Accuracy In Nutrition: How Accurate Are The Atwater Values
QUESTION: “I was wondering about your latest article about protein per 100 calories. The numbers don't add up to me when I do the math using protein and carbs at 4 calories per gram and fat at 9. The vegetables are counting up the ~120 calories and the steak at only 96. Is there something I am missing?
No, not at all. You just have to understand that the 4/4/9 figures are just a rough average and so most often, there is a little discrepancy. For those not familiar with this issue, it is often said that every gram of protein or carbohydrate yields 4 calories and every gram or fat yields 9 calories. So, the 4/4/9 is the approximate yield of calories from protein, carbohydrate and fat (P/C/F). So, often times, you will see a value given for the total calories of a food, but when you take the grams of protein, carbohydrate and fat and multiply them out and then add them up, they do not always equal the calorie value 100%. You can read a good discussion about the 4/4/9 and Atwater here...
QUESTION: Regarding the percentages of protein in your new chart. I assume you used the general conversion factors (4 cal/gram), rather than the Atwater conversion factor (2.44) in the USDA Access database. I created a macronutrient chart of my own and I made the decision to use the Atwater factors. I am aware that Atwater’s work was done 100 years ago, and that both macro- and micronutrients can vary widely in different samples of the same food, so in a sense this is all silliness … but I’m curious to hear your perspective on this topic.
The only reason I created the chart, based on the percentage of calories from protein is to correct several charts that were circulating around the internet and social media that were way off. Those charts all used the same system (4/4/9) I used and it is one of the reasons why I used it. Some of their numbers had numbers that were just way to high, almost double or more, so it was not an issue of which factor they used.
However, in regard to your points, here are my comments..
1) I hope you do know that not only is Atwater 100 years old, it is highly inaccurate and may be as inaccurate as using the 4/4/9. So, it may be slightly better, but runs into many of the very same same problems.
2) the Atwater values were modified around 1998, but, in the end, the modified Atwater values may still not be any better.
3) Even if you use the value of 2.44 value for protein for all vegetables, that is only an average and does not represent each individual vegetable, which can vary greatly. So, we are back to dealing with the same issues again.
4) Even if we could identify a specific value for each and every individual food, it would still not be accurate as so many other variables impact the metabolizable energy including temperature, texture, amount of chewing, blending, along with how much fiber is in the food, what types of fiber, etc., etc.,
I raise some of these similar issues in my articles on counting calories.
So, is one inaccurate formula better than another?
For me, the real issue is simplicity and so
1) the main issue is that protein is not an issue, which I think is the most important message.
2) I really see no reason (or value) for most people to calculate any of these numbers once they get and understand point 1
3) We have to remember that, in the end, these are all arbitrary numbers so, whether we measure the calories with Atwater, modified Atwater, 4/4/9 or method X,Y or Z, it really doesn't make any difference. There will always be way to many variables to account for no matter how you measure it.
4) Therefore, since all these approaches have inaccuracies and not one of them is perfect, we either open a can of worms that can't be fully resolved, or we stick with a simple accepted formula that at least approximates the values and use it across the board for comparative purposes. This way, at least it is consistently inaccurate ;)
Otherwise, like with calories, we are only attempting to "nail jello to the wall" and trying harder is not going to yield any better results.
That's my perspective :)