Determining covid-19’s death rate is tricky; it’s a work in progress
By Paige Winfield Cunningham
The Washington Post
The novel coronavirus is clearly deadlier than the seasonal flu, despite President Trump’s ongoing efforts to downplay its risks.
But how much more deadly? Figuring that out isn’t easy.
That’s because the calculation involves an elusive number: the total people infected, not just those known to have tested positive. While testing has ramped up dramatically in the United States, researchers are still trying to figure out precisely how far and wide the virus has spread.
To complicate matters, there’s more than one way experts calculate fatality rates. The fatality rate would appear lower if deaths are measured against a projection of all estimated infections. If deaths are measured against only those with symptoms – a smaller group, especially since many with coronavirus are asymptomatic – the fatality rate appears higher.
“It’s confusing,” said Robin Patel, president of the American Society for Microbiology. “In many cases we use that term fatality rate, and it can mean a lot of different things.”
Conservatives have latched on to the lag in deaths as coronavirus cases spike to claim the illness’s lethality has been overstated. “What they’re not telling you is the death rate,” radio host Rush Limbaugh said on his show Monday. “The death rate is falling. But you don’t know that because they are simply reporting this massive increase in cases.”
Trump has been making similar claims, as he tries to paint the country as post-pandemic and on the road to recovery.
The World Health Organization estimated the infection fatality rate for the first time last week.
WHO chief scientist Soumya Swaminathan said the rate is 0.6 percent. While that’s a small percentage, it could translate to 1.7 million deaths in the United States should every single person become infected.
But the general infection fatality rate isn’t necessarily helpful for officials and individuals trying to better grasp the virus’s threat. The risk varies widely by age and health status, with older people with pre-existing health conditions at much greater risk of dying. For example, only 30 kids in the U.S. younger than age 15 have died of covid-19, out of 115,000 deaths tallied by the Centers for Disease Control and Prevention.
“We think about this as: ‘If any average person gets it, what is their chance of dying?’ ” Patel said. “Well, nobody is an average person, so that statistic is misleading.”
The CDC has focused on fatalities among only those with symptoms. That approach can make the virus seem more deadly, because it involves a smaller pool of people.
The agency has said people who develop symptoms have a 0.4 percent risk of dying overall, although the agency hasn’t explained how it calculated that rate. It also provided age-based estimates, saying that those over age 65 have a 1.3 percent chance of dying while those under age 50 have only a 0.05 chance of dying.
But the CDC says the fatality rate drops to 0.26 percent overall if it’s measured against all infections — an estimate considerably lower than the WHO’s estimate.
“It’s confusing because there are different denominators, different rates,” said Anirban Basu, professor of health economics at the University of Washington. “So it’s important to lay out the different subgroups we are talking about.”
The fatality rate also depends on knowing the total number of deaths, which may be underestimated in some places. Media reports have chronicled a rise in the number of patients dying at home in New York City and Houston amid case surges before they are ever tested for the coronavirus.
It is also hard to estimate the true fatality rate because many people aren’t tested.
Calculating the infection fatality rate requires researchers to make an educated guess about how widely the virus has been circulating. Thirty-five percent to 45 percent of cases may not result in symptoms, complicating the task.
Neerja Sood, a professor of health policy at the University of Southern California, estimates the infection fatality rate is probably 0.13 percent for people outside nursing homes and 0.26 percent when people in nursing homes are included. That estimate is based on a survey of Los Angeles County residents who were tested for antibodies to the virus.
But Sood told my colleague Joel Achenbach that the rate could easily change if any of the numbers used to calculate it are off: “Depending on how you make the assumptions, you can get different answers for the infection fatality rate.”
The fatality rate looks closer to 5 percent if deaths are measured only against known cases. Basu said this “case fatality rate” is typically between 4 percent and 7 percent in most countries.
In the United States, it’s 4.2 percent, according to the statistics website Worldometer. That’s the total number of people who have died of the virus divided by the total number of people who tested positive for it.
Different countries have reported vastly different case fatality rates, probably reflecting differences in the age of their population, access to medical treatment and testing rates. Based on the Worldometer figures, the case fatality rate is 3.9 percent in Brazil, 4.6 percent in Germany, 9.5 percent in Spain and 15.5 percent in the United Kingdom.
But when it comes to people who actually develop symptoms, far fewer of them — about 1.3 percent — actually die, according to Basu’s own research.
That’s much less lethal than what the case fatality rates suggest — but still far more lethal than the seasonal flu, which kills only about 0.1 percent of those who fall sick with it.