Sunday 21 January 2018

The variability diet



'Tis the time for dieting. Let me jump on the band wagon with a completely new diet that fits to this blog: the variability diet.

Antipodes can skip this post or stay for some interesting science ideas; countries in the Southern Hemisphere apparently do not have a seasonal cycle in dieting.




The National Weight Control Registry is proud they found ten thousand people who managed to keep the weight off for a longer time. Ten thousand people world wide sounds to me more as if most diets hopelessly fail on the long term.

It could naturally be that the obesity epidemic in the West is an epidemic of sloth, a failure to count calories as well as lions do.

Oh, wait.

I would venture: we need fresh ideas.

Most diets are defined by averages. The right amount of calories, the right macro-nutrients (carbs, fats and proteins), in the right number of portions per day. In observational studies, scientists compute the average number of calories eaten and calories burned, they compute the average fractions of every macro-nutrient, maybe even the average amount of vegetables and animal foods eaten, the average number of hours of sleep, the averages of this and the average of that. And even if they do not explicitly average, their analysis methods (for example, multiple regressions) will only notice differences in the average.

The main exception from prescribing the averages of a diet is the advice to eat a diverse diet or to eat many different colours of fruits and vegetables. For example in the USA one recommendation is: "A healthy eating pattern includes: A variety of vegetables from all of the subgroups—dark green, red and orange, legumes (beans and peas), starchy, and other."

To be honest, "a variety of" is a somewhat unsophisticated way to describe variability. First of all it is rather unspecific. Other formulations often ask for more variability, but just like for the average, you can have too much or too little variability and an optimal amount in the middle. But lets be generous and assume the typical variability is way below optimal.

The biggest problem is that for variability you always need to define an averaging (time) scale. The red and the blue symbols in the figure below have the same total variability (a standard deviation of one). However, as the thick line shows, the blue line has much more variability on longer averaging time scales, while the blue values vary less from day to day.



Daily eating a wide variety of vegetables of vegetables from all over the world is nowadays possible and produces a lot of variability on short time scales, like the red symbols. But on long time scales it lacks variability ($). My father complained that in his youth he had to eat beans for weeks on end when the beans in the garden were ready for harvest. That was a diet with not much variability on short time scales, but a lot on longer time scales, like the blue symbols.

There is a diet that mimics this, the seasonal diet. For out of touch city folks like me, there is even a seasonal food calendar; I have one on the inside of a kitchen cabinet. This diet is mostly promoted as a diet that saves you money and it good for the environment. Sometimes it is advised for ensuring the food is healthier and tastier because fresher, but maybe the variability is also helpful.



What is the right time scale? I do not know. One reason to write this post is the hope that a nutritional scientist is inspired to take variability into account. When it comes to carbohydrates (bread, pasta, rice) it may be good to once in a while empty the (glycogen; animal starch) stores and train your body to also use fat for fuel. The liver has about 10 to 12 hours of glycogen in store. I wonder whether this is related to the German advice not to eat after 6pm or the Italians who eat a lot in the evening, but hardly eat any breakfast. These two national recommendations seem to conflict, but have a fasting period of over 12 hours in common.

Intermittent fasting seems to be an effective way to lose weight, at least for people able to do so. There are many varieties. Some people only eat during a few hours a day, every day. Others once in a while do not eat for 24 hours (breakfast to breakfast; skip lunch and diner) or 36 hours (diner to breakfast). I like this method, at least in summer because I get cold after some time when my metabolism goes down. Women often do not like it.


"Ancient humans ate a large variety of foods, which is why we are adapted to so many. Human variation is high though, since our lineage has become so populous and geographically wide-ranging."
Melissa McEwen


Many diets have a honeymoon period of one or two months in which they work like a charm (raw food diet; low carb diet). A blog called 180 Degree Health claims to have success by breaking all the healthy eating rules. Apparently diametrically opposed diets work in the beginning. If you are lucky your diet works for about half a year, but typically the weight and the health problems come back.

From the variability point of view a new diet is initially a change, variability, in the long-term a diet often makes your food more monotonous.

This honeymoon period suggests that another interesting time scale is a month to half a year. This honeymoon period was what provoked my to start thinking about a variability diet. Why stick with a diet for a year, why not try 12 diets for a month and benefit from the honeymoon period 12 times?

The honeymoon period is also why diet studies without a control group are a problem.


Randomized Control Trials, difference in success after 6 months, similar and less success after 12 months.

Variability in sports

What is the right time scale? Maybe all of them!? Nature is characterized by variability on all time (and spatial) scales.

The importance of variability is broadly acknowledged in sports and sports training uses several time scales, from hours to years. Sports training is based on periodization. A simple [[periodization scheme]] for a beginning runner could be: one has microcycles of one week (one hard run a week), mesocycles of about one month (one more relaxed week a month) and macrocycles of a year (preparation, competition and transition (rest) phase). If you add to this a 4-year cycle for Olympic athletes and a career plan for a professional athlete you start to think in variability on all time scales.

In weight training a similar concept is to include a deloading period/week once in a while in which you take it easy, which helps to keep getting stronger in the long run. Weight training fans can talk for hours about how fast exercises should be performed, how many repetitions, how many sets and how often per week one should go to the gym. There we have four more time scales. Five if you make one exercise a week extra hard.

Intermittent fasting could be seem as weight training. A fruitarian diet may be the equivalent of lying in bed all day; if you can handle the sugar load, fruits normally eaten hardly contain any anti-nutrients you body needs to defend itself against. On the other side, a very low carb may for some be the equivalent of running a marathon every day. If you think in terms of exercise, you would set a stressor by eating very low carb for a short period and then recover and become stronger while eating a mixed diet including fruits. No idea whether this specific example works, but I wonder why people do not think in these terms when in comes to nutrition.

Reasons why variability may be healthy

A healthy gut microbiome, that is the ecosystem in your intestines, is important for health and weight loss. A human has more mass (and stores) than a microbe. Missing nutrients thus hurt microbes more than us and fasting gives our immune system the possibility to fight bad germs more easily. This is a reason why we often do not feel like eating when ill. This also suggests that the microbiome may be the most sensitive to variability and the main target of a variability diet.

The biodiversity of the microbiome is linked to being overweight. The causal relationship may go both ways. Transferring the microbiome of a fat mouse to a skinny mouse makes it fatter, but losing weight also changes the microbiome. The obesity epidemic and other metabolic deceases in the West may be related to a reduced intestinal biodiversity in the West. This biodiversity has a seasonal cycle and the diversity increases on a calorie restricted diet, which gives hope this may be reversible.

Microbes are an ecosystem and their composition will as such highly likely be determined by variability (pioneer populations, tidal plain versus salt marsh; cactuses need to be dry regularly to fight root decease; palm trees grow where the minimum temperature is above zero degrees, fire keeps the savanna open, ...).



Variability also makes it easier for the body to learn which food does what to your body, especially when the short-period diets are relatively simple to get more long-term variability. Without paleo blogs, I would never have noticed that I do not handle grains well because I never went without grains for a long enough time to notice the difference.

My experiment

How could such a variability diet look like? Except for mimicking the seasonal diet your grandma ate, I have no idea. But this is the experiment I tried myself. I picked a period of 4 days and relatively large changes in diet. The 4 days nicely fit to my 2 times a week grocery shopping rhythm. These are some of the short-term diets I tried. The diets probably look funny to most because I do not eat grains.
  • 4 Days low protein: Fried potatoes, fruit meals and a large vegetable stew without meat, but with butter.
  • 4 Days low fat: I was at a conference and thought that low fat would be the easiest diet variety. Luckily they had lots of fish. Otherwise, vegetable omelets, meat, greens, yogurt with fruits.
  • 2 Days low carb: Brisket, fried eggs with cream, too much cashews, pork chops with red cabbage.
  • 7 Days deloading: Whatever my body wanted.
  • I started this 4 day period planning to eat yogurt and fruit. But after four meals, I just could not eat it any more and put in a low carb day. Repeat not to waste the fruits.
  • 6 Days a diet which was intended to be (almost) all animal products, but without dairy, so I used olive oil. That turned out to be too low carb, so I have added one fruit to most of my meals. This period was a bit longer than the others. It felt good and I made a delicious soup from whole 2-kg chicken, which takes some days to finish, if you would like to eat more than just soup.
  • 2 Transition days.
  • Started with a potato diet, with some vegetables and nut butter for taste. I mainly baked the potatoes in a little butter or olive oil. Plus about 2 fruits a day.
  • 5 Days low carb, mainly animal products, meat, eggs, heavy cream, butter, bacon, and one or two fruits a day. I had planned to do this one week, but my temperature started to drop (cold hands, feet and ears).
  • 3 Days my normal diet.
  • A low protein week: backed potatoes with butter and cream and fruits.
  • 2 Weeks of meetings and a conference in sugar and flour capital Vienna, These weeks I planned to eat whatever is available. To my surprise you can also get very good food in Vienna, lots of offal and grass-fed beef was available in the restaurants (Universitätsbräuhaus (University brewery, Unibräu) & Servitenwirt) I visited.
  • 1 Week my normal paleo-ish diet to recover from the meeting.
  • 3 Starch days, a lot of rice pudding (rice cooked in hay milk and butter, served with a little sugar and cinnamon or backed apples) and one normal meal.
There are many more things one could try. Also trying longer periods would make sense.
  • High Acid (vinegar)
  • Low Acid (My mum used to have chalk for cooking, no idea how she used it.)
  • No dairy
  • Salty or salt free
  • Pig out
  • Salad, raw vegetables
  • Drink a lot, or little
  • A lot of vegetables or just empty carbs.
  • Avoid nightshades, citrus fruits or [[FODMAPs]]
The good news is that I lost weight and felt good. The bad news is that I was not able to continue the diet. Inventing one or two new diets a week is exhausting. Thinking so much about what to eat is not my thing. So I went on with just occasional intermittent fasting (which requires no planning) and stopped doing that in winter.

It would be great if there were a traditional fasting celebration in spring to remind you to start with intermittent fasting again.

So you could say my n=1 experiment failed, because the main thing of a working diet is being able to stick to it. But maybe that is fixable (is there an app for that?) and people are different.

It is some time ago I tried this. I think I will try again. 'Tis the time of the year.

Related reading

Is obesity bias evolutionary?

Davenport and colleagues: Seasonal Variation in Human Gut Microbiome Composition.  PLoS ONE, 2014.

Mark's daily apple: The Pitfalls and Limitations of Self-Experimentation.

Are Diets Just Placebos?

Dietary intervention impact on gut microbial gene richness. This study suggest that a low fat diet good for microbiome diversity. I could imagine it was just the change.

Richness of human gut microbiome correlates with metabolic markers.

Photos: Pumpkin recipes by zsoolt used under a Creative Commons Attribution-NonCommercial 2.0 Generic (CC BY-NC 2.0) license; yellow snap beans photo taken at Farmers Market by Alice Henneman used under a Creative Commons Attribution 2.0 Generic (CC BY 2.0) license; Pepper Medley by Jitze Couperus used under a Creative Commons Attribution-NonCommercial-NoDerivs 2.0 Generic (CC BY-NC-ND 2.0) license; Grande Ronde Wild and Scenic River by the Bureau of Land Management Oregon and Washington used under a Creative Commons Attribution-NonCommercial-NoDerivs 2.0 Generic (CC BY-NC-ND 2.0) license; Fennel by Alice Henneman used under a Creative Commons Attribution-NonCommercial-NoDerivs 2.0 Generic (CC BY-NC-ND 2.0) license; seasonal by Lisa Ouellette used under a Creative Commons Attribution-NonCommercial-NoDerivs 2.0 Generic (CC BY-NC-ND 2.0) license; Sampler tray of Starbucks new Mocha Toffee Latte by urbanbohemian used under a Creative Commons Attribution-NonCommercial-NoDerivs 2.0 Generic (CC BY-NC-ND 2.0) license.