The Glycemic Index ~ It Was Supposed to Be About Carbs!

As I have quite a few tangential thoughts going on, in search of a cohesive theme here, I've decided I'd just throw up a few short posts, and perhaps come back and put them together at some later date.

The Glycemic Index is credited to Canadian researcher David Jenkins, and originated, near as I can tell, with this paper:  Glycemic index of foods: a physiological basis for carbohydrate exchange.

INTRODUCTION:   Recent work has suggested that the carbohydrate exchange lists that have regulated the diets of many diabetics for over three decades may not reflect the physiological effect of foods. Such factors as food form, dietary fiber, and the nature of the carbohydrate have been shown to have a marked influence on the postprandial glycemia and allowances cannot be made for these in lists which take into account only the available carbohydrate content of foods. Currently, very good blood glucose control has been advocated for diabetics to reduce the incidence of long term complications. We have, therefore, fed a range of commonly eaten foods to healthy volunteers so that physiological data on the blood glucose response in man could be obtained to supplement tables based solely on chemical analysis.
DISCUSSION:  The results demonstrate great inequality in the extent to which different carbohydrate sources raise the blood glucose and indicate that simple carbohydrate exchanges based on chemical analysis do not predict the physiological response.
I'm going to bullet-point and break up the discussion with some paraphrasing:
  • Differences Within Classes of Foods: Cereals ranged from 72% for wholemeal bread while wholemeal spaghetti was 42%.  Root vegetables ranged from 97% for parsnips to 48% for sweet potatoes.  
  • No Significant Relationship with Fiber:  "Surprisingly".  May have been due to the fact that wheat fiber has little impact on glucose, little difference seen between wholemeal and white counterparts.
  • Legumes:  Legumes "were remarkable in how little they raised the blood glucose."  Mean GI for legumes about half that for grains.
  • Sugar:  Sugar content was not related to blood glucose response even though absorption may have been more rapid. This is presumably due to the very small rise (20%) produced by fructose and reflected in the response to sucrose.
  • Fat & Protein:  Both fat and protein showed a significant negative correlation with glycemic index. Fat delays gastric emptying, protein stimulates insulin secretion. "However, it is not clear whether these actions or a direct effect of fat and protein in reducing the digestibility of food were responsible for the negative correlation."   It may not be as simple as just fat or protein content.
  • Starch:  Research has shown that starch type matters. 
  • Preparation:  Cooking method and time may lead to some GI variations. 


I capitalized and bolded that to shout at you for a reason.  The GI was not intended to be a diet gimmick.  It was intended to help diabetics better control their blood sugar.   The GI was standardized to a 50 g carbohydrate dose, and fat and/or protein were only considered for their impact if they went along with the carb in a particular food/preparation.  

Whatever the utility of the GI, and I do believe that for diabetics there is some utility vs. straight exchanges, it has become hopelessly clouded by diet gimmickry.  Furthermore, the progression of Type 2 diabetes and varying hormonal responses in those we would call "prediabetic", makes the GI somewhat of a crapshoot for these people in managing blood glucose levels.  The GI was never about the blood sugar roller coaster.   Likewise, it was never about some advantage for normal (non-diabetic) people to having some sort of "stable" blood sugar vs. "spikes".

So where I'm going with this is two-fold.  First, I want to discuss the anomaly that is the hypoglycemic blood sugar "crash" and its relevance to real life eating.  Secondly, I want to stress that a lower "glycemic" response to protein and/or fat is NOT in keeping with the spirit of the glycemic index.  Too often "low GI" is used to describe a meal that is simply low carb (or lower in carb) -- essentially a low glycemic load (GL).  Whether or not GL is important for health is a separate issue, but GI comparisons need to be macronutrient matched.  Jenkins *did it right* in this short, small study (6 normal men, 2 weeks per diet arm):  Metabolic effects of a low-glycemic-index diet.   I've chopped together the dietary composition and glucose responses.

As you can see, the diets were well matched to control well for JUST the glycemic effect.  Now look at the three post-meal glucose profiles.  Perhaps glucose dipped between meals in the regions "split" out, but I would note the following:

  • Pre-meal glucose is identical between diets for each meal
  • Post-meal glucose is slightly more for high GI at breakfast, but returns to baseline in the same time frame
  • Post-meal glucose is virtually identical between high and low GI for the lunch meal
  • Post-meal glucose profiles differ most significantly after the evening meal and if anything, you have extended hyperglycemia with the high GI meal.
The plots don't extend past 90 minutes, so some "crash" is not ruled out, but there seems to be no indication of these hypoglycemic dips, and there would be a longer way to go after the evening meal.  What exactly the evening results mean I'm not sure.  I'm reminded of a study where folks lost more weight eating their carbs at night.  Relevance?  Don't know.

Let's contrast this study with one that is often cited in GI literature and in low carb advocacy circles:  High Glycemic Index Foods, Overeating, and Obesity.  It has become clear that it is America's GI Man (as I've dubbed him) who is most responsible for trying to tie the glycemic index to obesity.  It is his schtick.  I've talked about this study before, in  A Matter of Control.  It is an example of a poorly controlled study.  This study impressed J. Stanton so much he wrote:

This article could easily be subtitled “The Study That Tells You Everything You Need To Know About Insulin, Blood Sugar, Carbohydrates, Satiety, And Obesity”. Yes, I admit to a degree of hyperbole—but this study is so well instrumented and controlled, and its results so informative, that I believe it’s important for everyone to read it.

If only that were true.  The more I look at the details of that study, the more troubling it becomes.  The study is bad enough, but how it is used, by Ludwig himself and others, is downright maddening.  Case in point, in his definitive 2002 paper, The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease, while discussing satiety and GI, Ludwig wrote:
For example, obese children were given high-glycemic index instant oatmeal or low-glycemic index steel-cut oats with identical energy and macronutrient content at breakfast and lunch, and ad libitum energy consumption was monitored throughout the afternoon.  Energy intake was 53% higher after the high- compared with the low-glycemic index meals.
WOW.  Sounds ominous!  No wonder things are so bad!!  But in contrast to Jenkins' study, all of the GI meals were not well matched.  Furthermore, the conditions were rather more than a bit artificial.  They took twelve, almost 16 year old obese boys, averaging 234 lbs, and did the following:
  • The night before they were checked into the hospital and fed a "standard low GI meal" of 18.5% of RMR around 6 pm and given a 5% RMR fruit and cheese snack around 10 pm.   This amounts to 386 + 104 = 490 calories.
Let's stop right there.  This meal was not a usual meal for these kids, it was high protein (30%), moderate carb (40%), "low" fat (30%).  Why?   Wash-outs and run-ins are supposed to be something "usual".  You're giving these boys less than one-quarter of their baseline RMR for dinner and snack?  This is around 15% of their TDEE.  For dinner!  Absurd.  
  • On Test Day morning ~ 7 am, they get a test meal, consisting of, again, 18.5% of RMR calories (1.63 MJ = 386 Calories).

Sorry, but there is unforgivable confounding in these meals.  The low-GI meal is not a carbohydrate meal.   This meal contains almost twice the protein and 50% more fat than the other two meals.  This is a comparison of glycemic load masquerading as glycemic index, and even if it were GL, that should standardize total carb content of the meal.    According to the researchers, however: 
This meal (a vegetable omelet and fruit) contained more protein and fat and less carbohydrate than did the high-GI meal and was designed to increase the range of GI in the study beyond that which could be achieved by manipulating only food structure and carbohydrate type (as with the high- and medium-GI meals)
In other words, beyond which any choice of carbohydrates could reasonably produce in the real world.  They had to include this ridiculously mis-matched meal to make their point.   To be fair, Ludwig wasn't referring to this meal in his review article, he was actually comparing the mid to high-GI.

The hormone and hunger responses were reported only after the breakfast meal.  So keep in mind that these 230 lb teenage boys, after having a measly 490 cals for dinner the night before, were fed an equally measly 386 calorie breakfast some 13 hours later (oh wait, they did have about 100 of the night time calories at around 10 pm).  The "low-GI" meal contained roughly 30 grams of protein -- not a bad amount -- while the oatmeal meals contained roughly half that much.   So with the unfair matchup as it was, here are the results:

Apparently all went "hypoglycemic" by 4 hours after the meal, with the high-GI meal producing about a 10 point greater depression in glucose levels.  Unfortunately, baseline levels aren't given so one has no way of knowing if these levels were anywhere near true hypo.  Look at that hunger rating.  We see a significant drop in hunger from the higher protein (assuming time 0 is right after eating).  I would note that for all meals, insulin is back to baseline at 4 hours at which time glucose levels off.   But comparing the two oatmeal meals, there was no difference in insulin, lower glucose and no difference in hunger in that "crash" phase where refined foods are supposed to be at their most dastardly.  
  • At noon, they were given lunch, the same test meal as breakfast, another 386 calories.  The two oatmeal groups are now up to only 30 g protein for the day, while the "low GI" group has had 60 g.  
  • They were then encouraged to engage in relatively quiet activities and ask for an ad libitum meal plate when they were "very hungry".  Time to request and amount eaten were recorded.
  • The ad libitum platters are described as a mix of foods but there is no indication of the overall macronutrient ratio, nor is there any indication what types of foods were consumed.  That might have been interesting.
Is it any surprise to anyone that all of the boys ate a lot following this semi-starvation? (Within 5 hours after noon).  Ad libitum intake ranged from 765 to 1385 calories!   The high protein group took longer to get hungry.  Who knew?  Perhaps the take away from a study like this is that if you want to restrict calories in obese teenage boys, keep their protein intake relatively normal.  Other than this, one can try to read more into this study than is there ... clearly Ludwig has.  Neither oatmeal meal was a normal meal.  The mid-GI was sweetened with fructose (so much for 30 g of that lighting up the reward centers and causing uncontrolled eating!).  The high-GI with dextrose (glucose) AND lactase treated milk.  Since they are so dastardly, why not test potatoes.  Something that a kid, or anyone for that matter, *might* eat as a meal.

By now, everyone should know where Ludwig is coming from:
A potential adverse consequence of the decrease observed in mean fat intake in recent years is a concomitant increase in dietary GI. A reduction of dietary fat tends to cause a compensatory increase in sugar and starch intake.
Too bad this is NOT what he tested in his study. If you want to test this, keep protein constant at around 15-7% (as it has been at for decades in the American diet) and have four groups of all combos between  high vs. low GI, and "high" vs. "low" fat (which would necessarily be lower or higher in carb).  I use the quotes because you will have to compare 40% to 30% fat, because that is a stretch of what may have happened.    Assessing hunger?  Try using normal caloric levels for the meals as well.  Just a thought!

It may seem like I'm going on a bit of a tangent here, but this is in sync with the evolution of the whole carb-insulin hypothesis (or what I call TWICHOO).   It seems the advocates are retreating to the fall-back position of refined carbohydrates as culprit.  Which in some ways, per Brand-Miller's work, is how the GI journey began in terms of obesity and type 2 diabetes.  

While everyone else seemingly moves on, a "low GI" diet will again be tested in groundbreaking research funded by NuSI!  This time protein will be kept constant at 20% while fat:carb will range from 65:15 to 40:40 to 20:60.   Too bad they won't match the food quality between diets to just look at the macros.  Too bad their Low GI diet is misnomered.  No, despite the fact that LF is to emphasize veggies and whole grains and limit sugar, the LGI will replace some refined carb (how much will there be?) with legumes and fruits (it's not that any low fat diet ever was heavy on legumes, and/or fruit, right?)  As such, where there could at least be some information gleaned from the middle group, it will be confounded from the get-go ... which makes one wonder why NuSI chose Ludwig's group for their most expensive ($14 million) study.    Oh ... and don't look now, but they are primarily looking at energy expenditure -- aka the elusive metabolic advantage.  It is strange that this was the outcome of Ludwig's 2006 devised study published in 2012, when we are still hearing about satiety and blood sugar swings and insulin trapping energy.  Make up your minds?


macrauchenia said…
Was there nothing in the Ludwig study to suggest that steel cut was more satiating?

The steel-cut vs. instant oats seems consistent to me with some Barbara Rolls study I read about once comparing apples to apple sauce to apple juice. Maybe that one was flawed too?

Maybe I've got it all wrong, but I've often been dubious about the satiety studies and the ones where they change plate sizes and such. The reason is that I question whether over the long term (weeks and months) people don't unconciously adapt to the tricks and gimmicks and tend back to a homeostatic weight.
Gordon said…
This study showed that food texture matters, and that powdered forms of food are obesogenic. Specifically, powdered mouse chow made the mice fat and the same chow in pellet form did not.

Rolled oats are certainly a step toward powder from steel cut oats. The processing that makes them cook faster makes the calories more available ...
MacSmiley said…
Good place as any to post this interview of David Jenkins discussing his own work.
StellaBarbone said…
Changing the shape of a food doesn't really change the amount of energy in the food, but chewing has an effect on satiety. Blending fruit into a smoothie also makes the fruit appear to be smaller -- one piece of fruit makes a tiny little shot -- so a satisfying smoothie includes several pieces of fruit, usually including one of the more calorically dense fruits like banana or mango because of their pleasant texture. You probably wouldn't put 4 or 5 pieces of fruit on a plate and sit down and eat them. It would appear to be a lot of food. Run those same 5 pieces through a blender and it looks like a reasonable smoothie, but it may be a 300-400 kCals beverage which you can easily slap back in a couple of minutes with very little decrease in hunger.
charles grashow said…
Glycemic Index, Glycemic Load and Glycemic Response: Scientific Consensus Statement
Glycemic Index, Glycemic Load, and Glycemic Response Definitions and Consumer Tip Sheet
scientists who spoke and PARTICIPATED in creating the Consensus Statement.
charles grashow said…

Q.A lot of the recipes in your book use fruit to add sweet flavors. Was this a way to limit refined sugar?

A.Exactly. People always say to me, “What about fruit? It has sugar.” But I have nothing against fruit, because it comes with its inherent fiber, and fiber mitigates the negative effects. The way God made it, however much sugar is in a piece of fruit, there’s an equal amount of fiber to offset it.

There’s only one notable exception: grapes. Grapes are just little bags of sugar. They don’t have enough fiber for the amount of sugar that’s in them. But I have nothing against real food, and that includes real fruit. Eat all the fruit you want. It’s only when you turn it into juice that I have a problem with it, because then it loses its fiber.
charles grashow said…
You have to add some fat to the smoothie - say 1/2 an avocado or some full fat kefir - that will satisfy you completely
Bris Vegas said…
Smaller plant food particles are digested far more efficiently. This means that more energy is recovered. If you swallow a grape or peanut whole (almost) NONE of the calories will be available. That is why herbivores chew their food so thoroughly.

Carnivores only chew food enough to swallow it. Meat is very easily digested.

It is very easy to eat 4-5 pieces of fruit. It just takes a little bit of chewing. A friend of mine once ate 10Kg (22lbs) of very ripe pears in a 24 hour period.
Bris Vegas said…
Humans evolved from frugivores. Our closest relatives chimpanzees and bonobos get the vast majority of their calories from fruit. Wild chimpanzees tend to avoid other foods if sufficient fruit is available

Whole fruit has almost no effect on triglycerides, blood glucose or insulin levels in humans.

What is even more surprising (to most people) is that whole fruits do not cause tooth decay in humans or other apes.
Jane Karlsson said…

Dr. David Jenkins is a world leader in nutrition research. He is the founder of the Glycemic Index (GI) that has paved the way for many of today’s popular diets. In this article, he names the top 10 healthy foods based on his research findings.

1. Whole wheat pasta (A low glycemic index whole wheat cereal)

2. Parboiled brown rice (Another low glycemic index whole grain cereal)

3. Whole barley (A low glycemic index food with cholesterol lowering fibre and a possible substitute for “sticky” rice)

4. Whole Grain Products (Reduce the caloric content of food and, for reasons not well understood, appear to be protective against both diabetes and heart disease)

5. Beans, Peas, and Lentils (Also known as 'pulses', these low glycemic index foods improve glycemic control and serum cholesterol, and are an excellent source of protein and minerals)

6. Nuts such as almonds, walnuts and pistachios (Shown to help lower cholesterol and reduce heart disease risk)

7. Tofu/Soy (An excellent source of protein and shown to lower cholesterol and reduce risk of heart disease)

8. Oranges/Tangerines (A lower glycemic index fruit shown to help improve glycemic control in Type 2 Diabetes)

9. Berries (A lower glycemic index fruit shown to help improve both glycemic control and blood pressure in Type 2 Diabetes, and is an excellent source of antioxidants)

10. Apples (Shown in our studies to independently lower the risk of heart disease)
Lighthouse Keeper said…
Let's not forget about cooking here, the one thing that truly separates us from everything else in the animal kingdom and has done for tens of thousands of years.
billy the k said…
"Whole fruit has almost no effect on triglycerides, blood glucose or insulin levels in humans."
Mr High-carb/Extremely Low-fat/Vegan himself disagrees with you: "Fruit, fruit juice, honey, and other simple sugars will cause the tryglycerides and cholesterol to rise in many people. Avoid these sweet tasting foods when trying to lower these blood values and your risk of heart disease." [John A> McDougall, MD. The McDougall Quick and Easy Cookbook. (1997) Dutton. p.252]

(some really yummy very-low-fat recipes from wife Mary in this book!)
billy the k said…
"Carnivores only chew food enough to swallow it. Meat is very easily digested."

Wrong again:

"Raw wild meat from game animals is tough, which is partly why cooking is so important. Advocates of the meat-eating hypothesis have themselves noted that humans differ from carnivores by our having small mouths, weak jaws, and small teeth that cannot easily shear flesh.

The way food moves through our bodies compounds the problem. In carnivores, meat spends A LONG TIME in the stomach, allowing intense muscular contractions of the stomach walls to reduce raw meat to small particles that can [THEN AND ONLY THEN] be digested rapidly.
Dogs tend to keep food in the stomach for TWO to FOUR HOURS, and cats for FIVE to SIX HOURS, before passing the food quickly through the small intestine. By contrast, humans resemble other other primates in keeping food in our stomachs for a short time, generally one to two hours, and then passing it slowly through the small intestine. Lacking the carnivore system of RETAINING FOOD FOR MANY HOURS IN OUR STOMACHS, we humans are inefficient at processing chunks of raw meat."
billy the k said…
Quite so!
"[Fruit] just takes a little bit of chewing."?
Just a little bit?

"Chimps in Gombe National Park spend more than six hours a day chewing,—which may seem high considering that most of their food is raw fruit...Most [forest] fruits
have to be chewed for a long time before the pulp can be detatched from the pieces of skin or seeds, and before the solid pieces are mashed enough to give up their valuable nutrients." [Richard Wrangham. Catching Fire. (©2009) Basic Books. p.139]

Señor Bris Vegas' friend was eating something quite different: Luther Burbanked/totally-domesticated "fruit"
StellaBarbone said…
Oh absolutely! No one ever overeats guacamole....
Bris Vegas said…
There are plenty of big soft juicy wild fruits.

Chimpanzees co-exist with gorillas in marginal environments. That means they are frequently forced to eat less palatable foods.

Captive chimps only spend about 30-60 minutes per day eating.
carbsane said…
Without going out and buying the ingredients, I'm left with my memory of instant oats from my college days. The meals were essentially two packets prepared with roughly 2 cups of fluid. Steel cut oats is going to suck that water up andmake a much heartier cereal.
billy the k said…
"marginal environments?" Nature is (naturally, of course) full of UNCOOKED foods, that is, foods that are raw and tough and "less palatable."

Captive chimps can bolt the foods they're fed just as quickly as humans can because the
foods are either a. domesticated [and so would never survive in Nature without man's protection] or b. COOKED.