Day after day, news organizations track the death toll for the novel coronavirus, often relying on tallies maintained by universities or researchers. The figure for the United States crossed 1,000 deaths on March 25 — a sharp increase from the 100 or so reported just nine days earlier.
Still, President Trump has taken comfort in often citing death tolls for the 2009 swine flu pandemic and the seasonal flu, appearing to suggest the death toll for covid-19, the disease caused by the coronavirus, is much smaller. But the numbers he cites are not comparable to the covid-19 statistics.
The daily tracking number for covid-19 is substantially below the real figure of deaths that can be attributed to the coronavirus. It represents only the bare minimum — confirmed deaths reported by hospitals, medical providers and state health authorities as caused by covid-19. The actual number is substantially higher, but that will only become apparent after statistical modeling on excess mortality during this period.
Here’s an explanation.
The Centers for Disease Control and Prevention has estimated that from April 12, 2009, to April 10, 2010, there were 12,469 deaths in the United States because of the H1N1 (swine flu) virus. That was the midpoint of an estimate that ranged from a low of 8,868 to a high of 18,306.
But that was an after-the-fact report. What was the CDC reporting as the swine flu pandemic unfolded? Here’s a comparison of the first 65 days of the 2009 swine flu in the United States and first 65 days of the 2020 covid-19 outbreak.
- April 15, 2009 — first infection detected
- June 19, 2009 — 21,449 cases, 87 deaths
- Jan. 20, 2020 — first infection detected
- March 25, 2020 — 69,344 cases, 1,050 deaths
The 2009-2010 swine flu pandemic is considered to have been relatively mild — and in the United States, it hit harder in a second wave in the fall. Clearly, covid-19 is striking with more force — and faster.
“We know that if you just look at laboratory-confirmed deaths, it is an undercount, especially in the case of flu,” said Cécile Viboud, senior research scientist in the Division of International Epidemiology and Population Studies of the Fogarty International Center at the National Institutes of Health, where she is part of the Multinational Influenza Seasonal Mortality Study (MISMS).
Viboud said that many deaths are triggered by the flu but occur later, via secondary bacterial infection or exacerbation of underlying chronic diseases, even as flu symptoms may have disappeared. So a death might be recorded as a heart attack, not as having been caused by the flu, even though the flu led to the cardiac arrest. The CDC, which tracks weekly flu deaths, also closely tracks pneumonia deaths because pneumonia is so often a consequence of the flu.
“Only counting deaths where influenza was recorded on a death certificate would be a gross underestimation of influenza’s true impact,” says the CDC in a website explanation of how it estimates deaths caused by seasonal influenza.
Researchers such as Viboud use a time series analysis to estimate how many excess deaths in a given period can be attributed to a viral disease such as the flu but are not counted in the official death statistics.
For instance, in the 2009 swine flu pandemic, 2,096 laboratory‐confirmed deaths were reported in the United States from Aug. 30, 2009, to April 6, 2010, the second wave. But a 2013 study that Viboud participated in concluded that the excess deaths that could be attributed to the pandemic in that period amounted to about 14,800. In other words, the official lab count in the United States missed 6 in 7 deaths.
Another 2013 study, examining data across the globe, concluded that the World Health Organization laboratory-confirmed count missed 9 in 10 deaths that could be attributed to the swine flu pandemic.
There is debate within infectious-disease circles about how to structure such statistical models, including accounting for seasonal mortality patterns and determining what types of deaths to include. That’s one reason that there can be such wide ranges in the estimates.
But there already is anecdotal evidence that many deaths that stem from covid-19 are not being recorded as such. Doctors and nurses have told BuzzFeed News that deaths in the United States are not being correctly reported, in part because overwhelmed hospitals and a shortage of tests meant people died before it was determined whether they had contracted the virus.
And in Corriere della Sera, one of Italy’s leading newspapers, the mayor of the northern Italian town of Nembro reported that even though 31 people are recorded as having died of covid-19 from January to March, 158 people are listed as dead in the town — 123 more than the average. “The difference is enormous and cannot be a simple statistical deviation,” wrote Claudio Cancelli, suggesting that the actual death toll was four times higher.
Viboud said that the Nembro figures were “not entirely surprising” but that they would need to be studied more closely, given that they were a relatively small data set in a limited period. She said the death reporting system in Europe is quite fast, in contrast with that in the United States, so the actual covid-19 death toll might be more quickly estimated there. In the United States, she said, the collecting of death certificates can take two or more years, so a time series analysis of the current pandemic probably could not be completed until at least 2022.
Meanwhile, the United Kingdom’s numbers jumped when officials began counting deaths outside hospitals, including those in homes and care homes. Radio Free Asia and Bloomberg News reported on a large discrepancy between the official death toll in Wuhan, China, and cremation figures. The official toll showed about 2,500 deaths from covid-19 in the city, while a count of cremation urns indicated far more people had died.
“In the case of a new cause of death (like covid-19), there are additional issues about incomplete capture of deaths due to the new cause, including a lack of an ICD-10 code,” said Keri N. Althoff, an epidemiologist and associate professor at the Johns Hopkins Bloomberg School of Public Health, referring to a WHO medical classification list. She noted the medical classification code for covid-19 was only released the week of March 23. “There will likely be a lot of work on the ground and with statistical methods to estimating the number of deaths from covid-19 prior to the release of the ICD-10 code.”
The Bottom Line
Trump is comparing apples and oranges. The figures he cites for the swine flu and seasonal influenza are after-the-fact estimates that try to account for all excess deaths that could be, in some way, attributed to a flu infection. Studies indicate that the official lab counts may be missing as many as 9 in 10 deaths.
The numbers so far for covid-19 are only laboratory-confirmed deaths. So, depending on how the United States handles this public-health emergency, the future estimates of the impact of covid-19 may be grim reading indeed.
Send us facts to check by filling out this form
Sign up for The Fact Checker weekly newsletter
The Fact Checker is a verified signatory to the International Fact-Checking Network code of principles
The Washington Post Fact Checker is working with the CoronaVirusFacts/DatosCoronaVirus Alliance, a coalition of more than 100 fact-checkers who are fighting misinformation related to the COVID-19 pandemic. Learn more about the alliance here.