Composition Techniques: Hydrostatic Weighing, Skinfold Thickness, and ~ioelectric Impedance

Transcription

1 Field Comparison of ody Composition Techniques: Hydrostatic eighing, Skinfold Thickness, and ~ioelectric Impedance DONALD T. KIRKENDALL, PhD,' JOHN. GROGAN, MS, PTK, PhD,2 RICHARD G. OERS, PhD3 Journal of Orthopaedic & Sports Physical Therapy ody composition and appropriate playing weight are frequently requested by coaches. Numerous methods for estimating these figures are available, and each has its own limitation, be it technical or biological. A comparison of three common methods was madeunderwater weighting (H,O, the criterion), skinfold thicknesses (), and commercial bioelectrical impedance analysis (IA). Subjects were 29 professional football players measured by each of the three methods after an overnight fast. Data was collected 10 weeks preceding the players' formal training camp. There was no difference for percentage of weight as fat between (15.8O/0) and H20 (14.2O/0). ioelectrical impedance analysis significantly (p c.05) overestimated percent fat (19.2O/0) compared to H20. Error rates when regressing on H20 were favorable, whether expressed for the whole sample (3.04"/0) or by race (1.78% or 3.56% for whites and blacks, respectively). Regression of IA on H20 showed an elevated, overall error rate (14.72O0) and elevated error rates for whites (11.57Oh) and blacks (73.87 /0). Of the two estimates of body composition on a racially mixed sample of males, provided the best estimate with the least amount of error. In the application of sports physiology services, one of the most requested procedures is body composition analysis and estimation of appropriate playing weight. Typical methods used to obtain such information include skinfold thickness (), hydrostatic weighing (H20), or newer methods using electrical impedance (IA). The choice of method, in many cases, depends on patient availability, number of patients, equipment availability, setting (laboratory or field; research or individual application), and support help. The measurement of skinfold thickness at various sites is routinely used to estimate body ' Section of Sports Medicine. John A. ergfeld. MD. Head. Cleveland Clinic Foundation. Cleveland. OH. Current address for Dr. Kirkendall is the Physiologtcal Assessment Laboratory, llllnois State University. Normal. IL ' Director. Center for Athletic Performance. Section of Sports Medicine. Cleveland Chic Foundation. Department of Physical Education. owling Green lhiversity. owling Green. OH /91/ $03.00/0 THE JOURNAL OF ORTHOPAED~C AND SPORTS PHYSICAL THERAPY Copyright Q 1991 by The Orthopaedic and Sports Physical Therapy Sectlons of the American Physical Therapy Association composition in a field setting. This technique can have large interinvestigator variability (1 2.1 mm) (2). However, there are numerous, inexpensive calipers on the market shown to be accurate in the hands of an experienced technician (1 3, 17). Many centers and universities routinely use hydrostatic weighing, which is the criterion for other estimation methods. Density calculations for underwater weighing require a correction for lung volume. Methods vary from measurement at a known or estimated (1 8) residual lung volume to data taken at total lung volume (16). Measured residual volume requires pulmonary function laboratory equipment either during or separate from the hydrostatic weighing procedure. Other methodological concerns that may increase variances include the conversion equation from density to percent fat. For example, specific conversion equations are published based on age (7) and race (1 5). Currently, ease of operation, reliability, minimal patient and technician time, and commitment make bioelectric impedance analyzers an attractive method. The validation studies of these tools JOSPT 13.5 May 1991 FIELD ESTIMATION OF ODY COMPOSITION 235

2 of ages (1 759) were studied, yet the mean height (177 cm) and weight (84.9 kg) were narrow enough to exclude many competitive athletes. One must also consider the fact that equations supplied by the manufacturers have not performed well (5, 6, 10, 1 1). Commercial impedance analyzers have been shown to be insensitive at extreme body compositions (10). In addition, racial variations are rarely reported; yet there a p pears to be a need for equations for varying ethnic groups (5, 22). The purpose of this investigation was to compare the results of a bioimpedance analyzer, accepted skinfold estimations, and underwater weighing in a mass screening setting for body composition in a group of men who had significant mass and a mixed racial makeup (professional football players). METHODS Subjects Thirtyfive professional football players visited the clinic for their annual preseason physicals in early May These physicals preceded a spring minicamp and were performed about 10 weeks preceding training camp for the competitive season. A station was set up for performing a variety of evaluations (orthopaedic, medical, dental, body composition, spirometry, flexibility, EKG, blood, TALE 1 Subject frequency.. and means (SD) of descriptive characteristics by race andposition Position N Aae Heiaht eiaht IN I LINt t wt LTE RD O Kicker INT LINE, offensive and defensive lines; LTE, linebackers1 tght ends; RD, wide receiverldefensive back; 06, offensive backs. t, blacks;, whites. eye screen, and others) for a large number of people in a reasonable time frame. The players were tested after an overnight fast. ody composition was estimated from skinfold thicknesses by the threesite (pectoral, umbilicus, thigh) equation of Jackson and Pollack (4). Routine methods for underwater weighing were used. Due to time constraints, residual volume was estimated from vital capacity obtained during clinical spirometry screening. Impedance was measured with the Valhalla system (Valhalla Scientific; San Diego, CA). Each of the procedures was performed by one of the coauthors to minimize interinvestigator variability with each method. Density was converted to percent of fat by the rozek et al method (3). Due to poor technique in underwater weighing, six players were excluded from the analysis, leaving a total of 29 players tested. Results were analyzed by routine descriptive statistics. Differences between methods and groups (race, position) were analyzed by repeated measures analysis of variance. A significant F ratio was followed up by the Scheffe procedure. The data were also subjected to bivariate correlation and linear regression techniques (21). RESULTS Table 1 lists the descriptive statistics of the subjects by race and position. lacks made up 65 T ALE 2 Mean (SD) percent body fat by race and position Position N H.0 I A INT LINE LTE RD O KICKER OVERALL Note: see table 1 for abbreviations. P <.05 IA VS ; IA vs H20 P = NS. vs H,O 236 KIRKENDALL ET AL JOSPT 13:5 May 1991

3 percent of the sample, while 35 percent of the sample were offensive and defensive interior linemen. There were no significant differences in age, height, or weight between races. Some obvious positional differences existed but were not analyzed due to empty cells (Table 1). The results of estimated percentage of weight as fat are listed in Table 2. ecause of empty cells, a full race by position and method analysis was not performed. There were no overall differences due to race nor was there a significant race by method interaction. There was a significant difference due to method, with IA results significantly higher than results of the other two methods. Table 3 lists a correlation matrix of the data by race and a matrix for both races combined. Correlations between skinfold thickness and hydrostatic weighing were in excess of.908 in all three matrices. The IA correlated most highly with the in whites (r =.88)., the correlations of IA with and H20 (Table 3) were quite variable. If two methods give equivalent results, a regression of one on the other would create a eight IA N19 r >.456, ' p <.O5 eight I A r >.632, ' p <.OS eight I A N = 29 r >.367, ' p <.OS slope of 1.0 and an intercept of zero; therefore y = x. Table 4 contains the linear regression analysis results of the three pairs of comparisons. In Table 4A (y = H20, x = ), the results of regression analysis on whites show that the intercepts were indeed quite near zero and the slopes were close to 1.0 with strong correlations. Table 48 shows the relationships of (X) + IA (Y) for blacks. Note that the intercept was nearly 11, with a slope well below 1.0 and a moderate but significant correlation. The error, however, is nearly three times greater than the error for whites, which itself was three times greater than the error from the and H20 analysis (Table 4A). For Table 4C, the major differences of Table 48 are not as visible. The slopes were still below 1.O, the intercepts were elevated, and the errors were among the largest in the sample. DISCUSSION TALE 3 Correlation matrices by race and the overall matrix The intent of the study was to field test an instrument on a group of men whose mass was outside the ranges cited in previous studies. In addition, eight I A lacks '.763' ' hites '.877'.636'.771* '.821*.505'.778' Critical levels for the correlation coefficient from Young and Veldman (21). appendix C, page 663. JOSPT 13.5 May 1991 FIELD ESTIMATION OF ODY COMPOSITION 237

4 TALE 4 Linear regression equations by method and race Race Intercept Slm Eror r A) y = HzO, x = lacks hites ) y = IA, x = lacks hites c) y = IA, x = lacks hites the authors hoped to obtain information based on a typical application of the IA technique. Reliability and validity of the instruments has been reported to be respectable (14). In these studies, a wide range of ages was reported, but the upper extremes in weight and height were limited. At first glance, the current percent of fat data might appear above that expected for professional football players. However, these results are not appreciably different from the frequently cited studies of ilmore and Haskell (20) and ilmore et al (19). In these two projects, the time frames within the competitive season were not stated. The present study was performed about 10 weeks prior to training camp and 15 weeks prior to the season. It is entirely likely that more impressive results would not occur until closer to or during the season. One obvious finding is the overprediction of percent of fat by IA in this sample. ioelectrical impedance analysis has been reported to overpredict percentage of weight as fat in lean subjects and underpredict it in subjects carrying excess fat (1 0). The other obvious finding is that body fat in blacks predicted from IA resulted in greater variability from the criterion H20. In addition, the error terms for predicting percent of fat from IA were larger than would be expected. ioelectrical impedence analysis has been reported as a reasonable alternative to skinfold tests. If so, then the results of IA versus the other methods should be similar. In this unique sample, IA overpredicted percent fat as estimated by and H20 (p <.05). This is consistent with the literature (5, 10). In addition, within the black race, results of regression analysis were poorer than that for whites. Schutte et al(15) have suggested that the fat free body of blacks is denser than whites; therefore, a different conversion equation from density to percent fat is needed. The factors that make up the differences between races (increased bone and muscle density, equal total body water) ap pear to have had an influence on these results as measured by skinfolds. However, because a similar trend was not evident between H20 and IA, one might speculate as to technique variation. Yet there was no technical crossover among the investigators or between races. One possible reason for the added error and displaced regression line of H20 on IA (Table 4C) could be due to using estimated residual volume (RV) instead of measured RV. Due to time constraints and numbers of players, the authors did not measure RV. It has been reported that variance in RV measurement technique adds to the error in body composition (12). Errors ranging from.2 (18) to 2 (12) absolute points in percentage of weight as fat in athletes have been reported. One might also consider ilmore's (18) suggestionwhen body fat is to be used for diagnostic or classification purposes, method of residual volume measurement matters little. The current commercial IA equations result in a greater error of estimation than that found in estimation equations using skinfold thickness measurements. The increased error found when using IA to measure blacks might indicate that current commercial IA equations produce greatest error when measuring black subjects. Future directions for IA study should focus on improved generalized equations using the impedance figure, singularly or in combination with other anthropometric measurements, to predict body composition. Results of such studies should allow for a broader application of the technique. SUMMARY For percent of weight as fat, there was no significant difference between the skinfold () and hydrostatic weighing (H20) methods. The bioelectric impedance analysis (IA) method overestimated percent fat compared to the H20 method. Using the racially mixed sample, the method provided the best estimate of fat with the least amount of error. 0 The authors express their appreciation to Dr. John A. ergfeld. MD, head of the Section of Sports Medicine. Cleveland Clinic Foundation, and team physcian of the Cleveland rowns, and ill Tessendorf. ATC. head trainer of the Cleveland rowns, for access to the study's players. The authors also thank the Cleveland rowns' players who participated in the study. REFERENCES 1. aumgartner RN. Chumlea C. Roche AF: i i impedance for body composition. Exerc Sport Sci Rev 18:193224, ouchard C: Reproducibtlity of body composit~on and adipose tissue measurements in humans. In: Roche AF (ed). ody Compos~tion Assessments in Youth and AdultsReport of the Sixth Ross Conference on Medical Research, pp 913. Columbus. OH: Ross Laboratones rozek J. Grande F, Anderson JT. Keys A: Densitometric analysis of body compos~tii: Revision of some quantitative assumptions. Ann NY Acad SCI llo:ll3l4o.l KIRKENDALL ET AL JOSPT 13.5 May 1991

5 4. Jackson AS. Pollack ML: Generalized equations for predicting body density of men. r J Nutr 40: Katch FI. Keller. Soloman R: Vahdity of IA for estimating body fat In cardiac and pulmonary patients, black and white men and women match for age and body fat. Med Sci Sport Exerc 18:S Lawlor MR, Crisman RP, Hodgdon JA: i impedance analysis as a method to assess body composition. Med Sci Sports Exerc 17: Lohman TG, o~leau RA. Slaughter MH: ody composition in ch~ldren and youth. Adv Pediat Sport Sci 1: Lukaski HC. olmhuk. Hall C. Siiers A: Validation of tetrapdar bilectrical impedance method to assess human body composition. J Appl Phys1ol60: Lukask~ HC. Johnson PE. olonchuk. Lykken GI: Assessment of fatfree mass using bioelectric irnpedance measurements of the human body. Am J Clin Nutr 41: Miles DS. Stevens AG: ody composltlon measured by biodectric irnpedance and underwater we~ghing. Med Sci Sports Exerc 17: Miranda E. Lombard VP, Troxdl RK. Menon J: Percentage body fat measurement via bioektric impedance, hydrostatic weighing. and skmfold assessment. Med Sci Sports Exerc 18:S Morrow JR. Jackson AS. radley P. Hartung GH: Accuracy of measured and predicted res~dualung volume on body density measurement. Med Scl Sport Exerc 18:647652, Reed RM. Moss RF. Jensen DG. Kirkendall DT. ~lson PK: An thropometric determination of body density in prepubescent males. aged 71 1 (abstract). Med Sci Sports Exerc 12: Segal KR. Gutln. Presta E. ang J, van ltallie T: Estlmation of human body composit~on by electrical impedance methods: a comparative study. J Appl Physiol 58: Schutte JE. Townsend EJ. Hugg J. Malina R: Density of lean body mass IS greater In blacks than whites. J Appl Physiol 56: Timson F. Coffman JL: ody composition by hydrostatic weighing at total lung capacity and residual lung volume. Med Sci Sport Exerc 16: Vlasek I. Hartung GH: Analysis of the Fat0Meter skinfold callper for body composition field testing. Med Sci Sports Exerc 12: 103 (abst) llmore JH: The use of actual predicted and constant residual volumes in the assessment of body composition by underwater we~ghmg. Med Sci Sports 1: llmore JH. Pan R. Haskell L. Cost~ll DL. Milbum. Kerlan RK: Football pros strengths and weaknesses charted. Phys Sportsmed 4(10): llmore JH. Haskell L: ody composition and endurance capacity of professional football players. J Appl Physiol , Young RK. Veldman DJ: Introductory Statistics for the ehavioral Sciences. 4th ed. New York: Holt. Rinehart and inston Zill~kens MC. Conway JM: Estimation of lean body mass in black adults by total body Impedance. Fed Proc 46: Journal of Orthopaedic & Sports Physical Therapy JOSPT 13.5 May 1991 FIELD ESTIMATION OF ODY COMPOSITION