COVID-19: The BIG Question
With all the noise going on about COVID-19 at present, there's one question that we would all like to know: am I going to die? Closely followed by: is anyone I know going to die? I'm not going to answer those questions today, but I am going to talk a little about what we do - and don't - know about mortality from COVID-19.
When coronavirus was first identified, it wasn't clear what the mortality was: two articles were published online on the 24 January 2020 by the medical journal, The Lancet:
One article described a family cluster of six people, four of whom had travelled to Wuhan but five of whom had developed a severe pneumonia and all with the same viral DNA detected, thus confirming person-to-person transmission.
The other article described 41 people who were admitted to hospital of whom six (15%) had died.
This second article was what's known as a case series: a select group who had required hospital admission; nothing was known about the general population or anyone else who may also have been infected. It was therefore not possible to provide an estimate of what proportion of people were dying, merely that people were dying and that they all had the same, new virus underlying their condition. The virus was initially called "2019-nCoV", but has been subsequently renamed to "2019 severe respiratory distress syndrome coronavirus 2" (SARS-Cov-2) which produces the illness now widely known as COVID-19 - itself short for COrona VIrus Disease 2019.
Further papers appeared. On the 19 February 2020, the BMJ published an article describing another case series, this time involving 62 patients admitted to hospital in a different province of China, Zhejiang, and none of whom died. However, it wasn't until the 24th of February that we were really able to get better information: the Chinese had made an enormous effort in trying to track down all the people they could possibly find who might have been infected and had been incredibly successful at it. As of 11 February 2020 - the latest date for which data were available at that point - they'd managed to identify a total of 72,314 people, and evidence suggested that the spread of SARS-Cov-2 had started at some point back in early December.
As there had been such extensive tracing of potentially infected people, much better data were available to examine mortality rates. But due to the sheer number of people who were infected, there were also problems with having enough laboratory tests everywhere, meaning that there was only laboratory evidence for 44,672 patients. These data were published in the Journal of the American Medical Association and showed that there were 1,023 deaths among the patients with a confirmed laboratory diagnosis - thus giving an overall mortality rate (also known as a case fatality rate or CFR) of 2.3%.
Now, you've probably heard people talk about an overall mortality rate of 1%. That's quite different from 2.3% - which one is correct? Truthfully, no one knows: there are many problems with the 2.3% figure, and "best estimates" seem to congregate around one percent. There are three big issues which affect the estimate:
- the age distribution of the population,
- the choice of denominator, and
- whether treatment is needed.
Age distribution
Most importantly, the 2.3% figure relates to the group of patients who had confirmed disease in China. But these patients come from a different population than wherever you live (well, unless you're living in China). And different populations have different numbers of people at each age. Consider the following example. When I was doing my "care of the elderly" attachment at medical school, most hospitals considered elderly people to be over 75 years old and those were the patients we helped look after on our attachments. But for the students who went to Torquay - well, Torquay is well known for being a nice place by the sea on the south coast of England where people go to retire and there were lots of older people there. So the students who went to Torquay for their care of the elderly attachment found that the patients they were involved with were restricted to those over 90 years old. Similarly, there are more elderly people in Italy - and the mortality rates that were being reported initially in Italy from COVID-19 were higher than those that were initially reported from China.
Denominator
The next big issue is the denominator. A mortality rate is simply the number of people who die divided by the number of people who have the illness for whatever the given disease is. Take a very simplified heart attack example: if someone has a heart attack, they either die immediately (in which case the person who signs the death certificate normally writes that they died from a heart attack) or they survive and go to hospital and get treated. There's not usually too much of a delay between having the heart attack and dying so it's easy to see what the mortality rate is from a heart attack (well, OK, it's more complex than that - but this is quite a simplified example).
With an infection, there's often a time delay. You get the infection, then you wait a while (for COVID-19, the incubation period is around 5 days as I described previously), then you get sick and you may get admitted to hospital. And then, you either get sicker and die, or you get sicker, stay sick for a while, and then start to recover - and eventually get better.
So, if you're calculating the mortality rate using all the people who currently have the illness, you're using the wrong number. For COVID-19, you really need to be using the number of people who got infected at the same time as the people who are dying - in other words, from 10 or 14 days ago, or maybe even longer. This was done for COVID-19 using a 14 day lag period and gave a much higher rate of death - 5.6% compared to 3.6% for China (they used a different range from that I've discussed above, hence why the numbers are different).
The other important factor relevant to the denominator is of course knowing how many people actually have the infection. Given that testing has been problematic in many countries, we don't really have any good data on this, and thus are reliant on mathematical models.
Treatment
The final important consideration is treatment. Most people are not going to suffer too badly - indeed, some are not going to have any symptoms at all. But others will require hospital admission. Obviously doctors are going to give the best treatment they can, and for a small proportion of patients (it's currently estimated around 5% of everyone who is infected - but of course this figure might be wrong as we don't know how many people are being infected) that's going to mean intensive care, specifically, ventilation (when a machine is required to support breathing). And this is what the current crisis is all about: there are not enough ventilators to help all the people who need them, and if you don't have a ventilator when you need one, you die.
What is the mortality rate?
So, what is the mortality rate? The answer is, I do not know - but nor does anyone else. Most people are saying around 1%. Some, even respectable people like John P.A. Ioannidis who is professor of medicine, of epidemiology and population health, biomedical data science, and statistics (I don't know how he manages to do all that, he must have no life!) at Stanford University, are saying that it's much lower . I don't agree with that analysis, although I do agree with his point that we do not know what the mortality rate is and that better implementation of testing is required. Others, as i've described, are suggesting the mortality rate might be higher. But there is one point we can all agree on: it doesn't matter what the overall mortality rate is, if the hospitals are overwhelmed and we run out of ventilators, it's going to get worse. It is therefore imperative that we do anything we can to slow down the spread of this virus through the population; we must enforce social distancing as much as possible - our only hope is to buy time to increase our knowledge about the virus and what we can do to combat it.
