Hacker News Reader

Hacker News

Comment by nostromo | Hacker News Reader

Parent Full thread

nostromo

More testing = more cases.

Actual deaths are down and have been trending down for months.

https://i.imgur.com/83sw5pd.png

https://www.worldometers.info/coronavirus/country/us/

I have no idea why this isn't front page news.

View on HN

Deaths are trending upwards, however, in Florida, Texas, Arizona, and all the other Southern hotspots.

This is an example of Simpson's paradox. You're looking at the sum of two different timelines, the virus in the Northeast (deaths declining) and the South (rising, but slow).

https://en.wikipedia.org/wiki/Simpson's_paradox#/media/File:...

https://www.nytimes.com/interactive/2020/us/arizona-coronavi...

https://www.nytimes.com/interactive/2020/us/florida-coronavi...

https://www.nytimes.com/interactive/2020/us/texas-coronaviru...

Isn't Simpson's Paradox when something is true of both parts but not of the whole? Wouldn't that make it not apply if the death rate is falling in the Northeast but not in the South?

Correct, this isn't Simpson's Paradox.

Because you are being misleading?

1) Avg age of infected is 36-37 now instead of 60+ as in April, so there is a lower death rate (plus, infections today will only translate into deaths in three weeks or so)

2) Test positivity rate is UP, so testing has gone up "a little" while cases have gone up a lot.

This is bad analysis.

It was impossible to get tested unless you were high-risk in April because there were so few tests. High-risk for Covid-19 means old.

Now we have more tests than we know what to do with -- so the average age is of course going to trend lower.

Also, test positivity is not up -- it's down:

https://pbs.twimg.com/media/Eb7_BMOU4AEdrLm?format=jpg&name=...

And yet, test positivity goes up? That makes no sense.

Younger people aren't locked down and are more likely to be social ATM. Since they are more likely to carry without showing symptoms, test positivity going up makes some sense, it is plausible at least.

No. Stop spreading disinformation and inform yourself.

It’s pretty easy to compute normalized infection rates for populations based on samples of data, which haven’t changed materially with the testing ramp up.

Additionally, antibody testing allows for an assessment of past infection rates at a community level.

You're completely skipping over the positivity rate; many countries are testing a heck of a lot more than the US is; and they're showing positivity rates in the 1-5% range.

If your hypothesis is correct, and infection rate is going up just because of more testing, then we should be pacing with other countries that are also testing more. We're seeing positivity rates of almost 25% in Arizona and 19% in Florida and Texas; which way out strips countries that have this more under control. At that rate of 15% or more, how many people are being missed (or are you also assuming we're catching 100% of all cases)? The more likely interpretation of the data is that not only are we testing more, but there are more cases; which fits with the overall data trend of BOTH more tests AND more infections.

I'm going to nip your goalpost moving in the bud here - since I'm sure you'll come back with "yeah, but deaths are going down."

Just because mortality has been going down doesn't mean that it isn't affecting the younger group and they're just magically recovering back to baseline. There's good evidence that there are lasting lung and immune response issues post "recovery" if it doesn't kill you. Further, deaths are a lagging indicator, and unfortunately what timeframe on that is unknown now because the age group is skewing younger and a large portion of our dataset is skewed towards the 65+ group due to how the initial wave of the disease spread.

By your own source, test positivity is up over the last 4 weeks.

> Now we have more tests than we know what to do with

A coworker related about a month ago that when his wife, who works in a hospital and showed some symptoms of covid-19, asked to be tested, she was told that the nearest place with available tests was a 2.5 hour drive away.

More young people going out = more cases. Those people infecting their parents and grandparents may lead to more deaths in due time.

And it has been news. https://www.nytimes.com/2020/07/03/health/coronavirus-mortal...

"just wait two weeks!"

People have been saying this for months now, and it hasn't been true at any point. Daily deaths have been falling as cases rise.

They will continue to say “just wait two weeks!” two weeks from now.

That’s a particular challenge with this virus: it takes a long time between being exposed and actually suffering from the disease. Which makes it really easy to dismiss the impact because the side saying “it’s a big deal” has to wait two weeks or more while the side saying “it’s not a big deal” gets to claim victory for two weeks before being proved right or wrong.

The fact is we don’t know a lot about how this virus operates or why it does what it does. What we do know is that there are fates much worse than death that come from this disease (reduced lung/heart/organ function, long hospital stays with very expensive bills at the end, etc). We know it spreads quickly and we know it’s spreading uncontrolled right now. We know at least 130,000 people in the US have died and many more will be permanently injured. We know the countries who handled the response right, because they are opening back up. We know the countries who handled it poorly because their cases are still going up.

All of the things we don’t know are problems and we need to keep learning. All the things we DO know tell us this is bad and it’s going to get worse before it gets better unless people start taking it seriously. Unfortunately many people are taking the opposite approach and claiming victory based on the unknown factors we still don’t know answers to.

As far as I can tell, nothing you wrote addressed the post you responded to.

The claim was that deaths having been going down. They have for 90 days. We know that deaths lag diagnosis by 7-14 days.

The out of control growth "side" has lost it's credibility.

>All the things we DO know tell us this is bad and it’s going to get worse before it gets better unless people start taking it seriously.

Deaths is the most accurate data we have and it shows us things have been continuously getting better for 90 days. Of course we are not out of the woods, and should continue to implement reasonable controls.

Doesn't this indicate that the information we used in weeks and months past to open up and make decisions about the fall was flawed, since we may not have understood the full severity of the infection rate?

I'm glad that death rates are down, but unless universities completely waive all attendance policies (and adjust grading systems accordingly) infections can't be ignored. Additionally, not dying doesn't mean a quick infection: there are long-term effects.

Not really. We’ve known the cases were much more widespread for months now from anti-body testing. Some estimates are 10-20x higher. CDC now quotes an infection-fatality-rate (IFR) of 0.3%, or about 3x the flu.

Initially the lockdown was enabled to flatten a curve of a disease we had little information about. Months later, we have lots more info about it. Treatments are better. Flattening the curve won’t change the area under the curve, and the majority of the US did not sign up to lockdown until a vaccine was released.

"Treatments are better" very much means the area under the serious illness/death curve is smaller.

And deaths are a lagging indicator. It takes up to two weeks, median around one week to show symptoms, and then another two weeks, median a week and a half or so, to become seriously ill.

The outcome after that depends on predisposition and quality of care. But one developer I know spent forty days on a ventilator.

So if lockdown isn't being observed and masks aren't being worn, it's going to take 4-6 weeks for that to start significantly increasing deaths.

There are rumours in the UK of another national lockdown in September/October, but I suspect local lockdowns will become a thing long before then.

> the majority of the US did not sign up to lockdown until a vaccine was released

False dichotomy. Responsible mask-wearing allows many (admittedly not all) areas of society and the economy to function without a full lockdown.

Because the number of positive cases is rising faster than the increase in testing.

Also, make sure to split out the data per state/region. The decline from NYC for example made up a huge amount of the decline. Otherwise you run into Simpson's Paradox.

Miles Beckett did a decent Twitter thread on the topic: https://twitter.com/mbeckett/status/1278750652160634880

It’s not “front page” news because it is inaccurate.

Deaths have declined for various reasons, including improved treatment (rolling patients and avoiding ventilators). Also we have not had the overwhelming of hospitals that took place in NYC and Italy repeat yet. Between the protests and people cutting loose over the holidays, I certainly hope we don’t see a repeat of that.

For what its worth, NYC had large protests through all of June and saw no spike. Mask usage was also high through them + regular use of hand sanitizer (handed out by organizers in some cases). It would appear that spikes are a function of either indoor activity or large crowds without masks.

>>More testing = more cases.

Please for the love of god stop repeating this extremely misleading claim. The increase in new cases cannot simply be explained by the increase in testing.

https://www.washingtonpost.com/outlook/2020/06/22/no-more-te...

>> Deaths, which happen regardless of how many tests you run, are down and have been trending down for months.

Deaths are a lagging indicator. Cases started to spike recently. Deaths will start to increase soon, especially when ICUs fill up and non-covid patients can no longer be treated in intensive care either.

Reply on news.ycombinator.com