Category Archives: Grains

Paleo vs. carbs (per se), Op. 68

The Paleo diet:

A)     the next big thing

B)      Atkins-lite

C)      Fail

D)     None of the above

While proponents of the Paleo diet take a page out of nutritionism‘s book and argue it’s about food choices, not macronutrients, my reductionism mandates inclusion of a comparative breakdown by protein, fat, and carbs.  In a recent publication, Lindeberg (a Paleo pioneer) compared Paleo to the Mediterranean diet in a cohort of CHD patients (Lindeberg et al., 2007 Diabetologia).  To make a long story short, Paleo came out on top in a variety of endpoint measures after 12 weeks.

Divide and conquer

The Paleo diet consisted of lean meat, fish, fruits, vegetables, potatoes, eggs, and nuts; grains and dairy were off-limits (Paleo is GFCF-friendly).  Paleo carbs include fruits, veggies, nuts, and beans… no starches, cereals, whole grains, added sugars, etc… FYI Atkins is very similar to Paleo but includes a lower absolute amount of Paleo carbs.  The Mediterranean dieters ate whole grains, low-fat dairy, vegetables, fruits, fish, oils, and margarines.  Both diets exclude processed junk food and both are relatively healthy diets.  

As such, both groups lost weight; slightly more on Paleo but this was probably due to reduced caloric intake (not uncommon for Paleo dieters; see below and also Osterdahl et al., 2008 EJCN):But the benefits of Paleo were much more robust WRT insulin sensitivity, which was markedly improved on Paleo but not Mediterranean.

Paleo: 1

Mediterranean: 0

With a 4% weight loss, why didn’t glucose tolerance improve in the Mediterranean dieters?  … weight loss is almost always accompanied by improved glycemic control…   The biggest difference in “foods” consumed by the two groups was cereals: 18 grams per day on Paleo vs. 268 on the Mediterranean diet… over 14 times more!  As I’ve discussed at length with gravitas, a high intake of cereals (aka grains aka fibre [in the figure below]) does not bode well for insulin sensitivity, inflammation, and outright all-cause mortality:

As such, Paleo does well to exclude grains.  Furthermore, Paleo is higher in protein and fat and lower in carbs- all good things.  A more interesting analysis showed that waist circumference (visceral fat) was associated with grain intake even when controlled for carbohydrates.  In other words, the detrimental impact of whole grains goes beyond their intrinsic carbohydrate content. (whole grains … insulin resistance … visceral fat)

Back to those calorie data for a moment, given that they were probably just as important as cereal exclusion in determining the results.  Why did Paleo dieters spontaneously eat so much less?  In a follow-up publication, Jonsson and colleagues assessed leptin and satiety in both groups (2010 Nutrition & Metabolism) and showed that despite eating less and losing more weight (things that should increase hunger and decrease satiety), Paleo actually did the opposite (hint: something to do with whole grains, perhaps?).

While the Paleo meals were smaller (5th and 6th rows) and contained fewer calories (3rd and 4th rows), they were just as satiating as Mediterranean diet meals (7th through 9th rows), leading the authors to conclude Paleo is more satiating calorie-for-calorie and pound-for-pound.  And if that isn’t enough, Paleo dieters also experienced a significantly greater reduction in leptin! (probably caused by their reduced food intake and body weight loss)  While the general consensus is that such a change in leptin should enhance hunger, as discussed previously I think lower leptin in this context reflects enhanced leptin sensitivity, which also helps to explain the improved insulin sensitivity.  Last but not least, WRT the Mediterranean diet I suspect reduced calories explains the weight loss, but the abundance of whole grains explains the blunted glycemic improvements.  (hint: whole grains … leptin resistance … insulin resistance) … (whole grain exclusion … leptin sensitivity … insulin sensitivity)

Paleo, the next big thing?  I’m holding out for a one-on-one with low-carb proper to exclude the role of Paleo’s lower carb content.  The whole grains issue requires no further confirmation IMO (e.g., Burr et al., 1989 LancetJenkins et al., 2008 JAMA, etc.).

The Paleo diet:

A)     the next big thing

B)      Atkins-lite

C)      Fail

D)     None of the above

might be considered “Atkins-lite,” probably not “the next big thing,” definitely not “fail.”

+1 for excluding grains


calories proper

USDA vs. nutrition, round II

The school lunch program is screwed.

First the USDA modifies the definition of a vegetable to include pizza.  Now they significantly altered their standards for school lunches to include fewer healthy foods and more USDA-approved ones (see report at the USDA’s website).  In brief, this move further reduces the nutrition of school lunches and will likely do more harm than good.  Here’s why:

In this cross-sectional Swedish study, parents recorded 7-day food diaries for their 4-year old children who then went in for a regular checkup.

Metabolic markers in relation to nutrition and growth in healthy 4-y-old children in Sweden (Garemo et al., 2006 AJCN)

On a 1,400 kcalorie diet, these children were consuming roughly 15% protein, 33% fat, and 52% carbs (about 20% of which came from sucrose).  That seems like a lot of calories, but besides playing all day, 4 year old children are also growing at an incredible rate.

Interesting finding numbers 1 & 2:  Children who got most of their calories from fat had the lowest BMI (i.e., they were the leanest), and the opposite was observed for carbs.

When divided into groups of normal weight vs. overweight and obese, some interesting and non-intuitive patterns emerged.  For example, lean kids don’t eat less food; but they do eat fewer carbs and less sucrose (and make up the difference by eating more fat and saturated fat).

Some of the weaker correlations showed:
-total calorie intake was associated with growth (logical)
-total carbohydrate intake was associated with increased fat mass (unfortunate yet also logical)
-total fat intake was associated with decreased fat mass (interesting)

And those who ate the most saturated fat had the least amount of excess body fat. (more on this below)

Fortunately, in a young child, a poor diet hasn’t had enough time to significantly impact their metabolic health; as such no macronutrient was associated, either positively or negatively, with insulin resistance [yet].

In a more appropriately titled follow-up, Swedish pre-school children eat too much junk food and sucrose (Garemo et al., 2007 Acta Paediatrica), Garemo reported that most of their carbs came from bread, cakes, and cookies, while most of the sucrose came from fruit, juices, jam, soft drinks, and sweets.  And WOW, go figure- most of the fat came from meat, chicken, sausage, liver, eggs, and dairy; NOT vegetable oils.

And in a mammoth dissertation, Eriksson (2009) confirmed many of these findings in a larger cohort of 8-year old Swedish children and had this to say about dairy fat:

The open boxes represent overweight kids, the closed boxes are lean kids.  Going from left to right, in either the open or closed boxes, BMI declines with increasing intake of full fat milk (perhaps parents should reconsider skim milk?).  Eriksson also confirmed that saturated fat intake was strongly associated with reduced body weight.  Interestingly, she mentioned that food intake patterns are established early in life, so it might be prudent to remove sugars and other nutrient poor carb-rich foods, and introduce nutritious whole foods as early as possible.  I’m not exactly sure how she assessed patterns of food intake establishment, but it seems logical.  Especially in light of the following study… we’ve seen 4 year olds, 8 year olds, and now we have 12-19 year olds.  The relationship between diet and health is consistent across all age groups.

Virtually all of the above data in Swedish children seem to suggest dietary saturated fat, whether it’s from beef, sausage, eggs, whole fat dairy, or liver (i.e., WHOLE food sources; NOT hydrogenated vegetable oils), is associated with reduced fat mass.  Metabolic abnormalities were not present, probably because the children were simply too young (although body weight seems to respond relatively quickly, other downstream effects of poor nutrition take years to accumulate before symptoms develop).

An American study about nutrient density and metabolic syndrome was recently published.  These kids were exposed to poor nutrition for just long enough to experience some of those malevolent effects.

Dietary fiber and nutrient density are inversely associated with the metabolic syndrome in US adolescents (Carlson et al., 2011 Journal of the American Dietetic Association)

The figure below divides fiber (a proxy for good nutrition; i.e., leafy vegetables, beans, etc.) and saturated fat into groups of least and most amounts comsumed. The lowest fiber intake was 2.9 grams for every 1,000 kcal, and 9.3% of these kids already had metabolic syndrome; the highest fiber intake was 10.7 grams / 1,000 kcal and 3.2% had metabolic syndrome.  Thus, consuming a fiber-rich [nutrient dense] diet is associated with a significantly reduced risk of metabolic syndrome.

The next rows are saturated fat.  The lowest saturated fat intake was 6.9 grams / 1,000 kcal and 7.2% had metabolic syndrome; the highest saturated fat intake was 18 grams / 1,000 kcal and 6.7% had metabolic syndrome…. huh?  While it didn’t reach statistical significance, the trend for saturated fat paralleled that of a “nutrient dense” diet.  Is it possible that saturated fat might be part of a nutrient dense diet?   if saturated fat comes in the form of red meat, liver, eggs, etc., then yes, it is part of a nutrient dense diet.  This conclusion evaded both the study authors and the media.

In 4 and 8 year old Swedish children, those who ate the most saturated fat had the least excess fat mass.  In 12 – 19 year old American adolescents, those who ate the most saturated fat had the lowest risk for metabolic syndrome.

Is it too much of a stretch to connect these ideas by saying that in the short run, a low saturated fat (nutrient poor, carb-rich) diet predisposes to obesity; and in the long run it predisposes to metabolic syndrome  ???

Collectively, these data suggest a diet based on whole foods like meat and eggs, including animal fats, with nutrient dense sources of fiber (e.g., leafy vegetables) but without a lot of nutrient poor carb-rich or high sugar foods, may be the healthiest diet for children.  

Flashback: recap of “USDA vs. nutrition, round I”
Nutrition: 0
They made pizza a vegetable and insiders suspect that next they’ll try to make it a vitamin.

USDA vs. nutrition, round II

USDA: replacing normal milk with low fat milk
nutrition: full-fat milk was associated with lower BMI in both lean and obese children (see the Eriksson figure above)

USDA: increasing nutrient poor carb-rich options
nutrition: this was associated with increased fat mass in children (Garumen et al., see figures above)

USDA: reducing saturated fat as much as possible
nutrition: reduced saturated fat was associated with excess fat mass in children and metabolic syndrome in adolescents.

Such changes will have an immeasurable long-term impact if children grow up thinking these are healthy options.  Finally, this blog post does not contain a comprehensive analysis of saturated fat intake and health outcomes in children, but the USDA’s new regulations should have been accompanied by one.  In other words, these regulations should not have been based on the studies discussed above, but the studies discussed above should have been considered when the USDA was crafting their recommendations.  Obviously, they weren’t.

calories proper

pizza on the docket

they’re all crooks!


a slice of pizza does not count as a serving of vegetables. Period.

not the worst thing for you, really just a bunch of empty calories.  definitely NOT a serving of vegetables.

The government-sponsored school lunch program is designed to provide nutrition and improve the health of our children.  And they get around 11 billion dollars (i.e., $11,000,000,000) every year to do so.  Due to the recent surge in obesity, Congress acted fast!  School lunch programs do not closely follow the dietary guidelines.  To us taxpaying voters, $11,000,000,000 of our taxes are being wasted AND our kids are suffering.   Therefore, Congress quickly changed the status of pizza to “vegetable.”  Many schools serve pizza, and thus are now more closely in line with the dietary guidelines; so our taxes are being less-wasted and our children are healthier because they are eating more vegetables! To be clear: now that pizza is a vegetable, your children are healthier.

You can’t make this shit up – it is what happens when government gets involved in nutrition.  Please, ignore the Dietary Guidelines, they are horribly misguided.  And be extremely wary of electing anyone who wants to control nutrition; or vote with your dollars, don’t buy processed food!  The message is almost always wrong and both our bank accounts and our health suffer the consequences.  I would suggest supporting nutrition education programs, but NOT IF THEY SAY PIZZA IS A VEGETABLE.  If anything, a slice of pizza should count as dessert plus 3 servings of grains :/

Isn’t it bad enough that French fries, or crisps, count as vegetables?

Admittedly, claiming “the Dietary Guidelines are horribly misguided” is a strong statement, especially when said guidelines direct how a portion of our taxes are spent AND which foods are made available to our children.  This is important.


calories proper


Become a Patron!




QLSCD II (or Grains IV)

WRT the Quebec Longitudinal Study of Child Development (QLSCD), I failed to adequately emphasize one major implication of their findings.  It is a point that completely and wholly illustrates the disconnect from data, empirical science, and all common sense exhibited by mainstream beliefs in calories and dieting.   gravitas

Higher intakes of energy and grain products at 4 years of age are associated with being overweight at 6 years of age (Dubois, Porcherie et al., 2011 Journal of Nutrition)

Divide and conquer

Exhibit A


The table above shows the percentage of underweight, normal weight, and overweight children consuming the recommended number of servings for each food group.  15.5% of underweight children, 19.1% of normal weight children, and 42.6% of overweight children meet the recommended ?5 servings of grains per day.  Grains comprise [sic]: “breads, pastas, cereals, rice, and other grains”

There is a direct relationship between body weight and the percentage of children consuming ?5 servings of grains per day, i.e., more grains equals greater chance of being overweight.

Exhibit B


This table shows the odds for being overweight at 6 years of age in increasing quintiles of how many calories consumed daily two years earlier.  The crude odds risk (first column) shows a poor relationship between calorie intake at 4 years old and risk of being overweight 2 years later.  I say “poor” because the risk is non-significantly lower in the second quintile, higher in the third, lower in the fourth, but much higher in the fifth quintile (3.15x more likely to be overweight for the biggest eaters compared to the littlest eaters).  These data are unadjusted and could be confounded by a variety of factors.  Thus, the significance level of the trend is high p=0.0007.

The second column is similar to the first, but is adjusted for many known confounders: birth weight, physical activity, mother’s smoking status during pregnancy, annual household income, and number of above normal weight parents.  As such, the degree of statistical significance was reduced from 0.0007 to 0.001.

The third and most important column is further adjusted for body weight at 4 years of age, and shows that calorie intake is no longer associated with body weight at 6 years of age.  In other words, being overweight at 4 years old predicted being overweight at 6 years old better than calorie intake (and physical activity).

In the authors’ own words [sic]: “The only food group significantly related to overweight was grains.”  No association was observed for overweight risk with vegetables and fruits, milk products, or meat and alternatives.

IMHO, the observation that being overweight at 4 years old was the best predictor for being overweight 2 years later is remarkable… body weight status at 4 years old is a more important risk factor than both physical activity and calorie intake.  The only ‘controllable’ variable  is grains; i.e., you can’t change whether or not your child was overweight at 4 years of age, and physical activity and calorie intake doesn’t matter.  But grain consumption seems to matter, and it is something that can be controlled.

What is it about grains?  I don’t know, exactly, but it’s not simply that they’re carbohydrates because elevated carbohydrate intake didn’t increase risk for being overweight.

Exhibit C



“Eating less and moving more” is not the answer.  Nutrition matters, not the guidelines.


calories proper



Grains III

Grains, gluten, and kids.  And I go WAY overboard on Table 4.

This topic has special relevance because grains provide more calories (31%) than any other food group.  And they are probably the most detrimental.

Higher intakes of energy and grain products at 4 years of age are associated with being overweight at 6 years of age (Dubois, Porcherie et al., 2011 Journal of Nutrition)

The Quebec Longitudinal Study of Child Development (QLSCD) assessed food intake and lifestyle variables in ~1,000 Canadian kids born in 1998 for 2 years.  The data came primarily from mothers but also daycare attendants when necessary, and their method for assessing food intake was pretty good- “multiple-pass” 24-h dietary recall interviews conducted in the home, and they double-checked by re-questioning a huge subgroup (~50%!, kudos).

Methodological peculiarities:

1)      Food groups consisted of

  1. Grains (e.g., breads, pastas, cereals, rice, etc.)
  2. Fruits and vegetables
  3. Dairy
  4. Meat and alternatives (e.g., meat [duh], lentils, tofu, and peanut butter)

What is the rationale for grouping lentils, tofu, and peanut butter 1) together, and 2) with meat?  IOW, data regarding the consumption of “meat and alternatives” will be difficult to interpret.

2)      Divisions between underweight, normal weight, and overweight were based on percentiles as opposed to absolute values.  For adults, BMI<20 = underweight, 20-25 = normal weight, 25-30 = overweight, and >30 = obese, regardless of the weight of their friends, colleagues, and neighbors.  By using percentiles: if the entire cohort is heavier than average, then overweight kids will be classified as normal weight because they are “normal” relative to the rest of the kids in the study, who are heavier than average.  So it’s not a debilitating methodological peculiarity, it just changes the definitions with which we are accustomed… so when they start out their results by stating [sic]: “20% of the children were overweight,” it doesn’t mean they have an unusually lean cohort, it actually tells us nothing.

Divide and conquer

Here’s what these kids were eating, in total and broken down by body weight groups:


Heavier kids ate more carbohydrates and less fat, and protein intake was relatively constant.  No big surprises, except that none of this reached statistical significance despite being true across all three quintiles… the lack of statistical significance is most likely due to the small sample size, and I suppose we’ve been spoiled lately with studies that included much larger subjects.  FTR, the carbs and fat data are probably the most relevant finding WRT feeding your kids.

Table 2 showed macronutrients and total energy, while Table 3 shows the breakdown by food groups (see Methodological Pecularity #2 above).


THIS is troubling.  Grain consumption is highly adherent to the guidelines, but the more the guidelines were adhered to, the fatter the kids got.  Combined with the amount of calories grains contribute to overall energy intake, this provides a fairly clear explanation for the childhood obesity epidemic.  IOW, these data strongly suggest the guidelines are wrong.

The long-awaited Table 4.  (did you feel the suspense?)


This table shows the odds for being overweight in increasing quintiles of total calorie intake.  The first and second columns show what everyone normally expects: more calories consumed = more chance of being overweight.  And it’s highly statistically significant.  But here’s the kicker: the third column adjusts for body weight at 4 years of age and the association is abolished.  !!!  That means being fat at 4 years old was a more important predictor of being fat at 6 years old than calorie intake.  Chubby 6 year olds were overweight because they were chubby when they 4 years old, NOT because they ate too much !  Excessive inactivity is ruled out because these data were adjusted for physical activity.

“Eating less and moving more” is not the answer.  Nutrition matters.  Don’t feed your kids grains, regardless of the guidelines.


calories proper


Nutrition under attack

Global nutrition is in a state of emergency

Tax this:

Not this:


If you catch a whiff of anyone talking about a dietary fat tax here in the states, attack!  Hold no bars.

Passage of the Danish saturated fat tax confirms the shit hit the fan.  They should’ve taxed people for being fat (to offset the increased medical and healthcare costs associated with obesity), or sugars (for making people fat).  Instead, the food companies, famous for crafting thousands of varieties of Danish pastries, lobbied for the taxes to be levied against dietary fat.  This “inadvertently” encourages people to consume more Danish pastries with only 7 grams of saturated fat yet 39 grams of sugar!  The tax will favor Pop Tarts over eggs, and this is supposed to make people healthier?

A dietary fat tax is particularly troublesome because it strikes an expensive blow against real whole foods like eggs, butter, and meat, while leaving unscathed processed foods like doughnuts, refined grains, and SUGAR.  This disproportionately affects healthy foods that are in no way responsible for the obesity epidemic.

My suggestions:

1)      Leave people alone.

2)      A better target, which would entail markedly less collateral damage, is “added” sugars.  Taxing “added” sugars would hit soda, processed food-like products, snacks, and junk food… lots of bad guys, few good guys

3)      Make a tax based on empty calories: foods with a higher ratio of calories to nutrients get taxed more than nutrient-dense foods… thus, people would be eating fewer calories but more nutrients!  That wasn’t too hard?

4)      Tax food in direct proportion to its shelf-life.

5)      The revenue from any of these alternative options should be put toward nutrition education programs in elementary schools.  And a portion of the money saved in healthcare costs should be redirected into funding research in the nutritional sciences.  And the rest can be used to pay off the National debt.


Calories proper


Empty calories III

Empty calories.  Nutrient density.  Empty calories.  The ANDI score?

With the advent of the ANDI, it is safe to say the phrase, school of thought, and cult following to “a calorie is a calorie” is fading.  All calories were not created equal; some make you fat, others make you strong.  Cantankerous old biochemists and low-fat diet proponents will likely remain loyal, however, with the former citing heat production in a bomb calorimeter (mumbo), and the latter citing the equivalency of nutrient density and animal fat scarcity (jumbo).

But for the rest of us, there is gravitas in this concept       empty calories.

A good place to start might be a critical view of Fuhrman’s  Aggregate Nutrient Density Index (ANDI).  In brief, ANDI is an index of healthiness and is calculated by dividing the amount of nutrients in a given whole food by the calories.

Pro’s and con’s

1)      it only applies to whole foods.  This is convenient because most processed foods would score miserably low unless they’ve been industrially fortified with synthetic vitamin-like chemicals.  Perhaps Fuhrman restricted ANDI to whole foods because something like Diet Coke Plus would score about a million (lots of vitamins, few if any calories), rendering ANDI utterly meaningless to the masses and downright offensive to people like me.


2)       “nutrients per calorie” is a far more biologically meaningful and physiologically relevant concept than “calories per gram.”

  1. “Calories per gram” can be too easily manipulated.  E.g., one ounce (~28 grams) of soybean oil has 248 kilocalories: 248 kilocalories / 28 grams ? 9 kcal/g.  Add it to an ounce of water and you get 248 / 56 ? 4.4 kcal/g.  It’s still the same nutritionally, but the “caloric density” was halved by trickery.
  2. “Nutrients per calorie” is relatively unchangeable.  Let’s say there are 14 grams of omega-6 fatty acids in an ounce of soybean oil; that would be 14 grams per 248 kcal.  Add it to an ounce of water and it’s still true.  Drink it on the moon and it’s still true (relatively).


1)      ANDI selectively quantifies only one aspect of a food’s nutritional value.  It is an important aspect, but please note that a diet based on high ANDI foods would be nutritionally inadequate.  Furthermore, there are highly relevant health parameters that ANDI completely ignores.  More questions:

  1. Shouldn’t more important nutrients be given a higher score?
  2. Shouldn’t excess amounts of a nutrient detract from the score?
  3. What about other non-nutritive health-promoting properties of a food?  E.g., foods that are healthier than indicated by their ANDI score:
    1. i.      foods that have some as-of-yet undiscovered nutrients
    2. ii.      foods that indirectly promote health (like pre- or pro-biotics)

2)      Furthermore, ANDI is fundamentally flawed in its application to foods whose value is based at least partially on the actual calories themselves.

  1. Fuhrman uses the ANDI score on fats, which score dismally low because they contain few “nutrients” and a lot of calories.  Thus, industrially-modified, partially-hydrogenated trans fat-rich soybean oil has the same ANDI score as olive oil.
  2. Animal proteins, including grass-fed beef, wild salmon, and pastured eggs, also score incredibly low.  These foods are far more healthful than many most others, essential for life (unlike kale, which is the highest-scoring ANDI food), and much of their value is contained in the quality of their calories.
    1. i.      the fatty acids in salmon are healthy in and of themselves; they don’t contain any nutrients per se; they ARE the nutrient.  But ANDI doesn’t take this into account; it views all fatty acids as empty calories, a grave mistake.
    2. ii.      the same goes for animal proteins.  Eggs, for example, are higher in protein quality than any other food on the ANDI scale yet they score quite low.  And getting a bio-equivalent amount of protein from lower quality plant proteins would require consuming many more calories.

The failure of ANDI to incorporate any measure of fat or protein quality is its demise; why it is unable to stand alone as an indicator of healthiness… a diet consisting exclusively of high ANDI foods is incompatible with life.  A protein deficiency would be vastly more severe than a low ANDI diet, and on a lighter note, the fish oil fatty acids would provide much greater benefits than a high ANDI diet.  These nutritional factors play too big a role in determining healthspan and quality of life to be ignored.

BUT, ANDI is nice in its simplicity, and it works very well for most plant-based foods.  E.g., spinach and cabbage have very high ANDI scores; rice, grains, and pasta have very low ANDI scores.

The diets of many cultures are based almost exclusively on low-ANDI foods.  This is largely because it is much easier to produce enough calories to feed a village than to produce enough nutrients.  Starvation is deadlier than dermatitis.  In the Western world, however, we are fighting a different battle: you need to eat a LOT of empty calories in order to get enough nutrients, but then you get fat.


calories proper



n = 1,570,808

the correlation between time spent watching TV and body weight may have nothing to do with the common thought that if kids aren’t watching TV, they’re out playing.  No, it turns out that kids who watch less TV eat healthier.  They sit around just as much … just not watching TV.  Is there something about mindless sitcoms and cartoons that make us want to eat junk food … or snack … wait a minute … subliminal commercial advertising?

Is any of this true?  probably not.  But if I were to test it, I’d like an intervention study with a couple randomized groups including: one whose TV commercials were strategically replaced with equally fun commercials that don’t promote snack foods; a group of kids who don’t watch TV; and maybe a group who doesn’t watch TV but is exposed to subliminal snack-promoting advertisements…   food for thought.

Changes in diet and lifestyle and long-term weight gain in women and men (Mozaffarian, Hao, Rimm, Willett, and Hu, 2011 NEJM)

This is one of the biggest and longest running prospective studies on diet and lifestyle behaviors.  That’s not to say it’s the best study; epidemiological, observational, and prospective studies are subject to a variety of crippling limitations; but this one is big.

It is a compilation of three big studies that I’ve blogged about in the past:

The Nurses’ Health Study (NHS, n=121,701, est. 1976)

The Nurses’ Health Study II (NHS II, n=116,686, est. 1989)

The Health Professionals Follow-up Study (HPFS, n=51,529, est. 1986)

All in all, a total of 1,570,808 person-years were analyzed (person-years: 16 people x 2 years = 32 person-years).  The biggest finding?  everybody gains weight, about 0.835 pounds per year.  It doesn’t sound like much but after 20 years it means you’re 15 pounds fatter.

The second biggest finding?  Drumroll please…

First, to set the mood: people don’t gain weight magically.  This study was unique in that there were multiple dietary assessments, performed over a very long period of time, in the same group of people.  And one way people gain weight is by eating more.  So these authors were able to see which foods, when increased in the diet, correlated with weight gain.

And the winner was, somewhat surprisingly, potatoes!  Actually number 1 was potato chips, and number 2 was potatoes and French fries or crisps.  The surprise, IMHO, was that soda and junk food was much further down the list.  Furthermore, I might have thought potato chips could be number 1 because of their high trans-fat content, but the presence of potatoes (which lack trans-fats) at number 2 means that trans-fats are not the obesogenic component of potato chips.

The figure (kudos for data presentation):

To quote the study [sic]: “Strong positive associations with weight change were seen for starches, refined grains, and processed foods.”  WRT study design, such conclusions cannot be interpreted to mean: eating less “starches, refined grains, and processed foods” will prevent weight gain.  It means body weight was determined by changes in the amount of “starches, refined grains, and processed foods” were in the diet… if any of those foods were increased, body weight increased; and if any of those foods decreased, body weight decreased. (note: this study was not designed to determine causation).

Exercise was also included in their analysis.  Another good-looking figure, albeit a little more complicated:

Think of the bottom left (front?) as the sum of good dietary changes (associated with weight loss [or less weight gain]).  People in the first “Quintile of Dietary Change” increased their consumption of “starches, refined grains, and processed foods” and gained the most weight, while those in the fifth quintile ate less and lost weight or stayed the same.  A similar trend for physical activity is displayed on the bottom right part of the figure.  People in the first quintile increased their physical activity, while those in the fifth quintile reduced their physical activity.  If you compare the amount of weight gain from the fifth to the first quintiles of dietary change (all the purple bars, for example), the amount of weight gained is highly dependent on diet.   However, within any given quintile of dietary change, not much weight is gained or lost by changing the amount of exercise (follow any set of bars from gray to purple to yellow to green to blue).  IOW, at any given quintile of physical activity change, diet predicts much larger changes in body weight.  Heck, people in the second quintile of dietary change actually gained weight by increasing physical activity … that’s a bit convoluted, but it demonstrates the case that in this enormous data set, diet was a better predictor of weight gain than physical activity (which in some cases didn’t matter at all).

Eat less and move more?

Calories proper


GB Grains blog, take II

GB Grains blog, take II

The effect of increasing consumption of pulses and wholegrains in obese people: a randomized controlled trial. (Venn et al., 2010 JACN)

I like this study because of its thorough dietary intervention.  The researchers provided a lot of the food, had frequent meetings, checkups, and dietary counseling sessions.  They even sponsored cooking lessons and supermarket tours!  Those are all definitely strengths, in addition to the ultra-long study duration of 18 months.  Both groups were advised to eat low fat diets, but the intervention group was specifically instructed to eat more whole grains.  To supplement their diets, the intervention group was given rolled oats and rye, wholemeal flour breads, etc., while the control group received cornflakes, cans of fruits & vegetables, refined grain breads, etc.

How’d they do?  As seen in figure 1 (below), the diet was followed quite well.

Figure 1.  Everybody tried to eat healthier in this study, so whole grains increased in both groups.  But it was significantly higher for most of the time in the intervention group.  Everybody also ate fewer calories.  And since whole grains are both carbohydrates and fibrous, consumption of these increased in both groups, but more so in the intervention group.

To make a long story short, both groups lost approximately equal amounts of weight with the treatment group losing slightly more than control.  The interesting thing is that we would have expected these weight losses to be accompanied by all-around improvements in health.  But they weren’t (reminiscent of the Orlistat trials).  Fasting glucose is a surrogate for insulin sensitivity.  Fasting glucose increased in both groups:

Figure 2. Metabolic outcomes.

Both groups lost weight.  Dietary carbohydrates are linked with insulin resistance, and although the % of calories from carbohydrates increased in both groups, the absolute amount decreased because of the large reduction in calories.  So they were eating fewer grams of carbohydrates and losing weight… So WHY did blood glucose increase?  I’d be willing to bet whole grains had something to do with it.  Whole grains increased significantly in both groups.  There’s something creepy about whole grains, like how every correlation between them and good health is attenuated after adjusting for confounding lifestyle and dietary factors.  Healthy people eat whole grains, but whole grains don’t healthify.  Possible suspects include lectins and gluten.

Just like DART, the Venn study was a randomized controlled intervention study, which is very powerful study design.

However the Venn study was a weight loss study, which is very different from free-living individuals eating ad libitum in ‘energy balance.’

Enter: the Jiangsu Nutrition Studies.  These epidemiological observational studies have been going on for a while and their goal is to identify dietary patterns that are associated with weight gain.

disclaimer: in general, when coming upon a study of “dietary patterns” I turn around and run away.  The data are usually so manipulated that they no longer reflect what a person actually eats.  I’m making an exception here because Jiangsu  demonstrates an interesting point.  Briefly, they were able to differentiate “diets devoid of whole grains” from “diets rich in whole grains,” and two other dietary patterns that couldn’t be characterized by their whole grain content.

Vegetable-rich food pattern is related to obesity in China. (Shi et al, 2008 International Journal of Obesity)

Dietary pattern and weight change in a 5-year follow-up among Chinese adults: results from the Jiangsu Nutrition Study. (Shi et al., 2010 British Journal of Nutrition)

They somewhat humorously defined four major dietary patterns:

Divide and conquer.


Table 1.  Dietary patterns.  Focus on the foods with the biggest “Factor loading,” as these are the most important foods that define each pattern.  In the traditional diet, for example, presence of rice (0.78) and absence of wheat flour (-0.75) are the two most important factors that distinguish the traditional dietary pattern.  Presence of whole grains (0.56) is what most defines the vegetable-rich pattern.  Those are the two I think are of most interest: traditional dietary pattern is defined by an absence of wheat flour, while the vegetable-rich diet is defined by an abundance of whole grains.

In 2002, the food intake data were collected and analyzed.  For each dietary pattern, subjects are divided into quartiles based on their adherence to each respective dietary pattern.  IOW, every subject is ranked on their adherence to each dietary pattern.  For example, you might rank very high for macho, intermediate for vegetable-rich, and low for traditional and sweet tooth. You are ranked by your adherence to each dietary pattern.

To analyze the effect of a dietary pattern on a specific health parameter, investigators compare the prevalence of that parameter outcome across quartiles of each dietary pattern.  If there is no association between a specific dietary pattern and the health parameter, it would be similar across quartiles.  If, OOTH, the parameter increases or decreases across all 4 quartiles, then there is a correlation.

At baseline (2002) and follow-up (2007), the subjects were weighed.  The figure below depicts weight change between 2002 and 2007 and is divided into quartiles of each dietary pattern.

5-year weight change across quartiles of each dietary pattern.  Can you spot which two of the four dietary patterns were significantly associated with weight change?


Traditional diet, defined by the absence of wheat flour (top left).  People who were the most adherent to the traditional diet (“Q4”), meaning they never touched wheat flour, gained the least amount of weight over those 5 years.  Conversely, people who were the least adherent to the traditional diet (“Q1”), i.e., those who ate the most wheat flour, gained the most weight over those 5 years (~2.0 kg).

Vegetable-rich diet, defined by an abundance of whole grains (bottom right).  People who were the most adherent to the vegetable-rich diet, meaning they ate plenty of whole grains, gained the most weight over those 5 years (“Q4,” 1.6 kg).  Conversely, people who ate the least whole grains gained the least weight over those 5 years (“Q1,” 0.4 kg).

It gets worse.

The prevalence of frank obesity (BMI > 30) according to adherence to the vegetable-rich (high whole grains) diet:

Obesity is far more prevalent among those consuming the most whole grains compared to the least.  To make a stretch, people who ate the most whole grains were twice as likely to be obese (bottom row, first [6.9] compared to fourth [15.0] quartile).

Whole grains are associated with frank obesity in the total population, but they are really really associated with obesity in folks between 31 and 45 years of age:

People aged 31-45 with the highest intake of whole grains were 3.66x more likely to be obese than people with the lowest.

The Jiangsu Nutrition studies are observational, but prospective.  The Venn study (above) and DART are randomized intervention trials.  Obesity (Jiangsu), elevated fasting glucose despite weight loss (Venn), and all-cause mortality (DART)… Collectively, these findings suggest that whole grains should be abandoned, or at least demoted to “consume sparingly.”  But their elite status among dietitians and health advocates prohibits this.  Divide and conquer?


calories proper


Become a Patron!