Transcript

Doctors still have questions about the UK's assisted dying bill

From The BMJ · Episode page · Watch on YouTube

· 38 min · 6,600 words · 59 timestamped sections

This is the full, timestamped transcript of "Doctors still have questions about the UK's assisted dying bill", an episode of The BMJ published September 22, 2025 (38 min, about 6,600 words). It mentions 19 people, works and ideas, including assisted dying, Kim Lebea, palliative care, Elizabeth Mahay, Tobias Good.

What this episode is about

The UK's assisted dying bill has advanced through its second debate in the House of Lords, where significant concerns remain regarding the safeguarding roles of doctors and the financial logistics of implementing such a service. Labour MP Kim Lebea, the bill's sponsor, discusses key issues surrounding the legislation amidst ongoing debates characterized by a mix of respect and contention among lawmakers.

Key takeaways

  • The assisted dying bill sparks heated debates, revealing deep societal divisions over life and choice.
  • Correlations in research may mislead; it's not always a straight line from data to causation.
  • Safeguards in assisted dying are crucial, yet many current practices lack any form of protection.
  • Respectful disagreement is key; even MPs find it tough to reconcile personal beliefs with public duty.
  • Vulnerable individuals deserve dignity in death, not just fear-driven policies limiting their choices.

Who speaks in this episode?

What is discussed, and when

  1. – 5:00: House of Lords, assisted dying, Kim Lebea, Theresa May
  2. – 10:00: assisted dying, Gareth Owen, Jalani Hussein
  3. – 15:00: assisted dying
  4. – 20:00: assisted dying, Gareth Owen, Jalani Hussein, King's College London
  5. – 25:00: Tobias Good, Tim Feny, Duncan Jarvis, Jeremy Lere
  6. – 30:00: target trial emulation
  7. – 35:00: Tim Feny, Cameron Abassi
  8. – 38:06: Elizabeth Mahay, Cameron Abassi

Notable quotes from "Doctors still have questions about the UK's assisted dying bill"

  • Hello and welcome to the Medicine and Science podcast from the BMJ. I'm Elizabeth Mahay standing in for Cameron Abassi. This week we have two very different stories from the world of medicine.
    about Elizabeth Mahay · at
  • Questions still remain around safeguarding the role of doctors and the financing of the actual service that will help people end their lives.
    about assisted dying · at
  • Yeah, I mean I've always been really respectful of different views on this issue and I understand the nervousness.
    about Kim Lebea · at
  • There are still issues to work out around assisted dying. Just before the hearing last week, the BMJ hosted a parliamentary roundt to hear the concerns of paliotative care doctors and psychiatrists.
    about palliative care · at
  • Professor of psychological medicine, ethics and law at King's College London and work there as part of the complex life and death decisions research group.
    about Gareth Owen · at
  • Hi, I'm Tobius Good. I'm a professor of public health and epidemiology at the Shiitesman in Berlin, Germany. I'm also an adjunk professor of epidemiology at the Harvard.
    about Tobias Good · at
  • Around causal inference that we'll be publishing soon. Tim, I'm Tim Feny. I'm a research editor at the BMJ. I'm a clinical editor at BMJ medicine. I'm a researcher at UNC…
    about Tim Feny · at

Who and what is mentioned in this episode?

People

  • Kim Lebea — Person · Main guest (interviewee) / Bill sponsor · 2 mentions · first at
    Interviewed as the MP sponsoring the assisted dying legislation; explains safeguards, autonomy framing, and implementation/budget concerns.
  • Elizabeth Mahay — Person · Host (stand-in presenter) · 2 mentions · first at
    Presents the episode, introduces both segments, and conducts the interview with the bill sponsor.
  • Tobias Good — Person · Main speaker (methods segment interviewee) · 2 mentions · first at
    Explains target trial emulation and the shift toward causal inference from observational data; also described as a BMJ/BMJ Open methods editor/consultant.
  • Jalani Hussein — Person · Referenced expert / Segment speaker · 2 mentions · first at
    Provides expert concerns about safeguarding and inequalities (racialized communities, domestic abuse victims, autism/learning disability) in the assisted dying bill.
    My name's Jamila Hussein. I'm a palative care consultant working in inner city Bradford in Yorkshire. I'm also an academic focusing on inequalities at the end of life. So, I…
  • Gareth Owen — Person · Referenced expert / Segment speaker · 2 mentions · first at
    Explains ethical/legal and clinical-process issues: importance of preliminary discussion with palliative care assessment and the 'demedicalized' nature of eligibility assessments.
  • Tim Feny — Person · Main speaker (methods segment interviewee) / BMJ editor · 2 mentions · first at
    Provides editorial perspective on evaluating evidence beyond simplistic hierarchies; discusses comparing emulations to RCTs and evidence synthesis cautions.
  • Duncan Jarvis — Person · BMJ producer / Segment interviewer · 2 mentions · first at
    Introduced as the BMJ producer who interviews methods experts about target trial emulation and causal inference from observational data.
    This novel methodology is called target trial emulation. And Duncan Jarvis, our producer, talked to some of our research team about how it works.
  • Jeremy Lere — Person · Main speaker (methods segment interviewee) · 1 mention · first at
    Discusses language/clarity around causal questions in observational research and authors an upcoming BMJ methods/reporting piece on causal inference language.
    Yeah. Hi, I'm Jeremy Lere. I'm assistant professor of epidemiology at Arasmus MC in Rotterdam in the Netherlands. and you are the author of an upcoming research methods and reporting about the language.
  • Theresa May — Person · Referenced politician / Critic of bill (quoted) · 2 mentions · first at
    Cited as a prominent opponent raising concerns about abuse and pressure on vulnerable people; her phrasing ('license to kill') is discussed and rebutted.
    It a license to kill and she said that it effectively says suicide is okay and she raised some concerns about how it could make vulnerable people feel that they're kind of…
  • Cameron Abassi — Person · Regular host (referenced; absent this episode) · 3 mentions · first at
    Named as the usual presenter; referenced during the episode and in the closing as returning next episode.
    Hello and welcome to the Medicine and Science podcast from the BMJ. I'm Elizabeth Mahay standing in for Cameron Abassi. This week we have two very different stories from the world of medicine.

Companies & orgs

  • BMJ — Company · Host · 1 mention · first at
    The publisher and host of the podcast, providing a platform for discussing medical and scientific topics.
    Hello and welcome to the Medicine and Science podcast from the BMJ. I'm Elizabeth Mahay standing in for Cameron Abassi. This week we have two very different stories from the world of.
  • Harvard TH Chan School of Public Health — Company · Affiliation · 2 mentions
    The academic institution where Tobias Good holds an adjunct position, highlighting his expertise in public health and epidemiology.
    Public health and epidemiology at the Shiitesman in Berlin, Germany. I'm also an adjunk professor of epidemiology at the Harvard Chen School of Public Health in Boston, Massachusetts.
  • King's College London — Company · Affiliation · 1 mention · first at
    The academic institution where Gareth Owen works, providing context for his expertise in psychological medicine, ethics, and law.
    Hello, my name is Gareth Owen. I'm a professor of psychological medicine, ethics and law at King's College London and work there as part of the complex life and death decisions research group.

Concepts & topics

  • assisted dying — Concept · Core topic · 11 mentions · first at
    Central theme of the episode, discussing the legislation and ethical considerations around assisted dying in the UK.
  • palliative care — Concept · Key argument · 4 mentions
    Discussed as an alternative or complementary approach to assisted dying, focusing on patient care and comfort at the end of life.
  • target trial emulation — Concept · Framework · 5 mentions · first at
    A novel methodology discussed in the episode for establishing causal relationships from observational data, challenging traditional research hierarchies.
    This novel methodology is called target trial emulation. And Duncan Jarvis, our producer, talked to some of our research team about how it works.

Places & events

  • House of Lords — Place · Setting · 3 mentions · first at
    The legislative body where the assisted dying bill is being debated, highlighting the political and legal context of the discussion.
    Firstly, the assisted dying bill has just passed through its second debate in the House of Lords. I spoke to the Labour MP who's sponsoring the bill, Kim Lebea.
  • Sun Valley — Place · Setting · 1 mention · first at
    The constituency represented by Kim Lebea, providing context for her political role and perspective.
    I'm Kim Levita, the member of Parliament for Spen Valley in West Yorkshire and I am the sponsor of the terminally ill adults end of life bill.
  • West Yorkshire — Place · Setting · 1 mention · first at
    The region where Spen Valley is located, adding geographical context to the discussion.
    I'm Kim Levita, the member of Parliament for Spen Valley in West Yorkshire and I am the sponsor of the terminally ill adults end of life bill.

Full transcript with timestamps

Hello and welcome to the Medicine and Science podcast from the BMJ. I'm Elizabeth Mahay standing in for Cameron Abassi. This week we have two very different stories from the world of medicine. Firstly, the assisted dying bill has just passed through its second debate in the House of Lords. I spoke to the Labour MP who's sponsoring the bill, Kim Lebea. And maybe the old phrase correlation does not equal causation isn't so simple after all. The BMJ's Duncan Jarvis spoke to the authors of some new research on how far exactly we can infer causes from observational data alone.

Firstly, the assisted dying bill has passed its second reading in the UK's House of Lords and will now go through another committee stage to finalize what has been a very contentious piece of legislation. Questions still remain around safeguarding the role of doctors and the financing of the actual service that will help people end their lives. In the days leading up to the bill's second hearing in the House of Lords, I went to Parliament to talk to Kim Lebbiter, the MP who proposed the bill in the first place to ask her about some of these key issues.

I'm Kim Levita, the member of Parliament for Spen Valley in West Yorkshire and I am the sponsor of the terminally ill adults end of life bill.

And do you watch the debate on Friday? So it was the first half of the Lord's debate. I was here in the Lords all day other than toilet breaks and sandwich break. Um yeah, I watched the entire debate last week.

What did you think of it? I thought it was really powerful. The debates in the commons have been extremely powerful, respectful, civilized for the main part. There have been some exceptions sadly, but generally speaking, people have treated each other in a very civilized manner irrespective of the different views there are on this subject. And I thought that was also the case in the Lords on Friday. And I think you know we have people in different categories on this issue. There are people who are very passionate about changing the law often because of their own personal experience. We have people who are fundamentally opposed to changing the law and we saw both sides of that debate on Friday and we also saw people probably somewhere in the middle uh many of whom think that there are problems with things as they stand but just want to make sure that we get this right as everybody does including myself of

course. So it was interesting to hear those different perspectives. I wanted to ask you about a couple of comments um that were made on Friday particularly around the kind of fears of the way that this could be abused. I think that's the kind of main aside from maybe religious reason that's the main kind of one that we're we're seeing a lot of. Um and Theresa May specifically said she called it a license to kill and she said that it effectively says suicide is okay and she raised some concerns about how it could make vulnerable people feel that they're kind of under pressure to end their lives. I wanted to ask you a bit about um do you understand where people are coming from with that and also if you have a response to those kind of fears.

Yeah, I mean I've always been really respectful of different views on this issue and I understand the nervousness. I understand the concern about such a big change and it is a big change although we are not trailblazers as I've said on this issue. Lots of other countries have been offering this choice to dying people in a very compassionate and very safe way for quite some time now and there were a number of comments made on on Friday which I disagree with. Uh but I think one thing that we've got to get better at in this country is disagreeing well. Um, and I do think somebody like Theresa May who I've got a huge amount of respect for and she's actually very kind to me when I got elected and uh we've had lots of conversations about lots of other things where we actually agree but she is fundamentally opposed to a change in the law and that puts her in a very different position to me as as with some

other colleagues as well. I think the language around suicide is a very delicate discussion to have. Certainly the terminally ill people I have met would never describe themselves as suicidal. They do not want to die. Now I also understand people who are suicidal often do not want to die. But so it's it's a complex area of discussion and debate and it's something that I think we need to explore and unpick. But I'm quite defensive about it because of those conversations I've had with terminally ill people. They want to live but they are dying and therefore that choice has been taken away from them and actually one of the only things that they can choose and take control over is how they die and when they die to some degree. So I think almost saying, you know, we're giving a license to kill.

It's not about that that that that framing of it I find is is a bit insensitive if I'm honest. I think it's about that person having the choice, having the autonomy, having the dignity in their dying days that I think we should be able to offer them in terms of a system that could be abused. Look, two two really important points on that. One, we have very few safeguards now in terms of what happens at the end of people's lives. So we have people, as I've said, who are taking their own lives out of desperation and fear. We have people who are going to other countries for an assisted death with very little legal regulation around that. We then have prosecutions of families and loved ones because the police have got to do their job. And this is the fundamental legal failing that we have. And we've got five former directors of public prosecutions who have said, you know, we have these cases

on our desk where we have a duty to have some sort of investigation. So, so there's that situation at the moment which is very unsafeguarded I would suggest. Um, and what the bill does, it puts a very robust legal framework around a very thorough process. And the other criticism, you know, that the bill has had and and we had a little bit of this on Friday as well has been it's actually too difficult. And I have had terminally ill people contact me and said, "Goodness me, Kim, we've got a few months left to live and you're making us jump through all these different hoops."

Now, I find that a bit uncomfortable, but I also don't really apologize for it because I would rather have a very robust process that, you know, is foolproof and means that only someone who really wants to make this choice would make this choice than have a very quick process where it would potentially be open to um abuse. So, I am very confident in the safeguards within the bill. But equally, if the laws have got further ideas to refine the legislation, I've always been very open-minded to that as well.

West Streeting has said there's no budget for an assisted dining service. What's your response to that? And do you think that there is a way that this can be rolled out?

Yeah, I mean, one thing I did struggle with a little bit at committee stage in the Commons was extending the implementation period from two years to four years, but I've reflected on that quite a lot and I think actually it's very important that we get this right. It's very important that we get the structures in place to facilitate this choice for people. Um, so I would rather do it well than do it quickly. That gives us four years. Now, did we come into government with a budget for assisted dying and this this bill? No, we didn't. But actually, if the law is passed, um, which I hope it will be, and I've got no reason to believe it won't be, then government will make this work.

The other thing I would say is there are costs involving setting it up and there was a very thorough impact assessment done by the government but there are also savings as well. What I've tried not to do is distill an issue of such significance and emotional importance to pounds and pence and money because that that feels doesn't feel right. But I also understand we do have to accept that you know there is a a financial side to it. These are not new patients. These are patients who are already in the system by very virtue of the fact that they have a terminal illness. So yes, there will be the infrastructure that needs to be put in place to give people this choice, but we are not suddenly going to have lots of new patients coming forward to access assisted dying um because they're already being treated. So I think we need to yes be aware of costs savings around the process but actually we also

need to keep a very patient centered person- centered approach to this. I I think

there are still issues to work out around assisted dying. Just before the hearing last week, the BMJ hosted a parliamentary roundt to hear the concerns of paliotative care doctors and psychiatrists. Two of those, Jamila Hussein, a paliotative care consultant in Bradford, and Gareth Owen, a psychiatrist in London, spoke to us about what they think is left to be worked out in the bill. Firstly, here's Jamila.

My name's Jamila Hussein. I'm a palative care consultant working in inner city Bradford in Yorkshire. I'm also an academic focusing on inequalities at the end of life. So, I firstly did a consultation with racialized communities living in disadvantaged areas in Bradford. Um I also spoke to 24 victims of domestic abuse and also parents of young people aged 18 to 25 um who are autistic or have a learning disability to get their views on how we could make the bill safer for them. One of the main things that came from the consultation with racialized groups was firstly there was a diversity of views about the principle of assisted dying and you know I'd spoken to Pakistani community, Bangladeshi, black African, black Caribbean, Roma communities. They all said the same thing. The main thing they were worried about is this is going to be something that's disproportionately going to be affect their community

especially postcoid where the stats really bear out that those communities suffered much more and you know they pointed to the cemetery down the road to say you just walk down there Jamal and you will see what happened to our community and what they made clear to me was that they'd also because of this be scared by accessing paliative care services but also acute hospital services because they were so worried that the same thing would happen that these people would come into hospital and a sister dying would be a way of um disproportionately impacting them and it's important to know that I was really careful in making sure they understood their safeguards they understood that that isn't possible you can't just go into hospital and a doctor decides that assisted dying could happen but because of this fear especially postcoid that is how people are thinking about this and any person who's listening to this podcast who

comes from those communities like I do know that this misinformation is already increasing. What worries me though is that within the bill there is the potential to make sure that there are public health approaches so we go out and proactively ensure misinformation is addressed, ensure that we build trust with these communities and we ensure that they get access to the palative care we already know they don't access as equitably as other groups. But when I've been speaking to legislators and parliamentarians, what I often find is when I make this point that it's really important that we address this misinformation is they will reframe this as an issue about equity. That the issue for racialized communities is they do not have equitable access to assisted dying. Now I am not against this. It's the dining principle and I I've always advocated for equitable access that this shouldn't be something

that people have to pay for that anyone should be able to access it if this is their wish. But if we do not understand the context of UK and health and social care, the fact that racialized communities have poorer access to housing, poorer access to good education and employment, poorer access to health and social care from the moment they're born right to the end of life. as deep-seated inequalities that if we then disproportionately target these groups for education or what some people would see as promotion of assisted dying that is deeply problematic. So the key thing is public health approaches to education about what is assisted dying and how is it different to palative care? How is it different to other health and social care so we don't worsen these already embedded inequalities.

The main thing that parents of the young people said was that they have had to fight and we see this in our practice don't we that you will have I see this in palative care I I'm an adult palative care physician that you'll have people with autism and learning disability and their elderly frail parent will be by their side still advocating for them as an adult because again our health and social care system hasn't been able to embed um adjustments and support that's needed that those parents aren't needed. um to to advocate for their um child and the key thing that they came to focus on was the importance of family advocacy.

So at the moment in the bill in terms of family involvement because this is framed around autonomy which I don't disagree with is that it would be up to the person to tell their family and if only if the doctor felt it was appropriate the assessing doctor and that they would ask the person whether they had told their family and perhaps support them to inform their family. Now there's a huge danger in that um for people with who are autistic and those with learning disability that the parents u made really clear that actually you know their cognitive processing is different their communications is different the way that they interact socially a key thing was masking the fact that they will not want to they'll be so scared of um advocating for themselves and and will want to please the health care professional who might be um assessing them for this um so there's real dangers um for this group um by not including family

but again you know I have been raising this since January and what I hear back is that this is an issue of autonomy that this is adding unnecessary bureaucracy and we're um cailing um the autonomy of some groups but it's really important for us to understand from an inequalities perspective what is bureaucracy for one person is an essential safeguard for another person. And at the moment, I don't think the bill adequately addresses these young people with autism and learning disability. When I spoke to the victims of domestic abuse, what they impressed on me is firstly, domestic abuse goes up towards the end of life because the woman can no longer physically keep herself safe. So, if there's a certain level of expectations around cleaning and chores, etc., she can't physically do that. So, she's more vulnerable. But also importantly, her ability to disclose is much harder because she's physically more dependent on that

person, but also the risks to her and her ability to keep herself safe is reduced. What they um said was actually the biggest risk and they didn't really even frame it as a risk was that the woman would choose a sister dying as her way out because she knows the abuse is going to go up because she knows she can't keep herself safe because she doesn't trust that health and social care services will be able to keep her safe and give her safe housing, give her financial supports that she would need. the sister Daniel was probably the best option that she could take and that's really bears out in national statistics that the largest cause of death due to domestic abuse now in the UK is suicide.

So this isn't a theoretical risk, it's a real risk. And that was Jamal Hussein talking about vulnerable groups in this legislation. Now over to Gareth Owen.

Hello, my name is Gareth Owen. I'm a professor of psychological medicine, ethics and law at King's College London and work there as part of the complex life and death decisions research group. The bill makes it clear that the process starts with a preliminary discussion. That preliminary discussion I think is very important to think about in terms of it being a normal sort of health and social care discussion actually about ordinary health and social dis care matters and most importantly about in this context of terminal illness about palative care. What we what you don't want is a situation where people are going into thinking about assisted dying because they haven't been able to have the opportunities to think through just ordinary palative care type matters. So we think in the complex life and death decision group that this this preliminary discussion really needs to have a palative care um assessment baked

in. In fact, we actually think it is a requirement of the process. Uh not because people um need to necessarily accept uh everything that's being offered uh in that palative care context, but just to sort of enable that people have the access equitably um when we know it can it can be so important. But of course after that what the bill is is is is doing is going through a process of assessments for eligibility. So there are assessments of the capacity of somebody to make a decision to end their own life. They're assessments that somebody is in fact um got a six-month life expectancy in the context of terminal illness and assessments of absence of coercion or pressure.

Now those assessments are not inherently medical assessments. I think that's very important to get across. So when doctors are involved, they're they're doing something that's not a conventional medical role because the consent processes here are not conventional consent processes for treatment. I mean assisted assisted dying is not inherently a treatment. It's um a decision that's a personal choice. It's a decision of a very existential personal kind about ending one's own life. And the other thing of course about these assessments and the consent processes that they involve is the doctor is not making a recommendation.

Of course the doctor is not recommending to um somebody that they seek an assisted assisted death. That would be quite quite odd to think think and and concerning in fact to think that a doctor was doing that. Whereas of course in a treatment context that's what you ordinarily would expect a doctor would be would be making a recommendation. So doctors in this role are taking on very unconventional role not not conventional medical roles. They're not inherently medical roles at all. I would argue although medical expertise is going to be important to bring to bear on these assessments. And then the other thing that's very important to understand of course about the consent processes here is that they are they're driven by the patient. The whole overarching um idea here is around patients autonomy. So the patient is making a request to end their own life and then the question for these assessment steps is eligibility. That's

that's the that's the primary the primary matter. So I think it's it is going to be very important in um the committee stages now in in the laws given that it goes through to this select committee that the bill makes it clearer uh that the the importance of the preliminary discussion that this is a palative care assessment that is offered and that if the patient wishes to proceed the assessments that follow are in an important sense demedicalized. Actually, importantly, they're not treatment matters. They're not assessments of treatment that that's being recommended by any doctor. They're um non or demedicalized assessments of eligibility.

Next, we've been taught that you can't establish a causal relationship from observational data. But if you've been reading the BMJ's research recently, you may have noticed that we're publishing more articles that do just that. This novel methodology is called target trial emulation. And Duncan Jarvis, our producer, talked to some of our research team about how it works.

Hi, I'm Tobius Good. I'm a professor of public health and epidemiology at the Shiitesman in Berlin, Germany. I'm also an adjunk professor of epidemiology at the Harvard Chen School of Public Health in Boston, Massachusetts.

And to be honest, you're also uh helped the BMJ with our research evaluation. That's true. So, I'm a consultant method editor for the BMJ and the BMJ Open.

Great. Jeremy. Yeah. Hi, I'm Jeremy Lere. I'm assistant professor of epidemiology at Arasmus MC in Rotterdam in the Netherlands. and you are the author of an upcoming research methods and reporting about the language around causal inference that we'll be publishing soon. Tim,

I'm Tim Feny. I'm a research editor at the BMJ. I'm a clinical editor at BMJ medicine. I'm a researcher at UNC Chapel Hill and my uh area of focus is clinical epidemiology and epidemiologic methods. I would have a rule of thumb that I've always had that you can't make causal inference from observational data. Now over time being at the BMJ that's been challenged. Mandelian randomization came along and I had to to learn about that. But this is yet another kind of step away from from there. Maybe to be if I can start with you. What can you take us through a little bit of this history?

How have we gone from saying no you can't make a a causal inference to to where we are now?

For many years also in epidemiology I got taught that with observational data you can only find associations uh but not uh really proving cause and effect. And I think um that uh we have moved away from this over the last decade or so more specifically um to to really um allowing us to to say well actually we can use what we call observational data. So pretty much like if you have the same exact same study question and study this in the same target population with the same allegibility criteria of participants, you will find very similar findings in a randomized control trial or an observational setting. So I think um that has been uh proven in the recent years with uh with study results as well as conceptional work. And so this has convinced the journals now to encourage researchers to to think causally even in an observational setting. And hopefully also we are in the future allowed to to

say what we are doing if we have a causal aim to actually explicitly say this. that I'm sure that Jeremy has um and also some some thoughts about this.

Yes, Jeremy, please. Yeah. So, when I get asked this question, I often answer by saying that, you know, we've been doing observational or associational research for a long time. And what people usually say is like, okay, this is an observational study or an associational study. uh you know the the past in the past people have said okay my this is not estimating a causal effect but you know maybe it's evidence that we should do a in quotes causal study in the future and so right away to me that's people admitting that there is some connection between what they found in their observational study and something they're interested in that's causal. So it's admitting right away that there is a connection. I think the problem is that yeah there are uh it's it's hard to do an observational study or rather some observational studies are not believable because or because the assumptions are unlikely to

be to to be satisfied to an extent where it's really hard to get anything out. Um but really the goal has always been to do a causal study and there is we I think everybody in that way admits that there's a connection to a causal study. So we've we've always admitted the connection there. Now we're just asking people to be more to be clearer about the question they're asking. So you are trying to answer a causal question. Um maybe your answer is not causal. Maybe it's bad. That's perfect. You know that can happen. You can ask a question. You can answer get an answer that you don't believe. But still the question itself is causal. And when people admit that their question is causal, then they might be more likely to use the the right methods.

I have a feeling I want to take this back to to being taught stats and taught by how to write a paper just to hold it up because that it feels very kind of intuitive to to someone who's starting out on on on that uh journey. Tobias, you mentioned there, you know, the specific conditions um about it and Jeremy, you mentioned specific language. this has kind of come together in this um target trial emulation framework uh a little bit. So perhaps um you could explain what that is.

Yeah, it's it's a very intriguing framework. So in epidemiology we we we used certain aspects of avoiding systematic biases, systematic errors in analyzing in designing and analyzing observational studies and and there was now a tendency to bring this together and then formally it has been formulated um 10 years ago or 12 years ago. So uh in this target trial framework targetal emulation. So pretty much what we are aiming to do is to come up with a hypothetical protocol to run a randomized control trial in an observational setting. So we write up a protocol as if we would run a hypothetical trial then using observational data to emulate it. So it it sounds almost too good to be true but by doing so by defining clearly who should be in your study. So the so-called allegibility criteria, exclusion criteria by clearly defining what is the intervention or what is the exposure precisely how is the

intervention being given once and then you take it forever or you take it for a week or and so on and define what you're trying to estimate and then finding the data to allow you doing this emulating this starting the treatment for those in the intervention group to those in the reference group at the same time to avoid certain biases, have the same length of follow-up, have the same length of outcome definitions. So really having a protocol that is um necessary for running randomized control trial, applying this to an observational setting and using hopefully the best data available to run it. And obviously we are not assigning the intervention at random. We cannot do this. But with certain analytic techniques and enough information to avoid or to reduce the amount of what we call confounding uh will allow to get us very close to the effect that we would have seen if we could have conducted a randomized

control trial. So that's the target trial framework in a nutshell. And you say very close. Are there lots of caveats around that? Well, there are papers now that compare the result of an existing randomized control trial with observational settings. And sometimes they're close, sometimes they're exact, sometimes they're overestimating, sometimes they're underestimating. Uh because that the data that you have available have certain issues certainly and you need to understand them. The question is if you don't have a result of a randomized control trial but you run a target trial framework a target trial emulation is that giving you the uh effect that you're after? There's some evidence that this is actually true to this. So there are randomized control trial that are confirming the observational target trial uh data but there's no guarantee obviously that you can have to say like this is the effect

that I'm after. How much you believe it's causal or not comes down to in that specific case are the assumptions believable? Is it believable that the analytic technique for example that's being used can get uh can account for all of the leftover confounding or selection bias or whatever that bias may be.

Tim, what do you think? You know, I will add that um there are attempts to try and see how close the emulations are to the actual trial uh when comparing the results. And I think that's I think that's the best you can do quite honestly is say is the observational target trial as close as possible to a real trial and that's typically what people use to determine whether or not they think the result should be uh well aligned or not. um you know very frequently in these papers that compare the outcomes between a target trial and an actual trial uh that has uh been performed they find that those that emulated the trial poorly tend to have effect estimates that are deviating pretty far from the trial estimate. Uh whereas when they're very closely emulated, they don't see that much of a deviation. But I mean that's separate from what Jeremy was talking about which is the assumptions um underlying all these uh things we do.

So Tim we've got very clear or I seem to think we had a clear hierarchy of evidence before RCTs at the top of observational below. What does this do to that kind of hierarchy of evidence? Yeah, it's interesting to talk about the hierarchy of evidence given the causal revolution that's happened even since I've been in med school, which I don't think was that long ago, but actually was almost 20 years ago. Jeez. And we were all given these little pyramids that say, you know, at the top is, you know, u RCTs and systematic reviews of RCTs and at the bottom is case control studies or case studies or whatever is at the bottom. But it turns out that this hierarchy is not very helpful because you can have really good observational studies that give you better answers than really poorly done RCTs.

And as a research editor, I get to see the gamut uh all all these assumptions in all these different ways being violated because you have people turning in papers all the time that they may or may not realize they're making mistakes in. Uh and sometimes those mistakes are fixable. Uh and that's fine, too. But I think it's important for research to move forward for both editors to understand that this hierarchy of evidence doesn't really exist. Uh and that you have to modify your heristic about how you evaluate papers that are coming in and how you make judgments on what is going to help your readers, clinicians, and patients learn the most that they can from information that they have.

Uh so it's just it's an exciting time but it also requires a lot more um investment in understanding what's going on because you have to be always vigilant even when you get an RCT you can't step back and trust that everything is just perfect because it was randomized you know there's you still have to pay attention and I think some of the some of the reason that that hierarchy of evidence came about is because people wanted a simple heristic they're like oh RCTs or meta analysis of RCTs I'll just whatever it says, I'll take that word for it. Uh, and observational studies, I'll, you know, I'll look at it, but I won't take much stock into it. But I think we're learning that with, uh, rigorous methods and attentive research personnel and researchers, you can get very good estimates from even the most routine data.

I'm going to clip that bit out, Tim. The BMJ editor says the hierarchy of evidence is non-existent. I can imagine. Uh,

oh, that causing some debate on social. That's what Cameron said. Uh, direct your emails to me.

Could I add something to that? Of course, Jeremy. Uh, so yeah, I agree with everything Tim said. Uh, and I just wanted to say that it could be that the hierarchy of evidence, the pyramid is true on average that like maybe RCTs on average do better than other designs that are below it on the hierarchy. But when you need to make a decision about something, you have one study in front of you. and are you going you know even if 75% of RCTs end up you know being believable you don't know if your study is part of that 75 or not so the hierarchy of evidence becomes not very useful to you if you have a one paper in front of you that you need to make a decision about about just because it's higher up on the pyramid doesn't shouldn't mean that you should believe this paper more or less

you mentioned earlier Tim the you know meta analysis of RCTs when it comes to something like a target emulation trial with apparent causation but with you know um some caveats around it. I mean how does one go to putting all of this kind of data together and perhaps synthesizing um something out of that?

Yeah, that is a very good question. Um well I'm not a metaanalyst. uh so that is not in my area of expertise but I will say that you know I think people don't um you know narrative reviews can help a lot particularly if uh the research field is manageable in those terms for research fields that are huge meta anal well done meta analysis can definitely uh inform but I think it's important to understand the assumptions that you make when you combine that data and the data that are going into these and if they're, you know, really comparable and really whether or not you can actually lump them into groups that can be metaanalyzed. Sometimes people do meta an analysis come hell or high water, but the groups aren't well comparable. The studies don't meld well. Uh, and so you're metaanalyzing and then the input is not great. So the output doesn't give you a whole lot of new information. And so I think there's I I would always

exercise caution when you're evaluating whether or not studies are metaanalyzable, whether or not they can be combined, whether or not the the aspects that are being um thrown into the meta analysis are actually useful to uh get uh information about a group of papers. And that's where I think, you know, narrative reviews are a lot more useful. Uh just because sometimes it just takes somebody with, you know, a brain to sit down and read these papers and and synthesize the information and then summarize that in a way that is informative and not just regurgitating facts. Uh so that people can help can help uh uh clinicians and patients read something coherent that is um that can create a synthesis of this information.

Tobias, you're agreeing there. Yeah, absolutely agree. And then and the the the other piece that that is often forgotten um and Jeremy sort of um talked in this direction already. There isn't such a thing as the effect of something. So if even if you compare results of perfectly conducted randomized control trials being conducted in different settings let's say in the meta analysis and then you show the effect that effect isn't true for everyone because in each setting you may have a difference in effect and that's true for that setting. So it's what we call hydrogenity of effects if you go across different settings and so and so this is important to understand and so also then if you're doing a target trial in different settings you will not find the effect you find the effect that is specific for the setting even if everything is done perfectly. So combining the evidence in meta analysis

is a good thought but in the end it may also be misleading because we're showing one effect, one pulled effect and people say oh that's the effect but isn't such a thing as the effect. So that's something what what what we should also keep in mind when reading these papers and then what uh what Tim said like you know we have to it's always a good idea to think think about these aspects and have people trying to combine the evidence with with all these pieces and providing guidelines um for conditions for policy makers for um health insurance companies or whatever is needed to really provide the best uh um health care to provide the best evidence for people who treat and to prevent disease and and to hopefully reduce burden of diseases. I mean, ironically, I think what I'm taking from this is having that skeptical, you know, view of of anything you read, even the RCTs that perhaps you gave a little bit of a pass

to uh before, you know, applying that um critical thinking there as well. Yeah, I think the well I think the takeaway should be you can be more skeptical of the RCTs which probably on balance do a good job generally because it requires a ton of time and effort so people aren't just doing this you know as a side project as a resident but also I think the takeaway should be that people should not dismiss observational research it can give us a lot of information and maybe the only thing we have to learn about effects because you can't always randomize Thanks.

That's it for this episode. Cameron will be back with you in a fortnight with more medicine and science. I'm Elizabeth Mahay. Thanks for listening.

Questions about this episode

Who speaks in "Doctors still have questions about the UK's assisted dying bill"?

Elizabeth Mahay (Host), Kim Lebea (Guest), Jamila Hussein (Guest), Gareth Owen (Guest), Duncan Jarvis (Guest), Tobias Good (Guest), Jeremy Lere (Guest), Tim Feny (Guest).

What is "Doctors still have questions about the UK's assisted dying bill" about?

The UK's assisted dying bill has advanced through its second debate in the House of Lords, where significant concerns remain regarding the safeguarding roles of doctors and the financial logistics of implementing such a service. Labour MP Kim Lebea, the bill's sponsor, discusses key issues surrounding the legislation amidst ongoing debates characterized by a mix of respect and contention among…

What is mentioned in this The BMJ episode?

19 people, works and ideas are mentioned, including assisted dying, Kim Lebea, palliative care, Elizabeth Mahay, Tobias Good, House of Lords. Each one on this page links to its own page with every episode that mentions it.

How long is "Doctors still have questions about the UK's assisted dying bill"?

The episode runs 38 min; the transcript is about 6,600 words in 59 timestamped sections.

More from The BMJ

About this transcript: the episode's spoken words as captioned on YouTube, split into timestamped sections. Reclip.Me locates every person, work and idea in the text, keeps the time it is first named, and quotes the words around it verbatim. Last updated January 24, 2026.