Intermittent Fasting: What I do and why

I have been fasting for 13 years. But what, exactly, does that mean? “Intermittent fasting” encompasses a range of related practices. I am often asked what exactly I do, and why.

Let me tell you.

My fasts have evolved a little over time, but my basic practice has remained the same: three one-day fasts each week.  For the last few years, this has meant Tuesday, Thursday and Saturday. I’m flexible (for example, if I’m meeting friends for lunch on Saturday, I’ll fast on Sunday instead).

Fasting means no calories between dinner one night and dinner the next night (with minor exceptions, detailed below). The fasting period is typically about 23 hours, but I don’t pay attention to precise timing. I eat dinner at the time I would normally eat based on other considerations. There’s a lot of variation (from 19 to 28 hours).

What are the exceptions to my complete fast? First, I typically have a glass of wine after dinner. Whether that’s an exception during the fast or an extension of dinner (meaning that the fast starts later) is a matter of definition; I think of it as an exception. I also take daily vitamins at bedtime, and a small amount (about 5-10 mls., or a teaspoon) of cod liver oil first thing in the morning. I also drink lots of water and black coffee, but don’t consider that an exception, since there are no calories or macronutrients involved. That’s it!

This is a more complete fast than recommended by Krista Varady (in “The Every Other Day Diet”) or Michael Mosley (the 5:2 diet). It’s also more strict than when I started out. For many years, I would have a latte in the afternoon on fast days, but I’ve stopped doing that. I would recommend that beginners eat 200 to 500 calories during the fasting period, at least until they become comfortable with fasting, but I like the simplicity of a complete fast.

So, that’s how I fast. What I do on eating days is not, technically, part of intermittent fasting, and I don’t follow any strict rules, but I’ll tell you anyway. I try to follow a healthy diet. To be consistent with the science behind my reasons for fasting, which are very much about reducing insulin signaling, I avoid anything that will cause a spike in blood sugar. That includes sugar and most starches. People differ considerably in their glycemic response to different foods (Zeevi et al. 2015), so I measure my blood sugar after meals (and snacks), and avoid foods that cause a spike. I hope to write another post about my findings later. If you want to follow my example without going to the trouble of measuring your blood glucose after meals, avoid foods that are reported to have a high glycemic index, and choose unprocessed foods that are high in fiber.

Recently, I’ve started making it a point to have breakfast on non-fast days, and I never eat after dinner, whether it’s a fast day or not. I eat lots of vegetables (broccoli, cauliflower, brussel sprouts, Chinese broccoli, squash), and usually start dinner with a salad, dressed with oil and vinegar. I can’t claim to eat a low carb diet, but I’m definitely eating many fewer carbohydrates than a typical American. I eat a lot of nuts and cheese, and choose fish or chicken over red meat. My favorite desserts (flan, shortbread, ice cream, dark chocolates) have added sugar, but I never eat foods with gratuitous sugar (for example, sweetened drinks or yogurt), and prefer plain versions (e.g. seltzer or plain yogurt).

That’s what I mean by intermittent fasting. It’s a healthy and simple lifestyle. I recommend it.

Fiber Factor for fats and total calories

In an earlier post, I described the "Fiber Factor" as I calculate it, which is essentially

Fiber factor = (% DV dietary fiber)/(% DV total carbohydrate)

Here, I'd like to extend that definition, which is focused on carbohydrates, to total calories.

Fiber factor (fat) = (% DV dietary fiber)/(% DV fat)

and 

Fiber factor (total) = (% DV dietary fiber)/(% DV calories)

In each case, the concept is that one can only reach 100% of the daily value for fiber in one's diet by eating foods with a fiber factor greater than 1.0 (on average). 

Intermittent fasting is sustainable and compatible with a well-balanced diet. Bon appétit!

The Bon Appétit blog recently posted on intermittent fasting under the title “Wellness Tips from Bon Appétit: Intermittent Fasting – Not So Fast.” I appreciate their attention to intermittent fasting, and I agree that caution is always appropriate when considering a new diet. However, they overstate the problems, and conclude that intermittent fasting “is unlikely to be a sustainable way of eating and living.” As someone who has been fasting three days each week for 12 years, I feel confident disagreeing with that. Furthermore, the fact that it is so sustainable is a big advantage relative to most other diets.

They add that “It isn’t as trendy, but eating a well-balanced diet full of plant foods, exercising regularly, managing stress, and sleeping well is your best bet for long-term weight management and overall health.” But I do all of those things while fasting, and I don’t see how intermittent fasting makes them any harder.

They also express concern about the interference of fasting with one’s social life. "if you can’t eat after 5:30 pm, how will you enjoy a dinner out with friends or family?" However, what I do (three fasts of about 24 hours each week) is quite flexible, and I almost never have to decline a social occasion involving food. If my fast is 22 hours instead of 24 one day, it’s no big deal. If I can't fast on Saturday, when I normally would, I fast on Sunday. If I fast only two days instead of three one week, that’s also fine. Freedom from interference with the rest of my life is indeed important to me, and the flexibility of my style of fasting gives me that.

Finally, they are concerned about the effect of fasting on mood. "Mood swings, a.k.a. becoming 'hangry,' along with lack of concentration are side effects of long periods between meals and very low calorie diets." In fact, regular fasting leads to some sort of adaptation (in most people, including me, but perhaps not everyone) so that these side effects never occur. Contrary to what they say, freedom from "hanger" a big benefit of fasting. I never have to worry that I will become irritable if dinner is delayed.

I agree that people should think about how fasting will affect their quality of life. I find that fasting improves mine.

What I do eat: the fiber factor

I do not count calories, but I do read nutrition labels and I pay attention to what I eat and how much.

Intermittent fasting does not require any particular diet on eating days, and the alternate day fasting that I do is, in principle, compatible with any diet. That’s one great thing about it. However, I do eat according to the same principles that motivate my diet, and this post lays out what those principles are and how I apply them.

What are those principles? Genetic studies in model organisms indicate that longevity and healthspan are improved when insulin (and insulin-like growth factor, or IGF) signaling is reduced. Insulin is induced by blood sugar, so reduced insulin signaling can be achieved by fasting, and by avoiding foods that raise blood sugar significantly. I also seek to improve insulin sensitivity so that less insulin will be produced. I have tremendous respect for the body’s many complex systems for telling me what would be good for it. The key here is to listen to what the body is telling me, which usually means eating food that I enjoy.

What do I look for?

Total. I do not attempt to count total calories consumed in a day or in a meal because I think that would be a fool’s errand. The uncertainty is simply too high, and errors accumulate throughout the day. A difference between calories consumed and calories burned sufficient for a dramatic weight change (in either direction) might be only 100 calories per day, much less than the likely difference between my best estimate and reality. But I don’t have to count calories because my body does it for me. I pay attention to hunger. I avoid eating when I’m not hungry, and I try to eat healthy foods when I am hungry. Also, if there is a nutrition label, I look at total calories, and I try to estimate calories when there isn’t a nutrition label. Even though I’m not keeping a total, I ask myself whether I would rather have this food or more of the alternative. The amount that I’m likely to eat each day amounts to a budget, and I’m deciding where to allocate those available calories.  I also keep a log of what I eat each day. It’s not terribly detailed. I just write it down quickly (usually without a lot of detail).

Fat and alcohol. I’m generally neutral on fat, but it’s complex, and much of the science is controversial. While some fats are clearly toxic (e.g. trans fats), others are likely to be healthy (e.g. fish oils). It also seems clear that dietary fat is less likely to lead to body fat than dietary carbohydrates, and there are many nutritionists (the LCHF, low carbohydrate, high fat, school) who recommend a high fat diet. Furthermore, eating more fat generally leads to eating fewer carbohydrates, which is better for reducing blood sugar. What’s more important to my diet is that fats (and alcohol) consumed at a meal with carbohydrates can mitigate the effect of the carbohydrates on blood sugar.

On the other hand, fat is extremely high in calories; when researchers want mice that are obese, diabetic and dying of coronary artery disease, they feed the mice a high fat diet, and it works. So, I try to focus on healthier fats (olive oil, fish and fresh nuts), and I avoid trans fats completely, but I don’t worry too much about the fat in cream, butter, eggs or chocolates if it enhances my enjoyment of food.

Carbohydrates and added sugar. Since my goal is to reduce the glycemic response to food (spikes in blood sugar), I try to eat a low carbohydrate diet and avoid added sugar as much as possible. Ideally, that would mean little or no sugar. However, many foods that I truly enjoy (fruits such as blueberries and peaches, chocolates, custards, shortbread) have significant sugar, and I let myself enjoy them in moderation. I completely avoid sweetened drinks, and overly sweet foods, of all kinds. This includes artificially sweetened drinks; a preference for overly sweet food is a taste that I do not wish to acquire.  

Fiber. A diet high in fiber reduces the effect of carbohydrates on blood sugar, and generally improves insulin sensitivity. Fiber is the primary factor in maintaining a healthy gut microbiome, and many epidemiological studies find that people, and populations, that eat more fiber are healthier and live longer. So, I try to eat foods that are high in a variety of fiber, particularly vegetables, whole grains and some nuts.  I also consult nutrition labels when they are available. When I do I always calculate what I call the fiber factor, which is a ratio between the percent of daily value (%DV) for fiber compared to %DV for total carbohydrate.

Fiber factor = (% DV dietary fiber)/(% DV total carbohydrate)

        = ((g. fiber)/(g. total carbohydrate)) x 12

Higher is better. The maximum possible value is 12, which is what you would get if all of the carbohydrates in the food were from fiber.

For example, if the label lists:

Total Carbohydrate    8 g.    3%

Dietary fiber               2 g.     8%

the first formula gives 8%/3% = 3.67 while the second gives (2 g. / 8 g.) x 12 = 3. The difference between these results is rounding error. The second formula is slightly more accurate, but the difference between the two values is not so great that you should worry about it.

Fiber factor is completely independent of serving size (no need to count calories).

If you find it tedious to do arithmetic in your head while standing in the grocery aisle, just compare the two values.  You want the %DV for fiber to be higher than the %DV for total carbohydrate.

Most legumes have a fiber factor about 4. Nuts have a fiber factor in the range of 6 to 8.  Where the fiber factor is most useful is when assessing cereals, grains and vegetables, because there is a lot of variation.

The following table is informative:

Cereal	Fiber factor		Fruit or juice	Fiber factor
Nature’s Path Flax Plus	2.0		Oranges	3.0
Oatmeal (no sugar)	1.7		Orange juice	0.0
Oatmeal (usual sugar)	1.0		Apples	2.1
Nut and honey granola	1.5		Apple sauce	0.8 to 1.4
Cheerios	1.4		Apple juice	0.2
Frosted Mini Wheats	1.3		Strawberries	2.7
Honey Bunches of Oats	1.1		Blueberries	1.7
Corn Flakes	0.4		Prunes	1.3
Frosted Flakes	0.2		Dates	1.3
Special K	0.1		Peaches	1.0

I look for cereals with a fiber factor close to 2. They are hard to find. Fruits reliably have a fiber factor over 2, but juices typically have a fiber factor of 0. Fiber is lost upon juicing. I have stopped drinking fruit juices altogether.  A really striking case is carrot juice. Carrots have a fiber factor of 3, but the juice has a fiber factor of 0.

Fiber is lost when juice is made from fruits or vegetables. Here we can calculate that carrots have a fiber factor of three but carrot juice has a fiber factor of zero.

Bread or grain	Fiber factor		Vegetable	Fiber factor
Whole grain bread	1.2 to 2.0		Green beans	6.0
Brown rice	1.7		Broccoli	5.0
White rice	0.7		Lettuce	3.0
Hearty crispbread	2.0		Carrots	3.0
Rye bread	0.5 to 1.2		Tomatoes	3.8
Pasta	0.5 to 1.2		Kale	2.3
Sourdough bread	0.2 to 0.6		Corn	1.7
White bread	0.2 to 0.6		Potatoes	1.5

Once I started looking at fiber factor I noticed some surprising things. Some chocolates have a fiber factor as high as 3.6 while others (most) have a fiber factor of 0. The list of ingredients is invariant: cocoa liquor, sugar and cocoa butter, in that order. What varies is how much fiber is retained during the processing of cocoa to make the liquor. What really interests me is that the specific chocolates with high fiber factors are the highest quality (at least, they are the ones that I prefer).  I’ve used this to identify some great chocolate that I would not otherwise have purchased. The best chocolates are also the healthiest. It’s a win-win situation!

Good chocolates can have a fiber factor of 3 or 4, while most chocolates have a fiber factor of zero.

Postscript: I am aware that when the FDA changes its daily value of fiber from 25 g to 28 g, the fiber factor calculated by the two formulae will diverge by that factor (25/28). Fiber factors calculated using the new % DV will be about 90% of the values calculated by the formula using grams.  I don’t think this difference matters that much, because %DV values are often rounded off by at least 10%. So, I will stick with the existing formula based on grams because it is easier to calculate using 12 than 10.71.

Fiber factor = ((g. fiber)/(g. total carbohydrate)) x 12

What matters is that higher fiber factors are better. That always works!

Are the benefits of intermittent fasting settled science?

I have been on the IF diet for over ten years now. I do it thinking that the available evidence supports health benefits, but that the jury is still out. I was thus a little surprised to read the following:

Intermittent fasting regimens ... have similar healthspan-extending effects [as caloric restriction] ... [and] the molecular consequences of fasting, rather than weight reduction per se, may account for at least some of the observed positive effects of caloric restriction.
    - Madeo et al. 2014 "Caloric Restriction Mimetics: towards a molecular definition". 

The articles cited (Anson et al. 2003, Li et al. 2013, Heilbronn et al. 2005 and Klempel et al. 2012) were not new to me. Only the authors' certainty was.  In fact, these authors were simply introducing CR and IF as prelude to a discussion of drugs that might have the same benefit ("caloric restriction mimetics"). The article is a thorough review of the underlying science (with an emphasis on autophagy and protein deacetylation), and ought to be read by any scientist interested in this field.

I should have been skeptical. I should be skeptical. I am skeptical. However, I also found seeing the benefits of IFdiet described as settled science to be reassuring. They are telling me to put aside my doubts and carry on. I will.

Making lifestyle choices with imperfect knowledge about what is healthy

Gary Taubes ends his opinion piece in today's New York Times with a rhetorical question:

Before I, for one, make another dietary resolution, I’d like to know that what I believe I know about a healthy diet is really so. Is that too much to ask?

after having spent the article presenting a persuasive case for the conclusion that "Yes, it is too much to ask."  His key point is that

Because the hypotheses [in nutrition] are ultimately about what happens to us over decades, meaningful trials are prohibitively expensive and exceedingly difficult. 

When I was interviewed recently about intermittent fasting, I explained that we don't really know whether it's healthy or not, but I follow it because the idea that it promotes health makes a lot of sense to me, given what I do know about the underlying science. Gary Taubes makes his own decisions. ("My own vote [for what most of us are eating too much of] sugars and refined grains; we all have our biases.") I don't think he's actually biased. His "vote" is based on a career of looking into the matter. It is not so much a bias as an informed guess.

As a scientist, I am aware that much of what we spend our time thinking about is not proven. In fact, I have argued that nothing in science is completely proven (outside of mathematics, and even within mathematics, a lot of time is spent considering the implications of conjectures and lemmas). There are cases, especially in the physical sciences, when residual doubt is absurdly insignificant, and there are cases, including nutritional science, when there is a lot of doubt.

When policy is developed based on science, it has to be based on the best available evidence and the best possible estimate of how likely it is that the prevailing wisdom is wrong.  If appropriate, we want confidence intervals. It's not always true that when there is smoke there is fire, but when there is smoke, you should definitely stop what you're doing and have a look (or get out).

In any case, I very much enjoyed this presentation of what we don't know, and I recommend it (here). One of the better bits:

Since the 1960s, nutrition science has been dominated by two conflicting observations. One is that we know how to eat healthy and maintain a healthy weight. The other is that the rapidly increasing rates of obesity and diabetes suggest that something about the conventional thinking is simply wrong.

I want to soundly endorse his conclusion:

Making meaningful inroads against obesity and diabetes on a population level requires that we know how to treat and prevent it on an individual level. We’re going to have to stop believing we know the answer, and challenge ourselves to come up with trials that do a better job of testing our beliefs.

The ELF diet (Eat Less Food)

Intermittent fasting is a good way to lose weight, and in my case, I think that fasting is good for about 10 or 15 pounds.  However, I've been fasting since 2004, and after a while, I gained some weight back and thought that it would be a good idea to lose some.  For the last few years, I've been watching my weight using the ELF diet, which involves Eating Less Food. The person who introduced me to this acronym did it as a joke (as in "I'm following the ELF diet, where you Eat Less Food, haha"), but I've given it some thought and see two useful aspects.

Eat less than you want. The amount of food that most people eat is dictated by appetite, and their weight, whether it is more than they would like or exactly right, is relatively stable. The body has an exquisite ability to regulate appetite, so if you want to lose weight you have to be a little bit hungry much of the time. I find that paying attention to appetite, and eating just a little less than I want, is an effective way to lose weight.

Eat less than you have been eating. I've read recently that people who record what they eat using smart phone apps are more likely to lose weight. This makes sense to me (although I myself just record what I ate, in words, at the end of each day). If you eat seven cookies, that's a lot, but it might be less than the eight cookies you ate at a similar event the week before. I find keeping track and eating less to be much easier than counting calories. There is a lot less arithmetic, and fewer opportunities for wishful thinking to play a role.

I also find that it's useful to keep track of my weight, so that I know which foods are associated with weight gain and which are not.  Keeping track of what I eat and what I weigh each day makes it possible to predict the effect of different foods, and that is very useful.

Do artificial sweeteners lead to weight gain?

The personal genomics firm 23andMe recently released the results of a user survey indicating that the consumption of diet soda is very highly correlated with high body mass index (BMI).  You can read this surprising result here.

people who drink diet soda five or more times a day have an average BMI nearly 5 units higher than those who never drink diet soda. 

Of course, this could be because people who are obese are more likely to choose diet soda.  However, it fits very well with something I have suspected for some time now, that consumption of artificial sweeteners leads to weight gain.

A 23andMe survey revealed that consumption of diet soda is correlated with BMI.  (The cartoon is also from their site.)

Although I know of no data in support of this hypothesis, I will explain two reasons why I think it likely.

Taste.  Consumption of sweet foods leads to a preference for sweet foods.  This was directly addressed in a study by Sartor et al. (Appetite 2011), who found that "overweight/obese individuals are more implicitly attracted to sweet" and "one month of soft drink supplementation changed sweet taste perception of normal-weight subjects."  My own preferences affect my thinking on this.  While I truly enjoy the sweetness of a ripe peach, fresh sweet corn or dark chocolate, I find many popular foods (including soda and iced cupcakes) to be too sweet and truly unpleasant.  The idea here is that when someone consumes artificial sweeteners they get used to sweet flavors and come to enjoy sweeter tastes, which ultimately leads to the consumption of more calories.  For more on taste, including possible links to obesity, see the Outlook on taste in this week's Nature, especially a summary of research  on taste and obesity. 


Homeostasis. This is a similar idea, but involves unconscious processes.  The gut is filled with sweet taste receptors (Dyer et al. Biochem Soc Trans. 2005). Although I don't know what they are doing there, it seems likely that information from these receptors is used to regulate appetite, or metabolism, or both.  If your gut is constantly full of sweet flavors, then they will be ignored.  The sugars released from real food will not make you feel satisfied, or will be improperly dealt with in some other way.  The body's homeostatic mechanisms are complex and I don't claim to understand them, but I do suspect that artificial sweeteners disrupt the natural response to food and contribute to craving and binge eating.  

Liping Zhao on caloric restriction, gut microbes and aging

I just heard a very interesting talk from Dr. Liping Zhao, who is Director of the Laboratory of Molecular Microbial Ecology and Ecogenomics, and Associate Dean of School of Life Sciences and Biotechnology, at Shanghai Jiao Tong University. He is also associated with the new Shanghai Center for System Biomedicine. He described a consistent effect of diet (high fat vs. normal or normal vs. low fat) and caloric restriction on the composition of mouse gut microflora. I was impressed that even within a treatment, the composition of gut microflora was an indicator of lifespan. This correlation is interpreted by him in terms of a cumulative effect of anergy on health. His working hypothesis is that excess calories feed gut bacteria that cause inflammation, leading to a variety of health problems. A freely available paper from his group was published in PNAS in 2008 ("Symbiotic gut microbes modulate human metabolic phenotypes"). I expect that the results that he presented today will be published somewhere soon, and we're certain to see a lot more about the gut microbiome and health in the years to come.

"The Youth Pill," a new book on aging by David Stipp

I haven't read it, but a review in the Economist (July 24, pg. 80) alerted me to a new book on aging, "The Youth Pill: Scientists at the Brink of an Anti-Ageing Revolution"). If the review is to be believed, this book successfully avoids many of the traps that anti-aging books fall into (such as "embracing decrepitude"). David Stipp is a medical writer for the Wall Street Journal.

One meal each day vs. intermittent fasting

People in the intermittent fasting group on Yahoo are trying out many variations of intermittent fasting. The most popular appears to be daily (or frequent) fasts of greater than 18 hours, without caloric restriction. This normally means following the Fast-5 program of eating all food within a five hour window each day. The Fast-5 approach avoids longer fasts, but provides many more of them. Following this approach, one is fasting most of the time. There has been little scientific data on the relative benefits of different variants of intermittent fasting, so many of us are keen to learn which provide the greatest benefit. Indeed, some remain unconvinced that intermittent fasting provides any benefit at all.

The Mattson group has just published a study of the effects of reduced meal frequency on glucose regulation (Carlson et al. 2007, Metabolism 56:1729. "Impact of reduced meal frequency without caloric restriction on glucose regulation in healthy, normal-weight middle-aged men and women"). They found that people on a diet that involves a single meal each day show elevated fasting glucose levels and impaired glucose tolerance. This is in contrast to an earlier study (Johnson et al. 2007, Free Radical Biology and Medicine 42:665) that tested alternate day calorie restriction and found improved diabetes risk profiles (also see the Johnson upday downday diet, which is promoted by the same James B. Johnson). The authors suggest (in the Carlson et al. paper) that the key difference is an overall reduction in energy intake.

That may be true, but there are two other important differences between the Fast-5 approach and alternate day fasting (including my method of fasting three times each week from dinner one day until dinner the next day, which is less rigorous than a full bedtime-to-waking fast of 30+ hours). First, even a dinner-to-dinner fast is longer (roughly 23 hours, on average, as opposed to 19 or 20). If the benefits of fasting kick in after 12 hours or so, this difference could be more significant than one might otherwise think. Second, the consumption of all daily calories within a five hour window is very intense, and ad lib eating over 24 hours need not be. Between fasts I usually eat dinner, a late-night snack, breakfast, lunch and dinner. Less intense eating means less insulin secretion and less food in the stomach at the beginning of a fast, which would further increase the effective length of the fast.

Perhaps the longer fasts are more effective. Further research will tell, and it's being done.