When using cronometer, or any other calorie counter for that matter, do you have to weigh bananas with the skin and put that value in, or do you weigh only what you eat? What's the correct way to get an accurate calorie count?
I have the same problem with mangos and oranges. I don't know if I should weigh the entire fruit or only what I eat. Does anyone know? Thanks!

You count. I wrote extensively on this topic in my article on my raw food website here. Whatever you do, the error will be too large to deem the calculation as accurate enough.


RawGosia channel
RawGosia streams

Haha, rawgosia. That's a very advanced answer for a simple question.

weigh the edible part on the fresh fruit.

When things are dried or semi-dried, the watercontent will vary and you can choose "date, pitted" instead of choosing grams.

But yes, it's an estimate. If you're still hungry, eat more. Or go out and exercise. Or drink more. Or sleep more. Or think positive.

Yeah, you get it.

Good LUCK!

I meant to say "You can't". Sorry.

Yes, haha. Me being pedantic, aren't I.


RawGosia channel
RawGosia streams

rawgosia, I was reading your article when noticing the Harris-Benedict formula for Basil Metabolism -

"The result of such analysis for BMR predictions, is the following Harris-Benedict equation:
for men, P = 13.7516m + 5.0033h + 6.7550a + 66.4730
for women, P = 9.6354m + 1.8496h - 4.6756a + 655.0955

..I though whoa, that's a lot of significant digits. A piece of trivia on significant digits -

"...the decimal representation of pi truncated to 11 decimal places is good enough to estimate the circumference of any circle that fits inside the Earth with an error of less than one millimetre, and the decimal representation of pi truncated to 39 decimal places is sufficient to estimate the circumference of any circle that fits in the observable universe with precision comparable to the radius of a hydrogen atom." [en.wikipedia.org]

Maybe Harris and Benedict were in fact that precise though for the average clinician to calculate weight, height and age to 5 decimal places would take some doing.

Francis Benedict was a chemist by training with an interest in magic (even as a pro later in life), and I gather a brilliant scientist. [www.whonamedit.com]

The technical analysis of the Harris-Benedict equation exists somewhere, at least at the library of the Carnegie Institute in Washington, DC. -

"Measurements on 136 men, 103 women and 94 new-born infants have been
analyzed biometrically with the purpose of determining the statistical constants
(means, standard deviations, coefficients of variation, coefficients of
correlation, and regression equations) which may serve as standard constants
in work on human metabolism until those based on more extensive series of
data are available." [www.ncbi.nlm.nih.gov]

I get your point about the equations being inadequate for humans who have great variability (especially if not accompanied by standard deviations etc.), and most especially not for rawists, and just though I'd hit you with some trivia and observations.

Thanks for the article!



Edited 3 time(s). Last edit at 10/17/2010 03:21PM by loeve.

rawgosia, there had to be a story behind Harris in the Harris-Benedict formula. James Arthur Harris was a botanist and apparently an accomplished statistician, having published in the Journal of the American Statistical Association --

[www.jstor.org]



Edited 1 time(s). Last edit at 10/17/2010 06:18PM by loeve.

Loeve, I doubt H&B meant that people should measure their weight etc with an accuracy of 4 digits after the decimal point. The constants appearing in the equation have such accuracy so as to have a little bit more better estimate in the end.

Interesting trivia for sure!


RawGosia channel
RawGosia streams

Oh, that should have read significant figures, not "decimal places".

The significant figures of the constants should reflect the number of significant figures expected in the variables, in this case 5 significant figures. That would mean weight might be recorded at, oh say, 56.456kg, height at 177.82cm and age at 32.765yrs, to give a BMR of 1892 kcal (four significant figures obviously) or whatever the formula would make that come out to. The "+ 655.0955" in the woman's Harris-Benedict formula really gets me with 7 significant figures. I think it was for aesthetics (symetry) or the art of presentation. Thanks to co-author/statistician Harris maybe.

Anyway that's how I was taught.

Thanks for following it along.



Edited 1 time(s). Last edit at 10/18/2010 12:27AM by loeve.

Ah, I get what you meant now. Well, I think that there are seven significant figures in the constant, because the desired degree of accuracy is 4 decimal places, which is what is often assumed in calculations (for example the default shown accuracy in MATLAB for example is 4 digits).

To see this, note that given weight 56.456kg, height at 177.82cm and age 32.765yrs, if P1 is the value calculated using 9.6354m + 1.8496h - 4.6756a + 655.0955, and P2 is the value calculated using 9.6354m + 1.8496h - 4.6756a + 655.10, then P2-P1=0.0045, which may be considered a too big error.


RawGosia channel
RawGosia streams

I see, well then if the desired degree of accuracy in that last constant (655.0955) is 4 decimal places then that might imply the answer to be accurate to within 3 decimal places, so P might = 1204.361 kcal, or rounded to 1204.36 accounting for just the significant figures of that last constant, for the example given in wikipedea --

"According to this formula, the woman in the example above has a BMR of 1204 kcal per day. During the last 100 years, lifestyles have changed and a survey in 2005 showed it to be about 5% more accurate." [en.wikipedia.org]

Then wikipedia talks about the Harris-Benedict being "5% more accurate". Well how accurate was it considered in 1918? I checked and a 1931 study compared several BMR calculators of the time and found the Harris-Benedict according to its own statistical analysis to have a standard deviation (for women ) of about 7% with an error of about 5%, so the 1204 in the above example might have an SV of +/- 84 kcal and an error of +/- 60 kcal, which supports your article saying that these calculators are not that accurate even for the general population.

[jn.nutrition.org] (Journal of Nutrition -pdf)

MATLAB using 4 decimal places leaves it up to the user to go back through their equation and decide how many significant digits or figures to walk away with, so calculating the area of a triangle, 0.5 x l x h gives an accuracy dependent on the accuracy of the l and h measurements and not on the constant, and 4 decimal places may be too many or too little depending on the application?

So I'm still more inclined to see the 4 decimal place convention here as a presentation device to convey an impression of great accuracy, maybe Benedict the magician at work.



Edited 2 time(s). Last edit at 10/18/2010 11:25AM by loeve.

"The result of such analysis for BMR predictions, is the following Harris-Benedict equation:
... for women, P = 9.6354m + 1.8496h - 4.6756a + 655.0955


-just crunching the numbers in my earlier example:

P = 9.6354m + 1.8496h - 4.6756a + 655.0955

P = 9.6354 x 56.465kg/1kg + 1.8496 x 177.82cm/1cm - 4.6756 x 32.765yrs/1yrs + 655.0955

P = 544.1 + 328.9 - 153.2 + 655.0955

P = 1374

I'm not sure that's done correctly accounting for decimal places and significant figures, but hope it illustrates the point about appearance of accuracy versus real accuracy.

The Harris-Benedict standard deviation of about 7% would give the result -

P = 1374 +/- 96

Harris-Benedict also published their analytical "error" at about 5% -

Hmmm, am not sure how to present P with "error" and SD... P = 1374, Error +/- 69, SD +/- 96



Edited 1 time(s). Last edit at 10/19/2010 04:05PM by loeve.

<<Then wikipedia talks about the Harris-Benedict being "5% more accurate".>>

Loeve, you mixed up two types of accuracies here, one and the less important, the one related to the parameters in the equation, which we discussed earlier. This really is a minor issue. Also, H&B's use of 4 digits after the decimal place is merely a standard notation.

The second, very important, is how accurately the equation predicts a person's BMR. As decided by science (and detailed in my article, see the reference in there), the accuracy of the H&B is very poor, making it an unacceptable tool for individual predictions.


RawGosia channel
RawGosia streams

<<Then wikipedia talks about the Harris-Benedict being "5% more accurate".>>

rawgosia, in context the Wikipedea article was saying that due to modern lifestyle changes the Harris-Benedict equation is 5% more accurate (today than it was then), without mentioning how accurate it was considered in 1918. I looked it up and found the standard deviation used by H-B listed in a comparative study of BMR methods of 1931 which found some BMR calculators tended to underestimate and some overestimate. Our modern living apparently worked in Harris-Benedict's favor. Lucky for them.

That means instead of about 7% SD and 5% error circa 1918 there is now 6.7% SD and 4.8% error circa 2010, the way I understood the wikipedia article anyway - pretty obscure stuff, and has little to do with your article other than digressing into statistical analysis and standards of notation.

Again, thanks for following, the info and excellent article.

Ah sorry, I see my mistake. This is the wiki quote in context --

"It was the best prediction equation until 1990, when MD Mifflin and ST St Jeor introduced the equation... During the last 100 years, lifestyles have changed and a survey in 2005 showed it to be about 5% more accurate (than the Harris-Benedict BMR equation)."

Here's the thing, rawgosia, people understand a little statistics and know that an estimate has error and the Harris-Benedict equation is an estimate. They might not know the precise variability without seeking it out as your article notes is suggested for practitioners --

"ERROR

"So I ask, "What is the accuracy of the Harris-Benedict equation when used in predictions?" The answer is quite stark. According to the survey in [1], which analyzed several caloric estimators (Harris-Benedict, Mifflin-St Jeor, Owen, and World Health Organization/Food and Agriculture Organization/United Nations University [WHO/FAO/UNU]), in relation to Harris-Benedict equation:

"45%-81% of estimates are accurate; errors tend to be overestimates, Error range: Maximal underestimation by 23% to overestimation by 42%".

"I hope that the reader sees that the above statistical errors are completely unacceptable. For example, given that a BMR estimate of 2000kcal is an overestimate with an error of 40%, the actual BMR is 0.6*2000 = 1200kcal. The enormous magnitude of error implies that when one is using the estimator, there is a high chance that the estimated BMR is nowhere near the actual BMR. The error is too big and so practically, the estimator is useless. It is pointless to use it in order to predict the BMR of an individual.

"The authors of the survey politely conclude:
"Many dietetics practitioners were taught to use various formulas as the basis for energy predictions. Although these equations are widely cited in textbooks and software, when the data are analyzed, it is clear that the practitioner should become aware of the limitations in the use of these equations."


Yes, the equations are presented often without the margins for error, and notation varies depending on the source. For example the 1931 study referenced in one of my earlier posts uses BMR single standard deviation (about 7%) and "error" (about 5%), as supplied by Harris-Benedict in their 1918 analysis, to compare various BMR calculators.



Edited 1 time(s). Last edit at 10/20/2010 01:29AM by loeve.

Oh my... I feel stoopid! *lol*

Rawseeker... food ready to eat. Peel the nanas and oranges. Measure what you eat and put THAT in the calculator.

Yep, I would not rely always on everything Wikipedia says. Looking up source references can be helpful. "Minor" details can mean a whole lot!

I like this quote:

"As far as the laws of mathematics refer to reality, they are not certain,
and as far as they are certain, they do not refer to reality."
Albert Einstein

Now, to answer the question
"What's the correct way to get an accurate calorie count?"

I would say, if you want to be really accurate, then the analysis of the food in the lab would be required, but for that one would need to give up the food...


RawGosia channel
RawGosia streams

"Yep, I would not rely always on everything Wikipedia says."

It's like trying to solve a puzzle.

Nothing against Wikipedia, they gave a live link to a so-so 2005 study comparing different RMR methods, "so-so" because it had just one fair/good (Frankenfield, 2003, 83 subjects) RMR study on normal weight subjects using the Mifflin-St Jeor method, compared to a number of relevant and good studies on the Harris-Benedict method on many more subjects, where greater minimal and maximal errors would be expected when cherry picking from several studies.

"...The panel found that in healthy nonobese and obese individuals, the performance of the commonly applied Harris-Benedict equation is surpased by the Mifflin-St Jeor equation in terms of both accuracy rate and lower magnitude of error. However, the Mifflin-St Jeor equation still carries a clinically relevant error rate (20%) relative to actual measurement of RMR that cannot be distinguished without measurement." (page 787)

[www.adaevidencelibrary.com]

So, 80% of the time the Mifflin-St Jeor formula is as good as common clinical calorimetry (consistent with calorimetry mearsurement errors of 5% or less (Frankenfield, p. 778)). The Wikipedia article then suggests to me that 75% of the time the Harris-Bennedict formula is as good as calorimetry, fair enough considering the convoluted way in which their reference, Frankenfield presents the data. Add a bit of possible conflict of interest/bias where Frankenfield, 2005 is here reviewing Frankenfield, 2003...

All these online nutrition calculators like fitday and nutritiondata.com must use formulas like the Mifflin-St Jeor and Harris-Benedict? What else? So it looks to me like they give a fair approximation most of the time, though off by a clinically significant margin about 25% of the time.

Just another way of looking at it.



Edited 2 time(s). Last edit at 10/20/2010 01:11PM by loeve.

75% or 80% is not good enough. Nice quotes.


RawGosia channel
RawGosia streams

There is reduced metabolizable energy in a high fiber diet and especially if it is also low in fat. So if you are eating a great deal of unprocessed f+v but not a lot of fat, you are probably overestimating your intake by at least 15%. If you are eating lots of fat, you might be right on target or underestimating. If you desire accuracy, the most critical measures are for the fatty foods which have the highest calorie density and also the lowest fiber content aside from juicing. One guy online didn't understand why he was still underweight, he was weighing and measuring all his food, everything in his calcs was according to his expensive software, he even weighed and measured all of his stools and brought everything to the lab for processing, and the culprit was the very high fiber content of his diet. A calorie in the bomb calorimeter is not as available to the bloodstream if it is tightly bound to a fibrous matrix in the food.

a paper without too many significant figures:
[jn.nutrition.org]

Good point, arugula; I'd forgotten about that!