Studies Point to Big Drop in Covid-19 Death Rates | Hacker News Reader

Studies Point to Big Drop in Covid-19 Death Rates

npr.org

24 pointsAndrewBissell27 comments

I am not surprised by this one bit. I've been looking at these charts daily, and it's a bit silly to need confirmation from a study: https://www.worldometers.info/coronavirus/country/us/

Agreed, however the media has been singularly focused on new cases as a measure of the pandemic's severity. To me the news here is as much "NPR reports on the falling COVID death rate."

Even here they seem insistent that the lower death rate is attributable to "flattening the curve" even though there's almost no correlation at this point.

What are you referring to? The bulk of the article is about hospitals having more experience and standardizing protocols. Then they mention masks and social distancing as possibly reducing the viral load. They never mention "flattening the curve", but one of the last things they mention is keeping hospitals under capacity.

Yes, and it's still at 4%, but, true, it fell down from over 20% in the early days, but the sample is big enough to know that it's not less than 1%.

If low-risk groups engaged in more risky behavior and no other factor changed, you'd expect death rates of hospitalized patients to drop because a larger share of the hospitalized would be from groups at lower risk of death if hospitalized with COVID.

Also, if hospitals learned more about interventions (like that ventilation wasn't appropriate as early or in as many cases as initially thought), and nothing else changed, same result.

Or if at-risk populations became more cautious, and/or less exposed due to policy (like better approaches to nursing homes), same result.

So, this is not even a tiny bit surprising. Nor does it indicate any reduced need for control measures.

The article says the analysis controlled for age. Survival has improved in every age range.

> The article says the analysis controlled for age

Even if age was the only risk factor (in which case controlling for age would negate the first and third points), the second point would cover that. But age isn't the only risk factor, so all three points are still relevant.

I don't think your second point is up for argument. We're not trying to disambiguate whether doctors are getting better at treating COVID or COVID is itself getting easier for doctors to treat.

Age is a strong predictor of COVID fatality, either directly or through correlates. In every US county whose stats I've spot-checked, deaths go up with age. It's also one of the best predictors, given that advanced age is much more common in populations than stronger, but more selective factors.

Thus, controlling for age and nevertheless observing a reduction in mortality should be taken as an indicator that COVID is becoming comparatively more medically tractable.

Probably because governors are no longer deliberately putting COVID patients into nursing homes. This is why NY and NJ have 3 times (!) the death rate of Italy.

I'm shocked how many people think Cuomo is a good leader when he (along with Murphy) made perhaps the most catastrophic choice of the pandemic so far.

Most people will believe whatever they read in the newspaper, and newspapers are unanimous in their opinion in this regard. As seniors were suffocating to death in the old folks homes, newspapers were pushing Cuomo as a _presidential candidate_.

True. Cuomo is imo as close as you can get to a typical Democrat politician and we’re told that it’s obvious a Democrat president would have handled the pandemic better.

The stats of hardest hit states are coincidentally blue states.

By what metric? The top states by total cases per 100k are all red and purple.

https://covid.cdc.gov/covid-data-tracker/#cases_casesper100k

Arbitrary stat but ok. Deaths are kinda the most important otherwise we wouldn't be shutdown

Arbitrary!?

This is nonsense. Are you suggesting that if this illness had all the same characteristics -- long hospital stays, long-lasting damage to the body, highly transmissible -- but didn't kill, then we would not be "shutdown"?

Yes Arbitrary. Why 100k? Why not 10, 100, 1000, 10,000, 1,000,000? Oh because 100k makes the stats look the best for your argument.

It's per 100k people in the state. It makes no difference if it's 100 or 100k. The comparison was state-to-state. The absolute numbers don't matter.

Wrong hill to die on, bud. 60M people got infected with Swine Flu, and nobody even thought of shutting anything down. Why? Because it was about as deadly as regular flu. So yes, deaths are the right metric to track.

Completely irrelevant. There were an estimated 274k hospitalizations from swine flu, over the course of that year. So swine flu did not have the characteristics I specified.

https://www.cdc.gov/flu/pandemic-resources/2009-h1n1-pandemi...

"estimated 274k hospitalizations" - remind me again how widespread the swine flu tests were in comparison to COVID?

Measure by lives lost?

A case only really matters to clinical medicine if there’s some type of symptom.

Of course it benefits the “concerned” side to couch their argument in the idea of some murky, unobservable effects.

Are deaths the only thing that matters to you? Shouldn't we be just as concerned about hospitalizations? Is there any reason that you're aware of for cases to not be a good proxy for hospitalizations?

My point is that: hospitalizations != cases

It would make sense that an observed drop in deaths would correspond to a drop in hospitalizations if the case number is rising significantly.

That's good imo.

edit: CDC graph shows hospitalization rate steadily dropping.

When you say "observed drop in deaths" are you referring back to the original article? Because the first sentence is "Two new peer-reviewed studies are showing a sharp drop in mortality among hospitalized COVID-19 patients." so I'm having trouble making sense of what you're saying.

Reply on news.ycombinator.com