Options
A MODIFIED LOCALLY WEIGHTED METHOD FOR DEVELOPING REFERENCE STANDARDS FOR HEIGHT, WEIGHT, AND BODY MASS INDEX OF BOYS AND GIRLS AGED 4 TO 18 IN TAIWAN
Resource
HUMAN BIOLOGY v.75 n.5 pp.749-770
Journal
HUMAN BIOLOGY
Journal Volume
v.75
Journal Issue
n.5
Pages
749-770
Date Issued
2003
Date
2003
Author(s)
PAN, WEN-HAM
Abstract
The objectives of this study are to modify the locally
weighted method to obtain reference values for height,
weight, and body mass indices (BMI); to describe the median
growth velocity of children aged 4 to 18 in Taiwan; and to
compare those values with data from other countries. We
modified the LOCAL method to generate empirical percentiles
and used the locally weighed regression to smooth the
percentiles. We examined the smoothed percentiles against
original data to ensure that smoothed percentiles were
within 1% of errors. We used numerical differentiation of
the 50th percentiles of height and weight to obtain the
velocity of growth in height and weight, respectively. We
applied our method to the data of the Nutrition and Health
Survey in Taiwan, 1993-1996, which was a multistaged
stratified sample. The sample included 2351 boys and 2355
girls aged 4 to 18. Our results indicate that (a) the
proposed method, which utilizes the original data,
guarantees the percentiles within 1% errors; (b) before
puberty, the height, weight, and BMI of Taiwanese children
are not much different from those of their counterparts in
Western countries; and (c) after puberty and at the stage of
young adulthood, the height, weight, and BMI of Taiwanese
youth are below the levels of those in Western young adults.
Genetic makeup and dietary patterns may contribute to these
phenomena.
weighted method to obtain reference values for height,
weight, and body mass indices (BMI); to describe the median
growth velocity of children aged 4 to 18 in Taiwan; and to
compare those values with data from other countries. We
modified the LOCAL method to generate empirical percentiles
and used the locally weighed regression to smooth the
percentiles. We examined the smoothed percentiles against
original data to ensure that smoothed percentiles were
within 1% of errors. We used numerical differentiation of
the 50th percentiles of height and weight to obtain the
velocity of growth in height and weight, respectively. We
applied our method to the data of the Nutrition and Health
Survey in Taiwan, 1993-1996, which was a multistaged
stratified sample. The sample included 2351 boys and 2355
girls aged 4 to 18. Our results indicate that (a) the
proposed method, which utilizes the original data,
guarantees the percentiles within 1% errors; (b) before
puberty, the height, weight, and BMI of Taiwanese children
are not much different from those of their counterparts in
Western countries; and (c) after puberty and at the stage of
young adulthood, the height, weight, and BMI of Taiwanese
youth are below the levels of those in Western young adults.
Genetic makeup and dietary patterns may contribute to these
phenomena.
Subjects
refference values
local method
locally weighted regression
height
weight
BMI
Type
journal article