More or Less: Behind the Stats - Covid plasma therapy

Donald Trump says allowing the emergency use of blood plasma therapy for coronavirus patients will save “countless lives” and is “proven to reduce mortality by 35%”. We look at the evidence. Amid talk of coronavirus being back on the rise in the UK, what does the data show? Could screening for breast cancer from the age of 40 save lives? And can it really be true than one in five women in 18th century London made a living selling sex?

  • Play Speed:
Content Keywords: cancer doctor number people Donald Trump nothing
00:00:00
This is the BBC.

00:00:03
This podcast is supported by advertising outside the UK.

00:00:09
sounds music radio podcasts

00:00:13
hello and welcome to more or less to show the tracks down suspect statistics traces who's been using them and isolates them from the solid majority of data points, which a healthy and reliable. I'm Tim Harford this week Grand claims have been made about treating kovats uppers with a blood plasma. If kovats aivas, we pick the scraps of evidence out of a steaming heap of exaggerations. If the virus spreading again in the UK because the data the BBC says early breast cancer screening saves lives. Does it take to investigate an astonishing Claim about the number of sex workers in Georgian London before the end of the program?

00:00:56
Fruitfast lost Saturday US President Donald Trump took to Twitter to complain about how the country's Food and Drug Administration the body that authorizes drugs and Medical Treatments was part of what the president called The Deep state which is not a compliment. He added that the FDA was making it difficult for drug companies to test vaccines and Therapeutics. He wrote they are hoping to delay the onset until after November 3rd the date of the presidential election. He tagged the commissioner of the FDA Stephen hung in his tweet as if by Magic the next day there was a press conference at the White House Donald Trump appearing alongside. Dr. Hahn today. I'm pleased to make a truly historic announcement in our battle against the John Tavares that will save countless lives. The FDA has issued an emergency use

00:01:56
Authorization and that's such a powerful term emergency use authorization for a treatment known as convalescent plasma blood plasma talk about how if your blood volume enough to fighting off a virus. It should contain the antibodies that we used in that fight convalescent plasma therapy uses donated plasma from recovered patients Street. Someone else who sick with the same virus to help them fight it off. It is not you from it. It's been used for over 100 years to treat illnesses from influenza to Ebola and sauce and according to Donald Trump. It is very effective in treating the virus that causes covid-19 over 100,000 Americans have already enrolled to receive this treatment and it is proven to reduce mortality by 35%

00:02:50
It's a tremendous number a tremendous number indeed if true and it wasn't just the president using it. He is Stephen Hahn. So let me just put this in perspective many of you know, I was a cancer doctor before I became FDA commissioner in a 35% Improvement in survival is a pretty substantial of clinical benefit what that means is a van in if the data continue to pan out a hundred people who are sick with covid-19, 35 would have been saved because of the ministration of a plasma that is a huge claim tools who makes no sense generally have a hundred people sick with covid-19 e expects wrong person to die. How could the deaths of 35 people be prevented then but even if we give them to him the benefit of the doubt and assume he meant to say out of a hundred people who might have died from

00:03:50
I'm the evidence about the effectiveness of plasma originates from a study by researchers at the Mayo Clinic. That study is only published as a preprint, which means it hasn't yet been peer-reviewed one of the main office at the study and said he doesn't know whether 35% number came from anyway to try to get to the bottom of this. I spoke to talk to Charlotte Summers cheese and intensive care doctor on the respiratory research at the University of Cambridge has made this quite a few things wrong with them thing is that actually therapy is notes therapy Vestige. No therapy in this study. So actually they have a go to study or any Deja from this study that supports the given convalescent plasma is better than not giving complex in plasma cuz there's no control breaking that study. That means it's impossible to estimate the effect of receiving this treatment compared to not receiving it instead they compared those

00:04:50
To receive plasma with a low level of antibodies with those who receive plasma with a higher level of antibodies. The next problem is that the patients would not randomly assigned to receive the high or low antibody plasma. It's an observational study which simply means the researches observe the results of decisions made on a case-by-case basis. If I don't trust these could easily by us the results because doctors will tend to treat people who I feel are most likely to benefit benefit its demise and there is no control group. So where did the FDA on President Trump's claim of a 35% reduction in mortality come from comes from the fact that in the 354 Hugo giving plasma with a high antibody content at 7 Days 8.9% of 5.

00:05:50
Where was people he got giving plasma but had a low antibody content 13.7% of them dies now 8.9 is 35% less than 13.7. But that's a relative risk reduction 8% You have a hundred people receiving plasma with a low level of antibodies you'd expect about fourteen of them to die. And if they were receiving plasma with a high level of antibodies it expects around 9 if them so although 9 is 35% less than 14. Does not mean that 35 lives were saved rather five more people survive longer than 7 Days among those receiving the high gray plasma compared to those cutting the week is tough. What's more all of this is based on people who was still alive after 7 days. They also look up

00:06:50
Who's still alive after 30 days and the difference is smaller, but the problem isn't really the size of the alleged fact, which is certainly big enough to celebrate. The problem is that there are so many reasons starting with the fact that the trial wasn't randomized to think that we don't yet have any good evidence that plasma works at all as a treatment The Cove it?

00:07:50
I meant he tweeted I have been criticized for remarks. I made Sunday night about the benefits of convalescent plasma. The criticism is entirely Justified. What I should have said is not the day to show a relative risk reduction. Not an absolute risk reduction doesn't mean that convalescent plasma doesn't work. It just means that we don't know yet and there are attempts to look at base in a more rigorous way. For example, the recovery trial in the UK has a control group and it's randomized and it hopes to provide Samantha's by the end of the year suicide attempts to when the data are Income a solid study that we will see a benefit like 35 lives saved out of a hundred for the reason that at 3:35. 101 saved in this study and randomized controlled trials using have similar effects on Isis then as he didn't observational studies like this one.

00:08:50
The chances of us having fatty Family Feud full safe every hundred I pretty much non-existent. My patients have been dying of Saul's kvsi having it that to therapy is important to me an important to my colleagues. But this noise therapy that doesn't have downsides. I'm so knowing that the therapy sway giving twelve patients have got benefits. I'm not just rest is really important and at the moment we don't know that yet talked to Charlotte Summers of the University of Cambridge you listening to more or less.

00:09:29
Some European countries note of this pain of seeing a distinct rise in cases. It's the same thing happening here in the UK. It's not completely straightforward to work as we discussed a couple of weeks ago at the rising number of confirmed cases might be because of a rising number of tests or the back to focusing of tests in areas without brakes or of course, the rising cases might reflect a warrior in reality. So how worried should we be long-suffering friend of the show Catherine mcconway Emeritus professor of Applied statistics at the open university has been digging into the stunts. There are three key source is published each Friday and Kevin began with a survey conducted by the office for National statistics. This is a big National survey. It takes a pretty large sample. Now of people from what they call the community population in England, they're not testing people because they go to symptom. They just sampled.

00:10:29
That kind of representative of the population so biases that might come from looking at people have. Symptoms don't to rise and that's very important. It's not being driven. For example by was a spike a case is in Leicester. So we're going to do. Smell to testing in Leicester and we'll find lots of cases because we're looking for the cases where we expect to find them. But the disadvantage is that we're actually pretty small number of cases in absolute terms. They estimate that's in a hundred thousand people pulling Fur Day in England that beat just over for new infections and the kind of the whole population and say that's a by 2400 new infections, but day they could be off that could be 1200 could be as big as 4,000 today because of the sampling uncertainty that sounds like quite a lot of people compared to the millions of people in the country. It's really not very many the advantages it removes all the bias the disadvantage.

00:11:29
Is the required wide are about it. What is that telling us about the level of infections and whether they going up or down level of people infected and also the level of new infections those increases do seem to be leveling off. They can't be absolutely certain that they are increasing their not increasing fast now reviews about test in hospitals that posting test says mobile testing centers a little leaf places where people are being tested for various reasons many many more tests, but not not a random sample. What do we learn from those tests? I don't want to take tested because they think they go to symptoms. Somebody else thinks they go to a symptom. They've been reported through the test and Trace operation and so on.

00:12:29
Unfaithful it's the way the people are brought in by those procedures changes over time and Holbrook positive tests and might be also changed over time without necessarily being a real increase the number of cases. It's just we Mayweather becoming better at finding them. If you don't get the test and traced reports again for England, they do show a kind of slow increased. It's still going on, but roughly speaking than no a long way out of line with what's coming up in the ons survey and my feeling is that we can conclude from that that there is a national increase going on it again is consistent with things being roughly level several pictures and it'll answer you go to Kanto definition was all kinds of different problems with the data, but there are a lot of different types of services and my arrow pointing towards the idea that

00:13:28
The current infection rate is pretty low and almost threatening Us increasing quickly. Take it easy. This won't they just do it? It's a results of the official pronouncements about the famous on number also come out once a week. The one that came out last week on Friday is on the face of it the scariest one because what they do which week is they give a range of values for the honor and remember that are number one means that it's kind of like a critical value if it's below one than overall. The number of cases is falling if it's above warmth and overall the number of cases increasing. So once again during the week before it was there to warn this week they put up that rain so it's no no point nine-two 1.1.

00:14:28
How do you say just now saying that they sent me, be confident that it's below 1 however, it is certainly nothing that says it is definitely going off just because they might have gone up it might not have gone up and those numbers are pictures but to interpret because I also have time lapse involved along the calculations are based on how many people are going to the hospital and numbers of deaths statistician Professor Karen McKeown. My wife is arising in some European countries, but apparently not in the UK is an important question. It's just a matter of time maps not I'm very sure we will be revisiting the question before this series is done. If you have a question or comment for us, please email more or less BBC. Co. Uk

00:15:19
The BBC recently reported that breast screening women in their forties could save lives routine breast cancer screening is already offered women in the UK between the ages of 50 and 70. The question is what happens if you offer it to women 10 years earlier from age 40 the onset came from a search published in the Lancet based on a study called the UK age trial this trial and rolled over a hundred and sixty thousand women between 1990 and 1997 and followed them for an average of 23 years some of them but given screening every year throughout my forties others chosen at random receipts screening at 50. Is it standard practice? It's a large-scale and well-designed randomized control trial, but did it really show that screening women in their forties saves lives and if so how many lives I also professor of Public Health

00:16:19
University of Sydney Alexandra berent, I think that as the lawyers say reasonable Minds could come to different conclusions about this so strictly speaking the child doesn't provide evidence separate screening saves lives because although in some analyses that showed that it reduces the risk of dying from breast cancer. He didn't have a big enough affect to impact significantly on the risk of dying from any cause so what we call the old clothes for tality was not reduced. So this is interesting and a bit puzzling the early screening provide some evidence of preventing deaths from breast cancer, even that depends on exactly how you slice up the data, but in any case the early screening doesn't demonstrable you prevent that's a rule. It's questionable then to claim that the early screening save lives.

00:17:17
So what's going on with mortality is a measure of death from any course and I wasn't estatistica a significant difference in all cause mortality between the Moon and the study who was screen early for breast cancer. And those who weren't. Breast cancer death World explanation is the screening and follow-up treatment had fatal side effects for some women some treatments do but there wasn't any evidence of that in this case. Another explanation is that some deaths from breast cancer were prevented and then the woman died from some other shortly after but the basic story is probably the breast screening simply didn't save enough lines overall to be detectable amidst the noise. It's just hard to conduct a study large enough to pick up changes in all-cause mortality unless the true affect the very large. It's because breast cancer particularly for women who are age forty is a

00:18:17
Relatively rare cause of death. Sorry in this child breast cancer accounted for 7% of the Dead set by observed over the follow-up. And you reduce that by little tiny fraction. You can say that the impact on all calls V is going to be even Tanya it's going to be really really small difference. And therefore you would need to have an enormous trial hundreds of thousands of women over a similar number of women ended up dying overall in the two groups. The study did find some evidence of fewer deaths from breast cancer among those who'd been screened in their forties by how much was the risk of dying from breast cancer reduced in relative terms. It's a 25% reduction in the risk of dying from breast cancer about an absolute terms. It's still a very small reduction. It turns out to be about one fewer deaths from breast.

00:19:17
How to put 1,000 of women who actually participate and screaming toddler activity to prevent one death on the other hand you are so obsessed with Carlos blacktivity contestability if you like lies because as you say if you are the one in 1,000 women whose life is saved is breast cancer death is prevented by participating regulating screening then obviously that's a really important benefit and I guess the question that seems the most important to me than is what what happens to the other 999 women. So what does happen to the best of my said that about 18% had at least one false positive that's just under 200 women for every 1,000 women screamed. So that's a false alarm where the screening test.

00:20:17
Does an abnormality and the woman has to go back and have more tests to sort out whether there is a breast cancer. They are will not and we know I from other research that women describe. This is a very scary time and that it hasn't some women quite long lasting psychological effects screening older women those between 50 and 70 does offer a much larger benefits in terms of reducing breast cancer deaths it prevent five deaths for every thousand women screened before these old women that also higher risks of overdiagnosis. This is different from a false positive a false positive only detect a cancer that is na in the end of the same anxiety more scans and tests the women would eventually be reassured that she didn't have cancer overdiagnosis is different. That's a cancer that could have been left align undetected untreated and it would never have caused any problem.

00:21:17
To the individual within the last time and it's always a risk when you go looking for disease early. So you would have a woman who cancer was identified. She would have to get to Willow Treatment and she would sinks at the Sweeney had saved my life. But in fact I had that be no screening. Nobody would ever have known and she would never come to any harm. It's one of those things that can only be measured epidemiologic Leo statistically. So what you have to do is look at a population of people who has screamed and then compare that to a similar group who are not screened and follow them up all the time to see if there's an excess number of cancer is found in the strange group Professor. Parrot says the best estimate is that around 20% of all breast cancers in women who is screened overdiagnosed. I'm like a lot of healthcare.

00:22:17
Training is done to people who a well by definition. We doing something offering something to people who a well and therefore it is always the potential to make them less. Well than they are at that moment Professor Alexandra Barrett of the University of Sydney.

00:22:38
Can I defrost?

00:22:44
Guido cavalcanti today

00:22:51
Historical drama Hallett tells the stories of women navigating the life and the sex industry of 18th century London strength topic for a current affairs programme such as moral as maybe but loyal listener Sarah got in touch and we live to serve but one in five women in 1760s London was a prostitute that seems extraordinarily high with me to help answer. This question is are historical sex industry correspondent. Lizzie McNeill had a lazy long time. No see ya my new correspondent job isn't getting me on Iris often is a tight. But today I've got big time to look at it appears in large letters on screen at started the series.

00:23:39
1763 London is booming and one in five women make a living selling sex. Is it true? Well, I got in touch with Dan Cruickshank historian who wrote The Secret History of Jordan London how the wages of sin shaped the capital opinion women had no rights professions information about life in the city is Wimbledon Esa consistency consistently.

00:24:24
That one woman in 5 in Section Street perception that can Jordan X people believe the claim of one in five or less. We know that just because something is repeated. It doesn't mean it's true. So, what do we know about the evidence tonight? There are several contemporary sources of a congratulatory epistle from a reformed rake of 1758 who wrote

00:24:53
Amongst every five women there is at least one or as I can with the greatest variety of stud that among all my female acquaintances. I could never think of any five but what I either had or could have Lain with one of them, but let's just say it doesn't sound like I know a misogynist in a misogynistic Society. I mean just because he wanted to stay or about the morals of the women he knew it doesn't really tell us much about sex with his assistant Dan Seitz is Patrick Calhoun a Scottish statistician and magistrate down looks of what we know about the population of Georgia in London and found that 50000 roughly quotes to 1 and 5.

00:25:53
Roughly $20,000 and the other $30,000 I owe a broken down two women who occasionally resort to prostitution when times are tough or so could kept women and Mistresses women who lived with men that they weren't married to with clearly sex workers by modern standards and some of them definitely know that is regarded as equivalent sex with these days that is another figure out which is probably the closest thing to an official statistic that we've got the estimated that the around 3,000 full-time prostitutes operating in London, which seems to be based on a rest records but wasn't actually go during this time. So the woman counted would have been arrested for something else.

00:26:53
Which would be just on the one in ten women seems to be our best estimate. I didn't women who did some sex what part time and you probably still don't get to one in five women, but the claim was widely made at the time that is not completely as mad as it sounds please send us your questions and your comments are more or less bbc.com UK. We will be back again. Same time same place next week until then. Goodbye.

00:27:31
More or less was produced by Simon Maybin with Josephine kassly and Lizzie McNeill. The production coordinator was Brenda Brown a studio engineer was James Beard editor, and I'm Tim Harford.
Translate the current page