• Your life and health are your own responsibility.
• Your decisions to act (or not act) based on information or advice anyone provides you—including me—are your own responsibility.


Can You Really Count Calories? (Part V of “There Is No Such Thing As A Calorie”)

Caution: contains SCIENCE!

(This is a multi-part series. Go back to Part I, Part II, Part III, or Part IV.)

We’ve already proven the following in Part II, Part III, and Part IV:

  • A calorie is not a calorie when you eat it at a different time of day.
  • A calorie is not a calorie when you eat it in a differently processed form.
  • A calorie is not a calorie when you eat it as a wholly different food.
  • A calorie is not a calorie when you eat it as protein, instead of carbohydrate or fat.
  • Controlled weight-loss studies do not produce results consistent with “calorie math”.
  • And, therefore:

  • Calorie math doesn’t work for weight gain or weight loss.

However, let’s suppose that we’re stubborn and want to count our “calories” anyway. What happens then?

How Accurate Is Our Data? Garbage In, Garbage Out

Computer scientists have an old saying: “Garbage in, garbage out.” (Commonly abbreviated as GIGO.) If a program’s input is inaccurate or misleading, its output will be meaningless—no matter how pretty the set of graphs we can draw from it.

How Accurate Are Calorie Counts In Chain Restaurants?

Given the popular emphasis on counting calories, it shouldn’t be surprising that calorie counts might be, er, fudged a bit. Scripps News Service ran a famous expose in 2008, showing that the few chain restaurants which volunteered the calorie and fat content of their dishes tended to dramatically underestimate both…with some entrees containing more than double their listed calorie count!

Partially as a result of these repeated exposes, and partially because it’s now a legal requirement in some states (and, soon, across the entire USA), calorie counts have indeed become more accurate—on average. However, the variation is still quite wide:

JAMA. 2011 Jul 20;306(3):287-93. doi: 10.1001/jama.2011.993.
Accuracy of stated energy contents of restaurant foods.
Urban LE, McCrory MA, Dallal GE, Das SK, Saltzman E, Weber JL, Roberts SB.
(Fulltext available here.)

Let’s skip to the punchline, from Figure 2:

Figure 2 of Urban 2011.

Figure 2 of Urban 2011.

I’ve added red lines to show +10% and -10% estimation errors—a range of 1800-2200 calories for a 2000-calorie diet. Note that over half of the dishes sampled lie outside these lines!

As we can see by the downward slope of the linear regressions, the lower in calories, the more likely an entree is to have more calories than advertised:

“…Among entrees obtained in sit-down restaurants, those with a lower stated energy content (ie, the most appropriate choices for individuals trying to lose weight or prevent weight gain) systematically contained more energy than stated, whereas foods with higher stated energy contents had lower energy contents than stated.” –Ibid.

This paper comes to similar conclusions, showing that restaurant entrees advertised as “reduced-calorie” underestimate their calorie content by an average of 18%:

J Am Diet Assoc. 2010 Jan;110(1):116-23. doi: 10.1016/j.jada.2009.10.003.
The accuracy of stated energy contents of reduced-energy, commercially prepared foods.
Urban LE, Dallal GE, Robinson LM, Ausman LM, Saltzman E, Roberts SB.
(Fulltext available here.)

The accuracy of stated energy contents of reduced-energy restaurant foods and frozen meals purchased from supermarkets was evaluated. “Measured energy values of 29 quick-serve and sit-down restaurant foods averaged 18% more than stated values…”

Returning to Urban 2011, the categories most likely to contain extra calories were salads, soups, and “carbohydrate-rich foods”…again, precisely those entrees that people on a calorie-counting diet are most likely to order.

The carbohydrate-rich foods averaged 24% more calories than claimed. In contrast, the “meat” category was the most underestimated, averaging 9% fewer calories. (See Table 2 of Urban 2011.)

Finally, Figure 3 shows that these errors are consistent over time, which dashes our hopes that errors will “average out”:

Figure 3 of Urban 2011.

Figure 3 of Urban 2011.

“The mean for the original sample was 289 kcal/portion (95% confidence interval, 186 to 392 kcal/portion) and the mean for the repeat sample was 258 kcal/portion (95% confidence interval, 154 to 361 kcal/portion). Both of these were significantly greater than 0 kcal (P <.001 for both) and they were not significantly different from each other (P = .37).” –Urban 2011

Conclusion: Calorie counts in restaurants are typically off by over 10%…and the lower-calorie and carb-heavy choices are more likely to contain more calories than advertised.

How Accurate Are Calorie Counts In Independent Restaurants?

Chain restaurants—particularly fast food—are frequently blamed for making America fat. However:

JAMA Intern Med. 2013 Jul 22;173(14):1292-9. doi: 10.1001/jamainternmed.2013.6163.
The energy content of restaurant foods without stated calorie information.
Urban LE, Lichtenstein AH, Gary CE, Fierstein JL, Equi A, Kussmaul C, Dallal GE, Roberts SB.

The mean energy content of individual meals was 1327 (95% CI, 1248-1406) kcal, equivalent to 66% of typical daily energy requirements. We found a significant effect of food category on meal energy (P ≤ .05), and 7.6% of meals provided more than 100% of typical daily energy requirements. Within-meal variability was large (average SD, 271 kcal), and we found no significant effect of restaurant establishment or size. In addition, meal energy content averaged 49% greater than those of popular meals from the largest national chain restaurants (P < .001) and in subset analyses contained 19% more energy than national food database information for directly equivalent items (P < .001).

Apparently McDonalds and Applebees aren’t the ones stuffing us with extra food…and even if we look up the calorie counts afterwards on our spiffy new smartphone calorie app, we’ll still underestimate by about 20%. Quoth a co-author of the above study:

“Small restaurants that don’t report calories appear to be the worst restaurants of all,” said study coauthor Susan Roberts, director of the energy metabolism laboratory at the USDA Human Nutrition Research Center on Aging at Tufts University. “They make fast food look like health food.”
Boston Globe, “Small eateries better than fast food? Think again,” May 20, 2013

(We’ll ignore, for the moment, the concept that a Happy Meal is more healthy than an entree of wild salmon with grilled vegetables, herbed butter, and a side of sweet potatoes because it contains fewer calories.)

Conclusion: Independent restaurants serve far greater quantities of food than chain restaurants…and our best estimates will still underreport calorie content by ~20%.

How Accurate Are Calorie Counts For Packaged Foods?

Now let’s look at nutrition labels on packaged foods. According to US law, calories can be underestimated by up to 20% over an average of 12 samples:

“A food with a label declaration of calories, sugars, total fat, saturated fat,trans fat, cholesterol, or sodium shall be deemed to be misbranded under section 403(a) of the act if the nutrient content of the composite is greater than 20 percent in excess of the value for that nutrient declared on the label.”
Code of Federal Regulations, Title 21, Sec. 101.9(g)(5)

Since weight must be >99% of stated weight over 48 samples (USDA Compliance Policy Guide, Sec. 562.300), it seems likely that calorie counts will be slightly overestimated. From Urban 2010, again:

J Am Diet Assoc. 2010 Jan;110(1):116-23. doi: 10.1016/j.jada.2009.10.003.
The accuracy of stated energy contents of reduced-energy, commercially prepared foods.
Urban LE, Dallal GE, Robinson LM, Ausman LM, Saltzman E, Roberts SB.
(Fulltext available here.)

“…Measured energy values of 10 frozen meals purchased from supermarkets averaged 8% more than originally stated.”

The range was from -10% to +31%. If we throw out the highest and lowest value, it still ranges from -5% to +28%. (See Table 1.) Note that these were all reduced-calorie meals: Lean Cuisine, Weight Watchers, Healthy Choice, etc.

Labels on junk food are more accurate:

Obesity (Silver Spring). 2013 Jan;21(1):164-9. doi: 10.1002/oby.20185.
Food label accuracy of common snack foods.
Jumpertz R, Venti CA, Le DS, Michaels J, Parrington S, Krakoff J, Votruba S.

“We tested label accuracy for energy and macronutrient content of prepackaged energy-dense snack food products. […] When differences in serving size were accounted for, metabolizable calories were 6.8 kcal (0.5, 23.5, P = 0.0003) or 4.3% (0.2, 13.7, P = 0.001) higher than the label statement.”

Apparently TV dinner calorie counts are more accurate than both fast food and sit-down restaurant meals—and junk food labels are the most accurate of all.

Conclusion: The worse a food is for you, the more likely its calorie count is to be accurately labeled.

How Accurate Are Our Estimates Of Portion Size?

Most of us eat the majority of our food at home, so it’s important to ask: how accurate are our estimates of portion size? Apparently the answer is: wildly inaccurate.

Am J Clin Nutr. 1982 Apr;35(4):727-32.
Estimates of food quantity and calories: errors in self-report among obese patients.
Lansky D, Brownell KD.
(Fulltext available here.)

The quantity was overestimated for all foods (mean 63.9%). The errors ranged from 6% (cola) to 260% (potato chips). The percentage error in calorie estimates was also substantial, ranging from an underestimate of 4.5% (cottage cheese) to an overestimate of 118.5% (green beans). The mean error in calorie estimates, calculated by averaging the absolute value of overestimation and underestimation errors, is 53.4%.
“Averaged across foods, 26% of the quantity estimates were within ±10% of the foods’ actual values; 32% of the estimates were in error by ±11 to 50%; and almost half the quantity estimates, 42%, were in error by more than 50%. Of the calorie estimates, 14% were in error by 10% or less; 46% were in error by ± 11 to 50%; and 40% were in error by ± 50% or more of the foods’ actual values.”
Inaccurate calorie estimates could have resulted from incorrect quantity estimates, even if judgments regarding calories per unit serving were correct. To test this, the error in number of calories per unit was calculated (Table 1). The subjects ranged from an underestimate of 49.4% (potato chips) to an overestimate of 206.4% (orange juice); mean error, calculated by averaging the absolute value of under- and over-estimates, was 53.8%.

Yes, you read that correctly. When given an unmarked portion of common foods, people overestimate both the quantity and the calorie content by over 50%.

Several studies show that obese people tend to underestimate calories more than lean people. Note, however, that Lansky 1982 demonstrates consistent overestimation of calorie content for individual servings, not underestimation…so the non-obese, if anything, ought to be even less accurate in their estimates.

Result: unless we weigh all our ingredients on a gram scale prior to cooking or eating, our estimates of how much we’ve eaten will be wildly inaccurate. Using that cute little smartphone app to count calories doesn’t help either, because our estimates of quantity are even more inaccurate than our estimates of total calories!

Then, just in case we forget to record all that calorie information right away, as we eat…

The results of study 2 indicate that only 53% of entries in daily food records were specified enough to permit objective estimates of the calories consumed. In study 3, blind raters could not predict weight loss based on subjects’ self-recorded behavior changes. Collectively, these results question the utility of food records for estimating energy intake or predicting weight loss.

Conclusion: our estimates of both how much we eat, and how many calories it contains, are off by over 50%.

(A bonus observation from Lansky 1982: “One-way analyses of variance were used to test calorie and quantity estimates of subjects who viewed foods in large and small containers. Except for one food (cottage cheese), there were no significant differences between estimates made from large and small containers. For cottage cheese, subjects estimated the smaller plate contained fewer calories than the large plate.”)

It Gets Worse: Errors Multiply, and What About Those Free Side Dishes?

Here’s another confounding factor: when eating out, what about the free table bread or tortilla chips? How many pats of butter did we use? And how many calories were in that salsa, anyway?

More importantly, we don’t always clean our plates. Whether we’re eating at a restaurant, eating a prepackaged meal, or eating our own cooking, we have to ask: how much of it did we actually consume? This is important because error terms multiply.

Stated plainly: The inaccuracy of calorie counts is multiplied by the inaccuracy of recalling how much of it we managed to eat, and the inaccuracy of treating all “calories” as equal.

Counting Calories Causes Greater Consumption of Packaged Non-Foods

Counting calories—even inaccurately—is both taxing and discouraging. Trying to recall everything you ate, estimating portion sizes, trying to assign a value in calories or “points” or “blocks”…”Only 53% of entries in daily food records were specified enough to permit objective estimates of the calories consumed.” (Lansky 1982)

Hypothesized result: calorie-counting motivates us to eat less real food and more processed junk. Nutritional shakes, energy bars, TV dinners…

Am J Med. 1997 Mar;102(3):259-64.
Divergent trends in obesity and fat intake patterns: the American paradox.
Heini AF, Weinsier RL.

“In the adult US population the prevalence of overweight rose from 25.4% from 1976 to 1980 to 33.3% from 1988 to 1991, a 31% increase.
“There was a dramatic rise in the percentage of the US population consuming low-calorie products, from 19% of the population in 1978 to 76% in 1991.

Conclusion: calorie-counting appears to motivate us to eat more processed foods…and get fatter.

Conclusion: Garbage In, Garbage Out…Or, When Your Error Term Is Far Larger Than The Change You’re Measuring

We’ve already established, in Part II, Part III, and Part IV, that foods containing the same amount of “calories” produce dramatically different weight gains and losses—and that controlled weight-loss studies do not produce results consistent with “calorie math” (the widely-quoted “3500-calorie rule”.)

Meanwhile, we must recall that, according to “calorie math” (otherwise known as the “3500 calories per pound of fat” rule), the entire obesity crisis—in which the average American has gained 19 pounds—is due to Americans eating six extra calories per day. (See Part II.)

In this article, we’ve demonstrated the following:

  • The typical calorie count for food eaten away from home is off by over 10%.
  • The lowest-calorie and most “healthy” menu items are most likely to be underreported.
  • The only foods whose calorie count approaches accuracy (< 5%) are packaged snack foods—precisely the foods we should avoid.
  • No matter whether we cook our own food or eat prepared food, our estimates of portion size and calorie content, both immediate and retrospective, are wildly inaccurate. The average error exceeds 50%.
  • Error terms multiply. The inaccuracy of calorie counts is multiplied by the inaccuracy of recalling how much of a food we managed to eat, and the inaccuracy of treating all “calories” as equal.
  • Therefore:

  • Unless we prepare all of our own food and weigh every portion on a gram scale, the errors in estimating our true “calorie” intake exceed the changes calculated by “calorie math” by approximately two orders of magnitude. (That’s 100x, or 10,000%, which equals GIGO: Garbage In, Garbage Out.)
  • Additionally:

  • Calorie-counting appears to motivate us to eat more processed foods…and get fatter.

We’re not done yet! Continue to Part VI, “Calorie Cage Match! Sugar (Sucrose) Vs. Protein And Honey”

Or, you can refresh your memory by going back to Part I, Part II, Part III, or Part IV.

Live in freedom, live in beauty.


Start a discussion by sharing this article! Use the widget below to spread it. And since I neither run advertising nor beg for donations, you can support gnolls.org by making your Amazon.com purchases through my affiliate link (which costs you nothing)—

—or by buying a copy of The Gnoll Credo. You already know how to eat like a predator. Instead of yet another book telling you to not eat gluten grains and soybean oil, consider the value of an “Astounding”, “Life-changing”, “Wonderful, inspired, original”, “Raw, powerful, and brilliant” (and much more) story.

More Peer-Reviewed Evidence That There Is No Such Thing As A “Calorie” To Your Body
(Part III)

Caution: contains SCIENCE!

Even after the previous installment of this series, there are still people who believe that calorie intake—and calorie output via exercise—are the only factors that affect weight loss. Apparently my work is not done!

(This is a multi-part series. Go back to Part I, Part II.)

Empirical Evidence: A Calorie Is Not A Calorie When You Add Lots Of Coconut Oil Or Butter To Your Regular Diet

Take three groups of Wistar rats. One group gets free access to standard low-fat rat chow; the others get free access to both standard chow and a “high-fat chow”, 2/3rds of which is butter or coconut oil. (Hat tip to George Henderson for this one.)

Nutr Metab (Lond). 2007; 4: 4.
Long term highly saturated fat diet does not induce NASH in Wistar rats
Caroline Romestaing, Marie-Astrid Piquet, Elodie Bedu, Vincent Rouleau, Marianne Dautresme, Isabelle Hourmand-Ollivier, Céline Filippi, Claude Duchamp, and Brigitte Sibille
(Note: link is to fulltext.)

A fourth group of rats in this study ate a methionine- and choline-deficient diet, which was the primary subject of the study (a successful attempt to give rats fatty liver). Short version: deficiencies caused fatty liver, but massive fat ingestion (and “calorie surplus”) did not.

Unsurprisingly, the rats with free access to the rat version of buttered popcorn ate it. By the end of the diet, both the coconut and butter groups were consuming slightly more high-fat chow than regular chow, the butter group was consuming 30% more “calories” than the chow-only group, and the coconut oil group was consuming 140% more “calories” than the chow-only group!

If a calorie is a calorie, we would expect the rats to gain fat roughly in proportion to their calorie intake. Here’s what actually happened, from Figure 1:

Figure 2 from Romestaing et.al.

Figure 1 from Romestaing et.al.
The open triangles and dashed line represent the chow-only rats, the gray circles and solid line represent the butter+chow rats, and the black circles and solid line represent the coconut oil+chow rats.

Results: “Surprisingly, in spite of a larger energy intake, body mass was not affected in rats fed the high fat diets.” The chow+coconut oil rats ate 2.4 times as many “calories” as the chow-only rats—

—and gained exactly the same amount of weight.

Even the butter+chow rats ate 30% more “calories”, but gained only a non-significant amount of extra weight.


Note that the graph above is partially incorrect: Table 3 gives calorie counts for each group, which agree with the figures quoted in the Results section but disagree with the graph. Apparently the calorie curve for the chow-only rats is shifted upwards, and the calorie curve for the butter+chow rats is just plain wrong! (Or Table 3 is wrong…I’ll pass on any additional information I find.)

Why It’s Important To Report Absolute Change, Not Just Relative Change

The study makes much of the extra WAT (white adipose tissue) gained by the coconut oil+chow rats—62% more—but as the rats started with very little fat, the total gain was approximately 8.4g versus 5.6g for the chow-only rats, for a difference of appx. 2.8g of fat on a 450-gram rat.

In human terms, that’s a 0.6% difference in bodyfat percentage…just under a pound for a 160-pound human.

This, gentle reader, is why it’s important to look at absolute percentages, not just relative percentages…a 62% increase in almost zero is still almost zero. (And this is why so many drug trials report relative risk…a 40% decrease in mortality sounds great until you discover that your absolute risk dropped from 1 in 200 to 1 in 333. Meanwhile, the chance of harmful side effects has stayed the same—and it’s usually far greater than the chance of being saved.)

Conclusion: A calorie is not a calorie when you add lots of coconut oil or butter to your regular diet.

Empirical Evidence: A “Calorie” Of Almonds Does Not Equal A “Calorie” Of Complex Carbohydrates

Take 65 obese and insulin-resistant people. Divide them into two groups, and place each group on a different 1000-calorie starvation diet for 24 weeks. (Another hat tip to Kindke for bringing this one to my attention.)

Int J Obes Relat Metab Disord. 2003 Nov;27(11):1365-72.
Almonds vs complex carbohydrates in a weight reduction program.
Wien MA, Sabaté JM, Iklé DN, Cole SE, Kandeel FR.
(Fulltext available here.)

The study subjects were in bad shape. Mean BMI: 38, weight: 250# (113kg), fasting blood glucose: 152 mg/dl, fasting insulin: 46 ulU/ml (320 pmol/l). Note that a reasonable fasting glucose measurement would be <100 mg/dl, and reasonable fasting insulin would be <9 ulU/ml...so these subjects exhibit classic signs of the metabolic syndrome in addition to being obese. Now, here comes the interesting part: Just over half the 1000 calories were fed as either "self-selected complex carbohydrates" ("peas, corn, potato, pasta, rice, etc.") or as unsalted, unblanched almonds. I'll skip to the punchline: [caption width="400" align="aligncenter"]Figure 2 of Wien et.al. Figure 2 of Wien et.al.[/caption]

That’s 43 pounds lost (19.5kg) for the almond group versus 26.6 pounds lost (12kg) for the complex carbohydrate group.

The authors quote, with typical scientific understatement: “The difference in weight loss was unexpected, given the study design featuring a matched prescribed total calorie intake and equivalent levels of self-reported physical activity between the groups.”

Furthermore, we can see that the “complex carbohydrate” group had plateaued by week 16 (92% of total weight loss after 67% of the time), whereas the almond group was continuing to lose weight at the end of the study (only 77% of weight loss after 67% of the time).

“Calorie math” says that to lose 16.4 more pounds, the almond group would have to have eaten 340 fewer “calories” per day…that’s 2/3rds of the “calories” in the almonds!

Even if we only count the 11.1 pound difference in fat mass lost (see Table 3), “calorie math” requires the almond group to have eaten 230 fewer “calories” per day.

Yet the subjects were voluntary inpatients at a medical clinic, where access to food was controlled. Additionally, “Subjects did not differ in their self-reported evaluation of the acceptability of their assigned dietary intervention in terms of satiety, palatability and texture at weeks 0, 8, 16 and 24,” and “Both groups had equivalent levels of noncompliance…during the 24-week intervention.” So cheating by either group seems unlikely, unless you posit that almonds give you the magical ability to jog for half an hour every day without anyone else noticing—and lie about it.

There were dramatic improvements in health markers for the almond group, which I’ll leave as an exercise for my readers. (Hint: see Table 3.)

Conclusion: A “calorie” of almonds does not equal a “calorie” of complex carbohydrates.

Our Story So Far

Using peer-reviewed science and publicly available population-level statistics, we’ve proven that:

  • A calorie is not a calorie when you eat it at a different time of day.
  • A calorie is not a calorie when you eat it in a differently processed form.
  • A calorie is not a calorie when you eat it as a wholly different food.
  • Controlled weight-loss studies do not produce results consistent with “calorie math”.
  • And, therefore:

  • “Calorie math” doesn’t work for weight gain or weight loss.

The juggernaut continues to roll! Continue to Part IV, Protein Matters…and feel free to stir up some controversy by sharing this article with the widgets below.

Live in freedom, live in beauty.


(This is a multi-part series. Go back to Part I, Part II.)

Thanks to everyone who places their Amazon.com orders through my affiliate link: you’re supporting gnolls.org at no cost to yourself. So next time you’re placing an order for books, music, video games, TVs, cooking implements, or anything else, please consider starting here, or at the link in the right sidebar. (Which takes you to the page for The Gnoll Credo—but you can order anything you want, and it’ll still help support gnolls.org.)

To those worried about privacy: no, I have no way to find out who ordered what.

The Calorie Paradox: Did Four Rice Chex Make America Fat? (Part II of “There Is No Such Thing As A Calorie”)

Caution: contains SCIENCE!

It’s possible to “prove” just about anything via a blizzard of citations and a few carefully-placed appeals to authority. It’s also easy to become seduced by a plausible and elegant biochemical pathway. Presto: science!

However, when formulating a hypothesis, it’s most important to constrain it by observed reality.

(This is Part II of a series. Click here for Part I.)

Empirical Evidence: “Calorie Math” Doesn’t Work

“ERS data suggest that average daily calorie intake increased by 24.5 percent, or about 530 calories, between 1970 and 2000.” (Source: “Profiling Food Consumption In America”, USDA Economic Research Service.) In absolute terms, the average American was consuming roughly 2150 “calories” per day in 1970, 2260 “calories” per day in 1980—and nearly 2700 “calories” per day in 2000.

Source: USDA ERS

Note that the shape of this curve roughly parallels the prevalence of obesity in America, which increased slowly before 1980 and took a steep upturn afterwards:

It's late and I'm out of witty alt tags.

Note that the upturn in obesity coincides with the US Government’s advice to eat less fat and cholesterol, and more whole grains.

Edible fats contain roughly 3500 calories per pound. Therefore, assuming that people were close to their mythical “daily maintenance calories” in 1970, “calorie math” tells us that the average American gained approximately 800 pounds between 1970 and 2000…and has been gaining one pound per week ever since!

If “calorie math” worked, we would all look like this.

Meanwhile, back in reality, the average adult American gained approximately 19 pounds between 1971 and 2000. (Source: Mean Body Weight, Height, and Body Mass Index, United States 1960–2002, Centers for Disease Control.)

The same “calorie math” says a 19-pound gain in 30 years should require a surplus of just six calories per day. That’s nearly two orders of magnitude smaller than the observed 530 calories per day.

Yes, six calories is enough to stop the obesity crisis! All Americans have to do in order to stop gaining weight is to pull four Rice Chex out of the bowl each morning.

Some say two extra teaspoons of milk are to blame...but that's just plain silly.

I blame the ones hiding under the spoon.

Clearly, “calorie math” doesn’t work.

These are back-of-the-envelope calculations, and are not meant to be exact. And I know some might be tempted to quibble about potential errors in the ERS data: keep in mind that we’re not speaking of a 12% difference, or even a 100% difference. We’re speaking of a nearly 10,000% difference between predicted and observed weight gain.

Yes, I weighed the Rice Chex myself.

Empirical Evidence: “Calorie Math” Doesn’t Work, Part II

We’ve established that the 3500-calorie rule is off by roughly two orders of magnitude for weight gain. It’s also wildly inaccurate for weight loss.

Int J Obes (Lond). 2013 Apr 8. doi: 10.1038/ijo.2013.51. [Epub ahead of print]
Can a weight loss of one pound a week be achieved with a 3500-kcal deficit? Commentary on a commonly accepted rule.
Thomas DM, Martin CK, Lettieri S, Bredlau C, Kaiser K, Church T, Bouchard C, Heymsfield SB.

Despite theoretical evidence that the model commonly referred to as the 3500-kcal rule grossly overestimates actual weight loss, widespread application of the 3500-kcal formula continues to appear in textbooks, on respected government- and health-related websites, and scientific research publications. Here we demonstrate the risk of applying the 3500-kcal rule even as a convenient estimate by comparing predicted against actual weight loss in seven weight loss experiments conducted in confinement under total supervision or objectively measured energy intake.

Their Java applet simulates the average of all the weight vs. time curves extracted from the studies they selected: you can download it here. While it doesn’t account for the differences caused by macronutrient composition (e.g. Ludwig 2012), meal timing (see below), meal composition (see below), or the host of other significant factors, you can amuse yourself by turning on “Show 3500 Calorie Rule” from the Options menu.

Clearly, “calorie math” doesn’t work.

Empirical Evidence: A “Calorie” At Dinner Does Not Equal A “Calorie” At Breakfast

Here’s a fascinating controlled study:

Obesity (Silver Spring). 2011 Oct;19(10):2006-14
Greater weight loss and hormonal changes after 6 months diet with carbohydrates eaten mostly at dinner.
Sofer S, Eliraz A, Kaplan S, Voet H, Fink G, Kima T, Madar Z.

Seventy-eight police officers (BMI >30) were randomly assigned to experimental (carbohydrates eaten mostly at dinner) or control weight loss diets for 6 months.
Greater weight loss, abdominal circumference, and body fat mass reductions were observed in the experimental diet in comparison to controls. Hunger scores were lower and greater improvements in fasting glucose, average daily insulin concentrations, and homeostasis model assessment for insulin resistance (HOMA(IR)), T-cholesterol, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, C-reactive protein (CRP), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) levels were observed in comparison to controls.
A simple dietary manipulation of carbohydrate distribution appears to have additional benefits when compared to a conventional weight loss diet in individuals suffering from obesity.

How about that?

The officers were eating the same number of “calories”…they were even eating the same balance of protein, fat, and carbohydrate. Yet, the “carbs for dinner” group lost an additional 2.5kg (5.5 pounds) after six months.

Furthermore, this was a deeply restricted diet (1300-1500 “calories”), so we’d expect all the participants to be extremely hungry. (The Minnesota Starvation Experiment fed its volunteers more: 1600 “calories” per day.) Yet the “carbs for dinner” group rated themselves as less hungry and more sated…so real-world results, in which food intake is only restrained by one’s willpower, would be even greater.

Finally, the “carbs for dinner” crowd were healthier in all measured respects: lower abdominal circumference and fat mass, lower fasting glucose and HOMA(IR), lower LDL, higher HDL, and lower whole-body inflammation (CRP, TNF-α, IL-6). All this from a standard “healthy” high-carb diet (20% protein, 30-35% fat, 45-50% carbohydrate), tweaked so that the carbohydrates were eaten mostly at dinner!

Conclusion: a “calorie” of carbohydrate eaten for breakfast is not equal to a “calorie” of carbohydrate eaten for dinner.

According to “calorie math”, the additional weight loss would equal 107 fewer “calories” per day. Apparently you can change the number of “calories” in food just by eating it at a different time of day…

…or perhaps the concept of “calories” is flawed. (I’ll leave the additional problems this experiment poses for reward-based hypotheses of obesity as an exercise for the reader.)

Also note the dramatic alterations to the hormonal milieu: the same authors explore this in more detail in a followup study.

Nutr Metab Cardiovasc Dis. 2012 Aug 14. [Epub ahead of print]
Changes in daily leptin, ghrelin and adiponectin profiles following a diet with carbohydrates eaten at dinner in obese subjects.
Sofer S, Eliraz A, Kaplan S, Voet H, Fink G, Kima T, Madar Z.

Empirical Evidence: A “Calorie” Of Powdered Food Does Not Equal A “Calorie” Of Regular Food

Hat tip to Kindke for this excellent and well-controlled study:

Br J Nutr. 2013 Apr;109(8):1518-27. doi: 10.1017/S0007114512003340. Epub 2012 Aug 6.
Diet-induced obesity in ad libitum-fed mice: food texture overrides the effect of macronutrient composition.
Desmarchelier C, Ludwig T, Scheundel R, Rink N, Bader BL, Klingenspor M, Daniel H.

“The most striking finding was that all mice fed the different powder diets developed obesity with similar weight gain, whereas among the mice fed the pellet diets, only those given the HF and W diets became obese.

(Note that all mice were fed ad libitum, which means they could eat as much as they wanted.)

Two instructive graphs:

Weight change on standard pelleted diets.

See? High-fat diets cause obesity! (In C57BL/6N mice genetically-engineered to quickly become obese.)

Weight change on powdered diets.

Except when you grind them all into powder—at which point all diets become equally “obesogenic”.

Another fascinating fact: the mice who became obese on the powdered chow were eating the same amount of food that kept them lean when it remained in pellet form! Yes, they were eating the same number of “calories”…

…which made them fat in powder form, but not in pellet form.

We’re not just talking about a little bit of extra fat, either: the mice got 80% heavier on the powdered food, versus 18% on the pelleted version of the same food.

Furthermore, the mice who ate the “high-fat” diet consumed 19% fewer “calories” worth of powdered food—but became just as fat as before. And the mice eating the “Western” diet also consumed 19% fewer “calories”—but became even fatter than before!

Conclusion: a “calorie” of powdered food does not equal a “calorie” of regular food—particularly when the powder is primarily carbohydrate.

Some More Observations On Desmarchelier et.al.

As Kindke notes, flour is powdered carbohydrate. So is sugar. So is almost anything that ends up packaged in a brightly-colored box…processed foods are almost entirely comprised of grains ground into powder, pressed into shape, usually doused with sugar, and baked or fried. Bread, cereal, pasta, donuts, cookies, corn chips, crackers, “instant” anything…yet another reason that Step 1 of “Eat Like A Predator” contains “Do not eat anything made with ‘flour’.”

This study also poses several problems for reward-based hypotheses of obesity. The “high-fat” diet became less “rewarding” when ground into powder, but resulted in the same weight gain. The “Western” diets came in three different flavors, but produced identical results…and all became less “rewarding” when ground into powder, yet resulted in more weight gain. And chow was apparently just as “rewarding” in powder form as in pellet form, yet caused much greater weight gain. (For a demystification of the current state of hunger science, watch my AHS 2012 presentation.)

Finally, here’s a bonus observation. Quote from the paper: “Irrespective of the food texture, the W diet induced a more severe hepatosteatosis and higher activities of serum transaminases compared with the two other diets. In conclusion, diets differing in macronutrient composition elicit specific pathophysiological changes, independently of changes in body weight. A diet high in both fat and sugars seems to be more deleterious for the liver than a HF diet.

There’s much more—including an indictment of the entire field of obesity research, which has based much on the idea that “high-fat” diets cause obesity. Head over to Kindke’s article to read it.

Conclusions: Our Story So Far

  • A calorie is not a calorie when you eat it at a different time of day.
  • A calorie is not a calorie when you eat it in a differently processed form.
  • Calorie math doesn’t work for weight gain or weight loss.

And we haven’t yet discussed the effects of nutrient partitioning, the mysteries of acronyms like REE, TEE, and TEF, or the myriad other ways in which a calorie is not a calorie. Click here to continue to Part III!

Live in freedom, live in beauty.


(This is Part II of a series. Click here for Part I.)

You. Yes, you. The one who doesn’t yet own a copy of The Gnoll Credo.

You saw Fight Club, right? Everyone did. Well, in addition to being “Raw, powerful and brilliant,” “Funny, provocative, entertaining, fun, insightful,” and “Utterly amazing, mind opening, and fantastically beautiful,” The Gnoll Credo also inspires reviews such as “You must read this forthwith—it’s more life changing than Fight Club“…an assessment with which I agree.


Thank you.