Dear Editor,
We read with interest and appreciation the article by Hetzler et al, “A comparison of bioelectrical impedance and skinfold measurements in determining minimum wrestling weights in high school wrestlers” ( J Athl Train. 2006;46–51). Although we applaud the authors on a well-done and important research endeavor, we would like to emphasize some concerns relative to applying their results to certifying minimum safe competition weights in wrestlers.
The question being asked was whether skinfold (SF) and bioelectrical impedance (BIA) methods can be used interchangeably for the prediction of minimum wrestling weight (MWW) in high school wrestlers. Skinfold measurements (3-site Lohman equation for males, 4-site Jackson equation for females) and BIA reactance/resistance measurements (RJL Quantum II instrument, hand-foot technique) were obtained for all participants in the same session. Euhydration was assured by eliminating data from any subject with a urine specific gravity >1.020. Percentage of fat was estimated by (1) converting SF sum to body density using the appropriate equation and converting density to percentage of body fat using the Brozek equation and (2) using 4 equations to convert reactance/resistance values to percentage of body fat. Each equation had been normalized previously to specific populations. From percentage of body fat, predicted MWW was calculated under the standard assumption that lean body mass does not change during the period of modification. The authors used an analytical technique not commonly applied for comparison of body composition data: plotting the data according to the Bland-Altman technique (see their Figures 1 through 8) rather than the more common relational scatter plot. In the authors' technique, the relationship between the difference of the 2 methods and the predicted MWW is shown. They then calculated a “total error” for the procedure. In general, if one looks only at the averages of the MWW values, little difference is evident between SF and BIA values in both male and female wrestlers. However, when predicting a safe amount of weight for the individual wrestler to lose over the season, fairness in wrestling weight classification and, more importantly, the wrestler's eventual health depend on individual values, not average values. It is therefore important to look at the scatter of the data, not the average values. For males, the mean prediction error ranged from 3.66 kg (8.1 lb) to 1.51 kg (3.3 lb). For females, this range was 9.16 kg (20.2 lb) to 0.27 kg (0.6 lb). The implications of these findings are that (1) SF and BIA data cannot be used interchangeably in predicting MWW in high school wrestlers; (2) choice of the prediction equation is critical in using BIA data, and no single equation properly serves all wrestlers; and (3) use of BIA data rather than SF data leads to large MWW discrepancies for many wrestlers and can, at best, result in unfair weight classifications and, at worst, lead to dangerously overestimating the amount of weight the wrestler can safely lose.
A total of 208 high school wrestlers (151 males, 57 females, aged 13 to 18 years) from Oahu, HI, public schools participated in this study. The wrestler subject pool presumably consisted of Hawaiian adolescents. The ethnic breakdown of the participants was not included in the methods. We know that the distribution of total fat as adipose fat is different in adults of Polynesian descent than in other racially diverse adults and white adults. Can we also expect this difference in total fat distribution among racial groups to be present in adolescents? What fraction of the subject pool was of Polynesian descent? This factor does not seem to have been addressed in the handling of the data.
The data (from these authors and from previous researchers) clearly indicate that BIA, if used carefully, can be a helpful assessment tool for relative body composition assessment. But it should not be accepted as a method for attempting the absolute assessment of body composition. This is particularly true for foot-foot technology, as used by Tanita, Taylor, and other “bathroom”-type scales. Yet hand-foot instruments are also affected by similar error sources. The manufacturers of BIA equipment, and Tanita in particular, argue strongly against this conclusion, but their contentions are all argued from pooled data and do not address the errors generated when trying to apply single measurements to individuals.
Although the mean prediction error was in the range of 5 to 10 lb between the SF and BIA methods, significant numbers of wrestlers are outside this range. In the lower weight classes, the difference from one weight class to another is approximately 2 to 3 kg. If the BIA MWW for an individual wrestler is outside the ±2 SD range, it becomes possible to classify the athlete in any of 5 weight classes, depending on the direction of the error. This is a huge and unacceptable disparity. The Bland-Altman plots in this study indicate that the magnitude of the difference in predicted MWW among methods is not a function of the size of the athlete; if it were, the differences would taper outward toward the right side of the graphs. Thus, the expected error (in absolute lb or kg) is as large for lighter wrestlers as it is for heavier ones. A 5-lb error in assessing how much weight a 112-lb wrestler can safely lose has a much greater effect than the same 5-lb error applied to a 225-lb wrestler, so this reality is a significant implication of the authors' results. In addition, the distinction between absolute and relative measurements of body composition was not made clear in the article. The question of fairness may be addressed on a relative basis by ensuring that only one method (SF or BIA) is used for determining MWW. The issue of safety for an individual wrestler's weight loss calculation depends on the accuracy of an absolute measurement of lean body mass.
The SF values in this study were replicated by having 3 certified athletic trainers measure each athlete. The intertester validity (test objectivity) for SF measurements is notoriously poor, and it requires experience to learn how to take SF measurements with high reliability. The degree to which one can ascribe absolute accuracy to single-assessor measurements by incidental assessors has not yet been well documented. Many state organizations are strongly opposed to “central” assessment, forcing the assessment into the hands of certified athletic trainers, school nurses, or school administrators who do not perform these procedures routinely and (perhaps) do not maintain the necessary skills to perform them accurately.
One last comment should be made regarding the prediction of a safe amount of weight loss for any individual wrestler. If we allow the use of both SF and BIA in the same state association (at present, the National Wrestling Coaches Association protocols allow both SF and BIA), which value do we use? Wrestlers as a group tend to want to lose as much weight as possible; if this is the case and if we allow use of the highest percentage of body fat, we multiply the chances for error and unsafe weight loss.
We thank you for allowing us to share our thoughts, and we once again congratulate the authors on their research endeavor regarding a very important area of health maintenance with our wrestlers. We encourage further research in this area, and we look forward to the comments of the authors regarding our thoughts.