How bad is it?
A question that has been at the forefront of my mind since I first heard of an outbreak of a new coronavirus in China has been, “how deadly is this virus?” I knew that the other novel coronaviruses in recent history were both quite deadly — SARS with an overall case fatality rate (CFR) of around 11% and MERS with a CFR of about 34%. So first off, what is a case fatality rate? The case fatality rate is the number of deaths divided by the total number of infections. So if 100 people get an infection and five of them die of it the CFR is 5%. The CFR is one measure of the severity of a disease. Of course it’s important to keep in mind that there are other ways of measuring the badness of something like COVID-19 — for instance the number of people requiring care in an ICU or a ventilator, long-term debility after lengthy ICU admissions, how contagious it is and, of course, the economic impacts. It’s also important to keep in mind that while the CFR sounds simple it can actually be quite complicated. With a disease like COVID-19 where many people with severe illness remain on a ventilator for weeks before dying, the CFR could be falsely low early in an outbreak when many people who will eventually die are still alive in the ICU. Another complicating factor is that the CFR for COVID-19 varies widely by age group and comorbidities, so in a country with an older, less healthy population the CFR likely will be much higher than in a country with a younger, healthier population. This extreme variability in CFR between age and risk group has caused me to question whether calculating an overall CFR is even worthwhile at all. If the overall CFR is 1% what does that really mean to a 10-year-old for whom the CFR may be 0.01% or to a diabetic 80-year-old for whom the CFR is 12%?
This question of “how bad is this?” has also loomed large in the media in recent weeks. On a daily basis I’ve seen editorials suggesting that perhaps the measures we are currently living under such as closing of restaurants, restriction of social gatherings and shelter-in-place lockdowns may be an overreaction. Perhaps this virus really isn’t as bad as the media has “hyped it up to be.” I’ve read commentary suggesting that perhaps COVID-19 may even be less deadly than seasonal influenza, which has had a CFR of 0.1–0.17% in the past five flu seasons. Elsewhere in the media a CFR of 3–4% has been reported — a figure consistent with reports from the WHO and easily arrived at if you divide the number of deaths reported worldwide by the total number of cases that have been reported. Interestingly and unsurprisingly I’ve also noticed that analysis of the pandemic has become politicized with outlets on the right tending to lean toward the current measures being an overreaction and those on the left generally criticizing the government for not doing enough and also not suggesting that current measures are an overreaction.
Another interesting thing about the CFR of COVID-19 is the wide variation between different countries. As of this writing I am calculating the Italian CFR at 12%, the German CFR at 1.2%, the U.S. CFR at 2.4% and the Chinese CFR at 4%. Iceland currently has a CFR of 0.15%. Possible explanations for such a wide range of severity include such factors as the age and health of a population, which cohorts are being infected in a population (if it gets into the nursing homes the CFR will be much higher than if it’s mainly circulating in college students), the quality of healthcare infrastructure, the explosiveness of an outbreak leading to overwhelming of the healthcare system, availability of potentially effective drugs among many other factors. The possibility of different strains of the virus with different degrees of severity circulating has been raised but I don’t believe there is any good evidence of this. Likely most importantly in determining the CFR in a population is the number tests that have been done, the proportion of the population that has been tested. It is clear that many cases of COVID-19 are mildly symptomatic and many are likely asymptomatic. The vast majority of these asymptomatic or minimally symptomatic people are not tested and therefore not included in the count of total infections. When a country is mainly testing those who are sick enough to be admitted to a hospital this will falsely elevate the CFR. Therefore, it can be assumed that the true CFR is lower than what is currently being reported — how much lower is hard to say. Conversely, it’s also been reported that many people have died in northern Italy without ever having gone to a hospital or being tested — this could theoretically falsely decrease the CFR although I think there are far more minimally or asymptomatic people not being tested than there are deaths being under-reported. My guess about the overall CFR last week would have been around 0.5%, based on additional information I am now thinking it might be closer to 1%, but time will tell.
To arrive at the true CFR we would need to know the total number of cases in a population and the total number of deaths attributable to COVID-19 after everyone had either recovered or died. Obviously we don’t have that data. Eventually, serologic surveys (documenting the percentage of population with antibodies showing they were exposed) will likely give us a good idea of the true overall CFR. A major factor in how much of an overestimation the currently-calculated CFR is are the number of asymptomatic cases. Based on my own experience of having tested asymptomatic people who ended up being positive for COVID-19, I know that asymptomatic cases are probably not uncommon. A study of passengers from the Diamond Princess cruise ship suggested that as many as 18% of those infected may be asymptomatic. Conversely, in a Chinese study, asymptomatic close contacts of proven COVID-19 cases were tested. Fifty-five individuals who tested positive were admitted to a hospital for observation. All 55 ended up developing symptoms which would suggest that the asymptomatic rate may actually be much lower than 18%. There is also good evidence that children and teenagers are much more likely to be asymptomatic or mildly symptomatic. So it is hard to say what percentage is truly asymptomatic but my guess is that among adults probably at least 5–10% are asymptomatic.
As mentioned above, there are other ways of looking at the “badness” of COVID-19 other than the CFR. The one I will touch on is its transmissibility — how contagious is it? There are a number of ways of looking at this but for me the most intuitive is the R0, pronounced R-naught. The R0 is the reproductive number — the average number of people who will catch the disease from a single infected person. For influenza the R0 is 1.3 so on average a single person with influenza will infect an additional 1.3 people. This can be hard to conceptualize since a human can’t be broken into a fraction but it means that some people with the flu will infect 2 others, many will only infect one other person and some will not infect anyone else. For COVID-19 the R0 is estimated to be about 2.2 with a range of 2–3. So COVID-19 is significantly more contagious than influenza (likely at least twice as contagious). It is important to note that social practices drastically affect how contagious a virus is — if people are washing hands and social distancing the R0 can be markedly decreased to less than 1 which is when the number of new cases begins to decrease — as has been seen in China, South Korea and some other countries. From personal experience I’ve seen how contagious this virus is, having diagnosed someone with COVID-19 who spent one night together in a room (different bed) with the person this patient caught it from. Another problem with COVID-19 is that there is likely a lot of transmission of the virus that occurs before the first symptoms develop — one study suggesting that someone could be contagious 1–7 days prior to symptom onset.
The U.S. government is projecting 100,000 to 240,000 deaths from COVID-19 if current social distancing measures are maintained. I have no reason to doubt these numbers. I’ve seen memes floating around social media suggesting that COVID-19 is comparable to an influenza season. The CDC estimates that between 12,000 to 61,000 deaths have occurred annually due to seasonal influenza since 2010. The H1N1 pandemic in 2009 resulted in an estimated 12,469 deaths. Based on those numbers alone, COVID-19 is significantly worse than a flu season. It may not be scientific but when I read about hospitals in multiple cities being overwhelmed, ventilators being rationed and talk with colleagues about the dire conditions they are facing I know that what we are dealing with something different. I don’t think the world has seen anything like this since 1918–1919 with the Spanish flu.
So we can’t sugarcoat things. We are facing one of the greatest challenges that has been faced in recent generations. But there is a light at the end of the tunnel. Other countries have been able to get a handle on things and we can do it too. It remains unclear what the end-game looks like — will an effective vaccine be developed? Will effective antiviral agents be developed or discovered? The effectiveness of agents we are currently using — remdesivir, hydroxychloroquine, Kaletra, tocilizumab and others remains to be seen. Will COVID-19 become a seasonal virus like influenza or other common cold viruses? I hope not. Based on the above analysis, I do think that current social distancing measures are sensible and will likely need to be continued for an extended period.
