Title: Anthropometric Assessment

Key words: sonic height gauge, scales, triceps skin thickness, standing height, demi-span, surrogate measurements, limb length measures, undernutrition, sources of error

Date: August 1999

Category: 14. Measurement

Type: Article

Author: Dr van Rhijn

Anthropometric Assessment

i) The reproducibility of the different methods


The individual methods are not described as they are assumed to be familiar and are described in the literature.

Measuring height (using a sonic height gauge) and weight (scales) gave consistent results in the group.

Mid-arm circumference (cm tape), Triceps skinfold thickness (Holtain skinfold calipers), Knee height and Demispan measurements revealed variations from the same operator as well as between operators. The inter- and intra-coefficients of variation (CV%) were calculated*.

This exercise emphasises how easy it is to make errors by not adhering to exact instructions if the operator is untrained. The Triceps skinfold thickness and Knee height measurements were most prone to variation. Precision is essential to allow subsequent calculations to carry any validity.

* All measurements are noted in the attached worksheet for reference.

As part of any study, an anthropomorphic estimate of measurement error is required. The most commonly used of these measures are the technical error of measurement (TEM) and reliability (R).

A number of repeated measurements on the same subject (the same variable) can be expressed in two equations, catering for within-observer and between-observer measurements1. The coefficient of reliability, R, is then calculated to reveal what proportion of variance in a population is due to measurement error. The precision of each measurement (reproducibility or reliability) can also be expressed as a coefficient of variation (CV% = SD x 100% / mean). The sample size requirements in anthropomorphic surveys can then be estimated. It is preferable to have a trained operator carrying out the same serial measurements in order to eliminate inter-operator errors in longitudinal studies.

ii. The use of surrogate markers of height and the conventional measure of standing height in the elderly.

The methods are not described as they are assumed to be familiar and described in the literature.
The measurement was 174 cm, and consistent between the 2 observers.
  • Demi-span: Height was calculated for a female subject. Males require a different equation.
  • Height (cm) = (1.35 x demispan (cm)) + 60.1 (R = 0.74 SEE = ± 3.3 cm)

    = (1.35 x 81.5) + 60.1 = 170.1 cm

  • Knee height: Multiple measurements are taken.. Height was calculated for a young female subject. Older females and male subjects require different equations.
  • Height (cm) = (knee height (cm) x 1.86) – (age (years) x 0.05) + 70.25 (SEE = ± 7.20 cm)

    = (52.6 x 1.86) – (30 x 0.05) + 70.25 = 166.5 cm (Calliper)

    = (54 x 1.86) – (30 x 0.05) + 70.25 = 169.2 cm (Sonic height gauge)

    * All measurements are noted in the attached worksheet for reference.

    The two surrogate measurements gave acceptable approximations for height, but are prone to potential measure errors resulting in inter- and intra-coefficients of variation*. It is much more accurate to use the sonic height gauge, provided the patient can sit, to minimise error risk. They are established proxy (surrogate) measures for calculating standing height, for the latter can be difficult (unreliable) to measure in bedridden and ill or injured elderly2. Conditions such as kyphosis, vertebral collapse and loss of disc height contribute to the difficulties encountered in the elderly. Limb length measures are now regarded as superior (reproducibility & reliability) to skinfold measures.

    iii. The difficulties of using anthropometric measures in the hospitalised elderly.

    Undernutrition is prevalent and largely unrecognised in hospital in-patients3 and anthropomorphic measurements primarily infer information about body size, amounts of skeletal muscle and fatness. These inferences are only meaningful as long as the body’s systems are homeostatic, but can change with disease4 and instability. The transferability of certain models (specific populations) from one patient group to another is limited, and may not quantify risk for individual patients. Appropriate anthropomometric reference data need to be established using sound experimental design and adequate sample size for well-defined populations.

    iv. The difficulties of using anthropometry to assess nutritional status/risk in older people living in the community.

    It may be difficult to obtain accurate and recent measures of height and weight. The reference values used may not be appropriate for the population to be assessed. Severe osteoporosis may cause the bones to bow and result in spinal deformities, making measurements of height unreliable in some elderly. There are also no standardised procedures for the nutritional assessment for the elderly and the definition of normality and referral values (genetic, racial and cultural differences) is difficult5.

    v. Advantages of using anthropometric assessment.

    Anthropometry is usually non-invasive, portable, cheap, practical, quick and relatively easy to do once trained. Anthropomorphic measurements are good predictors of ill health, functional impairment and mortality.

    vi. The three major sources of error in nutritional anthropometry

    The three major sources of measurement errors are imprecision, undependability and inaccuracy6. Imprecision is due to (within- and between-observer) measurement differences. Non-nutritional factors, such as variability in height of an individual as measured at different times of the day, can be a source of undependability. Inaccuracy is a function of instrument error. Anthropomorphic measurements (eg. skinfold thickness) require skill and adequate training, and it is difficult to obtain consistent measurements in obese patients. Chronic disease and dehydration7 can affect measurements, making it difficult to quantify body composition among the elderly8.

    vii. The criteria to apply when selecting an appropriate marker for identifying malnutrition

    The value of a diagnostic test for malnutrition lies in its ability to consistently identify patients with malnutrition (Sensitivity)9 and consistently exclude those patients without malnutrition (Specificity). It should be nutrition specific (unaffected by non-nutritional factors), which is generally a tall order for any test. The test should also demonstrate normalisation after nutritional support, which has not always been the case. Single markers of nutritional status (serum albumin) have a high prognostic sensitivity, but are only statistically valid for large selected populations in epidemiological studies. The use of multivariate prognostic models (PNIs), based on the simultaneous assessment of several nutrient markers, may overcome this problem.

    Until recently, no instruments to evaluate nutritional status have been available. Two initiatives, the Public Awareness Checklist10 and the Mini Nutritional Assessment11, developed and validated in the last decade, attempt to fulfil this purpose. The requirements for an appropriate marker include being a reliable, accurate and validated scale, with defined thresholds, compatible with the skills of a generalist assessor to minimise bias introduced by the data collector12 and easily repeatable over time. It should also be specific for over 65 year olds, non-invasive and acceptable to patients, inexpensive and practical13 (quick and simple to perform and record). Standardised techniques, calibrated, serviced equipment and correctly identified body sites need to be used as well as established reference values.



    1. Garrow, J.S. Composition of the body. In: Garrow, J.S & James, W.P. Human nutrition and Dietetics. Ninth Edition. Churchill Livingstone. 1996; Ch. 2. pp. 12 - 23.
    1. 2.Tierney, A. Undernutrition and elderly hospital patients: a review. J. Ad. Nursing. 1996; 23, 228 – 236.
    2. McWhirter, P.J. & Pennington, C.R. Incidence and recognition of malnutrition in hospital. BMJ. 1994; 308, 945 – 948.
    3. Cederholm, T. & Hellstrom, K. Nutritional status in recently hospitalised and free-living elderly subjects. Gerontology. 1992; 38, 105 – 110.
    4. Gariballa, S.E. & Sinclair, A.J. Nutrition, ageing and ill health. Review article. Br. J. Nutr. 1998; 80, 7 – 23.
    5. Ulijaszek, S.J. Anthropometric measures. In: Margetts B. & Nelson M.; Design Concepts in Nutritional Epidemiology. Second Edition. Oxford University Press. New York. 1998; Chapter 10 pp 290 – 311.
    6. Weinberg, A.D. et al. Dehydration: evaluation and management in older adults. JAMA. 1995; 274, 1552 – 1556.
    7. Chumela, W.C. et al. Malnutrition or healthy senescence. Proceedings of the Nutrition Society. 1998; 57, 593 – 598.
    8. McNeil, et al. 1975. Primer on certain elements of medical decision making. New. Eng. J. Med. 293, 211 - 215
    9. Rush, D. Evaluating the nutrition screening initiative [editor’s note]. Am. J. Pub. Health. 1993; 83: 944 – 945.
    10. Guigoz, Y. et al. Mini nutritional assessment: a practical assessment tool for grading the nutritional state of elderly patients. Facts and Research in Gerontology. 1994; 4, Suppl.2, 15 – 59.
    11. Garry, P.J. & Vellas, B.J. Aging and Nutrition. In: Ziegler E. & Filer L. Present knowledge in Nutrition. Seventh Edition. ILSI Press. Washington. 1996; Chapter 40 pp 414 - 419.
    12. Charles, R.P. Nutrition assessment methods for the older Irish adult in the clinical and community settings. Proc. Nutr. Soc. 1998; 57, 599 – 602.