Hacker News Reader

Comment by m0llusk | Hacker News Reader

This is turning a flawed study into a conspiracy theory and ignoring important related epidemiology. The fact is that the large number of asymptomatic cases led to initial undercounting of infections so the virus appears to be less deadly when looking at numbers that include previously missed cases. This doesn't make the pandemic any less serious.

At this point we have data from places with no lockdowns, mild lockdowns, strong lockdowns, and recently ended lockdowns. All of this verifies the long held epidemiology that points out that once a highly infectious agent is loose in the general public lockdowns have extremely limited benefit. This is not a political issue and does not relate to managing airlines or running clickbait web news sites.

View on HN

"once a highly infectious agent is loose in the general public lockdowns have extremely limited benefit."

Huh? What data supports this? It looks like the top dozen or so countries in terms of new cases are almost completely different from a month or two ago. Isn't it obvious this is because the US and western Europe that had the lockdowns slowed it relative to others?

Here is a list in descending order of new cases from a day or two ago, see how the US and UK are the only remaining of the first wave. Spain and Italy aren't even there! I don't see how your comment could be more at odds with the information I'm seeing.

USA

Brazil

Russia

Peru

India

Chile

Pakistan

Mexico

Iran

Saudi Arabia

Turkey

That's mostly a list of most populous countries in the world. Is that data per capita?

https://ourworldindata.org/grapher/total-daily-covid-cases-p...

there you go (may 9)

  df=pd.read_csv("~/Downloads/total-daily-covid-cases-per-million.csv")
  df.sort_values(['Date', 'Daily new confirmed cases per million (cases)'])[-20:][['Entity', 'Daily new confirmed cases per million (cases)']].to_string(index=False)
  Out[14]: 
  Entity  Daily new confirmed cases per million (cases)
  Saudi Arabia                                    48.860
  Armenia                                         48.933
  Belgium                                         50.994
  Gabon                                           52.118
  Portugal                                        54.233
  United Arab Emirates                            55.913
  Gibraltar                                       59.363
  Sweden                                          63.569
  United Kingdom                                  68.482
  Chile                                           72.765
  Russia                                          73.314
  United States                                   81.440
  North America                                   84.775
  Sao Tome and Principe                           95.820
  Belarus                                         98.737
  Peru                                           100.722
  Singapore                                      131.274
  Bahrain                                        143.984
  Kuwait                                         150.097
  Qatar                                          455.041

That is interesting, but per capita data is not the answer to what I was interested in originally - what countries is the epidemic growing fastest in is different from in what countries is most of the epidemic's growth happening in.

No, it's based on the actual numbers of new cases.

Per capita data tends to have very small countries at the top because they have a larger urban proportion and a larger amount of interaction with foreigners.

To expand on my other post, since I can't edit it any longer, the fact that it's many of the largest countries with the most cases is a point of note - the first wave wasn't the largest countries so if that's where we're headed, things are changing.

There are a number of analysts comparing the effectiveness of attempts to mitigate this pandemic and finding lockdowns to be high cost and risk for low impact as compared to distancing, handwashing, masks, testing, and tracking. One of the most prominent is Isaac Ben-Israel https://www.timesofisrael.com/top-israeli-prof-claims-simple... out-after-70-days/

I'm sorry you don't like the actual math, but the reality is that the initial wave of this pandemic has been burning out in all the effected areas in very similar ways regardless of lockdowns or not. Personally I find detailed studies by experienced professionals like Isaac Ben-Israel more compelling than what you present here.

"I'm sorry you don't like the actual math"

How is that a response to widely reported data? What are you disagreeing with? Are you saying my data is incorrect or that I'm interpreting it wrong?

My list is based on numbers from https://www.worldometers.info/coronavirus/ - is there an issue with their sources?

not disclosing a financial conflict of interest is a mayor breach in how to conduct science, and there is exactly zero clickbait let alone conspiracy theories here.

The study was thoroughly flawed at the most basic levels of sampling and analysis. The article supposes that the authors of this flawed study did what they did because of who they knew and what the political implications of their published results might be. That is a conspiracy theory and it is presented for emotional impact which makes it clickbait.

Meanwhile increased testing is showing increased numbers in the population who are infected with only minor if any symptoms. This has major implications for public health and correct policy, but what is being asserted is that anything that causes a change to the overall mortality rate is likely about politics.

their study is flawed, the scientists involved went almost exclusively onto right-wing programs to hype their biased results, and they didn't disclose important financial ties - particularly the jetblue guy that wrote an op-ed for the ben shapiro website (damning imo). you can whine about this article being clickbait but it's incredibly well-written and sourced.

>The whistleblower complaint made note of the authors’ media appearances, observing that they “coordinated the online publication of a non-peer-reviewed pre-print with multiple press appearances” and “continued to promote the results of this and subsequent studies despite severe criticisms of their pre-print from multiple statisticians.”

Indeed, scientists immediately attacked the high-profile study’s recruitment of participants and data analysis, with one calling it a textbook example of “how NOT to do statistics” and another saying he was “alarmed at their sloppy behavior.” One major cause for alarm was that the authors were overly confident in their test’s false-positive rate.

They explained that by saying they went to those sources because left wing news outlets and left win financial source black balled them. But that's not even the issue. How do you explain the fact that their numbers fall right in line with the study from Bonn Germany, or New york, or all the other random sampling studies. Are they all lying?

New York has far more deaths per capita, it makes no sense for their % positive numbers to fall in line with New York. It is similarly inconsistent with recent numbers out of Spain/Madrid, which was hit quite hard and is still floating around only 10% positive.

There have been, conservatively, at least 20,000 COVID-19 deaths in NYC alone. NYC has a population of 8,300,000. 20,000 / 8,300,000 puts a hard lower bound on IFR of ~0.24%. That is, if we assume that the entire population of NYC has already had SARS-CoV-2, then the lowest mathematically possible IFR for NYC would be ~0.24%.

Ioannidis' paper says 2.49%-4.16% of Santa Clara County has already been infected. Santa Clara County has a population of 1.9 million. That means we should expect (1900000 * 0.0249) * 0.0024 == 113 deaths at the lower bound, and (1900000 * 0.0416) * 0.0024 == 189 deaths at the upper bound. Santa Clara County has already recorded 134 deaths, so that number seem to fit. But it fits with the hard, categorical data from NYC only if we assume 100% of NYC to have already been infected.

TL;DR: Unless you're prepared to believe that

1) most of NYC residents have already been infected, or

2) that IFR is significantly greater than 0.24% and the actual number of COVID-19 deaths in Santa Clara County is multiples of what's reported

then it's clear that Ioannidis' paper is wrong. #2 seems far more plausible and may even be true, but the controversy is the implication of his paper on derived IFR. And if #2 is what's happening, the IFR is definitely greater than what people applying the paper (including, AFAIU, Ioannidis' himself) predict.

I will not defend his study. Pretty sure there are enough holes in it.

I am not sure why we think the IFR would be constant by region, though. Just doing an age banded calculation is enough to show that it is variable for populations. Such that it seems safe to assume other things could cause variation, as well.

An early thought I had was age being a proxy to lung damage from long term exposure to air pollution. This still holds for the population in NYC, as many of them are from there. Same for Italy. Many elderly are retired locals. Notable that both places have some of the worst air in the world. Contrast with California, which has been very progressive with emissions. I don't have data on age ranges in California. And I confess I have stopped following the news on this. (Peek in the occasional story that makes it here. Starting to pay more attention again.)

The NYC data is the outlier. There may be contributing factors which are causing that 20,000 number to be inflated, not the least of which are the financial incentives for emergency funding for COVID diagnoses.

Which numbers? The inferred IFR?

Reply on news.ycombinator.com