Thursday, March 16, 2006

Revisiting the Glycemic Index

Last month the British Journal of Nutrition published a study, Towards understanding of glycaemic index and glycaemic load in habitual diet: associations with measures of glycaemia in the Insulin Resistance Atherosclerosis Study, that concluded "The present results call into question the utility of GI and GL to reflect glycaemic response to food adequately, when used in the context of usual diet."

Today, an article about the study is found in The State that opens with "It’s trendy, it’s diet-y and it’s hard to understand. So naturally, the glycemic index has inspired all kinds of “expert” commentary and plenty of confusion among consumers. But a USC researcher says the glycemic index is so flawed, it’s pretty much useless for people trying to lose weight."

So, just what did this study investigate and what were the findings? More importantly, are the findings useful?

As I read through the full-text of the study, something immediately popped-out - the calculations for the food values were based on faulty assignment of glycemic index values. For example, cheese was arbitrarily assigned the same value as milk. Milk in the United States has a glycemic index value of 40 - low GI. Cheese hasn't been assigned a value, and you'd be hard pressed to find someone who would consider cheese as having a GI of 40 due to its protein and fat content, which makes it nearly impossible to test.

The reason it is difficult to test cheese is that consuming enough to reach the "test-load" intake - that is, eating 50g of digestible carbohydrate to reach the level to test - means you'd have to find enough people able to actually eat 8.6-pounds of cheese! So, that alone makes the data suspicious in my mind - assigning a value to an untested food, bad science!

But, that wasn't the only issue I had with the design of this study. The researchers used data from Food Frequency Questionaires (FFQ) not designed to measure glycemic index and/or glycemic load. Whether the assigned values are correct or not (I think they were inaccurate), those considered to be consuming a "low glycemic index" diet were in fact still eating a high level of carbohydrate at baseline and after five years.

One red flag that makes this an easy conclusion is the low amount of fiber in the diets - at baseline 16g on average and at the five-year evaluation 15g on average. This level was in the context of a high intake of carbohydrate - at baseline the average intake was 235g of carbohydrate, at five years it was 222g of carbohydrate.In looking at the assigned values from low to high for glycemic index - each level was HIGH, therefore the data was really only evaluating the effect of a high GI diet with a higher GI diet with an even higher GI diet, not a low GI against a high GI diet.

That's an important distiction here - and it is further supported by looking at the Glycemic Load (GL) levels assigned and assessed.

It is believed that GL is much more important at the end of the day than the GI of any one food or an entire day's GI measure. That's because the GI is not based on serving size of a food, but rather consumption of a specific level of digestible carbohydrate - 50g - in one sitting. With some foods a very small portion reaches this 50g intake, for others, like the cheese example above, the portion is so enormous it makes testing the food impossible and unnecessary. GL, on the other hand, is a calculation based on actual serving size of a particular food. For example, carrots have a high GI, but because you'd have to eat such a large portion to consume 50g of digestible carbohydrate, the actual serving size of a portion is measured and carrots are ranked as having a low GL.

I was happy to see that the research team tried to evaluate both the GI and GL because it provides support to my contention that few were following a low GI diet since only a small number actually fell into the low GL category - of the five groupings from low to high for GL, only one was calculated to consume less than the GL target of 100 to be considered consuming a low GI/GL diet. All the other groups ranked well above that 100 GL mark, so they were all consuming a higher GL diet. So again, we find comparison between high GL with high GL for the vast majority of those evaluated and for the one group that did seem to consume below the target GL level of 100, there is question with that due to the arbitrary assignment of GI values for untested foods as mentioned above.

My own opinion about the glycemic index is that it is very complicated and not really a very useful tool for the average dieter trying to lose weight. There is a lot of picking and choosing based on GI or GL numbers instead of nutrient-density. That said though, if one does follow it correctly - strive to maintain a low glycemic load each day by picking nutrient-dense selections - you're much better off than just eating a calorie restricted diet consuming whatever with only calories as your focus.

I personally think there are easier ways to plan what you're going to eat each day - simple basics - like real whole foods as the foundation of each day with lots of non-starchy vegetables, limit sugars and skip the processed junk food. Odds are high that if you're doing that, you're consuming a low GI diet anyway!

Here's an idea for researchers serious about investigating the efficacy of a low GI/GL diet - deisgn a study specifically to measure two or more groups consuming different levels over a period of time. Give the low GI/GL group specific foods allowed with instuctions on how to be compliant with the diet, the next group same thing, but higher value for GI/GL allowed and then have a control group eating whatever they want. Follow them for five years and then get back to us with REAL DATA!

4 comments:

  1. Regina, thanks so much for looking at this! One question...my issue with this study was that they didn't actually look at blood glucose levels in response to a particular type of meal. My understanding is that one of the benefits of a GL diet is reduced insulin response after a meal due to a slower, more tapered blood glucose.

    Didn't this study essentially test whether five years of a low-GI/GL diet had a longtime effect on glucose ingestion?

    If so, all the media reports completely missed the boat (IMO) on whether or not a low-GI/GL diet is useful for weight loss, since the point of the diet is to reduce cravings due to high insulin response.

    Am I missing something?

    ReplyDelete
  2. Didn't this study essentially test whether five years of a low-GI/GL diet had a longtime effect on glucose ingestion?

    Since none of the groups actually ate what is defined as a low-GI diet, and only one sub-group technically qualified as low-GL, the data IMO is worthless to assess the value of using the GI/GL as a diet planning tool.

    Add to that the originial collection of FFQ's wasn't intended to measure GI/GL and the arbitrary assignment of GI/GL values for untested foods - what a nightmare this data is!

    The researchers basically attempted to go back, re-use collected data in a way not designed for and come to a conclusion afterward. Now, when they actually get at least two groups together and design to the study to actually measure between two or more groups (including a control group) the effects of different GI/GL values, then we might have valuable data - right now, this data is worthless IMO.

    ReplyDelete
  3. Great post Regina! I have question. The glycemic index talks about the shape of the insulin spike caused by the carbohydrates in a particular food. 50 gms of carbohydrate from white bread will cause a higher narrower spike in insulin than 50 gms of carbohydrates from beans, which will have a low, broad spike. But the area under the curve will be the same, meaning the same amount of insulin will be secreted, but over different periods. Will these differently shaped spikes make the same contribution to insulin resistance?



    It also strikes me that insulin resistance will queer the glycemic index anyway. If someone with high insulin resistance eats low glycemic foods, they’re still going to produce more insulin than someone who eats the same food and has lower insulin resistance. But everyone who has been eating the standard American diet has some degree of insulin resistance. The only way to lower insulin resistance is to lower carbohydrates absolutely, not just substitute carbohydrate from beans for carbohydrate from white bread.

    ReplyDelete
  4. >>>The glycemic index talks about the shape of the insulin spike caused by the carbohydrates in a particular food. 50 gms of carbohydrate from white bread will cause a higher narrower spike in insulin than 50 gms of carbohydrates from beans, which will have a low, broad spike. But the area under the curve will be the same, meaning the same amount of insulin will be secreted, but over different periods. Will these differently shaped spikes make the same contribution to insulin resistance?<<<

    My understanding is that because the glucose response is lower - the spike not as high in blood sugars - a higher than needed amount of insulin isn't going to be secreted to stablize the blood sugars, which often leads to hypoglycemia and triggers hunger to raise blood glucose back up again. I would think that how this affects insulin resistant would be highly individualized and dependent on one's metabolism and complications in their metabolism.

    I'm not a big fan of the GI/GL mainly because it looks at foods in isolation and once you consume those foods as part of a mixed meal, basically all bets are off - you're going to have a different response with, say lentils, beef and brown rice than you will with lentils, turkey and brown rice.

    >>>It also strikes me that insulin resistance will queer the glycemic index anyway. If someone with high insulin resistance eats low glycemic foods, they’re still going to produce more insulin than someone who eats the same food and has lower insulin resistance.<<<

    Yes, and the testing is done on healthy individuals for the most part - at least the major lists available are based on healthy subjects, although some studies have used those with diabetes or IR as part of their study design.

    >>>The only way to lower insulin resistance is to lower carbohydrates absolutely, not just substitute carbohydrate from beans for carbohydrate from white bread.<<<

    LOL - preaching to the converted!!!

    At the end of the day I really think what's most important is to choose nutrient-dense foods, some of which may be a bit higher in carbs, but dense with essential nutrients....and whatever amount of carbohydrate is being consumed throughout the day MUST be spread out to limit spikes.

    ReplyDelete