Assessment of length data from preschool children for use in epidemiological research

Abstract

The purpose of this study was to assess the quality of the length data from a large national nutrition survey for use in an investigation into the relationship between indicators of macro level food availability and child stature. The lengths and weights of children from a 5% subsample of clusters from the Third Kenyan Child Nutrition Survey were remeasured between two and five months after the initial survey. A comparison of the Z-scores of length from the two measurements on the same children revealed a positive bias overall of .10 Z-score units (or approximately .5 cm) in the original length measurements in comparison to the resurvey measurements. Bias in length also showed a curvilinear relationship with age, with three-year-olds showing the largest bias at +.31 Z-score units (or approximately 3.0 cm). The results of the investigation in~ ,random errors in length data were less clear due in part to the incorporation of growth related variation. However"a comparison of the standard deviations from the kenyan-length data with the standard deviations from the WHO growth standards indicated a much higher non-nutritional variation-in Kenyan length data.