Anthropometry and Body Composition - Essay Example

Anthropometry and Body Composition Anthropometric measurements are important constructs to assess nutritional status. Among the available constructs body mass index (BMI), body fat (%) and total fat mass are considered. These variables are defined using height, weight and circumferences of various parts of the body, such as waist, hip, neck and mid-upper arm. To develop a universally applicable construct, several proxy variables for height are considered in the literature. A gender-specific validation study is performed and it is seen that none of them are performing well in predicting height, especially for females. There is significant difference between the heights actually measured and heights predicted from the proxy variables. As BMI is a function of height, BMI constructed using the proxy heights also differ significantly from BMI computed using measured heights. Fat amount measured from skinfolds has been used to get an estimate of body fat. Two different calipers are in use to measure skinfold fat. It is observed that measures available from the two instruments differ significantly. Finally interdependence among various variables is studied. Using recommended cut-offs contingency chi-squares are computed. They are found significant, indicating neck circumference may be taken as a measure to assess obesity. Introduction: Nutritional status of an organism is the result of a complex relationship between consumption of nutrients and their expenditure in activities. Nutritional status has been established in the literature as an indicator of health (WHO Technical Report Series, No 797, 1990; The National Diet and Nutrition Survey, 2004). Thus accurate measurement of nutritional status is important to assess an individual’s well-being. Four measures are commonly employed to assess nutritional status, (Gibson, 2005; Lee & Nieman, 2007) of which anthropometric measurements are non-invasive, require least time and cost and place comparatively less burden on the respondents. Anthropometric measurements are an important screening tool to assess whether an individual is underweight, overweight or obese (MUST Tool; WHO Technical Report, 1995). One of the limitations of anthropometric measurements is that they require to be validated against a population specific standard. Another limitation is that not all anthropometric measurements have high degree of sensitivity and specificity (Gorstein et al., 1994). However, benefits of using anthropometric measures outweigh their limitations. Accuracy of anthropometric constructs depends on the experience of the assessor in identifying correct sites and proper use of the tools. The aim of this study is to gain experience in performing measurements as well as study their inter-dependence. Method and Material: The sample consists of 204 subjects. More than 20 different variables are measured on each of them, many of them in duplicate. Measurements include height, knee-height, demi-span (tip and web) and ulna-length; weight and several surrogate measurements of weight such as circumferences of waist, hip, neck and mid-upper arm; and skinfolds at 4 different sites. Weighing scales and stadiometer were used to measure a person’s weight and height. Slim Guide Caliper and Fat Track Caliper (digital) were used to measure skinfold thickness. Ordinary measuring tape was used to measure knee-height, demi-spans, ulna-length and all circumferences. For exact locations and methods of taking measurements refer to Practical Handout (pages 8-23). For final analysis average of the duplicate measurements was used. Other than the recorded raw measurements a number of standard anthropometric constructs was used in the study, e.g. BMI, waist-hip ratio, body density, and body fat percentage. All the formulae used for computation are available in the Practical Handout (pages 8-23). The aim of this exercise is to determine to what extent anthropometric measures are good for predicting body composition. Data are analyzed using SPSS statistical software. Since body composition varies between males and females, gender-specific analysis is done. For model building regression analysis has been used. To determine whether group means differ, one-way ANOVA has been applied. To assess dependency of categorized variables, contingency chi-square tests have been used. All the statistical tests are done at 5% significance level. The main features of the data and results of the data analyses have been presented in the following section. Results: In the sample of 204 subjects there are 126 females and 78 males. A comparative study of the main variables describing body compositions shows significant gender-specific differences1 (Table 1). Table 1: Comparison of body compositions by gender (mean ± s.d.) Variable Female Male Height * 1.63 ± 0.0 1.81 ± 0.1 Weight * 63.60 ± 11.3 73.90 ± 8.2 BMI * 23.81 ± 4.1 22.72 ± 2.4 Waist-Hip Ratio * 0.76 ± 0.1 0.83 ± 0.0 Body Density * 1.04 ± 0.0 1.07 ± 0.0 Body Fat * 26.81 ± 5.4 10.79 ± 5.2 Total Fat Mass * 17.28 ± 5.8 7.80 ± 3.8 Total Fat-free Mass * 46.32 ± 7.9 65.91 ± 7.9 Several proxy variables, such as knee height, demi-spans and ulna length, are commonly used to estimate a person’s height (Hogan 1999; Weinbrenner, Vioque, Barber & Asensio, 2006). However, in our sample the differences between measured heights and estimated heights are significant in all cases for females and in two cases for males (Table 2A). Since regression equations are used to predict height from knee-height (Practical Handout p 10) we have looked at the adequacy of regression also. All regression equations are significant, implying height is dependent on the proxy measures. However, in case of male population regression seems to be performing better than females as is evident from the coefficients of determination (Table 2A). For both genders demi-span tip is the best predictor of height among these four. Table 2A: Adequacy of surrogate measures for predicting heights Predictor Mean ± S.D. Regression Equation (Females) R2 Knee height * 1.61 ± 0.05 Av Ht = 0.789 + 0.0169 knee height 44.7% Demi-span tip * 1.71 ± 0.06 Av Ht = 0.854 + 0.009 demi-span tip 46.5% Demi-span web * 1.62 ± 0.07 Av Ht = 0.998 + 0.008 demi-span web 42.5% Ulna length * 1.66 ± 0.04 Av Ht = 1.01 + 0.0245 ulna length 34.3% Regression Equation (Males) Knee height 1.80 ± 0.07 Av Ht = 0.221 + 0.0275 knee height 86.4% Demi-span tip * 1.91 ± 0.07 Av Ht = -0.009 + 0.0192 demi-span tip 91.2% Demi-span web 1.80 ± 0.07 Av Ht = 0.034 + 0.0206 demi-span web 80.3% Ulna length * 1.83 ± 0.07 Av Ht = 0.278 + 0.0529 ulna length 69.1% BMI computed using the above mentioned proxy measures differ significantly from BMI computed using height. Here also demi-span tip works best among the four alternatives, for both males and females (Table 2B). The regression equations also perform better than those in Table 2A. Table 2B: Comparison of BMI computed in different ways Surrogate BMI Mean ± S.D. Regression Equation (Female) R2 Knee height * 24.33 ± 4.09 BMI = 1.19 + 0.930 BMI knee height 86.4% Demi-span tip * 21.81 ± 4.28 BMI = 4.07 + 0.905 BMI demi-span tip 89.8% Demi-span web * 24.45 ± 4.92 BMI = 4.85 + 0.776 BMI demi-span web 87.1% Ulna length * 22.91 ± 3.84 BMI = 1.30 + 0.983 BMI ulna 85.2% Regression Equation (Male) Knee height 22.72 ± 1.73 BMI = -3.44 + 1.15 BMI knee height 70.9% Demi-span tip * 20.26 ± 1.49 BMI = -5.08 + 1.37 BMI demi-span tip 74.7% Demi-span web 22.77 ± 1.57 BMI = -4.33 + 1.19 BMI demi-span web 61.7% Ulna length * 21.96 ± 1.62 BMI = 0.53 + 1.01 BMI ulna 47.8% Recall that all the measurements were taken in duplicate to increase accuracy although the above analyses were performed using the average of the two sets of observations. To assess inter-measurement disagreement the duplicate sets are compared. Paired-t test was applied and four of the variables (Weight, Ulna length, Demi-span tip and Neck circumference) were found to differ significantly at 5% level (Table 3). In case of skin-fold measurements, different results are observed depending on the caliper. With Slim Guide Caliper subscapular and suprailiac skinfolds are significantly different in the duplication2 while biceps and suprailiac are significantly different when measured with Fat Track calipers3. Table 3: Variability in duplicate measurements Variable Measure1 Measure2 Variable Measure1 Measure2 Weight * 67.52 ± 11.6 67.55 ± 11.6 Demi-span web 80.27 ± 6.8 80.26 ± 6.8 Height 1.70 ± 0.1 1.79 ± 0.1 Waist circ 73.93 ± 8.9 73.90 ± 8.7 Knee ht 53.02 ± 4.8 52.98 ± 4.7 Hip circ 94.06 ± 8.5 93.95 ± 8.6 Ulna length * 26.81 ± 2.2 26.90 ± 2.3 Mid-upper arm circ 28.15 ± 3.9 28.14 ± 3.9 Demi-span tip * 87.88 ± 7.2 87.98 ± 7.2 Neck circ * 33.54 ± 3.8 33.62 ± 3.8 Table 4: Variability in duplicate skinfolds Variable Slim Guide Caliper Fat Track Caliper Measure1 Measure2 Measure1 Measure2 Triceps 11.08 ± 5.8 11.17 ± 6.0 10.46 ± 7.0 10.33 ± 7.3 Biceps ! 9.57 ± 6.47 9.57 ± 6.5 7.86 ± 6.5 8.05 ± 6.4 Subscapulars * 12.31 ± 6.6 12.20 ± 6.6 10.81 ± 6.5 10.93 ± 6.8 Suprailiac * ! 10.56± 5.6 10.40 ± 5.8 8.93 ± 5.5 9.30 ±5.9 Hence it is required to test whether there is any significant difference between the two sets of average measurements in skinfold, measured by two different calipers. As seen in Table 5 performance of the two calipers are distinctly different. Similar observations are noted between the two constructs in measuring body fat (Table 6). Table 5: Variability in average measurements taken with two different calipers Slim Guide Caliper Fat Track Caliper Triceps * 11.128 ± 5.9 10.394 ± 7.2 Biceps * 9.569 ± 6.5 7.955 ± 6.4 Subscapulars * 12.253 ± 6.6 10.867 ± 6.6 Suprailiac * 10.480 ± 5.7 9.115 ± 5.7 Table 6: Difference in body fat measurements by two constructs Slim Guide Caliper Fat Track Caliper Body Density * 1.05 ± 0.02 1.06 ± 0.0 Body Fat * 20.68 ± 9.4 18.42 ±10.6 Total Fat Mass * 13.73 ± 6.8 12.20 ± 7.6 To study the extent of association between BMI, body fat (%) and the measured proxy variables, their Pearson correlation coefficients are computed (Table 7). BMI is significantly correlated to all variables of interest, except neck circumference. Body fat (%) shows significant correlation with all variables except waist circumference. These observations seem counter-intuitive and may have been caused by inter-relationship among all the anthropometric variables. To study the relationships in greater detail the variables are categorized to see how they influence BMI and body fat. Table 7: Association among variables BMI Body Fat Waist Circ Hip Circ Waist-Hip Ratio Body Fat 0.28 * (< 0.01) Waist Circ 0.51 * (< 0.01) - 0.06 (0.39) Hip Circ 0.47 * (< 0.01) 0.15 * (0.03) 0.74 * (< 0.01) Waist-Hip Ratio 0.22 * (< 0.01) - 0.28 * (< 0.01) 0.64 * (< 0.01) - 0.03 (0.67) Neck Circ 0.07 (0.31) - 0.32 * (< 0.01) 0.66 * (< 0.01) 0.31 * (< 0.01) 0.62 * (< 0.01) Neck circumference has been recommended in the literature as a simple measure to screen overweight patients (Ben-Noun et al., 2001). Cut-off points were determined for neck circumference (Practical Handout p5) and its effectiveness in identifying overweight or obese persons was studied separately for males and females. Table 8 bears evidence that while for females the three groups determined by the neck circumference cut-offs do differ among themselves in all the 4 variables studied, for male population neck circumference may not be that effective. The underlined numbers indicate that mean of that group is significantly different from the other two means. If all three numbers are underlined it indicates that all three means differ significantly among themselves. Table 8: Differential distribution according to Neck Circumference Group Females (All significant) Males Group 1 Group 2 Group 3 Group 1 Group 2 Group 3 Waist Circ 68.33±7.2 81.27±9.4 93.12±2.1 * 75.55±3.4 78.74±3.3 83.09±3.5 * Waist-HipRatio 0.74±0.1 0.83±0.0 0.85±0.0 * 0.83±0.0 0.81±0.0 0.82±0.0 * BMI 22.76±3.7 26.98± 2.3 30.80±0.7 * 22.56±2.8 23.23±1.7 22.03±1.6 Body Fat 25.37±4.5 30.67±4.7 37.96±0.3 * 9.94±5.1 11.57±5.3 11.94±5.5 Finally cut-off points were defined for BMI and waist circumference also, separately for males and females and contingency tables were produced. Table 9 shows the cross-classification counts for females. For both genders the association between BMI and neck circumference groups is significant. Table 9: Association between BMI Groups and Neck Circumference Group Females Males BMI Group Neck Circ Group Neck Circ Group 1 2 3 1 2 3 1 3 0 0 3 0 0 2 78 3 0 26 22 11 3 13 17 0 11 5 0 4 6 0 6 0 0 0 In case of males, however, the given cut-off points for waist circumference are not valid since 100% of the sample have waist circumference less than 94 cm. For females the association between waist circumference and neck circumference is significant (Table 10). Table 10: Association between BMI Groups and Waist Circumference Group (Females) Waist Circ Group Neck Circ Group 1 2 3 1 93 9 0 2 7 4 0 3 0 7 6 Discussion: From the above analysis it is clear that gender specific analysis is to be recommended. The sample has almost twice the number of females than that of males. This anomaly in sample size may have contributed towards some discrepancy of results between the two genders. The proxy variables for heights have severe limitations and must be used only where direct measurement of heights is not available. Demi-span (tip) seems to be the winner among the 4 proxy variables considered. The fact that the duplicate measurements differ may not be a serious limitation of the study. For most part, the measurements are taken by inexperienced persons. With time and experience accuracy of measurement will improve. However, the discrepancy of measurements taken by two different calipers is a serious problem. More investigation is required to see if one caliper is consistently better than the other, or there are situations where one set of measurement is more to be relied upon than the other. Before a caliper is used, its validity is to be ensured. The interdependence of the anthropometric constructs is to be expected. However, simple correlation may not be powerful enough to capture the total scenario. All the variables studied here have more complicated relationship among themselves than simple binary relationships. More study is required to determine the conditional distribution and partial correlations of different variables, given levels of other variables. In fact, age may play a role here. In this sample the age distribution is skewed with a few high values in female sample (female maximum age = 54.32, male maximum age = 29.72). Another limitation of this sample is that, it is not a representative sample of any population. As mentioned before, all anthropometric constructs require golden standards, which is gender and race specific. The race information for this sample is not available. To fully develop anthropometric constructs it is necessary to determine a gender and race specific sample with clearly mentioned inclusion-exclusion criteria and validate the anthropometric measures by other nutritional measurements, e.g. dietary intake and laboratory assessments.4 References 1. World Health Organization. (1990). Diet, nutrition, and the prevention of chronic diseases. WHO Technical Report Series, No. 797. Geneva. 2. Food Standards Agency (2004) The National Diet & Nutrition Survey: adults aged 19 to 64 years (Survey carried out in Great Britain on behalf of the Food Standards Agency and the Departments of Health by the Office for National Statistics and Medical Research Council Human Nutrition Research), Web address - http://www.food.gov.uk/science/dietarysurveys/ndnsdocuments/ (check Volume 4) 3. Gibson, R.S. (2005). Principles of Nutritional Assessment. Oxford University Press, Oxford. 4. Lee, R. D., & Nieman, D. C. (2007). Nutritional Assessment. McGraw-Hill, New York. 5. Malnutrition Advisory Group. (2010). Malnutrition Universal Screening Tool. Web address - http://www.bapen.org.uk/must_tool.html. 6. World Health Organization. (1990). Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. Technical Report Series, No. 854. Geneva. 7. Gorstein, J., Sullivan, K., Yip, R., de Inis, M., Trowbridge, F., Fajans, P. & Clugston, G. (1994). Issues in the assessment of nutritional status using anthropometry. Bulletin of the World Health Organization, 72, 273-85. 8. Hogan, S. E. (1999). Knee Height as a Predictor of Recumbent Length for Individuals with Mobility-Impaired Cerebral Palsy. Journal of the American College of Nutrition, 18, 201-205. 9. Weinbrenner, T., Vioque, J., Barber, X. & Asensio, L. (2006). Estimation of Height and Body Mass Index from Demi-Span in Elderly Individuals. Gerontology, 52, 275-280. Read More