← More or Less
More or Less
Deaths at Home, Supermarket Infections and the Cobra Effect
00:28:24
Where are people catching Covid-19?
Loading summary
Transcript80 lines
- [00:02]
A
BBC Sounds Music Radio Podcasts.
- [00:06]
B
Hello, and welcome to More Or Less. We're the programme that tries to make your world add up without taking anything from the multiplying difficulties we face in these divided times. I'm Tim Harford. This week, are people more likely to die at home rather than in hospital these days? And is that good news or bad news? David Spiegelhalter is on the case. The government has a cunning plan to encourage people to self isolate. Is there a risk that it's a bit too cunning? Cobras and feral hogs may be involved. The UK's in lockdown, but tens of thousands of people a day are still testing positive for coronavirus. Where are they catching it? And grim data on drug deaths in Scotland have been called into question. On social media, we ferret out the truth. But first, Covid case numbers are still falling. The number of patients being admitted to hospitals also seems to have started slowly falling. But deaths remain high. On 20 January, headlines said that the UK had reported 1,820 deaths. We've mentioned many times on More or Less that reported deaths are not the same as deaths which have taken place that day. But that didn't stop many people claiming that 1,820 people had died that day. For example, Professor Alice Roberts at the University of Birmingham tweeted that 1820 people died today and that we should all be angry. And nearly 10,000 people were angry. Angry enough to retweet anyway. But as I keep trying to tell people, angry retweets don't necessarily help us think more clearly about the world. And while the death toll is appallingly high, it. It's not as high as the headlines and the tweets claim. David Spiegelhalter is the chair of the Winton Centre for Risk and Evidence Communication at Cambridge University.
- [02:05]
C
I find these headlines very distressing. I mean, we have been speaking since the start of the pandemic in March about the dangers of reporting the number of deaths or even the number of cases because of the reporting delays. And so these numbers depend so much on the day of the week. For example, Wednesday is almost always a high day, and last Wednesday was reported 1820 deaths, which got a lot of headlines. Now, so far we know that actually at least 1028 people died on that day. As more cases are reported, it may go up to about 1100, but nowhere near 1820. And so, you know, all these headlines, I feel, were deeply, deeply misleading.
- [02:51]
B
But I realized that the statistical pedant in me wants people to be precise and if Wednesday was when the deaths were reported, that is not necessarily when the deaths occurred, but overall, is there any reason to believe that deaths are being systematically misstated or just that the media is being a bit vague when it should be precise?
- [03:12]
C
In the end, the total numbers will match up. It's just that the headline figures on particular days receive such attention, and days on which far fewer Deaths are reported, for example, yesterday was 600 or so. That doesn't get any coverage at all. And so I do feel that these headlines are misleading. The numbers are high enough and bad enough anyway without making inappropriate comparisons.
- [03:36]
B
So what is the situation now, compared with the first wave? It seems to be even worse than the first wave in terms of deaths. Is that the case?
- [03:45]
C
Not in terms of the peak. The problem is that the number that's reported and is on the main dashboard is using this criterion of a death of somebody who once tested positive and then died within 28 days of their positive test back in April, in the first wave, there wasn't anything like the same amount of testing that was done. And so there's a systematic undercount in, in the actual number of deaths. So, for example, the peak in the first wave, according to the dashboard, is 1,073 who died on 8 April. But in fact, if we look at ONS data about who actually died on that day and then was registered as a death with COVID is 1,456, you know, much, much higher. And our current daily death rate is not approaching that. I don't believe we'll get near that figure. And so actually all those comparisons by saying, oh, we're higher than the peak of the first wave, also are incorrect.
- [04:47]
B
We've been focusing on COVID deaths, but of course Covid can kill people indirectly by making it harder or more frightening to get help for other conditions, physical or mental. Some of the people highlighting this risk have been arguing against lockdowns. But similar warnings have come from the heart of the British medical establishment. Here, for example, is England's chief medical officer, Chris Whitty. People die for many reasons in epidemics. They include the direct cause of death, but they can also include people who died, for example, if they stayed at home because they were worried that going into hospital was dangerous for them, and actually they have a heart attack or a stroke or something important like that. Clearly that's a risk in theory. But do we have any evidence? David Spiegelhalter thinks we do. He's been combing through the data which show where deaths have taken place from
- [05:40]
C
the start of the pandemic up to January 8. In England and Wales, there are 42,000 excess deaths registered in homes and only 4,000 of these were Covid. And it actually matches very closely what happened in hospitals. There were 42,000 fewer non COVID deaths than normal. And what this indicates is that a huge migration of deaths from hospitals to homes.
- [06:06]
B
Those numbers are from the UK's Office for National Statistics, but we have more detailed data from Scotland which breaks down deaths both by location and cause. And that gives us a greater insight into what's been going on.
- [06:19]
C
So they revealed that In Scotland in 2020, there were about 7,000 extra deaths and about 6,000 COVID deaths. So, you know, roughly matching. But overall in Scotland over the year, there were fewer deaths than normal in hospitals. And so what happened? Well, it's the same as England and Wales, a huge increase in the number of deaths at home. But because of this data, we can see what were these extra deaths. And in fact, in Scotland they had 1900 extra cancer deaths at home and about 1200 extra heart disease and stroke deaths at home.
- [06:56]
B
So we don't think that that is an increase in non COVID deaths, or at least not a large increase in non COVID deaths. We think it is a shift in the location of deaths that probably would have happened anyway.
- [07:06]
C
Yes, over the whole of Scotland, cancer and heart attack and stroke deaths are not raised very much. So it suggests that most of these people actually would have died anyway, but they would have died in hospital rather than at home. And so that really raises the question, what has been the quality, the end of life care?
- [07:27]
B
Yes. I mean, the idea of dying of cancer at home, surrounded by family, rather than dying in a hospital doesn't sound too bad. The idea of dying at home of a heart attack sounds really bad. Although maybe I've, you know, I'm just imagining something that doesn't reflect reality. Do we know anything about the quality of these deaths and whether we should be worried or perfectly content with the shift in location?
- [07:53]
C
I mean, those of us, including myself, who have experienced the deaths of somebody from cancer at home know that it can be as good an experience, I think, as one could hope for. But as you said, heart attacks and strokes, which are much more sudden and less, in a way, planned for, could be a very different matter. It would be very good to know much more about what's been going on in Scotland and of course over the whole country, because there's been a similar pattern across the whole country.
- [08:19]
B
And I'm Assuming the reason we're seeing, what we're seeing is that people are either not going to hospital because they're afraid to go to hospital because of COVID or the ambulances are too busy to take them to hospital or the hospital won't take them for one reason or another. It is Covid that is causing this shift.
- [08:40]
C
One can assume so, but this pattern occurred right over the summer when in fact there was not a huge burden of the epidemic in the country. And so we can I think, assume this is not just due to lack of ambulances and so on. People could be being sent home to die, which again this would be a very positive move because most people would prefer to die at home. And so I just think it's something which the statistics raises but cannot answer all these questions.
- [09:08]
B
Professor Sir David Spiegelhalter, You're listening to More or less on BBC Radio 4 with me, Tim Harford. The UK government has been trying to solve a problem.
- [09:24]
A
So there's growing evidence that most people don't completely follow the self isolation rules at the moment. So for example, a government backed study from September 2020 found that only 11% of people remained indoors for the full time that they were required to quarantine.
- [09:44]
B
This is Abby Adams Prasl, a professor of economics at Oxford University. And people not adhering to quarantine rules is at least in part because of economics. If you're a freelancer or you're on a zero hours contract, then you will take a serious financial hit. If you miss work and if you've tested positive but you're feeling ok, then we're asking you to bear that cost for the sake of everyone else. So perhaps the rest of us should be chipping in to help. And indeed, last year the government introduced a 500 pound payment for those on universal credit or other benefits who are asked to self isolate and couldn't work from home. Those payments are also available in some other circumstances. But then last week the Guardian published leaked notes suggesting a new policy extending that £500 payment to everyone who was asked to self isolate, whether they were claiming benefits or not. But as an economist, it made me think of something we call the cobra effect.
- [10:48]
A
The cobra effect is a term that basically refers to a scenario where the incentives designed to solve a problem not only don't work, but they actually backfire completely and end up making the whole situation worse than it was to start with. The story goes that in colonial India, Delhi was suffering from this proliferation of venomous snakes. And so the government placed A bounty on the snakes. So, you know, if you turned up with a dead cobra, you would get paid for it. Now, this seemed like a good solution. You're incentivizing people to go out and kill the snakes. And initially it seemed successful, but actually what ended up happening was as the population of cobras started to fall and so it became more difficult to go out and catch a cobra and get the bounty, some entrepreneurial citizens decided to start breeding cobras so that they could derive an income from going out and collecting those bounties. So when the government then realized what was going on, they decided to completely scrap the scheme. But then that meant that, of course, completely reasonably as well, all of these new cobra breeders decided to release all their cobras onto the streets. And what you ended up with was an even bigger problem with cobras than you had to start with. This might seem like a crazy story, but something very similar happened in around 2007 in the US with wild pigs in Georgia.
- [12:22]
B
You have my attention. Tell me about the wild pigs. I've not heard this story.
- [12:26]
A
Well, in Georgia they did the same thing. So they were having this issue with all these wild pigs, and so they started paying people to go out and shoot the pigs. Except for what happened was people started actually going into neighbouring states, buying pigs from pig farms in collecting bounties in Georgia for pigs which were never in Georgia to start with.
- [12:49]
B
Each time the cobra effect comes into play, it seems a government is trying to solve a problem by rewarding something subtly different. Trying to reduce the number of cobras by paying for dead cobras, or trying to encourage people to stay at home by paying for a positive test. And some commentators have pointed out that there is a logical possibility that some people may take less care if they know there's a financial reward for testing positive.
- [13:17]
A
Certain individuals who might face a relatively low risk from COVID 19 younger people, for example, actually, potentially their perceived cost of going out and catching the virus might seem relatively low compared to that £500 payment. So, I mean, really, the thing with the cobra effect in this context, I think it depends on which effect you think dominates the positive effect that incentivizes sick people, say, at home, or the negative effect that incentivizes healthy people to go out and catch the virus.
- [13:52]
B
I mean, do we have any evidence on the balance of risks there on this positive side?
- [13:57]
A
There have been some studies which have looked at changes in eligibility for sick pay. And the evidence there would suggest that more generous sick pay does result in more people claiming those sick pay days, but also that it does stop people from going back to work too early, especially for people with infectious diseases.
- [14:20]
D
On the.
- [14:21]
A
To what extent is there, however, this potential for people who perhaps might be in lower risk groups to take less care? I haven't actually seen any empirical evidence, to be honest, Tim. I'd probably say that the weight of evidence is more suggestive of there being a net positive impact.
- [14:40]
B
Is there a way that we might reform the payment to keep all the positive aspects but to remove the COBRA effect possibility? Or is that just. That would just in itself become far too complicated?
- [14:55]
A
It's a really good question and I've been thinking about it like all morning, trying to think like, how would you reform this? Because you kind of want to target how much care someone is taking in their day to day life. But of course that's fundamentally unmeasurable. Perhaps there is this space for a middle ground with some targeting whereby you can perhaps exclude. Exclude teenagers, for example, but still be able to actually get the benefits to those people who actually are really on the margin of compliance because of these poor financial incentives.
- [15:29]
B
Professor Abby Adams, Prasl. You're listening to More or less on BBC Radio 4 with me, Tim Harford. We've had many, many people email More or less@BBC.co.uk saying you enjoyed our sea shanty last week. There once was a woman, like many
- [15:46]
C
MPs, thought she understood our fisheries.
- [15:50]
B
Her name was Penny. So, coming soon from More or Less Enterprises, a compact disc the whole family can enjoy. The More or Less team sing mathematically correct Sea shanties featuring four squared men on a dead man's chest.
- [16:11]
C
Yo ho ho.
- [16:12]
B
That's an incorrect sum. And a never heard before special duet between me, Tim Harford and everyone's favourite number night, Sir David Spiegelhalter.
- [16:22]
C
What shall we do with the rounding error?
- [16:25]
B
What shall we do with the rounding error? What shall we do with the rounding error early in the morning? Order now to receive a free bonus disc of David Spiegelhalter reading vaccine delivery statistics in a slow and calming voice.
- [16:45]
C
21st of January, 409,855. 22nd of January, 478,248. 23rd of January, 491,000.
- [17:02]
B
One of the biggest questions of the pandemic is where are people catching the virus? This MP thinks she knows the answer.
- [17:10]
A
I don't know why the government haven't been harder on the supermarkets. Because back in November we knew that the top cause of catching coronavirus was in our supermarkets.
- [17:20]
B
This is Labour MP Siobhan McDonagh on the BBC's Politics Live recently.
- [17:25]
A
And if they want to issue some £20,000 fines, I think they'd probably start with our major supermarkets.
- [17:32]
B
Well, this is an alarming notion. For those of us lucky enough to be able to work from home through lockdown, a supermarket may be one of the only places that you're indoors with a bunch of strangers. That's certainly true for me. And I always thought that wearing a mask, washing your hands was probably fairly safe. Am I wrong? A man's got to eat. And I'm not the only one to be concerned. Several loyal listeners have emailed more or less@BBC.co.uk with a particular statistic they want us to investigate, which is that 20% of people with COVID 19 had recently been to a supermarket. And Charlotte MacDonald is here to walk us through this. Hello, Charlotte.
- [18:12]
E
Hi, Tim.
- [18:12]
B
Is it true?
- [18:13]
E
Well, this 20% figure appeared in news reports back in November, but I'm afraid it was based on a misunderstanding of some data put out by Public Health England. Let's suppose for a moment, Tim, that you have tested positive for COVID 19.
- [18:26]
B
Okay.
- [18:26]
E
I am supposing your details get sent to NHS Test and Trace and then they contact you to find out where you've been and who you've been hanging out with. If you've been to the supermarket recently, you'd obviously tell them.
- [18:38]
B
Obviously I would.
- [18:40]
E
Now, there are a handful of tables put out weekly by Public Health England which list the locations that people with confirmed Covid tests have said they went to in the days running up to the test places, such as GP surgeries, public transport and supermarkets. But the tables don't claim to be proof of where someone caught the virus, they just list the types of places people have been frequenting.
- [19:01]
B
I suppose it isn't really surprising that someone with COVID has been to a supermarket, because lots of people go to supermarkets. So not only do we not know that there's a causal link, we don't even know there's a correlation. Maybe people without Covid are more likely to go to supermarkets.
- [19:16]
D
Yeah.
- [19:16]
E
Although the journalistic confusion goes deeper because the original source of this claim was a table which analysed a quarter of the cases in detail. Now, these were instances where two people had gone to the same place and also both developed Covid.
- [19:30]
B
So if you've got two people who went to the same place and they also both got Covid, you might say, ah, there's a possibility that that's where the transmission occurred. We should look into it.
- [19:42]
E
Yeah. But if you're talking about a town with a big supermarket and two people in the town, both got Covid and both went to the supermarket, you've not really learned anything. And that's what this table was showing. Among the locations where several people with COVID visited, supermarkets made up 20% of them.
- [19:57]
B
Wow. So we've really got ourselves tangled up here. So, of the quarter of the confirmed cases who'd been to a place where another confirmed case had been 20%, or 1/5 of those locations were a supermarket.
- [20:11]
E
Correct. So basically the 20% figure was lifted from a table and misinterpreted. There isn't a table which gives the proportion of people who went to a supermarket in the run up to testing positive for COVID 19. And even if there was, that wouldn't tell us that they caught Covid at supermarket. We just don't know that.
- [20:29]
B
So what do we know? It would be really useful to know where people are catching this illness, wouldn't it?
- [20:34]
E
It would, but it's not easy to do that for such a sneaky virus. That's why the SARS CoV2 virus is hard to control and explains why the disease has spread so widely and caused so much suffering. I spoke to Hayley Thompson from Imperial College London, who co authored a meta analysis of contact tracing studies from around the world.
- [20:53]
D
These studies identify and test all contacts of known positive COVID 19 cases to determine the proportion go on to become infected. And while it's difficult to pinpoint precisely where and when someone caught an infection, pooling this contact tracing data together allows us to gain insights into the settings that are potentially higher risk for onwards transmission. They identify a known case and trace all their contacts. Once they found their contacts, they then test all of these contacts to know the proportion that go on to become infected. And they test them at multiple time points for around up to 14 days. We found that prolonged household contact and social contacts with family and friends had the highest risk for onwards transmission compared to other settings such as travel or healthcare or casual close contacts. Unfortunately, the data surrounding transmission potential in specific locations such as supermarkets, gyms and workplaces and was really limited.
- [21:54]
E
Hayley says more data is needed on cases connected to specific locations to be able to say much about whether the virus was being spread there.
- [22:02]
B
I guess one of the biggest problems is knowing for sure who you got the virus from. For example, clusters might be identified as a workplace, but maybe it didn't spread there. It might have been workers traveling on the same public transport network or living in the same community or possibly visiting the same supermarket. It must be very hard to prove when and how someone caught it.
- [22:22]
E
Exactly. You'd have to do a full genomic sequencing of each confirmed case to know if they'd caught the same version of the virus. The fact is, we just don't have enough data to track people and their interactions. So I'm afraid, Tim, we just don't know.
- [22:36]
B
Thank you, Charlotte. Scotland's statistics agency, the National Records of Scotland, recently published its latest data on drug related deaths. They don't make for cheery reading. The number of deaths has more than doubled over the past decade, up 132%. The most recent figure, 1,264 deaths in 2019, is 6% higher than the previous year, the highest number since records began 25 years ago and relative to population, higher than any other European country. Many commentators criticised the Scottish government for not doing enough, but on social media a number of posts suggested the numbers were not what they seemed the truth about Scottish drug Scotland is one of
- [23:29]
C
the very few nations to include all
- [23:31]
B
deaths where all types of drugs are found after death. The posts claimed Scotland's figures look high because their definition of a drug death is particularly broad and includes a number of deaths other countries wouldn't count. For example, one post states that people who died in a car crash would be included if they had any type of drug in their system, whether or not that drug contributed to their death. Is that true?
- [23:57]
F
No, that's not right. Car crashes are explicitly excluded from this definition.
- [24:03]
B
This is Alastair Bryan from the Scottish fact checking website the Ferret Fact Service. Alistair has checked a list of these claims.
- [24:11]
F
Other things which are not included include AIDS related deaths where the risk factor was believed to be needle sharing, for example, or deaths from drowning under the influence of drugs, falls, other accidents and even deaths due to assault by somebody who was under the influence of drugs. These are all things which are not included in statistics.
- [24:29]
B
Not only are these things not included in the Scottish drug death figures, they're explicitly listed in the report from the National Records of Scotland and as not being included. Alastair thinks someone misinterpreted that list. Instead, these figures include things such as accidental or intentional overdoses. So are there any differences in the way Scottish statistics are collected?
- [24:53]
F
There's a slightly different way in which Scottish statistics and UK statistics are collected, so the Scottish statistics might be slightly more up to date. They might be comparing the whole of a year with the midpoint of the year and the midpoint of the next year, for example, and there's one example which is used that if a person in Scotland overdoses on illegal substance but has illegal drugs in their system, that would be counted in our statistics but wouldn't be counted in English statistics. The report there says that that accounts for about three deaths on average per year. So wouldn't greatly affect any comparison between
- [25:26]
B
the UK and Scotland, nor any comparison between Scotland and the rest of Europe. Because when you look at a table of drug related deaths in European countries, Scotland's figure relative to population is by far the biggest.
- [25:39]
F
So if you look at Scotland's drug deaths per million population using 2018 statistics, which are the comparable statistics with the majority of European countries, Scotland's sitting on 295 deaths per million population. The nearest highest is Sweden, which sits at 81. The UK is sitting at 76 drug deaths per million and the European Union's is 24. So Scotland is far higher than any of the other European countries which have sort of comparable data.
- [26:09]
B
Scotland's rate is three times higher than Sweden. In second place, it's nearly four times higher than the UK as a whole, it's more than 10 times higher than the European average. This isn't the first time claims that Scotland's drug related deaths were artificially inflated have been doing the rounds. They were circulating last July too. Alastair says there is a political dimension to it.
- [26:33]
F
There are definite advantages for some supporters of Scottish independence to try and make claims that when a situation occurs where the Scottish record might be unfavourably compared to the uk, then claims like this might be used in order to kind of offer some level of defence for the Scottish government.
- [26:51]
B
In this case, attempting to explain away the large number of drug related deaths in Scotland attempts to paint a picture that the problem isn't as bad as it looks. So there may be some ways Scottish statistics include more deaths than other countries in their drug related figures, but they don't change the big picture, which is worrying. Drug related deaths in Scotland have risen to their highest since records began in 1996. And the drug related death rate is very high when compared to other European countries, thanks to Alistair Bryan of the Ferret Fact Service. If you have a question for us, please email more or lessbc.co.uk, but that is all we have time for. We will be back next week. Until then, goodbye. More or Less was produced by Kate Lamble with Nathan Gower, Chloe Hajimothayou and Charlotte Macdonald. The programme was mixed by James Beard. Our editor is Richard Varden. And I'm Tim Harford. A new podcast series from BBC Radio 4.
- [27:54]
C
In the first stage of a poltergeist haunting, the entity will confine itself to making noise as if it's testing its victims. The Battersea Poltergeist.
- [28:08]
A
My name is Shirley Hitchens. I'm 15 years old. I live with my mum, dad, brother Granite and Donald.
- [28:19]
B
Subscribe to the Battersea Poltergeist on BBC Sounds.