I had a conversation with someone earlier this week about how we should respond to COVID-19. During the conversation she said to me, “Bill, you just don’t understand, people are dying.” Of course, I do understand that people are dying because that happens every day. In fact, about 165,000 people die each day somewhere in the world. That is about 65 million per year.
My friend’s reaction was based on the heightened sense of empathy humans have evolved compared to other species. It became part of our survival strategy as it spurred greater cooperation and aid between humans. But empathy evolved while we were still tribal, so it is based on personal connections. As a result, we are moved by stories of personal tragedy but less so by impersonal statistics about tragedy.
For example, UNICEF estimates that about one million children die from malaria each year with virtually no global attention or awareness. If the media went to Africa and brought individual stories about those children into our living rooms and we had a counter on our television screens showing how fast the deaths were stacking up each day, we would have an entirely different reaction.
However, to effectively manage this epidemic we must act based on reason and analysis, not emotion. To do that we must stay focused on the hard data. So, I thought this might be a useful time to do a primer on the statistics for fatalities in the U.S.
The National Center for Health Statistics is part of the CDC and maintains the official vital statistics for the U.S. Since 2013, the NCHS has published a dataset tracking all fatalities on a weekly basis as part of its flu surveillance system.
Over that time, the U.S. has averaged about 2.7 million fatalities per year with the number slightly trending up with population increases.
There is a distinct seasonality to fatalities, which increases during the flu season every year.
The latest data the NCHS has completed for this year is for the week ending April 11 (Week 15). Here is how 2020 compares to the previous years through Week 15.
Ironically, through April 11, U.S. fatalities are down from last year and are almost exactly equal to the average for the previous six years.1 However, I would discourage anyone from jumping to any conclusions from this data. As of April 11, only the first week of New York’s elevated COVID fatalities had hit the numbers and revisions keep coming in. So, the 2020 total will grow over the next few weeks.2
Nonetheless, the fact that total fatalities have not increased seems counter-intuitive. As of April 11 almost 17,000 people died from COVID. Why aren’t the total fatalities for the year up by that amount instead of being down by 17,000 deaths?
There are, at least, three possibilities. First, the numbers will be updated notwithstanding that NCHS shows they are complete. I have seen them do that in recent weeks. Second, fatalities from other causes are probably down. Automobile accidents come to mind. But also, it may be because some COVID deaths are not incremental. That is, some of the people dying with COVID would probably have died even if they had not been infected with the virus.
Of course, this data will be endlessly interpreted. Some will see in it as evidence that the threat from COVID has been exaggerated. Others argue that it is evidence that the containment strategies are working. But it will take a lot more detail about the data to support any definitive conclusions. And there will probably be some questions we will never be able to answer.
Nonetheless, I think it is important continue to monitor this data for purpose of perspective. Because my friend was right, people are dying. But, at least so far, not in numbers that are much different from previous years.
I have downloaded the NCHS data to an Excel spreadsheet here and will periodically update it. If anyone sees any problems with the spreadsheet, let me know.
1 The NCHS dataset also includes the number of fatalities which are attributed to the flu and to pneumonia. There is a sharp increase in pneumonia fatalities in Weeks 13-15, which must be COVID-19 beginning to show up in the numbers.
2 To get some idea how much they might grow, the IHME model is now predicting about 68,000 deaths by August 4. As of April 11, there were only about 11,000 COVID deaths. So, that would leave another 57,000 to be spread over May-August, say fifteen weeks, which would add on average about 3,000 additional fatalities per week. But even at that, it appears unlikely the 2020 fatalities will reach the previous weekly highs posted during the 2018 flu season when fatalities exceeded 60,000 per week for seven weeks.