ESPE Abstracts (2018) 89 P-P1-170

Evaluating Cut-offs for Automatic Growth Screening in Swedish Children - Using the Finnish Growth Monitoring Algorithm

Lars Gelandera, Aimon Niklassona, Anton Holmgrenb, Antti Saaric, Leo Dunkeld & Kerstin Albertsson-Wiklande

aDepartment of Pediatrics, Institute of Clinical, Gothenburg, Sweden; bDepartment of Pediatrics, Halmstad Hospital, Halmstad, Sweden; cDepartment of Pediatrics, Kuopio University Hospital, Kuopio, Finland; dCentre for Endocrinology, Queen Mary University of London, London, UK; eInstitute of Neuroscience and Physiology, Gothenburg, Sweden

Background: Growth charts provide excellent help to the pediatric team in identifying abnormal growth patterns. However, the evaluation is highly dependent on the skills of the clinician. A computerized automatic screening system will add quality and patient safety in finding children with disorders affecting growth. Such screening system has been developed and tested in Finland and resulted in earlier detection of growth disorders1-3.

Aim: To examine the proportion of Swedish children that will be identified as growing abnormally using the Finnish algorithms for growth monitoring when using present Swedish reference values4.

Material: The study population was selected from the GrowUp1974Gothenburg cohort of 5111 final grade school children who were born in Sweden around 19744. The 2432 children who had longitudinal measurements around school entry, i.e. at age 5–8 years, and information about their father’s and mother’s heights, were included in the analysis. The mean age of these children were 5.9 years (S.D. 0.47) with foregoing measurement at 4.5 years (S.D.=0.58).

Methods: Height measures were analyzed using 3 screening algorithms and 99.5% screening specificity1,2: (1) against population-based height references using heightSDS ≤−2.8070; (2) for distance from target heightSDS (calculated from parental heights) ≤−2.8070; and (3) for ΔheightSDS ≤−2.8070, i.e. difference between selected heightSDS and foregoing measurement.

Results: The Finnish algorithms identified 20 children (0.8%) using heightSDS. 67 children (2.8%) were identified by ΔheightSDS. 26 children were identified by the difference from mid parental heightSDS (1.1%). Combining the selection criteria identified 74 children (3.0%).

Conclusion: In total, 74 (3.0%) out of 2432 Swedish children were identified as being short, growing slowly or being short in relation to mid parental height at school start at age 5–8 years, applying 99.5% specificity of the Finnish algorithm1,2. Using the Finnish algorithms for growth screening applied on the Swedish reference identifies a larger population for referral than what has been published from Finland1.


1. Sankilampi U, et al. JAMA. 2013 Sep 11; 310(10):1071-2.

2. Saari A et al. J Clin Endocrinol Metab. 2012. 97: E2125-32.

3. Saari A et al. JAMA Pediatr. 2015 Mar; 169(3):e1525.

4. Albertsson-Wikland K et al. Acta Paediatr 2002, 91:739-754.

Article tools

My recent searches

No recent searches.