You may or may not like Dr Prasad. Like everyone he has his biases, but he generally has a very scientific approach. I look forward to this series of videos.
2 Likes
chatGPT:
Tidy transcript
Title: Medicine Unpacked — Longevity Books, Evidence, and Sleep
I read these four longevity books, and hundreds of the underlying scientific publications, so you do not have to. I am Dr Vinay Prasad, and you are listening to Medicine Unpacked.
In the next few episodes, I am going to walk through four books on longevity science. I will go through the claims they make and the evidence underlying those claims. I am especially interested in what they have in common and where they differ. I will appraise the evidence for all of those claims in this series.
The four books span the gamut. We have a professor and provost at an Ivy League university, all the way to concierge medicine and functional medicine doctors. So we get a range and breadth of opinion. By no means is my list exhaustive, but I hope to capture the range of voices in this space.
Whenever the authors cite a reference that I think is of sufficient interest, I dive deep. I read that paper and usually half a dozen related papers on the topic. That is the framework with which I approach this topic.
The four books are:
- Eat Your Ice Cream by Dr Zeke Emanuel
- Super Agers by Dr Eric Topol
- Outlive by Dr Peter Attia
- Young Forever by Dr Mark Hyman
We are going to go through these books and the underlying claims in this series.
Some of the topics we will cover include sleep, which is the topic of this video; nutrition science; exercise science; cancer screening; what blood work you need; cholesterol management and cholesterol medications. I will also do a video on some very odd things mentioned in these books, and another on things they did not mention — curious omissions. Finally, I will do a video on blue zones, regions of the world where people live longer than average, and centenarians, people who make it to 100 years of age. What can we learn by restricting our analysis to these cohorts? I will talk about the evidentiary challenges with such analyses.
Why longevity? Why is my first video back on longevity science?
The simple answer is that we all want to live as long and as well as possible. At the same time, we crave new ideas and new suggestions for how to do that, but we also want to make sure that what we are doing is evidence-based. In this video series, I will try to put together these recommendations, show where the commonalities are, and show what the underlying evidence is or is not.
A couple of disclaimers.
I am very interested in advice and recommendations for healthy people. As a physician, oncologist, and doctor, I am also interested in how you treat people who are sick or who feel unwell. But that is not the topic of this video series. We are interested in longevity advice given to healthy people who may be reading these books.
So, as much of an enthusiast as I am for CAR-T therapy, which is primarily used for haematological malignancies and now some non-malignant indications, I am not going to include that in this series. We are talking about healthy people living longer and better, not how to treat disease, which is a separate category of medicine.
I am also interested in modifiable behaviours, exposures, and things you can change. For example: if you are sleeping six hours, should you sleep eight? If you are doing two sets of curls, should you do a third set? These are things you can modify.
As a doctor, I am interested in genetics and genomics, but for this series I am interested in them only insofar as they can be altered or changed. What is modifiable is within scope. What is not modifiable is out of scope.
If you ask a group of general internists what you should do to live longer and better, they would give you a list that sounds like this: sleep well, eat healthy foods, keep your weight down, do not drink too much, do not smoke, stay active, have a good sex life, enjoy the company of other people, and do not do silly things like drugs, riding a motorcycle, or becoming a recluse.
That is stuff we broadly agree on, but you are not going to make a book out of that. What interests me in these four books is how you make a book out of it. Broad advice like “eat healthy” or “exercise more” begs the question. What is healthy? Which foods? How much? What kind of exercise? When? How often? I am not interested in banalities. I am interested in specific recommendations.
Examples of specific recommendations include:
Introduce yourself to a new person each day, perhaps in the library or coffee shop. Take a multivitamin. Lift weights three times a week. Do three sets of 10 reps. Get a mammogram when you are 45.
Those are claims we can appraise.
If an author says they do something — for example, “I like to fast for 24 hours,” or “I like to wear a loaded rucksack and walk around my neighbourhood” — I am going to treat that as a tacit recommendation. They are not just saying they like to do it; they are implying it is a good thing to do. That is one convention of this series.
Adherence is part of the recommendation.
This is a very big idea. How often and how well people stick with your advice is baked into the advice itself.
Take an absurd example. Suppose someone who is overweight asks me what kind of diet they should follow. I say: “Follow the just-don’t-eat-anything diet. Literally do not eat anything. Drink water, and I will monitor your blood levels. If your electrolytes go out of whack, I will give you IV fluids.”
You would lose weight. But the problem is that this has no real-world adherence. Nobody can sustain that. It is a ludicrous diet.
When we ask a doctor what a good diet is, we are really asking: what is something I can do and sustain? What is a diet I can live with, tolerate, and thrive under?
That matters for all recommendations.
For example, pet ownership is associated in a number of studies with longer survival. Do I think that is causal? Do I think owning a pet makes you live longer, or do I think the types of people who own pets are otherwise more likely to live longer? I tend to fall into the latter category. But I love dogs, so I want to believe that owning a dog makes me live longer.
My first question is: of 100 people who go out and get a new dog, how many still have the dog a week, month, or year later? According to ASPCA survey data, around 90% of people retain a dog one year later, and around 85% retain a cat. So I am happy to concede that getting a new pet is likely to have very high adherence — although, interestingly, none of the books actually advocates for it.
What about the recommendation to introduce yourself to one new person every day? That sounds nice. But I have doubts about adherence. There is no published study showing how often people keep doing that a week, month, or year later. My common sense is probably the same as yours: it is likely to die off very quickly. Awkward interactions will make it hard to sustain.
My rule of thumb is not that high-adherence interventions must work and low-adherence interventions must fail. Rather, if adherence is low, it is prima facie hard to believe the recommendation can improve survival or health outcomes. If adherence is high, it may or may not work, but at least it passes a prerequisite for taking it seriously.
Some recommendations in these books have extremely high adherence. Others have adherence so low that they are barely worth talking about. If no one is doing the recommendation a year later, you need to go back to the drawing board. Perhaps incentives or structure are needed. You cannot just tell people to use willpower.
Another issue is the right recommendation for the wrong reason.
Suppose someone says you should do squats and leg exercises so you are less likely to fall and have a hip fracture. That seems plausible. Strengthening the legs could reduce falls and hip fractures.
But if someone says you should do bicep curls to prevent Alzheimer’s disease, I might agree bicep curls are not a bad idea, but I would question the reason. What is the evidence that bicep curls prevent Alzheimer’s disease?
Then there is the right recommendation, the right reason, but the wrong evidence.
Suppose someone says eating a handful of nuts is good for longevity. I am probably willing to concede that a handful of nuts may be good for you. But if they then say an observational study shows it lowers cardiovascular disease risk by 20%, I become more cautious. Are the people who consistently eat nuts week after week the same kind of people who rarely eat nuts? Probably not. That kind of evidence may not robustly support such a precise claim, even if the general advice is harmless.
A thorny issue across all the books is the distinction between avoiding untimely death and true life extension.
Avoiding untimely death means living to 70, 80, or 90 instead of dying at 40, 50, or 60 from diabetes, cardiovascular disease, or other preventable causes. Life extension means living to 108, 128, or 148 instead of 88. Some people in the longevity space even imagine living indefinitely, with each medical advance arriving in time to keep them alive.
I am interested in whether these recommendations are about avoiding untimely death or truly extending maximum lifespan. Many of us know the general advice for avoiding untimely death. But what actually gets us to longer lifespans? One video in the series will cover the science of lifespan extension and what we know and do not know.
With that prelude, let us talk about sleep science.
Sleep is probably the least controversial topic. Broadly, all these books agree: most Americans do not sleep enough. I agree with that. The questions are: how much should you sleep? Should you sleep at the same time every day? Should you use sleep aids? What should you do about the room you sleep in?
There is widespread agreement across the four books that most Americans do not sleep enough. They differ slightly on how much sleep to recommend. Three books — Eat Your Ice Cream, Outlive, and Young Forever — begin with anecdotes about how the authors did not sleep enough when they were younger, thought they could “sleep when they died,” and later realised sleep was important.
The fact that less than seven hours is bad for you is shown across many studies. But exactly how much sleep you should get is more debatable.
One paper cited in the books comes from the UK Biobank, a longitudinal cohort study of hundreds of thousands of participants with genomic and health data. Some participants have brain scans and complete questionnaires, including on sleep. One prominent paper in a Nature-family journal shows a relationship between self-reported sleep duration and outcomes including cognition, mental health, depression, and mania.
The data show a J-shaped curve: worse outcomes below a certain point and worse outcomes above it, with a sweet spot in the middle. The sweet spot appears to be around seven to eight hours.
But is that right? To appraise the data, you need to know exactly how the question was asked. The sleep duration question was: “About how many hours of sleep do you get every 24 hours? Please include naps.”
That is vital. If people report nine, 10, or 11 hours of sleep, are they getting that all overnight, or are they sleeping six hours at night and taking a three-hour nap? Those may be very different people. Someone taking a three-hour nap may be more likely to have chronic illness, be on disability, or be out of the workforce.
Both people are coded the same in the dataset. Therefore, I have a lot of problems concluding that sleeping more than eight hours is bad for you. I am happy to concede that sleeping too little is terrible for you. Many datasets and short-term physiological and performance studies support that. But I question whether seven to eight hours is truly the universal sweet spot.
Outlive is more open-minded on this issue. Dr Peter Attia cites a study of Stanford basketball players who increased sleep from about 7.8 hours to 10.4 hours and improved their performance. He also cites LeBron James, who reportedly likes to sleep nine to 10 hours a day, sometimes with a nap.
I am open-minded to the idea that the right amount of sleep may be higher for some individuals, especially high performers. I am sceptical of UK Biobank-style data suggesting that more than eight hours is harmful, because it may include people whose long sleep reflects illness, poor night sleep, or chronic napping.
Eat Your Ice Cream says seven to eight hours is ideal. Super Agers also says seven to eight hours and is critical of sleeping more than that. Outlive is more open-minded, and that is where I fall.
How should sleep science be studied better?
First, distinguish people already sleeping seven to eight hours from those sleeping less than seven. For those sleeping five to seven hours, randomise them into three groups: stay the course, increase sleep by one hour, or increase sleep by two hours. Then follow them at one month, two months, six months, and one year to see whether adherence occurs.
If recommending more sleep does not actually increase sleep, the intervention has failed. You need to rethink how to help people sleep more.
If it does work, then measure short-term endpoints such as blood glucose, A1C, cortisol, cognition, performance, job performance, cardiovascular outcomes, stroke, and eventually mortality.
A different study could be done in people sleeping seven to eight hours, asking them to add one hour, add two hours, or perhaps sleep one hour less. Similar studies could be done in people sleeping eight to nine hours. If high sleep durations are truly deleterious, reducing sleep might improve outcomes. Studies must also distinguish night sleep from naps.
Existing studies on this topic appear to be small, short, and poor at tracking adherence. This is an important question and deserves large studies, perhaps thousands or tens of thousands of people.
My takeaways from the books are that they broadly agree. Poor, disrupted, or short sleep — especially less than six hours — may increase untimely death and may also be a sign of chronic health issues. In the absence of credible data, sleep should perhaps be optimised to subjective wellbeing and performance.
There is a paucity of good randomised studies and good sleep advice. If someone goes from six hours of sleep to eight, nine, or 10, I am happy to believe they are less likely to suffer untimely death. But I am sceptical that this will make them live to 120 or 140. I put sleep mainly in the “avoid untimely death” bucket, not the “radical life extension” bucket.
Some further pearls: using alcohol to fall asleep is suboptimal. Sleep medications generally do not work very well; their effect sizes are marginal or minuscule. They may help you fall asleep a few minutes sooner, but the sleep may be more disrupted.
Some recommendations in the books have high adherence. For example, putting black electrical tape over small light-emitting diodes in your bedroom. I am not sure it will make you live longer or sleep more deeply, but adherence is likely to be high. The tape will stay there unless it peels off or you buy new devices.
By contrast, “do not look at your phone before bed” is sensible advice but probably low-adherence advice. Everyone has already heard it, and many people disregard it. Without adherence, it is a non-starter.
What about sleeping at the same time each day?
The UK Biobank has a study in eLife on sleep regularity and mortality. It uses the sleep regularity index, from 0 to 100. A score of 100 means sleeping at exactly the same minutes each day. A score of zero means sleeping at entirely different times. The study finds that more regular sleep is associated with lower all-cause, cardiovascular, and cancer mortality.
But I put almost no stock in such analyses. The people who can sleep at the exact same hours every night are fundamentally different from those who cannot. We may be comparing shift workers with people on stable schedules, people with children and many commitments with people who have more control over their time, people with different resources, self-discipline, and health awareness. This may be less about sleep itself and more about the kind of person who can maintain such regularity.
Sleep trackers are another area where the books differ. Some say trackers can cause anxiety and be inaccurate. Outlive says some find them anxiety-provoking, but Attia finds them helpful.
The claim that one should avoid sleep trackers because they can cause anxiety seems flawed to me. By that logic, you should not get on a bathroom scale or look in the mirror because you might not like what you see.
Some books cite a 2018 Journal of Sleep Research study of 60 participants. Everyone wore sleep trackers, but regardless of how much they actually slept, they were randomised to being told they slept well or poorly. Those told they slept poorly felt less rested. The argument is that if a tracker tells you that you slept poorly, you may feel worse even if you slept fine.
My interpretation is different. The study shows the power of suggestion immediately after waking. The first thing you encounter when you wake up can influence how you feel about your sleep. A sunny morning, birds chirping, and a dog beside your bed might make you feel well-rested. A cloudy day, a crash in the other room, or shouting down the hall might make you feel you slept badly.
This calls into question whether subjective perception is a faithful measure of sleep quality. Studies should include objective metrics such as cognition and performance, not just subjective sleep quality.
I do not think this invalidates sleep trackers. I am closer to the view that if you find them useful, you should use them. Sleep trackers may also motivate behavioural change. Randomised studies could test whether giving people trackers improves sleep, and whether multiple trackers agree with each other.
So, this is an introduction to longevity science through the topic of sleep. There is broad agreement that seven to eight hours is sensible. There is disagreement about whether nine or 10 hours may be beneficial. I am open-minded that more sleep may help some people, provided it is not a marker of illness or poor nighttime sleep.
Much sleep research is poor quality. We need better prospective randomised studies. Adherence is key. Tape over LED lights is unlikely to transform your lifespan, but the tape will probably stick. Telling people not to look at their phones before bed is sensible, but often ineffective because people do not change their behaviour.
That is the kind of analysis this series will apply to longevity science.
Summary
Dr Vinay Prasad introduces a planned series evaluating four popular longevity books: Zeke Emanuel’s Eat Your Ice Cream, Eric Topol’s Super Agers, Peter Attia’s Outlive, and Mark Hyman’s Young Forever. His aim is to compare their recommendations, assess the evidence behind them, and distinguish general health advice from true lifespan-extension claims.
He sets out several methodological principles. First, he is focused on advice for healthy people, not treatment of disease. Second, he is interested in modifiable behaviours rather than fixed genetic factors. Third, he argues that vague advice such as “eat healthy” or “exercise more” is not useful unless translated into specific, testable recommendations. Fourth, adherence matters: advice that people cannot realistically sustain is weak advice, even if it might work in theory.
The first substantive topic is sleep. Prasad says all four books broadly agree that most people sleep too little, and that poor or short sleep is linked to worse outcomes. However, he is cautious about claims that seven to eight hours is the exact optimum. He criticises observational studies, especially UK Biobank analyses, because self-reported sleep duration may include naps and because long sleep may be a marker of illness rather than a cause of poor outcomes.
He is more open-minded about longer sleep durations, especially for high-performing individuals, citing examples such as athletes who reportedly benefit from more than eight hours. He argues that sleep advice belongs mainly in the category of avoiding untimely death, rather than radical life extension.
He also discusses sleep regularity, sleep trackers, alcohol, sleep medication, bedroom light exposure, and phone use before bed. He is sceptical of observational claims about sleep regularity because regular sleepers may differ systematically from irregular sleepers. He thinks sleep trackers may be useful for some people, despite imperfect accuracy and possible anxiety. Overall, he calls for better randomised studies that measure adherence, objective performance, metabolic markers, cardiovascular outcomes, and eventually mortality.
Main claims
| Topic | Claim made in the talk |
|---|---|
| Longevity books | Popular longevity books often agree on broad principles but differ in specifics and evidentiary standards. |
| Scope | The series will focus on healthy people and modifiable behaviours. |
| Specificity | Advice should be specific enough to appraise, not vague slogans like “eat healthy.” |
| Adherence | A recommendation’s real-world adherence is part of its validity. |
| Evidence quality | Observational studies often cannot separate causation from selection effects or confounding. |
| Untimely death vs life extension | Avoiding early death is different from extending maximum lifespan. |
| Sleep duration | Less than six or seven hours is likely bad, but the exact optimum is uncertain. |
| Long sleep | Sleeping more than eight hours may not be harmful in itself; it may reflect illness, napping, or confounding. |
| Sleep regularity | Regular sleep correlates with better outcomes, but the causal interpretation is weak. |
| Sleep trackers | Trackers can be inaccurate or anxiety-provoking, but may still be useful for some people. |
| Sleep medications | Sleep drugs have small effects and may not improve sleep quality meaningfully. |
| Practical sleep advice | High-adherence interventions, such as covering bedroom LEDs, may be more realistic than telling people not to use phones before bed. |
Critique
Strengths
The strongest part of the talk is the insistence on separating general health advice from true longevity claims. Many longevity discussions blur the difference between avoiding premature death and extending maximum lifespan. Prasad is right that sleeping better, not smoking, controlling weight, and exercising may help someone avoid dying early, but that does not mean they will live to 120.
His emphasis on adherence is also valuable. Much lifestyle advice fails because it assumes people will behave ideally. A recommendation that nobody follows is not a practical intervention. This is especially relevant for advice like “do not use your phone before bed,” which is widely known but poorly followed.
He also does well to challenge overinterpretation of observational sleep studies. Long sleep being associated with worse outcomes does not prove long sleep causes worse outcomes. It may instead be a marker of illness, depression, unemployment, frailty, disrupted sleep, or daytime napping.
The distinction between subjective sleep quality and objective performance is another good point. If people’s perception of sleep can be altered by suggestion, then sleep studies should not rely only on self-reported restfulness.
Weaknesses
The talk sometimes swings from appropriate scepticism into dismissiveness. For example, Prasad says he puts “almost no stock” in sleep regularity studies. Confounding is a serious problem, but it does not make the finding worthless. Shift work, circadian disruption, and irregular sleep schedules have plausible biological mechanisms linking them to metabolic and cardiovascular risk. The observational evidence should be treated cautiously, not discarded outright.
His critique of advice like “do not look at your phone before bed” is also a little too narrow. He is right that adherence may be poor, but that does not make the advice useless. A better framing would be to convert it into a more adherent intervention: charge the phone outside the bedroom, use app blockers, set an automatic wind-down mode, or replace the habit with a fixed routine. Low adherence should lead to better behavioural design, not necessarily omission.
The proposed randomised sleep studies are conceptually sensible but practically difficult. Randomising thousands of people to sleep one or two hours more for years would be expensive, hard to blind, difficult to enforce, and vulnerable to contamination. Measuring mortality would require very long follow-up. Intermediate outcomes such as glucose, blood pressure, mood, reaction time, and work performance are more feasible, but they would not fully answer lifespan questions.
The discussion of sleep trackers could also be more balanced. Trackers may motivate some users, but they can also worsen insomnia or create “orthosomnia,” where people become preoccupied with achieving perfect sleep metrics. The fact that scales and mirrors can cause anxiety does not automatically mean trackers are harmless; the relevant question is whether they improve behaviour and outcomes on average, and in which subgroups.
Overall assessment
This is a strong introductory episode because it lays out a useful framework for evaluating longevity advice: demand specificity, ask whether the recommendation is modifiable, assess adherence, distinguish observational association from causation, and separate prevention of early death from radical lifespan extension.
The sleep section is mostly persuasive. The central conclusion is reasonable: short and disrupted sleep is probably harmful; seven to eight hours is a sensible general target; more sleep may benefit some people; and the evidence is not strong enough to claim a precise universal optimum.
The main limitation is that Prasad sometimes treats weak causal evidence as though it leaves us with little actionable guidance. In public health and lifestyle medicine, evidence is often imperfect, but decisions still need to be made. The best position is probably: sleep enough to function well, keep a reasonably consistent schedule where feasible, avoid alcohol as a sleep aid, be cautious with sleep drugs, use trackers only if they help rather than stress you, and design the environment to make good sleep easier.
1 Like