Construction of height and weight growth charts for Iran, with an investigation of appropriate statistical methods
Data on weights and heights of children 2-18 years old in Iran were obtained in a National Health Survey of families in 1990-2, with a sampling ratio of 1 in 1000. In total 10660 households were surveyed in random cluster samples of households in all 24 provinces of Iran. The data are hierarchically structured with children in families, families in clusters and clusters in provinces. The main aim of the thesis is to use the survey data on height and weight to model growth pattern of children and adolescents in Iran and the construction of growth charts for height and weight. After removing the outliers in the data by multivariate analysis, regional variation in growth patterns were studied by constructing multilevel models. The results of these analyses showed that the data from Urban Tehran can be used as a reasonable baseline for the country, and further investigations of distributions of different centiles confirmed this. Three recently developed techniques of chart construction were compared with a chart constructed from a multilevel model to see which method produces centiles which fit the data best. the data structure has little effect on estimates of population centiles. The HRY method using spline procedure is shown to produce the best fit, and this technique has been used to construct the growth charts of weight and height for Iranian boys and girls. Checks confirm that all these curves fit the data well. However, growth of rural children differs significantly from that of children in urban Iran; a practical solution enabling one set of charts to be used for both groups of children is proposed. In view of the difference between the Tehran charts and the NCHS reference centiles, it is concluded that charts presented here should be adopted as the new reference curves for children in Iran.