Short-term and long-term variability of standard deviation scores for size in children

Abstract

Primary objective : To quantify long-term and short-term variability in the standard deviation scores (SDS's) for six skeletal size variables and body mass index (BMI) in children and to compare average values of these quantities for boys with those of girls and to make comparisons across variables. Methods and procedures : The analysis is based on measurements made regularly for 120 boys and 112 girls from 1 month until 20 years for seven variables (standing height, sitting height, leg height, arm length, biiliac width, bihumeral width and BMI) as part of the first Zurich longitudinal growth study. Variation in these scores due to variablity in the timing of the pubertal spurt (PS) is separated out by rescaling the age axis on an individual basis and comparing children with the same developmental age rather than the same chronological age. For a given child, the relationship between the value of its SDS and age is modelled as the sum of an arbitrary (child dependent) smooth function plus an error term. The long-term variability for that child is defined to be the mean square of the departures of this smooth function from its mean level while the short-term variability is defined to be the variance of the error term. Main outcomes and results : Girls' SDS scores have significantly more long-term variability than those of boys, while there is no significant difference between the sexes for short-term variability. Bihumeral width, BMI and sitting height have significantly more long-term variation than the other variables. Bihumeral width and BMI have the largest short-term variability and standing height has the smallest. Correlations between long-term variability and adult size and timing and intensity of the PS were small. Conclusions : A useful way of assessing long-term and short-term variability of SDS's, which is widely applicable has been described and applied to data relating to the growth of children. The results of this analysis are intriguing. Why is the underlying growth process of girls more variable than that of boys? Differences across skeletal parameters are also interesting and deserve further consideration.