Tag Archives: Mat Lalonde

building a better nutrient density index

  • Nutrient dense foods can increase satiety by providing adequate nutrition and reduced cravings with less energy.
  • Some approaches to nutrient density focus on vitamins and minerals while others use a broader range of nutrients that include essential amino acids and essential fatty acids.
  • This article outlines a new system for prioritisation of foods that focuses on essential nutrients that are more difficult to obtain.

why nutrient density matters

Dr Joel Fuhrman has done some great work developing and testing his dietary approach based on high nutrient density foods.[1]


Fuhrman’s research suggests that a high nutrient density approach (HND) to food selection leads to a range of benefits including improved:

  • blood sugar control,
  • weight loss,
  • blood pressure, and
  • blood markers.[2]


People following a high nutrient density approach tend to feel more satiated with fewer calories and are able to skip meals more easily.[3]


Harvard researcher Dr Christopher Gardner has also shown the benefits of a high fibre, nutrient dense dietary approach with his recent paper Weight loss on low-fat vs. low-carbohydrate diets by insulin resistance status among overweight adults and adults with obesity: A randomized pilot trial.[4]


In this study all participants were encouraged to eat nutrient dense, higher fibre, unprocessed foods.  While the participants who were insulin resistant benefited more from a low carbohydrate approach and insulin sensitive people benefited more from a low energy density / low fat approach, everyone lost weight and improved their blood markers without having to consciously count calories!

calories or nutrients?

It’s generally accepted that people will lose weight if they consume less calories, however the real challenge is managing appetite in the long term.

“Appetite is a dragon.  Losing weight is brutally tough.  Harder than particle physics.”

says RD Dikeman (pictured) who has made some great progress via tight blood glucose control (using the process outlined in the article how to use your blood glucose meter as a fuel gauge), avoidance of processed carbs and intermittent fasting.[5]


Most people find that appetite and metabolism win out over willpower or conscious calorie counting in the long run.  Either we end up binging on the foods we were craving or our metabolism slows down to cope with the reduced energy intake.[6]

But what if satiety is influenced by the quantity of nutrients rather than the calories in our food?  Paul Jaminet in his Perfect Health Diet books says:

“A nourishing, balanced diet that provides all the required nutrients in the right proportions is the key to minimising appetite and eliminating hunger at minimal caloric intake.”

But how do we know if we are getting the required nutrients in the right proportions?   Which foods will help us maximise our chance of achieving nutrient density while minimising energy?

The chart below (click for a larger image) shows the percentage of the recommended daily reference intake (DRI)[7] of the various nutrients that you would obtain if you ate a little bit of ALL of the 7000+ foods in the USDA foods database.

  • Without following any particular dietary approach it seems from this that it’s fairly easy to obtain the recommended amounts of most of the amino acids, iron, phosphorus, selenium, niacin and Vitamin B-12.
  • However, without paying attention to the nutrient density of your diet or supplementation you will have to consume well beyond 2000 calories to obtain the recommended daily intake of calcium, magnesium, potassium, copper, vitamin E, vitamin D, pantothenic acid, choline and the essential fatty acids EPA and DHA.


Satiety is a complex and controversial topic.  There are many factors including, palatability, mouth feel, smell, protein, fibre, mood, insulin resistance etc etc etc.   Obtaining adequate nutrients may not be the only thing that influences appetite, but it just might be a significant piece of the complex puzzle.  As you will see below, nutrient dense foods are typically also unprocessed whole foods that you would be less likely to binge on than a packet of Pringles, pizza and a bottle of coke (i.e. ‘foods with no brakes‘).

The slide below from a presentation by Bruce Ames demonstrates that there might be some room for improvement in the nutrient density of most people’s diets.[8]


This slide shows how many people are lacking in a range of key micro-nutrients.  Very few people are getting adequate omega-3 essential fatty acids.


limitations of daily reference intake values

The daily recommended intake (DRI)[9] values are typically conservative.  You may do fine with much lower levels than the recommended intake levels.  The only way to really know if you are lacking in a particular nutrient would be to get blood tests to see if you are deficient in any nutrients.[10]

In lieu of regular blood testing of all the essential nutrients you can use the DRI values as a guide to understand if you are getting a ‘balanced diet’ with adequate amounts of the essential nutrients.  Some people use apps like cronometre to see if they are meeting their minimum levels of various nutrients, but how do you know which foods will give you the best chance of maximising your nutrition?

There are meal replacement shakes (e.g. Soylent, Optifast, Ambronite etc) that enable you to theoretically meet the DRI values with a minimum amount of calories.  However the safest approach is probably going to be to focus on nutrient dense unprocessed foods that contain all the essential nutrients that we know about as well as the other nutrients that we don’t yet know about.

Your metabolism may not have read the World Health Organisation’s research on the daily reference intake of the various nutrients however, if appetite is at least partially driven by obtaining adequate nutrition you can see why we are less likely to binge on nutrient dense whole foods.

The slide below from Bruce Ames shows the commonly accepted essential vitamins, minerals, amino acids and fatty acids that we require.


how to calculate nutrient density

Calculation of nutrient density is far from a precise science.  Different people have taken different approaches and arrived at different food rankings.

Joel Fuhrman’s take on nutrient density uses vitamins and minerals with no consideration of amino acids or fatty acids.   Fuhrman’s ANDI index also includes phytosterols, glucosinolates, angiogenesis inhibitors, organosulfides, aromatase inhibitors, resistant starch, resveratrol and Oxygen Radical Absorbance Capacity (ORAC) in the scoring.  These additional parameters are not available in the USDA food database and are not part of the generally accepted list of essential nutrients, so it’s hard to include them in a comprehensive analysis.[11]   The highest scoring foods with or without these additional parameters are similar (i.e. green leafy veggies) so I don’t think omitting these parameters will materially change the overall outcome.

Based on his analysis Fuhrman recommends a diet high in vegetables and fruit with a minimum of animal products and processed carbohydrates.  Fuhrman recommends eating animal products only occasionally, ideally fish to provide omega 3 fatty acids.  It’s not hard to see how restricting yourself to non-starchy veggies would help you to reduce your energy intake.

More recently Dr Mat Lalonde developed an alternative approach to analysing nutrient density which also includes essential amino acids and essential fatty acids.  With the inclusion of animal products this approach tends to prioritise high protein animal based foods.  Lalonde’s approach is based on nutrients per weight of food which may be useful for an athlete wanting to quickly refuel, however Fuhrman’s nutrietns per calorie may be more useful for someone wanting to lose weight.

The low carb community’s criticism of Fuhrman’s approach is that it is too high in carbohydrates and that it is unnecessarily biased towards plant based foods.   Meanwhile the vegan community’s criticism of Lalonde’s approach is that the higher protein and fat levels are unnecessary and even dangerous.[12]  They claim you can get adequate amounts without going out of your way to make it a priority.

As detailed in the optimal foods for different goals I previously had a go at developing a nutrient density ranking system that includes forty three (43) beneficial nutrients including vitamins and minerals as well as beneficial amino acids and fatty acids.   While this ‘belt and braces’ approach to nutrient density will ensure that you maximise the nutrient density of your food there is also a risk that it will prioritise nutrients that are easy to obtain at the expense of nutrients that are less common in our food system.

So which approach is optimal?  Vitamins and minerals only, all beneficial nutrients, or perhaps something else?  Which approach will enable you to obtain a nourishing, balanced diet that provides all the required nutrients in the right proportions to minimise appetite and eliminate hunger with a minimal caloric intake.

comparison of approaches to nutrient density

The chart below (click for a larger image) compares the nutrients we obtain for the following approaches:

  • all foods,
  • top 500 foods prioritised using vitamins and minerals, and
  • top 500 foods prioritised using all 43 beneficial micro-nutrients.


We can see from this analysis that:

  • Following either approach to maximising nutrient density provides an immense improvement compared to the average of all of the foods in the USDA database.
  • The vitamins and minerals only approach does better in terms of most of the vitamins and minerals.
  • The most nutrient dense approach using forty three micro-nutrients does better when it comes to amino acids (protein), essential fatty acids (DHA and EPA), vitamin B-12, zinc, selenium and niacin.
  • There is a lot of variability in the amounts of nutrients in terms of percentage of the DRI.

So if our goal is to avoid malnutrition with the minimum amount of calories, which approach is optimal?

Perhaps what we need, rather than amplifying all nutrients, is to prioritise the foods with the nutrients that are harder to obtain?



removing the overachievers

The chart below shows the proportion of the population that consume less than the recommended amount of various essential nutrients.  From this it seems we should, as a minimum, prioritise vitamin D, vitamin E, magnesium, calcium, vitamin A and vitamin C.


Starting with the full list of forty-three beneficial nutrients I have progressively removed the ‘overachievers’ so we only prioritise the harder to obtain nutrients.  The nutrients that you could obtain more than 500% of the daily recommended intake (DRI) with 2000 calories have been removed from the system.

I have also removed the fatty acids that could be considered contentious in a minimalist food ranking system.  So rather than 43 nutrients we end up prioritising only the 27 hardest to obtain essential nutrients.


  1. Choline
  2. Thiamine
  3. Riboflavin
  4. Niacin
  5. Pantothenic acid
  6. Vitamin A
  7. Vitamin B12
  8. Vitamin B6
  9. Vitamin C
  10. Vitamin D
  11. Vitamin E
  12. Vitamin K


  1. Calcium
  2. Copper
  3. Iron
  4. Magnesium
  5. Manganese
  6. Phosphorus
  7. Potassium
  8. Selenium
  9. Sodium
  10. Zinc

amino acids

  1. Cysteine
  2. Isoleucine
  3. Leucine
  4. Lysine
  5. Phenylalanine
  6. Threonine
  7. Tryptophan
  8. Tyrosine
  9. Valine
  10. Methionine
  11. Histidine

fatty acids

  1. Docosahexaenoic acid (DHA) (22:6 n-3)
  2. Eicosapentaenoic acid (EPA) (20:5 n-3)
  3. Docosapentaenoic acid (DPA) (22:5 n-3)
  4. Alpha-linolenic acid (18:3 n-3)
  5. Arachidonic acid (20:4)
  6. Oleic acid (18:1)
  7. Lauric acid (12:0)
  8. Capric acid (10:0)
  9. Pentadecanoic acid (15:0)
  10. Margaric acid (17:0)

The chart below (click for larger image) shows the outcome of the moderated approach compared to the other approaches (i.e. all foods, vitamins and minerals only and all 43 nutrients).  A number of the nutrients that were lower using the “all nutrients” approach have improved (i.e. calcium, magnesium, vitamin A, vitamin C, selenium, vitamin E and vitamin D).


which one is best?

The moderated approach does pretty well across the board.  The problem  is that it’s hard to make sense of all this data to confirm which approach is optimal.  How do we simplify the decision process?

In engineering we often think in terms of reliability statistics.[13]

Let’s say Acme brand widget is really strong on average but highly variable.  If you buy a box of Acme widgets most of them will be strong, but you might get some low strength duds.  Acme of widget not reliable so we have to be conservative when it comes to the design assumptions.   In the design we might assume that a widget is only as strong as the average minus one or two standard deviations of the strength to make sure our design is conservative.


However if we can decrease the variability by improving the manufacturing process and produce a box of widgets that are not quite as strong on average but less variable we can assume a lower factor of safety and assume more capacity in a design using that bolt.

Perhaps we can use a similar analysis approach when it comes to nutrient density.  What we ideally want is a diet that has high levels of all of the essential nutrients without any nutrient deficiencies that would require supplementation.

The chart below plots the average of all the nutrients as a proportion of the DRI (blue bars).  We can see that all three approaches to ranking nutrient density do better than the average of all foods in the USDA database, with the “43 micro-nutrients” approach scoring the best.  However we know from the chart above that this high score is largely due to very high amino acid scores for the “all 43 micro-nutrients” approach.

The vitamins and minerals only approach also does well, however we also know that this is due to the higher score in the vitamins and minerals with lower scores in some of the other nutrients such as the proteins and essential fatty acids.


The orange bars in the plot represent the average minus 0.8 times the standard deviation of the nutrients as a percentage of the DRI requirement.  Using this approach to comparison it appears that the moderated nutrient density approach is better because we have less variability across the nutrients, with some lower highs and lots of higher lows compared to the other approaches.

What this means in practice is that the moderated approach will more reliably provide you with the essential vitamins, minerals, amino acids and fatty acids that you require without needing to supplement or overeat to provide the missing nutrients.   The moderated nutrient density approach seems to give us a better outcome in terms of nutrient density.

most nutrient dense foods

Listed below is a summary of the top 1000 foods prioritised by the moderated nutrient density system detailed above.

In addition to nutrient density score (note: 0 is average and a score of 2 means that a food is two standard deviations above the mean) I have also included a number of other parameters that may be of interest.

  • The percentage of insulinogenic calories and net carbs per 100g of food will be of interest for someone who aiming for a high fat therapeutic ketogenic diet.
  • The insulin load may be of interest for someone who is insulin resistant and wanting to consume a diet that their pancreas can keep up with.
  • Net carbs will be useful for someone doing standard carbohydrate counting.
  • The energy density (calories per 100g) will be of interest for someone looking to decrease the energy density of their diet for weight loss.

I have also shown the vitamin, mineral and protein plots for some of the highest ranking foods in each category to get a feel for the nutrition provided by each of these foods.

Choosing nutrient dense whole foods typically ensures that the other relevant parameters are favourable, though these other factors may be of interest depending on your situation.

Future articles will look at how we can fine tune our food selection to suit people who are insulin resistant and wanting to normalise their blood glucose levels or who are insulin sensitive and still looking to lose weight.  In the meantime you can check out these summary food lists that are based around these ideas:


If you look down the nutrient density (ND) scores of all the foods you will see that the vegetables do really well compared to the other food groups.  If you were aiming to maximise nutrient density you could simply focus on eating as many vegetables as you could with perhaps some supplemental seafood for essential fatty acids. image09

Celery tops the list of nutrient dense dense foods because it has a lot of vitamins and minerals with very few calories.  The chart below from Nutrition Data shows that we would obtain 81% of our required vitamins and minerals from 1000 calories and 52% of the protein.   The is that we would need to eat 100 celery stalks to obtain that 1000 calories!  However you can see how in terms of nutrients per calorie celery is amazing and you wouldn’t go wrong trying to fill up on these high nutrient density low calorie density foods.


food ND % insulinogenic net carbs/100g insulin load  (g/100g) calories/100g
celery 2.63 49% 1 2 17
rhubarb 1.46 57% 3 3 21
turnip greens 1.31 39% 1 4 37
lettuce 1.34 52% 2 2 17
winter squash 1.22 80% 7 8 39
broccoli 1.21 57% 4 6 42
asparagus 1.12 46% 2 3 27
Chinese cabbage 1.02 60% 1 2 16
summer squash 1.00 65% 2 3 19
okra 0.94 57% 4 5 37
bamboo shoots 0.90 52% 3 4 28
bell peppers 0.86 64% 6 7 43
artichokes 0.83 33% 3 4 54
cabbage 0.81 53% 3 4 30
kale 0.75 74% 8 10 56
parsnip 0.73 38% 7 7 76
seaweed (kelp) 0.74 43% 4 5 50
snap green beans 0.74 47% 4 5 40
peas 0.69 58% 5 7 51
radishes 0.70 50% 2 2 19
mushrooms 0.65 70% 2 5 30
sweet potato 0.51 82% 17 18 87
onions 0.52 77% 7 8 41
jalapeno peppers 0.52 54% 4 5 35
pinto beans 0.44 60% 16 21 142
sweet corn 0.43 47% 10 13 111
collards 0.44 46% 2 5 40
dill 0.42 30% 2 4 52
eggplant 0.39 67% 7 7 41
beets 0.34 44% 4 5 48
shallots 0.27 60% 46 56 377
mung beans 0.33 46% 1 3 26
thyme 0.27 21% 14 19 359
black pepper 0.24 36% 24 29 327
bay leaf 0.21 37% 34 38 406
chives 0.27 34% 1 3 37
mustard greens 0.27 45% 2 3 30
Brussels sprouts 0.24 54% 5 7 52
shiitake mushrooms 0.20 68% 51 59 349
paprika 0.19 17% 8 16 389


The list of nutrient dense fruits is shorter than the vegetables due to the higher amount of calories and sugar in proportion to the amount of nutrients.

The plot below shows that we get 57% of the vitamins and minerals and 44% of our protein from 1000 calories of mandarin orange.

Nutrition Facts and Analysis for Oranges, raw, all commercial varieties - Google Chrome 16052016 54708 AM.bmp

food ND % insulinogenic net carbs/100g insulin load (g/100g) calories/100g
cherries 0.72 84% 10 11 54
orange 0.49 77% 10 11 55
apples 0.48 77% 10 10 53
grapes 0.45 80% 15 15 77
figs 0.37 81% 16 17 82
blueberries 0.32 72% 16 16 91
mandarin oranges 0.31 63% 9 9 59
honeydew melon 0.30 88% 8 9 40
passion fruit 0.24 54% 13 15 109
raisins 0.20 84% 68 70 336
litchis 0.20 80% 14 15 73
dates 0.17 72% 54 56 308


Legumes tend to have a higher energy density than the vegetables and thus may be useful if you need some more calories to support your activity and can’t fit in any more celery, lettuce and broccoli.


The Nutrition Data plot below for lentils shows that 1000 calories will provide 58% of your vitamins and minerals and 86% of your protein.



food ND % insulinogenic net carbs/100g insulin load  (g/100g) calories/100g
lima beans 0.56 71% 16 23 129
navy beans 0.47 55% 15 20 143
lentils 0.35 62% 12 18 118
hummus 0.26 32% 8 14 175
peanuts 0.17 18% 7 28 605


The nutrient dense grains tend to be the least processed.  Unfortunately most grains are consumed in a highly processed form.


The plot below shows that oats will give us minerals and a substantial amount of protein, but are not as high in the vitamins compared with a number of the other foods. image11

food ND % insulinogenic net carbs/100g insulin load  (g/100g) calories/100g
oatmeal 0.77 58% 8 10 67
teff 0.70 54% 11 14 101
spelt 0.58 54% 14 18 135
rice noodles 0.54 87% 22 23 105
quinoa 0.45 55% 14 16 120
oat bran 0.35 57% 29 38 264
millet 0.34 76% 20 22 118
rye bread 0.30 64% 37 45 282
rice bran bread 0.25 54% 31 37 273
wheat bran bread 0.24 68% 37 44 257
oat bran muffins 0.23 48% 29 35 288

dairy and eggs

The nutrient density score for eggs and dairy is not as high as the vegetables, however the proportion of insulinogenic calories and net carbohydrates is lower which will mean that these foods have a minimal impact on blood glucose levels.


The plot below shows that we would get half of our required vitamins and minerals and 136% of our protein requirements from 1000 calories of eggs (i.e. 14 eggs).


food ND % insulinogenic net carbs/100g insulin load  (g/100g) calories/100g
parmesan cheese 0.18 30% 3 31 411
goat cheese 0.17 22% 2 25 451
edam cheese 0.17 22% 1 20 356
gruyere cheese 0.17 21% 0 22 412
Swiss cheese 0.17 26% 5 25 379
egg yolk 0.17 19% 4 15 317
gouda cheese 0.17 23% 2 20 356
provolone 0.17 24% 2 21 350
blue cheese 0.16 20% 2 18 354
cheddar cheese 0.15 20% 1 20 403
limburger cheese 0.16 18% 0 15 327
camembert cheese 0.16 20% 0 15 299
Monterey 0.15 20% 1 19 373
muenster cheese 0.15 20% 1 18 368
Colby 0.15 20% 3 20 394
whole egg 0.16 29% 1 10 138

nuts and seeds

Nuts and seeds are more energy dense but lower in carbohydrates due to their higher fat content.  While nuts and seeds will help someone achieve more stable blood glucose levels it is common knowledge in low carb circles that you need to watch your intake of nuts, seeds and dairy if you’re trying to lose weight.


food ND % insulinogenic net carbs/100g insulin load  (g/100g) calories/100g
coconut water 1.51 66% 3 3 20
sunflower seeds 0.18 20% 11 24 491
tahini 0.17 16% 13 26 633
pine nuts 0.16 11% 9 18 647
pecans 0.15 5% 4 9 762
pistachio nuts 0.16 23% 19 34 602


Omega 3 fatty acids are important but hard to get in the diet, so it’s worth going out of your way to ensure you are getting enough.


The plot below shows that we can get more than half of our vitamins and minerals and 148% of our protein requirements from 1000 calories of sardines.


food ND % insulinogenic insulin load  (g/100g) calories/100g
anchovy 0.34 42% 21 203
caviar 0.30 32% 22 276
tuna 0.30 50% 17 137
oyster 0.31 57% 14 98
rainbow trout 0.28 43% 17 162
mackerel 0.28 45% 17 149
swordfish 0.28 41% 17 165
lobster 0.30 69% 14 84
herring 0.26 34% 18 210
salmon 0.28 50% 15 122
whitefish 0.27 67% 17 102
octopus 0.26 69% 27 156
halibut 0.27 63% 16 105
Pollock 0.27 66% 17 105
sturgeon 0.26 47% 15 129
sardine 0.24 36% 18 202
shrimp 0.26 66% 19 113
crab 0.26 69% 13 78
snapper 0.25 64% 15 94
haddock 0.24 67% 18 110
mussel 0.22 61% 25 165
whiting 0.21 63% 17 109
crayfish 0.21 64% 12 78
abalone 0.21 76% 19 99
haddock 0.21 69% 15 85
clam 0.20 71% 24 135

animal products

When it comes to animal products the lower fat cuts tend to rank higher when it comes to nutrient density.


Liver ranks the highest overall and the vitamin and minerals score as well as the protein score is substantial.


food ND % insulinogenic insulin load  (g/100g) calories/100g
beef liver 0.46 58% 24 169
chicken liver 0.43 48% 20 165
ham 0.26 55% 20 146
pork 0.25 54% 21 154
veal (leg) 0.25 56% 25 174
emu 0.24 63% 25 159
beef 0.22 50% 25 197
chicken breast 0.22 56% 25 178
turkey breast 0.22 70% 22 127
bacon 0.18 23% 30 522
ground turkey 0.19 37% 19 203
ostrich 0.19 46% 19 168
veal (sirloin) 0.18 38% 19 195
pork 0.18 46% 21 182
chicken drumstick 0.17 36% 22 238
goose 0.17 37% 21 230
duck (meat only) 0.17 36% 17 195
beef steak 0.16 28% 21 305

should everybody eat just these nutrient dense foods?

As a general rule most people would do well eating from this list of nutrient dense whole foods.  Unprocessed nutrient dense foods would be a major improvement for most people.  There is however opportunity to further refine this for specific goals such as weight loss or diabetes.

In future articles we will look at how we can use the concepts of energy density and insulin load to further refine this list for people who are looking to lose weight and for people who have diabetes and need to control their blood glucose levels.  In the mean time you may be interested in these summary food lists:



[1] http://www.wholefoodsmarket.com/healthy-eating/andi-guide

[2] https://www.drfuhrman.com/members/m_library/OJPM20120300014_73341742.pdf

[3] https://www.drfuhrman.com/library/changing_perceptions_of_hunger.pdf

[4] http://onlinelibrary.wiley.com/doi/10.1002/oby.21331/full

[5] RD also happens to be a physicist and a chief scientist with defence contractor Lockheed Martin.  He is also an admin on the TYPE ONE GRIT facebook group for people with type 1 diabetes (his son has type 1 diabetes) and produces Dr Bernstein’s Diabetes University.

[6] http://www.amazon.com/Obesity-Code-Unlocking-Secrets-Weight/dp/1771641258

[7] https://ods.od.nih.gov/Health_Information/Dietary_Reference_Intakes.aspx

[8] https://www.youtube.com/watch?v=ZVQmPVBjubw

[9] http://www.mydailyintake.net/nutrients/

[10] http://www.lifeextension.com/vitamins-supplements/blood-tests/nutrient-testing

[11] http://www.wholefoodsmarket.com/healthy-eating/andi-guide

[12] http://www.vegsource.com/articles/protein.danger.htm

[13] https://en.wikipedia.org/wiki/Reliability_(statistics)

nutrient dense superfoods for maintenance

These foods will help you maximise nutrient density and sustain health with adequate energy for weight maintenance.

“A nourishing, balanced diet that provides all the required nutrients in the right proportions is the key to minimising appetite and eliminating hunger at minimal caloric intake.”

Paul Jaminet

2016-07-06 (11)

The downside of the most nutrient dense foods is that they have such a low energy density that it can be hard to maintain your body weight.   The chart below shows that these foods still rank at #4 of the 13 approaches analysed in terms of nutrient density despite containing a little more fat!


These foods still have a fairly low energy density (#4 of 13), but not as much as the more aggressive weight loss approach.  The addition of some nuts and dairy brings means it will be easier to maintain body weight and satiety, especially if you are active.


The chart below shows that the nutrients provided by these foods is well above the average of all the foods in the USDA database.

2017-02-27 (9).png

The tables below show the

vegetables and spices


food ND % insulinogenic insulin load (g/100g) calories/100g MCA
endive 14 23% 1 17 2.7
coriander 14 30% 2 23 2.6
chicory greens 13 23% 2 23 2.5
asparagus 15 50% 3 22 2.4
brown mushrooms 17 73% 5 22 2.3
escarole 11 24% 1 19 2.2
spinach 14 49% 4 23 2.2
watercress 16 65% 2 11 2.2
alfalfa 10 19% 1 23 2.2
basil 13 47% 3 23 2.1
beet greens 11 35% 2 22 2.0
portabella mushrooms 13 55% 5 29 2.0
zucchini 11 40% 2 17 2.0
arugula 11 45% 3 25 1.9
chard 12 51% 3 19 1.9
white mushroom 14 65% 5 22 1.9
mustard greens 10 36% 3 27 1.9
parsley 11 48% 5 36 1.8
lettuce 11 50% 2 15 1.8
paprika 8 27% 26 282 1.8
shiitake mushroom 12 58% 7 39 1.8
Chinese cabbage 11 54% 2 12 1.7
okra 11 50% 3 22 1.7
dandelion greens 11 54% 7 45 1.7
broccoli 10 50% 5 35 1.6
curry powder 6 13% 14 325 1.6
summer squash 9 45% 2 19 1.6
yeast extract spread 11 59% 27 185 1.6
banana pepper 8 36% 3 27 1.6
pickles 8 39% 1 12 1.5
cucumber 8 39% 1 12 1.5
celery 9 50% 3 18 1.5
collards 7 37% 4 33 1.4
chives 8 48% 4 30 1.4
cauliflower 8 50% 4 25 1.4
turnip greens 7 44% 4 29 1.3
sage 5 26% 26 315 1.3
cloves 6 35% 35 274 1.3
artichokes 7 49% 7 47 1.2
thyme 5 34% 31 276 1.2
eggplant 5 35% 3 25 1.2
mustard seed 4 27% 37 508 1.2
seaweed (wakame) 10 79% 11 45 1.2
jalapeno peppers 5 37% 3 27 1.1
sauerkraut 5 39% 2 19 1.1
cabbage 7 55% 4 23 1.1
marjoram 4 31% 27 271 1.1
radishes 5 43% 2 16 1.1
chayote 5 40% 3 24 1.1
bamboo shoots 7 60% 5 27 1.1
radicchio 8 67% 4 23 1.1
red peppers 5 40% 3 31 1.1
dill 7 59% 8 43 1.1
edamame 5 41% 13 121 1.0
seaweed (kelp) 9 77% 10 43 1.0
poppy seeds 2 17% 23 525 1.0
caraway seed 3 27% 28 333 1.0
turnips 6 51% 3 21 1.0
snap beans 6 58% 3 15 0.9



food ND % insulinogenic insulin load (g/100g) calories/100g MCA
fish roe 11 47% 18 143 1.9
crab 13 71% 14 83 1.8
salmon 11 52% 20 156 1.7
caviar 9 33% 23 264 1.7
oyster 12 59% 14 102 1.7
trout 9 45% 18 168 1.6
anchovy 9 44% 22 210 1.6
lobster 12 71% 15 89 1.6
halibut 12 66% 17 111 1.6
crayfish 11 67% 13 82 1.5
sturgeon 9 49% 16 135 1.5
mackerel 4 14% 10 305 1.4
sardine 7 37% 19 208 1.4
cod 11 71% 48 290 1.4
flounder 9 57% 12 86 1.3
pollock 10 69% 18 111 1.3
rockfish 9 66% 17 109 1.2
herring 5 36% 19 217 1.2
perch 8 62% 14 96 1.2
shrimp 9 69% 19 119 1.1
haddock 9 71% 19 116 1.1
cisco 4 29% 13 177 1.1
whiting 7 66% 18 116 1.0
octopus 8 71% 28 164 0.9
tuna 6 52% 23 184 0.9

animal products


food ND % insulinogenic insulin load (g/100g) calories/100g MCA
lamb kidney 11 52% 15 112 1.8
lamb liver 10 48% 20 168 1.7
turkey liver 10 47% 21 189 1.7
chicken liver 9 50% 20 172 1.6
beef brains 6 22% 8 151 1.5
ham (lean only) 10 59% 17 113 1.5
veal liver 8 55% 26 192 1.3
beef kidney 8 52% 20 157 1.3
beef liver 9 59% 25 175 1.3
chicken breast 8 60% 22 148 1.1
turkey heart 6 47% 20 174 1.1
lamb brains 4 27% 10 154 1.1
turkey ham 6 45% 14 124 1.1
pork chop 7 57% 23 172 1.1
pork liver 7 59% 23 165 1.1
beef heart 5 41% 16 165 1.0
chicken liver pate 4 34% 17 201 1.0
ground turkey 3 30% 19 258 1.0
roast pork 5 41% 20 199 1.0
turkey drumstick 6 52% 21 158 1.0
turkey meat 6 52% 21 158 1.0
turkey (skinless) 4 40% 16 170 1.0
ground pork 6 54% 25 185 0.9
pork shoulder 6 56% 22 162 0.9
pork loin 4 41% 19 193 0.9
pork ribs 4 39% 21 216 0.9
roast ham 4 41% 18 178 0.9
ham 2 29% 11 149 0.9
leg ham 5 56% 22 165 0.8
lamb sweetbread 4 43% 15 144 0.8
lamb heart 4 48% 19 161 0.8
lean beef 6 61% 23 149 0.8
pork 4 44% 22 209 0.8
chicken (leg with skin) 3 42% 18 184 0.8

dairy and egg


food ND % insulinogenic insulin load (g/100g) calories/100g MCA
egg yolk 6 18% 12 275 1.6
whole egg 6 30% 10 143 1.4
whey powder 10 95% 82 339 0.8
kefir 6 64% 7 41 0.7
cheddar cheese -2 20% 20 410 0.4
feta cheese -2 22% 15 264 0.4
Swiss cheese -2 22% 22 393 0.4
mozzarella -1 34% 26 304 0.4
gruyere cheese -2 22% 23 413 0.4
limburger cheese -2 19% 15 327 0.4
blue cheese -2 21% 19 353 0.4
sour cream -3 13% 6 198 0.4
camembert -2 21% 16 300 0.4
cream -4 6% 5 340 0.3

legumes, nuts and seeds


food ND % insulinogenic insulin load (g/100g) calories/100g MCA
sunflower seeds 4 15% 22 546 1.3
brazil nuts 1 9% 16 659 1.0
flax seed 1 11% 16 534 1.0
pumpkin seeds 1 19% 29 559 0.9
almond butter -0 16% 26 614 0.7
almonds -0 15% 25 607 0.7
walnuts -1 13% 22 619 0.7
sesame seeds -1 10% 17 631 0.7

other dietary approaches

The table below contains links to separate blog posts and printable .pdfs detailing optimal foods for a range of dietary approaches (sorted from most to least nutrient dense) that may be of interest depending on your situation and goals.   You can print them out to stick to your fridge or take on your next shopping expedition for some inspiration.

dietary approach printable .pdf
weight loss (insulin sensitive) download
autoimmune (nutrient dense) download
alkaline foods download
nutrient dense bulking download
nutrient dense (maintenance) download
weight loss (insulin resistant) download
autoimmune (diabetes friendly) download
zero carb download
diabetes and nutritional ketosis download
vegan (nutrient dense) download
vegan (diabetic friendly) download
therapeutic ketosis download
avoid download

If you’re not sure which approach is right for you and whether you are insulin resistant, this survey may help identify the optimal dietary approach for you.


post last updated April 2017

optimising micronutrients and macronutrients for different goals

  • This article looks at the macro and micronutrient split for the most nutrient dense foods using a multi criteria analysis tailored for different goals (i.e. diabetes, therapeutic ketosis, weight loss and athletic performance).
  • High protein foods are typically much more nutrient dense than high carbohydrate foods.
  • The optimal protein intake for athletes, people who are metabolically healthy and people who are trying to lose weight is virtually unrestricted, particularly as the blood glucose impact from gluconeogenesis is not a concern and the amount of protein someone can eat is self-limiting.
  • People with diabetes and metabolic syndrome may benefit from a more moderate protein intake in order to reduce their insulin load to the point that that their pancreas can keep up and maintain normal blood glucose levels.

the setup

In a number of previous articles I have discussed minimum glucose and protein requirements based on starvation experiments.  Thinking that these are recommendations for protein a number of people responded noting that 0.8g/kg lean body mass (LBM) or 8% of calories is too low. [1]

My typical response to these comments has been along the lines of:

  • We need to get adequate amino acids from protein, just like we need adequate vitamins and minerals.
  • It’s all about finding the balance between adequate nutrition while keeping your dietary glucose load low enough so your pancreas can function efficiently.
  • The optimal approach is going to be different for each individual, depending on your needs, activity levels, personal situation and goals.

Microsoft Word Document 23082015 20658 PM.bmp

But what does this look like in practice?  What is the optimal macronutrient ratio that balances the competing goals of maximising nutrition while not overloading our body’s capacity to process those nutrients?  How do we get adequate protein for growth, repair and mental function while avoiding high blood glucose and excess body fat?

How do you find a way through all the contradictory health advice from different health camps saying ‘you can’t eat too much protein’, ‘maximise your veggies’, ‘eat fat to lose fat’, or others saying ‘intermittent fasting is the rediscovered cure for everything’?

different strokes for different folks

Context matters when it comes to designing the optimal diet.

In the article Your Personal Food Ranking System I outlined a multi criteria system to help prioritise food choices for different situations.  The system uses different weightings for a range of parameters available from the USDA food database.  I then used these parameters and weightings to develop lists of optimal foods for different situations (i.e. weight loss, diabetes / nutritional ketosis, therapeutic ketosis, and athlete / metabolically healthy) highlighting foods from the top 25% of the foods ranked using those weightings.

In developing these lists of optimal foods I also included cost considerations as a factor.  However in the discussion of optimal nutrition I don’t think cost is as important.  I also wanted to use the full USDA database of nearly 8000 foods rather than the smaller database of around 1000 foods that I have cost data for.

So we are left with the following parameters from which to develop a food ranking system:

  • nutrient density / calorie,
  • nutrient density / weight,
  • fibre / calorie,
  • fibre / weight,
  • percentage of insulinogenic calories, and
  • calories / weight.


Fibre can be useful to help feed and increase the diversity of our gut bacteria which appears to be good for a range of benefits including insulin sensitivity and autoimmunity.

The amount of fibre per calorie also seems to be a good proxy for the degree of processing involved in carbohydrate containing foods.

  • Highly processed foods = bad.
  • Minimally processed foods = good.

calorie density

Regardless of calories, it seems we eat about the same weight of food per day. [2]

Calorie density is related to fibre content.  Decreasing calorie density is a useful way to spontaneously manage energy intake.

A low calorie density typically means that your food contains a lot of water, so by itself it isn’t particularly useful;  a fortified energy drink or a slice of watermelon will end up scoring highly when we only consider nutrient density per calorie.

Considering both calorie density and fibre together ends up being a much more useful approach.

proportion of insulinogenic calories

If you are insulin resistant or obese then managing the insulin load of your diet will be an important consideration.  The percentage of insulinogenic calories concept builds on the food insulin index testing which demonstrates that both carbohydrates and protein require insulin. [3]

Keeping your percentage of insulinogenic calories low (as defined by the formula below) means reducing net carbohydrates, increasing fibre and possibly moderating protein.


This formula biases against non-fibre carbohydrates.  As noted earlier, one of the major concerns people have with this approach to identifying optimal foods is that we still need our protein.  So the question is, “to what extent would protein be penalised if we also try to maximise our nutrients?”

nutrient density

The nutrient density concept builds on the excellent work of Mat Lalonde which he discusses in this video from AHS 2012.

Lalonde shortlisted the following essential nutrients [4] which have been used in the nutritional analysis as part of the multi criteria analysis:

Essential fatty acids [5]

  1. alphalinolenic acid (omega-3) (18:3)
  2. docosahexaenoic acid (omega-3) (22:6)

Amino acids

  1. cysteine
  2. isoleucine
  3. leucine
  4. lysine
  5. phenylalanine
  6. threonine
  7. tryptophan
  8. tyrosine
  9. valine
  10. methionine
  11. histidine


  1. choline
  2. thiamine
  3. riboflavin
  4. niacin
  5. pantothenic acid
  6. vitamin A
  7. vitamin B12
  8. vitamin B6
  9. vitamin C
  10. vitamin D
  11. vitamin E
  12. vitamin K


  1. calcium
  2. copper
  3. iron
  4. magnesium
  5. manganese
  6. phosphorus
  7. potassium
  8. selenium
  9. sodium
  10. zinc

It is worth noting that the list of essential amino acids, vitamins and minerals has about equal number of elements (i.e. eleven amino acids, twelve vitamins and ten minerals).  Hence it is unlikely that we would end up biasing towards high protein foods just because we have more amino acids being counted than say vitamins or minerals.

The nutrient density for a certain food is based on a relative score calculated by comparing the amount of a particular nutrient in each food with the all of the foods in the database.

For example, if a food has an average amount of vitamin C compared to the 8000 other foods in the database it will get a score of zero because it is zero standard deviations from the mean.  If it has a large amount of a certain nutrient then it will receive a high score.  If it is two standard deviations from the mean then it gets a score of two for that nutrient.  If however it is five standard deviations from the mean it gets a maximum score of three in order to avoid prioritising foods that have massive amounts of one nutrient versus foods that have high amounts of a number of nutrients.


One example of where this limitation comes into play is kale, which has a massive amount of vitamin K, [6] versus spinach which has a high amount of vitamin K but also has a range of other nutrients and ends up with a higher overall nutrient density score.  Because of the upper limit on the score for a single nutrient, this system would give a higher ranking to spinach which has a more well-rounded nutrient profile rather than simply being an over achiever in one or two nutrients.


No weighting of individual nutrients has been applied.  Weighting one nutrient as more important than another could be useful for a particular person with a particular goal or health condition.  However at the same time it’s nice to keep the analysis ‘clean’ to avoid arguments about bias. [7]  This unweighted approach highlights foods that have a broad spectrum of nutrients at significant levels.

minimum protein

So what is the minimum amount of protein that we require?

  • According to Nuttall and Gannon [8] the body requires between 32 and 46g of high quality dietary protein per day to maintain protein balance in a starvation situation. This equates to around 6 to 7% of calories in a 2000 to 2500 calorie diet but in starvation this is how much your body will cannibalize off your muscle per day if it gets no food, with the rest coming from body fat stores.
  • The recommended daily intake (RDI) of 0.8g/kg LBM is based on an estimated average requirement (EAR) of 0.66g/kg LBM which is the amount estimated to ensure that 50% of the population are not deficient.
  • Ron Rosedale, an advocate of ketogenic diets and minimising insulin as far as possible, recommends 1.0g/kg LBM for most people, 1.25g/kg LBM if you exercise and 0.8g/kg LBM for people with diabetes. [9] [10] [11] [12]

typical protein intake

As a reference point, the American diet typically consists of between 65 and 100g of protein per day.  According to NHANES fat and protein intake has decreased over the last few decades as carbohydrate content has increased. [13]


maximum protein

Most people will find it difficult to eat more than 30% of their calories from protein from whole foods, though you can push to higher levels by using protein powders.

Protein is very satiating so our intake is typically self-limiting.  The upper limit of protein intake for our liver is said to be 200 to 300g per day, or 35 to 40% of calories from protein. [14]

The term ‘rabbit starvation’ refers to a situation where people who only have lean rabbits to eat starve because they just can’t process any more protein. [15]

This video from Dr Donald Layman gives an overview of the benefits of protein.  He says that the range of safe protein intake varies between 0.8g/kg LBM and 2.5g/kg LBM.


Steve Phinney’s WFKD triangle suggests that people will find it hard to achieve nutritional ketosis with protein levels greater than 30% of calories or 2.4g/kg LBM, even if they have very low carbohydrate levels.


In The Myth of 1 g/lb: Optimal Protein Intake for Bodybuilders[16] Menno Henselmans argues that there is a limit to how much protein we can metabolise to grow muscle.  More protein does not necessarily mean more “gainz”.  As shown in the figure below:

  • people who are sedentary will be unlikely to gain more muscle with more than 0.8kg/kg lean body mass;
  • an endurance athlete may not gain more muscle with more than 1.3g/kg lean body mass; and
  • a strength athlete will not activate more protein synthesis by eating more than around 1.8g/kg lean body mass.


Interestingly Luis Villasenor (aka Darth Luiggi, pictured below), who runs Ketogains and is studying under Henselmans, recently changed the upper limit on his Ketogains macro calculator to include an upper limit of 1.8g/kg total body mass for protein to reflect Menno’s findings.


Luis said his rationale behind the protein levels in the Ketogains macro calculator:

  • lower limit of 0.8 to 1.0g per pound of lean pound of body weight, and  .
  • upper limit that will not affect ketosis = 1.8g per total kilogram.

Luis says that protein grams over 1.8g/kg total body mass should be counted as 50% toward daily carb intake and that higher levels of protein can hinder ketone production.  Higher levels of protein are not necessarily going to be a problem if your carbs are low and / or you are insulin sensitive and have your blood glucose levels under control.  This approach certainly seems to be working for Luis!

While there is a limit to the amount of amino acids we can use for muscle growth and repair, it’s important to note that amino acids are also important for mental health.  Julia Ross’s Mood Cure details how nutrients can be targeted to address depression and a range of other mental health issues. [17]  So protein in excess of what is required for our muscles is not necessarily wasted.

To summarise, the table below shows a comparison of these minimum and maximum levels in terms of percentage of daily calories, grams per kilogram of lean body mass and grams per day for a 2250 calorie per day diet for a man with 74kg lean body mass (LBM).

scenario PRO (g/day) PRO (%) g/kg LBM
minimum (starvation) 32 6% 0.4
RDI / sedentary 59 11% 0.8
typical 90 16% 1.2
strength athlete 133 24% 1.8
WFKD max 169 30% 2.3
maximum 197 35% 2.7

the Goldilocks glucose zone

I believe that consideration of optimal protein intake needs to have some regard for the carbohydrate level, given that we can get the glucose we need from both protein and carbohydrate.  If we increase the glucose load (from carbohydrates and / or protein) above what our body can tolerate then we drive up blood glucose and insulin levels, thus ending up with obesity and diabetes.

In the Goldilocks Glucose Zone article I looked at how we can obtain glucose from both carbohydrate and protein (through gluconeogenesis).  If we are more towards the bottom left of the plot of protein versus carbohydrate chart, our diet is more likely to be more ketogenic.  However, the more we minimise protein and carbohydrates the more we risk not obtaining the fibre and nutrients that might be harder to find on a higher fat diet diet of butter, coconut oil and avocado.

What this means in practice is that the more ketogenic our diet is, the more intentional we have to be about achieving adequate nutrition and maximising the nutrient density of the foods that we eat.

Microsoft Word Document 23082015 22135 PM.bmp

therapeutic ketosis

Let’s first look at the most extreme end of the spectrum, therapeutic ketosis.  This approach is based on the theory that a low insulin load diet will help control conditions such as cancer and epilepsy.

The table below shows the weighting criteria for therapeutic ketosis, with a low percentage of insulinogenic calories being the primary goal.  Minimal weighting is given to fibre and calorie density with moderate weighting given to nutrient density.

ND / cal fibre / cal ND / weight calories / 100g insulinogenic (%)
10% 5% 10% 5% 70%

People who are aiming for therapeutic ketosis in an effort to conquer cancer or epilepsy are typically motivated to reduce the insulin load of their diet, sometimes at the expense of nutrition.

Check out Andrew Scarborough’s story for a fascinating example of someone who has worked very hard to maximise nutrition while still having a very low insulin load.   Andrew focuses on maximising omega-3 fats and has to avoid ketogenic favourites such as avocado and coconut oil due to his intolerances.

I’ve sorted the foods in the USDA food database using these weightings and plotted the highest ranking 5% of these foods against the protein versus net carbs chart with the various levels of ketosis.  On average these foods have a very low 1.4% net carbohydrates and a fairly low 11% protein (i.e. approximately the RDI minimum level for sedentary people).


In the bottom left hand corner of the chart butter, cream and oils make the list due to their very low percentage of insulinogenic calories.  However what I find most interesting is to look at the foods that make the list due to their higher nutritional value such as spices, olives, bacon and cheese.

You can download the list of the top 500 foods ranked using this system here to review the outputs of the system in more detail.  Because nutrient density and fibre don’t play a big part in this ranking there are lot of high fat foods here that don’t have a lot of amazing nutritional properties, so you’ll need to use your discretion to find foods from this list that suit your goals other than just being high fat.

diabetes and nutritional ketosis

The table below shows the weighting to identify optimal foods for diabetes and nutritional ketosis.  This approach is less extreme than the therapeutic ketosis approach with only half the weighting in the multi criteria analysis being given to the insulinogenic properties with some of the weighting spread to nutritional density and calorie density.

ND / cal fibre / cal ND / weight calories / 100g insulinogenic (%)
10% 10% 10% 10% 60%

The plot below shows the top 5% of foods in the USDA database ranked using this weighting in terms of protein versus net carbohydrates.  On average these foods have 2% net carbohydrates and 20% protein which aligns reasonably well with what we see practiced in the low carbohydrate and diabetic community.

Protein levels are moderate in order to ensure adequate levels of amino acids while managing the insulin load of the diet.  With such low levels of carbohydrates this approach is certainly still ketogenic!

To give some context it’s worth noting that the typical western diet is about 15% of calories from protein, [18] so most people would need to eat more protein to achieve these levels.

Again, it’s interesting to look at the outliers that make the cut due to their nutritional density such as black pepper, parsley, rosemary, spinach and liver.  Although these foods have a higher percentage of insulinogenic calories it would be hard to overeat them.

You can download the top 500 foods using this system here.


weight loss

This approach to the weighting of the multi criteria ranking is designed for someone who already has their blood glucose levels under control, but still wants to lose more weight.

Some people find that they can achieve their target weight on a ketogenic diet.  However others don’t have the same degree of success and perhaps may need more help to reduce their calories lower through food choices to help them lose weight.

While many people find that a higher fat diet will naturally lead to increased satiety with the reduced insulin load enabling stored body fat to be released for fuel, it is still possible to overdo your calories on a high fat diet.  Too many calories, even with ketogenic macronutrient ratios, can still mess with your insulin sensitivity and cause weight gain. [19]

Chris Gardner’s A to Z trial [20] highlighted that people with insulin resistance lost the most weight if they reduced their insulin load, while people who were not insulin resistant could lose weight on any diet as long it was low enough in calories. [21] [22]


The table below shows the weighting for the suggested optimal foods for weight loss which prioritises high fibre, high nutrient density and low calorie density foods rather than focusing as much on insulin load.

ND / cal fibre / cal ND / weight calories / 100g insulinogenic (%)
15% 10% 15% 20% 40%

This approach might be appropriate for a person that still has weight to lose but has gained control of their blood glucose levels and meets the following criteria:

  • HbA1c < 5.4mmol/L
  • fasting blood sugar < 5.0mmol/L (90mg/dL)
  • average blood sugar < 5.4mmol/L (100mg/dL)
  • post meal blood sugar < 6.7mmol/L (120mg/dL)

Being overweight and / or having a larger than desirable waist line is usually a good sign you are insulin resistant, so most people wanting to lose weight should typically start with the diabetes / nutritional ketosis approach (possibly with some intermittent fasting) until they get their blood glucose levels and insulin resistance under control.

As indicated by the chart above from the analysis of Gardner’s study, the low carb approach worked the best regardless of insulin resistance status.  Designing a diet that is high in fibre, has high nutrient density,  while still being fairly low in carbohydrates and calories, may help automatically limit food intake and manage calorie intake. This way we are focusing on food quality rather than having to count calories!

The plot below shows the top 500 foods using the weight loss ranking.  You will note that there are a lot of foods that sit well outside the ketogenic triangle.  On average we have 4% net carbohydrates and 36% protein.

Once we reduce the emphasis on glucose load the system will strongly prioritise protein rather than carbohydrate-based foods to source nutrition!


With the priority on high fibre and low calorie density, this approach would simply make it physically difficult for someone to overeat.  The insulin load will naturally be reduced because you just won’t be able to binge on calorie dense foods!

If you can manage to eat 2250 calories per day (i.e. typical intake[23]) of the top 10% of foods prioritised using these weightings you would be eating 150g of net carbs but also 130g of fibre!  Most people would struggle to eat this much fibre and volume of food.  If they were able to consume only 1500 calories per day they would be getting 120g of protein, 100g net carbs and 85g of fibre.

While 100g of net carbohydrates is high by low carb standards it is much less than the 300g per day consumed by most people and is unlikely to drive insulin resistance in most people because the carbohydrates being consumed would have an extremely low glycemic index!

If you find your blood glucose levels are drifting up using this approach (i.e. post meal blood sugars of greater than 6.7mmol/L or 120mg/dL) you could revert back to the diabetic / nutritional ketosis approach to make certain you are keeping your insulin levels low enough to be able to successfully lose weight.

You can download the top 500 highest ranking foods using this approach here.

athletes and metabolically healthy

As shown in the weightings in the table below, this approach maximises nutrient density without trying to minimise calorie density.  If you’re a lean athlete trying to refuel quickly you won’t want to be eating a pile of low calorie density lettuce and spinach to fill your calorie needs.  You will be interested in maximising nutrient density to maximise health and athletic performance.

ND / cal fibre / cal ND / weight calories / 100g insulinogenic (%)
30% 10% 30% 10% 20%

The top 500 foods using this ranking are plotted below.  On average these foods are 4% net carbohydrate and 36% protein.  Most people will find it difficult to eat greater than 30% protein from whole food sources, so as with the weight loss approach, this scenario would enable you to basically eat as much protein as you wanted from real food sources.

If you’re lean and insulin sensitive then it’s hard to eat too much protein.  Again, if you found that your blood glucose levels were driving up with this approach you should revert to foods with a lower insulin load.

You can download the top 500 highest ranking foods using this approach here.


without consideration of insulin load

To see if the insulin load was adversely biasing the analysis against carbohydrate and towards protein, I thought that it would be worth running the analysis without the percentage of insulinogenic calories as a consideration.

The table below shows the weightings used for this scenario which heavily bias towards nutrient density and with a smaller weighting towards fibre and calorie density.

ND / cal fibre / cal ND / weight calories / 100g insulinogenic (%)
35% 15% 35% 15% 0%

The chart below shows the most nutrient dense 500 foods plotted on the protein versus net carbohydrates chart with an average of 54% protein and 12% carbohydrates.  While the carbohydrates comes up a little compared to the other scenarios the protein is very high.

If the insulin load is not considered the macronutrient split of the highest ranking foods is 54% protein and 12% carbohydrates.  However this isn’t realistic as we can’t physically eat that much protein, so we will end up eating either more fat or carbohydrate rather than all that protein.  If you have some level of insulin resistance, fat is the logical choice rather than carbohydrates in order to maintain normal blood glucose levels.


You can download the highest ranking 500 foods using this criteria here.


The table below lists the average protein and net carbs for the highest ranking foods for the four dietary approaches.  The figure below shows this graphically.

approach protein (%) net carbs (%) insulinogenic (%) protein (g) net carb (g)
therapeutic ketosis 11% 1% 8% 62 6
diabetes / nutritional ketosis 20% 2% 13% 113 11
weight loss (2250 cal) 36% 4% 24% 203 22
weight loss (1500 cal) 29% 3% 19% 109 11
athlete / metabolically healthy 36% 12% 38% 203 67
typical western diet 16% 50% 59% 90 281


When it comes to weight loss it’s also useful to consider the fat that is coming from your body.  I have shown the macronutrient split for the weight loss scenario both in terms of a 2250 calorie per day diet and as a 1500 calorie per day diet assuming that the calorie deficit is coming from fat stores.

For athletes and the metabolically healthy, protein is largely unrestricted when it comes to maximising nutrient density.  The maximum amount of protein for people who are insulin sensitive will come down to how much they can physically eat.

Even if blood glucose levels are not a concern the most nutrient dense foods are still quite low in net carbohydrates.

If you have some degree of insulin resistance or elevated blood sugar it may be useful to moderate protein levels to some extent as well as cutting carbs.  If you are watching your blood glucose levels and / or ketones you can wind the amount of protein back until you achieve your target levels.

In all scenarios nutrient density was maximised, with much less carbohydrate relative to the western diet.


[1] See comments section in https://optimisingnutrition.wordpress.com/2015/06/15/the-blood-glucose-glucagon-and-insulin-response-to-protein/ and https://optimisingnutrition.wordpress.com/2015/06/22/why-Regardwe-get-fat-and-what-to-do-about-it-v2/

[2] http://www.precisionnutrition.com/what-are-your-4-lbs

[3] It’s worth noting that Metformin works by limiting gluconeogenesis (i.e. the conversion of protein into glucose) in order to help the body better manage blood glucose levels (see http://www.ncbi.nlm.nih.gov/pubmed/24847880).

[4] http://ketopia.com/nutrient-density-sticking-to-the-essentials-mathieu-lalonde-ahs12/

[5] The omega 6 fatty acids are also classed as essentially however it is generally recognised that we have more e

[6] Possibly more than the body could ever use especially in a low fat environment given that it is a fat soluble nutrient.

[7] For example see Chris Masterjohn’s review of Joel Furhman’s ANDI index at http://www.westonaprice.org/book-reviews/eat-to-live-by-joel-fuhrman/

[8] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3636610/

[9] http://www.ketogenic-diet-resource.com/daily-protein-requirement.html

[10] http://drrosedale.com/blog/2011/11/21/ron-rosedale-protein-the-good-the-bad-and-the-ugly/#axzz3gVOJ69np

[11] http://www.meandmydiabetes.com/2010/05/07/ron-rosedale-protein-the-good-the-bad-and-the-ugly/

[12] While Rosedale has his reasons for reducing protein such as longevity and avoiding the mTOR metabolic pathway, this minimal protein approach may not be ideal if your goal is to maximise nutrition.  Check out this interesting discussion between Robb Wolf and Jamie Scott from the 52 minute mark in this podcast about the balance between health and longevity when it comes to optimal protein intake levels.  – http://robbwolf.com/2015/07/21/episode-279-jamie-scott-the-state-of-paleo-and-ahsnz/.  I tend to agree with the approach to maximise health and vitality now rather than eating for theoretical longevity with reduced health and vitality now.

[13] http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=9146789&fileId=S1368980012005423

[14] http://www.paleoplan.com/2011/04-22/meat-is-not-the-devil-high-protein/

[15] https://en.wikipedia.org/wiki/Rabbit_starvation

[16] http://bayesianbodybuilding.com/the-myth-of-1glb-optimal-protein-intake-for-bodybuilders/

[17] http://www.moodcure.com

[18] http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=9146789&fileId=S1368980012005423

[19] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1819381/

[20] http://jama.jamanetwork.com/article.aspx?articleid=205916

[21] http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3504183/

[22] http://profgrant.com/2015/07/21/5-reasons-sugar-not-fat-is-the-problem/

[23] http://www.nhs.uk/chq/pages/1126.aspx?categoryid=51

your personalised food ranking system

  • A number of attempts have been made to develop food rankings.
  • We can combine the concept of insulin load with nutrient density to help us make optimal food choices based on our goals, situation and budget.
  • This article looks at other ways to prioritise our our food choices quantitatively to design a food ranking to suit your situation, goals and budget.

Mat Lalonde’s nutrient density

Dr Mat Lalonde developed a ranking of foods based on nutrient density in terms of nutrients per gram using the USDA food database. [1]  This analysis identified organ meats as one of the more nutritious foods, with vegetables coming in second.  Fruits and grains landed much further down the list.

Lalonde noted that people wanting to lose weight may wish to prioritise in terms nutrient density per calorie, however he had chosen to analyse nutrient density in terms of weight as that might be more relevant for athletes (Lalonde is a CrossFit athlete as well as a biochemist). [2]

I was left excited, yet a little unsatisfied, wondering what the ranking might look like in terms of calories, or maybe some other measure.

Aggregate Nutrient Density Index (ANDI)

Joel Fuhrman’s Aggregate Nutrient Density Index (ANDI) ranks foods based on micronutrients per calorie [3] but excludes a number of essential vitamins and minerals while placing extra emphasis on the oxygen radical absorbance capacity.

This approach heavily biases plant foods and seems to ignore the nutritional benefits of animal foods. [4]  Kale ranks at the top of the list, largely due to its massive amount of vitamin K.

Unfortunately a massive dose of vitamin K isn’t much use to us in the context of a low fat given that vitamin K (along with vitamin D and E) is a fat soluble vitamin.  It’s also not much use having a food that ranks off the chart in one nutrient but it’s that good in a number of other areas.   Vitamin K is important but you can only absorb so much in one day.

Another criticism that has been levelled at ANDI is that simply using nutrition per calorie prioritises very low calorie density foods that may not be viable for anyone doing a significant amount of activity.

Dave Asprey’s Bulletproof Diet

Dave Asprey developed the Bulletproof Diet Infographic [5] which is a simple ranking of foods to avoid and preference based on both nutritional density and toxins.

The downside of this is that it shows only a select range of foods and doesn’t explain why each of the foods has the ranking that is has been given (though there is a good discussion of the toxins and various issues in his book [6]).

Most people would be happy with this visual list of foods to preference and avoid, and I recommend you check it out, however I wanted to see the numbers to understand why one food ranked above another.

nutrient density per dollar

I also came across a food ranking system in terms of nutrient density per dollar.  Dale Cumore of the blog Solving Nutrition [7] had created a ranking based on nutrient density per dollar cost of that food to arrive at the cheapest way to get nutrition for around 1000 foods that he could find cost data for.

Dale included a link to his  spreadsheet on his blog (in which he has mimicked Lalonde’s analysis [8]) for people to have a play with.  So I downloaded it to see what I could do with it. [9]     After dropping out the fortified products, we get the following list of foods based ranked on nutrient density per dollar.

  • bagels
  • French rolls
  • croissants
  • muffins
  • lentils
  • tortillas
  • rice
  • parsley
  • beef liver
  • spaghetti
  • Chinese cabbage (Bok Choy)
  • sunflower seeds
  • White bread
  • chicken liver
  • peanut butter
  • skim milk
  • peanuts
  • chives
  • whole eggs
  • brown rice
  • sweet potato
  • cabbage
  • orange juice

Grains are actually a cost effective way to get nutrients, however not necessarily the most healthy.    People believe that most if not all grains should be avoided. [10]  My ten year old daughter knows that if she eats bread she will end up tired, with a stomach ache and dark circles around her eyes.  However if  cost is your number one priority you might find this list useful.

cost per calorie

Cost will always be a consideration to some degree.  Some people may not have the finances to buy grass fed organic while others will have the means to invest in food as preventative medicine.  Listed below are the cheapest foods in terms of cost per calorie.  Again, grains (including white rice), candy and sugar rank up there with some of the cheapest ways to get calories. [11]

While it’s true that grass fed beef, salmon and organic vegetables can be more expensive than boxed cereals and sugar, it’s also worth noting that obtaining significant proportion of your calories from fats such as coconut oil and butter can actually be very cost effective on a per calorie basis.

  • pumpernickel rolls
  • croissants
  • bagels
  • canola oil
  • French rolls
  • margarine
  • what muffins
  • coconut oil
  • granulated sugar
  • rice
  • brown sugar
  • mayonnaise
  • doughnuts
  • tortillas
  • cake mix
  • peanut butter
  • cranberry juice
  • spaghetti
  • sausage
  • corn starch

nutrient density per calorie

Nutrient density per calorie is a useful measurement for someone wanting to lose weight while maximising nutrition.   One line of health and weight loss thinking says that once the body obtains adequate nutrients it will stop searching for food and overeating will be minimised. [12]  Using this approach vegetables shoot to the top of the list with things like spinach, liver, seafood oysters, kale and broccoli rank really well.

  • spinach
  • chicken liver
  • beef liver
  • beet greens
  • veal liver
  • pork liver
  • duck liver
  • goose liver
  • turnip greens
  • mustard greens
  • parsley
  • chard
  • oyster
  • coriander
  • dandelion greens
  • basil
  • caviar
  • kale
  • broccoli
  • All bran
  • collards

fibre per calorie

One of the more exciting concepts in the diet space recently is the concept that what you eat could possibly change your gut bacteria for better or worse.

While this area is still in its infancy the thinking is that lean people have a higher bacteriodes : fermicutes ratio and that this can be influenced by eating more fibre and taking prebiotics.

Typical daily fibre intake is around 17g for those of us in western civilisation. It is said that African hunter gatherer children obtain more than 150g of fibre per day from eating unprocessed foods in their natural state [13] and before the invention of fire and cooking our ancestors were eating more than 100g of fibre per day. [14]

Fibre in carbohydrate-containing foods neutralises the insulinogenic effect of the carbohydrate.  Fibre is not digestible by the human gut and hence it does not provide energy or cause a rise in blood sugar or insulin.

The typical western recommendation is to get at least 30g of fibre per day to improve your blood sugar and cholesterol levels.  Most people don’t achieve these levels even when eating “healthy whole grains”, largely due to the high level of processing in most popular foods.

It’s also worth noting that it’s better to lightly steam your veggies rather than cooking them until they’re soft so that the fibre remains intact.

Ironically the number one recommended source for fibre is from “healthy whole grains”.  While whole grains will be marginally better than processed grains such as white bread, they also have a high glycemic load and will be much more insulinogenic than other options such as non-starchy vegetables.  The end result of eating the whole grains is increased blood sugars and cholesterol, which is exactly what “healthy whole grains” was meant to help us avoid!

If we rank for fibre per calorie we end up with a few spices such a cinnamon, curry powder, or cocoa at the top of the list along with vegies such as turnip, artichoke, sauerkraut, cauliflower.  All Bran features in the list but only because it has been fortified with extra fibre.

  • cinnamon
  • turnip greens
  • artichoke
  • curry powder
  • sauerkraut
  • cauliflower
  • raspberries
  • lettuce
  • blackberries
  • lemon peel
  • All Bran (w/ added extra fibre)
  • oregano
  • wheat bran
  • eggplant
  • basil

practical application

These lists of foods ranked based on one measurement or another are interesting, however they are not particularly useful by themselves.  If we went by Lalonde’s system we’d be eating bacon and organ meats all the time.  If we went by the ANDI system we’d be living off kale.  And if we just looked at the proportion of insulinogenic calories we would be living off butter, cream and oils.

But it gets interesting though when you can combine the various measurements to highlight foods to suit your individual goals.

In my previous articles on diets for weight loss, blood sugar management and athletes I provide a list of optimal foods for using different weightings for the following:

  • nutrient density per calorie,
  • fibre per calorie,
  • nutrient density per dollar,
  • nutrient density per 100g,
  • proportion of insulinogenic calories,
  • calories per 100g, and
  • cost per calorie.

Listed below are the weightings that I’ve devised for each situation.

I’ve also developed a suite of ‘cheat sheets’ to highlight optimal food choices to suit your goals, whether they be weight loss,  normalising weight loss or or athletic performance.

Why not print one out and stick it to your fridge as a helpful reminder or use them for some inspiration for your next shopping expedition?

In the next article we’ll look at how we can use this style of analysis to identify diabetic friendly, ketogenic, nutrient dense meals.

weighting for blood sugar control and ketosis

ND / calorie fibre / calorie ND / 100g ND / weight insulinogenic (%) calorie / 100g $ / calorie
15% 5% 5% 10% 50% 10% 5%
weighting for weight loss
ND / calorie fibre / calorie ND / 100g ND / weight insulinogenic (%) calorie / 100g $ / calorie
15% 10% 10% 5% 20% 30% 10%
weighting for athletes and metabolically healthy
ND / calorie fibre / calorie ND / 100g ND / weight insulinogenic (%) calorie / 100g $ / calorie
15% 10% 10% 30% 20% 5% 10%
weighting for theraputic ketosis
ND / calorie fibre / calorie ND / 100g ND / weight insulinogenic (%) calorie / 100g $ / calorie
5% 5% 5% 5% 70% 5% 5%


[1] http://ketopia.com/nutrient-density-sticking-to-the-essentials-mathieu-lalonde-ahs12/

[2] https://www.youtube.com/watch?v=HwbY12qZcF4

[3] http://www.wholefoodsmarket.com/healthy-eating/andi-guide

[4] http://www.westonaprice.org/book-reviews/eat-to-live-by-joel-fuhrman/

[5] http://www.bulletproofexec.com/wp-content/uploads/2014/01/Bulletproof-Diet-Infographic-Vector.pdf

[6] http://www.amazon.com/The-Bulletproof-Diet-Reclaim-Upgrade/dp/162336518X

[7] http://blog.paleohacks.com/ultimate-guide-paleo-diet-budget/

[8] The analysis considers the relative amount of calcium, iron, magnesium phosphorus, potassium, zinc, copper, manganese, selenium, vitamin C, thiamine, riboflavin, niacin, panto acid, vitamin B6, choline, vitamin B12, Vitamin A, vitamin D, Vitamin E and Vitamin K across more than 1000 foods.  No weighting of these vitamins based on a view of their relative importance, though this refinement could be made to the analysis for a specific need.  This unweighted approach however highlights foods that have a broad spectrum of nutrients at significant levels.

[9] The statistical analysis in the spreadsheet downloaded compares the value of a nutrient in each food to the average of the full database of foods and gives it a score based on the number of standard deviations from the mean.  I also modified the spreadsheet such that a score for one nutrient could not be greater than three (i.e. three standard deviations from the mean).   Just because Kale has an inordinate amount of Vitamin K doesn’t mean that it ranks at the top of the list on the basis of just one nutrient.

[10] https://www.youtube.com/watch?v=VvfTV57iPUY

[11] If you wanted to view this cynically you could say that the fact that grains and sugars have the lowest cost per calorie enables food manufacturers to place the largest mark up on these foods when reselling them in cardboard boxes in the supermarket.  It’s harder to put a bar code on generic vegetables and meat products that are already relatively expensive.

[12] See discussion in chapter 17 Nutrient Hunger in Paul Jaminet’s Perfect Health Diet where he notes that a nourishing, balanced diet that provides all nutrients in the right proportions is the key to eliminating hunger an minimising appetite and eliminating hunger at minimal caloric intake is a key to weight loss.  

[13] http://www.abc.net.au/catalyst/stories/4067184.htm

[14] http://www.gregdavis.ca/share/paleo-articles/academic/The%20Ancestral%20Human%20Diet%20by%20S.%20Boyd%20Eaton.pdf