Introduction Predicting body mass from skeletal frame size is important in disciplines including forensics, physical anthropology and archaeology, where partial skeletal remains can be used to reconstruct a complete person. Dimensions of athletes are of particular relevance as their physiques "more closely approximate the probable conditioning of earlier humans" (Ruff, 2000). Such estimates produce reasonable predictions of mass with few predictor variables (e.g., stature and bicrystal breadth), but inevitably fail to address the variation in skeletal proportion between individuals which would make estimates more accurate in less typical individuals. In living humans, unlimited skeletal measurements are possible, and the purpose of this study was to produce a prediction of total body mass based on multiple skeletal dimensions of a large sample of athletes and controls. Methods This study examined the prediction of scale mass via anthropometric skeletal measurements in 536 elite Caucasian athletes and controls measured by criterion anthropometrists from Australia, New Zealand and the UK. All participants were measured for anthropometric profiles according to the methods of Marfell Jones et al. (2006). A total of 18 skeletal lengths and breadths were measured and log transformed to enable resulting regressions to be biologically plausible because they constrain a zero stature to a zero mass. These, along with gender were entered into a best subsets exploratory regression, using Ln mass as the dependent variable. Results Based on exploratory regression results, prediction was optimised as follows: Ln mass = 0.6071(ln head girth) + 0.65303(ln femur breadth) + 0.65062(ln ankle girth) +0.26724(ln chest breadth) +0.17492(ln A-P chest depth) +0.32479(ln femur length) +0.31324(biacromial breadth) - 5.4611. Standard error of the estimate = 8.3% , R2 (adjusted) = 84.5%, P<0.0001. Significant correlations were found between all log-transformed dimensional measures and mass, ranging from 0.42 - 0.81 at femur length and femur breadth respectively. Discussion / Conclusions Due to the high correlation between skeletal dimensions and body mass, any one of the variables in question could be used to produce a useful prediction of body mass, justifying the approach of Ruff (2000). In the regression analysis, gender was not significant once shoulder breadth had been input, suggesting that shoulder dimensions explained the variability due to gender, and predicted mass more closely. These results have broadly similar error to the equations used the study of Ruff (2000) whose predictions used only stature and bi-iliac breadth based on 56 population and sex-specific sample means and applied to other groups. His work noted that the prediction errors were considerably greater with male US marine recruits whose mass was underestimated by 9%. This effect is not altogether surprising as the likelihood of extra muscle resulting from conditioning programmes adding weight, but not altering the skeletal structure appreciably, is high. The exponents of the model of the present study total 2.99, almost exactly as predicted from geometric similarity scaling. This finding is interesting, because we have previously demonstrated that inflated mass exponents of girths at the thigh and calf (corrected for overlying adipose tissue) reflect Newton`s law of acceleration, where a larger mass requires disproportionately larger and stronger muscles to accelerate and decelerate the body (Nevill et al., 2004). With the large variation of physiques of participating athletes, muscle mass variability is the probable cause of the discrepancy from the mass estimates from skeletal measures. The authors offer these new mass-prediction equations to augment previous work where skeletal remains can be used to estimate mass.
© Copyright 2008 2008 International Convention on Science, Education and Medicine in Sport: Proceedings, Vol. II. Veröffentlicht von People´s Sports Publishing House. Alle Rechte vorbehalten.