ARMSPAN AND HALFSPAN AS ALTERNATIVES
FOR HEIGHT IN ADULTS: A SAMPLE FROM GHANA

Tayie FAK*¹, Agyekum S², Owusu-Ahenkora M²,
Busolo D³, Adjetey-Sorsey E⁴, Armah J⁵ and E Imaya³

ABSTRACT

The suitability of armspan and halfspan as alternatives for height in BMI (body mass index) calculation was studied using a sample of 761 Ghanaians. Armspan looks promising as a substitute for height in elderly persons and others whose height cannot be obtained. Our confidence to assess nutritional status of older persons using regular BMI cut-off limits developed for younger adults is limited by the senescent changes that occur during ageing. Weight, height, armspan and halfspan were measured to obtain anthropometric data which enabled the development of regression equations that can be used to predict height. Background data were collected via one-on-one interview using a study-specific semi-structured questionnaire. Results showed that armspan significantly correlated with height in both males (r = 0.85) and females (r = 0.86) (P<0.0001). Predictions of height using the developed regression equations were strong (r = 0.91) (P<0.0001). The mean armspan to height ratio (armspan/height) for males and females were 1.07 and 1.06 respectfully and the mean difference between armspan and height was about 11.0 cm. BMI correlated significantly with BMA (body mass index calculated from armspan) (r = 0.95 0.97) (P<0.0001). Both BMA and BMH (body mass index calculated from halfspan) were lower than BMI values by about 3 kg/m² in both males and females. The values of armspan and fullspan (halfspan doubled) were similar and resulted in a high correlation between BMA and BMH (r = 0.98) generated from the two measures respectively. Older age groups had significantly less armspan values (P<0.0001) than younger adults. There seemed to be a slower rate of reduction of armspan than height as age advanced. Apparently, armspan may not remain unchanged during aging in this population. We concluded that a high correlation exist between height and armspan to allow prediction of height from armspan or halfspan for assessment of nutritional status in this population.

Key words: Body mass index, armspan, halfspan, height.

French

ENVERGURE ET DEMI-ENVERGURE DES BRAS EN TANT QUE MESURES ALTERNATIVES DE LA TAILLE/HAUTEUR DES ADULTES:
UN ECHANTILLON DU GHANA

RESUME

La pertinence de l’envergure et de la demi-envergure des bras comme moyens alternatifs de calculer la taille/hauteur de l’IMC (indexe de la masse corporelle) a été étudiée en utilisant un échantillon de 761 Ghanéens. L’envergure semble prometteuse comme substitut de la taille/hauteur chez les personnes âgées et d’autres dont la taille ne peut pas être mesurée. Notre confiance pour évaluer l’état nutritionnel des personnes âgées en utilisant les limites régulières minimum de l’IMC mises au point pour les adultes plus jeunes est limitée par les changements qui s’opèrent au cours du vieillissement. Le poids, la taille/hauteur, l’envergure et la demi-envergure ont été mesurés en vue d’obtenir des données anthropométriques qui ont rendu possible l’élaboration des équations de régression qui peuvent être utilisées pour prédire la taille/hauteur. Des données de base ont été collectées au moyen d’une interview individuelle en utilisant un questionnaire semi-structuré spécifique à une étude. Les résultats ont montré que l’envergure des bras avait de fortes corrélations avec la taille/hauteur aussi bien chez les hommes (r = 0,85) que chez les femmes (r = 0,86) (P<0,0001). Les prédictions de la taille/hauteur en utilisant les équations de régression qui avaient été mises au point étaient fausses (r = 0,91) (P<0,0001). Le rapport de l’envergure moyenne des bras sur la taille/hauteur (envergure des bras/hauteur) chez les hommes et chez les femmes était 1,07 et 1,06 respectivement, et la différence moyenne entre l’envergure des bras et la taille/hauteur était à peu près 11,0 cm. L’IMC avait de fortes corrélations avec le BMA (indexe de la masse corporelle calculé à partir de l’envergure des bras) (r = 0,95 – 0,97) (P<0,0001). Le BMA et le BMH (indexe de la masse corporelle calculé à partir de la demi-envergure des bras) étaient de 3 kg/m² inférieurs aux valeurs de l’IMC aussi bien chez les hommes que chez les femmes. Les valeurs de l’envergure des bras et de l’envergure totale (la demi-envergure doublée) étaient similaires et ont eu comme résultat une corrélation élevée entre le BMA et le BMH (r = 0,98) découlant des deux mesures respectivement. Les groupes de l’âge avancé avaient des valeurs de l’envergure des bras de loin inférieures (P<0,0001) à celles des adultes plus jeunes. Il semblait y avoir un taux plus lent de réduction de l’envergure des bras par rapport à la taille/hauteur au fur et à mesure que l’âge avançait. Apparemment, l’envergure des bras ne peut pas rester inchangée au cours du vieillissement dans cette population. Nous avons conclu qu’une grande corrélation existe entre la taille/hauteur et l’envergure des bras pour permettre de prédire la taille/hauteur à partir de l’envergure des bras ou de la demi-envergure en vue d’évaluer l’état nutritionnel de cette population.

Mots-clés: Indexe de la masse corporelle, envergure des bras, demi-envergure, taille/hauteur.

INTRODUCTION

Body mass index (BMI) calculated using height is a useful tool for the diagnosing of obesity and chronic energy deficiency (CED) [1], important routines in elderly populations in developing countries. In some cases in all age groups, height cannot be measured accurately because of clinical problems such as limb contracture, deformity or amputation. The measurement of height in older people may however not be accurate due to age-related height loss. The height loss has been attributed to shrinking of the inter-vertebral cartilages and overall spinal curvature, a condition termed kyphosis [2]. There is therefore a need for an alternative height measure which is not affected, to a significant degree, by aging to use as the denominator in the BMI equation, Weight (kg)/height (m²). Due to senescent changes in height, important questions have been raised about our capacity to assess nutritional status of older persons using regular BMI cut-off limits developed for younger adults [1].

Armspan was significantly different (p < 0.01) from height in two ethnic groups, the Afro Caribbean male and females, and Asian males, which indicates the importance of gender, age and ethnicity in the estimation of armspan and height relationships of a population. They suggested the need to consider ethnic differences in using armspan as an alternative for height [3]. The relation between armspan and height also varies between African-Americans and Caucasians. The differences between means of armspan and height are larger in African-Americans than in white adults [3]. In a report by Steele and Chenier [4] using subjects in North Carolina, strong but different correlations between armspan and height for African-Americans (r = 0.852) and white (r = 0.90) females were observed.

Armspan has been advanced as a surrogate for height to correct for age related loss of height [2]. Armspan approximates height at maturity and is relatively less affected by aging, a practical alternative to height in elderly persons who show extensive spinal curvature [5]. Substituting height for armspan to compute BMI, termed BMA [1], tends to overestimate chronic energy deficiency and underestimate obesity if cut-off limits developed from height are used [1]. Hence relationships must be developed that can be used to predict height from armspan or halfspan. Even though armspan measurement is a practical alternative for estimating height, it is necessary to establish a firm relationship in relation to ethnicity or race and gender.

Measurement of halfspan (from the midline of the sternal notch to the tip of the middle (longest) finger of the outstretched arm) [2] also called hemi-armspan, becomes important in the situation when the subject has limited movement of one shoulder, sometimes due to osteoarthritis. In such cases, halfspan can be measured and doubled, termed fullspan, to obtain a close estimate of armspan [6]. The BMI calculated from fullspan, termed BMH, can then be used for the assessment of nutritional status [7,8].

The objectives of our study are thus, (i) to study the relationship between height and armspan, and height and halfspan among Ghanaians as a sample of West Africans, (ii) to generate regression equations that can be used to calculate height from armspan and halfspan, and (iii) to assess BMA, BMH as practical alternatives to BMI.

MATERIALS AND METHODS

Subjects, locality and sampling
The participants of the study were Ghanaians resident in Accra. A total of 761 subjects participated in the study, 379 females and 382 males. Adults within the age group 20-85 years were randomly sampled from residences in Accra and the University of Ghana campuses. All participants were apparently without spinal curvature or arthritis. Subjects showed consent and demonstrated willingness to participate before they were included in the study.

Background and anthropometric data collected
Interview data on age, gender and ethnic group, were obtained by means of a study specific semi structured questionnaire. Anthropometric data collected included body weight, height, armspan and halfspan. These parameters were measured in duplicate using standard procedures as described below.

Body weight measurement
Body weight was measured in accordance with a standard procedure [2] using an adult weighing scale (SECA 890), minimum sensitivity limit 0.1 kg. Subjects in ordinary light minimum clothing stood on a weighing scale placed on a smooth level surface and without footwear. Unsupported with feet together and hands by side, looking straight at eye level, their weights were recorded [2,9,10].

Measuring armspan
Total length of the outstretched arms was measured as the armspan [2]. Subject was made to stand straight with the back against a wall or a smooth upright support, and wearing light clothing. Subject was made to stretch out arms fully straight with palms facing forward. A trained assistant supported the right arm and elbow. With wrists and fingers of subject all straight in line, and looking straight at eye level, a flexible non stretch steel tape was placed at the end of the middle (longest) finger and held in place by the assistant. The tape was then extended straight across the chest and along the left arm to the tip of the middle finger of the left arm. With all arms straight and horizontal, the armspan value was recorded to the nearest 0.1 cm in duplicate. Values varying for more than 0.5 cm for the same subject were rejected and measurement re-done [2].

Halfspan measurement
The halfspan or hemi armspan was measured under similar conditions as that for armspan. The length of the left arm was measured from the mid point (lowest point) of the sternal notch to the tip of the straight middle finger. Measurement was done in duplicate and the average taken [2].

Height measurement
Heights of subjects were measured under standardised conditions using a Leicester height measure (Leicester, UK) set up against a firm wall. Subjects were asked to stand on the platform of the instrument with feet together, back of heels, buttocks and head touching the vertical frame of the height measure and looking straight ahead in the Frankfurt plane. The headpiece of the height measure was lowered until it loosely but firmly touched the top of the subject’s head and the height was recorded to the nearest 0.1cm. The procedure was repeated for a duplicate value.

Analyses of the data collected
EPI-INFO version 6.04 (Centre for Disease Control and Prevention, WHO, Geneva) enabled data storage while Stata 5.0 (Stata Corporation, Texas, USA) was used for data analysis. Categorisation was by age group, BMI group and gender of subject. Fullspan values were obtained by doubling the halfspan values. Pearson’s correlation and linear regression analyses were used to ascertain relationships between variables. Students’ t-test for independent samples was used where comparison between two variable means was required. There was no need for data transformation because assumptions for normality and equal standard deviation were satisfied with no outliers. All findings were statistically tested for significance at P < 0.05.

RESULTS

We studied the suitability of armspan and halfspan as alternatives for height in BMI calculation using a sample of 761 Ghanaians. The sample population comprised of 382 (50.2%) males and 379 (49.8%) females. The mean age of the study population was 48 years (range; 20 – 85 years). Subjects belonging to the Akan tribe were about 54 %, while those of the Ga tribe were about 26 %. Ewes (12%) and Northerners (7%) made up the rest. These proportions reflect the generally observed tribal populations distribution in Ghana. Most of the participants (75%) have had at least 10 years of formal education. Vocational workers (40 %), students (32%) and office workers (18%) were the majority in the study.

Our results showed that the mean height of this study sample, sexes combined, was 163.605 cm (Table 1). The mean armspan and fullspan (halfspan doubled) for this study population, sexes combined, were similar (armspan, 174.598 ± 0.376; fullspan 174.483 ? 0.375) (P>0.05). In both genders, mean armspan was significantly higher than height (P<0.0001). The mean difference between armspan and height was about 11.0 cm (Table 2). A similar difference was observed for fullspan. The differences between height and armspan, and height and fullspan for females tended to be lower than males (P<0.001). Armspan and fullspan for females therefore correlated better with height than males. Armspan and fullspan significantly correlated with height in both males and females (P<0.0001) (Table 3). In both genders, correlations between 0.83 0.85 were observed for height with armspan (Figure 1) as well as height with fullspan (Table 3). Correlations of BMI with the calculated indices, BMA and BMH, were strong ranging from 0.95 – 0.97 (Table 3). Due to the closeness of the values of armspan and fullspan, values for BMA and BMH, which are generated from these two measures respectively, were also similar (r = 0.98) (Table 3). On average, BMI values were higher than both BMA and BMH values by about 3 kg/m2 in both males and females (Table 2). In this study, the mean armspan to height ratio (armspan/height) for males was 1.07 (range; 0.997 – 1.153) and for females was 1.064 (range; 0.982 – 1.184). The observed armspan to height ratio was similar to what has been reported for Afro Caribbeans (1.04) [3]. Similar values were found for height to fullspan ratio: males; 1.060 (range; 0.976 – 1.155) and females; 1.063 (range; 0.930 - 1.140).

Figure 1 The relationship of armspan with height among the study populaiton (n=761, sexes combined)

Figure 2 Original height versus height predicted from armspan to show their relationship (n=761, sexes combined)

Prediction of height using the developed regression equations (Table 4) was accurate with high correlation between the actual and the predicted values (r = 0.91) (Figure 2).

There were statistically significant (P<0.0001) differences in height between younger adults (below 65 years) and the elderly (above 65 years) in both genders. Differences ranging from 3.5 – 6.5 cm were observed between these two groups. Contrary to general expectation, armspan also reduced significantly (P<0.006) as age advanced (Figures 5 and 6).

DISCUSSION

This study furnishes data on West Africans, in terms of the suitability of armspan as a proxy for height. The anthropometric values of the study participants (Table 1) reflect the usual anthropometric characteristics of Ghanaians. Armspan was significantly greater than height in this study sample (Table 2). Steele and Mattox [11] reported a significant relationship between armspan and height in African-Americans and Caucasian females aged between 23 – 28 years and observed that on the average, armspan exceeded height by 8.3cm for African-Americans and 1.8cm for whites. The armspan and height difference for African-American females in their report is similar to what we have found (Table 2).

Armspan and fullspan were significantly greater than height, thus both BMA and BMH were significantly less than BMI. A direct substitution of armspan for height in the BMI equation will therefore tend to overestimate CED and underestimate obesity. However, the significant association between armspan and height makes armspan an excellent predictor of height in this population. Versluis et al. [12] in a study to assess the usefulness of armspan as a substitute for height in detecting vertebral deformities in women, reported a correlation of 0.83 between armspan and height. A study by Rabe et al. [1] showed the correlation of height with armspan among Indonesian elderly to be r = 0.83 and r = 0.81 for females and males respectively. These correlations between armspan and height are slightly lower than what we have observed among Ghanaians, though Steele and Chenier [4] had observed a stronger association (r = 0.90) between armspan and height among 293 African-Americans and 298 Caucasian females in North Carolina. A study of 50 adult African-Americans found a correlation of 0.87 between armspan and height [13], which is similar to our observation for Ghanaians.

Although reference cut-off limits for BMI are frequently documented, reference standards for both BMA and BMH are scanty. Fortunately, a strong association exists between BMI and BMA or BMH (Table 3) among this study sample. BMI values can thus be predicted from BMA and BMH for use in the assessment of nutritional status (Table 4). Inserting the armspan or halfspan values in the appropriate equation (Table 4), one can also easily predict height for BMI calculation. Due to the wide age range of participants in this study, our prediction equations can be applicable to all adults in this study zone.

James et al.. [14] proposed in their classification of chronic energy deficiency (CED), that BMI (kg/ m2) between 18.5 and 30 is classified as normal; 17.0 - 18.4 as grade I, 16.9 16.0 as grade II, and below 16.0 as grade III or severe CED. Using this BMI classification [14], a CED prevalence of 9.07% was detected using our predicted values as compared to a prevalence of 9.46% using the actual BMI values (Figures 3 and 4). This observation makes our regression equations useful in predicting height when it cannot be measured (Table 4). Similar equations were developed by Steele and Chenier [4] for African-Americans and Caucasian females in North Carolina as follows:

White women:
Height (cm) = 29.58 - (0.04 × age) + (0.81 × armspan).
African-American women:
Height (cm) = 37.72 - (0.01 × age) + (0.73 × armspan).

Figure 3 Body mass index classification using observed values

Another set of equations relating armspan to height were developed for estimation of height in Afro-Caribbeans [3] but without age component as follows:
Males:
Height (cm) = 0.66 × armspan + 54.9
Females:
Height (cm) = 0.57 × armspan + 54.9

Figure 4 Body mass index classification using predicted values

DeGroot et al.. [15], using a representative samples from 12 European countries, termed the EC/SENECA study, observed mean BMI ranges of 23.9 - 30.5 7 kg/ m² for females and 24.4 -30.3 7 kg/ m² for males. These ranges are narrower than what we have observed in this Ghanaian sample, which ranges from 12.98 – 38.50 (Table 2). Reasons for these differences could be due to age range, racial and environmental factors [1].

Results of our study indicate strongly that armspan can be used to predict height for this population. However, older age groups also had significantly less armspan values than younger adults. There seemed to be a slower rate of reduction of armspan than height as age advanced (Figures 5 and 6). Thus armspan may not remain unchanged during aging in this population. A longitudinal study of armspan during aging is required to validate this observation.

Figure 5 Mean height of the study participants according to age group (n = 761, sexes combined)*

*Age group
1 = 20-40 years (n = 386)
2 = 41-64 years (n= 124)
3 = 65-74 years (n = 144)
4 = >75 years (n = 107)

Figure 6 Mean armspan of the study participants according to age group (n = 761, sexes combined)*

*Age group
1 = 20-40 years (n = 386)
2 = 41-64 years (n= 124)
3 = 65-74 years (n = 144)
4 = >75 years (n = 107)

CONCLUDING REMARKS

In conclusion, there is a high correlation between height and armspan to allow prediction of height from armspan among this population.

REFERENCES

1.      Rabe SL, Thamrin MH, Gross R, Solomons NW and W Schultink Body Mass Index of the Elderly Derived from Height and from Armspan. Asia Pacific J. Clin. Nutr. 1996; 5: 79-83.

2.      Manandhar M and I Ismail Better Nutrition for Older People. Assessment and Action. HelpAge International. London, 1999: 12-25.

3.      Reeves SL, Carakamin C and CJK Henry The Relationship Between Armspan Measurement and Height with Special Reference to Gender and Ethnicity. Eur. J. Clin. Nutr. 1996; 50: 398.

4.      Steele MF and JW Mattox Short Report: Correlation of Armspan and Height in Young Women of Two Races. Ann. Hum. Biol. 1987; 14: 445-447.

5.      Kuzmarski RJ Need for Body Composition Information in Elderly Subjects. Am. J. Clin. Nutr. 1989; 50 : 1150-1157.

6.      Kwok T and MN Whitelaw The Use of Armspan in Nutritional Assessment of the Elderly. J. Am. Ger. Soc. 1991; 6: 226.

7.      Agyekum S and FAK Tayie Armspan as an Alternative Index for Height in the Adult Ghanaian. Department of Nutrition and Food Science, University of Ghana, Legon, 1999: 4-42.

8.      Owusu-Ahenkora M and FAK Tayie The Use of Armspan and Knee‑height as Alternative Anthropometric Indices for Assessing Nutritional Status. Department  of Nutrition and Food Science, University of Ghana, Legon, 2001: 5-35.

9.      Gibson RS Nutritional Assessment. A Laboratory Manual. Oxford University. New York, 1993: 59-68.

10.  Jelliffe DB and EFP Jelliffe Community Nutritional Assessment, Oxford University Press, Oxford. 1989.

11.  Steele MF and TC Chenier Armspan, Height and Age in African-American and White Women.  Annals of Human Biology, 1990; 17: 533-541.

12.  Versluis RG, Petri H, van de Ven CM, Scholtes ABJ, Broerse ER, Springer MP and SE Papapoulos. Usefulness of Armspan and Height Comparison in Detecting Vertebral Deformities in Women. Osteoporosis International, 1999; 9: 129-133.

13.  McPherson R, Lancaster DR, JC Carrole Height Changes with Aging in African-Americans. Amer. J. Gerontol.  1978; 33: 20-25.

14.  James WPT, Ferro-Luzzi A and JC Waterlo Definition of Chronic Energy Deficiency in Adults; Report of a Working Paper of the International Dietary Energy Consultative Group. Eur. J. Clin. Nutr. 1988; 42 :969-981.

15.   DeGroot PC, Van Staveren WA and HJ Euronat EC/SENECA. Nutrition in the Elderly in Europe. Eur. J. Clin. Nutr. 1991; 45 (3): 180.

*Corresponding Author Email: ftayie@iastate.edu
¹Department of Food Science and Human Nutrition, Iowa State University
²Department of Nutrition and Food Science, University of Ghana
³HelpAge International- African Regional Development Centre, Nairobi, Kenya
⁴HelpAge Ghana
⁵Nutrition Division, Ministry of Health - Ghana