The NetScut Anthropometry Calculator

John J. Welch
Georgetown University Medical Center
Department of Pediatrics
Washington DC 20007

Overview

The Netscut Anthropometry Calculator (NAC) can be used to compare individuals or groups to a growth standard.  These results can be useful in determining the nutritional status of subject and aid in determining if the subject has a growth abnormality.  Data are entered via an HTML form on the user’s computer, submitted over the Internet to a UNIX server, processed there by a PERL script, and results are returned to the user in an HTML document in tabular format.

Introduction

The primary application of anthropometry is assessment of nutritional status by comparison of observed measurements to a standard.  There have been several standards for growth, but the most widely accepted is the 1977 data set from the NCHS/Fels Institute (1,2).  This data set, while not ideal, is the present international growth standard (3).  The data include measurements of length/stature, weight, weight-for-length, and head circumference.  The NetScut Anthropometry Calculator (NAC) can be used to calculate these parameters when given the height, weight, head circumference and age of a subject.  When a new set of standards is released by the NCHS, the program will be modified to permit choice of the old or new data set for use in calculations.  The advantage of the NAC is that it can be used from anywhere in the world and has a minimum hardware requirement for the user.  Its source code is freely available and, although copyrighted, can be used, modified, and incorporated into other software according to the terms of the GNU Public License.

Input

Two versions of the individual input form exist, a Javascript-enhanced form and a plain HTML form.  The Javascript version incorporates several features not found in the plain form, but the plain form is retained for universality of use.  The Javascript form performs data-entry validation and reads the current date from the user’s computer.  Age can either be calculated from this date or entered directly.  The date-based method is preferred because an exact date is calculated from the subject’s birthday and is later converted to fractional months.  The user-provided age will usually be based on the calendar age provided by the subject, and may be rounded up or down.  This source of error can significantly distort the results (4).  The Javascript version also allows users to enter measurements in English units, but converts them to metric units prior to submission of the form.  The plain HTML form does not provide data-entry validation, cannot compute the age from birth date, and accepts only metric units. A third form exists for the processing of batched data, for instance, from an epidemiological survey. This form accepts a comma delimited ASCII file for input and outputs in similar format.

Processing

The heart of the program are the data sets which describe the length, weight, weight-for-length and head circumference as coefficients to a cubic equations.  The coefficients for length, weight and weight-for-length were published by Dibley et al (5) and the coefficients for head circumference were calculated for the NAC using the same methods.

The original data were published by the NCHS as cubic spline-smoothed percentile curves (1).  Each percentile was actually a composite of several segments, since a single spline would not describe the entire curve well.   Length as used in this program is a combination of recumbent length for subjects less than 24 months and stature for older subjects.

The height and weight curves were not normally distributed, making comparisons at different ages difficult.  Dibley et al (5) derived normalized curves from the original curves and published the coefficients for these curve.  The coefficients for normalized head circumference curves are available at the NetScut site.  Using these normalized curves, it is possible to calculate Z-scores (multiples of the standard deviation) for any given observation.  These scores are easily comparable at different ages, and are more useful as the extremes of measurement, as might be found in areas where malnutrition is present (6,7).  All other statistical parameters are based on the Z-scores.

For each observation, the median (50th percentile) value is calculated.  For instance, for a male subject of 15 months age, the median height is about 50 cm.  If the observation is greater than the 50th percentile, a Z-score is generated based on a the relative distance between the Z-scores for the 50th and 97th percentiles.  If the observation were less than the 50th percentile, the Z-score would be based on the 50th and 3rd percentile.  Both 97th and 3rd percentiles are about 1.88 standard deviations from the median, and the distances from median are computed proportional to this value.  In deriving the 3rd and 97th percentile curves from the original data, the 97th was based on the averaged standard deviations of the 75th, 90th and 95th percentiles, while the 3rd percentile curve was based on the 5th, 10th and 25th percentiles to help counter the skew present in the original curves.

If a Z-score cannot be calculated for a given observation, the program is unable to calculate the derived values, but will attempt to determine the value for which the observation is the median.  For example, if a male subject has a head circumference of 48.5 cm, but is four years old, the Z-score cannot be calculated because it is undefined above 36 months of age.  However, the program can determine that the 48.5 cm is the median value for a subject of about 18 months of age, hence this subject has a small head for age.

Several other values are based on the Z-score.  These include the percentile, the absolute value relative to the median, and the percent of median.  The percentile is calculated using an algorithm developed by Hill (8).  The weight and length curves are defined only up to about 18 years, and head circumference curves are defined up to 36 months.  Above these ages, the script will generate error messages.  The weight-for-length curves are defined between a lower limit of 49 centimeters and a gender-specific upper limit of about 140 cm.  For very large or small observed values may not correspond to any median value within these ranges, and an error message will be generated.

Output

An HTML document is returned to the user.  It is prefaced by a summary of the submitted values.  All measures are converted to metric units, and dates are converted to fractional months.  These and other values in the summary are limited to two decimal places, but were internally represented at the default precision of the PERL script.

The program is designed so that errors will be trapped.  Any value which cannot be calculated is represented in the output as an asterisk, and an error message appears below the entry.

Difference from similar programs

The program EPIANTH is part of the Epi Info program which can be downloaded from the CDC; this program supercedes the previous program CASP. Excellent documentation is available online which describes EPIANTH and some of the considerations regarding computation of anthropometric indices (9).  This program can either calculate the indices from individual subjects or from batched data.  It provides many of features of the NAC, but does not calculate the value for which the observations are median and does not perform calculations related to head circumference.  The program ANTHRO, available from the CDC/WHO, performs similar calculations on flat ASCII or dBase files.  The program FORANTH calculates some head circumference data, but relies on linear interpolation between the most adjacent smoothed percentile curves.  The NAC, however, uses normalized growth curves for all of its calculations.

Acknowledgements

I am grateful to Jeffrey R. Backstrand, Ph.D. of  New York University who developed a similar Z-score calculator in Javascript and provided useful advice at the start of this project (10).  I would also like to thank Kevin Sullivan, Ph.D. at Emory University for his helpful comments and the program and source code for the FORANTH programs.  The coefficients of the smoothed percentile curves describing head circumference are found in the head.for file.  The coefficients for length, weight, and height are from the Dibley et al.  A formatted ASCII and a Microsoft Excel-97 format file containing all of the coefficients for the normalized 3rd, 50th and 97th percentile curves (including the normalized head circumference curves derived for this project) are available on NetScut.

Links

An excellent discussion of anthropometry and its application is found online at the Anthropometry Resource Center (11).

References

  1. National Center for Health Statistics.  NCHS Growth Curves for Children 0-18 Years, United States.  Vital and Health Statistics. Health Resources Administration.  Washington, D.C.: United States.  Government Printing Office. 1977; Series 11-no. 165.

  2. Hammill PV, Drizd TA, Johnson CL, Reed RB, Roche AF, Moore WM.  1979.  Physical growth: National Center for Health Statistics percentiles.  Am J Clin Nutr 1979;32:607-626.

  3. WHO working group. Use and interpretation of anthropometric indicators of nutritional status. Bull WHO 1986;64:929-41.

  4. Gorstein J. Assessment of nutritional status: effects of different methods to determine age on the classification of undernutrition. Bull WHO 1989;67:143-50.

  5. Dibley MJ, Goldsby JB, Staehling NW, Trowbridge FL. Development of normalized curves for the international growth reference: historical and technical considerations. Am J Clin Nutr 1987;46:736-48.

  6. Dibley MJ, Staehling N, Nieburg P, Trowbridge FL. Interpretation of Z-score anthropometric indicators derived from the international growth reference. Am J Clin Nutr 1987;46:749-62.

  7. Waterlow JC, Buzina R, Keller W, Lane JM, Nichaman, MZ, Tanner JM.  The presentation and use of height and weight data for comparing the nutritional status of groups of children under the age of 10 years.  Bull WHO 1977;55:489-498.

  8. Hill ID.  Algorithm AS 66. 1973;Appl Stat 22:424.

  9. Sullivan K, Gorstein J, Coulombie D.  Epi Info Online Manual. Chapter 23, Example: Programs for Nutritional Anthropometry.  URL = "http://www.cdc.gov/epo/epi/intro/manual/manchp23.htm"

  10. Backstrand, AR. Z-scores on the Internet.  1996;URL="http://www.nyu.edu/education/nutrition/software/zscore.htm"

  11. Bender B, Remancus S.  Anthropometry Resource Center.  1999;  URL="http://www.odc.com/anthro/"
Nutrition
Net Scut Home
Please direct all comments to: addy
Last modification: June 24, 1999