I had the idea of doing a nutrient density index in my head for almost half a decade now. I didn't like the method of the ones I had seen done that much. It was also a lot of work to do, so I've always delayed it until now.
What I call nutrient density is how nutritious a food is. To be healthy, beyond the basic advice of exercising, sleeping well and eating well, you also need to make sure you're eating enough vitamins and minerals. But isn't this included in the eating well advice? For the most part, yes. If you eat a decently healthy diet with a variety of foods, you will likely get plenty of vitamins and minerals, and worrying about specific ones is not worth your time. However, there are some caveats with this as we will see.
The first problem is that getting all the vitamins and minerals required isn't as easy as you think. Indeed eating a varied diet with plenty of veggies helps a ton, but it does depend on which veggies you're eating.
Vitamin K isn't a hard vitamin to get. It comes mostly in veggies. But let's say that your favourite vegetables are mushrooms, tomatoes, onions and peppers. These are all pretty healthy, yet you would have a hard time getting vitamin K. This is because vitamin K is mostly found in dark leafy green vegetables. So unless you're eating those (or some other foods like chia seeds), you're not going to get enough. One pound of all the vegetables mentioned combined doesn't even get you to 10% of the RDA. Another example is Vitamin E. It's abundant in nuts, avocados, and some vegetables. And most sources don't have a ton. Unless you're consistently eating foods very high in Vitamin E, you won't get to the RDA.
My point isn't to focus on specific foods nor specific nutrients. What is important to realize is that it's not as simple as eating vegetables, because not all vegetables have the same nutrients. This is obvious for people who have bothered to actually track their micro-nutrients. When I first started doing so many years ago, I was somewhat annoyed that I often ended up not getting enough of a specific vitamin or mineral, even though I considered my diet to be decently healthy and varied. It was almost a matter of chance of not having the right combination of foods. This isn’t to say that I was deficient, which is another matter, but I was likely not getting a sufficient intake for ideal health.
Getting enough nutrients
The best way to solve this problem is by eating as many different foods as you possibly can. However, most people don't even eat a healthy diet in the first place. Having them making sure that they're eating many different foods is an extra burden that many simply aren’t going to follow. The way to make this more practical is to rotate foods. For example, you can have beef, eggplant, zucchini and squash in one week, and in another week have salmon, leek, brussels sprouts, and asparagus. This variation allows for variety in the long term without the immediate burden of having countless different foods. Bags of mixed vegetables or mixed nuts are also great to get variety with little work.
But from my experience, despite these helpful tips to get variety, many still end up having a core set of foods which tends to result in suboptimal micro-nutrition. Furthermore, this emphasis on variety also seems to imply that foods are all more or less the same. All you have to do is to eat many of them so they average out with what some have and don't have. But this is wrong.
The idea is something like this: food X may have vitamin C and no vitamin E, and food Y might have vitamin E but no vitamin C. Thus you just have to eat both and you get the vitamin C and vitamin E. As a whole, this is a good heuristic. But if you dig deeper, you will see that some foods have a lot of vitamin C and E (in our hypothetical scenario), while some foods have almost zero of it. And they don’t necessarily compensate with other nutrients.
I believe this was a push-back against the notion of superfoods. Yes, popular media over-rates specific foods rather than one’s diet as a whole, and this “new” exotic berry isn’t going to increase your lifespan by 10 years or make you lose 20 pounds. But this does not mean that all foods are equally nutritious. So if you're going to have a core set of foods, to me it seems to make sense to at least pick the more nutrient-dense ones. But which ones? That's what we are here to find out.
I picked up a database of 1923 foods, which listed all their nutritional information. Out of those, I filtered out items that aren't relevant for our purposes, such as baby food, junk food, raw food that isn't eaten raw, spices and so forth. I ended up with a list of about 300 foods. This list is somewhat biased because I only included foods I'm familiar with. I'm sure there are foods that I never heard of that I excluded, especially foods from different cultures than my own (for example Asian or African). Nevertheless, most people reading this will be in the same boat as I am, and thus the selection is good enough.
What I did was first see the percentage of the RDA that any given food had for each nutrient in one 100g serving. Then I added all these percentages together. This sum is helpful because it tells you how nutrient dense a food is in general, and not any specific nutrient. So if Food A has 90% of RDA Vitamin X, while food B has 80% of Vitamin X and 80% of vitamin Y, the latter is clearly superior overall, and this would clearly show up in our number (90% vs 160%).
But there was one problem with this method. Vitamins and minerals generally only give you a benefit to a certain point. More isn't better as long as you're getting enough. Yet, some foods have a specific nutrient way beyond the RDA. For example, you only need 3000 IU per day for vitamin A. Yet, 100g of carrots have 16700 UI, 5 times as much (ignoring any absorption concerns for simplification). This boosts the nutrient density index massively, but it's not actually providing any benefit. So I capped all the nutrient numbers to a maximum of 100% of the RDA. If they went above it, it wouldn't affect the nutrient density index (NDI).
With that done, this is the following top 30, from the highest to the lowest:
This ranking isn’t wholly accurate because it was full of organ meats, but most people don’t eat them. Most were liver, so I clumped them all together under “liver”, averaging the highest 3 organ meats: beef liver, chicken liver and turkey liver. For organ meat that isn’t liver, turkey heart was in the 11th place. I also grouped sunflower, sesame and pumpkin seeds together. They were 3rd, 4th and 5th place respectively.
I standardized the scores to a maximum of 100 to make the numbers more intelligible. Meaning that egg yolk doesn’t score 100 in any objective sense, but the scale itself was based on the highest score (the egg yolk). This makes it really salient how nutrient-dense some foods are. Even though the database had 300 foods, just going from top1 to top10 already reduces its nutrient density by half.
The biggest problem with this list is that it is based on nutrient density per 100 grams. You may notice that there are barely any vegetables on this list. This is because vegetables are foods with very low densities. So this ranking method isn’t fair for vegetables and some other foods that don’t do well when analyzed by weight.
Accounting for calories
This is even more crucial for people that are dieting and their calories are limited. While egg yolks are incredibly nutrient-dense, they are 322 calories per 100 grams. With 300 calories you could have a massive amount of vegetables. For example, cooked spinach is 23 calories per 100 grams. So for 322 calories, you could have 1400 grams of spinach. Would 100g of egg yolks be more nutrient-dense than 1400g of spinach? Egg yolks are only more nutrient-dense by a factor of 2. So to compensate for a 14-fold weight discrepancy, it would have to be more nutrient-dense by a factor of 7.
The previous ranking isn’t wrong per se, but it has its limitations. So I made another NDI, but this time taking into account calories. We can call the first Weight NDI, and the second Calorie NDI. The top 30 foods for calorie NDI are the following:
As you can see, a vastly different list. Almost exclusively vegetables, although surprisingly some organ meats still show up, which is impressive given the caloric constraint. I didn’t omit or group any foods on this list.
I tried doing an averaging score of the two lists, but it mostly ended up being a mix of both, so it’s not worth including here. I would also have liked to make a list taking into account volume. For instance, when I made the comparison of eggs vs spinach, the way I framed it makes spinach look a superior option due to its low calories, but you can’t physically eat that much spinach. Unfortunately, my dataset did not have the data for a volume comparison, and I don’t have the time to look up individual foods nor match it with another database.
We asked what are the most nutrient-dense foods we can eat, and this provides an answer. However, this is an answer with a lot of constraints. Nutrients interact with each other. All nutrients here are considered in isolation as if they are independent. They are not. Furthermore, nutrients have different absorption rates depending on the food. For example, spinach has a decent amount of iron. But that iron has very low bioavailability. Roughly only 2% is actually used, compared to 25% or so from meat. That's a massive difference that this index cannot account for. This happens in a lot of nutrients, such as calcium, zinc, and many more. Usually with plant options being inferior.
Most RDA values are incredibly old, and how they were determined is not so clear-cut. It is likely that for many nutrients you need more (or less) than what is prescribed for optimal health. Even if the RDA was accurate, it’s accurate for who? It would only be an average. There are individual differences in how we respond to food and absorb nutrients. For example, some people absorb vitamin B12 so poorly so they are always deficient no matter how much of it they get from food, and they require regular B12 injections. I have 3 different SNPs (nucleotide polymorphisms) that negatively affect my choline metabolism, and thus my intake needs to be higher than the average person.
Some nutrients are also surprisingly hard to get. For example, 98% of Americans don’t get the RDA for potassium. While the average population doesn’t have a super healthy diet, no other nutrient gets even close to this percentage. If I don’t make an effort to consume rich-potassium foods, even if my diet is healthy, I rarely get to the RDA. On the opposite side, some nutrients are so easy to get that they don’t need much attention, think of the vitamin A case previously mentioned. Yet, in the NDI all nutrients are counted equally which is an oversimplification.
Lastly, the list is biased due to my methodology, certain ways of dealing with the data and which foods to count towards the ranking. While I made the decisions that I think provides the most useful list, someone else might have done it differently and get a different outcome.
Because of all of this, you should not take this index too seriously. Furthermore, this is a level of detail that most people shouldn’t be worried about. Don’t put the cart before the horse. If you don’t exercise often, sleep well, consume a variety of fruit and vegetables and minimize processed food, that’s what you should focus on.
If you eat a healthy and varied diet, you will likely get the most nutrients without paying too much attention to any of this. But as we covered, the concept of nutrient density can be useful, and variety (especially realistic variety) isn’t a guaranteed solution to ensure an ideal intake. Most people aren’t going to rotate food sources that much.
This is not a ranking of healthiest foods. There are more considerations to health than vitamins and minerals. And this does not invalidate the need for variety in your diet. There are other benefits to variety such as phytonutrients. But despite all its limitations the NDI has, it can serve as a guide if you want to make more nutrient-dense food choices.