We May Be Underestimating the Coronavirus Death Toll
bloomberg.com
bloomberg.com
The net result is most likely an underestimate (as the article describes -- the situation in Northern Italy pretty much assures this must be the case, at least there)
Examples of overestimate: So, the first COVID "caused" death in Pennsylvania was a fellow in Allentown who fell and hit his head. He died of head trauma. He had been diagnosed with COVID. He was counted as the first COVID death in the state. There are certainly hundreds of cases like this, where someone died with COVID but clearly not from COVID.
On the other hand, the underestimate: Everywhere that is having COVID problems is also having unseasonably high "pneumonia and influenza" death rates. (In the U.S., for example, we're running about 0.5 above seasonal baseline (that is to say, about 1 in 200 deaths "too many" are being caused by "pneumonia or influenza") as of 2 weeks ago.
In Northern Italy, as the article points out, all-cause mortality suggests that almost half of the COVID deaths are going recorded as non-COVID. This will surely include COVID-influenced non-disease deaths (like someone who dies of a car crash or heart attack that would have survived, except that the healthcare system is swamped and they got no treatment -- apparently the ambulance system is sufficiently overstretched that if you're over 85, the ambulance just won't come for you at all) Those aren't COVID deaths, and they aren't counted -- but they're deaths that wouldn't have happened otherwise.
It's all a tangled mess.
We're also definitely terribly underestimating the number of COVID cases. It is certainly the fact that, in hard-hit areas, we're detecting less than 1 case in 3. Quite possibly worse even than 1 in 10. (but we won't know for sure for a long time, if we ever do).
The WSJ article covering the same ground cited experts estimating the true number of cases as anywhere from about 400,000 to 6,000,000.