Coronavirus: Who’s Going to Die
Wash your hands and whatnot, sure, but let’s ask the question that people are actually asking: Who’s going to die from this thing?
Excuse my bluntness, but I’m an ER doctor and people ask me blunt questions all the time with the expectation of getting equally blunt and actionable answers, so that’s what I’m trying to address here. By this point in time, we’ve all heard of the “coronavirus” or the virus “SARS-CoV-2” that causes the disease “COVID-19” or whatever specific thing you want to call it. We all know what we’re talking about, and it’s obviously spreading quickly across the world. It also seems somewhat manifest by now that quarantine efforts are not going to control this infection as well as they were able to in recent prior battles with SARS or ebola.
We’ve all also seen a million local news reports telling us to wash our hands and stay home when possible, as well as countless expert spots explaining that the virus is actually an “RNA virus in a lipid bilayer that belongs to a family named ‘coronavirus'” and a bunch of other associated information that answers questions that nobody is asking.
So let’s answer the question that, in my experience, people are actually asking: Are we going to die?
Are We Going to Die?
First off, almost certainly not. The odds of death for you, the people you know and care about, and our community at large are generally not immediately catastrophic. There are particular at-risk demographics, however, and we will discuss those at length and with the gravity that they deserve. The risk of increased strain on the healthcare system is also real but not something I will focus on here.
The more detailed answer is that nobody can completely tell you who might die, because this virus is new and we’re just starting to learn about it. However, we have preliminary data that are, in my opinion, quite decent and sometimes reassuring, and I want to discuss them with you.
In my investigation, the best data I could find come from China’s version of the U.S. Centers for Disease Control and Prevention (the U.S. CDC), which is helpfully named the “China CDC” or the “CCDC.” This organization publishes a weekly report, similar to that of the United States’ CDC, and they released a ten-page rundown on the coronavirus on February 14th, 2020, which covers what they have learned from 72,314 COVID-19 patient cases.1 Importantly, 44,672 of these cases were “confirmed,” meaning that they were included based on advanced laboratory confirmation (nucleic acid testing) that proved that they were actually caused by the novel coronavirus.
Now if you’re asking, “Isn’t that the report which found a 2.3% death rate?!” Yes, it is, but that’s what we’re going to talk about. In my opinion, there are two very important factors which will affect your interpretation of that overall mortality rate, and they’re worth thinking about:
- The death rate is probably wrong
- The death rate isn’t the same for everyone
My strong suspicion is that the currently reported death rate is too high. I also believe that we have actionable information about the demographics who actually die from this virus, and we should appreciate just how stark the risk divisions amongst us might be.
1. The Death Rate Is Probably Wrong
What is a “death rate” or “mortality rate” or “case fatality rate” anyway? Well, it’s pretty straightforward:
(Number of known deaths) / (Number of known cases)
That is, take the number of people who die from a disease and divide it by the number of people who had the infection or disease to begin with. Simple.
Except each of these two numbers is prone to bias.
For the numerator, sometimes people take a long while to die from something. With a new (“emerging”) disease, you might know of many cases and not many deaths simply because there just hasn’t been enough time for people to die yet. Grim but true. This effect would artificially lower the death rate.
However, for COVID-19, this is almost certainly not the case. This new virus is a respiratory virus within a family that we understand quite well, and it seems to cause its trouble across the same few weeks that we all expect for respiratory infections. If you recover after that, you’re in the clear. This means that we likely have a pretty good read on the “known deaths” number; grim or not, if you’re dying from COVID-19, you’re probably coming to medical attention, getting tested, and getting recorded correctly. Overall, it’s probably a reasonable numerator.
The denominator, on the other hand, is likely wrong.
In the above equation, the “number of known cases” depends on people coming to medical attention and getting tested. In order for that to happen for any given person, four things must occur:
- They must be sick enough to seek medical attention
- The medical provider must suspect the illness
- The medical provider must have a test for the illness that they choose to use
- The test must accurately detect the infection
For any infectious disease, COVID-19 included, a lapse in any of these steps means one more actual infection that doesn’t get recorded as a “confirmed” case. To use influenza (the “flu”) as an example, the official CDC reporting leads off with an immediate asterisk which clarifies that “influenza surveillance does not capture all cases of flu that occur in the U.S.” and that some of the agency’s estimates are “adjusted for the frequency of influenza testing and the sensitivity of influenza diagnostic assays.”2
In the current case of COVID-19, we have every reason to believe that many people are experiencing mild or asymptomatic disease. These people are almost certainly not seeking medical attention, even though they have true infections and would be “confirmed” cases were they to be tested. There are also certainly patients who see a physician and don’t get tested (often because we have no common or quick test available), as well as patients who get tested but have a false-negative test result. None of these people will be recorded as a “confirmed” case, even though they rightfully should be.
The net result of all of these “missed” diagnoses is that the denominator for the current COVID-19 death rate is almost certainly too low. And this means that the death rate we are seeing is almost certainly too high.
For comparison, the CDC is currently estimating 32–45 million flu illnesses in the U.S. this year but only 14–21 million flu medical visits and 18–46 thousand flu deaths.2 First of all, that’s notable because TENS OF THOUSANDS of people die from the flu every year in the United States, and I think we forget about that. Second of all, it’s notable because the CDC is telling us that the actual disease burden is higher than the number we get from just “medical visit” data, meaning that they’re accounting for people who aren’t getting tested.
With this in mind, if you calculate the U.S. flu death rate this year using only “medical visits” as the denominator, you end up with a death rate of 0.09%–0.33%, with an overall estimate of 0.21%.2,3 Remember those numbers, because we’re going to revisit them later.
The take-home message here is that the COVID-19 death rate that is currently being reported is probably not too low or even accurate. It is probably too high.
Data from the recent WHO-China Joint Mission report may also support this idea, as the report’s authors noted a reduction in death rate over time: “In China, the overall [case fatality rate] was higher in the early stages of the outbreak (17.3% for cases with symptom onset from 1-10 January) and has reduced over time to 0.7% for patients with symptom onset after 1 February (Figure 4).”4 The study doesn’t discuss why this might be happening, other than noting that “the standard of care has evolved over the course of the outbreak,” but it seems unlikely to me that changes in supportive care alone could account for such a dramatic drop. I would instead suspect at least some contribution from changes in disease identification and reporting.
The WHO report, unfortunately, does not provide its raw data for us to run our own analyses. This makes it difficult to investigate deeper questions, such as whether recent severely ill cases simply haven’t been present long enough to show up as mortality figures, which could be depressing any newly observed mortality rates. The report does mention that the median time from onset to clinical recovery for mild cases is only two weeks, whereas it is three to six weeks for patients with severe or critical disease. The authors also report that the time from disease onset to death “ranges from” two to eight weeks for patients who eventually die; this is potentially a long lag time, but the report does not provide median time to death or other population statistics that could help us understand how much impact any possible lag might be having on mortality calculations. Hopefully future reporting will clarify this and will make note of possible limitations of any reported mortality numbers.
2. The Death Rate Isn’t the Same for Everyone
Again, the China CDC found an overall death rate of 2.3%: 1,023 deaths among 44,672 confirmed cases.1 We have already established that this is probably too high, but that’s not the end of the story. To go deeper, let’s look at how the CCDC reports the death rate by age group:
I don’t know if you can see it, but the bars start to go up toward the end of that graph. Even among people who were sick enough to seek medical attention, get tested, and have a positive test result, the only people who died at any alarming rate were 50 years old or older. And I don’t say that lightly. As a physician (and a non-psychopath), I care very deeply about people of all ages, and I want to see completely flat death bars for all of you. However, I don’t control this, and the graph shows what it shows.
Also notably, this death rate wasn’t some weird byproduct of people being more likely to get the disease at older ages:
We can see in the above graph that the bulk of disease incidence was in the young adult to upper-middle-age brackets, and it’s pretty well distributed across a number of those brackets. Yet the mortality rates were quite separate from this. Wikipedia says that the Chinese population age and gender distribution looks like this overall:
At a bird’s-eye level, this general population demographic chart mirrors the COVID-19 incidence graph above it, with the clear exception that people younger than 20 years old just aren’t getting this virus (or at least weren’t getting tested for it). As always, youth is wasted on the young.
So now I have some new data processing to show you. Using the CCDC data, I recalculated the Chinese coronavirus disease incidence (how many people get and test positive for the virus) and absolute death counts, and grouped them by age buckets, using age 50 as the cut point. Here are the results:
You’re probably already doing the mental calculation of death rate here, but let me save you the time and just show it to you. This is fairly stark:
So take a look at that. The currently reported COVID-19 death rate for people younger than 50 years old is approximately 0.3%, which as you’ll recall from earlier in this article, is basically the same as the seasonal flu death rate, were we to calculate the latter using the uncharitable denominator that we’re currently using for COVID-19.
If you’re wondering what you get if you analyze flu mortality by age instead of just overall numbers, then good for you, because that’s a great question. Let’s look at some recent CDC data:3
And that same data bucketed by the age groups we used for COVID-19 above:
Now, you might be looking at these charts and thinking, “Hmm. It sure looks like COVID-19 is about 10x as deadly as a typical seasonal flu, adjusting for age bracket,” and it’s possible that is correct. However, in my opinion, this is when we get back to the first point from way above, which was that the currently reported death rates for COVID-19 are almost certainly too high. It does seem directionally to be a more deadly disease than a typical influenza, but I doubt that it’s truly ten times as deadly. Probably the truth is somewhere lower than that, within an order of magnitude.
Another important thing to keep in mind is that the true inherent deadliness of the flu is masked by the effectiveness of the flu vaccine (the flu shot) on a population level. Were it not for the flu vaccine, the typical seasonal mortality rate for influenza would be higher, as the vaccine reduces not just the number of flu cases but also the severity of those that do occur.5 The commonly used drug oseltamivir (Tamiflu) might also be reducing the death rate of the flu. It’s not unreasonable to wonder if a person not vaccinated against the flu and not treated with oseltamivir might have a risk of death from flu that is elevated and more similar to that which any of us has for coronavirus, which has no vaccine or treatment available.
The Death Rate Isn’t Just About Age
The CCDC data also suggest that general health status affects your likelihood of dying from COVID-19. Again, we don’t have perfect data, but the results are suggestive:
The chart above tells us that people with no underlying conditions (comorbidities) have an overall death rate of 0.9%. That doesn’t account for age, so it may just be showing us that young people tend not to have comorbidities and also tend not to die from coronavirus, but it at least generally suggests that being healthy at baseline is protective against death from this new virus. On the other hand, having diabetes or heart and lung disease puts people into a riskier category. It’s not unreasonable to assume that age and underlying health are somewhat independent actors, and that their combination likely predicts death rate from COVID-19.
Pregnancy and Breastfeeding
This section is not based on the CCDC data, but I’m hearing questions about COVID-19 and pregnancy/breastfeeding, so I wanted to include it. Based on current CDC reporting, here is the limited consensus opinion (which MIGHT BE WRONG BECAUSE IT IS PRELIMINARY):6
- Pregnant women can be more susceptible to viral respiratory infections in general, which means that they might be more likely to be affected by COVID-19
- Related coronaviruses (e.g., SARS and MERS) have been linked to pregnancy loss (e.g, miscarriage and stillbirth), but there is no data so far for whether COVID-19 can cause this
- There is limited evidence that the COVID-19 virus does not transmit from a mother to a fetus or newborn any differently than it might transmit to any other person
- There is limited evidence that coronavirus neither exists in nor is transmitted by breast milk
Go in Peace
I really don’t think we’re going to die. The odds are strongly against it. The likelihood is that you’ll get a cold. We do have to take extra care of the vulnerable among us, however, and the high death rates for the elderly especially are terrifying and no joke.
The advice to wash your hands regularly, thoroughly, and with soap and water is good. Alcohol-based hand sanitizer is a close second. And make sure to use moisturizer regularly so that you don’t end up with dry, cracked, or bleeding hands; your skin is a wonderful immunologic barrier, and harming it would be a mistake. Don’t touch your eyes or your nose. Stay home when you can, and help older people in your community to avoid exposure whenever possible. Zinc lozenges might help, just like they do for generic common colds, but there’s certainly no coronavirus-specific evidence for them. Wearing a mask will help prevent you from spreading the virus if you’re infected, which is very important, but they aren’t a silver bullet in terms of preventing you from becoming infected by other people, especially if you’re not using a high-quality mask with perfect technique.
In any case, we don’t have all the data yet, but my physician and public-health colleagues in China have done a lot of hard work and have provided us a wonderful service and head start in their reporting. I’m grateful to them, and I hope that my conclusions based on their work are correct and that they provide you some comfort and some actionable direction.
Good luck, stay safe, and I hope with all my heart that we all come out the other side of this quickly and in one piece.
- The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team. The Epidemiological Characteristics of an Outbreak of 2019 Novel Coronavirus Diseases (COVID-19) — China, 2020. CCDCW 2, 113–122 (2020)
- CDC. Preliminary In-Season 2019-2020 Flu Burden Estimates. Centers for Disease Control and Prevention https://www.cdc.gov/flu/about/burden/preliminary-in-season-estimates.htm (2020)
- CDC. Estimated Influenza Illnesses, Medical visits, Hospitalizations, and Deaths in the United States — 2018–2019 influenza season. Centers for Disease Control and Prevention https://www.cdc.gov/flu/about/burden/2018-2019.html (2020)
- WHO-China Joint Mission. Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19). https://www.who.int/docs/default-source/coronaviruse/who-china-joint-mission-on-covid-19-final-report.pdf (2020)
- Arriola, C. et al. Influenza Vaccination Modifies Disease Severity Among Community-dwelling Adults Hospitalized With Influenza. Clin. Infect. Dis. 65, 1289–1297 (2017)
- CDC. Frequently Asked Questions and Answers: Coronavirus Disease 2019 (COVID-19) and Pregnancy. Centers for Disease Control and Prevention https://www.cdc.gov/coronavirus/2019-ncov/specific-groups/pregnancy-faq.html (2020)
- Cover image: “Novel Coronavirus SARS-CoV-2“; transmission electron micrograph of a SARS-CoV-2 virus particle, isolated from a patient. Image captured and color-enhanced at the NIAID Integrated Research Facility (IRF) in Fort Detrick, Maryland. Credit: NIAID. February 28, 2020.