Exactly how deadly COVID-19 is remains up in the air. Limited testing and undetected cases — people with no symptoms or ones so mild they don’t seek medical attention — make it hard to pin down how many are infected. And that number is crucial for calculating the ratio of people who may die from COVID-19.
Enter the Diamond Princess cruise ship. Quarantined at sea off Japan after a passenger tested positive for SARS-CoV-2, the ship became a natural data lab where nearly everyone was tested and few cases of infection were missed.
Infections and deaths onboard suggest that the disease’s true fatality ratio in China is about 0.5 percent, though that number may vary from place to place, researchers report March 9 in a paper posted at MedRxiv.org.
That 0.5 percent is far less than the 3.4 percent of confirmed cases that end in death cited by the World Health Organization, but troubling nonetheless. The WHO’s number has come under fire because the true number of people infected with the virus worldwide is not known.
“How worried should we be?” says Timothy Russell, an epidemiologist at the London School of Hygiene and Tropical Medicine who does mathematical modeling of disease outbreaks. “Well, it’s more severe than the flu.” Flu — which annually kills hundreds of thousands worldwide — has an estimated 0.1 percent fatality rate.
As of February 20, tests of most of the 3,711 people aboard the Diamond Princess confirmed that 634, or 17 percent, had the virus; 328 of them did not have symptoms at the time of diagnosis. Of those with symptoms, the fatality ratio was 1.9 percent, Russell and colleagues calculate. Of all infected, that ratio was 0.91 percent. Those 70 and older were most vulnerable, with an overall fatality ratio of about 7.3 percent.
Extrapolating those numbers to China, the team estimates that 1.1 percent of symptomatic cases there turned deadly. Considering asymptomatic cases drops that ratio to about 0.5 percent in China, the team calculates.
These ratios depend on available health care and public health measures, Russell cautions. And there are still uncertainties in the data, he notes. For example, some patients initially counted as asymptomatic may later develop symptoms, or even die. So the true fatality rate may be somewhat higher, “0.6 or 0.7 [percent], but it’s still a good ratio.”