Endocrine Abstracts

Background: Screening algorithms for the identification of growth-disorders are routinely used in several countries. In the UK, the use of the Coventry consensus for the referral of children with suspected growth-disorders performs poorly compared to more sophisticated screening mechanisms used elsewhere. We aimed to test an algorithm developed to screen for growth-disorders in 2- to 8-year-old UK children.

Methods: The algorithm was tested using heights at referral for 153 children with growth hormone deficiency (GHD), Turner Syndrome (TS), small-for-gestational age with no catch-up (SGA), Noonan Syndrome, SHOX deficiency and hypopituitarism attending a London paediatric growth clinic. Randomly selected heights between the ages of 2 and 8 years old from 8603 children participating in the Born in Bradford (BiB) study were taken as the reference population. BiB children with known growth-affecting conditions were excluded. Standardized height-for-age (HSDS) and the child’s distance to their mid-parental height (DMPSDS) were derived. The total sample available for DMPSDS analyses was smaller due to missing parental heights. Cut-off values for HSDS and DMPSDS were selected to identify around 1.5% of children for referral, and the sensitivity and specificity to identify children with growth-disorders were obtained.

Results: Both HSDS (n=8756) and DMPSDS (n=7588) had good diagnostic accuracy with cut-offs of −2.5 for HSDS and −2.1 for DMPSDS having high specificity (99% and 99%) and acceptable sensitivity (75% and 65%). Applying either cut-off selected 1.4% of BiB children and identified 81% of children with growth disorders. An analysis of children with GHD only (n=86) had good diagnostic accuracy for the same cut-offs with sensitivity of 72% for HSDS, and 60% for DMPSDS.

Conclusions: Results for the use of a growth-screening algorithm in a UK clinical sample are promising. However, there is a need to obtain data from children before a growth-disorder is suspected, rather than at referral, to understand the true potential of this algorithm to identify children for further investigation for growth-disorders. Although children with known health conditions were excluded from the BiB sample, it is likely some children will have subsequently developed growth-disorders. Longer follow-up would help improve estimates.