Forum Topic

Ferguson model may be spectacularly wrong

The epidemiological model used by several governments globally predicted massive death counts and overwhelming of health services.  It was a major input into population wide house arrest policies.These sort of models tend to be unverifiable and unfalsifiable (therefore not science) as their creators can always claim actual outcomes which differ from their predictions are down to their advice being taken.But now we have many variations of policy globally giving a unique opportunity to validate the model.  Swedish researchers at Upsala University have done just that, feeding parameters representing Sweden's light touch policies into the model and finding that it massively over predicts deaths and ICU utilisation.Here is an article about their work with a link to their paper:https://hectordrummond.com/2020/05/05/the-swedish-version-of-fergusons-model-turned-out-to-be-hopelessly-wrong/This is just one of many studies casting doubt on the validity of the ICL model.

Andy Riley ● 2060d29 Comments

There are just so many issues. David Spiegelhalter is big on the topic of excess deaths and there are several videos of him talking about the issue.Not so much in this country but I believe that elsewhere there are emerging shortages of donated blood. I don't think that there is any evidence that CV can be transmitted via donated blooid but I guess that many people simply want to avoid going into any sort of clinical situation.There's always a silver lining! Experts in the field are saying that the current situation is an unprecedented opportunity to dramatically reduce or even eliminate some sexually transmitted infections.  Professor Pantsdown has weakened yet another message by converting the length of his social distancing to a negative figure.

Jim Storrar ● 2057d

Good point Susan,And another effect (which I believe has already been seen - I'll try and find the reference) will be displacement of deaths from other causes such as flu to Covid.  Also the excess deaths already seen in the stats from lock down related deaths i.e. people missing cancer treatments etc.  This will continue for some time onwards as people die earlier than otherwise because their treatments were interrupted by lock down.

Andy Riley ● 2057d

Here is a possible interpretation of events.  Prof Pantsdown realises the game his up with his dodgy epidemiological model as it comes under scrutiny from software engineers and critical scientists.  He needs to get out of the scientific limelight so arranges a smokescreen for his escape in the form of being caught in an illicit tryst with his very photogenic lover.  (Note that this is not a conspiracy theory as he didn't have to conspire with anyone).On the other hand it may have just been a wayward libido at work.

Andy Riley ● 2057d

I think the most important statistic when this crisis eases will be how many extra deaths there were than normal in any given country. After all, 600,000 people die in the UK in a typical year: that's one every minute.The median age for women dying of Covid-19 in the UK is 84. Anyone want to take a guess as to what the median age for women dying of all causes in a typical year is?That's right.

Susan Kelly ● 2057d

The code for the guts of the ICL model has now been released and is the subject of much online discussion.  IMHO as a former programmer, the code stinks - it is the work of a dabbler.The code is stored in an open repository called GitHub where issues can be raised and discussed.  There is a lively debate going on between software engineers and the scientsist:https://github.com/mrc-ide/covid-sim/issues/165

Andy Riley ● 2058d

Apologies folks - one could post on this topic ad infinitum.If you haven't already seen it I have at last found the link to the meeting of The Independent Sage:https://youtu.be/L7uBwyr0sdgThis gig lasts for almost 3 hours but it's well worth dipping in and out.

Jim Storrar ● 2058d

I predict that the s--t will really hit the fan when someone called Jennifer loses her earlobe. Mark my words!

Jim Storrar ● 2058d

I had thought MD was BG too, but he's not. He uses his real name in the podcast but I'm as useless remembering names from podcasts as I am in real life.Jim lists a good bunch of candidates. I could probably add some others but I think a period of reflection and respect for his missing earlobe is called for.

Guy Lambert ● 2058d

Guy,Thanks for the links, I'll listen to the podcast later - MD is, I believe Ben Goldacre, someone I have followed in the past when he did a column in the Observer.  Really worth looking at his book Bad Pharma to understand how what appears to be science is often not.I hope you are looking at the Z-score graphs to compare mortality.  As I posted elsewhere, comparisons between countries are fraught with difficulties and will take researchers some time to work through the data to give meaningful comparison, which then will allow some sort of multi factorial analysis to indicate underlying causes for the differences. However, with caveats, the data is still useful to indicate trends within a country, and I note that Sweden is still not an outlier, providing continuing support for the hypothesis that the ICL model is flawed.I do appreciate you "having a lot on" - you clearly take the Councillor role seriously and put a lot into it.5G - don't go there!  Except regarding the decision to hand over our parts of our telecoms infrastructure to Huawei.

Andy Riley ● 2058d

Yes, the Private Eye podcast is well worth a listen. I see that the venerable organ has revealed that we may have reached "peak graph" and that we may see a flattening of the curve followed by a gradual reduction in graph numbers.The international comparison of Z-scores for excess deaths illustrates very strikingly the poor performance of England. It's fascinating to read elsewhere that Greece may have implemented the most effective public health strategy in Europe. Many will argue that this is not the right time to look back and pick over the what, how, and who of where we got things but we can make a start on a list of candidates e.g austerity, failure to integrate health and social care, failure to plan and prepare, dropping the testing programme.I see that there are estimates of 18,000 to 50,000 additional cancer deaths due to lack of early intervention. These and other deaths will be appear in the statistics for years to come.The other day my wife gave me my first haircut since lockdown began (no, since Christmas) and I can honestly say that it is the finest haircut I have ever head. I didn't even feel the need to give a tip. However I have noticed that my left earlobe is now missing. Does this matter? Should we cut off the other one? Is this another widespread example of collateral damage?

Jim Storrar ● 2058d

England's 'excess deaths' are far and away the highest in Europe - only Spain comes close:https://www.euromomo.eu/graphs-and-mapsThis tracks how many have died against how many might have been expected to die and is thus very hard to fiddle.On that basis the performance of England is pretty catastrophic (note that Scotland, Wales and particularly Norn Iron are nothing like as bad).This Private Eye podcast https://www.private-eye.co.uk/eyeplayer/play-370 is well worth a listen. The doctor who leads it admits he underestimated the impact of COVID at the start (and in fact because he is someone I follow and respect I fell into the same trap)You're fair to criticise me for not reading the works of 'Drummond'. I have a lot on my plate at present and a lot of people look for Councillors to endorse dubious theories from dubious 'professors' - I wasted a lot of time engaging the the 5G conspiracy theory a little while ago so this has made me selective. In any case, at the stage we have got to  I don't see any realistic option but to pursue a gradual easing with maximum test and trace and analysis of what went wrong in the first place can wait.

Guy Lambert ● 2059d

The UK's death toll in relation to how many per million is way below most countries in Europe, Belgium faring the worst if this measure is employed.China's death toll per million is even lower, as you might imagine. Good stats. collection, fantastic health care facilities or major porkie pies from the orient? You decide.

Keith Iddon ● 2059d

Jim,Thanks for the link - I'd already watched about 15 minutes, I'll finished it later.I do like Peter Whittle's interviews.  Always level headed, calm and informative.

Andy Riley ● 2059d

And he said he believed himself immune after infection.  This contradicts other experts who say the question is open about immunity.

Andy Riley ● 2059d

Keith,Excellent question!I used to work in health care information so have some experience of the complexity of gathering health data.  In the UK we are fortunate in having a well developed set of standards governing definitions and coding, plus commonality across the sector.  Some of these standards are international e.g. ICD - the International Classification of Diseases and many are not. Also with differing health delivery models data quality may be stronger in some areas than others. Think of the possible biases introduced by an insurance based system vs ours.As to the differences in counting Covid cases I've not seen anything that directly answers your question across all countries - I suspect in time studies will be done that address this.Links below give a bit of insight into some of the general issues regarding health data comparisons:https://www.statslife.org.uk/features/4474-a-statistician-s-guide-to-coronavirus-numbershttps://www.statista.com/statistics/1104709/coronavirus-deaths-worldwide-per-million-inhabitants/https://www.kingsfund.org.uk/publications/articles/big-election-questions-nhs-international-comparisons

Andy Riley ● 2059d

I agree, Andy; and it seems that Prof Neil Ferguson of ICL didn't believe his model, having had to resign after being caught sneaking a lockdown whore into his place.

Michael Brown ● 2059d

It might help if all countries counted the number of deceased due to the virus in the same way. I don't know what the different methods used in different countries are. Does anyone on here know?

Keith Iddon ● 2059d

Andy - Damn! That was one of my favourite quotes, but thanks for putting me right.I should have been much more explicit. What I intended to say was that it will take a long time to fully assess the relative success of different approaches to Covid. One of the best examples of this is the international comparison of excess deaths. This will be no simple task because of the need to take account of numerous intervening factors such as age structure, culture, population density, ethnicity, income etc.etc.I always enjoy listening to Brendan O'Neill. It's hard to beat a good Marxist libertarian! This is a few weeks old now but worth a look:https://youtu.be/3qs5ryRCBmM

Jim Storrar ● 2059d

However, we don't have the luxury of waiting for years, or even months. We need to make decisions now or soon. I think many have become insulated from the fact life is thoroughly dangerous and risky - humans have always had to make decisions based on meager data between bad choices with uncertain outcomes.It was Zhou Enlai and he was talking about the French riots of 1968.Damn! Another cherished quote shot down - I too believed it was about 1789.https://marginalrevolution.com/marginalrevolution/2011/06/it-is-too-soon-to-tell-the-real-story.htmlI suspect this issue roughly follows the libertarian/authoritarian axis (see the Political Compass web site) and here are some interesting thoughts from Brendan O'Neill (libertarian left)https://www.spiked-online.com/2020/05/01/unlock-the-people/

Andy Riley ● 2059d

When Chairman Mao was asked what he thought about the French Revolution he famously replied that it was far too early to say. Perhaps the same should apply to the various approaches to tackling Covid.Surely it's a good thing to listen to the views of those with whom you would disagree, if only to sharpen your own beliefs and arguments? If you Google for - Hector Drummond podcast Delingpole - you will find a podcast interview with "Hector Drummond". I certainly found that he can't be dismissed as an eccentric.However it is always fascinating to find how a view on one subject tends to coalesce with other subjects and individuals such as climate change, Brexit, the Daily Mail, Dominic Cummings, Toby Young,Peter Hitchens, population control, anti-BBC etc.etc.

Jim Storrar ● 2059d

Watch the Sweden trends - and don't be distracted by the within week variation, I believe that's because they don't register deaths at weekends so there is a pattern of dip at the weekend and catch up midweek.https://www.worldometers.info/coronavirus/country/sweden/New Cases is, to my mind the best indicator of recent trends.  It looks like it has plateaued since mid April. Daily Deaths shows a decline since April week 3.

Andy Riley ● 2059d

Guy,You might like to reflect on this:"The only way in which a human being can make some approach to knowing the whole of a subject, is by hearing what can be said about it by persons of every variety of opinion, and studying all modes in which it can be looked at by every character of mind."John Stuart Mill, 1859.(I await your dismissal of Mill as a dangerous libertarian)

Andy Riley ● 2059d

Alex,The question of differing death rates between countries is hugely complex, encompassing demographics, social habits, pop densities, data standards etc etc.  Essentially, the headline figures we keep seeing in the mainstream media seem to be driven (in part) by a motive of denigrating Britain's performance (which is not to say it is good).  We won't know the actual comparable figure until studies have been done to normalize them and adjust to common standards.  So Sweden's figures reflect a different society across all of those factors above.  But the point of my original post was not this - it is that the Uppsala study gives us a comparison between the ICL model's predictions and reality.  So it tells us something about the model.My view is that is too early to draw firm conclusions about this - the Sweden death rate could potentially take off soon - but is fits into a pattern with other evidence which indicates the lock down policy as it stands is too severe.  Particularly when viewed along with the increasing risk of global economic melt down, social disorder and lock down induced health impacts.

Andy Riley ● 2059d

Guy,Forget Drummond, forget AIER read the paper and comment on that.And regarding the right strategy - if the consequences of the paper hold out, then we are most definitely following the wrong strategy based on a flawed model.  If you continue to support universal house arrest then the world economy will take even greater hits as time goes on and small/medium businesses fail.  Are you prepared to own the consequences?It's interesting how you dismiss ideas based on the label you attach to their advocates.  What does it matter if Drummond is a right wing libertarian - engage with his ideas and critique them.  It is precisely this sort of closed mindedness that has contributed to my drift from being a Labour supporter to now classifying myself as somewhere on the socially conservative spectrum.And did you do the political compass questionnaire? I did it a couple of years ago and found I was somewhat libertarian and somewhat left.  Would love to hear about your result.

Andy Riley ● 2059d

Thanks Andy.Unfortunately people too often prefer to go with the main stream, looking for promoted names rather than thinking and ignoring facts to the extent that people often mix science with religion. That article is a bit in the scientific researches that may offer a sort of explanation to some facts. For example:Why are the figures of cases and deaths per person in Sweden are so much better than in the UK?Why is that Nightingale hospital underutilised (or even dismantled?)?

Alex Shpinkov ● 2059d

I am very much open to argument as to whether we have been pursuing the right strategy and I'm much more interested in the opinions of reputable people like King than people like 'Drummond' who are pursuing a right wing libertarian agenda. He claims to be an ex-professor - but we can't establish if that's true because he hides his identity, and he could have been a professor of ancient mythology even if it is.More to the point is what is the right strategy now. It seems our strategy thus far has been pretty ineffective, but I think it would be highly dangerous to take the brakes sharply off now: in any case our economy (in Hounslow, the UK and the world) has already taken an enormous hit and recovery will be slow, at best.And we need to build a different world, very far from that which 'Drummond' or indeed the AIER would advocate.

Guy Lambert ● 2060d

Yesterday I whiled away almost 3 hours of self isolation by watching the first webcast of "The Alternative Sage". This is chaired by David King, the former Government Chief Scientific Adviser. I was spellbound by the discussion although I find that I am often spellbound (as well as housebound) during the lockdown.I can't find an online recording of the meeting but this video interview gives a flavour of some of the points of discussion:https://www.youtube.com/watch?v=azR2tW6tgYYI see that the list of Sage members has been published:https://www.gov.uk/government/publications/scientific-advisory-group-for-emergencies-sage-coronavirus-covid-19-response-membership/list-of-participants-of-sage-and-related-sub-groupsNo shortage of planet-sized brains but there are a lot of them. I trust that they hold virtual meetings - an awfully big room would be needed to maintain social distancing.

Jim Storrar ● 2060d

Did you actually read the paper?It matters not a jot whether Drummond is a pseudonym (lots of authors use pseudonyms) I used his blog to give a short summary with a link to the actual paper. I could have linked to the same story at AIER although I expect you would have scurried off to Google to find some reason to disbelieve them too. https://www.aier.org/article/imperial-college-model-applied-to-sweden-yields-preposterous-results/?utm_source=CapX+briefing&utm_campaign=b5527fa3d8-EMAIL_CAMPAIGN_2017_07_17_COPY_02&utm_medium=email&utm_term=0_b5017135a0-b5527fa3d8-241892017If you or others can critique the findings I will be interested to hear what you have to say.You can download the ICL code from here:https://github.com/ImperialCollegeLondon/covid19modelAnd your dismissive comment regarding him being a libertarian is a tad worrying.  What are you Guy? An authoritarian?  Libertarian covers a wide range of ideas within traditional left and right - see the Political Compass website to find out where you fit:https://en.wikipedia.org/wiki/The_Political_Compass"someone whose opinions are likely to trouble me too much." tells me something about the state of the Labour party today.

Andy Riley ● 2060d

A bit hard to get much info on supposedly 'ex-professor Hector Drummond' which appears to be a pseudonym. But it's clear he's a libertarian, therefore not someone whose opinions are likely to trouble me too much.

Guy Lambert ● 2060d