Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Body Composition and Physical Performance 1992. Pp. 185-193. Washington, D.C. National Academy Press 11 Critique of the Military's Approach to Body Composition Assessment and Evaluation Henry C. Lukaski There are many reasons for assessment and evaluation of body compo- sition of military personnel. One purpose is to provide objective standards for recruitment and retention of personnel. Other purposes include the maintenance of appropriate physical appearance, optimal performance un- der combat conditions and health. Thus, body composition assessment and evaluation are important and necessary to meet the duties and responsibili- ties of the Armed Forces. APPROACHES Because of the large numbers of military personnel that require body composition assessment, any approach must acknowledge and balance the factors of practicality, reliability and accuracy of measurements, time re- quirements, and skill required by the test administrator. These constraints led to the use of weight-for-height tables. Currently, each branch of the Armed Forces uses gender-specific weight-for-height tables both for re- cruitment and retention. Interestingly, the target values are different for recruitment and retention, except for the U.S. Air Force (Table 11-1~. Whether these discrepancies reflect true differences in requirements for physical de- mands or historical precedent is unknown. If an individual fails to meet the weight-for-height guidelines, an evalu- ation of body composition is performed by using either body circumference 185
186 HENRY C. LUKASKI TABLE 11-1 Weight Standards for Recruitment and Retention of a 70-Inch Man in the U.S. Armed Services Body Weight (lb) Recruitment Retention Difference Army 215 192 23 Navy 215 192 23 Marine Corps 211 194 17 Air Force 194 192 2 SOURCE: Adapted from U.S. Department of Defense (1981). measurements (AR 600-9, 1986; Hodgdon and Beckett, 1984a,b; Vogel et al., 1988; Wright et al., 1981) or the combination of skinfold thicknesses and body circumference measurements (Clark, 1976~. Each of these anthro- pometric approaches relies on regression equations to predict percent body fat (BF). As shown in Table 11-2, similar variables (neck and abdominal circumferences) are found in the currently used equations. The estimated percent BF values are then compared to BF standards to determine whether an individual has excess BF. The U.S. Army BF stan- dards are presented in Table 11-3. More stringent Department of Defense TABLE 11-2 Variables Used to Predict Body Composition of U.S. Military Personnel Source Gender of Sample Variables U.S. Air Force Clark (1976) Men Lengths of humerus, radius, acromion, iliac crest, patella and tibia; circumferences of flexed biceps, forearm, chest, waist, buttocks, thigh, and calf; skinfold thickness at triceps, scapula, supra-iliac crest, and calf; and fat density multiplied by number of limbs measured U.S. Navy Wright et al. (1981) Men Neck and abdominal circumferences U.S. Navy Hodgdon and Men Neck and abdominal circumferences; height Beckett ( 1 984a,b) Women Neck, abdominal and hip circumferences; height U.S. Army Vogel et al. (1988) Men and Neck and abdominal circumferences; height Women
BODY COMPOSITION ASSESSMENT AND EVALUATION TABLE 11-3 Maximum Allowable Percent Body Fat Standards in the U.S. Army Age Group (years) 17-20 21-27 28-39 >40 Men Women 20 22 24 26 28 30 32 34 SOURCE: Adapted from AR 600-9 (1986). 187 guidelines indicate goals of 20 percent BF for men and 26 percent BF for women (AR 600-9, 1986~. DISCUSSION The military program of body composition assessment and evaluation is ambitious and very challenging. Any critique of the current program needs to address issues that are philosophical and technical. It is unclear from the available literature whether the military body composition program intends to establish norms and standards for the indi- vidual or for the armed forces as a whole. With the current system of weight-for-height tables, body circumference measurements, and an allow- able increase of 2 percent BF standards per decade of age, it appears that population assessment methods are used for screening, and individual stan- dards are used for evaluation. Thus, the basis for establishing the percent BF norms needs detailed examination and probably revision. Weight-for-height tables have gained considerable use by the civilian American population. To generate national weight standards requires infor- mation on a large group of individuals. One approach was to use data on weight and height from the insurance industry (Society of Actuaries, 1959, 1980a,b). Although these surveys supply data on weight and height for nearly 5 million people, they suffer from the extreme bias of self-selection. A second data base has been generated by the National Center for Health Statistics (Abraham et al., 1983), which developed weight standards for height by plotting the normal distribution of weight-for-height. This distri- bution was arbitrarily divided into overweight and severely overweight. Overweight was defined as those persons exceeding the eighty-fifth percen- tile of weight-for-height and used as a reference the weights of men and women between 20 and 29 years of age. Severely overweight was consid- ered as greater than the ninety-fifth percentile. The major drawback of this approach is that the standard may change as the weight distribution of the population changes.
88 HENRY C. LUKASKI Weight-for-height standards can also be based on the lowest overall risk to health. For example, the minimal death rate in several prospective studies was associated with a body mass index (BMI) of 22 to 25 kg/m2; also, the BMI associated with the lowest risk of death increased with age (Andrea, 19851. A World Health Organization (1987) group suggested that a BMI range of 20 to 30 kg/m2 was associated with a modest risk of mortality. Another approach to defining healthy weights was taken by a Canadian review group (Health Promotion Directorate, 1988~. They labeled as "good weights for most people" the body weights associated with a BMI of 20 to 25 kg/m2. Individuals with a BMI of less than 20 kg/m2, as well as those with a BMI of 25 to 27 kg/m2, were considered to have an increased health risk. Currently available weight-for-height tables do not take into account ethnic or racial differences, morbidity, and mortality in the distribution of weight-for-height. Efforts are in progress to develop race-specific weight- for-height data distributions for Black, Hispanic, and Asian Americans using the limited data available. The bases for the derivation and application of weight-for-height tables in the military need examination. What criteria have been used to establish the tables currently in use? If the tables were constructed from statistical analyses assuming normally distributed weight-for-height data and by using arbitrary cutoff points, the ranges of acceptable weights are biased by changes in the secular distribution of weight-for-height. Furthermore, these estimates may not include any consideration of the criteria of health, ethnic- ity, or performance. The current weight-for-height standards differ for recruitment and re- tention. The differences are large (see Table 11-1) and represent unrealistic goals for weight loss, independent of body composition change, that are attainable during recruit training. It is reasonable to suggest that these differences be resolved. Any attempt to revise weight-for-height tables for military use needs to Include such factors as gender, ethnicity, performance, appearance, and health. Realistic consideration of attainable changes in body weight and body composition during recruit training should be included in deriving weight estimates for recruitment and retention of military personnel. Evidence from the military application of anthropometric approaches to predict densitometrically determined body composition variables indicates that models for predicting percent BE by using either skinfolds and body circumferences (Clark, 1976) or neck and abdominal circumferences (Hodg- don and Beckett, 1984a,b; Vogel et al., 1988) yield biased estimates of body composition. That is, the equations overpredict body fatness for the lean individuals and underpredict fatness for the obese. This bias or error i]
BODY COMPOSITION ASSESSMENT AND EVALUATION 189 may be attributed to either errors in the biological assumptions associated with the densitometric and/or anthropometric methods, technical errors of the measurements or a combination of these two factors. One critical issue for establishing normative standards for the military is ethnic or racial differences in body composition. There is accumulating evidence that distinct differences exist in body composition both within and among ethnic groups, and these observations indicate the need for race- specific standards. To date, this approach has not been adopted, but it appears to be necessary for the development and validation of useful body composition prediction equations. The potential impact of the problem of ethnic or racial differences in body composition is magnified by the use of inadequate reference and can- didate measurements of body composition. Currently, underwater weighing or hydrodensitometry is the reference method used in body composition surveys of military personnel to develop anthropometric models. This ap- proach uses the two-compartment model to assess BF content (Lukaski, 19871. Unfortunately, bone mineral density or content is an unmeasured variable that has the potential to significantly bias the BF estimate. Bone mineral density, which has been shown to be greater in Blacks than in Caucasians (Cohn et al., 1977), greater in men than women (Cohn et al., 1977), and possibly reduced in Asians, has not been measured in any of the previous surveys. Using extrapolations from data on children (Lohman et al., 1984), the estimate of this error can be as high as 5 percent. Thus, failure to correct body density measurements for individual differences in bone mineral density can result in overestimates of BF. With regard to the densitometric equipment used in previous surveys, investigators should modify existing apparatus to perform measurements of residual lung volume while the volunteer is immersed in the water. It is well established that conditions such as obesity are associated with a signif- icant reduction in lung compliance and reduced pulmonary ventilatory ca- pacity (Bray et al., 1977~. The principal ventilatory variable that is reduced is the expiratory residual volume, whether expressed as a whole number or as a fraction of the vital capacity (Bartlett and Buskirk, 1983~. Because this impairment appears to be a continuum over the range of body fatness from lean to obese, it would be prudent to measure residual lung volume rather than estimate it using standard equations or tables. Failure to do so may result in an overestimation of body volume, an underestimation of body density and an overestimation of BF (Lukaski, unpublished observations). The selection of appropriate anthropometric measurements (body cir- cumferences and bone diameters) and skinfold thickness sites is a challeng- ing process. However, the availability of a current reference manual (Loh- man et al., 1988) should be useful. Nevertheless, an important issue is
190 HENRY C. LUKASKI the biological basis for using skinfold thicknesses and anthropometric measurements. Although measurements of bone diameters, limb circumferences, and skinfold thicknesses have been used to derive prediction models for esti- mating body density and percent BE, this approach has generally been limit- ed by population-specific prediction models (Lukaski, 19871. This point was recently reinforced by the findings of Hodgdon and Beckett (1984a,b) and Vogel et al. (1988) with military groups. The limitation of using skinfold thicknesses to predict BE is found in the basic assumptions of this approach. It is generally assumed that the subcutaneous adipose tissue reflects a constant proportion of the total body adipose tissue and hence fat. Also, the sites selected for measurement represent the average thickness of the adipose tissue and thus are the best predictors of BF. Neither of these assumptions has been validated (Lukas- ki, 1987~. Furthermore, the validity of such assumptions is dubious because of the extremes in distribution of body adipose tissue in the population. In addition to the theoretical limitations of using skinfold thicknesses to predict BE, there also exist some practical concerns. The within- and between-observer variability in determining skinfold thickness can be great- er than 5 percent (Burkinshaw et al., 1973; Jackson et al., 1978~. Thus, trained and certified specialists are required. In addition, most prediction equations based on skinfold thicknesses are population specific (Edwards, 1951; Jackson, 1984; Lukaski, 1987~. These factors limit the use of skin- fold thickness measurements for precisely and accurately estimating body composition in the heterogeneous military population. In contrast to the interobserver error in skinfold thickness measure- ments, the measurement of body circumferences is more reliable (Lohman et al., 1988~. Unfortunately, this approach still suffers from population specificity in the development of prediction equations. Statistical approaches for the development of prediction models need some consideration. Using power analysis to assess sample sizes for various racial groups based on estimates of both technical errors of the instrumenta- tion and biological variability in the chemical composition of the fat-free mass (FFM) would enhance the probability of developing valid prediction equations. Furthermore, stepwise multiple regression analysis and factor analysis are needed to describe the most important predictor variables in the model. An appropriate design for cross-validation of the candidate model is also needed. It is necessary to develop the prediction equation in one sample and then to cross-validate it in an independent sample. This ap- proach has been used in previous cross-validation trials of equations de- rived in military personnel (Hodgdon and Beckett, 1984a,b; Vogel et al., 1988). _
BODY COMPOSITION ASSESSMENT AND EVALUATION TABLE 11-4 Variability Estimates for Prediction Models and Cross-Validation Trials for Estimation of Percent Body Fat in U.S. Military Groups Standard Error of the Estimate (percent body fat) Sex of Source Sample Model Validation Hodgdon and Beckett (1984a,b) Men 3.52 2.7* Women 3.72 4.36* Vogel et al. (1988) Men 4.02 3.7* Women 3.60 4.4 *Statistically significant (p < 0.05) difference between predicted and measured values. 191 Another statistical analysis that would be appropriate is a determination of the directional bias of the error relative to the magnitude of the measured and predicted variable. This approach for cross-validation of values whose accuracy is unknown was proposed by Bland and Altman (1986~. It in- volves the graphical representation of the residual scores plotted against the mean of the measured and predicted values. This is the appropriate statisti- cal approach for cross-validation of the derived model. The variability of the distribution of the relationship between measured and predicted percent BF values from the military trials using neck and abdominal circumference measurements and height is summarized in Table 1 1-4. It is clear that the standard errors of the estimate of percent BF are quite large and exceed the theoretical precision of the densitometric method (Lohman, 1981~. These data indicate that the models are adequate for assessments of percent BF in population groups but are inadequate for individuals. CONCLUSIONS AND RECOMMENDATIONS Presently, the available anthropometric equations for estimating percent BF in the U.S. military are not valid for assessing body composition of individuals. This conclusion may be due to technical errors in the densito- metric method, differences in the chemical composition of the fat-free body, the lack of specificity of the anthropometric measurements used in the pre- diction model, or a combination of these factors. In retrospect, the major limitation of using regression equations to pre- dict human body composition is the reliance on a mathematical equation derived in one group to predict a variable in another individual who may be
92 lIENRY C. LUKASKI a member of another group. This approach is susceptible to factors that can adversely influence its validity for estimating body composition of the indi- vidual. If factors such as ethnic differences in the bone density and BE distribution can be assessed and improvements in the technical measure- ment of the body can be made, perhaps an improved and more sensitive assessment and evaluation of body composition in the military population can be achieved. These difficulties can be addressed and controlled by the following recommendations. · Use current and technically accurate methods and equipment for den- sitometry and anthropometry. · Use a multicompartmental model of body composition, and include measurements of bone mineral density (regional and total body) to correct apparent whole body density obtained by underwater weighing. · Use appropriate statistical methods to determine appropriate sample sizes for model development and cross-validation. Calculations for sample sizes need to include estimates of technical and biological variability of measurements. . Use stepwise multiple regression analysis and factor analysis to de- velop the prediction model. · Establish the need or lack of need for race-specific prediction models. · Ascertain the validity of the model or models to determine change in body composition after weight loss. · Establish practical and valid criteria for implementing the new modelts) in the U.S. military environment. REFERENCES Abraham, S., M. D. Carroll, M. F. Najjar, and R. Fulwood. 1983. Obese and overweight adults in the United States. Vital and Health Statistics, Series 11, No. 230. DHHS Publication No. (PHS) 83-1680. Hyattsville, Md.: U.S. Department of Health and Human Services. Andres, R. 1985. Mortality and obesity: The rationale for age-specific height-weight tables. Pp. 311-318 in Principles of Geriatric Medicine, R. Andres, E. L. Bierman, and W. R. Hazzard, eds. New York: McGraw-Hill. AR 600-9, 1986. See U.S. Department of the Army. 1986. Barlett, H. L., and R. R. Buskirk. 1983. Body composition and the expiratory reserve volume in lean and obese men and women. Int. J. Obes. 7:339-343. Bland, J. M., and D. G. Altman. 1986. Statistical method for assessing agreement between two methods of clinical measurement. Lancet 1:307-310. Bray, G. A., B. J. Whipp, S. N. Koyal, and K. Wasserman. 1977. Some respiratory and metabolic effect of exercise in moderately obese men. Metabolism 26:403-412. Burkinshaw, L., P. R. N. Jones, and D. W. Krupowicz. 1973. Observer error in skinfold thickness measurements. Human Biol. 45:273-279. Clark, D. A. 1976. A physical model for estimating body fat. U.S. Air Force School of Aero- space Medicine. Report SAM-TR-76-41. Brooks Air Force Base, San Antonio, Tex.
BODY COMPOSITION ASSESSMENT AND EVALUATION 193 Cohn, S. H., C. Abesamis, I. Zanzi, J. F. Aloia, S. Yasumura, and K. J. Ellis. 1977. Body elemental composition: Comparison between black and white adults. Am. J. Physiol. 232:E419-E422. Edwards, D. A. W. 1951. Differences in the distribution of subcutaneous fat with sex and maturity. Clin. Sci. 10:305-315. Health Promotion Directorate, Health Services and Promotion Branch. 1988. Canadian Guide- lines for Healthy Weights. Ottawa: Ministry of National Health. Hodgdon, J. A., and M. B. Beckett. 1984a. Prediction of percent body fat for U.S. Navy men from body circumferences and height. Report No. 84-29. Naval Health Research Center, San Diego, Calif. Hodgdon, J. A., and M. B. Beckett. 1984b. Prediction of percent body fat for U.S. Navy women from body circumferences and height. Report No. 84-29. Naval Health Research Center, San Diego, Calif. Jackson, A. S. 1984. Research design and analysis of data procedures for predicting body density. Med. Sci. Sport Exerc. 16:616-620. Jackson, A. S., M. L. Pollock, and L. R. Guttman. 1978. Intertester reliability of selected skinfold and circumference measurements and percent fat estimates. Res. Q. 49:546-551. Lohman, T. G. 1981. Skinfolds and body density and their relation to body fatness: A review. Hum. Biol. 53:181-225. Lohman, T. G., M. H. Slaughter, R. A. Boileau, J. Bunt, and L. Lussier. 1984. Bone mineral measurements and their relation to body density in children, youth and adults. Hum. Biol. 56:667-679. Lohman, T. G., A. F. Roche, and R. Martorell, eds. 1988. Anthropometric Standardization Reference Manual. Champaign, Ill.: Human Kinetics. Lukaski, H. C. 1987. Methods for the assessment of human body composition: Traditional and new. Am. J. Clin. Nutr. 46:537-556. Society of Actuaries. 1959. Build and Blood Pressure Study, vol. 1. Chicago, Ill.: Society of Actuaries. Society of Actuaries. 1980a. Blood Pressure Study of 1979. Chicago, Ill.: Society of Actuaries/ Association of Life Insurance Medical Directors of America. Society of Actuaries. 1980b. Build Study of 1979. Chicago, Ill.: Society of Actuaries/Associa- tion of Life Insurance Medical Directors of America. U.S. Department of the Army. 1986. Army Regulation 600-9. "The Army Weight Control Program." September 1. Washington, D.C. U.S. Department of Defense. 1981. Directive 1308.1. "Physical Fitness and Weight Control Programs." June 29. Washington, D.C. Vogel, J. A., J. W. Kirkpatrick, P. I. Fitzgerald, J. A. Hodgdon, and E. A. Harmon.1988. Der- ivation of anthropometry based body fat equations for the Army's weight control pro- gram. Technical Report 17-88, AD-A 197 706, U. S. Army Research Institute of Environmental Medicine, Natick, Mass. World Health Organization (WHO). 1987. Measuring obesity: Classification and description of anthropometric data. Report on a WHO Consultation on the Epidemiology of Obesity, Warsaw, October 21-23. Copenhagen, Denmark. Wright, H. F., C. O. Dotson, P. O. Davis. 1981. A simple technique for measurement of percent body fat. U.S. Committee on Military Nutrition Research (CMNR) Med. 72: 23-27.
