Development of normalized curves for the international growth reference: historical and technical considerations

Abstract

The World Health Organization recommended in 1978 that the National Center for Health Statistics/Centers for Disease Control growth reference curves be used as an international growth reference. To permit the expression of growth in terms of standard deviations, CDC developed growth curves from the observed data that approximate normal distributions. Because of significant skewness, standard deviations for weight-for-age and weight-for-height were calculated separately for distributions below and above the median. Standard deviations below the median were calculated from the 5th, 10th, 25th, and 50th observed percentiles while those above the median were based on the 50th, 75th, 90th, and 95th observed percentiles. Height-for-age distributions did not show significant skewness, thus, the standard deviations were calculated based on all six of the above observed percentiles. The normalized reference curves provide a highly useful data base that permits the standardized comparison of anthropometric data from different populations. The World Health Organization recommended in 1978 that the National Center for Health Statistics/Centers for Disease Control (CDC) growth reference curves be used as an international growth reference. To permit the expression of growth in terms of standard deviations, CDC developed growth curves from the observed data that approximate normal distributions. Because of significant skewness, standard deviations for weight-for-age and weight-for-height were calculated separately for distributions below and above the median. Standard deviations below the median were calculated from the 5th, 10th, 25th, and 50th observed percentiles while those above the median were based on the 50th, 75th, 90th, and 95th observed percentiles. Height-for-age distributions did not show significant skewness, thus, th standard deviations were calculated based on all 6 of the above observed percentiles. The normalized reference curves provide a highly useful data base that permits the standardized comparison of anthropometric data from different populations.