The Problems with Food and Exercise Studies
nytimes.com
nytimes.com
To be clear what I am saying, I do not think that other sciences are a waste of time in general. But I think that their methods are too much inspired by physics. I think that we need a much more cautious approach to the scientific method and a lot more robust tools. In a way, it looks to me like building a car by taking inspiration from Formula 1, when the task at hand is crossing rivers and navigating rough terrain.
[1] Previously I assumed that it is just complicated, that all I see from studies is the headline "Coffee is/is not healthy" and if one would look into the studies I would find that a better picture is Coffee has some positive and some negative effects.
[2] An anisotropy in the arrival directions of ultra-high-energy cosmic rays. https://arxiv.org/abs/0711.2256
The first issue is that it's extremely difficult to run a long term controlled study on nutrition. Adherence to diet interventions is difficult to measure and it drops off over time. As a result most nutrition studies are epidemiological cohort studies which are not particularly good at determining causality.
The second issue is a question of funding. Diet and exercise interventions are not something that are typically monetized be the pharmaceutical industry. As such there is little private sector funding going into nutrition studies. Couple that with the fact that they're extremely expensive and you have a serious deficit in studies.
There's meat and also fat that doesn't (taste good). It's not just the meat/fat itself, it's also the combination with other foods and the preparation that can change completely how my body perceives a food. Since those "feelings" have developed for a reason - we didn't have science to tell us what is good to eat for the 1st million years of humans (and pre-humans) - I dare claim it is significant for ones long-term health.
But studies don't make any such distinction. To assume everybody eats what tastes best is a stretch, not just because circumstances in our stressful lives, but also because the brain can only ask for food that it knows, and I remember the stories from that British chef who showed American kids tomatoes and potatoes and they had no clue what that was, they only knew processed food.
So how can you seriously leave it at "they ate x amount of meat" when such a person eats "meat" compared to someone who had opportunity to taste "traditional" (pre-industrial) foods and prepares him- or herself a good piece, including a good combo with other stuff and well-prepared?
Am I seriously to believe a McDonalds burger's "meat" is correctly put into the same place - as "meat" - as a real meal?
Of course, if you acknowledge this difficulty you have to give up on studies because I see now way to reliably collect that information on a scale large enough and long-term enough for a useful study. As the article suggests, actually (to give up going down this path). The story of that guy looking for his car keys under the street light even though he lost them somewhere else comes to mind, in the desire to study this at all this seems to be what happened.
It isn't about giving up as some readers took away from that paragraph, it's about choosing another path. For example, get down to understand what's actually happening instead of high-meta-level studies and see if that gives us new ideas.
Oh and here's a nice article related to the subject: http://www.vox.com/2015/3/23/8264355/research-study-hype (I hope nobody tries to count the dots on each side of the plot to use that to draw conclusions... it's an example and incomplete and even if it had dots for every single study could not be used for that purpose)
I get that the laws requiring calorie content on labels have lead to finding an efficient way to measure and provide the numbers, but it is obvious to me that the measuring system is completely broken or very inaccurate at best.
http://www.scientificamerican.com/article/how-do-food-manufa...
That's like saying "once I learned that voltage was measured by the deflection of a coil in a current (a practice started in the 1800s)" or "once I learned that heat was measured by the expansion of a liquid in a glass tube..."
You could argue that the heat definition of calorie doesn't really reflect the health function of the food (for example, maybe glycemic index is more useful), but I'm not sure why complaining that a calorie isn't a calorie is really insightful.
Bioavailability / metabolizable energy varies not only in the small, ie composition and cooking of food, but in the large, as long run diet choice changes gene expression.
I used to work in aquaculture, where we measured the amount of food going into a fish (via the bomb calorimetry that we're talking about, directly), the amount of energy spent by the fish in respiration (by measuring carefully increases in water temperature in their tank), and then we measured how much energy the fish had at the end (by burning the fish).
And quite simply, calories in was equal to calories out. It was a good relationship within very acceptable experimental error.
Now of course, the fish were growing very rapidly and we could control their food exactly. Humans, growing over a longer period, can be subject to different metabolic pathways, and we can't measure their respiration directly of these pathways [Edit: this is where the gene expression you mention comes in] - this is a far, far bigger source of error than the calorimetry, so calories alone can be less predictive of outcome. But this doesn't take away from the basic definitions of thermodynamics.
It's like we didn't burn the wood or store it for fuel, we used it to repair the walls of the house.
Right? Or am I confused.
Since fat and protein each have different amounts of calories stored per gram, that's why if you want to get really accurate about weight loss/gain, you track your body's %fat & %muscle and not just your weight.
Edit: to clarify your analogy, just because we use wood to build a part of the house, doesn't mean we won't take it and burn it if we're desperate for heat and have used all the other fuel.
2) All other things being equal, changing "calories in" will make you lose/gain weight. The trouble is, all other things are not equal - simply changing how much you eat will significantly change those other things, it has an effect on your metabolism and eagerness for physical activity, thus directly affecting also "calories out" if you don't carefully monitor that and work to keep that stable.
3) All other things being equal, changing "calories out" will make you lose/gain weight. The trouble is, all other things are not equal - simply starting/stopping working out will significantly change your natural appetite, how hungry you are and how often you're hungry, and which types of food you have cravings for, thus directly affecting "calories in" unless you carefully track what and how much you eat everything and actually do keep that schedule exactly the same.
Are we accepting that the complex biochemical mechanisms of the cells are identical to burning things in an open flame?
And, that every living human's body processes all foods in exactly the same way, regardless of age, ethnicity, gender, time of day etc? Just like the open flame would?
I get that it's the best scientists could think of 100+ years ago. But putting averaged numbers derived from this absurd method on labels as if they are how my (!!) body works, and deriving all kinds of "food science" conclusions and recommendations from it is imo willful ignorance in the name of convenience.
That is, it's assumed it does, but we just can't pin it down because of poorly done studies.
My guess is that that's true, but the other very real possibility is that our bodies are designed to process a very wide variety of foods in a very robust manner, such that as long as you're not eating something obviously poisonous, it doesn't matter as long as you're not depriving yourself of something important nutritionally (as in scurvy).
Similar arguments can me made with regard to weight and exercise, although there it seems like the effects are clearer in certain ways (e.g., I doubt anyone would argue that being in shape isn't better than being physically deconditioned).
There are indications that various health conditions can impair glucose regulation, and those affected would tolerate less dietary carbohydrate than healthy people. I'm aware of two conditions, elevated cortisol [1] and elevated blood iron [2] (which can be caused by regular alcohol consumption, btw).
[1] http://www.ncbi.nlm.nih.gov/pubmed/11724664
[2] http://www.ncbi.nlm.nih.gov/pubmed/26618110
Also, type 1 diabetes of course.
I find the suggestion at the end of this article truly pathetic: "Some medical experts say the problems with lifestyle studies are so overwhelming — and the chance of finding anything reproducible and meaningful so small — that it might be best to just give up on those questions altogether." This is typical big media bullshit pandering to the inert.
I mean come on... How about we don't give up on questions? How about we learn to ask the right questions instead? How about we spend more time training and studying our own bodies than we spend reading our Facebook feeds?
> the majority of the population lazily look for simple formulas
Great! After all, it's not that different from "a good programmer is a lazy programmer" and "keep it simple, stupid" -- if at all possible, look for simple solutions and quick fixes.
> This is typical big media bullshit pandering to the inert.
"Epidemiology--is it time to call it a day?" (http://ije.oxfordjournals.org/content/30/1/1.full), an essay that comes to pretty much the same conclusion as this article, was published in 2001 in the International Journal of Epidemiology. As far as I know the International Journal of Epidemiology is not big media.
> I mean come on... How about we don't give up on questions?
Rhetorical questions don't count as arguments.
> How about we learn to ask the right questions instead?
Observational research is intrinsically very, very hard. Statisticians are working at making it more reliable (especially in the context of causal inference: https://www.hsph.harvard.edu/miguel-hernan/causal-inference-...), but there's only so much you can do with data that is full of noise and confounders.
> How about we spend more time training and studying our own bodies than we spend reading our Facebook feeds?
You don't get to decide other people's priorities in life.
Oh boy. I'm not OP but you and I could easily have the same altercation. Quick fixes and simple solutions are two different tools, and conflating the [two] leads to endless disagreements and hurt feelings.
Pascal and de Saint-Exupery both have famous quotes that apply to the distinctions between the two.
Exupery has the one about perfection being achieved when there is nothing left to take away. I confess you have to unpack that one a bit to turn it into a commentary on simplicity and how it's challenging to achieve.
There's the one attributed to Einstein about 'as simple as possible but no simpler', and for some reason I occasionally swap the two in my head. Any problem is easy to solve if you misunderstand it properly.
No, but let's not bullshit ourselves and claim half the population is not dangerously addicted to the mental equivalent of "junk food", ex: Facebook.
Don't want to be fat? Then don't be fat. Want to be fat? Wasting hours on Facebook will help you achieve your goal.
Some people try to go their whole lives without showing up for anything. You should be encouraged to get off your ass and try.
Convincing yourself that everything is pointless is a self fulfilling prophecy.
I understand what you're saying, and they are certainly valid thoughts. I might even agree, largely. But there's not much objective in what you expressed.
What would it take for you to change your mind on this?
There are certainly limits to inductive reasoning. And there's probably a limit when it comes to understanding complicated systems with observation. It's valid to ask whether we have reached some of those limits and how we can know in general.
The same smart researchers (and the excitement of the general public) could be focusing on more tractable problems. Maybe carbon capture or inexpensive desalination.
I don't know about population level effects but the Internet (more random blogs than the big media) has delivered a super simple formula (cutting carbs, lifting weights Starting Strength-style) which works every time I stick to it and in mere weeks. Perhaps it's not the demand for simplicity driving this nonsense but the opposite -- trying hard to sound smart at the dinner parties and avoiding obvious answers.
Then again, my workout is not nearly as fun as CrossFit, and my conversations about it rarely last longer than two minutes.
I find people, in general, to be goddamned fucking annoying because of behaviors like this.
But a simple answer doesn't mean the solution doesn't require hard work or sacrifice.
It's not wrong to desire maximum results with minimum effort. But some things just don't have shortcuts as good as the others.
* Small sample sizes
* Self-reporting and diary studies
* Many uncontrolled variables
* Inconsistent interventions (not isocaloric, not volume-equivalent or tonnage-equivalent etc)
* Impossibility of performing blind trials to control for placebo
First, "it's hard and we can't control everything" isn't the same as "it's impossible to control anything".
Second, large sample sizes are preferred exactly to help even out the variables that are uncontrolled or, especially, unknown.
That it's impossible to perfectly model reality is not news in any scientific field.
In any case, I was talking about grossly observable variables such as "does this person exercise at all?", or "do they smoke?" or "do they drink?". You'd be surprised how many studies don't bother to check simple factors like these.
For example, you could do a large study where you measured the correlation of wine drinking with obesity. This study would at least tell me interesting information, even though I wouldn't know exactly if there is a causative relationship. But if the correlation was high and I wanted to lose weight I would stop drinking wine and see what happens. Maybe there is a confounding variable, maybe there isn't, I'd take that chance if I wanted to lose weight. This is much better than the current state of food and exercise studies where none of the information is reliable because of cherry picking stats and low sample sizes. As detailed in the article, we don't know anything about food and exercise from studies besides the basic physics that to lose weight you need to use more calories than you take in.
For years tobacco companies argued that correlation is not causation.
On a slightly offtopic note; the first results of NuSI have not been published but in this video, one of the researchers discusses the findings which may (or may not) come as a surprise:
_No Metabolic Advantage for Ketosis Found_
There's also the mentality that every aspect of what you consume must be scrutinized intensely, or else you risk destruction. We probably get that from our more adventurous forebears eating poisonous stuff, through religion.
Where I get pissed off is that health insurance companies will create policy around research based on shitty metrics. They really only seem to be capable of understanding what might help when things go REALLY wrong - as in you require constant drug use or surgery, because hey, they have patient data on that. But in terms of preventive medicine, insurance companies can't figure out what to do. Thus, it's completely random as to what you may or may not get covered for preventative techniques - like doing physical therapy to prevent surgery instead of recover from it. And if you have to pay out of pocket for this sort of stuff, damn... it gets expensive these days.
But people lived before peer-reviewed research. They collected information somehow and used it somehow and it worked to varying degrees of success, which has been reflected in the rise and fall of various societies and in the quality of life in those societies. The challenge is not to abandon peer-reviewed research, but to integrate it effectively with other sources of knowledge.
That is - the outliers often differ from the mean in a biological system because they are fundamentally different; they have different underlying biology, and while we might be able to place everybody on the same normal distribution, any conclusions we draw based on that normal won't apply to individuals whose private variation makes them behave differently.
A multi-year study with thousands of participants is possible. If you aren't worried about perfect data, wait a decade and start consuming what I'm sure will be many public life-streaming feeds.
I guess livestreamers are another data point, so in a perfectly pure framework, they can't actually make your epistemic situation worse, no matter how bad the selection bias is, but I don't see how they're going to make the situation that much better.
While that population is going to skew wealthy if the study involves the participants making a purchase, that can be counteracted by seeking out under-represented populations and asking and/or incentivising them to participate.
Not exactly because we don't know all the factors to control for. You end up with a bunch of correlations but often it isn't clear why. Are people who eat X foods Y because they eat that, or because the common genetic background of people who eat X are Y? Or maybe it is because before they were born their parents did Z?
Track enough variables in a population and you're bound to get correlations.
What the hell does this mean?
I think a safe intuition is lots of cookies and chips are probably bad for you, and mostly vegetables, fruits and protein are probably better for you.
Don't worry about the best diet, just try to be sensible.
The way those concepts are fleshed out in the story is pretty compelling: http://mobile.nytimes.com/2007/01/28/magazine/28nutritionism...
- Avoid junk food (if it comes in a vacuum sealed bag, it's junk food)
- Avoid sodas, juices, and other high sugar drinks. If you choose to indulge yourself, don't finish the whole thing - the portion is probably too big anyway.
- Exercise at least a few times a week. Meaning, break a sweat and keep it up for at least 30 minutes, preferably an hour. Eat (or drink) protein afterwards.
- Only eat when you're hungry. Stop eating when you're full.
- Listen to your body.
I hate this advice. When I'm exercising, my body is telling me that I'm tired and that I should give up. When I'm eating chocolate, my body is telling me that it's delicious and I should have more. Sure, there are things that you probably should listen to your body about, but there are just as many that you should ignore and overcome.
> Reasons for this issue include mining operations that have left not much arable land; as a result, much of the local diet is of processed, calorie-dense, imported food such as Spam or corned beef, rather than fresh fish, fruit and vegetables
This actually isn't remotely true. It's far cheaper to buy more nutritious foods from the grocery store (especially in bulk) and then cook your own meals: http://www.leannebrown.ca/good-and-cheap.pdf
You can't have organic beef every night, but you can nutritionally and calorically out-gun fast food for the same (or a smaller) amount of money.
But if I'm poor, the difference between a ten-pound bag of potatoes and one fast-food meal can really matter.
A also believe that healthy meals can potentially increase my lifespan and quality of life within that lifespan. (I don't know if science backs this up). When I cook for myself I'm assured that cheap by unhealthy foods are not hidden in my meals - even the most expensive restaurants won't do that. (Trans fats and excess salt are two examples)
It's easy to see how people get out of shape.
"I don't have time" is a somewhat common situation for anything that's important but not #1 important, since a single priority area can easily sometimes take all the waking hours in a week and more than that (on account of e.g. "sleep debt"); and getting an hour for something you want means taking that hour away from something you need.
- Long work hours, mostly sitting
- Long commutes
- Portion sizes way too large (in restaurants I am often full after appetizers)
- Eating at the desk, eating late at night, skipping meals
- TV watching
- Nothing reachable by walking
- Not enough sleep
- Not enough vacation
- Most foods in stores that are labeled "healthy" are not healthy
You need a certain level of stubbornness to realize that the generally accepted lifestyle is so far away from reasonable that almost everything that's considered "normal" needs to be changed.
That said while the default is bad you really can live a healthy lifestyle without to much work. It just takes planning to avoid spending 10-15+ hours a week in a car etc.
A couple years in and they all hit the same wall I had, clothes stopped fitting, too much stress eating. No exercise, sleep problems.
By the time I left we occasionally had almost the entire team out walking to lunch together.
There have been studies in animal populations where certain members of the population are observed to keep returning to opium fields while others will come just once and then never return.
Many of us can think back to our own childhoods and recall that, depending on your geography, the choices of what you could buy at the store were fairly limited in terms of fruits, vegetables, and produce. Getting fat wasn't something people used to worry about.
Since lack of discipline is a character flaw, the confirmation bias is that since I can't follow through on my grand gestures that I'm a bad person and I should just give up. My fate is sealed, better luck next life.
The minority's view is that discipline is a lot like a muscle, and that you need to be able to identify ways to exercise it without injuring yourself. Start small, work up, be aware of your limits. And yes, you're going to feel ridiculous at first because what kind of a baby can't safely do X and Y?
I have no idea why people need to be told this or what it even means.
By definition, if you're full you can't eat anymore and you've already crossed the threshold of hunger some time ago.
It's mindbogglingly bad, non-science.
You probably only need n = 5 for a confident answer to the question "What happens when you are shot in the head?"
People are looking for smaller effects than before, and the incentive to publish any and all significant results is just so great that you get tons of noise.