Prediction of stature from knee height for black and white adults and children with application to mobility-impaired or handicapped persons

Abstract

Objective: No adequate methods exist for predicting stature to help assess the growth of handicapped (including mobility-impaired) children or to help in developing indexes of obesity or equations for estimating basal energy expenditure for adults. In this project, equations were developed to predict stature in white and black adults and children from nationally representative samples for application to mobility-impaired and handicapped persons.

Samples: Samples of healthy adults (n = 5,415) and children (n = 13,821) were selected from cycles I, II, and III of the National Health Examination Survey (NHES) conducted by the National Center for Health Statistics from 1960 to 1970. Balanced validation and cross-validation groups were created with regard to age, ethnic group, and sex.

Outcome measures: The NHES is the only national survey that contains body measurements biologically appropriate for predicting stature. These measurements include stature, sitting height, knee height, and buttocks to knee length.

Statistical analysis: Equations were computed from an all-possible-subsets of weighted regression procedure to select the predictor variables in the validation group based on the values of R2 and the root mean square error.

Results: Knee height predicted stature for white and black men, but the predictor variables for white and black women were knee height and age. For predicting stature in children 6 to 18 years of age, the predictor variable was knee height for all children.

Applications: The equations presented here were developed for use with mobility-impaired or handicapped persons, but the measurements were collected from ambulatory people by means of standard techniques. The use of recumbent anthropometric data from mobility-impaired or handicapped persons in the equations will expand the errors of prediction over those presented in this report. The standard error for a person is a guide to the range of probability within which a predicted value can occur.