Height and body mass are important parameters to measure in the assessment of adiposity in children and adults( Reference Burton 1 ). Given the well-documented limitations of BMI in the assessment of obesity, future basal metabolic risk and standardization of physiological parameters like muscle strength and glomerular filtration rate during human growth( 2 , Reference Kemper 3 ), waist circumference, mid-upper arm circumference (MUAC) and skinfold thickness are leading direct methods used to assess fat distribution( Reference Cook, Weitzman and Auinger 4 , Reference Cameron 5 ). There are established relationships between height and different body dimensions in adult human beings. The relationship between height and arm span is reported by some authors( Reference Lucia, Lemma and Tesfaye 6 – Reference Yousafzai, Filteau and Wirz 11 ) while other studies have reported the estimation of height from arm span( Reference Zverev and Chisi 12 – Reference Ofluoglu, Unlu and Akyuz 15 ). Furthermore, height has been estimated from the dimensions of upper and lower extremities( Reference Ozaslan, Ozaslan and Tugcu 9 , Reference Agnihotri, Agnihotri and Jeebun 16 , Reference Krishan, Kanchan and DiMaggio 17 ). However, in the fracture of vertebrae or paralysis situation, measuring height is difficult( Reference Brown, Feng and Knapp 18 ). In such a situation, measurements such as arm span, waist circumference, knee height, arm length, sitting height, MUAC and skinfold thickness could be used( Reference Kemper 3 , Reference Cameron 5 , Reference Ozer, Gultekin and Sagir 19 , Reference Gauld, Kappers and Carlin 20 ) to assess height.
There are few studies showing the associations between height, MUAC and waist curcumference( Reference Yousafzai, Filteau and Wirz 11 , Reference Krishan, Kanchan and DiMaggio 17 ). The purpose of the present study was to assess relationships between height and arm span, waist circumference, sum of four skinfold thicknesses and MUAC among Ellisras rural children aged 8–18 years.
Methods
Geographical area
The rural areas of Ellisras are situated in the north-western region of Limpopo Province, South Africa. Approximately 50 000 people live there in forty-two settlements, which are 70 km from Ellisras town, now known as Lephalale, adjacent to the border with Botswana. Iscor coal mine and Matimba electricity power station are the two major sources of employment for many residents. The remaining workforce is involved in subsistence farming and cattle rearing, while a minority works in education or the civil service. Unemployment, poverty and low life expectancy seem to be more prevalent in this rural area( Reference Monyeki, Monyeki and Brits 21 ).
Study population
The Ellisras Longitudinal Study (ELS) initially followed a cluster sampling method described previously by Monyeki et al.( Reference Monyeki, Cameron and Getz 22 ). In brief, the study was undertaken at twenty-two schools (ten pre-schools and twelve primary schools) randomly selected from sixty-eight schools within Ellisras area( Reference Monyeki, de Ridder and Steyn 23 ). For the purpose of the present analysis, data collected in November 2003 were used. The sample population comprised 1769 children (911 boys, 858 girls) aged 8–18 years.
The Ethics Committee of the University of Limpopo (Turfloop Branch) granted ethical approval prior to the survey and parents or guardians provided informed consent. The children and their parents signed assent forms after receiving verbal assent, which were translated into their mother tongue by the project’s principal investigator.
Anthropometrics
All children underwent a series of anthropometric measurements according to the International Society for the Advancement of Kinanthropometry. Body mass was measured on an electronic scale to the nearest 0·1 kg. A Martin anthropometer was used to measure height to the nearest 0·1 cm. Skinfolds (suprailiac, subscapular, triceps and biceps) were measured three times using a Slim Guide skinfold calliper. Waist circumference and MUAC were measured with a flexible steel tape to the nearest 0·1 cm. Waist circumference was measured laterally midway between the lowest portion of the rib cage and the iliac crest, and anteriorly midway between the xiphoid process of the sternum and the umbilicus. Arm span was measured by positioning the participant’s back against the wall with his/her arms spread against the wall at shoulder level and parallel to the floor with the palms facing forward. A steel tape was used to measure from the tip of the middle finger on one hand across the chest to the tip of the middle finger on other hand( Reference Norton and Olds 24 ). Participants below 8 years of age and above 18 years of age were of limited number. In order to make a meaningful conclusion to the study, participants were placed into age groups of 8–10 years, 11–14 years and 15–18 years.
Quality control
Training of all anthropometric measurements was done in compliance with standard procedures of the International Society for the Advancement of Kinanthropometry( Reference Norton and Olds 24 ). Reliability and validity of the anthropometric measurements are reported elsewhere( Reference Monyeki, de Ridder and Steyn 23 ). In brief, the absolute and relative values for intra- and inter-tester technical error of measurement ranged from 0·04 to 4·16 cm (0·2 to 5·0 %) for height, from 0·01 to 0·02 kg (0 to 0·3 %) for body mass, from 0·2 to 6·0 mm (0·4 to 6·8 %) for skinfold measurements, and from 0 to 3·4 cm (0 to 4 %) for arm span and girth measurements (waist and MUAC).
Statistical analysis
Descriptive statistics were calculated for absolute body size. Student’s t test was applied to test for significant differences between sexes. Pearson correlation coefficients were calculated to assess associations between height and arm span, waist circumference, MUAC and sum of the four skinfold thicknesses. Linear regression models were used to assess relationships between height and age, arm span, MUAC, waist circumference and sum of four skinfolds. All statistical analyses were done using the statistical software package SPSS version 10. The statistical significance was set at P<0·05.
Results
Table 1 presents the descriptive statistics for age and anthropometric measurements of Ellisras rural children aged 8–18 years. Ellisras rural girls the age group of 11–15 years showed significantly higher mean values than boys for height, arm span, MUAC and waist circumference (all P<0·001). Sum of four skinfolds was significantly (P<0·05) higher in girls (range from 26·2 to 40·4 mm) than in boys (range from 24·0 to 26·2 mm) throughout all age groups. Mean height (range from 137·6 to 162·2 cm) was found to be higher than arm span (range from 125·7 to 157·4 cm) throughout all age groups for both Ellisras rural boys and girls, with the differences ranging from 4·0 to 11·5 cm.
Table 1 Descriptive statistics for age and anthropometric measurements of Ellisras rural children aged 8–18 years, Limpopo Province, South Africa, November 2003
MUAC, mid-upper arm circumference.
*P<0·05, **P<0·001.
† Sum of triceps, biceps, suprailiac and subscapular.
Height and arm span showed a highly significant (P<0·001) relationship (correlation coefficient ranging from 0·74 to 0·91) in both boys and girls throughout the age range (Table 2). The correlation between height and MUAC, waist circumference and sum of four skinfolds was low and significant (P<0·05) in both Ellisras boys and girls, decreasing in magnitude in the older age group (correlation coefficient ranging from 0·15 to 0·47). The correlation (correlation coefficient ranging from 0·15 to 0·29) between height and sum of four skinfolds was low and significant (P<0·001) in the older age groups (11–18 years). The correlation between height and age was not significant in the age group 8–10 years and 15–18 years for boys, while for girls in all age groups it was significant (P<0·05; Table 2).
Table 2 Pearson correlation coefficients between height and age and other anthropometric variables of Ellisras rural children aged 8–18 years, Limpopo Province, South Africa, November 2003
MUAC, mid-upper arm circumference.
*P<0·05, **P<0·001.
Table 3 illustrates the linear regression analysis of various anthropometric measurements and age as predictors of height for Ellisras rural children aged 8–18 years. The coefficient of determination of the models was high in the age group 8–10 years (84 % for boys and 75 % for girls) and 11–14 years (83 % for boys and 75 % for girls) and decreased in the 15–18 years age group for both boys (59 %) and girls (56 %). Arm span was a significant (P<0·001 or P<0·05) predictor of height in all the age groups for both boys and girls, ranging from β=0·615 (95 % CI 0·55, 0·68) to β=0·755 (95 % CI 0·69, 0·90). Waist circumference, with a range from β=−0·144 (95 % CI −0·56, 0·27) to β=−0·177 (95 % CI −0·32, −0·04), and sum of four skinfolds, ranging from β=–0·040 (95 % CI −0·19, 0·11) to β=0·063 (95 % CI 0·01, 0·12), were statistically significant (P<0·05) predictors of height in the 15–18 years age group for girls. MUAC, with β=0·497 (95 % CI 0·17, 0·82), and sum of four skinfolds, with β=−0·076 (95 % CI −0·14, −0·02), were significant (P<0·05) predictors of height in the 11–14 years age group for boys (Table 3). Waist circumference was not a statistically significant predictor of height among boys in all the age groups.
Table 3 Linear regression analysis of various anthropometric measurements and age as a predictor of height for Ellisras rural children aged 8–18 years, Limpopo Province, South Africa, November 2003
SEE, standard error of the estimate; MUAC, mid-upper arm circumference.
*P<0·05, **P<0·001.
Discussion
The present study was aimed at assessing the relationship between height and other anthropometric parameters of schoolchildren from rural areas of Ellisras, South Africa. Linear regression analysis revealed that arm span was a significant (P<0·05) predictor of height in all age groups, while MUAC and sum of four skinfolds significantly predicted height in the 11–14 years age group for Ellisras boys. However, age could predict height in the 8–10 years age group for both boys and girls. Similar results were reported for Dhimals in West Bengal, India for children aged 10–18 years( Reference Banik 7 ).
Longer mean height than arm span among Ellisras rural children was recorded in present study. Similar results were reported by Lucia et al.( Reference Lucia, Lemma and Tesfaye 6 ) for the four ethnic groups Oromo, Ambara, Tigre and Somali in Ethiopia. Furthermore, Yabanci et al.( Reference Yabanci, Killic and Simesek 10 ) reported shorter arm span compared with height among Turkish children aged 7–14 years. The results of the present study exhibit different trends from earlier reports. In Malawian children and Indian children arm span was found to be higher than height( Reference Banik 7 , Reference Yousafzai, Filteau and Wirz 11 , Reference Tores, Martinez and Manqo 25 ).
A significant strong correlation was recorded between height and arm span, waist circumference, MUAC and sum of four skinfolds in the present study among boys and girls. Similar trends were reported among Turkish children aged 7–14 years for height, arm span and MUAC( Reference Yabanci, Killic and Simesek 10 ). In other studies too, height and arm span were reported to be strongly correlated( Reference Yousafzai, Filteau and Wirz 11 , Reference Zverev and Chisi 12 , Reference Marcato, Sampalo and Alves 26 ).
Arm span was found to be an impressive predictor of height in the present study. Sum of four skinfolds was a significant (P<0·05) positive predictor of height in the 15–18 years age group while waist circumference significantly (P<0·05) negatively predicted height in the same age group for girls. Yabanci et al.( Reference Yabanci, Killic and Simesek 10 ) reported MUAC to be a predictor of height in the age group 7–10 years, while waist circumference was not a predictor of height for both sexes. Even though sitting height was reported to be a better predictor of total BMI( Reference Burton 1 , Reference Marcato, Sampalo and Alves 26 ), there has been no study showing the correlation between height and sum of four skinfolds.
The findings of the ELS do not differ with other investigations worldwide. In Montenegrins the use of arm span was very effective to determine the height of female and male communities in the surrounding areas, even though it may differ in ethnic and racial groups( Reference Brown, Feng and Knapp 18 , Reference Coon 27 ). In all of the estimations of body height from various anthropometric measurements it is important to emphasize that arm span has been the most reliable indicator for predicting the body height of an individual( Reference Mohanty, Suresh and Sreekumaran 28 ). Arm span can be a reliable method used to measure height for disabled, old aged and deformed individuals( Reference Yousafzai, Filteau and Wirz 11 , Reference Hickson and Frost 29 ). A study by Yabanci et al.( Reference Yabanci, Killic and Simesek 10 ) stated that MUAC and waist circumference were found to be moderately correlated with height, while previous work recommended the use of MUAC as a criterion standard that can be used to estimate height( Reference Mozumdar and Roy 30 ). Others have argued that neither MUAC nor skinfold thickness has a good relationship ability with height( Reference Burden, Stoppard and Shaffer 31 ).
Conclusion
In conclusion, arm span was found to be strongly related to height in schoolchildren aged 8–18 years from rural areas of Ellisras, South Africa, as reported in many studies. Sum of four skinfolds and waist circumference were found to be significantly associated with height in the older age group for girls. With waist circumference, however, the association was negative.
Acknowledgements
Acknowledgements: The authors are indebted to the ELS administrators (P.S. Seleka and T.T. Makata) for providing technical support in preparation of this manuscript. Financial support: This work was supported by the Vrije University, Amsterdam, The Netherlands (grant number UNIN Health Project under VUA Foundation funds) and the University of the North, South Africa (grant number 1404); the National Research Foundation (grant number URD2002050400168) for the Ellisras Longitudinal Growth and Health study is also acknowledged with gratitude. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors, and therefore the above-mentioned funding sources do not accept any liability in regard thereto. Conflict of interest: None. Authorship: K.D.M. was responsible for study design, data collection, data analysis and write-up. M.M.S. was responsible for data analysis and write-up. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects/patients were approved by the University of Limpopo (Turfloop Campus). Written informed consent was obtained from all subjects/patients.
