| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
LETTER TO THE EDITOR |
Numedloc, Inc.
Webster City, IA
I am troubled by the paper by Tverskaya [1] on predicting basal metabolic rate (BMR) in obese children. Comments on some of the issues appear warranted.
The authors clearly state that it is inappropriate to use predictive equations in populations different from which they were derived. This is a basic statistical principle. But then inappropriate comparisons are made repeatedly by comparing data from their study population with other dissimilar populations.
The rather brief description of the indirect calorimetry procedure suggested that the study actually measured resting energy expenditure (REE) as opposed to BMR. One would hypothesize a 10% difference between REE and BMR measurements. The process used for defining the "measurement period" while a child was being tested wasn’t defined nor was the process used to calculate energy expenditure from oxygen consumption and carbon dioxide production. Presumably, protein oxidation was ignored in calculating 24 hour "BMR."
Studies over the decades have consistently shown that height, weight, age and sex are important parameters in predicting REE; and easily measured by all clinical staff regardless of location and resources. Hence, it would be interesting to know if FFM or FM significantly improved predictive performance, above and beyond these basic anthropometric parameters.
Because sex is a discontinuous variable, most investigators chose to develop predictive equations by sex, as opposed to testing for sex based interactions. The presented data does not allow one to assess the validity of a single formula. In addition, the data provided covers the 3 to 18 year age span but the predictive formula is limited to 6 to 18. Why? Also, the reason for splitting data at the 10 and under vs 11 and older is not clear. Growth curves and sex based development profiles suggest that other approaches may be more appropriate—for example, unisex 3 to 5; unisex 6 to 12; and M/F 13 to 18.
The statistical analysis suggests that fat-free mass and fat mass were independently acquired measurements since issues such as interactions were not addressed. However, the methods indicate that they were both derived from a single frequency impedance measurement (which presumably factored in anthropometric data such as height, age, weight and sex). In addition, a casual reading would suggest that for this article FM+FFM=WT. Does it?
My initial response to reading the Methods/Measurements section was to congratulate the authors on presenting independent data on the precision and accuracy of their indirect calorimetry measurements. Unfortunately, instrument differences do exist, and operator differences exist both in calibration and in their skill in performing this clinical procedure. An independent assessment of the precision and accuracy of the FFM and FM (or at least impedance measurement) would also have been appreciated.
The paper masks a interesting investigation which could answer clinically important issues such as: a) Can one accurately predict "resting" energy expenditure in obese children using basic anthropometric data? or b) Does the addition of a single body impedance measurement significantly improve predictive ability?
Presumably, REE data was collected on children prior to being placed on a weight control regime. It would be very interesting to know the ability of a derived formula to predict REE 4 to 7 days after placing a child on a calorie restricted diet (but before significant fat or proteil loss). That is, can we predict an individual’s metabolic adaptive response to a diet? What are the limits to use of predictive equations?
REFERENCES
Research Institute
Miami Childrens Hospital
Author’s Reply
CLARIFICATION OF THE METHOD FOR CALCULATING BMR
We appreciate your letter sent to us on August 11, 1998 regarding some comments on the paper recently published by Tverskaya et al in the Journal of the American College of Nutrition. In regards to the inappropriate comparisons, we made these comparisons just for illustrative purposes. We wanted to show that some of the published prediction equations, such as ones we tested, were not appropriate for use in obese children. Our goal was to prove that inappropriate use of prediction equations leads to potential errors in BMR.
The indirect calorimetry method was explained well. We referenced the validation study (ref #4) for the Deltatrac metabolic monitor. The operator of the system was well trained and observed by me during all tests. Furthermore, we mentioned that our BMR tests were "preprandial, early in the morning." Our patients were fasted overnight and "BMR" and not "REE" were measured. We express basal metabolic rate "BMR" throughout the manuscript.
Not accounting for protein oxidation will cause an error of about 2% in 24-hour BMR (Ravussin et al, J Clin Invest 1986;78:1568–1578). The error of indirect calorimetry is closer to 5%. Furthermore, to account for this error, it is necessary to collect 24 hour urine for nitrogen analysis. Our patients only came in the morning for testing and it was not feasible to obtain 24 hour urine collections. This would be especially difficult in the younger patients.
Body weight will be a major contributor to metabolic rate in any derivation. This is due to body weight being mainly composed of fat-free mass. Most of this is muscle which is the single major contributor to metabolic rate. However, by separating out the components of body composition only serves to improve the predictability of metabolic rate and reduce the amount of unaccounted for variation. For example, we derived another equation without fat-free mass and fat mass. Age, height, weight and sex were entered into the model. The R2 was only 0.70 as opposed to an R2 of 0.84, obtained with the addition of fat-free mass and fat mass. Using fat-free mass and fat mass in our equation accounted for another 14% of the variability of BMR. Fat-free mass and fat mass were found to be significant predictors of BMR in our equation along with age and sex. We did not include any covariates that were not significant predictors of BMR.
We did not separate the sexes because this would reduce the number of patients in each individual equation. There is no change in the final result whether there are two separate equations or a single equation with sex as a covariate. It makes sense that sex will be significant predictor of BMR because metabolic rate changes with puberty in girls. This is due to the two phases of the menstrual cycle (Ferraro et al. J Clin Invest 1992;90:780–784). This is why sex is an important covariate in any derivation.
The age breakdowns were based on the ones used by WHO. We will admit to a subtle error in the manuscript. Our youngest patients were 6 years old. However, the "3" in the physical characteristics table is an error. We do agree that using the growth profile curves may be a better way to divide up our patient age groups.
We agree that fat-free mass and fat mass equal weight when fat-free mass is obtained by bioelectrical resistance. Fat-free mass is determined from the calculation of total body water by the resistance of a 800 microamp current imposed by the bioelectrical resistance instrument. A prediction equation is used to calculate fat-free mass. However, it has been determined that muscle is very close to 73% water in young children and adults. The prediction equation we used to calculate fat-free mass included patients of similar backgrounds to our own. These equations have been refined over the years and include thousands of subjects in their derivation.
You suggest some interesting studies. At this point we do not know if our new prediction equation will appropriately estimate BMR in children just beginning a weight control program. Usually, prediction equations have to be used in populations for which they were derived. Going outside the "scope" of an equation will lead to potential errors.
I hope we addressed some of your concerns about the manuscript. You bring up a few good points that can only help to improve future metabolic studies.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
