Recently, a controversy arose when the CDC released its weekly statistical analysis of COVID fatalities that showed COVID was the sole cause of death in only about 6% of all fatalities.  Several social media outlets suggested the CDC has “quietly revised” its estimates regarding how many people had actually died from COVID.  The posts prompted a good deal of pushback such as this post by Politifact that gave the claim its “Pants on Fire” rating.

The news that over 90% of the COVID fatalities had comorbidities was not new.  Those of us who have been closely following numbers have known this since the beginning of the epidemic.  I have written about it several times and noted it in the City of Houston fatality data.  However, to completely discount the significance of the high incidence of comorbidities is equally misleading.

Any attempt to count the number of fatalities that are caused by COVID has two principal problems.  The first is that COVID symptoms are very similar to other respiratory diseases, like the flu, which account for the passing of many elders.  So, there is no question that some people died with COVID without being diagnosed, especially in the early going.

The second problem is that determining the cause of someone’s death is frequently as much of a metaphysical question as a medical one.   To illustrate, let’s suppose that a person has a serious heart condition and gets COVID and that the person would not have died from either, but did from the combination.  Should we say that person died because of COVID or from heart disease?  Suppose the same person got the flu instead of COVID.  Or a staph infection?  What about a person who was morbidly obese?

But debating about the actual cause of death is a bit like debating how many angels can dance on a pinhead. What we really want to know is what has the death toll from COVID been so far and what can we reasonably expect in the future.

The CDC attempts to get around these problems with what they call an “Excess Death” analysis.  They compare the fatalities for each week to previous years to see if there is a substantial variance.  The CDC uses death certificates from the states to perform this analysis.  Because of the lag time in getting the death certificates from the states, the analysis is only valid if you look back a couple of months or so.

What the CDC’s tabulation shows is that during April, fatalities were running about 40% above normal.  That was an unprecedented increase and was largely driven by the extraordinarily high fatality rate in New York.  After April, the excess deaths began to decline and have settled in over the summer at around 15%.

The chart below is the CDC’s current visualization of the excess death during this period. You can see the difference between the COVID outbreak and 2018 flu season, which was a particularly bad year. Any blue bar above the orange line indicates that fatalities are running higher than expected. Note that the chart goes through September 1. The CDC would not have received all the death certificates from the states through that date, so they project the number of likely deaths based on what they do have. That number is revised as the certificates are received. As a result, the farther left you go on the chart the more the entries will reflect actual as opposed to estimated values. You can access the CDC interactive chart by clicking on the image below.

There is one other issue that will ultimately affect the accounting for COVID’s toll.  It is a phenomenon epidemiologists call “mortality displacement.”  This is when a disease or event accelerates fatalities that would have occurred in the near future regardless of the disease or event.  For example, suppose that an elderly person near the end of their life died from COVID over the summer but would have died from the flu this winter.  But until we get well past the epidemic and can look back on a longer time frame, we will not know how much of an effect this type of displacement has had.

Just to provide some context, slightly under 3 million American die each year.  So, every 1% increase in fatalities equates to about 30,000 additional fatalities.

On a year-to-date basis, fatalities this year are currently running about 12% higher than the 2017-2019 average.  If that holds, there would be about 360,000 more fatalities this year.  BTW, that is about 50,000 below the current estimate for widely followed IHME model.  If the current rate of fatalities holds through the end of the year, the total would be about 280,000.

My guess is that the total fatalities for 2020 will come in quite a bit lower than either of these for a couple of reasons.  First, as I shared in this post the lethality of the disease is declining.  Secondly, I am convinced that naturally acquired immunity is steadily growing (as I suggested in this post) and that will dampen any surges in infections between now and the end of the year, even with the easing of suppression measures.  As nearly as I can tell most of the models are not taking the declining lethality or the growing natural immunity into account.

I would peg the likely increase in fatalities for this year in the 200,000-250,000 range or about 7-9% above recent years.  I think it is also reasonable to expect that we may see slightly fewer fatalities in 2021 than would otherwise be expected because of the displacement factor.  But, of course, only time will tell.