The reliability of height and height velocity in the assessment of growth (the Wessex Growth Study)

Both biochemical and auxological measurements can be used to assess growth. Quality control in routinely reported in laboratory studies, but the reproducibility of height measurements, and the height velocity data derived from them, is seldom considered. We have previously established our error and in this report we examine its implications for the screening of short children and subsequent monitoring of their growth. The 95% confidence interval for height for a 5 year old observed to be on the 3rd centile for height, spanned the 2nd-4th centile. However, the confidence interval for a 12 month height velocity appropriate to such a child spanned the 8th-52nd centiles, the lower limit pathological and the upper more than satisfactory. A single height velocity even over 12 months therefore lacks the precision to provide a reliable index of current growth in short children. Furthermore, serial height velocity calculations on a cohort of 78 short normal children showed no significant correlation from year to year, suggesting that velocity is also unable to predict future growth. Although the proportion of this cohort of short children lying beneath the 25th centile for velocity remained constant from year to year, the identity of the individuals comprising that proportion changed, a phenomenon which could be largely accounted for by the random error associated with height velocity. Our data suggest that, by the time a trend in abnormal velocity is reliably established, a deviation from the height centiles is clearly evident. Although velocity charts are attractive in concept, they seem to be no more discriminating than height charts in practice, and may be clinically deceptive unless interpreted with great care.