular testing of cast and crew members in order to start production again. The federal government has recommended that nursing homes test all staff once a week. Numerous colleges and universities have included repeated testing in their protocols for bringing students back to campus. Individuals are seeking assurance tests before visiting older family members.

The economist Paul Romer has argued for going further, saying that regular testing for every person in the United States should be our lead strategy for getting society back to normal. Those with a recent negative test would have permission to go without a mask; those with a positive test would be required to isolate for two weeks. The small Italian town of Vò, outside Padua City, took a similar approach. It offered testing to all its thirty-three hundred inhabitants at two points over two weeks during lockdown. Eighty-six per cent of the population came forward in the first round. Almost three per cent were infected, half of whom had no symptoms. In the second round, seventy-two per cent of the population was tested, with a 0.3-per-cent new-infection rate. With these last few cases isolated, the town was able to reopen long before the rest of Italy could.

But is any of this remotely feasible when we can’t even make sure that sick people can get tested in a timely way? The lunacy of our testing system is the lunacy of our health system in microcosm. We are now paying the price of our long, uniquely American resistance to making sure that everyone has proper health-care coverage, and to building an adequate public-health infrastructure. We have not fully grappled with the difficulties we’re up against. But, if we do, we have a chance to fix the problem before the worst of flu season hits.

The need for speed

The course of infection with the SARS-CoV-2 virus is relatively brief. It starts with perhaps as little as a few hundred viral particles getting into your respiratory passage; then, if they take hold, they multiply to a viral count so high that you are exhaling millions of viral particles per hour. You are typically at your most contagious around day five of infection. Symptoms usually begin about the same time. In more than eighty per cent of cases, adequate oxygen levels can be maintained without hospitalization, and the body’s immune system can reduce viral counts rapidly. In these mild to moderate cases, researchers have found almost no transmission a week after the onset of symptoms, although recovery takes upward of two to three weeks for a third of such patients. (In cases of serious illness, the reduction in viral counts can take a week or so longer.) The goal of testing is to detect the virus as early as possible during infection.

The standard diagnostic test involves, as tens of millions of people now intimately know, a swab of the cavity deep behind the nasal passage—or, as more recently authorized by the Food and Drug Administration, of just inside the nose. A few labs have been approved to test saliva. The sample from the patient is analyzed for evidence of viral genetic material. The method of analysis, real-time polymerase chain reaction, PCR, is a small wonder of science, and, if it detects the SARS-CoV-2 virus, the diagnosis approaches a-hundred-per-cent accuracy: false positives are rare. But false negatives are another matter. The miss rate among people with a symptomatic infection is upward of twenty per cent, owing either to sampling problems (the swab didn’t pick up enough of a specimen) or a viral count too low to be detected. That miss rate is even higher during the four days or so before symptoms begin—going from a hundred per cent on the first day of an infection to sixty-seven per cent on the fourth day, according to a Johns Hopkins study. A negative test is not a guarantee that a person is infection-free.

How often the testing must be repeated varies from once a day to once a week, depending on the baseline prevalence of infections in the group, among other factors. It’s especially helpful for groups that can be isolated from the larger community. The National Women’s Soccer League did just that, confining participants in its month-long Challenge Cup to an athletic village it created outside Salt Lake City and testing participants on arrival and then before every game. Not a single athlete tested positive after initial screening and admission to the village.

A few critics have argued that PCR testing for people without symptoms is overly sensitive, catching many people when their viral counts are too low to be contagious. Some of them, it’s true, could be on the downswing of infection, but most will be on the upswing (when it’s important to catch them), particularly when they’ve had a previous negative test.

An oddity about SARS-CoV-2 transmission is that many infected people, whether or not they have symptoms, won’t pass it along. (Contact-tracing studies find that, among people who live with someone diagnosed with COVID-19, sixty to ninety per cent never become infected.) And yet a particular person can, in the right settings—like a crowded bar or workplace—infect scores of others. There is a measure in epidemiology called the dispersion factor, which indicates how much a disease clusters. SARS-CoV-2 has an unusually low dispersion factor—an estimated eighty per cent of transmissions are caused by just ten per cent of cases. By contrast, the 1918 influenza had a very high dispersion factor: there was almost no clustering at all.

Perhaps just a subset of people are prone to spread SARS-CoV-2. Or perhaps everyone with the infection sheds the virus, but only in occasional eruptions during a short period of infectivity. You stop the spread by avoiding large indoor crowds that could include a person in the midst of an eruption, by masking everyone to contain an eruption, and by testing people so you can identify and isolate positive cases during the period when they can erupt. But that is possible only if you get the results fast.

The Kansas State University football team is a cautionary example in this respect. It set out to create a bubble of its own when the N.C.A.A. allowed practices to begin on June 1st. All players were tested upon arrival and isolated as a group from the rest of the community—so far, so good. The first ninety-six players to arrive tested negative. A last group of twenty-four players, largely freshmen, arrived a few days later and were tested on a Friday. But they didn’t get the results until Monday—and that window was all the virus needed. As the Times reported, teammates spent the weekend hanging out. One player who would test positive played video games at an apartment with up to fifteen players; another infected player joined a group that drove out together to a lake. Although the protocol was to wear masks and maintain distancing off the field, the rules slipped. Within a week, two cases had become fourteen, and the team’s practices were shut down.

Unused capacity

To fix testing, we have to accelerate two lines of operation: test collection and test processing. The primary challenge of test collection—sticking a fifty-cent swab in someone’s nose—is last-mile logistics, which can be more intricate than most realize. You’ve got to manage people flow; it can be a challenge just to find locations where potentially infectious people can turn up without infecting others. Then there’s the flow of supplies, which involves having adequate quantities of the swabs and tubes that your particular lab requires, plus the personal protective equipment required for the staff. There’s the information flow—gathering and linking a patient’s information, the provider information, and the bar code on the specimen tube in a way that the laboratory can deal with. And there’s the financial flow—figuring out the billing system required to get reimbursed by the correct payer, which, for a particular person, could be Medicare or Medicaid, a private insurer, an employer, the state, the patient herself, or any number of other sources. Running a test-collection operation can be a nightmare for scores of reasons. You have to navigate supply shortages, neighborhoods that object to having lines of cars turn up at a pharmacy or clinic parking lot, business-insurance plans that may not cover you if you provide testing, and building owners with liability concerns. You may be unable to use your label printer and computer system if there’s no Wi-Fi in the parking-lot drive-up location you’ve picked.

Processing tests is a very different enterprise. It is exacting. Most diagnostic tests are performed by putting a sample into a device and getting a result. But large-scale molecular diagnostic testing—in which segments of genetic material are read—isn’t done with off-the-shelf kits; these are laboratory-developed tests, meaning that running them is less like operating an appliance than like doing a procedure, with multiple steps requiring precision and tight controls. The basic coronavirus PCR test starts with a reduction process—a series of steps to wash the specimen from the swab, inactivate any virus present, and separate out the genetic material (both human and viral). What arrives at the lab as a six-inch specimen tube of material is reduced, by means of chemical reagents and various machines, to a few extracted drops of nucleic acids. Then comes the viral-detection phase. Enzymes and probes—special stretches of DNA—are added that, under the right conditions, recognize and bind to SARS-CoV-2 RNA. (The genome of coronaviruses is made of RNA, not DNA.) The binding of a probe triggers a reaction that creates a DNA transcription of the viral RNA template. That enables the final steps: the signalling process.

The mixture is now put into a PCR machine—in essence, a tiny oven that sequentially heats and cools the mixture to precise temperatures, triggering further reactions that generate new copies of the viral genetic material. Each cycle of heating and cooling doubles the number of copies. After thirty cycles, up to a billion copies of viral DNA are produced from each strand. A separate marker in the mix recognizes the DNA and releases a fluorescent dye that can be measured by an optical device in the machine; there’s a threshold above which the test is considered positive. The result is checked by a technician and released electronically back to the test site. Although labs are constantly making improvements to this chain of steps in order to increase efficiency and accuracy, the whole process, from receiving a sample at a lab to sending a result, typically takes between six and twelve hours.

Yet we have other laboratories with large amounts of untapped processing capacity ready to provide next-day results. In July, for instance, I visited the Broad Institute, a large academic laboratory affiliated with M.I.T. and Harvard, in Cambridge, Massachusetts, which has provided coronavirus testing for area hospitals, clinics, and others. As Sheila Dodge, the senior director of Broad Genomics, explained as she walked me through its molecular-testing operation, they had the capacity to process up to thirty-five thousand tests per day. With a few weeks’ notice, they could expand that to a hundred thousand a day—more than fifteen per cent of the nation’s current capacity. But, when I visited, they were receiving just a few thousand test specimens per day. I saw an entire room of machines standing by, mostly idle.

It’s the same story elsewhere. I spoke to leaders at the University of Minnesota’s Genomics Center, who reported that they have unused capacity to deliver up to twenty-five thousand tests a day. And there are numerous other university-based and independent molecular-diagnostics laboratories with the ability to expand the country’s testing capacity. Several companies—including Guardant Health and Helix, in California; Kailos Genetics, in Alabama; and Ginkgo Bioworks, in Boston—are rolling out advanced molecular-testing techniques that could enable them to collectively process hundreds of thousands of tests a day. Whereas the usual charge for test processing is a hundred dollars or more, most of these labs are willing to charge much less—from fifty dollars to as little as twenty dollars. What’s missing is the logistics operation to connect their supply of tests to the people who need them and to the entities that pay for them.

These remarkable scientific enterprises aren’t sitting on their hands. Each is making a herculean effort to connect to places seeking tests, one by one. The Broad Institute has built support for more than a hundred area colleges. Nonetheless, the labs are connected to only a fraction of those they could serve.

Decades ago, electric companies were organized in the same way that laboratory testing is organized today. They were vertical monopolies that ran their own power plants, transmission lines, and customer operations. That arrangement got the job done, but it meant that many communities endured brownouts and blackouts from a shortage of capacity, while others had an oversupply. And the companies impeded innovation such as cleaner and cheaper energy. The creation of a national electric grid that physically connected the electricity supply, plus the Energy Policy Act of 1992, which required transmission-line owners to allow electric-generation companies access to their power lines, opened the door for load balancing, increased supply, lower costs, and alternative energy production.

We have no national grid for the generation, transmission, or distribution of our testing supply—or, for that matter, the supply of ventilators, masks, intensive-care beds, or almost any other health-care resources. Now we’re paying the price. In power generation, the worry is that our national grid is aging; in health care, the worry is that we have no grid at all.

A grid is the goal

To see what a functional national grid can deliver when it comes to public health, consider South Korea. It supported test collection, test processing, and the connections between them. On January 27th, when the country had seen just three confirmed COVID-19 cases, officials from Korea’s equivalents to the C.D.C. and F.D.A. held an urgent meeting with clinical laboratories and medical manufacturers, telling them to develop PCR tests for the new coronavirus spreading out of control in China, and to coördinate manufacturing swabs, reagents, and other necessary supplies. The government created an accelerated authorization process. By February 4th, one lab had isolated viral specimens from the three cases and reproduced the genetic material from them. Government scientists had varying dilutions put into test plates and delivered to participating laboratories as a kind of final exam for their molecular tests. If a laboratory correctly identified all the positives and negatives, its test was approved. By February 6th, the South Korean government had validated and approved tests from forty-six laboratories. Afterward, it carried out periodic cross-checks to make sure that testing remained accurate.

At the same time, the government coördinated with hospitals, clinics, and public-health agencies to set up test sites, agreeing to cover the testing costs for patients (at a flat rate of sixty-seven dollars a test), and creating a common scheduling system. (Dial 1339 from anywhere in the country and you can schedule a test nearby that day.) That scheduling system also allowed them to balance the testing loads, connecting laboratories that had extra capacity to test sites where that capacity was needed. Today, South Korea, with as many people as Florida and Texas combined, has some six hundred test sites, served by a hundred and eighteen authorized laboratories, with, typically, a one-day turnaround time. That’s what a grid can get you. Other countries followed South Korea’s example.

We never did. In fact, the United States has stymied rather than accelerated the ability of laboratories to develop testing capacity. In the labs of my hospital system, the Mass General Brigham—as in other academic and commercial labs—scientists began developing a coronavirus test in January, concerned that the outbreak in Asia could become a danger here. But, through February, the F.D.A. authorized only the C.D.C.’s coronavirus test. On February 20th, the Department of Health and Human Services sent out an e-mail blast to labs nationwide summarizing the Administration’s policy.

It is a stunning document. It stated that the C.D.C. had the only authorized test protocol in the country, and that only it and select state and local public-health laboratories could use it. Other clinical laboratories were supposed to send their patients’ swabs to an already backlogged C.D.C. Not only did the notice provide no encouragement for laboratories to develop tests; it actually blocked one of the principal means for labs to do so, declaring that “clinical laboratories should NOT attempt viral isolation from specimens” collected from patients thought to be infected.

Neal Lindeman, the head of my hospital’s molecular-testing laboratory, could only shake his head as he read the notice. “I thought it was foolish and dangerous,” he said. “I decided to ignore the e-mail. I didn’t think this plan was going to work for the country. So we just continued working on our test in the background.”

And, as became widely known, the C.D.C.’s plan did not work; its test kit had contaminated probes, and its protocol was flawed. The virus spread undetected nationwide for a month while we delivered hardly any tests at all. On February 29th, the F.D.A. finally reversed course, announcing that laboratories could develop and deploy their own coronavirus testing programs, so long as they also applied for emergency authorization from the agency. But, unlike its Korean counterpart, it provided no plan to streamline the test-authorization process. On the contrary, the agency made clear that it was not enough for a laboratory to show that its test yielded accurate results on blinded specimens. A detailed review of protocols would be required as well. Even slight deviations from the C.D.C.’s protocol required the laboratory to submit a separate “bridging study”—a term that had not been defined. Nor did the F.D.A. have enough experts to do the detailed review the agency required.

Even with the substantial capacity of F.D.A.-authorized coronavirus tests that we’ve already accumulated, the government launched no major effort to balance testing loads among them. There is still unmet demand and untapped capacity. At the same time, we are undersupplied with test centers. And we have no defined approach at all for some hundred million essential workers, students, and others who need access to assurance testing that health-care plans won’t pay for.

Conservatives have long opposed the government’s playing a direct role in addressing major gaps in the supply of health-care services. A third of U.S. counties have no childbirth facilities, and shortages of psychiatric beds are even worse. So maybe we should not be surprised that we have no such system for testing. It’s not just that we lack a national grid; we haven’t even agreed that we need one.

Innovation’s impasse

Will improvements in techniques and technologies get us out of our testing debacle? They can help a lot—but only if we fix the underlying problem we have with delivery.

Consider, for instance, pooled testing, in which you combine samples from a group of people in a single tube for a test run. A negative result will clear the entire group. If a tube comes up positive, you go back to the original samples and run them individually. During the Second World War, draftees were screened for syphilis through pooled testing. The Red Cross has used the technique to screen donated blood since 1999. In May, the Chinese government used the technique to screen the entire city of Wuhan, testing more than nine million people in ten days. (Just three hundred tests came back positive.)

Pooled testing is useful when you’re screening lots of healthy people for an uncommon disease and you don’t have enough testing machines or reagents. But we’ve made no commitment to setting up the systems required to physically screen large numbers of healthy people at scale—and pooling doesn’t work if you can’t collect the specimens to pool. In communities where disease rates are high, meanwhile, pooling isn’t so valuable: too many tests will come back positive, which slows the whole production line down to a crawl. Plus, the approach doesn’t do anything about bringing online unused lab capacity, which would be needed to deploy any testing strategy efficiently.

Sewage testing, one type of pooled testing, has real potential as a means of cost-efficient, large-scale screening. People infected with SARS-CoV-2 shed the virus in their stool, and PCR testing of wastewater can detect the virus upward of four days before people typically obtain clinical testing. Several hundred communities have partnered with companies such as Biobot, a Boston-area startup, to monitor viral prevalence in their populations, and the C.D.C. has announced a National Wastewater Surveillance System to track the information. The technique could be especially useful when deployed in an institutional setting: for instance, to clear a nursing home or dormitory with a single daily PCR test. The University of Arizona prevented an outbreak when regular testing of sewage samples from twenty campus buildings turned up traces of the virus from a dormitory; swab-testing the three hundred and eleven residents the next day identified two with infections. The lower limit of detection still needs to be confirmed, but Biobot claims its system can pick up a single case in up to sixty-five-hundred people. And it’s comparatively cheap: the primary cost is installing a twenty-four-hour sampling system into a building’s wastewater outflow. But, again, the mere existence of the capability is not enough. Coördinated efforts and investment are required to deploy it. The supply of compositors, the sampling devices required, must be secured. And then you need a system to provide swift follow-up testing of individuals when a sewage test comes back positive.

What about at-home test collection? The F.D.A. has authorized a small number of companies—for example, LabCorp and the startup Everlywell—to give people home collection kits with specific, preapproved instructions for swabbing themselves or providing a saliva sample. Unsupervised self-collection can result in higher rates of missed infections, to be sure. And delivering and returning the tests by mail can add a couple of days to the testing process—although, as more labs gain approval for unsupervised collection kits, one can imagine pharmacies, workplaces, and schools facilitating pickup and drop-off sites that could speed it up. If we haven’t addressed the demand-capacity mismatch, however, the approach could still be slowed by lab-processing backups.

You could circumvent the lab bottleneck by means of tabletop “point of care” machines that can run SARS-CoV-2 tests in clinics and other authorized settings without specialized personnel. They’re a marvel of miniaturization—akin to having your own electric generator. The White House uses these devices to test staff and visitors. There are versions that can run a PCR test in less than forty-five minutes. For those who have the money, such machines provide a means to buy your way around our broken system. But demand for the devices and the cartridges they require has far outpaced manufacturing capacity. Because each tabletop device can typically run only one sample at a time, the technology is at best a partial solution for getting to the millions of low-cost tests we need per day—there’s a difference between a Xerox machine and a printing plant. To deploy these devices, the country would still need a focussed national effort to boost the supply and get it to the populations that most need them.

A cheaper type of point-of-care test has emerged that takes just fifteen minutes. Unlike PCR tests, which detect the virus’s nucleic acids, this is an “antigen” test, which detects the virus’s proteins. Most involve machines. But the Abbott antigen test card, which has just come on the market, uses the same technology as home pregnancy tests and costs labs just five dollars. It embeds a filter-paper-like strip with antibodies that trigger the release of a dye and make a line appear when they recognize and bind to SARS-CoV-2 proteins. The test is said to be almost as sensitive as PCR tests, but it has only been formally tested with (and authorized for) symptomatic patients. For people without symptoms, antigen tests are significantly less sensitive. The Abbott test card is still complex enough that it requires laboratory certification and a trained health-care worker to use it. But, with no machine needed, it’s easier to manufacture these tests at scale. The federal government has committed to buying a hundred and fifty million of these tests. How they’ll be distributed is unknown, however. With no grid to monitor need and link it with supply, distribution in this Administration has been left to the vagaries of political favor.

The weakness of antigen testing is not just its miss rate but also its rate of false positives. Two to three per cent of people without COVID-19 nonetheless turn up positive with these tests, unlike with PCR testing. The antibodies that these tests use to recognize and attach to SARS-CoV-2 proteins can also bind to unrelated proteins and cellular material. A quick, cheap test at the airport before you board a flight with a hundred other people sounds great. At present rates, one passenger might actually be unknowingly infected with the virus—current assurance-testing programs commonly find infection rates at or below one per cent. But two to three other uninfected passengers are likely to have a false positive result, be prohibited from boarding, and have to quarantine in whatever city they are in until they’re cleared by PCR testing. Today, most positive antigen tests will be false—an outcome that could prove untenable in many situations and inspire mistrust in the tests.

There are paper-strip molecular assays in development that, promising both precision and speed, could have the greatest transformative potential of any test. They recognize viral genetic material, typically using CRISPR-based technology, and are expected to avoid the false-positive problem of antigen tests. James Collins, an M.I.T. professor of biomedical engineering, is leading a team that, going a step yet further, is developing a face mask with a CRISPR-based sensor that signals if you develop a SARS-CoV-2 infection.

We are on our way to an emporium of testing technologies. The trouble is that none of them will allow us to get around the necessity to upgrade and simplify our system for securing public health. Home-use paper-strip technologies have not yet received F.D.A. authorization. It will be months more before they can be produced in mass quantities. And, again, technologies do not implement themselves: we have to make sure that people actually have the tests in hand, that they won’t run out of them in hot-spot communities, that public-health authorities are notified when tests are positive, that positive cases get confirmatory tests and self-isolate, and that their contacts are traced and tested, too. If the development of essential technologies requires effort, skill, and investment, so does the delivery of them.

The San Francisco model

There are places in the country that have recognized that reality. San Francisco, one of the first American cities to be hit by the pandemic, has also been one of the best at curbing it. A big reason was Mayor London Breed’s decision, starting on March 6th, less than twenty-four hours after seeing the city’s first two confirmed cases, to back the aggressive recommendations of her city’s public-health director, Grant Colfax, to halt large gatherings, suspend nonessential travel, explore work-from-home options, and rapidly deploy a citywide testing program that remains one of the country’s most effective.

San Francisco’s leaders recognized the importance of universally available testing; residents needed to have an easy way to get tested regardless of what insurance they had or didn’t have, regardless of what relationship they had or didn’t have with a clinic, and regardless of whether or not they had a doctor’s order. The city’s hospitals, labs, and clinics were not going to meet that need, because that isn’t their job. In America, health-care systems are responsible for the care of individuals, not communities. So the city’s public-health department decided to step in.

It took an approach that resembles South Korea’s. Officials worked with local providers and laboratories to establish a network of free drive-through and walk-in sites, open to anyone who lives or works in the city. The city established a scheduling system, which people could access online or by phone, and outreach testing for nursing homes and other vulnerable communities. Public-health officers signed standing orders for each site, eliminating the need for a doctor’s referral. The city negotiated and paid most test costs. And officials made sure that they could enlist a network of labs to provide adequate capacity and keep turnaround time low. They established, in effect, a public option for testing. Although there are scores of other test sites provided by hospitals, clinics, and pharmacies across the city, the CityTest SF network has accounted for two-thirds of the city’s daily volume of tests. The program has literally been a lifesaver. Elsewhere in California, places like Los Angeles have exploded with COVID-19 cases. San Francisco has kept the coronavirus contained.

All this, alas, makes San Francisco an anomaly. Historically, the work of public health has been separate from the work of health-care delivery, using measures outside the medical system to address controllable diseases like cholera, tuberculosis, and food poisoning. But even before the coronavirus crisis hit, the separation was making less and less sense. A wide range of diseases are now controllable with the modern tools of diagnostic testing, treatment, and prevention. As the coronavirus pandemic has demonstrated, the population-level distribution of medical tests, ventilators, and drugs matters as much to disease control as nonmedical measures such as masks and social distancing. Success requires integrating the systems focussed on individual care and those focussed on community needs.

We’ve long resisted that integration, though. The clearest indication is how we’ve spent our money. Spending per person on medical care was seventy-nine hundred dollars per American in 2008; it climbed to an estimated eleven thousand six hundred dollars in 2019. By contrast, the average spending per person allocated to state and local health departments was a measly eighty dollars per person in 2008, and fell to fifty-six dollars in 2019. Public-health agencies that are supposed to look after communities’ health have been forced to expand tattoo-parlor inspections while shrinking their programs to assure adequate maternal- and child-health services or screen for chronic illnesses like high blood pressure and diabetes.

Talk to city and state officials about what San Francisco accomplished, and you typically hear: “Just can’t do it.” So how has San Francisco been able to pull it off? In part, it’s because the last deadly pandemic it experienced—the H.I.V./AIDS crisis—taught the city that public health and medical care must work together. San Francisco has been a pioneer in guaranteeing that H.I.V. testing and treatment is readily available and affordable to all residents in need. Those reflexes, and the associated budgeting commitment, served the city well when the coronavirus arrived.

Some cities and even states—including Los Angeles, Illinois, Maine, and Tennessee—have created versions of San Francisco’s free coronavirus-testing model; the federal government set up test sites, too. Those sites received large numbers of people. But most programs don’t have sufficient funding to open enough sites—Illinois has only eleven, for instance—and the federal government opened just forty-one locations nationwide, most of which it shut down by the end of June, despite the accelerating spread of the disease. Some cities and states have run testing “blitzes,” providing public-sponsored, pop-up testing sites for a few days, with no scheduling required. But, again, because there isn’t the funding to open enough sites, lines are typically long. As cases spiked in July, almost a thousand people waited up to thirteen hours at such a site in Phoenix. Nor have these states created a system to shift testing loads to labs with capacity the way San Francisco did.



UPDATE:  June 24, 2020

Presented without comment…charts highlighting when each state below “re-opened.”


UPDATE:  June 17, 2020

On March 12, our local school district closed its doors after a fifth presumed case of coronavirus was detected in our Boston suburb of 28,000. At the time, 108 total cases had been identified in Massachusetts. No deaths had yet occurred. The US totals were 1,630 cases and 41 deaths (31 of which had occurred in Washington.)  Within the next week, most Massachusetts businesses and schools had been shut down.

Three months later, Massachusetts has reached 105,885 cases and 7,665 deaths. We see another 300-500 new cases and 40-50 new deaths every day. US cases are increasing at a rate of 20-25,000 per day and total over 1,225,000 – more than the next 4 countries on the list combined. About 1,000 lives are lost every day for a total of 120,000.

Three months later, most businesses have reopened in Massachusetts and around the country. Wearing masks is passé, especially if you’re under 50 or a Republican. The administration insists there won’t be a second wave – that it’s just the media trying to scare people (but, sign a waiver if you’re attending one of their rallies, just in case.)

If 1,630 cases and 41 deaths were enough to essentially close down the entire country, why are the politicians and titans of finance so excited about opening it up with 1,225,000 cases and 120,000 deaths?

UPDATE:  June 12, 2020

With little public discussion or debate, the US has embarked on a herd immunity approach to the coronavirus. No one’s really describing it as such, but the label is accurate just the same. Over the past month, scores of politicians and business leaders have declared (from the safety of their home offices) that “getting back to work” is more important than saving the maximum number of lives from COVID-19.

“We cannot let the cure be worse than the problem itself” tweeted President Trump. Before long, supporters appeared in statehouses across the country, sometimes toting weapons and often without masks or social distancing, protesting states’ shutdown orders and demanding their right to return to work, frequent their favorite restaurant, or host a pool party.

How most of us feel about this probably depends on their health, wealth or politics. Some consider reopening the country an economic and civil rights imperative that justifies the risk. While others, particularly those in front line jobs who can’t obtain or live on unemployment insurance, have little choice but to return to work, regardless of the risks involved. Many health care workers who have lived through the hell of a local outbreak are alarmed.

As one governor after another caved to the protestors and economic pressures, the nation held its breath. Would there be a resurgence in cases?  Would areas which hadn’t previously experienced serious outbreaks be affected?

Now that all 50 states have reopened to at least some extent, the answer is clear: there has been a resurgence of cases, particularly in those states that took fewer precautions in the early days of the pandemic, closed down late, or opened up early.Some states such as Washington, Vermont and Alaska, which had early success in limiting spread, are now seeing a resurgence. Others such as North Carolina, California, Texas and Arizona, which had been relatively unaffected a month or two ago, are now seeing rapid growth and new highs in cases.

Though the White House established certain gating criteria such as a declining rate of infections before opening up again, almost none have.  Consequently, about 20-25,000 Americans are being diagnosed every day, while about 1,000 die every day.  Other countries are beginning to catch up to the US in both cases and deaths.

Brazil, with about two-thirds the US population, now experiences more diagnosed cases and deaths each day than does the US.  Russia and India aren’t far behind.

It would be nice to be able to say that Americans are receiving mixed messages from the administration regarding the threat still posed by the coronavirus. Unfortunately, the message has been crystal clear: masks and social distancing aren’t that important.

The implicit bet being placed is that Americans, with more money in their pockets, will ignore the rising infections and deaths in their midst.

If the numbers become too alarming, the blame can be placed on Black Lives Matter protesters instead of those protesting to reopen the country or attending political rallies.

Sweden’s chief epidemiologist Anders Tegnell recently admitted that “there [was] potential for improvement in [Sweden’s herd immunity approach], quite clearly.” Asked if the country’s high death toll has made him reconsider having attempted herd immunity, Tegnell replied, “Yes, absolutely.”  Looking at the continuing climb in Sweden’s new cases, it’s not hard to see why…

…particularly when compared to its Scandinavian neighbors.

Time will tell what the damage is from America’s return to work.  There will obviously be winners and losers. A month ago, the losers would have been concentrated among those who are elderly, have underlying health conditions or work in crowded, unsafe conditions. With the country opening up COVID-19 will obviously infect and kill plenty of others, including those who chose to go back to work and those who had no choice.

The public debate over the wisest course of action could have included clergy, scientists and civic leaders. It never happened. Instead, the decision was made for us. Our politicians have decided we have nothing to worry about and the economy cannot be shut down again.  It was too expensive – though there was plenty enough money to offer billions to unidentified corporations.

Bottom line, the decision to reopen the country was made out of political and economic desperation with suspect motives and dubious scientific backing. While the strategy was purportedly for the benefit of the American worker, it is those very same workers who will end up paying the price.

Stay safe.


UPDATE:  May 10, 2020

As some of you know, my daughter was in Rome for a semester abroad when the coronavirus breakout occurred in northern Italy. I was able to convince her to return to the US on March 8 – the same day that 16 million in northern Italy were quarantined. Both her US college and the Italian institution waited until March 15 to recommend (not require) that students leave the country.

She wasn’t feeling well and had an obvious, nasty cough, yet passed through customs at both Heathrow and Boston Logan Airport without a single person asking her about her health, taking her temperature, or recording her name, contact information or destination.

She didn’t have a fever, but we decided to quarantine her at home anyway. For her, this meant staying in her room most of the time, washing her hands any time she entered a common area, using a separate bathroom, etc.

For me, it meant getting up early every morning to disinfect everything she had likely touched: every knob on every kitchen cabinet, the refrigerator door handles, bathroom faucets, door handles, stair banisters, etc. We also cooked almost all of her food to minimize her contact with kitchen items/surfaces and administered temp checks daily.

I put an hour or two per day into maintaining the house in this manner, wondering all the while whether it was even necessary.  I also spent another hour or so every day contacting various health care professionals to try and get her tested.  Without a high fever, it wasn’t allowed per the CDC guidelines – even though President Trump had declared two days prior that “anybody who wants a test can get a test.”

Thirty-two days after her return, during one of countless sleepless nights, I happened to notice that CVS had announced drive-through testing at a site only 45 minutes away. I snagged an appointment online moments before they filled up.

Two days later, I watched incredulously as she was handed a swab and told to collect her own specimen by sticking it an inch or so into her nose.

Based on what I had seen, a proper nasopharyngeal specimen requires inserting the swab several inches into someone’s nose, so far back that it reaches the back of their throat. If a person isn’t very uncomfortable, it hasn’t been done correctly.

Her results were negative but, given that the specimen wasn’t collected properly, I assumed that she could easily be one of the 30% of those who were given false negative results.  It’s now a month later, and for better or worse, I’m finally beginning to believe the test results.

But, there are two important takeaways from our experience. First, the US has utterly failed in its efforts thus far. By March 8, it had conducted a total of 1,707 tests. By contrast, Italy and South Korea had each tested hundreds of thousands.

Given that up to 50-60% of all infections are asymptomatic, how many of the current 1,350,000 known cases (1,035,000 of which are still active), 166,000 of whom were hospitalized and 80,000 of whom died could have been spared had widespread testing and contract tracing been more widely available?

My daughter had a persistent cough and arrived from a country with the second highest number of infections in the world, yet no one even jotted her name down for follow up. Nor was she able to get tested until 34 days later – 36 days after Trump insisted tests were available to anybody who wanted one.

The day she returned home, China led the world  with 80,735 reported cases. Italy and South Korea were in a close race for second and the US was far behind. Two months later, the US is very far ahead of the others on both an absolute and relative basis. According to experts, US cases could be 10-20 times greater as the shortage of available testing has contributed to a significant undercount.

Likewise, deaths in the US have multiplied much more rapidly than in Italy or South Korea.

On both an absolute and population-weighted basis, South Korea’s aggressive testing, tracking, and quarantining measures have made a huge difference with 212 cases and 5 deaths respectively per million citizens. There have been many days over the past two weeks when South Korea reported no new cases and no new deaths.

By contrast, the US has incurred 4,079 cases and 36 deaths per million citizens. It has averaged about 29,000 new cases and 2,000 new deaths per day over the last three weeks. The periodic lulls have been just that – lulls, not persistent declines.

Partisan rhetoric and bickering aside, the US has failed its most important public health test of the past 100 years – one measured in millions of illnesses and tens of thousands of deaths.

The second big takeaway is that things will likely get much worse in the coming months. With 1,350,000 known cases, about one of every 245 Americans has been diagnosed as infected. If undercounted by 10-20X as former FDA Commissioner Dr. Scott Gottlieb believes, it could be as many as one in every 12-24 Americans. About one of every 4,111 Americans has been killed by COVID-19 so far. These are numbers that should concern anyone considering whether to dine at a restaurant, attend a football game or send their kid back to school.

My youngest daughter’s high school has 1,500 students, so it’s likely there are at least a few students who have been infected (maybe half of whom are asymptomatic) and will come into contact with at least 100 other students every day. At the end of each day, those students will head home and risk infecting their family members.

Most families can and will exercise caution. But, if my experience is at all typical, it won’t be enough.  Anyone can be hyper-vigilant for a day or two. But, it’s extremely difficult to maintain that same level of caution day after day for weeks at a time.

I’ve been accused of being over-the-top cautious (my daughters prefer “paranoid.”) And, I can’t count the number of times a mistake was made which might have landed one of my family members in the hospital. There were times when my daughter coughed in a common area of the house and we nervously ignored it. Other times, one of us used the same faucet, toilet, door knob, kitchen utensil or remote as she did without first disinfecting it. Several mornings, I forgot to wipe down the banister or the handle of the water pitcher.

Despite the best of intentions and very significant time and effort, each passing day was filled with countless little mistakes – any one of which could have been serious or even fatal. I even knew this to be the case at the time — having recognized very early on that health care professionals with access to superior knowledge and safety equipment were falling ill and dying from COVID-19.

In short, we were as fortunate as we were simply because my daughter was not infected. Had she been infected, our story would be quite different.

A recent study determined that two-thirds of patients hospitalized with COVID-19 in New York had been observing lockdown restrictions. They weren’t eating out, working in a crowded office, riding public transportation, attending political rallies or welcoming their kids home from school every day. They were being careful, yet still became sick.

As more and more states open up, I suspect we’ll see a surge in cases and, after a brief lag, a surge in deaths – particularly in those states which have done inadequate testing, have large numbers of nursing homes, prisons and meat-processing plants, and/or have leadership which has downplayed the coronavirus threat.

There is some good news related to flattening the curve. The daily rate of growth in US cases and deaths continues to slow. The 10-day moving average is about 2.5% for cases and 3% for deaths.  At this rate, cases will reach 1,500,000 around May 14 and deaths will reach 100,000 around May 18.

Global cases should reach the 5 million milestone by May 20 and deaths should reach 300,000 by May 14.  Of course, the progression of global cases will depend largely on the US’ success or lack thereof in reopening the country.

Much has been discussed about the tradeoffs between saving the economy and saving lives. I’m neither a philosopher nor a religious scholar, so I won’t wade into the moral argument. I am troubled, however, by much of what I hear on the economic front.

To a family struggling to buy groceries or pay the rent, reopening the country is a matter of necessity rather than choice. Through no fault of their own, they must bear the economic brunt of their government’s failure to act in a timely manner to forestall the epidemic.

For many, working from home is not an option and buying groceries means exposing oneself and one’s family to serious illness and death. One particular story caught my eye this past week.

Sixty-four year-old Rafael Benjamin worked at a Cargill meat-processing plant in Pennsylvania. Many of his coworkers were falling ill, so on March 25 a daughter gave him a face mask to wear at work. Two days later, a supervisor ordered him to remove it as it made other plant employees nervous. On April 4, Rafael called in sick. The following day, unable to move, he was taken by ambulance to the hospital. He died on April 19.

On April 24, President Trump classified meat plants as critical infrastructure, preventing local and state health authorities from closing them down regardless of health concerns. The business community is working with the White House and Congress to protect companies like Cargill from customer and employee lawsuits related to coronavirus.

If passed, such legislation would allow bad actors to pressure workers to report to work regardless of their health conditions and without providing common sense safeguards. Because it would ultimately result in more illness and death, it would ironically apply additional economic pressures on the very companies seeking liability protection.

Another industry singled out for special treatment is the oil industry. If the White House and Congress were primarily concerned with unemployment in the sector, they could more vigorously subsidize payrolls instead of artificially driving the price of oil and gas higher.  First, artificial price support will do nothing to correct the imbalance between supply and demand. Second, higher fuel prices are a regressive tax on those who can least afford them.

As a disciple of the Cynical School of Economic Thought, I find the entreaties by many major corporations and billionaires (most of which have cheap, ready access to capital markets) to reopen the country now to be self-serving — focused more on inflating excessive equity valuations than ensuring full employment and economic well-being for society’s most vulnerable.

Perhaps quantitative easing and financial engineering have become so effective that the economic well-being of ordinary citizens is no longer relevant.  Things certainly seem to be headed in that direction.


UPDATE:  Apr 17, 2020

Word on the street is that the remdesivir story was planted by a rumor-for-hire outfit and that the drug has failed in previous trials in China. Gilead has since walked back the hype.

The other COVID-19 story to hit the wires this morning is that Moderna received $483 million from the government to fast track its vaccine efforts. It’s great news, as is all the other vaccine research going on at an unprecedented pace. In listening to the CEO’s interview, however, it’s obvious that this, like every other vaccine being developed, won’t be widely available until 2021.

I’m as hopeful as anyone else. But, in the absence of a therapeutic or vaccine, I believe opening the country back up will absolutely result in a resurgence of community transmission and a sharp rise in the number of deaths – particularly in those parts of the country which have yet to be affected in great numbers.

The problem continues to be testing. If all of us pack in to a basketball game or concert and even one of us is infected, many of us will become infected. We’ll then head home and infect our loved ones and go in to the office and infect our coworkers. Since 25-50% of infected people have little or no symptoms, there is absolutely no way to prevent this from happening unless we get a vaccine.

In the US, only 13% of the 680,000 diagnosed cases have been closed – meaning either a person got better or died. Of those which were closed, nearly 40% resulted in someone dying – 34,700 of them.The chart below shows that the death rate (the orange line) has remained very steady since March 26, bouncing back and forth between 35% and 43%. This is not progress, and it hardly argues for easing up on efforts to prevent community transmission.

The US remains in 44th place in the world in testing per 1mm people. The chart below, sorted by tests per 1mm, shows some of the countries which have done much more. Not a great argument for opening up the country. The raw number of tests, BTW, is irrelevant. In total cases per 1mm population, the US is in 20th place…

…while in deaths per 1mm, the US is in 14th place. Again, not a great argument for opening up the country. We have had periodic declines in new deaths and cases and the rate of growth in each has definitely subsided. But, both are still occurring at an alarming rate.

If the proportions hold, 1,500 of the 30,000 people diagnosed today will die — and that’s without opening up the country.

Compare the data to a country such as Germany, which imposed a stronger quarantine much earlier than the US. Sixty-two percent of all diagnosed cases have been closed, and deaths account for a mere 4% of those. Remember, the US is running about 40%.

Even though active cases have clearly rolled over…

…Germany is still seeing 3,000 new cases per day… …and deaths are still ticking higher.By comparison, the number of active cases in the US is beginning to flatten, but is far from rolling over.

The most frightening report I’ve seen in the past week was this one from the UN which reveals that 74 million people in the Arab region lack access to a basic handwashing facility (i.e. clean running water.) It’s safe to say that most of those 74 million also don’t have a bottle of Purell handy.

Consider the 1.2 billion people in Africa who have been so susceptible to past epidemics which were much easier to detect. The 2013-2016 Ebola outbreak in West Africa infected 28,646 and killed 11,323. Almost 20,000 cases of COVID-19 have been detected so far, and it’s still early days.

I give President Trump’s political advisors credit for convincing him to give governors the responsibility for deciding when/how to reopen their states. Those who are aggressive about it will pay the price, while those who don’t reopen things quickly will continue to pay an enormous financial price. Without widespread testing and an effective therapeutic/vaccine very soon, there will be no happy ending for either.


UPDATE:  Apr 13, 2020

I’ll start with the bad news, just to change things up.  The US now has almost 600,000 cases, 1 out of every 587 people and almost as many as the next 4 countries combined.  Deaths, now 3.9% of all diagnosed cases, total over 23,000.  Sadly, 40% of all closed cases result in someone dying.

Fatal outcomes now account for 4% of all diagnosed cases. Another way of looking at it: for every 100 people who have been diagnosed, 90 are still sick. Of the remaining ten, 6 recovered and 4 died. Were these percentages to hold, we might expect the 26,641 new cases diagnosed just today to result in 1,065 deaths.

But, since deaths lag new diagnoses, it would be 4% of a larger number. And, since the mortality rate is climbing every day (it was 1.2% on March 22) it would probably be a larger percentage of a larger number of people. If one of those were a friend or relative, you’d might disagree with opening up the country just yet – regardless of the trends.

Many countries have had much better outcomes, such as Switzerland where deaths are only 7% of closed cases and testing has been conducted on 22,993 per million citizens versus 8,816 per million (43rd in the world) in the US. Even Italy, to which the US is often compared, has tested 17,315 per million.

Now, for the good news. The daily rate of growth (ROG) in cases has slowed significantly  – presumably as a result of quarantining (aka social distancing.) The 10-day moving average for case growth has slowed from 15% a week ago to about 10%. The 10-day moving average for the daily ROG in deaths has slowed from 22% to about 14%.

More testing and continued quarantining should continue to bring the growth rates down. Opening up the country without more testing, however, will only serve to increase them. At a ROG of 14%, deaths will reach 100,000 by April 24. Slowing the rate to 10% (i.e. flattening the curve) would delay that milestone to Apr 28 and to 5% would delay it to May 13.  This would buy scientists more time to come up with effective therapeutics.

Getting tested?  It’s easier now but, based on my experience, not very accurate. I took my daughter (who returned from Italy with a bad cough) to a drive-through testing site run by CVS. It was fast and free, but they handed my daughter the swab and asked her to collect her own specimen.

She inserted the swab about 1 – 1.5 inches into each nostril, at which point the tech said “good job” and put the sample into a bag for assessment. This was completely inadequate and is probably responsible for the 30%+ false negative results being obtained.

An accurate nasopharyngeal specimen collection means sticking the swab so far back that most people would gag and tear up in the process. Here’s a video of the correct process.  Bottom line, don’t assume that a negative test is correct. Continue to exercise caution – particularly if someone is exhibiting symptoms.

And, even if some politician gives you the all clear, remember that nearly 30,000 new cases will be diagnosed today. Don’t be one of them.

UPDATE:  April 3, 2020

Out of curiosity, I took a look at mortality rate trends in a few other countries. While some are clearly better than others, every single one is rising consistently over the past week.

And, deaths are consistently rising faster than cases. Cases have risen an average of 2x over the past week, while deaths have risen an average of 3.4x.  In the US, the numbers are 2.9x and 4.7x – a ratio of 1.64 deaths per every new case diagnosed. In Germany, deaths are rising over twice as fast as cases.

I have very little medical training, but I believe this is why we’re hearing so much about flattening the curve. The experts know that our hospitals will be overwhelmed if the ROG doesn’t slow dramatically.  Overwhelmed hospitals and shortages of ICU beds and ventilators will translate into higher mortality rates.

In some cases, however, the higher mortality rate can be attributed to a drop in the rate of growth in new cases.  Due to the lag between diagnosis and death, case growth can roll over as deaths are still rising. A good example is Austria.  New cases peaked on March 26 while deaths peaked on March 30.

This has allowed its mortality rate to almost flatten: from 1.3% on March 31 to only 1.5% on April 3. And, its curve is definitely flattening. Note that Austria enacted nationwide restrictions on travel and gatherings on March 16 when there were 1,018 total cases and 3 deaths. This was 18 days after the first community transmission case was detected and two days after the first death.The US, by contrast, has still not enacted a nationwide lockdown despite over 300,000 cases and 8,000 deaths since the virus arrived over two months ago on January 20.

The 10-day moving average of the daily rate of growth in deaths in the US dipped slightly to 27.3%.  But at this rate we would still reach 10,000 deaths this weekend, 100,000 by April 14, and 1 million by April 24.

Some might feel this rate is too pessimistic, that something will happen to change the trajectory. I hope so. But, even if the rate were to plunge to 20%, we would still have over 1 million deaths before the end of April – one in every 323 people in the US.  And, remember, this number is understated – though not as dramatically as the number of cases.

Don’t wait for the government to tell you what we already know: the only way to be completely safe is to stay at home, away from anyone who might be infected (including asymptomatic family members) and to wear a mask if you absolutely must go out – especially around other people. Assume that everything you touch might harbor the virus.

Yes, you might feel like you look really paranoid. But, your family will thank you.  And, God willing, you can all laugh about it some day.

For those staying in this weekend, I just read that HBO is releasing 500 hours of programming for free, including every including every episode of The Sopranos, Veep, The Wire, Barry and Ballers.  You can download the HBO NOW or HBO GO apps or visit or #StayHomeBoxOffice.

Stay safe everyone.


UPDATE: March 31, 2020

Italy’s mortality rate continues climbing steadily: from 2.0% on March 1 to 11.4% as of March 30. I never see this data reported anywhere, but believe it’s very important as it confirms what should be obvious: An overwhelmed health system is not as effective in saving the lives of sick COVID-19 patients.

Those in the US who take comfort in its current 2% mortality rate probably shouldn’t. Reported cases – still an understatement of actual cases due to the continuing shortage of test kits – continue to increase at an average (10-day SMA) of 24% per day. But, deaths are climbing at about 30% per day – much faster in some states. Inexplicably, many states have yet to issue stay-at-home orders.

The top 25 states in terms of cases are shown below. Note that Louisiana, a relative newcomer to the top 10, is already at a mortality rate of 4.6% – topping Washington’s 4%. Virginia’s governor finally issued a stay-at-home order yesterday, as that state’s mortality rate reached 2.5%.

Several cities and counties in Georgia, which faces a 3.4% mortality rate, have issued orders; but, the governor has yet to announce a statewide order. Arizona’s governor, who had issued an order restricting the ability of cities and counties to issue their own orders without his approval, finally caved yesterday when cases topped 1,000.

As more hospitals across the nation are overwhelmed by cases that outnumber the available ICU beds and ventilators, the mortality rate is bound to rise. Those that restrict transmissions before cases get out of hand stand the best chance of not being overwhelmed. The rest will suffer the consequences. As Dr. Georges Benjamin, executive director of the American Public Health Association, said: “Waiting until you get a lot of cases is the wrong strategy.”

Currently, about 1/3 of closed cases – someone being either discharged or dying – result in a death. But, only 5% of total reported cases have been closed – again, a factor of the rapid growth in cases.  In Italy, where 26% of all cases have been closed, deaths account for 44% of all closed cases and, as mentioned above, 11.4% of total cases.

Applying the same percentages to US cases would indicate that 2,313 of the 20,297 new cases reported just on March 30 will result in someone dying. At the current daily rate of increase, the US will reach 10,000 deaths by April 4 and 100,000 on or around Easter. My projections indicate total cases could reach 2.4 million by then, swamping the 924,000 beds in the 5,198 community hospitals across the nation.

If even 5% of those cases were serious/critical, the resulting 120,000 cases would far outnumber the 77,000 ICU beds in the US.

Bloomberg pubished an interesting article earlier today citing an Italian study that offered a theory as to why Italy’s mortality rate is so high.  The study suggested that because Italy had such a mild winter, many of the deaths that might ordinarily have occurred due to flu were brought on instead by the coronavirus. The inference is that the mortality rate is much higher than it would otherwise be had the winter been colder.

While I certainly find the premise plausible, it doesn’t do much to move the needle.  The study suggests that “the total number of older Italians spared death directly from the flu this season may have been in the hundreds, based on an annual average of 8,000 nationwide flu deaths cited in the paper.”

Italy has seen over 12,000 deaths. Subtracting 500, for instance, would lower its mortality rate from 11.4% a still very high 10.9%. Having had a daughter in Italy between late January and March 8, I followed the situation very closely. I think the high death rate is much more likely to have been the result of some bad luck (getting hit with the first burst of infections) and a failure to lock down the entire country.

Early on, several small towns in northern Italy were quarantined.  Many who were able to simply left the area, returning to their homes in southern Italy or elsewhere in Europe.  As cases spread, Milan was included in the quarantine. Again, this didn’t stop anyone from returning to their home. As I wrote at the time, 20,000 people were taking a train from Milan to Rome every day.

Eventually, the entire country was put on lockdown. But, by then, the spread had occurred and quickly overwhelmed the available hospital facilities. While cases increased 59X between Mar 1 and Mar 30, deaths increased 341X.

This is why I believe a nationwide lockdown is essential for the US. We all know someone who left a heavily infected area to go someplace “safe” such as a friend or relative’s house, a second home or just a hotel. Unfortunately, some of those people are infected and transmit the disease to people in the other location. It only takes a few to get the community spread ball rolling. This is precisely what happened in Europe.

True, there are some very efficient and tightly controlled countries in Europe with low mortality rates (so far) such as Switzerland (2.7%), Austria (1.4%) and Germany (1.1%).  But, each of these took early, aggressive measures to limit the spread. Many other countries are slightly behind but on the same track as Italy: France (6.8%), Spain (8.9%), UK (7.8%), and Netherlands (8.6%.)

A March 30 study from Imperial College of London researchers draws a direct correlation between the number of deaths and the speed with which various countries took steps to intervene in the virus’ spread.  All but one (Sweden) ultimately ordered a lockdown – typically as each country’s leaders realized they would not be spared.

While identifying and limiting the spread of US cases is incredibly important, note that I have focused mostly on the trend in deaths — which distills all the theories and practices into one defining, indisputable and depressing number every day.  It’s where the rubber meets the road.

I printed a list of the top 25 states in terms of cases as of March 30 above. Mortality rates ranged from 0.5% to 4.6% with a median of 1.7% and a mean of 1.9%.

At the end of today, March 31, the range was from 1.1% to 4.6% with a median of 2.1% and a mean of 2.2%. While total deaths grew by 31% in that one 24-hour period, the average daily rate of increase among the 25 states was a stunning 47%.

Note that the order of the top 25 has changed. Florida has moved into the #5 slot, Illinois has moved ahead of Washington to #7, Louisiana moved from #10 to #9, Georgia moved ahead of Texas to #11, etc. I’ll continue to track these same 25 so we can stay abreast of the trend.

The important thing to realize is that deaths continue to rise sharply. As in Europe, the growth has tended to accelerate faster in states which were late to start testing, close schools and businesses, and issue stay-at-home orders. Those that have not restricted movement should expect to see deaths continue to outpace those that have.

With a total of 4,053 deaths as we goes to press, rising at a 10-day moving average of 29.8% per day, total US deaths should reach 10,000 by Saturday April 4, 100,000 by April 13, 500,000 by April 19 and 1 million by April 22.

I believe that this fate can be avoided only if: (a) everyone who should be tested is tested, (b) the entire nation is locked down, and (c) scientists can develop effective treatments in the next week or so. I hope to God I’m wrong.

A few reminders that I haven’t seen circulated nearly enough: Studies indicate that up to 50% of transmissions are from people who have little or no symptoms. Your kid or your neighbor who appears perfectly healthy can give you the virus.

And, studies now indicate that the coronavirus can remain aerosolized for up to three hours. You could be walking down a deserted aisle at the grocery store and stroll right through the viral mist emitted by someone who sneezed there 20 minutes ago. Wear a mask when out in public places and assume that everything you touch is infected.

 *  *  *

UPDATE:  March 22, 2020

Italy’s mortality rate increased to 9.3% on March 22, a steady increase from 2.0% on Mar 1.  Though, on a positive note, the daily rate of growth in cases and deaths has ebbed somewhat: Neither has exceeded 20% since March 13.  On a not so positive note, both are still doubling every 6-7 days.The daily ROG in US deaths ticked lower on March 20-21, but pushed back up to 39% on March 22.   At this rate, the 419 US deaths as of March 22 would exceed 1,000 by March 25, 10,000 by April 1, and 100,000 by April 8. Take a moment and let those numbers sink in.

Italy still has many active cases, 80% of the total. Fortunately, most are resolving as “recovered/discharged.” But, the 9.3% mortality rate – up from only 2% on March 1 – means that nearly as many die as are discharged/recovered.

Only a very small fraction of US cases are closed. Unfortunately, about 73% of closed cases have resulted in death rather than recovery.

This is up from 60% of cases just 4 days ago. Serious/critical cases as a percentage of active cases have more than quadrupled in that time. In other words, things are getting worse at an accelerating rate.

I regret to report that our original models posted on March 13-14 have not kept up with the actual results. The model which used the then 10-DMAs of daily rates of growth for cases and deaths (34% and 18%) had forecast 23,410 cases and 247 deaths for March 22.

The model which borrowed the 10-DMA of Italy’s daily ROG with an 11-day lag had forecast 23,410 cases and 481 deaths.

As of last night, total acknowledged US cases totaled 33,546 (still undercounted) and deaths totaled 419.

  *  *  *

UPDATE: Mar 17, 2020

For those interested, here is an update on the current case and death data for Italy and the US as of Mar 16.

In Italy, the current mortality rate has ticked up even higher to 7.7%. The 10-day moving average for mortality rate has increased to 6.3%. The 10-DMA for the ROG in cases and deaths has eased slightly to 21% and 28%.

   date    cases    deaths
21-Feb                21                  1
22-Feb                79                  2
23-Feb              152                  2
24-Feb              229                  6
25-Feb              322                11
26-Feb              400                12
27-Feb              655                17
28-Feb              889                21
29-Feb           1,128                29
1-Mar           1,701                34
2-Mar           2,036                52
3-Mar           2,502                79
4-Mar           3,089              107
5-Mar           3,858              148
6-Mar           4,636              197
7-Mar           5,883              233
8-Mar           7,375              366
9-Mar           9,172              463
10-Mar         10,149              631
11-Mar         12,462              827
12-Mar         15,113           1,016
13-Mar         17,660           1,266
14-Mar         21,157           1,441
15-Mar         24,747           1,809
16-Mar         27,980           2,158


This is largely due to a “good” day on March 16 where cases increased by only 13% and deaths by 19%.  We’ll pray that the decline lasts more than a day.

The US, unfortunately, has managed to keep pace with our forecast from March 13 which was based on a moving average of growth rates. On March 13, our model predicted 5,444 cases and 80 deaths for March 16. Actual results were 4,663 cases and 86 deaths.

Again, the continuing lack of testing kits renders the number of cases essentially meaningless. Many experts feel it could be off by a factor of at least 10X, possibly much more.

The number of deaths, unfortunately, is probably accurate. The 10-DMA for daily ROG in deaths has risen from 18% on March 13 to 19.2%. The chart below reflects these data and uses the most recent 10-DMA in daily ROG for each category to forecast future results (in italics.)

US: 10-DMA
    date     cases    deaths
3-Mar              124                  9
4-Mar              158                11
5-Mar              221                12
6-Mar              319                15
7-Mar              435                19
8-Mar              541                22
9-Mar              704                26
10-Mar              994                30
11-Mar           1,301                38
12-Mar           1,697                41
13-Mar           2,269                48
14-Mar           2,943                57
15-Mar           3,680                68
16-Mar           4,663                86
17-Mar          6,102             103
18-Mar          7,984             122
19-Mar        10,448             146
20-Mar        13,672             174
21-Mar        17,890             207
22-Mar        23,410             247
23-Mar        30,632             294
24-Mar        40,084             351
25-Mar        52,451             418
26-Mar        68,635             499
27-Mar        89,812             595
28-Mar      117,523             709
29-Mar      153,784             845
30-Mar      201,232          1,008

If we were to use the daily ROG exhibited by Italy on an 11-day lag as originally theorized, we’d see the following results in deaths (cases are the same as above, as this is a poor estimate anyway):

     date      cases     deaths
3-Mar              124                  9
4-Mar              158                11
5-Mar              221                12
6-Mar              319                15
7-Mar              435                19
8-Mar              541                22
9-Mar              704                26
10-Mar              994                30
11-Mar           1,301                38
12-Mar           1,697                41
13-Mar           2,269                48
14-Mar           2,943                57
15-Mar           3,680                68
16-Mar           4,663                86
17-Mar          6,102             114
18-Mar          7,984             135
19-Mar        10,448             213
20-Mar        13,672             269
21-Mar        17,890             367
22-Mar        23,410             481
23-Mar        30,632             590
24-Mar        40,084             736
25-Mar        52,451             837
26-Mar        68,635          1,051
27-Mar        89,812          1,254

As mentioned above, the mortality rate in Italy has risen to 7.7% from only 2.0% on March 1. This was no doubt impacted by the shortage of hospital beds, ventilators and medical personnel.

The US’ relative readiness in this regard has been the subject of many studies and articles. But, I have yet to find a credible source suggesting that the US is better prepared. According to the OECD, the US has 2.8 hospital beds per 1,000 people. China has 4.3, South Korea has 12.3, and Italy has 3.2.

If the number of current US cases were accurate, the current mortality rate would be about 1.8%. We know the number of cases reported is too low.  So, hopefully, the mortality rate is too high. But, even if it were accurate, an escalation in the mortality rate to Italy’s current level would result in over 15,000 deaths by the end of March.

This is why flattening the curve is so vital and why the current growth in cases and deaths is so troubling.

As in Italy, once hospital beds and ventilators are impacted the mortality rate will go up. By slowing admissions, we can reduce overcrowded conditions and keep the mortality rate lower. At least, that’s the theory.

It’s important to note that even a slight increase has a huge impact on resulting cases and deaths.  The chart below shows deaths at Day 5, 10 and 30 for different daily rates of growth.

ROG           Day 1           Day 5          Day 10              Day 30
5%                86              110               140                    372
10%                86              139               223                 1,501
15%                86              173               348                 5,694
20%                86              214               532               20,414
25%                86              262               801               69,470
30%                86              319            1,186             225,320


According to the American Hospital Association, the United States has 98,000 total intensive care beds, but 51,000 of those are for specialized use such as cardiology, pediatrics, neonatal, etc.  The difference between a 20% daily growth rate and a 30% growth rate, then, would make all the difference in terms of being able to handle the number of cases. But, that’s looking only 30 days out.

According to a Johns Hopkins Center for Health Security report conducted last month, a moderate pandemic would mean 1 million people needing hospitalization and 200,000 needing intensive care. A severe pandemic would require 9.6 million hospitalizations and 2.9 million ICU beds.

The even more serious problem is ventilators.  According to Johns Hopkins, the US has 160,000 in total, 98,000 of which “are not full-featured but can still provide basic function in an emergency during crisis standards of care.” Another 8,900 mechanical ventilators of uncertain capability are in a national stockpile. Any way you slice it, it’s woefully short of what will likely be needed and will, like Italy, result in a much higher mortality rate.

One last comment on an article which hasn’t received much air time: Stay away from your kids. Seriously. In this article from CNN, several studies are cited which indicate that infected but asymptomatic people might be more contagious than those with obvious symptoms.

In other words, your 20 year-old who just got back from a semester abroad but only has a cough could very well give you the virus that kills you – something to think about as you’re waiting for tests we’ve been promised to finally become available.  Bottom line, keep your distance but follow them around with a spray bottle of bleach. God willing, you’ll both laugh about it some day.





 *  *  *

March 13, 2020

I saw something quite stunning this morning: Treasury Secretary Mnuchin describing the COVID-19 pandemic as “a great investment opportunity.” Tone deaf as ever, but no surprise there.  He then went on to draw comparisons to South Korea’s experience where 200,000 people have been tested – only 4% of which were found to be infected, as though this meant there was nothing to worry about. Let’s unpack this, shall we?

Actually, 222,000 people in South Korea have been tested — about .44%, or one in every 225 people.  In the US, the number of people who have been tested was recently removed from the CDC website. But, many academics put it at approximately 13,600, about .0042%, or one in every 24,200 Americans.

In South Korea, the government can access data – CCTV footage, GPS tracking data from phones and cars, credit cards transactions, immigration records and other data to deduce who might be at risk of contracting the disease or passing it to others. Anyone who wants testing can get it, with drive-through testing in place since March 2.

The names of those Koreans found to be infected are made public so friends and relatives who might have been exposed can get themselves tested. Anyone who tests positive is self-quarantined and monitored through an app or by phone until a bed is available, at which point an ambulance delivers them to a hospital with air-sealed isolation rooms.

In the US, an acquaintance recently told me of arriving from an extended trip to Italy and breezing through customs with a noticeable, persistent cough. No one took their temperature, asked their health status or jotted their name down for follow-up. When they called their US doctor, they were told they couldn’t be tested unless their fever topped 100.4 degrees due to a shortage of test kits. The drive-through idea was just announced today.

All things considered, I consider South Korea a very poor comparison. For the past several weeks, I have focused instead on Italy. Like the US, they have a modern health care system, but were caught completely off guard when COVID-19 arrived from China. Italians are also used to a certain degree of autonomy and would presumably bristle at the intrusiveness of South Korea’s response.

About 11,500 Chinese travel to Italy each day compared to 14,000 who travel to the US. The US instituted a travel ban (with some exceptions) from China on February 4.  Foreigners were mostly turned away, but others had their temperature taken and were asked to self-quarantine. Italy instituted a travel ban with no exceptions, and no flights whatsoever, four days earlier.

Italy locked down the northern part of the country, putting 16 million people on lockdown on March 8. The lockdown was extended to the entire country – 60 million people – on Tuesday, with essentially every establishment except grocery stores and pharmacies closed.

In the US, not a single city or state has been quarantined.  Over the past few days, many schools, sporting events, concerts, conventions and the like have made the decision to close or operate online. I am not aware of a single enforced quarantine other than those who returned from a cruise or trip overseas and were clearly infected. Americans are asked to self-quarantine, which is fine in concept but doesn’t work very well given that the incubation period can be two weeks or longer, and few can afford to ditch their jobs for two weeks.

Italy has essentially been shut down, even though its response has arguably been much more robust than America’s.  In my opinion, this doesn’t bode well for the US. I have been making predictions for the past several weeks based upon comparisons between the two: schools closing, stores closing, travel restrictions, runs on cleaning supplies and groceries. Many readers, not to mention friends and family, who initially scoffed are no longer scoffing. I get many texts and emails every day saying something along the lines of “can you believe they just cancelled…?”

I’m a numbers guy, so I’ve been working diligently to come up with a formula that expressed the path the number of cases and deaths was taking. The problem is that only a small number of cases have resolved in both Italy and the US.

Total cases in the US are (supposedly) 2,269. This is no doubt very far short of the actual number due to the lack of available testing. But, the outcome data are telling. Only 79 of these 2,269 are closed: 39% due to the patients’ recovery and 61% due to their death. This leaves 2,190 in limbo, whose fate is as yet undetermined. But, we can make some educated guesses.

While the cases are almost certainly undercounted, the number of deaths is probably fairly accurate.  At 48 total deaths, the current mortality rate is 2.1%. Some might take solace in such a low number. But, that would probably be a mistake. The Italy data are below.

   date    cases    deaths
21-Feb                21                  1
22-Feb                79                  2
23-Feb              152                  2
24-Feb              229                  6
25-Feb              322                11
26-Feb              400                12
27-Feb              655                17
28-Feb              889                21
29-Feb           1,128                29
1-Mar           1,701                34
2-Mar           2,036                52
3-Mar           2,502                79
4-Mar           3,089              107
5-Mar           3,858              148
6-Mar           4,636              197
7-Mar           5,883              233
8-Mar           7,375              366
9-Mar           9,172              463
10-Mar         10,149              631
11-Mar         12,462              827
12-Mar         15,113           1,016
13-Mar         17,660           1,266


The mortality rate in the 10 days between February 21 and March 2 averaged 2.5%. In the 11 days since, it has averaged 5% – twice as high. It currently stands at 7.2% of total cases.

Other noteworthy data points: the daily rate of growth in total cases has averaged 22% over the past 10 days, and the daily rate of growth in deaths has averaged 32% over the same period.  Again, this is the actual daily rate of growth, not an annualized or theoretical number.

Let’s look at the official US data, flawed as they might be.

   date    cases    deaths
3-Mar              124                  9
4-Mar              158                11
5-Mar              221                12
6-Mar              319                15
7-Mar              435                19
8-Mar              541                22
9-Mar              704                26
10-Mar              994                30
11-Mar           1,301                38
12-Mar           1,697                41
13-Mar           2,269                48


First, note that the numbers of US cases and deaths are roughly at the same point that Italy was on March 2 (2,036 and 52.) For this and other reasons, the US data are lagging Italy in most respects. Thus far, assuming a 10-11 day lag has proven to be a very accurate predictor of cases, deaths, closures, and overall social upheaval.

Currently, the mortality rate in the US is considered to be 2.1%. In Italy, it was 2.0% on March 1 – at which point it began its sharp rise to 7.2%. The number of reported cases in the US has averaged a daily rate of growth of 34% – much higher than Italy’s actual ROG of 22%.

On the other hand, the US daily ROG for deaths has been lower – averaging only 18% compared to Italy’s 32%.  This early on, however, the data have been all over the map – from a low of 8% to a high of 27%. So, we’ll examine a number of potential outcomes based on a variety of assumptions.

The first outcome assumes that the number of cases continues to grow at 34% per day and the number of deaths continues to grow at 18% per day.  Under these assumptions, total US cases by March 24 would be over 56,000. Deaths would total 308.

 US – #1
   date     cases    deaths
3-Mar              124                  9
4-Mar              158                11
5-Mar              221                12
6-Mar              319                15
7-Mar              435                19
8-Mar              541                22
9-Mar              704                26
10-Mar              994                30
11-Mar           1,301                38
12-Mar           1,697                41
13-Mar           2,269                48
14-Mar           3,038                57
15-Mar           4,066                67
16-Mar           5,444                80
17-Mar           7,288                94
18-Mar           9,756              112
19-Mar         13,061              132
20-Mar         17,485              157
21-Mar         23,407              185
22-Mar         31,336              219
23-Mar         41,950              260
24-Mar         56,160              308


Again, the number of cases is no doubt already much higher than reported, so I have no problem with this forecast. If anything, it is too low. Obviously, we’ll want to revisit this number as time goes by, but keep in mind the US population is 5.5 times that of Italy.

Note that 56,160 cases works out to 1,560 infected people per each city with a population of at least 500,000 or 166 per each city with a population of at least 100,000. My daughter’s high school and all churches in town just closed down after 5 cases were reported in our suburb of 30,000. Yesterday the Boston Marathon, which runs very close to our house, was cancelled for the first time since the outbreak of the Spanish Flu in 1918.

If we had 50-60 cases, I can easily imagine every restaurant, shop, gym and public building being closed. Like Italy, we might well be limited to only groceries and pharmacies (and funeral homes) in another 10 days or so.

Forget about the number of cases for a moment. Let’s focus on deaths – the number that folks of my vintage really care about. If the daily ROG in deaths were to match the 32% Italy has averaged lately, the March 24 tally would be 1,048 instead of 308.

But, remember, Italy’s mortality rate has mushroomed from 2% to 7.2% in the past 11 days. If US cases grew to 56,160 by March 24, a 7.2% mortality rate would equal 4,044 deaths. I wonder how many yogurt shops would remain open.  Oh, and by the way, Italy has more hospital beds per capita than the US.

My point isn’t to scare people or incite a hoarding panic at the local Costco. The coming headlines should do a good enough job of that.  But, I can’t sit idly by and watch the politicians spin this. Tests are not available to anyone who wants one. The mortality rate is not way under 1%. We are not very close to a vaccine.  We are not the most prepared country in the world. And, COVID-19 will unfortunately not disappear one day, like a miracle.

If we’re lucky, the rate of growth will decline and cases and deaths will level off in the not too distant future. If we’re real lucky, a new vaccine or existing medicine will prove effective. But, it probably won’t be in the next 10 days, despite what you might read on Twitter.

At today’s press conference / pep rally, President Trump insisted on shaking hands with nearly every one of the corporate executives who got up to tout his company’s efforts – each of whom also reached out and adjusted the potentially infected microphone.

Trump then refused to admit he had shaken hands with or palled around with several individuals who were known to have tested positive, and was nonchalant about getting tested himself.  Not a great example, to say the least.  Do yourself a favor: Listen to the scientists and doctors and ignore the politicians.

I’ll continue to update the above numbers as conditions warrant. I hope I’m very, very wrong – but fear I’m not.  Watch this space, and please be careful out there.




COVID-19: What to Expect — 1 Comment

  1. It is scary. One week makes a big difference. Just for thought. What will happen 3 months or 6 months from now?