There has been a lot of anecdotal evidence of inflating the fatality rate of the coronavirus. Interestingly enough, there is a way to keep everyone honest. In its section on influenza tracking, the CDC breaks down influenza and other respiratory infection fatalities. The data includes not only deaths from influenza et.al., but total deaths going back 5 years. Any trend in coronavirus deaths should show up in the weekly fatality count. And it does.
To put the following charts together, I pulled in three datasets: the CDC weekly influenza data for the last 5 years (available here), the CDC coronavirus data (available here), and the NY Times github data (available here).
This first chart is the total weekly fatalities from the CDC influenza section. I also added a 5-year average and a 5-year max. It is easy to see the spike in fatalities from the current epidemic. It is also interesting to note there was an epidemic two years ago and no-one really cared. One note on the graph: I excluded plague months from the 5-year-average and 5-year-max trendlines.
From this data, we can create a coronavirus fatality trend by taking the area between the curves. By taking the difference of the Red Line (current fatalities) and either of the black lines, we can calculate how many excessive deaths have occurred and compare it to published coronavirus data. Note, there is a lag in the CDC data, so we can only track up to about one month ago. So doing this and comparing the ‘area under the curve’ with the CDC and the NY Times data produces this chart:
And, what we have is remarkable agreement between reported coronavirus fatalities and the ‘area between the curves’ using the max curve (not the 5-year-average). So not only do the numbers match, but unless there is a massive spike in suicides or homicides that is not reported, the numbers appear a little conservative. So, despite the anecdotal evidence, the overall fatality counts for the US appear to be solid, at least for one month ago. The CDC updates their data once a week, so I’ll be sure to check back and see if everyone has been staying mostly honest.