As you know, we’ve been focusing heavily on the ‘excess mortality’ question in the COVID19 pandemic across the world. Excess mortality is the number of deaths in a given region over a particular period of time which is in excess of the average number of deaths in previous years. We’ve looked closely at analyses from Italy and Spain which show dramatic discrepancies between the reported number of COVID19 fatalities and actual amount of excess mortality during the periods in question. In many cases, when this full excess mortality is calculated the number is two, three or even four times higher than the official COVID19 death toll.
What remains unknown in these other cases is how many of these additional deaths were ‘hidden’ COVID19 fatalities versus people who died as a result of the overall crisis but not the disease itself. This can range from an overwhelmed hospital system which creates a degraded level of care, to stress imposed by the totality of the crisis to people who do not seek medical attention for health crises they could have survived.
New York City has taken a different approach with the new numbers it released today. This included the number official COVID19 fatalities (i.e., lab confirmed test), those who are presumed to have died from COVID19 (symptoms and medical history but no lab confirmed test) and then total number of people who died in the city during the period in question between March 11th and April 13th.
It is that last number which we can compare to historical averages to establish an approximate excess morality figure. (A brief note on method. In this discussion I will often refer to an exact number for excess mortality. But these are derived by comparison to historical averages. So these are really approximations and estimates rather than concrete numbers.)
Let’s assume for the sake of conversation that the New York City clinicians have done a fairly good job catching the COVID19 cases that lacked a lab confirmed test. This allows us to distinguish between the hidden COVID19 mortality and deaths caused by the knock-on effects of the crisis rather than directly by the disease itself.
If we simply look at the official death toll as of yesterday and total excess morality, the number is roughly double: 6,589 deaths versus roughly 13,000. As noted, about half of the additional toll are presumed diagnoses and the rest are non-COVID19 cases or ones in which COVID19 was not obviously suspected. (See Josh Kovensky’s computations here.)
I was interested to compare this to numbers from Spain and Italy.
One comparison from the autonomous community of Castile and Leon in Spain showed a differential of 112 lab tested COVID19 deaths and a total excess mortality of 385 deaths over the period. So the total excess mortality was more than three times the number of official COVID19 deaths.
Another comparable analysis from Madrid found a total excess mortality of 6,613 deaths in the second half of March, of which 3,439 could be attributed to official COVID19 diagnoses. So 3,174 more people who died either from COVID19 or from the knock-on effects of the crisis. In this case, it’s a bit less than twice the number, which is in the ballpark of the statistics released today by New York City.
The other comparable analysis came from Nembro, the hardest hit town in Italy. That produced a total excess mortality four times the official number of COVID19 deaths. A look at nearby downs showed even high disparities.
The upshot of the comparison is clear. New York City had substantially less excess mortality not captured in the official death toll than two of these three other examples. One possible pattern is that New York’s numbers seem broadly comparable to Madrid’s. Madrid is substantially smaller than New York. But these are still both major global cities. I’m no expert on either but my impression is that Castile and Leon in Spain and the province of Bergamo in Italy are both much more rural or at least more sparsely populated. It seems worth considering – and would not be surprising – whether major cities with their dense public health and hospital infrastructures may be better able to capture the true scale of the epidemic than more rural or at least less urban areas. We’ll need to see comparable analyses from different countries and regions to confirm a pattern.
(Note that in these more sparsely populated areas the absolute numbers were much smaller. So the multiples or percentages may tell us less.)
Looking at data from New York, Italy and Spain, what we can say now is that we have the first indications of a baseline which suggests that the extent of excess mortality during COVID19 epidemics is often at least twice the scale accounted for in the initial, official COVID19 death tolls.
