Of all the things I thought I'd see on hackernews tonight, "Covid is fake" wasn't one of them.
The linchpin of this article is that the Germany death numbers are not substantially different: that is, Germany has not had excess deaths due to the virus when compared to death numbers from last year. Unfortunately, this is the worst possible conclusion you could draw from that graph. Germany went into lockdown, and had people stay home and social distance. Consider, for example, that Germany's traffic deaths are the lowest they have been in 100 years[1]. Influenze has spread less extensively, people have died less from violent crimes, and other, similar phenomena have all occurred as a result of their lockdown. In spite of all of that, Germany has still hit their average death number.
What could possibly explain this? Maybe, I dunno, the pandemic?!
That isn't the lynchpin of the article. Did you actually read it to the end? That graph is a throwaway at the top which it explicitly says immediately after, isn't the focus of the article.
The article is about false positive rates on the PCR test.
Germany went into a form of lockdown, but so did the UK, and the UK saw relatively high excess deaths. There has been a lot of analysis of the data on this, and lockdown duration/start/severity doesn't correlate with outcomes in any obvious way.
I know the article is about false-positive rates. I did read it. But the claim is:
1. There is not a significant uptick in deaths in Germany.
2. The test is not 100% accurate, and indicates some false positives.
3. From #1 and #2, we can conclude that the "pandemic" we saw there was the result of false positives, and not people actually getting sick.
If #1 isn't reliable, then neither is the conclusion. As hospitalizations and death climb in the US, writing a think-piece article pointing at how Spain is mis-accounting for false positives is irresponsible and dangerous. It pushes a "reopening is fine" narrative, when the people most-likely to read the article are living in places where that is patently false.
I don't believe (3) appears anywhere in the article. That's your inference, not the authors.
As for (1), the graph clearly shows there is no uptick in deaths in Germany. So I'm not sure what you mean by, if (1) is not reliable. The graph is right there.
As for your final point, you seem to be assuming the article was written by an American for a US audience. What's your evidence for that? COVID is bigger than the USA, you know.
Here is where (3) is heavily implied in the conclusion of the article:
...if something causes test numbers to rise then so will case numbers, which in turn will cause a further increase in testing, causing the rise to continue, triggering local lockdowns and pointless evidence free rituals, until people get depressed and stop trying to do things causing numbers being tested to fall again...
Health is run by people who suffer no consequences from policy over-reactions. Lockdown induced job losses won’t affect them, as they work for the government. A larger-scale case of “one rule for them and another for us” can’t be imagined. It’s thus no surprise when we read things like Public Health England defining a COVID death as anyone who has ever tested positive and then died, for any reason, at any time i.e. the UK being supposedly “one of the worst hit countries in the world” is a statistical fantasy. PHE officials defined it this way because they didn’t want to be accused of being a nasty libertarians who were underplaying the problem just to help capitalist workers. The idea that they’d create other, bigger problems simply didn’t occur to them — or worse, it did but they didn’t care.
This is, indeed, a bit of speculation on my part, but I do not know how else to interpret this beyond "the pandemic wasn't anywhere near as severe as the tests suggests, the majority of cases we saw are false positives, and lockdowns are much worse than the actual risk." What do you think I should take away fro this conclusion, instead?
--------------------
My top comment in this thread explains why (1) is not reliable. The graph is a trick of the eyes, due to a lack of consideration for externalities. Over the entire time period, ~10,000 people died in Germany of Covid. Amortized over a 10-week outbreak period, that's 1k / week, on a graph with a scale an order of magnitude higher. Combined with the citation I provided for fewer traffic deaths, plus additional drops in violent crimes, seasonal flu, etc., it is not unimaginable that "everyone staying home" saved a thousand lives a week, or more, during that same time period.
--------------------
My final point is that the assumed audience is members of the UK, and given the author's concluding thoughts are mostly concerned with their degradation in respect for the UK health establishment, I think that's sort of a fair assumption to make.
The most insightful chart in this article is the % of positive cases per number of tests. Surprised that this statistic isn’t more readily available to the public. As we increase testing capabilities in the US, if is inevitable that the number of detected cases will also rise - parsing out the difference between a higher detection rate, and a higher infection rate, is something that can easily provide clarification on the current situation.
> As we increase testing capabilities in the US, it is inevitable that the number of detected cases will also rise
Most people miss that aspect and just look at overall case count.
Personally, I'm more and more encouraged every day that the case count grows dramatically and the resulting deaths don't. That means it's far far less deadly than we thought in March.
Except the death count has followed cases count nearly exactly with about one month of delay. (R > 0.9 for dumb linear fit, better for other kinds.)
Unless you're alleging that testing people or quarantining them kills them, which would be laughable. Or some grand scale conspiracy theory to make it look like a pandemic worldwide.
In Spain the death graph lagged the cases graph by only a few days in the first wave. In the second wave of cases (positive tests) they're now experiencing, deaths haven't gone up at all, as apparently the patients are now younger and recover quicker.
Can you show some graphs and working for this claim that death graphs are basically just lagged versions of the case graph? Also why would deaths follow cases by a month when infection-to-death is supposedly about 18 days.
I'd be interested in whether false positives are statistically independent when you run the test multiple times on the same patient.
If they are, this suggests that at some point you ought to simply re-test everyone who tests positive the first time.
"At some point" likely only comes once the disease is well under control, maybe down to 3% of current active cases, just as a wild guess? When that happens, the prior of actually having COVID-19 gets really low. Such that even with a positive COVID-19 test the posterior probability is still somewhat low.
This all falls apart if multiple tests on the same patient aren't statistically independent in false positives. For example, I'm no biologist, but I'd guess it's possible there are a few people out there whose human genetics by chance have a short sequence in common with the viral genetics of COVID-19. If that's the case, that part of the patient's own genetic code might invariably get picked up and amplified by the PCR test. So these unlucky folks will always test positive for COVID regardless of whether they have it or not.
They would also need to have the antibodies tested for in the non-PCR tests. (ELISA specifically.) Which reduces the likelihood of exactly this error to incredibly low.
Cross-testing some similar coronavirus as positive is much more plausible mechanism of failure. However, different tests check different subsequences.
There's no point in testing, retesting if the test isn't precise. Are there any reliable tests available? Statistics points that tests aren't reliable.
Excess deaths aren't reliable either, because it can happen independent of any new disease.
The fact that strikes me most is that people have the same flu symptoms, but no "new" symptoms different from the diseases we already know.
Generally speaking ... If the false positive rate is low enough and the costs of performing the test aren’t prohibitive... running the test multiple times makes sense regardless of how precise or sensitive it is.
an example with math... if false negative is 75%, but false positive is 0%, then you just run the test multiple times to increase confidence in the results.
The article is only tangentially about Covid test reliability. It's actually a "reopen the UK" thinkpiece with some poorly-drawn conclusions to support that stance. While it points out the obvious insights you get with testing unreliabily (which anyone who knows anything about statistics will nod along with), it mistakenly uses it to reach this conclusion:
...if something causes test numbers to rise then so will case numbers, which in turn will cause a further increase in testing, causing the rise to continue, triggering local lockdowns and pointless evidence free rituals, until people get depressed and stop trying to do things causing numbers being tested to fall again...
Health is run by people who suffer no consequences from policy over-reactions. Lockdown induced job losses won’t affect them, as they work for the government. A larger-scale case of “one rule for them and another for us” can’t be imagined. It’s thus no surprise when we read things like Public Health England defining a COVID death as anyone who has ever tested positive and then died, for any reason, at any time i.e. the UK being supposedly “one of the worst hit countries in the world” is a statistical fantasy. PHE officials defined it this way because they didn’t want to be accused of being a nasty libertarians who were underplaying the problem just to help capitalist workers. The idea that they’d create other, bigger problems simply didn’t occur to them — or worse, it did but they didn’t care.
The key to this is the over-estimation of deaths. If cases are up, but deaths are not, the author argues, we should reopen everything and stop large-spread testing, because those cases are false positives. And that's a reasonable stand, in the abstract, and the discussion of false positives is worthwhile.
Unfortunately, the author uses this in combination with the possible (and unsubstantiated) over-counting of deaths (and some weird libertarian persecution complex) as a way to justify reopening the country and stopping testing. That starts to sound like a claim toward "Covid is not a big deal" at best (and "Covid is fake" at worst).
You might not like or agree with article, but I actually found it very well reasoned and provide alternative point of view to a very important discussion.
Thus I'm really puzzled why this community would flag it, it is disappointing frankly.
I think an article about false positive rates, and signal to noise, is indeed interesting. Discussing how most of the tests we have float around 1%-2%, and leaning on that to argue that reopening may be safe when tests are steady around those levels, may actually be worthwhile.
Unfortunately, the article doesn't do that. Instead, it:
- Calls into question the legitimacy of coronavirus death impact, consulting a flat death without considering the externalities of reduced risks that lockdowns bring about.
- Links to the article suggests healthy people shouldn't be wearing masks, even though many people are contagious well before showing symptoms.
- Calls into question Spain's second wave as false, the result of over-testing, even though [it isn't, the case count is rising far above any expected false-positive rate, and people are dying there](https://www.worldometers.info/coronavirus/country/spain/).
- Puts forth the sentiment: "The idea that [lockdowns] create other, bigger problems simply didn’t occur to [Public Health England] — or worse, it did but they didn’t care."
If the article addressed this head-on and used that metric to determine which lockdowns might be necessary, that'd be one thing. These conclusions, however, are a bit inexcusable.
The linchpin of this article is that the Germany death numbers are not substantially different: that is, Germany has not had excess deaths due to the virus when compared to death numbers from last year. Unfortunately, this is the worst possible conclusion you could draw from that graph. Germany went into lockdown, and had people stay home and social distance. Consider, for example, that Germany's traffic deaths are the lowest they have been in 100 years[1]. Influenze has spread less extensively, people have died less from violent crimes, and other, similar phenomena have all occurred as a result of their lockdown. In spite of all of that, Germany has still hit their average death number.
What could possibly explain this? Maybe, I dunno, the pandemic?!
1. https://www.dw.com/en/germany-traffic-deaths-fall-to-lowest-...