Exercise isn’t necessarily good for people
I would appreciate it very much if anyone would take a close look at this—it looks sound to me, but it also appeals to my prejudices.
http://www.youtube.com/watch?feature=player_embedded&v=E42TQNWhW3w#!
My comments are in square brackets. Everything else is my notes on the Jamie Timmons lecture from the video.
Short version: 12% of people become less healthy from exercise. 20% of people get nothing from exercise. This is a matter of genetics, not doing exercise wrong.
****
Ask a hundred people about exercise, you’ll get a wide range of answers about what exercise is and what good it might do for health, and the same for health professionals.
You need to focus on the evidence that exercise affects particular health outcomes. Weight and health are not strongly correlated. BMI is problematic.
There’s a recommendation for 150 minutes of exercise/week, but this isn’t sound. People who *report* being active have better health. People who are fitter have better health. These are not evidence that having a person with low activity take up exercise will make them healthier.
Nothing but a supervised intervention study is good enough.
Improved lifestyle is better than Metformin for preventing diabetes. (Studies) Exercise + diet modification has a powerful effect of preventing and slowing the progression of Type II diabetes. People with Type II have more cardiovascular disease (heart attacks and strokes). However, it doesn’t follow that the lifestyle changes which help with Type II will also help with CVD. [I’m surprised]
Diabetes doesn’t kill, CVD does, and a major motivation for the NHS to care is that CVD is expensive.
[9:45] Two studies which find that lifestyle intervention has no effect on CVD in diabetics. [11:00] One study which found that lifestyle intervention prevents Type II but doesn’t affect microvascular disease (blindness and ulcers). [I’m not sure what this means. Maybe people can have the ill effects of Type II without the disease showing up in their blood sugar levels?] There are no supervised exercise-only intervention studies which show that exercise prevents long term disease progression.
[13:00] The usual advice on exercise from the NHS (pretty similar in the US): Aerobic exerise must raise your heart rate and make you sweat to be benefiscial. The more exercise you do, the better. Do a minimum of 150 minutes/week of aerobic exercise + strength training. If you do more than 150 minutes/week, you’ll gain even more health benefits. Using a skipping rope is an example of vigorous intensity exercise. People aren’t following this advice, and a major factor is the amount of time required. The advice is based on best guesses.
[15:55] Exercise will increase aerobic capacity in 80% of people (lowers all-cause mortality), improve insulin action in 65% of people (lowers type II diabetes by 50%), reduce blood pressure in >55% of people (lowers strokes 25%), increase good cholesterol in 70% of people (less vascular disease), promote muscle and bone mass (? less fractures and ‘aging’)
[17:40] Exercise response graphs. The average person gets a 15% increase in aerobic capacity, but a few get less capacity if they exercise. Insulin response—average of 20% improvement. Some people get better, some get worse. A high proportion, maybe the majority, have little or no change. The people in this chart were doing 150 minutes/week of supervised exercise.
[20:00] High-intensity exercise is exercise which depends on stored energy, there’s no way to take in enough oxygen to contribute. An athlete might be able to continue for 10 minutes. The average person can continue for more like 30 seconds to one minute.
[22:00] Experiments with high-intensity/rest intervals: 3 x 20 seconds of high intensity. [25:00] Charts showing flattened glucose spike (there probably was a peak, but the test missed the moment) and less isulin in the blood after only two weeks of 6 x 30 seconds interval training (total 7 minutes).
[30:54] “Advice has been based on what epidemiology methods can detect, not what is actually important or required.” Health questionaires don’t include things like 20 seconds of running for the bus.
[33:00] Ten days of bed rest will make healthy people insulin resistant.
[35:00] It looks as though modern hunter gatherers expend about as much energy/mass as Americans on the east coast do. [I found I could make sense out of the graphs by using full screen.] This evidence suggests that people are eating more rather than moving less. The evidence for 7 minutes of HIIT three times a week isn’t completely solid, but it’s at least as good as the evidence for 150 minutes/week.
[38:36] ….. Epidemiology of a sort—evidence that eating chocolate makes it more likely to get a Nobel prize. Beautiful corelation! The Swiss eat the most chocolate and get the most prizes. The Swedes are an outlier—they don’t eat as much chocolate as they should to get so many prizes. That the prize is given in Sweden might have something to do with this. Cocoa has flavenols which slow age-related cognitive decline, but the corelation is probably just a coincidence.
[40:00] 12% of healthy people make their blood pressure **higher** by exercising 150 minutes a week. 20% get little or no improvement. [42:00] Graphs of low responders for aerobic capacity, muscle mass, and insulin sensitivity. Exercise does slow progression of diabetes on the average, but that doesn’t apply to all individuals.
[44:47] There’s no obvious indicator to tell high responders from low responders in advance. You have to either check the genes or track the results of exercise. [45:00] Finding non- or adverse responders: change in aerobic fitness is 60% genetic, insulin sensitivity is 40% genetic, strength is 50% genetic. These are estimates from family studies, including twin studies. There are 10 million genes variants which might have at least a 5% effect.
[47:35] There’s a group of 27 genes which together can ‘predict’ gains in VO2max. It isn’t necessary to understand how the genes work to create their effect as long as that effect is predictable, and it’s possible that we will never understand something so complex. There may be drug combinations which can make exercise safe and effective for non-adaptors. There’s research happening. It’s possible to breed rats which are better at responding to training.
[53:52] A life-style program will *on average* reduce the risk of developing type II diabetes. We *don’t know* whether exercise-training on its own will reduce heart-disease, angina, etc. It does improve risk factors and symptoms. If *you* have a risk-factor for ill-health, we *can not* be sure that exercise will help. (12% *adverse* responders, 20% no effect)
[57:00]Public health (what advice should the government give?): 1 minute a day of high-intensity sprint cycling reduces major risk factors. [For what proportion of people?] People tend to like brief high intensity exercise better than longer low intensity exercise. North American study: 150 minutes/week of exercise increase one’s carbon footprint by 15% (food, laundry, showers).
Safety: 2 million marathoners have been studied. Very low fatalities. HIIT isn’t likely to be more dangerous. [Ack! Ack! Ack! What happened to all the care about evidence? Marathoning isn’t sprinting. Fatalities during the race aren’t the only thing that can go wrong. People who do marathons aren’t randomly selected.]
HIIT has be done safely by medically supervised diabetes and heart failure patients. It would take a billion dollars to do a thorough supervised intervention study. Some pieces of it have been done. This is much less than big drug companies spend, without much results. The current hope is finding the gene markers and then useful drugs for non and adverse responders. There are no average people!
**** http://www.medicalnewstoday.com/articles/242498.php
Summary of a TV show which has more details about High Intensity Interval Training.
I’m not sure where this statement comes from. Googling “exercise experiment health” in google scholar has the first 8 results be experiments, where they took an group of inactive people and made them exercise, and they showed improvement over people who did not. I didn’t look beyond the first 8, but I suspect there are many more
Yes there are. Google “exercise cvd”, you’ll find several that show improvement.
The only relavent result I could find says the opposite: http://www.nature.com/ijo/journal/v29/n1/abs/0802842a.html
Mixed results, not enough to show a firm conclusion either way.
Bottom line, the speaker appears to be heavily cherry-picking the studies he shows.
It’s not mixed results but mixed people. For the median person, exercise improves blood pressure. But for 12% of people exercise makes their blood pressure worse every time (and for 20% it’s always a wash).
For 12% of people it makes blood pressure higher not worse. The presenter chooses the terms ‘adverse’ and ‘worse’ for his own purposes.
My girlfriend has chronic hypotension, and can easily faint or become dizzy when standing or if she doesn’t have enough water. Anecdotally, regular exercise seems to help prevent that. It’s hard to find numbers, but the closest thing to solid I could find is that 26% of people with diabetes (couldn’t find general population) suffer from hypotension.
If the 32% of people receiving no change or an increase in blood pressure fall on the bottom end of the bloodpressure spectrum, they’re getting exactly what they need. Regardless of how the speaker labels the results.
Here is the study cited. There are four health markers that were chosen to be analyzed for this analysis. Actual physical fitness (ie VO2 max, strength, body fat percentage) were not measured. In all but one of the examined studies, diet was not controlled for, and they specifically only included endurance exercise, when virtually every recommendation is to combine endurance and strength exercise. Claiming that this population doesn’t respond to exercise is a vast overstatement. More accurately, you might be able to say that “Endurance exercise alone isn’t sufficient to improve health markers in a small fraction of the population.”
EDIT: The inverse to that is “Endurance exercise alone is sufficient to improve health markers in most of the population,” which doesn’t strike me as a good reason to not prescribe it.
And three out of their four markers are known to be strongly influenced by diet!
My understanding was that you can see exercise induced changes in risk factors (blood pressure etc) relatively easily, but that evidence for changes in health outcomes (like risk of death, lowered incidence of disease etc.) were on shakier ground.
He also claims that HIIT as more standard 150 aerobic exercise shows the same level of response for those same risk factors.
The guy claims that variable response to exercise (including ~10% negative responders) was a robust finding that most studies see. He didn’t explicitly discuss whether they had ruled our within person variation as the cause of that data.
From what I can tell, this seems to be true. Does he talk about this in the video?
I remember him doing so.
The idea that blood pressure is an adequate proxy for overall health is highly dubious and I wouldn’t draw conclusions about longevity from it. Even if the inference was valid blood pressure is notoriously unreliable because it is affected by so many variables.
Here’s a variable that might not be well known—doctors and hospitals don’t always have large blood pressure cuffs, so they use cuffs that are too small on fat patients, and get inaccurately high blood pressure readings as a result.
Seriously? This is really embarrassing. Like, how hard can it be to get some valid data around here?
Harder than you can imagine.
Decline in autopsy rates
Autopsies supply significant information
Related: math.ucla.edu/~scp/publications/mortality.PDF
Thanks.
This may imply some heuristics. Don’t trust linear corelations in biology. Don’t trust thresholds which are supposed to apply to everyone, or almost everyone. Be dubious about round numbers.
I agree blood pressure is a generally a poor predictor of health or mortality.
This is often measured because it is easy to measure, rather than it being particularly informative.
Aelephant—that’s a good paper with data on this. I needed to edit that link to http://www.math.ucla.edu/~scp/publications/mortality.PDF for the pdf download to work.
FWIW, one of my takeaways from taking subjective measurements of my happiness, motivation, etc. everyday was that by far the biggest correlation was that I was happiest when I exercised more. But again, that’s not really establishing causation.
It wouldn’t surprise me if you’re right about exercise being good for you, even though I can’t imagine any way to double-blind exercise.
If the lecture is right, then there are a fair number of people exercise isn’t good for, and a smaller number that exercise is bad for.
I’m beginning to think this is about what your priors for the effects of exercise should be. The lecture hooked me, partly because I think the culture has a halo effect around exercise and partly because if there are foods which are nourishing for the majority of people and poison for a minority, why shouldn’t there be similar variation for exercise?
There no reason why you would have to double blind it to measure whether engaging in it causes you benefits or harms. It’s irrelevant whether you improve your health because you think that exercise improves it or whether you improve it because of physical effects of exercise.
The problem’s with self experimentation are different. Subjective ratings can be off. Then you are measuring short term effect and ignore long term effects.
Diabetes and Heart Disease.
Type II diabetes predicts a 7 year shorter life, due to cardiovascular disease. No one knows why. It also destroys small blood vessels, leading to blindness and loss of feeling in the extremities (thence falls or infections). This is called microvascular damage and is fairly well-understood as the direct result of blood sugar. The cardiovascular disease is called macrovascular damage to imply it is similar, but it’s really a smokescreen.
Timmons cites two studies of diet and exercise improving blood sugar but failing to improve heart disease. One is a small Finnish study. The other is a big American study, Lookahead, but it hasn’t published yet; items 88 and 89 in its bibliography sound relevant. As background, there are two large past studies that have conflicting results on whether improving diabetes (mainly via drugs) improves CVD. 15 years ago, the UKPDS study compared aiming for 7% A1c is better than aiming for 8%. It reduced CVD, though not as much as it reduced microvascular outcomes. So 7% became the new baseline and follow-up studies were started that are only just reporting. A couple of years ago, the ACCORD trial finished comparing aiming for 6% to 7%. It found the more aggressive treatment had worse mortality (and worse CVD, though not statistically significantly). The Accord study had lower A1c targets in both arms than UKPDS, but I think it only achieved the same levels, probably because its patients were older and had more advanced diabetes.
One hypothesis that people throw around is that if you are too aggressive with drugs, you cause hypoglycemic events (too low blood sugar), and these are correlated with CVD. But if you lower blood sugar by diet and exercise, you don’t expect to cause hypoglycemic events, so there was some hope that this would improve CVD, despite the Accord study. But the recent studies falsify this loophole. (Also, I think it failed to explain the detailed Accord data.)
Still, the UKDPS suggests that for some people reduced blood sugar improves CVD. Maybe microvascular problems are due to very high levels of blood sugar, at the tail end of diabetes, while CVD is due to long exposure to lower blood sugar. Perhaps for the Accord patients it was too late, but starting earlier with the UKDPS patients worked.
Added: In other words, blood sugar might just be a symptom, not a direct cause of heart disease (but definitely a direct cause of diabetic morbidity). This is not too surprising, since the mechanism isn’t understood. Experiments in the past had mixed results. What’s new is lifestyle experiments, in contrast to drug experiments. It is surprising that they are worse, but type of patients probably matters.
Upvoted.
I just want to point out that this is entirely dependent on what kind of medication you take, it’s practically impossible (I have never heard of it at least) to have a hypoglycemic event without insulin being a part of your cocktail.
Yes, insulin is a prerequisite for a hypoglycemic event. But once on insulin, most drugs increase the rate of events.
From your summary health appears to mean ‘not sick.’ If you can clarify what health is rather than what it is not you might be closer to an answer on whether exercise and health are related. Spoiler: yes.
There’s the level of physical activity of most of my ancestors, and there’s the level of physical activity of my ancestors since the clerical explosion (1800s-present). The former group has a longer track record of success than the later, and for that reason alone I’m guessing they for sure got it right (with a provisional ‘not all bad’ grade for the inactivity of today). Regular moderate activity of some kind with bursts of heavy activity leads to physical strength and physical grace more often than not.
Length of track record alone is enough? By that logic, are CDs better than DVDs? We have higher life expectancy now then we did before 1800, so if we’re just going by post hoc reasoning, exercise is bad.
It seems the results show on that some actually get worse, but I wouldn’t say this is a firm conclusion that some should avoid exercise altogether. The problem with most experimental studies on exercise is that many use an across-the-board regimen for all participants that isn’t customized to the individual. So, if one participant gets 30 mins a day at 70% of HRmax then all do. It is probable that some who get worse may need less or more exercise, greater or lesser intensity, resistance as opposed to aerobic, etc.
This is what came to mind to me too. If you have someone who is thoroughly unhealthy & metabolically broken, adding a bunch of exercise on top of that is not a positive stressor, it is likely to knock them down even further. You have to approach things in order, like fixing the diet & addressing potential issues like hypothyroidism before you start training them to be an athlete.
You might want to add an indicator to the title that you’re not sure whether this is true or not. I think that people might treat the number of upvotes (not huge, but definitely positive) as an indication of the veracity of the claim in the title when I don’t think that they were intended as such.
Isn’t the double modal somewhat redundant, though? “I think it’s possible that it’s possible that exercise isn’t good”.
“isn’t necessarily good for people” could expand to “is known to not be good for a certain subset of people” or “is possibly not good for people (or a subset thereof)” The second reading seems intended, not the first.
A reasonable point. There are at least two sorts of leakage—one is whether I’ve given sufficiently careful attention to the lecture, and the other is whether the material in the lecture is accurate.
Hmmm. 27 genes? Assuming this was done as a data mining analysis, you can always find 27 genes to fit any function. I definitely want to see cross validation results before I accept this.
Do you have predictor they’re using with those genes? The original paper?
Has anybody looked into sarcopenia and resistance training can slow it’s progression?
Yes.
http://www.unm.edu/~lkravitz/Article%20folder/sarcopenia.html
Recommends resistance training for older people, recommends taking reasonable care. No mention of the possibility that some people just don’t benefit from exercise.
Most surprising detail: There can be loss of fast-twitch muscles in the sense that the innervation for fast twitch dies, but may be replaced with slow-twitch.
My hypothesis: this may relate to the risk of falling since keeping a potential fall (a slip or going off-balance) from turning into an actual fall requires the sudden use of strength.
Exercise science is a notoriously poorly executed field, so I’d take any studies done with a decently sized grain of salt. This study gets referenced very frequently as evidence for exercise non-responders. However, anyone with even a surface level knowledge of fitness knows that diet and training must work in synergy to achieve a goal, and this study did not consider diet, calories, or nutrition. If you are not eating enough food to recover from exercise, your strength and fitness will go down (along with your weight). Minimizing muscle loss is the reason why 1-2lb is the recommended rate for losing weight. This chart is pulled from the study, and it shows that a combination of strength endurance training is beneficial for the vast majority of the study population (older adults). The small groups that are negative or non-responders were likely just not eating enough to recover from exercise, rather than some group which mysteriously doesn’t benefit from exercise at all.
This article is cited by the video you’ve posted, and also does not mention diet, nutrition, calories, etc. I checked the studies analyzed for dietary controls, and only the STRRIDE mentioned bodyweight and nutrition controls. The meta-analysis seems to exclude non-endurance training, and the risk factors it identifies are related to disease, not fitness. At best it can claim that endurance exercise doesn’t improve disease markers in a tiny subset of the population. With so many factors not being controlled for, to me the probability that hypothesis “Some people don’t respond to exercise” is true is much lower than the hypothesis “Exercise alone isn’t sufficient to reverse negative health markers caused by unhealthy diet and weight in a small population.”
I believe that they don’t mention exercise non-responders because it’s unlikely that that is even real for normal population. Unless you have muscular dystrophy or some similar disease, you will gain strength, muscle mass, and bone density by weight training and eating a caloric surplus. People with weak bones, muscles, and connective tissue are more fragile and prone to injury, and as you get progressively weaker, the injuries become more devastating. By becoming stronger, you become more resilient to injury.
You’ve got a plausible hypothesis there, but it doesn’t seem to have been tested.
I notice that you go from probability of hypothesis to strongly phrased advice.
So, the conclusion here is exercise is 68% likely to do you good, and 88% likely to not harm at least.
Unless you have your relevant genome sequences. :-)
My take is that if exercise leaves you feeling cruddy or doesn’t seem to do you any good, then you should consider the possibility that exercise really isn’t helping you rather than you’re doing it wrong. And if you stop, it’s not because you’re a bad person for not pushing yourself enough.
The studies mentioned in the lecture were mostly about cardio. It’s possible that High Intensity Interval Training might do you better, or something subtle like Tai Chi, but the same caveat of looking for actual effects in your life applies.
I think this should probably read “if exercise leaves you feeling cruddy or doesn’t seem to do you any good after 4-6 weeks of doing it.” Exercise feels cruddy to everyone at the beginning when they’re completely out of shape.
Also, I’d be shocked if low-intensity exercise (i.e. an hour-long walk) was harmful to anyone.
I said “leaves you feeling cruddy” for a reason. It’s one thing to feel out of breath or whatever during exercise, and another to feel bad (not just sore muscles) hours later.
Maybe if exercise leaves you feeling cruddy you need to try changing the type of exercise first. Can’t generalize from just one type. Personal anecdote: I never started to like running, despite trying repeatedly, but for the last half-year I’m doing HIIT (pretty much just squats and push-ups) every other day and enjoy it so far.
Isn’t “exercise” an incredibly broad category? Saying “exercise isn’t necessarily good” is a bit like saying “drugs aren’t necessarily good”. Obviously, there are going to be combinations of exercise and people that don’t lead to good outcomes. Broken leg + running a marathon = bad.
I’ve been running up against the prototype vs. whole range issue on this, and I’m not sure how I want to handle it.
On the other hand, I think most people would think that 150 minutes of moderate cardio/week is beneficial (not just harmless) for anyone who isn’t obviously incapacitated.
It’s possible that there are different classes of responders, and it might be possible to detect this from the data they have. I doubt that they’ve done the right statistical test, though- you’d be looking for a mixture of three Gaussian peaks, rather than one Gaussian which has a large enough standard deviation to be represented both around zero and below zero. (Without a link to the paper, I don’t care enough to check.)
It also seems like you could check the test-retest reliability of this- record stats, have people not exercise for X amount of time, then exercise for Y amount of time, then not exercise for X amount of time, then exercise for Y amount of time. Are the people who get worse/better the first time the same as the people who get worse/better the second time? Or do the groups mix like you would expect from it being random noise?
The distribution is not multimodal, but very smooth. Timmons mentions test-retest reliability, but it’s not clear that he means trying the intervention repeatedly vs just measuring VO2Max repeatedly. And his citation is not very precise, but to the Heritage family study. Here they say that response is heritable, which requires that it be individually real.
I am also a little skeptical that they’ve done a good job of separating within person variation and between person variation. However I think a multimodal distribution is the wrong way to look for this (I think multimodal distributions are rare). Your second suggestion is much better.
I suppose; I was taking the view that there were different classes of responders, when it could be the case that each person has some “response to exercise” number that smoothly varies from -.2 to 1, or whatever, in which case you wouldn’t see a multimodal distribution (but you would still see high test-retest reliability).
As a general rule taking health advice from youtube video’s bad.
I would doubt that the 12% adverse responders is a controlled result.
“I would doubt that the 12% adverse responders is a controlled result.”
Exactly, especially when you consider a study that suggests a large portion of the population changes their dietary patterns, usually for the worse, after they begin exercise. So it is possible that the adverse responders began to up their caloric and fatty intake. Hopefully they controlled for change in diet and health at baseline, but considering the rigor of most studies, I doubt it.
Source: http://well.blogs.nytimes.com/2012/04/16/does-exercise-make-you-overeat/
Some of the results from the lecture weren’t connected to eating in an obvious way. What was checked was blood pressure, VO2max, muscle growth, and blood sugar/iinsulin.
It’s conceivable that the last would be affected by eating more high-glycemic food, and that muscle mass might be negatively affected if exercise was lowering some people’s appetites too much.
Actually, the study you cite doesn’t mention that some people ate worse food, just more, and it’s merely enough that they maintained a stable weight.
I didn’t say the study mentioned it, I said that dietary patterns can change due to exercise. The link didn’t say what they ate or didn’t to maintain their weight, and I can imagine that increasing your total calories consumed can’t be completely safe even if it is just for weight maintenance. That is why researchers need to control for changes in diet, changes in overall health unrelated to exercise (people do get diseases for various reasons, as I can imagine would be common to a group of previously sedentary participants), etc., to make sure adverse health changes are directly caused by exercise itself.
Of course, without a link to the papers themselves, it is hard to say how properly controlled the experiments were.
If these variables weren’t properly accounted for, then overall the data should give cause to concern, but is still inconclusive.
Another possibility is that the non-responders were on Statins: Can Statins Cut the Benefits of Exercise?
Diabetes in isolation doesn’t increase CVD that much for men(2.4X), though the relative CVD risk increase for women are 4.3, (in absolute numbers, CVD risk for diabetic men and women are about the same). Also worth pointing out is that type II correlates to a high degree with other risk factors due to the metabolic disturbance, the cluster is termed metabolic syndrome.
I’m curious how good of a job they’ve done separating within-person variation from between-person variation in exercise induced changes in risk factors. I would imagine that within-person variation is pretty large, and the sort of hierarchical modeling you would use to tell these sorts of things doesn’t seem that common.
I’m also very curious whether response vs. non-response to exercise correlates risk factors. If you don’t gain VO2 max from exercise, does that mean you also won’t see a drop in blood pressure?
I’m pretty sure the risk factors are separate from each other at least to some extent. There was some material near the beginning about needing to pay attention to what end point you were using to indicate improvement.