COVID-19's Excess Deaths

I’ve been seeing on facebook an argument that COVID-19 is overblown because deaths attributable to other causes are being falsely called COVID deaths. That is an important hypothesis to test.

The way to test it is by looking at the total deaths, not just those attributable to COVID. Look what is happening in NY. The monthly death rate is FOUR TIMES higher than it usually is:

We see the same pattern in other countries.

If it isn’t COVID-19 killing everyone, something else is, and we would be in in serious trouble because scientists are misdirecting all this effort at the wrong pathogen.

These numbers undermine the notion that many people who have died from the virus may soon have died anyway. In Paris, more than twice the usual number of people have died each day, far more than the peak of a bad flu season. In New York City, the number is now four times the normal amount.

The key point here, also, is that many of these deaths (probably more than 36,000) are NOT in the COVID death totals.

In the last month, far more people died in these countries than in previous years, The New York Times found. The totals include deaths from Covid-19 as well as those from other causes, likely including people who could not be treated as hospitals became overwhelmed.

The actually death toll is going to be painfully high. It is hard to express the degree of disruption this is causing to healthcare across the country. Anyone who needs a hospital to stay alive for any reason is at serious risk.


Apparently a big problem is that people are now so afraid of going to hospital (they don’t want to catch the virus) that they’re putting off going to hospital until they’re extremely sick. For example, people suffering who have a heart attack are waiting as long as days before seeking medical attention, resulting in much more extensive damage to their hearts than if they’d immediately gone to hospital when the first symptoms appeared.

This is probably a real contributor to some of these excess deaths, and I’ve seen doctors in the UK imploring people to come to hospital if they feel they need medical attention, no different to how they would have done before the pandemic.


This is a good way of tracking the net effect of covid, but it should be noted that here are some caveats. This would include all kinds of things: it would obviously include the people dying from other medical causes because they didn’t want to go to the hospital, but also people who killed themselves due to lockdown-induced suicide, those whose lives were saved because they didn’t get into a car accident the would have otherwise been in, and those whose health conditions have been exacerbated by the loss of their oil/gas related jobs.

I wouldn’t attribute them all directly to covid, but it’s fair to attribute a great deal of them to covid and all of its many, many secondary effects.


Excess Deaths are a good measure once all the data are in, IMO. It’s not particularly useful for real-time monitoring of the current situation.


Hi @swamidass,

I quite agree with you that the official figure of 58,000 deaths is probably off by tens of thousands, although I’m not sure that all the excess deaths are due to the new coronavirus. I suppose some fatalities could be due to people with other pre-existing conditions missing out on medical treatment, because they’re not able to get a hospital bed.

In other news, there’s also talk of Oxford University having a vaccine ready by September, but I don’t want to get my hopes up. More here for interested readers.

Meanwhile, here’s an article that may be of interest to some: The Forgotten Hong Kong Flu Pandemic of 1968 Has Lessons for Today. More here.

And for those who are wondering what’s going to happen to American capitalism, the forecast for the next four years doesn’t look good, if Scott Minerd, global chief investment officer at Guggenheim Partners, turns out to be correct. But who knows?


Vaccines are already ready. The issue is testing them. What are the estimate of when testing that vaccine will be completed?

Edit: reading the article, that vaccine is more than 18 months away from deployment. Other vaccines are farther along.


So that is what I’ve heard a lot of, that it’s the testing that takes time.

Are there any good ways of speeding up the testing? Are there unavoidable “wait times” ? Are there particular bottlenecks?

Yes, there are. The main biological limiting factors are the time it takes to generate an adaptive immune response and the time it takes to get exposed naturally.

As the name implies, the adaptive immune response involves a fairly Darwinian process of mutation and selection to get high affinity binding to the target (in this case, the spike protein of the virus). We can use adjuvants in the vaccine to try to make sure that this process is started as quickly as possible, and our bodies themselves speed up the process by increasing the mutation rate in the cells going through the process, but ultimately there are still limiting factors like the speed of cell division. The end result is a wait time on the order of a week or two between vaccine delivery and being able to see a robust immune response (or longer, if multiple doses are needed).

The other wait time is waiting for patients who have been given the vaccine to have enough time to get exposed so that that the efficacy of the vaccine can be tested. Perversely, this means that all of our other control measures–staying at home, wearing masks, etc.–actually work against faster vaccine development. The fewer infected people there are in the population, the longer you have to wait for your trial participants to have had opportunities to get exposed. There are statistical techniques that can be used to help spot the effects of the vaccine as quickly as possible, but they only get you so far if there are no effects to spot because no group has had enough exposure to get sick.

Hypothetically, we can speed up that process by directly exposing participants. We do this in animal trials. And we do this with human subjects for some vaccines. But the consensus so far on this virus seems to be that the risks to patients are too great and too unknown and so it would be unethical to knowingly expose humans.


Direct challenge studies, where we intentionally expose study participants, is the most likely way accelerate vaccine studies. Needless to say, this is ethically fraught. However, in this case it is possible we will see direct challenge studies, as the ethical calculus has also changed:

1 Like

Great info, thanks!

Do they also worry about potential side effects/bad reactions with the vaccine (I know next to nothing about vaccines)? Even a rare problem could cause major issues if millions of people get the vaccine. I don’t know if each vaccine needs it’s own “prep” of solutions/delivery or if that’s all been worked out before. I was just wondering if that also goes into the testing time.

1 Like

Yes, direct challenge studies have not been completely ruled out, and the case for them is public and nontrivial. I should have been more thorough.

1 Like

Yes. The first phase of human trials is focused on safety and identifying side effects and adverse reactions simply from receiving the vaccine. This phase is conducted with a small number of participants to limit the magnitude of problems of there are some, with the tradeoff that rarer events may not be detected.

The second phase of human trials also looks for possible adverse events from exposure to the pathogen following vaccination. In some cases, vaccines can actually exacerbate illness rather than preventing it. Dengue fever is a prime example. There are four strains of the Dengue fever virus, and infection with a second strain after recovering from a first strain can result in severe illness because of interactions between the immune response and the infection. Thus a Dengue vaccine needs to protect against all four strains equally well, and unfortunately a candidate vaccine did not elicit a sufficiently protective response to one strain, leading to worse outcomes after infection with that strain.

If a vaccine candidate passes those first two phases, it then moves on to the third phase, which is a larger scale trial to measure efficacy in a broader population. Safety issues may still be detected at this stage because of the greater number of participants involved, but the goal is to identify them earlier in the smaller scale studies.

I should add that this three phase approach is the US model. Other countries have different systems, but are still looking to answer the same questions about safety and efficacy.


Yes, and @AndyWalsh is perfectly correct. I serve on a Data & Safety Monitoring Board for cancer trials, and this is a big part of what we monitor.

Another slow-down in the trials process is the need to enroll subjects over a period of time. It’s desirable to observe side-effects in the first-enrolled subjects in time to halt the trial before the entire group is put at risk.


Here’s a really great article from NY Times that analyzes excess deaths in the last few months. I think this utterly destroys the argument that “it’s just a bad flu season”, “even normal deaths are being attributed to COVID!”, and so on.

Note that New York experienced 5.8 times the regular amount of deaths during the month of April, which is normally only possible in famine and war.