There are enough people out there that are accusing everyone of overreacting to COVID-19, calling it “just the flu,” that I decided to jump into some of the data to see how the two viruses really compare. Both are respiratory illnesses, albeit ones caused by completely different families of viruses, so they’re bound to act at least slightly differently. But for the most part, the two diseases act in a similar fashion: They both infect the lungs, causing fluid buildup that leads to pneumonia, which is the primary cause of death. Both also can spread to the rest of the body through the circulatory system, damaging organs and leading to other causes of death. It appears now—a few months into the pandemic—that the SARS-CoV-2 virus causes far more destruction, as well as more widespread destruction, to the blood system and organs periphery to the lungs than does influenza. It can cause clotting that leads to strokes, heart attacks, kidney failure, and much, much more. We don’t know yet the full extent of what the virus is doing to the human body, and there is growing evidence that it might be altering the very blood cells themselves.
Now, outside of general descriptions like this, it gets really problematic to try to compare the specific impact and the lethality of the two viruses, and trying to do so has led many people to rely on some bad logic and flimsy rationalizations in the process. The main culprit here is that with our lack of knowledge about the fledgling SARS-CoV-2, all of the numbers we see are the result of direct observation—testing and direct reports from doctors and hospitals. Flu numbers, by contrast, are not compiled through direct testing and observation but through statistical analysis. The Centers for Disease Control and Prevention gathers reports of all illnesses and deaths every year and sorts them according to symptoms, analyzing them against past and current trends.
A reliable statistical analysis like this is far superior to direct observation when it comes to a widespread illness—especially one like the flu which we experience every year, in similar forms but different extremes. We know the flu, we’ve been dealing with it for centuries, and thus we understand how it manifests and affects us, both individually and as a society.
The new SARS-CoV-2 virus is, of course, a different story. We know other coronaviruses, from the common cold to MERS, but we aren’t yet certain exactly how this one affects us. Every day, we seem to find new ways that the virus is attacking our bodies, which demonstrates that we still have a long way to go to fully understand the extent of its pathology. Direct observation and testing, thus, are the appropriate path to tracking it—for now.
Apples vs Oranges
But these two forms of data collection are not equal and thus not really comparable. Direct observation overlooks a lot of cases. The data acquired using testing and case-by-case counting is not an accurate total of cases and deaths. They are, instead, the absolute lowest amount. (I wrote about this in a previous article.) As the data on excessive deaths reported in April and May rolls in it should become clear that COVID-19 is even more destructive than has been reported. Deaths from heart attacks, strokes, pneumonia, and more, all of which had not been previously connected to the disease but that almost certainly are a result of it, will have to be included in the totals.
Thorough statistical analysis can fix gaps in testing and reporting like this, but it takes time. For a good analysis of a new disease like this, it can take years. The CDC analyzes flu illnesses and deaths each year, but the models adjust and update as new information is acquired, the same way that economic models and reports are. (Things like job reports and GDP reports are released initially as preliminary findings and then adjusted over the coming months before they are considered “final.”) For COVID-19, we don’t have past years to compare, so the experts have nothing to start. The model has yet to be built—although the process has begun.
Apples Here! Get Your Apples Here!
But I don’t feel like waiting years to try to think of a more proper comparison between the two viruses, so I figured I’d take a stab at it. We know the model from the CDC: They release the results for each flu season. (You can find the data for 2010 through 2019 on their website, which is what I cull from for my analysis here.) The CDC calculates the number of symptomatic illnesses, medical visits, hospitalizations, and deaths resulting from the flu.
In order to gauge the full scope of the problem, I also tried to calculate the number of asymptomatic and very mild cases of the flu. That’s more difficult. There’s a good reason why the CDC doesn’t try to include this information each year, other than declaring ranges of how large and widespread the entire nationwide infection could be. Without mandatory testing of every citizen, constantly, there’s no way to really know this for certain because … well … you can’t really count people who show no (or few) symptoms. But there are various studies that have tried to determine the rate of asymptomatic cases that result from the flu, and from those, I was able to set a range. It’s a wide range—there are somewhere from 33% to as much as 75% of all flu cases that are asymptomatic—but that reflects the size and scope that I could find from various studies. From that, I made calculations on the total viral spread of the flu each year. (See below for each individual year’s data.)
The Alchemy of Turning an Orange into an Apple
The hard part of creating an adequate comparison for COVID-19 is figuring out which of the data now being tracked is analogous to which flu stats. To put the COVID data into buckets analogous to the flu, I took the “medical visits” and “deaths” figures from the COVID-19 Dashboard from the Center for Systems Science and Engineering at Johns Hopkins University. The “hospitalizations” I got from The COVID Tracking Project’s US Coronavirus Hospitalizations.
(These COVID deaths numbers are almost certainly undercounted, but since we don’t have data on extra deaths for April and May, we can’t yet try to accurately calculate how much those numbers are off by. So I’m using the reported numbers here, for total confirmed cases, hospitalizations, and deaths, knowing full well that they are certainly the floor and not the ceiling, and are probably undercounted by significant amounts.)
Why did I put the total U.S. confirmed COVID-19 cases in the “medical visits” line instead of the “symptomatic illnesses” line? For the flu, the medical visits line is the portion of symptomatic patients who seek out help because the virus hits them badly enough to see a doctor or at least go get tested. Total symptomatic illnesses are determined by a calculation using the most conservative multipliers of past flu seasons. According to this CDC data, the average percentage of people with serious enough symptoms that go to the doctor each year with the flu stands at about 46%.
With the chaos of the past few months, combined with stay-at-home orders and the fear of going out and doing anything, people are understandably avoiding taking trips to the doctor unless absolutely necessary—and sometimes not even going then. We have ramped up testing somewhat over the past few weeks, which is probably mitigating a little of that undercount, but doesn’t change the underlying situation. The number of positive cases almost certainly only captures those who are the most sick, which is why I put the total COVID cases number that we’re familiar with in the medical visits line and not the larger symptomatic illnesses line.
Because of this fear of going to the doctor, the proportion of medical visits is probably lower than that of the flu. How much lower can’t be known, but it’s safe to assume that there were a bunch of people that should have gone to see a doctor but didn’t. To try to offset that, I lowered the 46% rate of the flu to 40% for COVID-19—basically saying that almost 1 in 10 people that got sick and probably should have gone to see the doctor, and normally would have, didn’t do so this spring. While I don’t know of sure how accurate that is, if feels approximate.
As you can see, using that 40% rate of medical visits, I calculated the total symptomatic COVID-19 cases in the U.S. through June 1st, coming up with a figure of nearly 4.5 million cases.
But … How Many People Have Gotten It, Really?
The big question of the hour is: How many people in the United States have already caught the virus? To calculate that total, we need to know how many people who catch the virus have either very mild symptoms or none at all. Figuring this out, for a brand-spanking new pathogen, is very difficult to decipher. It’s been one of the major questions about the SARS-CoV-2 virus since it emigrated from Wuhan, China, at the end of 2019. We have some clues that, if nothing else, should give us enough info to set a reasonable range, like we did with the flu.
The number of silent COVID-19 carriers has been a matter of debate, mainly because the long gestation period of SARS-CoV-2 makes it difficult to tell who is truly asymptomatic and who is merely presymptomatic at time of testing. Some surveys have been finding very high levels of asymptomatic positive tests, but follow-up rounds of testing days and weeks later reveal that many of those people do develop symptoms.
The story of Vo is possibly instructive in this instance. Vo, a village in northern Italy inland from Venice, was one of the first places hit hard in the epidemic. Like many places in Italy, it was quickly locked down—but then something important happened. Researchers at the nearby University of Padua used the isolated population as a chance to learn more about the virus. In what was maybe the first instance of anything resembling a scientific study of SARS-CoV-2 and its resulting COVID-19 disease, everyone in the village got tested—twice. In the first round of testing, researchers found roughly 10 times more positive tests than they had expected based on the number of people with symptoms. At first, it looked as though 90% of cases were asymptomatic. Reports over time revised that number down to somewhere between 75% and 50%, showing that many of the asymptomatic carriers were really just presymptomatic ones after all.
Other surveys in various countries have reported varying numbers too, as high as 90% and as low as 25%. The truth—which I emphasize, once again, we don’t know yet—is likely somewhere in the middle. All things considered, then, and for comparison’s sake, I think it’s safe to use the same range of asymptomatic + mildly symptomatic cases might be similar to the flu that we used earlier: 33% to 75%.
So, It’s Bad …
When those who are trying to mischaracterize the situation and say that COVID-19 is similar to the flu are not only comparing apples to oranges, they’re also trying to compare bins of fruit that don’t even match and hope you don’t notice. Often, they will try to compare the death rate from all reported COVID cases to only those of hospitalized flu patients, or to compare the death rate of symptomatic flu cases (the commonly referenced 0.1%) to a figure that tries to calculate all coronavirus cases.
All of these comparisons are being made in bad faith. In reality, if COVID-19 were only as bad as the flu, to have the number of deaths from COVID-19 that we already have, we would need to have already exceeded a total number of cases larger than the entire population of the United States, having left herd immunity in our rearview mirror long ago. That is clearly not the case.
As we can see from the two charts above, on fewer than half of the total number of hospitalizations (for the entire flu season), COVID-19 has already killed three times as many people (through June 1). Whereas the rate of deaths to hospitalizations for the flu is about 8%, for COVID-19 it’s a staggering 49%. Even in the best-case scenarios, where three-quarters of all people to contract the viruses show no symptoms, the death rates are 0.03% for the flu to 0.58% for SARS-CoV-2.
To put that in perspective, if the entire 330 million person population of the United States were to catch both diseases, the flu would kill 99,000 people, whereas COVID-19 would kill 1.914 million people. That’s nearly 20 times more dangerous.
Everything I’ve seen since February shows me that this disease is extremely dangerous, but this is worse than I thought I would find going into this little homework project. I expected to find a mortality rate that was about 10 times worse then the flu, bad but not necessarily apocalyptic, but which nevertheless was horrific because of how easily this virus spreads. Turns out I was wrong. It’s worse.
Data Dump
Wanna check my math? I culled all my calculations from the CDC’s yearly reports on the flu season. I used the average numbers in the chart above, but here are all of the individual years if you also want to see the variability in the individual flu seasons.
Resources
Wanna read more? Here’s a list of places that I got the bulk of my research from for this article, plus some other evergreen resources that will continue to be of interest throughout the pandemic.
Medical Visits data from COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU)
https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6
Hospitalizations from The COVID Tracking Project’s US Coronavirus Hospitalizations page
https://ycharts.com/indicators/us_coronavirus_hospitalizations
Deaths data from COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU)
https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6
How CDC Estimates the Burden of Seasonal Influenza in the United States
https://www.cdc.gov/flu/about/burden/how-cdc-estimates.htm
Image in the intro image by Alberto Giuliani, via Wikimedia Commons.