3 ms·

The Oxford study is based on blatantly, and easily demonstrably, bad data. For example it relies on only 1 out of every 1000 infections needing hospitalisation.

by jpxw 7y ago

The Oxford study is based on blatantly, and easily demonstrably, bad data. For example it relies on only 1 out of every 1000 infections needing hospitalisation. This is not supported by any current evidence. https://www.wired.co.uk/article/coronavirus-infections-oxford-study-immunity https://www.wired.co.uk/article/coronavirus-infections-oxfor...

gadders 7y ago
Conversely, the lead on the Imperial Study was also in charge of modelling for the UK foot and mouth epidemic which was found not fit for purpose and resulted in lots of livestock being slaughtered unnecessarily. And this is the one the UK and other governments seem to be basing their approach on. "He was behind disputed research that sparked the mass culling of farm animals during the 2001 epidemic of foot and mouth disease, a crisis which cost the country billions of pounds. And separately he also predicted that up to 150,000 people could die from bovine spongiform encephalopathy (BSE, or ‘mad cow disease’) and its equivalent in sheep if it made the leap to humans. To date there have been fewer than 200 deaths from the human form of BSE and none resulting from sheep to human transmission." https://www.telegraph.co.uk/news/2020/03/28/neil-ferguson-scientist-convinced-boris-johnson-uk-coronavirus-lockdown-criticised/ https://www.telegraph.co.uk/news/2020/03/28/neil-ferguson-sc...
taipan100 7y ago
To call the Oxford study "utter rubbish" is foolish. Your "example" of bad data is the only quoted data in that article and it doesn't even mean anything since the implicit assumption of the Gupta model is that we do not know how many asymptomatic cases of COVID exist (even WIRED concede this point). Testing in Italy is insufficient to tell us this. 1 in 1000 infections requiring hospitalisation could be a realistic number if a high percentage of infections are asymptomatic. > we just won’t know the true proportion of people who have contracted the disease without showing any symptoms, but it is likely a much lower number than the Oxford study assumes. Epidemiology is not done by "it is likely much lower than the study assumes" since that is pure guess work without the tests. What the Oxford study offers is a strong argument that antibody testing is vitally important and nobody is doing it. As I say this is a very long way from utter rubbish.
- jsnell 7y ago
  1/1000 was not realistic when the preprint was published, let alone now. It showed that one could fit a bunch of curves onto 14 days of data on deaths. Which is true, but also utterly unintersting. The reason that sorry excuse of a model got so much attention is that the authors did not stop there. They also made the totally unsupported assumption that one of the lower end curves matched reality. (That's right. They actually made exactly the kind of assumptions you're accusing others of, rather than just argue for antibody testing). The problem is that when the study was made, there was a lot more data available than just that 14 days of deaths from two countries. And a ton of it was totally incompatible with their modeling. Here's some of the conflicting data points as of ten days ago: https://news.ycombinator.com/item?id=22698584 https://news.ycombinator.com/item?id=22698584
- kgwgk 7y ago
  > Testing in Italy is insufficient to tell us this. It's not about "testing in Italy": "In Lombardy – despite the region being under lockdown since March 9 – more than one in every 1,000 of the entire population have already been hospitalised due to coronavirus."
  - lbeltrame 7y ago
    But we don't know the actual amount of people which were infected in Italy, because some of these (a sizable percentage) only reported fevers, and perhaps didn't even think about having SARS-CoV-2 in their bodies. An acquaintance's partner suspects having got it, because after two days (two days, not weeks) of very mild fever (~37.5C) he was hit by anosmia. An ex-coworker also mentioned "a horribly strong fever" with respiratory difficulties which lasted just a few days. Yes, anecdata, but shows that you can easily miss a large part of the infected cases if you only test those hospitalized.
    - kgwgk 7y ago
      You’re missing the point. We know for sure that in Lombardy the hospitalisation rate was higher than 1 in 1000 people infected.

by jpxw 7y ago

gadders 7y ago

Conversely, the lead on the Imperial Study was also in charge of modelling for the UK foot and mouth epidemic which was found not fit for purpose and resulted in lots of livestock being slaughtered unnecessarily. And this is the one the UK and other governments seem to be basing their approach on. "He was behind disputed research that sparked the mass culling of farm animals during the 2001 epidemic of foot and mouth disease, a crisis which cost the country billions of pounds. And separately he also predicted that up to 150,000 people could die from bovine spongiform encephalopathy (BSE, or ‘mad cow disease’) and its equivalent in sheep if it made the leap to humans. To date there have been fewer than 200 deaths from the human form of BSE and none resulting from sheep to human transmission." https://www.telegraph.co.uk/news/2020/03/28/neil-ferguson-scientist-convinced-boris-johnson-uk-coronavirus-lockdown-criticised/ https://www.telegraph.co.uk/news/2020/03/28/neil-ferguson-sc...

taipan100 7y ago

To call the Oxford study "utter rubbish" is foolish. Your "example" of bad data is the only quoted data in that article and it doesn't even mean anything since the implicit assumption of the Gupta model is that we do not know how many asymptomatic cases of COVID exist (even WIRED concede this point). Testing in Italy is insufficient to tell us this. 1 in 1000 infections requiring hospitalisation could be a realistic number if a high percentage of infections are asymptomatic. > we just won’t know the true proportion of people who have contracted the disease without showing any symptoms, but it is likely a much lower number than the Oxford study assumes. Epidemiology is not done by "it is likely much lower than the study assumes" since that is pure guess work without the tests. What the Oxford study offers is a strong argument that antibody testing is vitally important and nobody is doing it. As I say this is a very long way from utter rubbish.

jsnell 7y ago
1/1000 was not realistic when the preprint was published, let alone now. It showed that one could fit a bunch of curves onto 14 days of data on deaths. Which is true, but also utterly unintersting. The reason that sorry excuse of a model got so much attention is that the authors did not stop there. They also made the totally unsupported assumption that one of the lower end curves matched reality. (That's right. They actually made exactly the kind of assumptions you're accusing others of, rather than just argue for antibody testing). The problem is that when the study was made, there was a lot more data available than just that 14 days of deaths from two countries. And a ton of it was totally incompatible with their modeling. Here's some of the conflicting data points as of ten days ago: https://news.ycombinator.com/item?id=22698584 https://news.ycombinator.com/item?id=22698584
kgwgk 7y ago
> Testing in Italy is insufficient to tell us this. It's not about "testing in Italy": "In Lombardy – despite the region being under lockdown since March 9 – more than one in every 1,000 of the entire population have already been hospitalised due to coronavirus."
- lbeltrame 7y ago
  But we don't know the actual amount of people which were infected in Italy, because some of these (a sizable percentage) only reported fevers, and perhaps didn't even think about having SARS-CoV-2 in their bodies. An acquaintance's partner suspects having got it, because after two days (two days, not weeks) of very mild fever (~37.5C) he was hit by anosmia. An ex-coworker also mentioned "a horribly strong fever" with respiratory difficulties which lasted just a few days. Yes, anecdata, but shows that you can easily miss a large part of the infected cases if you only test those hospitalized.
  - kgwgk 7y ago
    You’re missing the point. We know for sure that in Lombardy the hospitalisation rate was higher than 1 in 1000 people infected.