The industry playbook to combat public health, and FUTURE-AI
34:03 Share

Welcome to the BMJ's Medicine Science Podcast. I'm Cameron Abassi, Editor-in-Chief. This week we'll hear about fast food giant McDonald's subverting attempts to stop it opening stores near schools.

And McDonald's arguments, they sort of claim that, you know, customers could order a salad from a drive-through. They could cycle or walk there. And they also stress how the company's sponsorship of local football teams sort of promotes health and well-being. Sticking with industry behaving badly, we'll learn lessons from the attempts to control big tobacco companies.

Never underestimate how sophisticated that playbook is. And it goes, you know, it's so much more than just denying of harm. It's influencing research, how that research is done, who does it, who uses it, what questions are asked. And finally, AI in healthcare. How can you tell which applications are trustworthy?

For these editors to be trusted, we need to generate evidence on different aspects holistically so they would be accepted and adopted later. Today I'm joined in the studio by Rebecca Coombs, Head of Journalism at the BMJ. Hello Bex. Hello. And also by Mae Van Schelkwyk.

from Edinburgh University. Hello, May. But you've just moved jobs, have you? Yes, yes, I've just moved jobs. So I've been researcher with colleagues at the London School of Hygiene and Tropical Medicine for several years now, but recently moved to Edinburgh University. And you were at London School when you wrote the paper we're going to talk about. Yes, that's right.

And Bex and May are here to talk about industry's effect on health, something that's a bit of a regular theme at the BMJ and in our podcast. So I'm going to start with Bex. Now, you've just published an investigation about McDonald's. Tell us about that. Yeah, that's right. So obviously a big topic at the BMJ is the way that major industries influence health for good and bad.

And we call this commercial determinants of health, which, May, I know you're an expert on. And we've done a series of investigations recently into how the economic power of multinationals can be used to influence public policy on food in ways that can hugely impact health outcomes.

And this latest story just published goes to the heart of this problem, actually. And it's about fast food giant McDonald's, which everyone's heard of, and how it's been thwarting local councils' attempts to create healthier environments.

They've been really successful actually in overturning council's attempts to prevent new outlets from opening. And they've been doing this by using a sort of playbook of arguments to win planning appeals. So you're saying the applications have been turned down? Yes, by local authorities, including those that have really important anti-obesity plans. And it's been part of that. They don't want new fast food outlets opening, particularly near schools. Yeah.

um and but madonna's been very successful in overturning these um decisions on appeal and some of these successful appeals have been in some of the most deprived parts of england some of the most vulnerable areas yeah with the like the poorest public health outcomes such as like mansfield and nottinghamshire where 28 of pupils leave primary school um

with obesity. So they have one of the worst rates in England. And McDonald's arguments, they sort of claim that customers could order a salad from a drive-through. They could cycle or walk there. And they also stress how the company's sponsorship of local football teams sort of promotes health and well-being.

And they've even deployed a specialist GP to argue that obesity is caused by over 100 factors other than fast food and that its menu contains nutritious and low calorie options. So that sounds pretty cynical and disappointing. Yeah, exactly. And I'd be really interested, May, to get your take on this. And, you know, does any of this surprise you? And where have we seen this playbook before? Yeah.

Well, I mean, thanks, Bec, first of all, for what's a really important article. And I'd like to say it shocks me to read, but doesn't surprise me at all. You know, these are classic textbook strategies that we see across a lot of these types of industries whose products are potentially harmful to health.

I think what's kind of striking here for me is that it really reveals the contradiction at the heart of a lot of these kind of strategies and the way these industries function. So on the one hand, they'll promote themselves as acting in the best interests of their customers or to protect children and part of the solution when it comes to addressing obesity. But

But yet, on the other hand, expending considerable resources actually blocking or delaying the very regulations we know that are actually needed to deal with this problem. So, I mean, May, just to kind of jump in there, which is that

It's one thing to say we've got healthy options on our menu. Well, possibly healthy options. You can cycle to us to McDonald's. I mean, I guess you're just walking from school or it's next to school. And if that's where they're citing these. But.

That's all well and good, but the evidence doesn't support that, does it? I mean, the evidence tells us something very different and it tells us that if you have these outlets near schools, for example, then that's bad for children's health. Yeah, that's absolutely right. And I think, you know, at the heart of a lot of these issues, we know that, you know, what increases consumption? Advertising, availability, having these options everywhere in your environment. And we know that those environmental factors are what drives consumption. Absolutely.

Bex, when you, with this investigation, well,

When you say that these were some of the responses that McDonald's gave, were they the responses they gave to you or were they, presumably they were the responses that they used in their planning applications? Yeah, and to get exactly, to uncover the basis of their appeals, we had to do freedom of information requests to the planning inspectorate, which deals with planning appeals. Okay.

And what we found is that, well, we identified 15 cases since January of 2020 in which McDonald's has objective to new proposals by local authorities to ban fast food outlets close to schools and in areas with high obesity.

And they win the vast majority of these appeals, going back to what May said about the huge resources. This is having a huge chilling effect on local authorities, by the way. If you're going to turn down an application, you know that an appeal is in the works and you may end up having to pay the costs of that appeal. You can imagine what effect that has on your decision making. No, I understand that a bit.

I'm thinking of when the council seems to be doing trying to do the right thing in the first place. So when they're saying, you know, we don't want to create these zones where we don't have any fast food outlets. Yeah, that's actually the right thing to do. Yeah.

Why aren't they being a little stronger? I appreciate there's the threat of legal action and there's a threat of having to pick up costs. And McDonald's presumably have clever arguments and clever lawyers and clever marketing people to make those arguments for them. But why aren't councils being a little stronger in this? Ultimately, it's the council that's deciding on whether or not to allow that planning permission application to go through. I mean, I think it's a financial challenge.

essentially. I mean, May, you probably know more about this than me, but I think it's interesting because, you know, this new Labour government is making all the right noises, isn't it? And it's saying that it wants to stop fast food outlets opening their schools and it wants to create healthier food environments. But you wonder what they're doing to better support these councils? What do you think, May? Yeah, I mean, I think you can't underestimate the amount of, you know, kind of time and resource that needs to go into unpicking a lot of these arguments, you know.

I've been very fortunate in spending years learning about these types of strategies and practices, but they are extremely sophisticated and on a surface are very appealing. You know, these references to evidence and use of kind of very evidential sounding terms. As Bex was saying, we can't then underestimate evidence.

what the hurdle that local authorities then face, given the considerable amount of other things they're having to juggle in a very resource restricted context. I suppose I'm also interested in what kind of pressure central government comes under from these kind of multinationals.

You know, it's something we see over and over again. We know a little bit about lobbying activities, what we can tell. They can only actually FOI ministerial meetings. Obviously, it's a huge way that goes on underneath that with civil servants. Yeah, I suppose if there's an environment created centrally, which says growth is the topic of the moment, growth is the most important thing, then...

There is a sort of downward pressure there, isn't there? And also councils are struggling to meet their budgets and control their costs as well, aren't they? Would you put all these allegations to McDonald's? Yeah, we did. And they actually told the BMJ that they're really proud of the 94.5 billion it has contributed to the national economy since opening its first UK restaurant in 1974. And

And they say that, you know, we've unlocked more opportunities across the UK by investing over one billion over the next four years. That's their plan, creating 24,000 new jobs. So it goes to what you were just saying about the growth mindset. And again, we're going to pick this up later when we talk about gambling and tobacco. It's one thing to think, well, you know, how much have we put into the economy or we've generated with the exchequer?

these companies rarely calculate, these industries rarely calculate the downside, the harms, the impact on people's health and well-being. May, can we talk about your paper? I think it's a good time to kind of break into that. I mean, you're looking at the role of the gambling industry and the harms it causes and what it might learn from

The tobacco industry? Well, not the tobacco industry, what they've learned from the tobacco industry and how we might combat that. Yeah, absolutely. Well, I suppose the kind of, you know, driving force or the thinking behind our paper is that, I mean, as a group of researchers, we study industry practices and how they impact public health. But what we know is that there's overwhelming evidence of what we've been talking about throughout this podcast.

you know, podcast so far is that there's this industry, these cross-industry practices that are used by not just tobacco but across a range of health-harming industries or those products who are potentially harmful to health.

And what we really wanted to emphasize is that while we've seen kind of action taken in the context of tobacco, you know, rethinking this industry as a vector of harm and taking action to address those practices. What we're not seeing is action, you know, that that evidence is informing how we interact and regulate other harmful industries and a particularity.

particularly notable one, a really important example is the UK gambling industry. And I think you can just see that as we were going along in 2005 into 2007, we were one of the leading countries getting involved in, for example, the Framework Convention on Tobacco Control, really taking action to address the tobacco industry, but at that same time completely deregulating the gambling industry.

we wanted to draw some parallels between those two industries. - Yeah, so I mean, the first thing is that there's a common playbook, isn't it, across these industries. So you've got big tobacco, big oil, big food,

100%. So what do we call gambling? Big gamble? It is. It's difficult. But I really, actually what you've captured there, Cameron, is spot on. What you said was big gamble and not gambling. We need to talk about the products, not people's gambling. So it's exactly, you know, big gambling products. Yeah. Yeah. Okay. But what's,

What I'm trying to get to is there is a common playbook here. Yes. What are the elements of that? I think, you know, one is to never underestimate how sophisticated that playbook is. And it goes, you know, it's so much more than just denying of harm.

It's influencing research, how that research is done, who does it, who uses it, what questions are asked, influencing every stage of the policy process from how we define the problem, what we think is a reasonable response, who's responsible, who's to blame and what's the role of government to influencing youth education.

shaping public debate by influencing the media, even journals are influenced and then importantly then representing the industry not as the driver of harm but as the solution and often able to self-regulate and also promote their favoured policy responses and also importantly present themselves as experts in the very health issue that they're actually contributing to.

Yeah, I think one of the issues we all face is that if you take tobacco and

I think people can imagine a clear link between smoking and harm to their lungs and probably don't imagine the link so strongly to heart disease and other problems that smoking causes, but certainly to respiratory problems. When you look at some of these other industries, and gambling is one of them, the evidence points to them having harmful effects on

But as an individual, you might not see them directly unless, of course, you end up bankrupt or something. How do you try to...

get over that, because I think that's a challenge. I do think it's a definite challenge and I think it's also very much plays into the favour of a lot of these industries in being able to either blame people for the problems that arise, to cast out on the burden of harm that's occurring and also to kind of continuously delay action on something by calling into question the links, which

It's the same with the food industry and we see that. Another important example is the indirect harms. Fossil fuel is exactly the same. It's very difficult to look at your own consumption of oil, for example, and link it to extreme weather events. But that's why it's so important to see this at that

at that population level as well and really just fundamentally rethink what is the driver of these harms and think about how we can by thinking about that in a very different way prevent those harms from happening in the first place and countering this kind of idea of it's all just down to individual responsibility and choice in particular because if you look at

You know, the outcomes, the health outcomes that we're seeing, gambling harm, obesity. These are apps. They're actually completely logical outcomes of environments that are designed to get people gambling and get people consuming unhealthy products. Yeah. I mean, this point about individual responsibility is one that we often pick up on, Bex, isn't it? Absolutely. And we've been doing some training with a group called Frameworks to sort of reframe how we think.

present obesity and and one of the one of the things that they suggest is that we talk about food related disease rather than obesity um and we've they've actually given us a really great uh free image library so that when we are um trying to illustrate these types of articles we don't use pictures of overweight children we use pictures of

children in unhealthy food environments so for example in a breakfast cereal aisle of a supermarket yeah and these are all these are all sort of sort of things we're trying to adapt at the BMJ actually to do exactly as you say to sort of focus on the problem okay yeah sorry I was just going to say I think the other thing is also talking you know about about about the systems as being you know we've got a food system that's that's a

problem or a sick food system, you know, and we've got an extremely unnecessarily harmful gambling system that's driven, you know, very much by the industry, its products and, you know, lacks regulation. I mean, one thing you mentioned, May, was about the role of, well, the way industry tends to create doubt about the science.

And that's a common theme across all these industries. I mean, there are examples for big oil, big tobacco, big food, whichever industry. And the same for gambling? Yeah, I think it's important to recognize that the creation of doubt comes in many different ways. So it can be casting doubt on otherwise independent, robust evidence. But then there are other ways in which doubt can be created or more...

influencing of the research agenda in ways favourable to the industry. But I think as well, you know, we know that it...

You know, it's one of these cross-industry tactics. You know, every single time asbestos claimed that they would warn the public. You know, tobacco... People forget the tobacco industry actually funded a health promotion in the UK and other countries. And, you know, we know that they would not ultimately inform the public. This was after they said smoking was healthy. Good for you. Yeah, yeah, absolutely. And, you know, even subtle things, you know, smoking then may be harmful. It just doesn't work. And yet we see...

this has played out in the gambling context is that it's been a major funder of gambling research for over 40 years and that's led to an entire international evidence base that is overwhelmingly favourable in their interests meaning not so much calling into question the link but

but very much avoiding studying it in the first place as opposed to not having kind of longitudinal studies or similar to what we saw in being able to establish a link between smoking and lung cancer, but also just completely problematizing people. So a real focus on what's wrong with people, why do they gamble the way they do. It's about behavior rather than... Absolutely, and about individual level interventions.

So it's a subtle way in which uncertainty around the harms is cultivated. Yeah. May, I want to talk a little bit about self-regulation. Often the industry, be it tobacco in the early days, be it gambling, being food, whichever industry it is, is often left to, they offer to regulate themselves. The pharmaceutical industry, we'll police ourselves.

I think the three of us will entirely agree that there's no evidence that ever works anywhere. Even if you take the police or doctors or lawyers, whichever, even a professional group, there are always problems with self-regulation, even though the principle might seem to be a good one.

How do we get it get out of this cycle with industry? Yeah, if you don't mind, Cameron, I would just say I don't think there's not just no evidence. It's not just no evidence that's effective. It's harmful. I'd say that there's enough evidence to say that it's actually harmful because it delays being able to act.

to prevent the harm. So I think all of these, what other scholars have really kind of referred to as discourses of delay, and that's really important to remember that it's not only ineffective, I think it's fair to say it's a harmful way of listening to it and allowing it to continue.

Okay, so we're agreed on that. In terms of the other lessons, my final question for you, what else can we learn from tobacco? In a nutshell, what can we learn from tobacco that we can apply to how we address the issues we're facing with gambling?

Yeah, so I mean, I suppose just building on what I was saying before is that then think, you know, what does the evidence tell us about how industry influences policy and science and public debate? So we do need to reclaim and reaffirm

and protect a lot of those core institutions and functions that we know are needed to protect health. So thinking about protecting the policy making process from undue industry influence are really key to making progress in tobacco, for example, protecting and maintaining the integrity of science and evidence. But other things like

youth education, training of medical professionals, that needs to be independent of industry influence. The other thing to say though is that a core thing to learn from tobacco control is there's no point in doing a little bit of this over here and a little bit of that. The industry will then just circumvent that and find new ways to influence.

It does need to be comprehensive and needs to be based on this type of evidence. You mean like the Framework Convention on Tobacco Control? Absolutely, yeah. It's not a pick one option. Although that hasn't been, let's be honest about it, I think it's probably the most effective piece of work in terms of

helping us control the behaviour of an industry. But even then, it hasn't stopped the tobacco industry. No, and I think it's about understanding, in my view, that this is, you know, it's not something that you can tick a box. It's being constantly vigilant of how the industry then adapts to your policies and making sure then, this is another really important point to make, is that

is that you evaluate what you're doing. You need to understand, is this being effective in preventing industry influence and the types of harmful practices? And staying abreast of the industry, it's a, you know, you have to in it for the long game. The long game. You mentioned game. I mean,

If I was to say one thing, I think we should stop shirt sponsorship of Premier League games, Premier League teams by the gambling industry. Absolutely. Do you think that would have an impact? Absolutely. I think comprehensive restriction on their advertising is fundamental. We know that from tobacco and we know that because of how much they invest in these extraordinarily sophisticated advertising campaigns. Stop free bets as well.

Yeah. Which really kind of captures people. Well, no, exactly. Sucks them in the first place. Yeah. I can see we've got more work to do, Bex. If you want more details of our McDonald's investigation, you can read the full story online at bmj.com. And May's article with others on learning from tobacco control, time to address gambling industry harms is also available on bmj.com.

Our final segment this week is about AI. We don't need to explain to you how topical and important that is. Even countries are getting involved with the US and China battling out to be the leader in AI technology. Amongst all that pressure to create a monetized AI, how can we in healthcare know what tools to trust?

An international team of tech researchers have come up with the future AI guidelines and Karim Lekadir tells Navjot Lada, our UK research editor, about the six guiding principles to evaluating artificial intelligence.

Yes, so I am a nuclear research professor at the University of Barcelona. I am part of the Department of Mathematics and Computer Science, so that's my background really. But my focus is really on artificial intelligence research for medical and healthcare applications. You're very well placed then to lead on your paper, which is what we're talking to you about. Thinking

quite broadly about the use of AI tools within healthcare, not just how they're studied, but how they're designed and how they're deployed as well. Can you tell us a little, just give us the quick summary of what your paper's about?

My paper was born from the observation that we needed almost a holistic approach to addressing trustworthy AI. It was a big puzzle, and each bit of the puzzle was important. But it was not possible only to focus, say, on development and forget other aspects of validation. It was not possible to focus on development and validation without having put some...

some best practices during the design phase and so on. So the problem of trustworthy AI and ethical AI spans the whole life cycle of AI. The idea was we needed best practices from

day one till the very end and beyond actually. It is not clear yet precisely how we can reach trustworthiness in healthcare AI. So that's why we started thinking that perhaps we need some type of code of practice

some best practices and we need some principles that we need to define that would help, that would provide guidance. Yeah, because when we think within healthcare of

You know, the tools we use, the medical devices, drugs, you know, those all have to go through quite a rigorous evaluation before we can use them within healthcare. And as far as I know, nothing similar exists for AI tools. So is that the kind of gap you were trying to fill? Exactly. So if we look at, for example, the drug development field, as you correctly pointed out,

The guidelines are quite established, they've been approved by different important agencies and we know fairly well how to work in this area. There's a lot of experience behind it.

and that's why there was so much success, for example, during the COVID-19 vaccine development phase. That was quite quick and efficient. In AI, and AI for healthcare applications in particular, where the stakes are high, where this could impact the health and the life of individuals involved, we don't have an equivalent. So this is how we start. We need a code of practice. We need some guidelines.

as clear as possible, as actionable as possible. Of course, the challenge here is to do something that can be applied in different medical specialties, in different types of application because IIC, even though it's healthcare, it's still broad.

And so really that was the goal when we started working on this. In this code of practice, the framework that you've developed, you, and when I say you, I mean you and your co-authors, because there's really a huge number, you know, it really does seem like this huge undertaking that you've done with so many collaborators. But there are these six principles that you outline, you know, that we should consider when we're approaching how we think about AI.

So future AI, what does it stand for?

So we have these six principles and they are the fairness, universality, traceability, usability, robustness, and explainability, and hence future. What we recommend is to have a multifaceted evaluation of the AI tools, not looking only at the performance, the technical performance, the accuracy, the robustness, but really go beyond, of course, looking at fairness, evaluating explainability, usability, acceptance,

looking also at evaluating the tool over time. So what we propose is really a new way of evaluating these AI tools with a lot more depth. And we are making these recommendations also to regulatory agencies. We have over time, this would be...

adopted because as it is today, a lot of these elements I just described are still not really part of the regulatory process. So we only started now saying that we need to evaluate fairness, but we don't say exactly how, what does it mean, exactly what are the attributes that are very important, how to define these attributes that should be included in the evaluation of fairness, for example. I don't even talk about evaluating explainability

trustability and so on. So I think that's really what we, the novelty about here is that we, for these air to be trusted, we need to generate evidence on different aspects holistically so they would be accepted and adopted later.

One question I have when I'm reading the paper is when you were writing it and developing it, who did you have in mind as the user for it? Because I can see that there's a wide audience potentially for this paper. Yes.

Yeah, the users of the guideline, first and foremost, are the developers in the broad sense of the term. A developer could be an engineer, but it could be a clinician that is involved in the development process. A data scientist could be any person that is involved in the process. An ethicist that is part of the development team.

So, could be a person within the development team that is in charge of regulatory aspect of certifying the tool.

So that person also should be aware of this guideline and should apply them to anticipate, for example, on the regulatory requirements. So it's really broad. It's not only engineers. I would say it's stakeholders that are involved in the co-creation of the tool. Well, I mean, I think this will be really useful for many people and is filling this gap.

But an area that a lot of, I know that there are lots of questions about, you know, how can we, you know, as you've just said, how can we ensure that these tools that we're using are fit for purpose and have that quality assurance behind them? Just thinking, you know, it touches on issues that we often think about, about the impact that AI might have in healthcare, you know, on society more broadly. But we were,

Talking about Sam Altman, who runs OpenAI, is just talking about redrawing the social contract and this feeling within the tech industry, certainly, that AI is so important and so transformative that it can remake the world. And sometimes that kind of tech culture butts up against the kind of prevailing culture within healthcare. So how do you think that those attitudes do fit within healthcare? What would be your view on that?

So, of course, introducing AI in healthcare, this induces changes, for example, to the working relationships, to the idea of skilling or de-skilling of the workforce, to the way the care models are defined in clinical settings, and so on. So, from our perspective,

It should be done in cooperation with the actors from day one that they should help in the design of these air tools in a way that it is something that will help them and that will improve things.

Rather than reversionize completely the way healthcare is organized. I think we think this is the way we have an impact. Yeah. So it's an assistive technology that is co-designed by different actors, including the healthcare professionals and healthcare managers and healthcare providers in a way that it adds value that can be integrated to the current care model rather than

change them completely. It's totally disrupting it, yeah. Exactly. We think that, of course, step-by-step things will change, but it's really a process. It's really a process rather than something that will completely disrupt the way things are organized today. That article is available on bmj.com.

That's it for this episode. We'll be back in a fortnight. The Medicine and Science Podcast is available wherever you listen to podcasts. Until then, I'm Cameron Abassi. Thanks for listening.