In this era of vast uncertainty, one thing that has remained constant is the innumerable questions. When is this going to end? Are children at risk to the virus? Is ordering fried chicken three nights in a row a terrible idea?*
And one really quite important query: is COVID-19 getting less lethal over time?
If you listen to the headlines, the answer is an emphatic yes. According to many news sources, recent evidence has shown that COVID-19 is now less lethal than it was earlier in the year, which may be down to better treatments, mutations in the virus, or whatever theory is most popular at the time of writing.
And don’t get me wrong: I also think that the answer to this question is almost certainly yes. Over time, with newer treatments and improved modalities of care, it’s pretty much a sure thing that COVID-19 will kill fewer of the people that it infects. However, there’s a huge caveat to all of this — we expect that the virus will get less lethal, but the timeline of that reduction is very much in the air. I would be perfectly comfortable saying that, in 10 years’ time, you are probably very substantially less likely to die from the disease. But today compared to March? Or April? That’s a much more challenging question to answer, despite the headlines.
So, how would we know if COVID-19 was becoming less dangerous for those infected?
The quickest and easiest way to examine death rates would be to look at the case-fatality rate (CFR) of COVID-19 over time. This is simply the ratio of deaths to the number of people testing positive to the disease, and if you look at the figures you immediately see a drop between March and now across the board.
The problem is that the CFR of COVID-19 isn’t really that informative. The denominator of the ratio is simply the number of confirmed cases of the disease, which means that testing rates largely define the CFR — if you test more people, there are more cases, and the CFR drops. Over time, testing has gotten better, so a falling CFR is to be expected.
A much better way of looking at whether mortality rates for COVID-19 are improving would be to calculate an infection-fatality rate (IFR) over time. Both of my studies into COVID-19 lethality are using IFRs rather than CFRs for precisely this reason, because it tells you the ratio of deaths to total infections, giving you the risk of dying if you get infected with the virus. Unfortunately, due to the nature of this calculation and the studies you need to estimate it, it’s really hard to know what the IFR has been over time, and so we can’t use this metric (yet).
However, there is yet another way of trying to figure out whether the death rate has been falling — looking at a specific group of patients over time. In this case, people who have been hospitalized or admitted to an intensive care unit with COVID-19, to see if they are less likely to die over time.
There have now been several studies that have done this, and all are relatively similar in nature. Basically, if you look at the group of people who are being admitted to hospital, or treated in an ICU, for COVID-19 over time, a smaller proportion of them are dying. In other words, compared to the first weeks of the pandemic, fewer people who were treated in these conditions died within a particular time period (usually about a month).
Now, that sounds like great news, but there’s a problem. You see, it’s really quite hard to compare the situation in March with now for any number of reasons. Hospitals are, generally, less overwhelmed, and thus the people that they are admitting are very different. Indeed, if you look closely at each of these studies, the average age of people with COVID-19 admitted to hospital/ICU drops quite a bit over the weeks, as do markers of disease severity. While we can correct for this to an extent, it’s very likely that people who are being admitted now are dissimilar in ways that can be quite challenging to capture compared to people admitted earlier in the year.
Yes, I’m talking about residual confounding, and it’s quite a big potential issue when we look at these numbers. We can correct for the things we know about, and measure, like age, but people who were admitted for COVID-19 in July are different to those going to hospital in March in ways that may be really hard to measure. Improvements in testing, for example, may have meant that people who previously would not have been hospitalized were, or were hospitalized earlier in their disease course and so less likely to die within 30 days. It’s just not easy to compare the situation in these disparate times, especially when you consider that the data is not consistent everywhere.
Even one of the studies cited above found that, while mortality overall decreased, at the start of March people were less likely to die, which got worse until mid-April, and then improved after that. This seems to indicate that the most likely explanation was changes in the denominator (i.e. who got admitted in this time).
What does this all mean? Well, firstly, it’s absolutely still probably right to say that the death rate from COVID-19 is decreasing over time and will continue to do so. You are almost certainly less likely to die from the disease today than you were in March, regardless of anything else.
That being said, it’s quite hard to know exactly how much your risk has decreased. While the news has been filled with massive reductions of 50–90% in mortality based on one study or another, the true figure is extremely problematic to capture. Hospital admission is, after all, not a concrete phenomenon, and exactly who gets admitted to which hospital changes over time and location. It’s not at all unlikely that much of the difference we’re seeing in hospital patients comes down to the people who have been admitted over time, rather than a reduction in the disease severity per se.
Without a constant denominator — ideally, a marker of the true proportion of people who have been infected in the population over time — it’s just not easy to say that the death rate has declined in any meaningful way.
Ultimately, I’d say that the news is still good. Getting sick with COVID-19 today is less likely to be a fatal experience today than it was when the coronavirus first emerged. The longer you delay getting ill, the better your chances are, and there’s every reason to believe that the death rate will drop enormously over the next few years.
However, it’s still worth being cautious about the very spectacular claims out there. Yes, the death rate is probably dropping, but whether it’s dropped quickly over the last 6 months, and how big a drop that represents, are still questions with open answers.
So is COVID-19 getting less lethal?
Probably, it’s hard to say how much less dangerous it might be.
*Note: the answers to these are, respectively, we don’t know, probably although clearly a lower risk than adults, and never.