ESPE2022 Poster Category 1 Bone, Growth Plate and Mineral Metabolism (46 abstracts)
Background: Adult height prediction models (AHP) were designed several decades ago based on the Caucasian population, hence they are not adapted to our population’s characteristics and secular changes. Technological advances have improved the accuracy of bone age (BA) reading through automated analysis, which has been incorporated into new AHP models, but have not been evaluated in the Mexican population.
Objective: Identify which AHP model best fits the average height of the Mexican population.
Methodology: This is an analytical cross-sectional study in which 1,173 healthy children were included between the ages of 5 and 18. AHP for each participant was performed using projected height (PH), Bayley-Pinneau (B&P), Tanner-Whitehouse-2 (TW2) and the automated BoneXpert model, with both manual BA reading taken by experts and the automated BoneXpert method. The best model was considered the one showing the least difference according to the average height of adults in Mexico City (men 167.1 cm and women 155.2 cm), considering the secular trend.
Results: It was observed that all AHP models tend to overestimate between 3.2 and 11.4 cm in men and between 1.7 and 6.2 cm in women. In the Table 1 we shown the differences between manual and automated readings for each AHP model.
|Mean ±SDS||Rank (cm)||Mean ±SDS||Rank (cm)||p Value|
|HP||177.3 ± 11.2||138.1-232.7||175.5 ±7.7||155.7-205.2||0.001|
|B&P||178.5 ± 8.0||151.2-211.5||178.5 ± 6.9||163.1-211.5||0.948|
|TW2||175.2 ± 6.7||155.3-204.5||175.1 ± 6.6||158.7-204.7||0.675|
|BoneXpert||170.3 ± 5.9||149.3-195.4||171.5 ± 5.9||151.2-192.3||0.001|
|Females n 490|
|Model||Mean ±SDS||Rank (cm)||Mean ±SDS||Rank (cm)||p value|
|HP||157.9 ± 7.0||132.9-189.0||160.3 ± 6.6||136.1-191.8||0.001|
|B&P||159.1 ± 7.2||138.6-179.7||160.8 ± 6.8||140.7-180.7||0.014|
|TW2||160.2 ± 6.7||134.8-181.3||161.4 ± 6.7||136.4 180.5||0.014|
|BoneXpert||156.9 ± 6.1||136.9-174.1||158.8 ± 5.8||138.1-175.0||0.001|
|HP, projected height; B&P, Bayley-Pinneau; TW2, Tanner-Whitehouse-2; cm, centimeters; SDS, standard deviation score.|
Conclusions: The BoneXpert model best fits the characteristics of our population. On the other hand, reading of BA by experts may improve model prediction compared to automated reading. However, inter-observer variability may cause significant differences in predictions that can be reduced by the higher accuracy of automated reading.
15 Sep 2022 - 17 Sep 2022