The international growth standard for preadolescent and adolescent children: statistical considerations

Food Nutr Bull. 2006 Dec;27(4 Suppl Growth Standard):S237-43. doi: 10.1177/15648265060274S507.

Abstract

This article discusses statistical considerations for the design of a new study intended to provide an International Growth Standard for Preadolescent and Adolescent Children, including issues such as cross-sectional, longitudinal, and mixed designs; sample-size derivation for the number of populations and number of children per population; modeling of growth centiles of height, weight, and other measurements; and modeling of the adolescent growth spurt. The conclusions are that a mixed longitudinal design will provide information on both growth distance and velocity; samples of children from 5 to 10 sites should be suitable for an international standard (based on political rather than statistical arguments); the samples should be broadly uniform across age but oversampled during puberty, and should include data into adulthood. The LMS method is recommended for constructing measurement centiles, and parametric or semiparametric approaches are available to estimate the timing of the adolescent growth spurt in individuals. If the new standard is to be grafted onto the 2006 World Health Organization (WHO) reference, caution is needed at the join point of 5 years, where children from the new standard are likely to be appreciably more obese than those from the WHO reference, due to the rising trends in obesity and the time gap in data collection between the two surveys.

Publication types

  • Review

MeSH terms

  • Adolescent
  • Child
  • Child Nutrition Disorders / diagnosis*
  • Child Nutrition Disorders / epidemiology
  • Cross-Sectional Studies
  • Female
  • Growth / physiology*
  • Humans
  • Internationality
  • Longitudinal Studies
  • Male
  • Models, Biological*
  • Obesity / diagnosis*
  • Obesity / epidemiology
  • Puberty / physiology*
  • Reference Standards
  • Reference Values
  • Statistics, Nonparametric