Quizzes & Puzzles21 mins ago
Why Is Covid-19 Less Survivable In The Uk?
According to a study by the Institute for Health Metrics and Evaluation (IHME) in Seattle released yesterday, the UK could be the European country worst hit by the coronavirus pandemic, suffering a death toll over 66,000 by August.
The IHME, which produces the Global Burden of Disease study, believes that deaths in the UK will peak with an estimated 2,932 deaths on 17 April, and predicts 66,314 total deaths in the country by 4 August.
In the EU, Italy, Spain and France are forecast to suffer the highest death tolls, at 20,200, 19,209 and 15,058 respectively by 4 August.
https:/
The IHME, which produces the Global Burden of Disease study, believes that deaths in the UK will peak with an estimated 2,932 deaths on 17 April, and predicts 66,314 total deaths in the country by 4 August.
In the EU, Italy, Spain and France are forecast to suffer the highest death tolls, at 20,200, 19,209 and 15,058 respectively by 4 August.
https:/
Answers
Best Answer
No best answer has yet been selected by Gromit. Once a best answer has been selected, it will be shown here.
For more on marking an answer as the "Best Answer", please visit our FAQ.
I don't know how you can reach that assessment, 10CS. The lockdown will certainly have had some impact, but during the first week or more everybody would have expected deaths to only increase due to those who caught it before the lockdown came into force. It's reasonable to suppose that, were we still behaving as before, the death rate would continue to rise until some of the more dire predictions -- those that put even IHME's pessimism to shame -- came true.
As it is, we might be thankful if deaths remain not much more than 20,000*, although I'm currently expecting us to log 900-1,000 deaths a day from/related to Covid for another week or so at least.
It will also be interesting to see what the next set of ONS figures for deaths reveals, as that should end up picking up deaths outside hospitals, which could add another 100-odd a day at its peak.
*If we assume that the crisis has already peaked, then the optimistic scenario is that total deaths will double by the end. So we're already looking at 20,000 deaths minimum now.
As it is, we might be thankful if deaths remain not much more than 20,000*, although I'm currently expecting us to log 900-1,000 deaths a day from/related to Covid for another week or so at least.
It will also be interesting to see what the next set of ONS figures for deaths reveals, as that should end up picking up deaths outside hospitals, which could add another 100-odd a day at its peak.
*If we assume that the crisis has already peaked, then the optimistic scenario is that total deaths will double by the end. So we're already looking at 20,000 deaths minimum now.
On April 8th (i.e. six days ago):
//Deaths in the UK are forecast to peak the third week of April, with an estimated 2,932 deaths on April 17//
Only three days to go. Last three day’s figures: 737; 717; 778.
//The projected figure for to-day is 1,233 (range 480 to 2,614) so we will see later to-day how close that is.//
It could hardly miss. Such predictions are ridiculous. I could produce such figures with no more than a pencil and paper. Here’s my prediction for UK deaths tomorrow:
803 – range 402 to 2493.
//…although I'm currently expecting us to log 900-1,000 deaths a day from/related to Covid for another week or so at least.//
That was Saturday, Jim. Since then:
Sun: 737
Mon: 717
Tue: 778
Close, but not close enough for the cigar.
I’m afraid that this thread demonstrates that mathematical modelling – whether undertaken by the amateur or the professional – is hopelessly inadequate to forecast the spread and mortality rate of a virus. Better to stick your finger in the air, declare it will be “a lot” then plan for the most you can possibly cope with. I spent a spell of my working life running multiple regression analysis models and their associated predictions. They were always inaccurate – sometimes disastrously so – mainly because they replaced forecasting the one thing that you had a handle on with forecasting a number of others about which you didn’t have a clue. Very often a finger in the air would have been just as useful and a lot less time consuming.
Personally I’m going to stick with what has actually happened. The predictions as far as I am concerned are bunkum.
//Deaths in the UK are forecast to peak the third week of April, with an estimated 2,932 deaths on April 17//
Only three days to go. Last three day’s figures: 737; 717; 778.
//The projected figure for to-day is 1,233 (range 480 to 2,614) so we will see later to-day how close that is.//
It could hardly miss. Such predictions are ridiculous. I could produce such figures with no more than a pencil and paper. Here’s my prediction for UK deaths tomorrow:
803 – range 402 to 2493.
//…although I'm currently expecting us to log 900-1,000 deaths a day from/related to Covid for another week or so at least.//
That was Saturday, Jim. Since then:
Sun: 737
Mon: 717
Tue: 778
Close, but not close enough for the cigar.
I’m afraid that this thread demonstrates that mathematical modelling – whether undertaken by the amateur or the professional – is hopelessly inadequate to forecast the spread and mortality rate of a virus. Better to stick your finger in the air, declare it will be “a lot” then plan for the most you can possibly cope with. I spent a spell of my working life running multiple regression analysis models and their associated predictions. They were always inaccurate – sometimes disastrously so – mainly because they replaced forecasting the one thing that you had a handle on with forecasting a number of others about which you didn’t have a clue. Very often a finger in the air would have been just as useful and a lot less time consuming.
Personally I’m going to stick with what has actually happened. The predictions as far as I am concerned are bunkum.
Such predictions are sometimes the best you can do. After all, how can anyone prepare for the worst that can happen if they can't quantify it?
The flaw in the model in this case is that it was based on too few data points to be taken seriously in the way that Gromit did, but the model itself isn't unreasonable. The curve the were showing has the correct expected behaviour, for example.
Meanwhile, yes, deaths in the last few days have been lower than I suggested, or at least official figures currently are. I'm fairly sure that's something to celebrate rather than mock, though. Moreover, doesn't it undermine the e tire tenor of your own earlier posts, which is that the lockdown has had no measurable effect? For daily deaths to apparently stall is encouraging and strong evidence that the approach we've taken is beginning to pay off.
Still, I'm not sure we're at the peak yet. Bank holiday weekend was expected, sadly, to lead to reporting delays. Hopefully in this too I am wrong, but there's a decent chance that we'll see an uptick later this week as the delayed announcements come in. And finally, as we saw in the ONS release, official figures may well be a significant underestimate of total deaths anyway.
The flaw in the model in this case is that it was based on too few data points to be taken seriously in the way that Gromit did, but the model itself isn't unreasonable. The curve the were showing has the correct expected behaviour, for example.
Meanwhile, yes, deaths in the last few days have been lower than I suggested, or at least official figures currently are. I'm fairly sure that's something to celebrate rather than mock, though. Moreover, doesn't it undermine the e tire tenor of your own earlier posts, which is that the lockdown has had no measurable effect? For daily deaths to apparently stall is encouraging and strong evidence that the approach we've taken is beginning to pay off.
Still, I'm not sure we're at the peak yet. Bank holiday weekend was expected, sadly, to lead to reporting delays. Hopefully in this too I am wrong, but there's a decent chance that we'll see an uptick later this week as the delayed announcements come in. And finally, as we saw in the ONS release, official figures may well be a significant underestimate of total deaths anyway.
And, finally, you are still drawing the wrong lessons. Mathematical modelling especially when it's not about the abstract but about reality is bound to be wrong, because the point about the prediction is that it's meant to be a warning of what to avoid, rather than a guarantee of what's inevitable. The Government was given projections that, without the imposed measures, the deaths in the UK could be well in excess of 100,000 and probably closer to half a million. Responding to that with an attitude that amounts to "modelling is all bunkum so I'm going to pay no attention to it" would be exactly the opposite response from the correct one. And what's even more troubling is that this is the response you always seem to express at any scientific warning based on sound scientific principles coupled with sound modelling for the future -- and you are always wrong to react this way. Yes, the error bar may sometimes be sadly huge, but the underlying message that "we have to respond to this and take the threat seriously" is what you're distracting yourself from, and any others who take your posts seriously.
//I'm fairly sure that's something to celebrate rather than mock, though.//
I'm not mocking the deaths, Jim. I wouldn't mock one, let alone hundreds. I'm mocking the notion that the progress of this pandemic can be forecast with even a modicum of accuracy. The figures I have quoted demonstrate that.
//Yes, the error bar may sometimes be sadly huge,//
Indeed. So huge as to about as useful as a finger in the air. For confirmation of this compare my prediction for tomorrow (which is a finger in the air and of a similar error range to those produced by the experts) when tomorrow's results are in.
//Moreover, doesn't it undermine the e tire tenor of your own earlier posts, which is that the lockdown has had no measurable effect?//
My earlier posts were commenting on the rate of new cases. I made no comment about the numbers of deaths. These continue to grow at about 8% per day (average for the last week). Of course we'll never know what the figure might have been without the lockdown. The government may well have predicted 100,000 to 500,000 deaths (nice wide range again). But they also predicted - as little as a week ago - almost three thousand deaths for next Friday and that's probably not going to be close.
I maintain my view that many aspects of the lockdown will see little benefits and I base this on Sweden. I know they have a much smaller population (about a sixth of the UK) and they have a lower density. But they have some large cities where population density is probably similar to large conurbations in the UK. But most significantly they have not closed schools, bars, restaurants or shops. Their new infections have not grown anything like ours despite beginning at around the same time and in fact being ahead of the UK for a few days in mid-March. Their infections per head of the population is lower than ours. If a lockdown here is said to have such a great beneficial effect then a lack of one there should have seen a rapid rise in new infections and it's been nowhere near as rapid as ours, so obviously other factors are in play.
I wouldn't dismiss the modelling if, as you suggest, it was sound. But it very clearly isn't. The forecasts vary wildly over short periods and simply react to the latest "actuals" which similarly vary wildly. That's no criticism of the modellers - it's a comment on the inadequacy of mathematical modelling to deal with a problem like this. It's rather like using a jewellers' screwdriver to hammer a fencepost into the ground. There are some jobs mathematical modelling cannot cope with and this is one of them.
I'm not mocking the deaths, Jim. I wouldn't mock one, let alone hundreds. I'm mocking the notion that the progress of this pandemic can be forecast with even a modicum of accuracy. The figures I have quoted demonstrate that.
//Yes, the error bar may sometimes be sadly huge,//
Indeed. So huge as to about as useful as a finger in the air. For confirmation of this compare my prediction for tomorrow (which is a finger in the air and of a similar error range to those produced by the experts) when tomorrow's results are in.
//Moreover, doesn't it undermine the e tire tenor of your own earlier posts, which is that the lockdown has had no measurable effect?//
My earlier posts were commenting on the rate of new cases. I made no comment about the numbers of deaths. These continue to grow at about 8% per day (average for the last week). Of course we'll never know what the figure might have been without the lockdown. The government may well have predicted 100,000 to 500,000 deaths (nice wide range again). But they also predicted - as little as a week ago - almost three thousand deaths for next Friday and that's probably not going to be close.
I maintain my view that many aspects of the lockdown will see little benefits and I base this on Sweden. I know they have a much smaller population (about a sixth of the UK) and they have a lower density. But they have some large cities where population density is probably similar to large conurbations in the UK. But most significantly they have not closed schools, bars, restaurants or shops. Their new infections have not grown anything like ours despite beginning at around the same time and in fact being ahead of the UK for a few days in mid-March. Their infections per head of the population is lower than ours. If a lockdown here is said to have such a great beneficial effect then a lack of one there should have seen a rapid rise in new infections and it's been nowhere near as rapid as ours, so obviously other factors are in play.
I wouldn't dismiss the modelling if, as you suggest, it was sound. But it very clearly isn't. The forecasts vary wildly over short periods and simply react to the latest "actuals" which similarly vary wildly. That's no criticism of the modellers - it's a comment on the inadequacy of mathematical modelling to deal with a problem like this. It's rather like using a jewellers' screwdriver to hammer a fencepost into the ground. There are some jobs mathematical modelling cannot cope with and this is one of them.
I've already commented on the flaws with the comparison to Sweden before, but another one worth noting is that maybe, if you compare to Norway -- in many ways a far more appropriate comparison -- it suddenly looks much worse. Four times higher rate of deaths per head, based on far less stringent response. And, again, the Prime Minister of Sweden has acknowledged that the approach there has been inadequate, it's been condemned by scientists both there and here, and indeed in the last few days it's introduced more stringent measures. So hardly a shining example on which to base a claim that the less aggressive approach is better than, or at least not significantly worse than, tougher social distancing measures.
Put bluntly, then, Sweden provides no evidence at all for your case. As to the modelling -- yes, there are deficiencies, but it's simply not reasonable to write off modelling altogether. It's a vital tool in trying to understand what can happen, and how to cope with it, and as long as it's understood properly where its strengths and weaknesses are then we benefit from trying to solve the impossible problem, rather than abandoning all hope of understanding it.
https:/
https:/
Put bluntly, then, Sweden provides no evidence at all for your case. As to the modelling -- yes, there are deficiencies, but it's simply not reasonable to write off modelling altogether. It's a vital tool in trying to understand what can happen, and how to cope with it, and as long as it's understood properly where its strengths and weaknesses are then we benefit from trying to solve the impossible problem, rather than abandoning all hope of understanding it.
https:/
https:/
The Countries we should be learning from, in terms of examples to follow rather than avoid, are in reality not Sweden (or, even more extremely, Brazil, where the worst is yet to come in part because Bolsonaro refuses to take the threat seriously), but:
1. South Korea: aggressive and widespread testing, contract tracing, and measures designed to ensure that the sick were isolated. Even when the disease threatened to explode, the country was able to get on top of it by taking the threat seriously and responding accordingly.
2. New Zealand: early commencement of extreme lockdown measures has turned something that threatened exponential growth into something already resembling a logistic curve and slowdown. Active cases has recently started to decrease.
3. Germany. Again, rapid rollout of widespread testing has enabled the disease to be at least somewhat controlled. Sadly it came too late to stop the disease spreading widely, but at least those who were sick were more quickly identified, and active cases are now decreasing. Possibly the closest comparison to the UK, and sadly is doing four times better where it counts.
In practice this means testing is even more effective than social distancing measures, since those can only be maximally effective if the sick are isolated, or know to self-isolate, whereas if the population is still somewhat free to mingle without being aware of who has or hasn't got it then the disease can still continue to spread, albeit slower than it would have. But still, the message the world is sending is that social distancing and lockdown measures have an impact.
1. South Korea: aggressive and widespread testing, contract tracing, and measures designed to ensure that the sick were isolated. Even when the disease threatened to explode, the country was able to get on top of it by taking the threat seriously and responding accordingly.
2. New Zealand: early commencement of extreme lockdown measures has turned something that threatened exponential growth into something already resembling a logistic curve and slowdown. Active cases has recently started to decrease.
3. Germany. Again, rapid rollout of widespread testing has enabled the disease to be at least somewhat controlled. Sadly it came too late to stop the disease spreading widely, but at least those who were sick were more quickly identified, and active cases are now decreasing. Possibly the closest comparison to the UK, and sadly is doing four times better where it counts.
In practice this means testing is even more effective than social distancing measures, since those can only be maximally effective if the sick are isolated, or know to self-isolate, whereas if the population is still somewhat free to mingle without being aware of who has or hasn't got it then the disease can still continue to spread, albeit slower than it would have. But still, the message the world is sending is that social distancing and lockdown measures have an impact.
For those who think the lockdown is overkill and should be lifted/scaled back considerably, the figures will always convince them they are right. If deaths are high despite the lockdown they can say it's not having the desired effect so should be withdrawn/scaled right back. If deaths are low they can say it's all been hype and overkill- all this pain for a few hundred deaths that might have happened anyway.
This problem arises because we are comparing actual deaths with forecasts of what they would have been- and no-one will ever know how accurate the latter figure is.
Overall I tend to agree with jim360 on most of the Covid issues, and it does seem in line with what most major governments are doing, but we'll not get a better idea for some time. We may never know who was right until we relax the lockdown and see what happens- but getting that wrong by doing it too quickly too soon and seeing a huge spike and ongoing increased levels is a risk no responsible government would take (I don't think Labour would do things much differently either)
This problem arises because we are comparing actual deaths with forecasts of what they would have been- and no-one will ever know how accurate the latter figure is.
Overall I tend to agree with jim360 on most of the Covid issues, and it does seem in line with what most major governments are doing, but we'll not get a better idea for some time. We may never know who was right until we relax the lockdown and see what happens- but getting that wrong by doing it too quickly too soon and seeing a huge spike and ongoing increased levels is a risk no responsible government would take (I don't think Labour would do things much differently either)
Latest Posts
Quizzes & Puzzles31 mins ago
Quizzes & Puzzles48 mins ago
Strands #262 “For Our Furry...
Quizzes & Puzzles128 mins ago
Is There An Option To Turn Of Reader
TV161 mins ago
ChatterBank169 mins ago
Related Questions
Sorry, we can't find any related questions. Try using the search bar at the top of the page to search for some keywords, or choose a topic and submit your own question.