If you work in health, Christopher Snowdon is supposed to be the enemy. He defends smokers, opposes minimum pricing on alcohol and argues against sugar taxes.
He thinks the wellness industry is “mostly a force for bad” and that nutritional epidemiology, the science underpinning nearly everything you have ever been told about food, is a “complete binfire of spurious correlations and p-hacking.”
As Head of Lifestyle Economics at the Institute of Economic Affairs, he is of course paid to make these arguments and the IEA’s historic tobacco funding gives his detractors an obvious line of attack.
So the caricature writes itself; man devotes career to undermining the health of the nation.
It is, however, a bit of a sloppy caricature, and the reason is uncomfortable for anyone in the health industry. Snowdon’s philosophical position is that adults should choose how they live and carry the consequences of those choices. Personal freedom. Personal responsibility and accountability. Which is, more or less, what the lifting community has promoted for decades.
On top of that, his argument about many prominent public health measures, regardless of what you think of them philosophically, is that they don’t work. So anyone who has ever rolled their eyes at a headline about red meat or taken the view that grown adults can decide what goes into their own bodies may have more in common with Snowdon than they might assume.
Do public health policies work?
Take minimum unit pricing, one of his longest-running targets. Scotland introduced it in May 2018 at 50p a unit, raised to 65p in 2024. The logic is simple enough: if you put a floor under the price of cheap, high-strength booze, the heaviest drinkers consume less and fewer of them die. Public Health Scotland’s evaluation, published in The Lancet, estimated a 13.4% reduction in alcohol-specific deaths and a 4.1% reduction in hospital admissions.
Snowdon’s response was that the 4.1% wasn’t statistically significant and the 13.4% isn’t a fall in deaths at all; it’s a modelled reduction against a counterfactual, an estimate of what would have happened otherwise.
What actually happened was that Scotland’s alcohol-specific death rate rose, from 20.5 per 100,000 in 2017 to 21.5 in 2020 and kept climbing to a fifteen-year high in 2023.
The evaluation was pre-registered, peer-reviewed and controlled against England, which is close to the best you can do when you can’t run a country twice. The death registrations are also real and they went the wrong way.
This is the thing that may surprise the health industry about scrutinising Snowdon’s positions. You go looking for the sleight of hand and find a real study, a real limitation and a genuine argument about what counts as evidence, with the government incentivised to read its own data generously, Snowdon incentivised to read it harshly and the truth sitting somewhere neither side finds convenient.
Which is why we wanted to put one of his key claims to him directly.
The claim is this, and it is his own wording, from the conclusion of his 2015 IEA paper Death and Taxes: it is healthy lifestyles, not unhealthy ones, that have driven up the cost of the NHS. Smokers, according to Snowdon, save the state money and non-smokers are the ones being subsidised. He has published the numbers, he says he has the studies and he has answered the obvious objections many times before.
So we did two things. We interviewed him and pushed on it. Then we went away and checked what he said. Every study he cited, every number he gave and the assumptions upholding the argument. Here’s what he said, vape in hand (naturally), during that conversation.
Is it fair to say that you’ve said healthy, not unhealthy people drive up costs to the government? That sounds insane.
Healthy lifestyles. Basically the argument, which is backed up by empirical evidence and common sense actually, I think, if you just think it through. Put it this way, has the cost of the NHS gone down since the smoking rate reduced from sixty per cent to ten per cent? No, clearly it’s gone up, right? It was half a per cent of GDP in 1948 [Editor: The Institute for Fiscal Studies puts it at 3.5% of national income in 1949–50, the first full financial year of the service] and now it’s eleven per cent of GDP.
Now, there’s all sorts of reasons for that, but one of the main ones is that we have an ageing population. One of the reasons we have an ageing population is we don’t have 60% of the population smoking. So if you take smoking as the most obvious example of this, but the same would apply, I think, to obesity and a few other things, the average age of lung cancer death is about 70. And obviously most lung cancer deaths are because of smoking. Life expectancy is about 80. So straight away, if everybody was dying of lung cancer, to use a crude exaggerated version of the argument, you would have a lot less to spend on, not just pensions, but also on healthcare.
And lung cancer and indeed heart attacks are generally not particularly expensive to treat, certainly no more expensive to treat than other forms of cancer and indeed just other general diseases, and they’re a lot less expensive to treat than a lot of chronic diseases such as dementia, things that tend to take the place of smoking-related diseases.
Now it’s one of the paradoxes in public health that although smoking causes cancer, you will have more people getting cancer if nobody smokes. Because an even bigger cause, or ultimately the biggest risk factor, is old age. And if you get rid of all the risk factors that are killing people at the age of sixty-five, seventy, then you have more people who live into the age of eighty-five, ninety, and they are costing the state, in the case of the UK, where we have state-run healthcare and pensions, a lot of money. And so it’s the ageing society that is costing.
| THE TEST |
| The claim |
| The NHS costs more than it did because we stopped smoking. Fewer people dying at seventy means more people reaching eighty-five, and those people are expensive. Ageing is, he says, one of the main reasons the bill has grown. |
| The evidence |
| He hedges at the start, there are all sorts of reasons, and then builds everything on this one. It’s the weakest angle in the argument, because the body that actually does these projections for the British government has looked at exactly this question and reached a different conclusion. Reviewing four decades of data, the Office for Budget Responsibility concluded that demographic change alone could not explain the rise in UK health spending. Not that it explains less than people think. That it cannot account for the trend. Their most recent long-term projections say that health spending is forecast to reach 13% of GDP by the mid-2070s. Strip out the non-demographic cost pressures (technology, relative prices, the rest) and the same projection, with the same ageing population, rises only to around 9%. The demographics are in both scenarios. The gap between them is everything else. The OBR also notes that health spending is one of the few areas assumed to rise faster than GDP in the absence of demographic change at all. Freeze the age structure of the country tomorrow and the bill keeps climbing. What’s driving it is a combination of rising demand as incomes grow and excess cost growth, particularly the fact that productivity in a labour-intensive health service tends to rise more slowly than in the wider economy. New treatments and technologies add further pressure. The European Commission’s estimates attribute somewhere between a quarter and three quarters of health expenditure growth in industrialised countries to technological change alone. |
| Where that leaves it |
| He’s right that we have an ageing population and right that we stopped smoking. He’s also right, and this gets less credit than it deserves, that age is the single biggest risk factor for cancer, so a country that stops dying of lung cancer at seventy will see more cancer overall, not less. Where Snowdon overstates the case is in describing ageing as one of the main drivers. In the OBR’s latest model, demographic change contributes around 0.4 percentage points of annual real health-spending growth. Income effects contribute three times as much, at 1.2 points, with another 1 point coming from other cost pressures. Ageing matters. But it isn’t what primarily explains the rising bill. |
Is that not basically saying, if people drop dead the day after they become pensionable age, that’s a great saving to the government?
That’s exactly right. From purely a financial point of view from the government, and the government has known this for decades, many, many decades, the ideal time for you to die is the day that you retire. Because after that you stop being a net contributor to the government coffers and become a net taker. And the older you get, the more you start taking.
Is that desirable though, for society? That doesn’t sound like the right way for it to…
No, because it’s not desirable. I’m talking about the economics of this. And the only reason I’m talking about the economics of it is a lot of people who are pushing for more regulation of people’s lifestyles use an economic argument. They use the economic argument that in the case of smoking, smoking-related diseases cost x billion pounds a year, therefore the government needs to clamp down on smoking and it won’t have to spend this money. But the basic mistake they make is they’re only looking at the cost. They’re not looking at the costs that these smokers would incur had they never smoked.
And it’s relatively easy to work out that the overall cost would be much greater if they lived to a long age, because, as I say, it’s older people who are overwhelmingly using the health service, being treated for chronic conditions, many of which exist for years and years and years, rather than dropping dead relatively quickly from a smoking-related disease. The same applies, as I say, to obesity.
The same applies actually to most… about eighty per cent of preventive health measures in general are not cost effective, because they extend people’s lives after the point that they’re retired. There are actually a relatively small number of preventive health measures that save money. Things mainly that affect younger people. So if you can prevent a motorcycle crash that would lead to someone being a paraplegic, that’s a huge saving to the state. If you can vaccinate children, that’s a big saving to the state. You’re stopping people who otherwise would be out of the workforce and claiming all sorts of disability benefits. Huge saving. But that’s a small minority of preventive health measures.
| THE TEST |
| The claim |
| That around 80% of preventive health measures are not cost effective, because they extend life past retirement. |
| The evidence |
| The source is almost certainly Cohen, Neumann and Weinstein in the New England Journal of Medicine, working from the Tufts cost-effectiveness registry. What that literature says is that most preventive measures do not save money. Not that they are not cost effective. Those are two different tests, and only one of them is how health systems decide anything. Cost-saving means an intervention pays for itself. You spend a pound and the pound comes back. Almost nothing in medicine does that. Cost-effective means you get good value in health for the money you spend, measured per quality-adjusted life-year. That is the standard NICE uses, and the standard by which the NHS approves treatment. |
| Where that leaves it |
| Apply his test consistently and you don’t just lose smoking cessation and screening programmes. You lose most of the hospital. Chemotherapy doesn’t pay for itself either. Neither does a hip replacement, or dialysis or a statin. Nobody argues we should stop doing those. The question was never whether health spending turns a profit. |
So we’re basically saying it’ll save us cash but it’s not actually how we want society to be.
Well, what we want society to be is about what individuals want their lives to be, right?
Society is made up of individuals and if individuals want to smoke and eat cream cakes and drink too much alcohol, then that’s up to them. And everybody really at least pays lip service to that. Even quite illiberal people will say, I don’t mind people drinking themselves to death, but why should I have to pick up the pieces?
And so the question is, are you actually picking up the pieces? Or are these people actually, if you look at it in the round, saving you money? And in most cases, although not actually in the case of drinking, but certainly in the case of smoking, obesity, if these people are dropping dead like we’re told they are, they are saving money, and it’s actually the non-smokers who are being subsidised.
The thing is, it seems like you’re saying a healthy person is a living one, an unhealthy person is a dead one. But it’s a study from 2008, right, that your argument largely rests on. Is that right?
No, no, no — there’s loads of studies looking at smoking. But the Van Baal study looks at obesity and smoking, yeah.
Regardless of the study, it doesn’t seem like your version is taking into account the actual health status of the individual whilst they’re alive. So for example, if someone lives to 85 but spends the last twenty-five years of their life in ill health, they’re going to cost a lot more than someone who only spends the last couple of years of their life in ill health, right?
Yep. Which only gives more power to my point, which is that the diseases that we associate certainly with smoking, but to a large extent with obesity, have traditionally tended to be relatively short and inexpensive to treat. Whereas the diseases of old age tend to be chronic and quite expensive to treat. And it’s not just things like dementia, which is becoming a major cause of death as societies age, but it’s the other little things, the hip operations, the cataract operations, the trips and falls, the brittle bones, the constant trips in and out of A&E.
But again, that’s talking about ill health.
Yeah, but that kind of ill health is kind of inevitable if you get into your nineties. That’s the thing.
What I’m saying is, what the entire health industry is focused on — no one out there is just for the sake of it trying to push up lifespan. Everyone out there who’s championing health is trying to increase healthspan. And that would reduce the cost to the government.
No, I don’t think it would, you see. This is about what they call the compression of morbidity. So the morbidity is the bit you get at the end of your life. About fifty per cent of the average person’s lifetime healthcare expenditure is in the last year of life. For some people it can be significantly more than that.
Now, ideally you would want people living a healthy life, living to the age of a hundred, using very little healthcare because, as you say, they’re healthy, they’re physically active and all the rest of it, and then they just die in their sleep. But people like that are really quite unusual. And you can’t compress that period of morbidity. That’s been the evidence, certainly so far in recent human history. That last year of life always ends up being very expensive. It doesn’t matter a great deal whether you have it at the age of 70 or 90 from a financial point of view. What matters is what’s leading up to that.
| THE TEST |
| The claim |
| That healthspan is beside the point. When I put it to him that someone with twenty-five years of ill health must cost more than someone with two, he agreed and said it strengthened his case. Smoking kills you relatively quickly and cheaply; the diseases of old age are chronic and expensive. And that kind of ill health, he said, is inevitable if you get into your nineties. |
| The evidence |
| Snowdon is right that healthcare costs are driven by illness, particularly towards the end of life. But that doesn’t establish that longer life itself is what makes healthcare more expensive. Research using NHS data found that healthcare expenditure is principally determined by proximity to death rather than chronological age, and that proximity to death is itself largely a proxy for morbidity. In other words, the expensive thing isn’t simply being old. It’s being ill. (Howdon & Rice, Journal of Health Economics, 2018) That makes the healthspan question economically important. If the period of ill health can be compressed, if people can remain healthier for longer rather than simply surviving for longer, then the cost implications of longevity change. And we don’t have to treat that as a thought experiment. The OBR now models it. In its July 2026 fiscal report, the OBR looked at what happens if gains in life expectancy are spent in good health and chronic disease does not continue rising beyond the effect of demographics. Under that scenario, healthcare spending reaches around 12% of GDP by 2075–76, compared with roughly 13% in its baseline and 15% under its lower-healthspan scenario. Better health also means more people working, higher earnings and tax receipts, and lower welfare spending. In the OBR’s broader modelling, the cumulative fiscal difference becomes enormous. |
| Where that leaves it |
| Snowdon hasn’t conceded my argument. But his argument depends heavily on what happens to morbidity as people live longer. If longer life simply extends illness, the economics are bad. If healthier ageing compresses or delays illness, they look very different. The latest OBR modelling explicitly recognises that distinction. I would disagree with that. You’re right in saying that there’s a disproportionate cost at the end of life, but there is also evidence to suggest that the later you die, the less expensive that final year is. |
Yeah, that’s true if you die over the age of a hundred, it tends to be a bit cheaper, because by that stage the doctors kind of go, well, this person’s over a hundred, there’s not a lot of point in spending a huge amount of money.
Even before that. So even if you’re comparing someone who dies in the second half of their sixties versus someone who’s eighty-five, ninety, it is more expensive for the person in their sixties. And there is evidence to back that up.
In that very last year, that’s not the evidence I recall. I did write a paper about this called Death and Taxes years ago where I looked at…
We’ll have to trade studies over email.
Sure. But even if that were true, even if a bit more money was spent on people in the late sixties in the last year of life than somebody in the late eighties, the person dying in the late eighties has still drawn a pension for an extra fifteen years, has still been in and out of hospital for an extra fifteen years, has still been getting all sorts of benefits that the rest of society pay for.
| THE TEST |
| The claim |
| Half of a person’s lifetime healthcare spending falls in their final year. And that this final-year cost is much the same whether you die at 70 or at 90, so extending life just moves the bill later without shrinking it. |
| The evidence |
| The first number appears to be wrong, and the correction comes from Snowdon himself. In Death and Taxes, the 2015 paper he mentioned during our conversation, he wrote that the final twelve months account for around a quarter of total lifetime healthcare expenditure. In our interview, he put it at around half. A large nine-country study published subsequently in Health Affairs doesn’t measure the same thing, but puts end-of-life spending in useful context. People in their final twelve months account for between 8.5% and 11.2% of aggregate medical spending. Extend the window to three years and it rises to between 16.7% and 24.5%. The second claim fails on the study he’d recognise. Lubitz and Riley, in the New England Journal of Medicine, found that people in their final year cost over ten times what surviving beneficiaries cost at ages 65 to 69 — but only four times as much at 85 and over. [The 16.6x/2.5x figures are from a different study — attribute separately or cut] |
| Where that leaves it |
| The person who stays well and dies at 85 doesn’t incur the same terminal bill they’d have run up at 65. They incur a smaller one. But the more interesting finding is the one the nine-country study reaches: high aggregate spending is driven not by last-ditch attempts to save lives, but by spending on people with chronic conditions. The deathbed was never where the money went. It went on the years of illness before it, which is the one variable his argument essentially treats as fixed and the one the entire health industry is trying to change. |
Again, you’re just talking about age and I’m talking about health. And actually at the start of this argument you’re saying it’s healthy lifestyles that cost more, and I’m saying it’s not healthy lives that cost more. Healthy lives cost less.
Yeah, but that’s where you’re wrong, you see. And I’ve got the evidence to prove it. And you’re also ignoring pensions, which, the pension pot is an even bigger source of government outgoings than healthcare. And you can’t deny that healthy people who live a long time are taking out more in pensions.
I wouldn’t deny that, no. But the thing is, lifespan at the moment in the UK is pretty flat, and healthspan is actually declining. So that suggests we’re getting less healthy.
The idea that healthy life expectancy has declined in recent years is purely an artefact of the fact that you have more young people claiming to have mental health disorders. They’re not physical health disorders.
So you disagree with the way that that data’s collected, or what it actually means.
Yeah. The Financial Times did a good story about this at the time, about a week after that sort of factoid came out last month I think, and they said it’s only mental health. And because it’s younger people, obviously that drags down the healthy life expectancy more than if it was older people saying they’ve got depression or anxiety or what have you. Now, not to completely dismiss anxiety and what have you and ADHD as problems but it’s not what people are thinking, I don’t think, when they hear the phrase healthy life expectancy. The idea that healthy life expectancy has declined implies that people are getting cancer at an earlier age, getting heart disease at an earlier age. They’re not.
| THE TEST |
| The claim |
| The fall in Britain’s healthy life expectancy is purely an artefact of more young people reporting mental health disorders, rather than evidence of a country getting physically sicker. |
| The evidence |
| The Financial Times examined the fall, from just over 63 years in 2019 to 61 by 2023, and found the measure blends two very different things. Life expectancy is the most precise statistic in demography, a simple tabulation of population and deaths. Self-reported health status is not. Drill into the underlying conditions and physical health has been stable or improving across all age groups. The deterioration comes almost entirely from answers about mental health, concentrated among young adults. The FT’s own summary is that Britons’ physical health is holding up while younger adults, and young women in particular, report rising mental health problems. What I’d contest is one word. He called it purely an artefact. The FT is arguing that a crude statistic has been misread, not that nothing is happening. Its opening position is that there are good reasons to be concerned Britain may have a problem with working-age wellbeing. The Health Foundation, whose analysis produced the original figure, went further. It accepts the measure is imperfect and still calls it a valuable signal of a growing problem, noting that the sharpest deterioration has been among women, people in the most disadvantaged communities and working-age adults. Its position is that whether the driver is mental or physical health, it remains a health problem, and the two are closely interlinked. |
| Where that leaves it |
| He is right about the measure and I was wrong to lean on it. But hold the two halves of his own position together. He dismisses the healthspan decline as an artefact of young people reporting mental health problems. Minutes later, asked whether the country should get healthier for the sake of the economy, he says you would sooner not have people off work with depression, anxiety and the rest of it and that the economy needs healthy workers. Both can be true at once. The statistic can be a poor instrument and the underlying problem can still be real. That is roughly what the FT argues. Which makes “purely an artefact” the wrong description of something he goes on to identify as an economic problem in its own right. It is also, by his own measure, the form of ill health that matters most, because it hits people who are still working, still paying tax, and still years away from the pension that makes them expensive. |
So for the sake of the economy, would you rather we collectively get more or less healthy?
The economy per se doesn’t really matter so long as people of working age are reasonably healthy, which I guess is a big caveat. But yeah, for the sake of the economy you would sooner not have a lot of morbidly obese working-age people, for sure. And certainly you don’t want to have people off work with depression, anxiety, ADHD and autism and all the rest of it. So the economy needs to have healthy workers, clearly.
But in terms of people who are in actual reality more likely to be ill and to have cancer or whatever, they’re mostly retired. So in terms of GDP, it doesn’t make a lot of difference. It does make a difference, as I’ve said, to government spending. So the rest of us do need to pay more tax in order to keep these people in pensions and to get them healthcare.
So best case all round is, healthy working life, drop dead the day after your retirement?
Well yeah. I mean, as I say, this is an argument I only use because people constantly use a spurious argument that we must do something about this because it’s costing us money. And the only reason they use that is because they don’t want to be seen as being nakedly paternalistic and saying, I don’t like people smoking and eating too much and I know best and this is for your own good. And so they dress it up in a spurious economic argument, which I then knock down. And whenever I knock it down, people say, you just want to kill people off for the sake of the economy. No, that’s not what I want. I just want the debate to be grounded in economic evidence.
