Personal SGIs

Hi Everyone,

An engineer friend, Derek Paice, came up with the idea of the "substance glycemic index," or SGI, which is essentially a GI but for all foods, not just carbs. He describes it in a free online booklet at Unfortunately, he's closed his website, which had some other information.

My idea was that there's tremendous power in the online community, and if there were a way for people to measure their own SGIs and then post them to TuDiabetes, eventually there might be enough data to help people who didn't want to measure their own.

For example, Derek measured the SGI of whole-wheat bread as 100, cheese pizza at 59, Kellogg's Bran Buds at 25, a Fugi apple at 19, steak at 17, cantaloupe at 12, cauliflower at 7, and cheese omelet at 6. He also measured a complete meal and found that the SGI of a meal containing steak, cabbage, Merlot wine, sugarfree gelatin dessert, 2 candies, and coffee with creamer had the same SGI (77) as 1.5 slices of whole-wheat bread.

The idea is that you eat a food or a meal and then measure the area under the curve (AUC) until it returns to baseline and compare that AUC with your AUC for a control substance, expressing it as a percentage. Derek used whole-wheat bread, but I think that's not a good control because the ingredients of whole-wheat bread can vary so much. Glucose would be better.

He used 60 grams of food, but one could use any amount. The point is to compare that food with an identical amount of the control food (bread or glucose).

The GI used 50 g of "available carbohydrate," not a certain weight of food, and some foods might have carbs with a high GI but not much carb. The glycemic load is supposed to correct for that, but you can't get GLs for foods for which no one has calculated GI, for example, meat and nonstarchy vegetables. I'm type 2, but meat can make my BG go up.

Also, GI only measured for 2 (nondiabetic) or 3 (diabetic) hours, and when you have type 2, you can stay high for longer than that, so measuring until you return to baseline makes more sense.

Clearly, this wouldn't work for anyone using bolus insulin, but knowing the SGIs that had been calculated by type 2s could help some type 1s with food choices.

Anyway, right now it's just a vague idea. I was wondering if anyone else thought it had merit.


I'm trying to figure out how this number would help me dose insulin. Does, for example, the substance glycemic index or SGI of 77 for a whole meal translate to a customized insulin dose for me?

I generally don't eat high glycemic index foods. I definitely don't drink carbs such as juice or milk. I enjoy a low-carb high-fat way of eating and dose my insulin based on the carb, protein, and fat content of any meal.

If I understand Derek Paice's SGI system, it appears designed to help people that consume a large amount of fast acting carbs as part of their daily intake. I admire his attempt to quantify and help people respond in a rational, more scientific way, to control post-meal blood sugar excursions. If this works dependably, it could help many people.

I wonder about the variance in food absorption, even for one person from meal to meal and day to day. There's also the insulin absorption variable. We all don't absorb the same dose of insulin the same metabolically form instance to instance.

I'm a big fan of the law of small numbers, which says that smaller inputs (read food, insulin) to any system will result in smaller output errors (read blood glucose response).

Terry, I agree with you about low carbing and variance depending on a myriad of factors. A list of SGIs would mostly help newbies, who didn't realize how much better it would be to have a cheese omelet than a slice of bread. Seeing the numbers might convince them. And the GI lists don't include the cheese omelet, just a lot of carby foods.

For insulin dosing, a person would have to test the SGI vs insulin given, just as one tests to find the insulin:carb ratio.

To be honest, I haven't found that the GI has much impact on Type 1's; only in the sense that high GI foods are often also high carb foods which of course affect both types! For example the pizza and cereal are listed above as significantly lower than whole wheat bread when pizza and cereal are prohibitive for many Type 1's.

I've been aware of the glycemic index concept for a long time but have never tactically used it. I avoid fast acting foods, except when I'm low or just want a taste.

Dietitians seems to like this concept but I don't value its utility except in a general way. And I'm skeptical, generally, about dietitians' advice due to their embrace of the Big-Food friendly tactic to eat whatever you want and dose insulin accordingly. This borders on medical malpractice.

I have to be honest, I think the glycemic index and load concepts are basically useless. At least in a general sense and certainly as it applies to anyone who is beta cell challenged and dependent on exogenous insulin. In the end those of us on insulin have to count every carb whether it is high or low glycemic index.

That being said, if you have T2 and still produce insulin but not enough to cover a high glycemic index high carb meal than eating low glycemic index foods will result in better blood sugar control. But this is a complicated function of your phase I and phase II response and it varies significantly from individual to individual.

And the biggest difficulty I have with the concept is that it doesn't scale. it is based on the assumption that you eat a "modest" level of carbs and that your phase II response can provide a reasonable level of control at 2 hours. If you eat a huge amount of low glycemic foods it can overwhelm both your phase I and phase II response and the benefit of a low glycemic food will just not be observed. And by the same token, a small amount of high glycemic food may be handled by your phase I response and hence small amounts of carbs will see no difference between high and low glycemic foods.

And finally, our responses to foods are very complicated, fat slows digestion, protein has an effect and we have responses to foods based on amounts (i.e. the chinese restaurant effect). I just think while these ideas may help use as individuals to understand our response to foods they don't generalize and should be understood as just general "trends" not hard and fast rules.

I tend to take lower insulin doses for lower index foods, like an apple, and a bit higher does for refined carbo. I actively overshoot my standard dose when I eat high glycemic foods, like pizza.

Here's a study with a formula for including protein and fat in type 1 insulin dosing:

There seems to be some confusion here. This isn't about the traditional glycemic index. It's about a new measurement that includes protein and fat.

In general, I don't pay any attention to the GI except for chana dal, which has a GI of 7. I've tested it, and it hardly raises my BG at all at 2 and 3 hours. However, at 4 and 5 hours there's a larger rise. Pasta also has two peaks. I think both are because of protein that is released later.

I understand what you are saying, you are right when I read the paper clearly there was an element that looked at the response from protein and fat. But my experience is that the response is just not additive and that seems to be a core principle of all this glycemic index stuff is that you can just add it up. I just don't think it works out that way. I have many meals where I have figured out exactly what insulin I need based on experience, but new meals with new mixes of ingredients always seems to be unpredictable.

Oh, Terry. Now I'm gonna bicker with you. You sound like my husband. What the heck do you eat?! Please don't say oatmeal, Cheerios, and popcorn.

So, your saying that you dose via experience and intuition. I think I do that more as I age. But, Brian, you never count? Never? There's been a lot of chatter about this fat and protein stuff, so I'll read the article, if its free. Perhaps they are close to curing our inability to eat pizza!!! The age old pizza riddle. It has plagued me for many years.

I eat eggs, nuts (almonds, walnuts, macadamia nuts, filberts, Brazil nuts), lower-carb veggies, onions, tomatoes, avocados, bacon, fish, sausage, ham, pork, ground beef, steaks, full-fat dairy including heavy cream and yogurt, cheese, leafy greens, beans in moderation, and berries (strawberries, raspberries, blueberries, blackberries).

I used to eat a lot of oatmeal, but 27 grams of carbs for a serving is too carby for me. I don't eat Cheerios or any boxed cereal or other cereals like cream of wheat or grits.

I still have the remnant of a sweet tooth and will make sparing use of sugar and maple syrup.

I'll drink a glass of red wine with dinner once or twice a week. A few times per year I'll treat myself to a pint of Guinness @ 19 grams of carby goodness. Otherwise I don't drink my carbs.

I only eat two meals per day, breakfast between 11 a.m. and 1 p.m. and dinner between 5 p.m. and 6 p.m.

I will have an ounce of nuts for a snack mid-afternoon. I don't like eating anything after dinner. I'm currently doing one 24-hour fast each week from dinner one day till dinner the next. This way of eating satisfies for me. I'm not hungry between meals. My current dietary goal is to introduce more variety to my diet. I'd like to make a habit eating sardines.

More than you wanted to know?

Here's the conclusion from your linked study:

A mixed meal effectively elevates postprandial glycemia after 4-6 h. Dual-wave insulin bolus, in which insulin is calculated for both the carbohydrates and fat proteins, is effective in controlling postprandial glycemia.

This has been my experience, too. I use a dual-wave bolus for every meal I eat.

I wont give you a hard time, unless you start in on those kipper snacks.

I'm off the eggs, cheese, booze, and meat...and the sardines, if you count those as meat. Like the cream, but off cows milk. Cholesterol.

Terry, Do you sling code for Medtronic? You gotta be kidding me. I hope to see your smiling face on one of their billboards.

How do you determine how much to use for the protein and fat?

I do count, I count all the carbs and half the protein. But after that I adjust based on experience. So I know that I actually can have a modest serving of lasagna which is high in fat and protein and I halve my bolus. But pizza has always been a disaster so I don't even try it anymore. For the life of me I don't know why they are that different. But I would not be able to predict that from the GI of the lasagna ingredients or the pizza ingredients. And that is where it all breaks down for me, I don't eat prepared and processed foods the vast majority of the times.