32.8 million Americans have had confirmed positive tests for COVID-19 as of April 26[1], and the current US population is 328.2 million, so at a minimum 10% of the people in the US have gotten COVID. A maximum of 100% of the people in the US could be infected[2], so that implies an upper bound of 10x worse or 5M deaths averted by all interventions combined. US CDC estimates that about 3 - 3.5x as many people have contracted as have tested positive[3], so a more realistic upper bound is that COVID-19 would have been 3x worse without any intervention (assuming that 100% of people would have been infected without intervention).

So that gives you an upper bound. I'm not sure how to get a lower bound but it might be possible for someone with sufficient statistical know-how.

[1] https://www.statista.com/topics/6084/coronavirus-covid-19-in...

[2] I'm ignoring reinfections here since, while reinfections happen, they are quite rare relative to first-time infections everywhere I have seen

[3] https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/burd... -- note this is as of January 15, when there were 24.2 million confirmed cases, and CDC estimated that there were about 83.1 million total cases.

There are a bunch of stuff that make this hard to calculate. One is that cases are undercounted, while deaths are almost certainly not. Another is that deaths are correlated to available healthcare. 3x cases would mean that the standard of care goes down, which means more deaths. More deaths would probably also mean increased "protective" measures. Risk tolerance (and risk of infection) is elastic. Seeing 3x more deaths would certainly mean at least some people would take it more seriously, which reduces risky behavior and slows down cases, etc.

So you're right, if worst case everyone in the US gets infected, we're in for a lot more deaths, but aside from an apocalyptical R=100 type disease, not everyone is getting infected. The more infections there are, the more preventative measures are taken, etc.

It's a hard problem.

I'm pretty sure what we saw this year demonstrates that protective measures are far less elastic than we might assume.

This calculation does not take into account a collapse of the medical system. The problem with covid is the speed of its spread. We see this in India right now. There are not many cases per capita but their limited health care system could not handle this and people suffocate to death.

This is very wrong because as another comment mentioned, we are constantly on the verge of collapse of the health care capacity. I have friends who work in hospitals and the measures a government in Europe seems to take seems to always have as loose as possible measures while keeping the health care system at near maximum strain. Many people recover from covid in the hospital and even bigger problem is that if many more people are sick from covid, the people with other unrelated issues are lacking health care attention. One number I've heard is that we will see an increase in deaths from unrelated health problems which would usually get detected on time but since it's hard to get an appointment

Great answer, thank you

> Did we really avert 50M deaths by social distancing?

No, not even close.

> this implies that COVID deaths are also likely to have been reduced to ~1% from what they would have been without said measures

I don't think it implies that. There is nothing that says that the anti-COVID measures are about as effective at preventing spread of COVID as spread of flu.

> 500K deaths

Closer to 600k :(

This is among 32MM cases reported to the CDC, probably more like 80MM cases total.

Ignoring lots of factors, our ceiling for COVID is more like 4-10x what we've experienced, not 100X.

I feel like you misunderstand the meaning of "implies", which is to suggest without explicitly saying. I think it is reasonable to assume that social distancing measures would have a similar effect, and in the absence of other data, the best estimate might be to assume they are proportional.

Fortunately there are a couple of other answers in thread that answered this at a level of detail that I could understand.

> To naive little me, this implies that COVID deaths are also likely to have been reduced to ~1% from what they would have been without said measures.

No, unfortunately this is not true. While the flu season disappeared entirely, covid obviously did not -- it grew significantly. As it stands, the United States has seen just shy of 100,000 confirmed cases per million people; if that represented 1% of a "status quo" infection rate then you'd have to expect everyone in the US to have covid 10 times over.

Instead, you have to think about the logistic model of infectious disease growth. As a very rough back of the envelope calculation, Covid has an R0 of about 2.5. It will spread throughout the population until about 60% is immune/recovered (which roughly matches the critical vaccination level seen in Israel, providing a reasonable check on this number).

That implies that absent public health measures (i.e. where covid quickly spreads to a level that gives herd immunity), the United States would have seen about 200 million cases (330m0.6), or 600,000 cases per million.

What that implies about deaths* depends on your estimates of how comprehensive case identification has been. If you think that we've tested and identified the vast majority of cases, then the "worst case" would be about 6x the current rate of infection and a similar increase in the total death rate. If on the other hand you think we've vastly undercounted cases, then maybe the US has in fact come close to that herd immunity level already, and the public health restrictions only succeeded in "flattening the curve" without changing the cumulative outcome much.

Working at this from another angle, the CDC estimates (https://www.cdc.gov/coronavirus/2019-ncov/hcp/planning-scena..., case 5) give an infection/fatality ratio of about 0.25% when you apply their demographic estimates to the US population writ large. If you back-fit an estimated number of covid cases from the currently-observed 500,000 deaths, you end up remarkably close to 200m total cases -- but you can also quibble here by noting that seniors' homes suffered a disproportionate share of infections and deaths, especially early in the pandemic.

This model greatly depends on the rate of actual infection.

Assuming a perfectly oblivious population, for example, where there are absolutely no protective measures in place† and people just act 100% as usual††, then the infection rate would have skyrocketed in a very short time and the death rate would have been much, much higher because of the healthcare system completely collapsing. You'd have people dying from tetanus because nobody would have been available to give them shots. People dying from easily stoppable bleeding, etc. So in this extreme case I think the total death rate caused by the pandemic could have easily been in the low millions.

† Self distancing, masks, etc.

†† Which would be quite irrational and hasn't actually happened even in places which otherwise ignored the pandemic: a certain percentage of the population self-isolated due to the survival instinct.

The difference might be in flu deaths being confounded by C19 deaths in vulnerable populations. An example might be a person with a weakened immune system. In an alternate timeline without C19 they would have died of flu. In the real world they caught C19 first and died of that instead.

This makes a lot of sense

I would not claim that it accounts for all flu avoidance, but some nontrivial portion of missing flu deaths.

Additionally I would not be surprised if flu deaths remained lower for several years since C19 appeared to be more severe than flu in vulnerable populations. Future flu deaths were likely preempted by C19.

Paralleling the term herd immunity, this would be called culling the herd.

That extrapolation only works if the flu is equally as contagious as covid and also has identical methods of transmission (for example, it may be that covid transmits through the air more readily than the flu). But we know that covid is actually quite a bit more contagious, so you can't really make a linear extrapolation like that.

If it's more contagious, would you assume social distancing would have a greater or lesser effect? It's not clear to me either way.

Other answers on here have convinced me however that 50M is not a reasonable number.

All else being equal, mitigation efforts are going to have less effect on the more contagious disease. 6 feet social distance with a mask might be plenty to nearly eradicate the flu while it only somewhat reduces covid. Washing your hands frequently may have a very profound effect on the flu but very little effect on covid since it seems to transmit more through the air.