The new serological data, which is provisional, suggests that coronavirus infections greatly outnumber confirmed covid-19 cases, potentially by a factor of 10 or more. Many people experience mild symptoms or none at all, and never get the standard diagnostic test with a swab up the nose, so they’re missed in the official covid-19 case counts.
Higher infection rates mean lower lethality risk on average. But the corollary is that this is a very contagious disease capable of being spread by people who are asymptomatic — a challenge for communities hoping to end their shutdowns.
The crude case fatality rates, covering people who have a covid-19 diagnosis, have been about 6 percent globally as well as in the United States. But when all the serological data is compiled and analyzed, the fatality rate among people who have been infected could be less than 1 percent.
But as infectious disease experts point out, even a seemingly low rate can translate into a shockingly large death toll if the virus spreads through a major portion of the population.
New York Gov. Andrew M. Cuomo (D) said Monday that the latest antibody numbers in New York City indicate that 25 percent of the population of 8.8 million has already been infected. The city has recorded more than 12,000 confirmed covid-19 deaths, and lists another 5,000 as probable deaths. That is an infection fatality rate between 0.5 and 0.8 percent, depending on which death toll is factored in. (A spike in all-cause deaths in recent weeks also suggests that some coronavirus-related deaths have not been captured by mortality statistics.)
“The death rate is much, much lower,” Cuomo said Monday, referring to the serology tests. He said the New York state rate appears to be 0.5 percent — which is one death per 200 infections.
That figure is still sobering to infectious disease experts. A rate of 0.5 percent “is way more than a usual flu season and I would think way more than the ’57 or 1968 [influenza] pandemic death toll, too,” Viboud said.
This is a novel, highly infectious virus, and so everyone who hasn’t had an infection is presumably susceptible. It is unclear whether and for how long a person who recovers from covid-19 will have immunity. Research shows that, in a completely vulnerable population that takes no precautions, each infected person will infect well more than two others on average, and some estimates put that reproduction number over three.
Epidemiologists have said somewhere between 40 to 70 percent of the population will likely become infected in the next couple of years if there is no vaccine and the public does not take aggressive measures to limit the spread of the virus.
“Do the math!” said Jeffrey Shaman, a Columbia University epidemiologist who has been studying the coronavirus since early in the outbreak.
Shaman and his colleagues have developed a model of the coronavirus spread that estimates that only 1 in 12 infections in the United States have been documented in official counts. That leads to an infection fatality rate of 0.6 percent, he said — a figure that roughly matches what has been seen in New York City.
At that rate, the United States could potentially experience 1 million deaths if half the population became infected and no efforts were made to limit the contagion through social distancing, a vaccine or proven therapeutics, Shaman said.
“That’s 20 times worse than a bad flu season,” he said Monday.
This is obviously a highly politicized pandemic, and some critics of the nationwide shutdown have seized on this early antibody data to argue that covid-19 isn’t all that deadly. A common refrain is that the disease is not significantly worse than the flu. In this view, the pandemic threat has been wildly exaggerated.
“Pandemic Is Over. Let’s Stop the Economic Suicide, and Get Back to Work,” declared a headline on a column by free market advocate George Gilder on the RealClearMarkets website. He said policies to control the viral spread “should be reversed summarily and acknowledged to be a mistake, perpetrated by statisticians with erroneous computer models.”
Kevin McCullough, a columnist for the conservative website Town Hall, wrote a column that ran last week with the headline “Antibody Testing: Proves We’ve Been Had!”
“Nearly everything we’ve been told about models, rates of infection, deaths, and recoveries was inaccurate,” he wrote. “The death rate in New York State isn’t 7.4%, it is actually .75%.”
But the two numbers describe different things: The first is a case fatality rate, reflecting deaths among people with confirmed diagnoses of covid-19. The second is the infection fatality rate, extrapolated from the antibody surveys.
In other words, both numbers can be correct, and useful.
“Those higher numbers of case fatality rates, they’re still true,” Viboud said. “It’s your probability of dying if you’re clinically sick with it, which is something that people may care about, too.”
Moreover, the fatality rate of a virus, however it is defined, is not an innate feature of the pathogen. It depends on many variables, including the age and health of the population and access to health care. Timing matters, too: In China the fatality rate was high during the initial phase of the outbreak, when hospitals were overwhelmed and doctors struggled to cope with the crisis.
There are so many numbers flying around that it is understandable that people may struggle to make sense of them. That’s become even more difficult in recent days due to serological studies that haven’t been peer-reviewed but have been presented to the public, typically with dramatic public health conclusions attached.
The most controversial such study came out of Santa Clara County, Calif., the heart of Silicon Valley. It claimed that the true number of infections in the county in early April may have been 50 to 85 times the official coronavirus case counts. Extrapolating from that, it gave an estimated infection fatality rate of between 0.12 and 0.2 percent.
That provided what seemed like reliable ammunition to those who believe the danger of covid-19 has been overstated. They said the numbers show this disease is not much different from seasonal flu.
A commonly cited statistic about seasonal flu is that it has a fatality rate of 0.1 percent, That, however, is a case fatality rate. The infection fatality rate for flu is perhaps only half that, Viboud said. Shaman estimated that it’s about one-quarter the case fatality rate.
Thus, even if the Santa Clara numbers are roughly accurate about the spread of the virus and its lethality, it would still be deadlier than the flu.
Statisticians and other epidemiologists have argued that there were also errors in data analysis in the Santa Clara report, which was led by researchers at Stanford University. The researchers used advertisements on Facebook to call for volunteers, and critics said that may have led to an unrepresentative sampling of the county. The critics noted that even a small rate of false positives can skew a survey in which the targeted disease has low prevalence.
Shaman, the epidemiologist, said he hopes the higher estimates for infections and commensurately lower fatality rates are correct. Everyone needs more testing and better data, he said.
For now, the preponderance of evidence describes an exceptional pathogen. For someone like Shaman, who wears a mask when he walks the dog in New York City and is conscientious about washing his hands, there is no doubt that the endgame is far away — that until a vaccine or reliable therapeutics are in place, or when so many people have been sickened and then recovered that the country achieves “herd immunity,” everyone needs to try to limit the spread of the virus.
“We have completely disrupted our society through our social distancing practices, shutting our businesses, closing our schools, and still this thing has already killed 53,000 people in the United States,” he said. “We have a very long way to go until we get herd immunity.”