Covid and other causes of death

Covid has killed about 220,000 people in the United States. If you measure by “excess deaths,” meaning deaths that have occurred this year in excess of what would have been statistically expected, and which are most likely under-counted Covid deaths, it is over 300,000.

Either way, that’s more Americans than died in the Vietnam War. On the other hand, it’s half or less than the number of people who die of heart disease every single year. Put one way, it seems like a whole lot; put another way - well, it’s still a lot, but maybe it’s not the apocalypse. Which is the better way of looking at it? How do we gain some intuition for just how serious a problem Covid is?

First, we can make the problem smaller and zoom in on just Wisconsin. The state just passed 2000 Covid deaths. How does that compare to other causes of death? Below is a table of deaths in Wisconsin in 2017, the most recent year for which there is a complete death report from the DHS.

Just looking at this table is enough to rebut two of the common arguments that Covid is no big deal. First, Covid is clearly worse than the flu - it has already killed twice as many people as flu and pneumonia did in 2017, and that’s before we have even gone through a winter.

Second, it seems extremely unlikely that suicides brought on by the social isolation of lockdown will outnumber Covid deaths. An increase in suicide may well be a serious problem; I have not looked into the data enough to know. But suicides would have to double compared to 2017 in order to equal the Covid deaths so far. My intuition is that, as difficult as this year has been, that large of an effect is not plausible.

A third argument, that Covid may be serious but that it’s not out of line with other risks that we routinely tolerate, has more force. Unintentional injury, for example, was responsible for nearly 4000 deaths in 2017, twice as many as Covid so far. Many of these accidents, like Covid, could probably be prevented through draconian lockdowns. If we are okay with 4000 accidental deaths a year, should we accept up to 4000 Covid deaths this year?

Maybe so? In fact, we will most likely have to - I do not think winter will end before we have seen 4000 deaths in this state. Very crudely: extrapolate the current 7-day average deaths per day (38), rounded down, to the end of the year. So 30 deaths/day times 60 days equals 1800 deaths; add it to the current 2000 and there’s a rough 4000.

The case for Covid prevention and action does not really rest on the number of deaths we have had so far, however, but on the number of deaths we may be able to prevent, compared to going back to life as usual. We cannot know this with much certainty, but we can estimate.

Assume a population-wide infection fatality rate (IFR, the percentage of infections that end in death) of 0.5%, and that 50% of the population would become infected before widespread acquired immunity, i.e. “herd immunity,” suppresses the virus. These are arguably middle-of-the-range values. With Wisconsin’s population of 5.8 million, that would result in about 15,000 total deaths.

The IFR and herd immunity threshold are not known with much precision, but we do know them within a range. IFR might be 0.3% or 1%; the herd immunity threshold might be 30% or 80%.¹ Using these numbers, a low estimate of expected deaths would be 5000, and a high estimate 45,000. Here is another table showing these estimates in comparison with yearly deaths from other causes.

The low estimate, again, is not so different from unintentional injuries. But even the medium estimate would be the single largest cause of death over a year’s time, eclipsing heart disease. The high estimate would be a disaster, nearly doubling the death count for the year.

Here is my conclusion from all of this. Covid is not smallpox or the Black Death; we are not looking at a third of the population dying. It is not even the Spanish Flu, where 0.65% of the US population died.²

But it is bad, and it is worth taking strong action to prevent. If our collective actions can prevent the medium estimate from occurring, and make the low estimate happen instead, it would save 10,000 lives. That would be like curing cancer for the year. There are certainly many questions as to the right actions and the right tradeoffs, but it is clear enough to me that the stakes are high.