Making sense of the Food Insulin Index (with recipes and food lists to stabilise insulin and blood sugars and help you to lose weight)

The Insulin Index of Foods provides powerful insights into how the food we eat affects our insulin.   

Unfortunately, many people find the data complex and confusing, so it tends to be underutilised and misinterpreted (even by the original researchers).   

However, once we understand the implications and the limitations of this data, the Food Insulin Index empowers us to make better food choices to fine-tune our blood sugars and optimise our diabetes management.   

What is the food insulin index?

The original 1997 food insulin index data was published in An insulin index of foods: the insulin demand generated by 1000-kJ portions of common foods by Dr Susanne Holt, Professor Jenny Brand-Miller and Professor Peter Petros at the University of Sydney.  You can download the free pdf of the original paper here.  

Small portions (i.e. 1000 kJ or 239 calories) of foods were fed to study participants.  The change in blood insulin and blood sugar levels were measured over the following two hours.   

The chart below shows the change in insulin levels in the two hours after consuming different meals.  Notice how, after eating pure glucose (black line) insulin levels return to baseline.  However, insulin levels are still elevated two hours after consuming high-fat meals like milk (aqua line).

White bread was assigned a score of 100%.  So any food with a score higher than 100 has a larger insulin response than white bread, while anything less than 100 has a smaller insulin response than white bread.  

The food insulin index chart from the original paper below shows jelly beans have the largest insulin response, while foods like eggs, butter, bacon and olive oil have a much smaller insulin response over the two hours of the test.  

This image has an empty alt attribute; its file name is insulin-index-of-foods-1.png

People with diabetes often use jelly beans to raise their blood sugars quickly.

Low carb foods like eggs, butter, bacon and olive oil maintain more stable insulin and blood sugar levels.  

Unfortunately, with only 38 data points, this information is difficult to make much sense of.  Thankfully, the food insulin index database has been expanded and now contains more than one hundred foods.  Dr Kirstine Bell’s 1994 thesis, Clinical Application of the Food Insulin Index to Diabetes Mellitus published the full set of more than a hundred data points which I have plotted from lowest (on the left) to highest (on the right) on the chart below.  

It’s hard to see the names of all the foods on the chart.  If you want to drill into the details, you can check out the Tableau version (on your computer, not your phone) here.   

Simply looking at all the data in this format still doesn’t tell us much.  But, with a little bit of analysis (as detailed below), we can make more sense of it to help us make more informed food choices to stabilise and reduce our insulin requirements.  

Which macronutrient raises insulin the most?

By plotting the different quantifiable parameters of our food (e.g. carbohydrates, protein, fat, fibre and fructose) versus the insulin index data, we can better understand how the food we eat affects our insulin response to various foods.  

We can then combine them to accurately predict our short term insulin response to any other food.  Not only does this empower us to make better food choices, but we can also more accurately calculate insulin doses for people with diabetes who need to inject insulin at every meal.   

Do carbs raise insulin levels?

The chart below shows carbohydrate vs the food insulin index of foods data (see more detail in Tableau here).  

In the top right corner of the chart, we can see that it’s the processed carbohydrates that tend to raise insulin the most (e.g. bread, pancakes, cereals).   

However, interestingly, to the left of the chart, we can also see that low carb high protein foods (e.g. fish, steak, prawns, low-fat cottage cheese and eggs) also raise insulin significantly.   

By analysing this data, we found that protein requires about half as much insulin (56%) compared to carbohydrates over two hours.   

People with Type 1 Diabetes typically find that they achieve better blood sugar control when they inject insulin to “cover” the protein in their meals.  This is particularly important when following a low carb diet where bolus insulin doses are smaller.  

Accounting for the insulin response to carbohydrates and protein

In the updated chart below (see in Tableau here), we can see that once we account for the insulinogenic effect of protein, we get a much better prediction of our insulin response to food with the high protein foods now much closer to the trend line.      

How do fructose, fibre and fat affect our insulin levels?

But after accounting for protein, we noticed that juicy fruits (like apples, oranges, grapes, raisins, etc.) that contain their carbohydrate as fructose were sitting well below the trend line.  Even Coca Cola made with high fructose corn syrup seemed to have a lower insulin response than predicted.  

We also noticed that high fibre cereals (e.g. all bran, porridge) tended to have a lower insulin response than predicted.  While fibre is a carbohydrate, it doesn’t tend to be digested for use as energy, so it doesn’t tend to raise our blood sugars as much.  

We also found that fat has a small impact on our insulin response even within the two hours.    

Once we account for these factors, we can calculate the percentage of insulinogenic calories which aligns much more than carbohydrates alone (see Tableau version here).  

This precise understanding of our insulin response to the food we eat enables us to:

  • rank foods based on their % insulinogenic calories (i.e. foods that are more or less ketogenic),
  • accurately calculate the insulin load of meals (to help you stabilise your blood sugars), and
  • more accurately calculate insulin dosing for people managing diabetes.

Should you be concerned that protein raises insulin?  

There has been plenty of confusion and controversy in low carb circles around the fact that protein is insulinogenic, with many people avoiding protein in an effort to stabilise their blood sugars.  

However, while protein does elicit an insulin response, it’s important to note that foods with more protein tend to require less insulin.  This is mainly because they contain fewer carbohydrates (see interactive Tableau version here).

If you aren’t consuming a lot of carbohydrates, most of the amino acids can be converted to glucose (via gluconeogenesis).  However, unless you are injecting insulin to maintain stable blood sugar levels, you probably don’t need to worry too much about gluconeogenesis or excess protein overflowing into your bloodstream.  

Gluconeogenesis is actually a highly energy-intensive process, and your body would much rather get its energy from fat or carbs than protein.  We lose about 25% of the energy from the protein we eat in the conversion to usable energy (ATP).  By comparison, we only lose about 8% of the energy from carbohydrates and 3% of the energy from fat.   So, although protein can be used for energy, your body would much rather not have to convert protein to energy.  

In fact, due to the strong satiety effect of protein, foods and meals with a higher percentage of protein tend to help us eat less, so we tend to lose body fat and actually reverse our insulin resistance over the long term when we consume foods that contain a higher percentage of protein!  

As shown in the chart from our satiety analysis, we tend to eat fewer calories overall when our diet consists of a higher percentage of protein.  We find that the people who have the best results in our 6 Week Nutritional Optimisation Masterclass tend to be the ones who are able to get their protein intake up to around 40% of total energy by reducing the refined carbohydrate and fats in their diet.  

It’s also worth noting that, as shown in the chart below from the analysis of our series of nutrient-dense recipe books, a higher percentage of protein (up to around 50%) tends to align with a higher nutrient density.  

What does insulin do?

We often think of insulin as an anabolic hormone (to build things up).  However,  it can be more helpful to think of insulin as anti-catabolic (to stop your body from falling apart).  

Your body uses insulin to control the flow of energy from storage (via the liver) into your bloodstream when there is plenty of food coming in from your diet.  When there is less energy available from food, glucagon stimulates the release of stored energy and balances the effects of insulin.  

When you consume more protein than you need for essential bodily functions (muscle protein synthesis, neurotransmitters etc.), your pancreas will raise insulin to shut off the release of stored fuel from your body while you use up the extra energy.  

Although fat doesn’t require as much insulin over the short term, all food triggers an insulin response.

Limitations of the food insulin index data

Before you assume fat is a free food because it doesn’t raise insulin, it’s important to note that the food insulin index testing was undertaken over only two hours, so it underestimates the insulin response from protein (which actually elicits a rise over approximately eight hours) while fat raises insulin for a much more extended period.  

People with Type 1 Diabetes need to inject bolus insulin with meals and long-acting basal insulin to cater for the portion of their meals that are processed over a longer period.  Someone on a low carb diet may have around 70% of their insulin requirements as basal insulin with only 30% of their daily insulin requirements as bolus doses around meals (these ratios may be reversed for people following a high carb diet).     

The chart below shows the changes in insulin levels across a 24 hour period.  If we only consider the insulin response in the two hours after meals we don’t account for the significant amount of insulin released across the other 16-18 hours of the day, particularly from low carb diets.  We can’t simply assume that because the insulin rise after meals is low that our total insulin secreted across the day will also be lower.  The reverse may actually be the case.  


Insulin and oxidative priority 

Our insulin response to the different macronutrients appears to be proportional to their oxidative priority (i.e. the order in which they are burned), how easy they are to keep in storage and how much storage capacity we have for each of them.

  • We only have limited space to store carbohydrates in our body.  It has to be burned off quickly, so the body raises insulin sharply to hold back your glycogen and adipose tissue in storage.  
  • Any excess protein needs to be burned off fairly quickly, so we get a lower insulin response.  
  • However, fat is easy to store, and your body is more than happy to store it, so we get a much smaller insulin response to fat over a more extended period.   

While a diet with more fat is undoubtedly helpful to stabilise your insulin and blood sugars, we can’t make the assumption that it will lower our insulin levels across the day or lead to fat loss.   

What is the difference between glycemic index and insulin index?

You may have heard of the Glycemic Index (GI) which was developed by the same team at the University of Sydney and has been a popular method to identify high carbohydrate foods that do not raise blood glucose as much over the short term.  

The GI is a measure of maximum increase in blood sugar in the two hours after eating a food that contains 50 g of carbohydrates.  The GI is useful to identify high carbohydrate foods that raise our blood sugars less than others.  

As part of the development of the Food Insulin Index, the researchers calculated the Glucose Score of all the foods they tested.  If you’re looking to stabilise your blood sugars, the Glucose Score is actually more useful than the Glycemic Index because it is based on the area under the curve glucose response for all foods, not just high carbs food.  The insights from the glucose score data are also very informative, as we will see below. 

Do carbohydrates raise your blood sugar?   

While the drivers of insulin are complex, the thing that raises glucose is pretty simple!  It’s carbohydrates (see Tableau data here).  

If your blood sugars are rising by more than 1.6 mmol/L or 30 mg/dL, then you should reduce the carbohydrates in your diet to get off the glucose rollercoaster to avoid the hangry feeling that comes from constantly rising and falling blood sugars.   

In our 6 Week Nutritional Optimisation Masterclass, we guide people to dial back their carbohydrate intake if their blood sugars are going too high after meals.  But, as with most things, it’s crucial to find the right balance between extremes.  

The chart below from our analysis of our series of nutrient-dense recipe books shows that we maximise nutrient density when we have 15 to 20% of our energy from non-fibre carbohydrates.  

There is no need for most people to be concerned about carbs from non-starchy vegetables that can actually be quite nutritious, and avoiding carbs altogether may make it harder to get the nutrients you need.  

How does fat affect your blood sugar?

Conversely, a higher percentage of fat tends to lower blood sugars after meals significantly (see Tableau data here).  

Fat is an excellent source of slow-burning fuel, but it can be easy to overdo.  If your goal is to lose fat from your body, then you want to prioritise foods and meals that provide more satiety and nutrient density rather than more fat from your diet.  

If you have diabetes, it can be challenging to find the right balance between stabilising your post-meal blood sugars (which is related to the amount of carbs you eat) vs lowering your fasting blood sugars and insulin across the day (which is more related to your body fat levels and your total daily energy intake).  

As shown in our chart below from our analysis of half a million days of MyFitnessPal data, foods with a higher percentage of fat can be easier to overeat.  

Further, as shown in the chart below from our analysis of our series of nutrient-dense recipe books, foods with more fat tend to be less nutritious.  

Does protein raise blood sugar?

As shown in the plot of protein vs glucose below (see Tableau version here), high protein foods will raise your blood sugar a little, but much less than low protein foods that contain refined carbohydrates.  As a general rule, consuming a higher percentage of protein will lower your blood glucose.

When you eat protein, you release both glucagon and insulin.  In most people, insulin and glucagon are balanced, so your blood sugars remain stable after a high protein meal.  But if you are insulin resistant, this signal becomes imbalanced and leads to a more dominant glucagon response from the liver and some elevation of blood sugars (see Why do my blood sugars rise after a high protein meal).  

If you are injecting insulin, you may need to top up with some extra insulin so you can metabolise the protein to repair your muscles while also keeping your blood sugars stable.  Avoiding protein can lead to a reduced metabolic rate and reduced satiety.  

The good news though is that increasing the percentage of protein in your diet tends to lower your overall energy intake and hence reduce your body fat levels.  This, in turn, will lower basal insulin requirements (i.e. the insulin that you need when you are not eating) because you have less body fat to try to keep in storage.

So, unless you require a therapeutic ketogenic diet (for the management of Alzheimer’s, dementia, Parkinson’s or epilepsy), there is no need for most people to worry about eating “too much protein”.  For people with a functioning pancreas (i.e. you are not injecting insulin), focusing on reducing carbohydrates should be adequate to stabilise blood sugar levels without worrying about cutting back on the protein in their diet.  

Insulin response to fat

Higher fat foods tend to cause a smaller insulin response.  However, there is still a small insulin response, even over the two-three hours (see Tableau version here).

Fat is the last macronutrient to be burned (after alcohol, ketones, excess protein and carbohydrate).  Fat is easily converted to energy or stored, so only a small amount of insulin is required to store any fat that is left over.

Some people like to believe that they can lower the insulin load of their diet to lose weight (a bit like someone with uncontrolled type 1 diabetes).  If you are making your own insulin, your pancreas will not produce any more insulin than is required to hold your fat in storage while you use up the energy coming in from your mouth.

As shown below, our insulin levels across the day are proportional to the amount of fat we are carrying.  

obesity vs insulin.png

Similarly, your fasting blood glucose levels are also proportional to your weight.  

The more body fat you have, the more insulin you will require to keep your body fat locked away in storage.  This is why many people who are obese also have high fasting insulin levels.   It is not that insulin is causing them to be fat but rather that their pancreas is working overtime to keep insulin high to hold their fat in storage. 

If your goal is fat loss, then you need to find a way of eating that allows you to be satisfied with less energy.  Your fasting insulin levels will reduce as your body fat levels decrease.

The misguided Carbohydrate–Insulin Hypothesis simplistically assumes:

carbs -> insulin ->  fat storage 

But in reality, it is a little bit more complex.  It’s actually:

low satiety nutrient-poor foods -> increased cravings and appetite -> increased energy intake -> fat storage -> increased insulin

Hence, the solution to managing your diabetes, blood sugar, insulin levels and avoiding the myriad of complications of metabolic syndrome is:

high satiety nutrient-dense foods and meals -> decreased cravings and appetite -> decreased energy intake -> fat loss -> lower insulin levels

Putting it into action 

So hopefully you can see that managing the insulin load of your diet is important to help you stabilise your blood sugars and insulin after meals.  However, if you want to reduce your insulin across the day and reduce your fasting blood sugars, you also need to prioritise high satiety nutrient-dense foods that will promote satiety and help you to lose weight.  To help you apply all this we developed a number of tools.

Get your free food list

Combining our understanding of the insulin response to food with nutrient density we have developed a range of printable pdf food lists to help you optimise your diet to achieve your goals that you can access for free here.  


Finally, to help people make sense of the food insulin index and find the right balance between stabilising insulin and blood sugars, we have created a series of 22 recipe books optimised for a range of goals.  

The table below will help you find the right recipe book that aligns with your goals, from therapeutic ketosis to aggressive fat loss.  

Approach Ideal for 
Therapeutic ketosis Designed to achieve therapeutic ketone levels (e.g. for epilepsy, dementia, or Parkinson’s).
Nutritional ketosis Designed for someone who enjoys a ketogenic way of eating (without therapeutic ketone levels).
Low carb & diabetes Stable blood sugars and weight maintenance on a nutritious low carb diet.  
Blood sugar and fat loss Ideal for anyone with elevated blood sugars and body fat to lose.
Fat loss Rapid fat loss with less hunger and cravings due to nutrient deficiencies and adequate protein to prevent muscle loss.
High protein:energy High protein:energy ratio for aggressive fat loss. 

Click here you’re interested in learning more about the recipe books or to purchase one.  

To kickstart your journey towards optimal get your free program and one of 70+ food lists personalised just for you!  

Marty Kendall

  • […] that protein appears to contribute to insulin demand I ran a number of scenarios with the food insulin index data to see if insulin requirement is better predicted by carbohydrate in a food plus some proportion of […]

  • Alec Collins says:

    But the foods which cause the least insulin secretion, according to the above graph, are low carb. So, “independent of calories or carbohydrates” seems a little misleading.

  • martykendall says:

    I had the same question Alec. But if you read the actual thesis and the associated papers you’ll see that they actually kept the calories and carbohydrates consistent. They weren’t really testing a low carb approach. The only variable was the insulin index of the foods. In effect they would have been choosing foods with more fibre and less protein. A much better improvement would be achieved by adopting a low insulin load approach. Hopefully they can repeat a similar study in the future with this approach.

  • […] [next article… the latest food insulin index data] […]

  • […] [next article…  latest food insulin index data] […]

  • stenbj says:

    Great information ! I would have liked to see “lite” or skim milk included. The Swedish “Keyhole marking” requires less than 0.7% fat and in my opinion it is a pure early disease and widow maker! A 12 year study of 1600 men showed that those with lowest intake of milk fat compared to those with most high fat dairy had 3x = 300% higher incidence of central obesity = lots higher risk for diabetes, heart disease and risk for death of all causes. The study is here:
    As also well known, skim milk was used to fatten pigs when pig fat was more important than lean pork. It would not surprise me if skim milk was also given to geese with grain to speed up fois gras. But for now its insulin index is most interesting!

  • stenbj says:

    Thanks for reply!
    Full cream milk comes out at insulin index 25% from the diagram in your “The most ketogenic foods”. What am I missing to arrive at 40% ?

    One more thing to address is probably the carbohydrate fructose as has a very low insulin response but produces liver fat and leads to insulin resistance in a different pathway than through (over-) exposure to insulin. It may deserve an exclusion and a special goodbye? Is it one reason raisins are coming out good in the insulinogenic index, its 28% fructose? Looking forward to your analysis!

    • Maybe you have fattier milk than we do.

      The 40% in the list is based on the USDA food database.

      I just ran the numbers using the full cream milk that we have in the fridge (4.4g fat, 3.2g protein and 4.6g carbs) and I get 36.1% insulinogenic calories.

      Here’s the link to the calculator…

    • That’s a really good observation about fructose meaning that the insulin response is very low. I will have to look into that in more detail.

  • […] being irritable all the time.  I came across the Optimising Nutrition blog discussing an “insulin index” – much more useful to me than a “glycemic index” since my goal is shifting […]

  • […] Kendall from Optimising Nutrition posted an incredibly dense set of tabled data. Luckily he is an awesome person and took the time to […]

  • […] are two graphs from Marty Kendal of The first shows the insulin response of some typical foods from the Standard American Diet. The […]

  • rothschild86 says:

    what are the dots without labels?

  • mstickles says:

    Found it interesting that in the study, the Australian version of Kellogg’s Special K has an FII of 48, while the U.S. version has an FII of 86. And there’s not much difference in the macros. Be interesting to find out why such a huge difference.

  • Leo Tat says:

    Hey Marty

    Excellent information.

    One major premise of a low carb diet is that carbs increases insulin.

    So instead of looking at how many carbs foods contain, and draw a correlation, its even better to look simply at how much insulin foods raises.

    Your table list of foods is particularly useful to diabetics.

    p.s. low fat yogurt could be renamed high insulin yogurt.

  • Danni says:

    Thanks Marty,
    Do you have list the new Food Insulin Demand (FID) not the Food Insulin Index (FII) ? Which Do you think the most accurate to determine insulin shots? GI, GL, FII, FID, or Carb content?


  • […] I came across the insulin index data, it was Robb Wolf and Mat Lalonde’s thinking on nutrient density that made me believe there might […]

  • fadiaboona says:

    Do you have a youtube channel? What’s more important to maintain healthy insulin level? Insulinogenic or Insulin load?

  • Nereda Boundy says:

    Hi Marty, have been on a low carb, high fat diet now for 4 weeks, but am flying by the seat of my pants re insulin dosing. I’ve printed out the superfoods table, but have no idea how to translate this info into insulin requirements. Also,blackberries seem the only fruit on the list, unless I’m missing something. Can you clarify?

  • Danni says:

    Hai Marty, I have read Bell Thesis 1,page 214, how to count FID of the white bread of 26 (1 slice white bread has FII of 73 and contains 310 kj so 73 times 310/1000 is 22.63)? FID, Food Insulin Demand. (FID = FII x kJ in food portion /1000) scaled using the FID and carbohydrate content of 1000 kJ of glucose powder (100/59)).

  • tk says:

    How do substances that trigger insulin, but which don’t contain any calories, factor in? How do you calculate the insulin load of an artificial sweetener like Aspartame that has an insulin response? Is there a way?

    • stenbj says:

      From The book Brain Maker: David Perlmutter.

      “We used to think that sugar substitutes like saccharin, sucralose, and aspartame didn’t have a metabolic impact because they don’t raise insulin, it turns out that they can indeed wreak tremendous metabolic havoc and cause the same metabolic disorders as real sugar. How so? They do this by changing the microbiome in ways that favor dysbiosis, blood sugar imbalances, and an overall unhealthy metabolism. And yes, the food and beverage industry has a splitting headache over this latest study, which was published in 2014 in the journal Nature.
      Just forget Coke’s Zero. It is like drugs. Always side effects.

  • Eric says:

    The research out of Israel and Stanford on personalised diets and gut micro biome see eran Segal’s ted talk one published research should help one size does not fit all.

  • Isabella says:

    Okay, I’m sorry if I sound stupid but I’m not getting it. According to the graphs withe pasta has a lower insulin response than beef. But that goes against everything everybody is saying. What am I missing? Why does it matter how far to the right the foods are on the x-axis? Surley the insulin response (y-axis) is the more important factor, which would mean all the foods towards the bottom of the graph are the ones to dig into. I’m completely confused. Can somebody please explain? I’d really appreciate it.

  • […] The latest food insulin index data, and […]

  • Eugene Leong says:

    Thank you so much for sharing this post ! =D

    It was really helpful

    I notice that most insulin indexes were done based on the 1000 Kj standard, it may startle people to think for example Potatoes which has 121 insulin index value is so much worse than brown pasta which has 40 insulin index value

    However when we do compare the actual caloric value per 100g for potatoes and brown pasta , we get 77 calories for potato and 242 calories for brown pasta

    Hence if we calculate to estimate the amount we need to eat in order to achieve 100 insulin index, potatoes and brown pasta is actually about the same with Potatoes being slightly ahead of brown pasta

    256.52g of Potatoes needed to activate 100 insulin while
    246.90g of Brown pasta is needed to activate 100 insulin

    so technically that means we can eat more potatoes before we reach the same level of insulin as brown pasta

    In my opinion, using this food weight per 100 insulin interpretation may be a more practical way of comparing food quality types for different foods

    I have taken the liberty to extrapolate a table of the foods for the purpose of comparison here

    I also written more on longevity here today

    Any feedback please feel free let me know =D

    Thank you so much everyone !

    God bless

    • Marty Kendall says:

      There are certainly pros and cons of using cals vs grams, but in the end I’ve chosen calories as most people need to eat about the same amount of calories per day to maintain their body weight.

  • Dohn Joe says:

    In the second graph is Special K so special because it is both low insulinemic and high insulinemic at the same time much like a quantum particle? Data with brand names is dubious at best.

  • […] insulin index data suggests that higher fat foods have a lower insulin response.  But is this simply because the […]

  • […] first, why was dairy a problem? I was studying about the insulin index on Marty Kendall’s outstanding website, a site I read often. Since I was eating minimal carbs, what else was raising my insulin? I drank […]

  • Krissy says:

    Is FII and controlling insulin load good for reactive hypoglycemia? Mine is non-diabetic. I spike then crash but my A1C is actually low.

  • […] I found this article: by someone everyone interested in insulin should read regularly. Marty Kendall Of Optimizing […]

  • […] being irritable all the time.  I came across the Optimising Nutrition blog discussing an “insulin index” – much more useful to me than a “glycemic index” since my goal is shifting […]

  • […] Óptima e completa  explicação da aplicação do índice insulínico (ou Insulinêmico), do ponto de vista de alguém que usa para gestão de diabetes tipo 1 […]

  • I’m 73, female and following LCHF most of my adult life (since 1972). My age group is largely ignored in all the research being done. The Science of ketogenesis and it’s long-term effects on health and mobility in postmenopausal women could use a bit of interest. We’re pretty much left to swing in the wind when attempting to refine our macros. Today I utilized your Nutrient Optimizer for the first time in hopes that I can stabilize my intake and output.

  • […] the core concepts of nutrient density, satiety and insulin load, the Nutrient Optimiser enables you to pinpoint the optimal foods and meals that enable you to […]

  • […] density and satiety.  Participants whose primary goal was to reduce blood sugar used a lower insulin load meal plan. […]

  • […] with nutrient density and insulin load, satiety is a key component of the Nutrient […]

  • […] carb or keto) can be helpful to enable you to stabilise your blood sugars and insulin demand (see Harnessing the power of the food insulin index for more […]

  • >