Year : 2008 | Volume
: 3 | Issue : 3 | Page : 94--99
Using arm span to derive height: Impact of three estimates of height on interpretation of spirometry
Cardiorespiratory Physiology, Clinical Research Centre, Vallabhbhai Patel Chest Institute, University of Delhi, Delhi, India
S K Chhabra
Department of Cardiorespiratory Physiology, Vallabhbhai Patel Chest Institute, University of Delhi, Delhi-110007
Background: When standing height required to calculate forced vital capacity (FVC) cannot be measured, it can be derived from arm span using different methods.
Objectives: To compare three different estimates of height derived from arm span and investigate their impact on interpretation of spirometric data.
Methods: In a cross-sectional study, 517 subjects aged 7 to 76 years, with various respiratory diseases underwent spirometry. Three estimates of height were obtained from arm span: (a) by direct substitution (Ht AS ); (b) estimated height (Ht est ), obtained from the mean arm span:standing height ratio; and (c) predicted height (Ht pred ), obtained from arm span by linear regression analysis. Predicted values of forced vital capacity (FVC) obtained from these estimates were compared with those obtained from actual standing height (Ht act ), followed by Bland Altman analysis of agreement in the patterns of ventilatory impairment.
Results: The arm span was 5%-6% greater than the height. The difference increased with increasing height. Ht AS and the FVC predicted from it were significantly greater than the other measures of height and the related predicted FVCs respectively. Compared to Ht act , Ht AS gave a misclassification rate of 23.7% in taller subjects (Ht act > 150 cm) and 14.2% in shorter subjects in the patterns of ventilatory impairment. Misclassification rates were 6%-8% with Ht est and Ht pred . Agreement analysis showed that FVCs predicted from Ht pred had the best agreement with the FVC predicted from Ht act .
Conclusions: Among several methods of estimating height from the arm span, prediction by regression is most appropriate as it gives least errors in interpretation of spirometric data
|How to cite this article:|
Chhabra S K. Using arm span to derive height: Impact of three estimates of height on interpretation of spirometry.Ann Thorac Med 2008;3:94-99
|How to cite this URL:|
Chhabra S K. Using arm span to derive height: Impact of three estimates of height on interpretation of spirometry. Ann Thorac Med [serial online] 2008 [cited 2021 Dec 6 ];3:94-99
Available from: https://www.thoracicmedicine.org/text.asp?2008/3/3/94/39574
Assessment of lung function by spirometry constitutes an important part of diagnostic work-up of a patient with respiratory disease. , Interpretation requires comparison of observed with normal values that are obtained from prediction equations. Standing height and age are the major determinant variables of forced vital capacity (FVC) and explain most of its variance. 
In patients unable to stand straight due to physical disability, structural defects such as kyphoscoliosis or neuromuscular weakness or leg amputation, standing height can be estimated from arm span measurements. ,,,,,, This may be done by direct substitution by the latter ,,, or by application of a fixed correction factor based on arm span:height ratio  or by estimating height from arm span using regression equations. ,, For reasons of simplicity, direct substitution by arm span has been favored, both in children  and adults. , The errors introduced by such substitution in interpretation of spirometric data have been reported to be small. , The recent joint statement of the American Thoracic Society (ATS) and the European Respiratory Society (ERS) has also recommended such substitution.  However, the agreement between arm span and standing height has been found to be poor. , Direct substitution of arm span for height was also questioned recently by Golshan et al. ,  who found estimation from regression models to be superior in healthy subjects. As these three methods may not provide similar estimates of standing height, the predicted values will differ and could have an impact on interpretation. An agreement analysis of predicted vital capacity values obtained from height estimated by different methods from the arm span, as well as an evaluation of the impact of predicted vital capacity values on interpretation of spirometric data in patients suffering from respiratory diseases, has not been done. Therefore, the present study was carried out.
Material and Methods
The study protocol was approved by the institutional ethics committee. Patients, both children and adults, suffering from different respiratory diseases and referred to the Pulmonary Function Laboratory of our institute were considered for inclusion in the study. The inclusion criteria were an ability to stand erect for measurement of standing height, absence of any spinal and other skeletal abnormality of the limbs, and performance of acceptable spirometry maneuvers as per the ATS recommendations.  It was a cross-sectional study. A total of 550 subjects were screened and 517 were found eligible. There were 287 male and 230 female subjects. The age ranged from 7 to 76 years. The purpose of the study was explained and consent taken.
Standing height (actual height, Ht act ) was measured with a stadiometer in barefoot condition with head in the Frankfurt plane. Arm span (AS) was measured while the patients were asked to stand with their backs against the wall and arms spread in a straight line parallel to the floor. It was measured from the tip of the middle finger of the right hand to the tip of the middle finger of the left hand across the chest at the clavicles. All readings were taken to the nearest 0.5 cm. All the measurements were performed thrice, and the mean was taken and rounded off to the nearest centimeter. Arm span used as a surrogate for height was labeled as Ht AS .
Besides Ht act and Ht AS , two additional measures of stature were derived. Estimated height (Ht est ) was obtained by dividing the AS with the mean AS:Ht act ratio. As the AS:Ht act ratio was not different in males and females, the mean of the whole sample was used. Predicted height (Ht pred ) was obtained by developing a linear regression equation with arm span as the independent variable.
Spirometry was performed on a dry rolling-seal spirometer of the Benchmark model lung function machine (P. K. Morgan, Kent, UK). Maximal expiratory flow volume curves were obtained as per the ATS 1995 recommendations.  Three acceptable and at least two reproducible curves were obtained in each subject. The highest values of FVC and FEV 1 were selected.
Predicted values of FVC were calculated using regression equations for healthy north Indian adults  and children.  Forced vital capacity (FVC) obtained from Ht act was taken as the reference and labeled as FVC-Ht act . Predicted FVCs were obtained from other three measures of stature and labeled as FVC-Ht AS , FVC-Ht est , and FVC-Ht pred . The lower limits of normal (LLN) for FVC were calculated as the difference between the predicted value and 1.645 times the standard error of estimate (SEE). An FEV 1 /FVC 1 /FVC ratio was labeled as 'restrictive disorder.' An FEV 1 /FVC 1 /FVC ≥70% and FVC above LLN was labeled as 'normal spirometry.'
Statistical analysis was done using SPSS 11.0 and GraphPad Prism 4.01 for Windows. Data was expressed as mean (sd). Comparisons between two groups were carried out using unpaired paired t test. Comparison of multiple groups was done using the analysis of variance followed by Bonferroni's multiple comparison test to identify significantly different groups. Coefficients of correlations in bivariate relationships were obtained using the Pearson's correlation test. Agreement between various sets of measurements was evaluated by calculating the kappa value for nominal variables and by the Bland Altman analysis for continuous variables. Kappa values were interpreted as recommended by Byrt et al.  Linear regression analysis was used to predict Ht pred from the arm span. Statistical significance of differences was assumed at P act ) and arm span (AS) is shown in [Table 1]. In an overwhelming majority of the subjects (97.8%), the arm span exceeded height by a mean of 9.66 cm and 8.09 cm in male and female subjects respectively. There was no gender difference (P > .05). The difference was 5.9% in males and 5.4% in females.
Ht act and arm span were well correlated (r = 0.95, P act against Ht act [Figure 2] revealed that the difference increased with increasing Ht act and reached a plateau at Ht act = 150 cm. The subjects were therefore divided into two groups: group A, Ht act ≥ 151 cm (n = 397); and group B, Ht act ≤ 150 cm (n = 120). With this division, the relationship between AS - Ht act and Ht act no longer remained significant in group A (r = 0.18, P > .05), while it remained significant in group B (r = 0.40, P act and Ht act , subsequent analysis was carried out separately in the two groups.
The linear regression analysis with arm span as the independent variable and Ht pred as dependent variable yielded the following equation in group A: Ht pred = 33.05 + (0.753 × AS) (r 2 = 0.78, standard expected error, 3.597). In group B, the equation was Ht pred = 30.47 + (0.745 × AS) (r 2 = 0.89, SEE, 3.51). The four measurements of stature in the two groups are shown in [Table 2]. ANOVA revealed significant differences ( P AS was significantly greater than Ht act , Ht est , and Ht pred ( P P > .05).
Predicted FVC (in L) from the four measures of stature revealed significant differences in both the groups ([Table 2], ANOVA, P AS was significantly greater than each of the other three FVCs (P P > .05).
Results of Bland Altman analysis to examine the agreement between FVC predicted from Ht act and that predicted from the other three estimates of height are shown in [Table 3] and [Figure 3],[Figure 4],[Figure 5],[Figure 6],[Figure 7],[Figure 8]. In both the groups, the mean difference between FVC-Ht act and FVC-Ht AS was large with wide limits of agreement. A trend of increasing difference was observed with increasing average of the two [Figure 3] and [Figure 6]. On the other hand, the mean difference between FVC-Ht act and FVC-Ht est and between FVC-Ht act and FVC-Ht pred was smaller in both the groups. The scatter around the bias line was uniform [Figure 4] and [Figure 5],[Figure 7] and [Figure 8]. The 95% limits of agreement were narrower and similar between FVC-Ht act and FVC-Ht est and between FVC-Ht act and FVC-Ht pred in group A. In group B, the 95% limits of agreement between FVC-Ht act and FVC-Ht pred were even narrower than those between FVC-Ht act and FVC-Ht est .
The interpretation of spirometry results using the four measures of stature to calculate predicted FVC is shown in [Table 4]. In group A, out of 397 subjects, 303 were interpreted similarly by Ht AS and Ht act , giving a misclassification rate of 23.7% (kappa value for agreement, 0.68, indicating a good agreement). With Ht est , 367 subjects were categorized similarly as done with Ht act , giving a misclassification rate of 8.4% (kappa value for agreement, 0.89, indicating a very good agreement). Finally, categorization was similar in 373 subjects by Ht act and Ht pred , giving a misclassification rate of 6% (kappa value for agreement, 0.88, indicating a very good agreement). In group B, out of 120 subjects, 103 subjects were interpreted similarly by Ht AS and Ht act (misclassification rate of 14.2%; kappa value for agreement, 0.8, indicating a good agreement). Using Ht est , 112 subjects were categorized similarly as done by Ht act , giving a misclassification rate of 6.7% (kappa value for agreement, 0.90, indicating a very good agreement). Finally, categorization was similar in 110 subjects by Ht act and Ht pred , giving a misclassification rate of 8.3% (kappa value for agreement, 0.88, indicating a very good agreement).
The results of the study show that arm span was about 5 to 6% greater than the standing height. The difference increased with increasing height and was similar in males and females. Height estimated from a fixed correction factor derived from the arm span:height ratio and/or predicted from arm span using a linear regression equation yielded a value closer to actual standing height than using the arm span as a substitute. Therefore, predicted FVC calculated by using arm span instead of actual height was substantially higher than that obtained using the other two estimates of height. This resulted in large errors in interpretation of spirometric data. Substitution of arm span for actual height resulted in erroneous interpretation in almost one quarter of the subjects taller than 150 cm and in about 15% of the shorter subjects. The errors in interpretation with the other two estimates of height were smaller (6% to 8%) and similar.
The correlation between standing height and arm span is usually excellent. , Therefore, when actual standing height cannot be measured or is affected by disease as in thoracic vertebral compression due to osteoporosis, and deformity in kyphoscoliosis, arm span has been recommended as a substitute for calculation of the predicted FVC. ,,, Parker et al.  observed that fixed arm span-to-height ratios may also be used to estimate height with reasonable accuracy, but there were errors at extremes of stature. Aggarwal et al.  observed that height obtained by substitution of arm span or estimation by ratio method resulted in similar errors in interpretation of spirometric data. The errors were however substantial, 16.22% and 14.04% subjects being wrongly classified respectively. They did not evaluate the utility of predicting height from arm span by regression analysis. In a recent study, Golshan et al.  found that arm span substitution for height over-predicted FVC in healthy subjects compared to FVC calculated from height obtained by regression equations. They did not evaluate the impact of different methods to obtain height on interpretation of spirometry data in patients.
A good correlation does not necessarily mean a good agreement. Torres et al.  and Hickson and Frost,  applying the Bland Altman analysis, observed a poor agreement between the arm span and height even though these correlate well. Hence substitution is not justified. Our observations corroborate these. In addition, we found that height estimated from arm span using a fixed ratio or regression gave similar FVC. This is the first study to report on the impact of these three different estimates of height from arm span on spirometric interpretation in patients with respiratory diseases.
The greater misclassification with direct substitution of arm span for actual height resulted from shifts from normal and obstructive categories to restrictive and mixed categories. This can be explained as arm span yields higher predicted FVC and observed values in patients are more likely to fall below the lower limits of normal and thus get classified as 'restrictive.' The misclassification was also greater in taller subjects. Yun et al.  have reported earlier that growth rates of arm span were greater than those of height in general. In agreement with this, we also observed the difference (arm span − actual height) to increase with height and peaking when the latter reached 150 cm. Thus the over-prediction of FVC on using arm span would be greater in taller subjects. This aspect of the relationship between arm span and standing height was not examined in earlier studies. , The misclassification has significant clinical implications. There is a risk of wrongly labeling a normal subject as abnormal, as well as overestimating the degree of impairment in those with a true restrictive pathology. Therefore, the practice of substituting arm span for height cannot be recommended.
The FVC predicted by the other two estimates of height was closer to that predicted from actual height, and thus overall agreement in interpretation of spirometry was also very good. Yet the limits of agreement between FVC obtained from these estimates and that obtained from actual height were fairly wide on Bland Altman analysis and therefore, the clinician needs to be careful in interpretation of spirometric data using estimated heights, especially when the values are closer to the lower limits of normal. From the results of this study, it appears that when actual height cannot be obtained, either of the two measures may be used in taller subjects. In shorter subjects, however, height predicted from arm span using regression agrees more closely with actual height than does that estimated using a fixed ratio and is therefore recommended.
The study has limitations. It was not blinded. The population of the study was divided into two groups after the study was completed. This, however, was not possible before the study as the relationship between (arm span − height) and height was not known. Further, generalization to other ethnic backgrounds may not be done as arm span and height correlations have been shown to be different among different ethnic groups. Similar studies therefore need to be carried out in other populations.
In conclusion, when actual height cannot be measured, prediction from arm span using regression analysis appears to be the most appropriate method as it predicts vital capacity closer to the actual and gives least errors in interpretation of spirometric data in patients with respiratory diseases. Direct substitution of arm span for height is not recommended as it yields greater than true predicted vital capacity, resulting in unacceptably high rates of misclassification of spirometric data.
The technical assistance of Mr. H. K. Vatsa, Mr. G. C. Chandrashekharan, Mr. Rajesh Mehta, and Mr. Ramveer Singh is acknowledged with thanks.
|1||American Thoracic Society: 1994 Update. Standardization of spirometry. Am J Respir Crit Care Med 1995;152:1107-36.|
|2||Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A, et al . Standardisation of spirometry. Eur Respir J 2005;26:319-38.|
|3||Pellegrino R, Viegi G, Brusasco V, Crapo RO, Burgos F, Casaburi R, et al . Interpretative strategies for lung function tests. Eur Respir J 2005;26:948-68.|
|4||Hibbert ME, Lanigan A, Raven J, Phelan PD. Relation of armspan to height and the prediction of lung function. Thorax 1988;43:657-59.|
|5||Leech JA, Dulberg C, Kellie S, Pattee L, Gay J. Relationship of lung function to severity of osteoporosis in women. Am Rev Respir Dis 1990;141:68-71.|
|6||Aggarwal AN, Gupta D, Jindal SK. Interpreting spirometric data: Impact of substitution of arm span for standing height in adults from North India. Chest 1999;115:557-62.|
|7||Zverev YP. Relationship between arm span and stature in Malawian adults. Ann Hum Biol 2003;30:739-43.|
|8||Torres LA, Martinez FE, Manco JC. Correlation between standing height, sitting height and arm span as an index of pulmonary function in 6-10-year-old children. Pediatr Pulmonol 2003;36:202-8.|
|9||Golshan M, Amra B, Hoghoghi MA. Is arm span an accurate measure of height to predict pulmonary function parameters? Monaldi Arch Chest Dis 2003;59:189-92.|
|10||Teramoto S, Matsuse T, Ouchi Y. Substitution of arm span for standing height is important for the assessment of predicted value of lung volumes in elderly people with osteoporosis. Chest 1999;116:1837-8.|
|11||Parker JM, Dillard TA, Phillips YY. Arm span-height relationships in patients referred for spirometry. Am J Respir Crit Care Med 1996;154:533-6.|
|12||Hickson M, Frost G. A comparison of three methods for estimating height in the acutely ill elderly population. J Hum Nutr Diet 2003;16:13-20.|
|13||Malik SK, Jindal SK, Jindal VK. Vital capacity and forced expiratory volume one second (FEV1) in normal healthy north Indian adults. Bull PGI 1987;21:179-86.|
|14||Malik SK, Jindal SK. Pulmonary function tests in healthy children. Indian Pediatr 1985;22:677-81.|
|15||Byrt T. How good is that agreement? Epidemiology 1996;7:561.|
|16||Yun DJ, Yun DK, Chang YY, Lim SW, Lee MK, Kim SY. Correlations among height, leg length and arm span in growing Korean children. Ann Hum Biol 1995;22:443-58.|