Disentangling predicted near-adult and measured adult height in pediatric growth assessment: The impact of bone age and prediction model accuracy

Background: Assessing growth outcomes in children/adolescents often relies on predicted adult height (PAH) and near adult height (NAH), but the distinction between these and true final adult height (FAH) is inconsistently addressed in the literature. Inaccuracies in height prediction and inconsistent endpoint definitions can lead to misinterpretation of physiologic growth or treatment efficacy, particularly in children with varying bone age maturity.

Objective: To critically evaluate the validity of predicted versus attained height outcomes across children with idiopathic short stature (ISS) and related pediatric conditions, and to explore how bone age status and model choice influence prediction accuracy and treatment assessment.

Methods: A structured review of 20 peer-reviewed studies (2000–2025) reporting on GH-treated and untreated children and adolescents was conducted. Studies were included if they reported NAH or AH or FAH and used standard prediction tools such as Bayley-Pinneau or Greulich & Pyle. Populations included ISS, small for gestational age (SGA), Turner syndrome, GHD, and PCOS. Data on bone age status (delayed, on time, advanced), height gains (PAH vs. NAH), and prediction discrepancies were synthesized. A forest plot was constructed to visually assess study validity based on prediction concordance, follow-up completeness, and methodological transparency.

Results: Predicted adult height consistently exceeded NAH by an average of 1.5–2.0 cm across studies. The discrepancy was more pronounced in children with delayed or advanced bone age. Prediction models performed best when bone age was on time; overestimation occurred in delayed bone age and underestimation in advanced bone age due to early growth plate closure. GH therapy response was also stratified by bone age, with delayed bone age groups showing the greatest gains. Terminological inconsistencies were noted, including misuse of subjective terms like “improvement” instead of “height gain.” Eleven studies exceeded the 0.80 validity threshold in the forest plot, indicating robust methodological alignment with growth outcome standards.

Conclusion: Adult height prediction in children and adolescents requires careful contextualization based on bone age and endpoint definitions. NAH should not be conflated with FAH, and Bone-age-adjusted models must be used to improve accuracy. Standardizing terminology and employing condition-specific prediction approaches can enhance both clinical decision-making and research clarity.