P0825 - Combined Skeletal Age using Hand, Elbow, and Pelvis X-Rays: A Pilot Study

Purpose: Knowing a patient’s skeletal age (SA) is important to treat diseases that are dependent on growth, such as adolescent idiopathic scoliosis (AIS). A left hand XR is the most studied and applicable to both sexes and wide age ranges. Although no gold standard exists for determining SA therefore there are several radiographic (XR) methods to assess how much a growth a child has left. Determining SA using hand, elbow, and pelvis XRs is common practice in our pediatric spine clinic but there are no published studies that evaluate the combined use of all three XR methods. Our primary objective was to determine if using three SA methods together, one method per hand, elbow, and pelvis XR, correlates better with chronological age (CA) than using any hand SA method alone.

Methods: IRB approved, retrospective, pilot study of 225 new patients (2012-2018) was performed. Inclusion criteria consisted of diagnosis of AIS, females 9-15-years-old, males 11-17-years-old, and obtained hand, elbow, and pelvis XRs on the same date. Patients were excluded if their medical history affected osseous structure formation or XRs were inadequate. There were three blinded evaluators, a senior attending, junior attending, and senior resident. Hand XRs were assessed using the Greulich and Pyle, Tanner-Whitehouse 3 (TW3), and Sanders methods. TW3 further included 2 methods, the radius, ulna, and short bones of the hand (TW3RUS), and the carpals (TW3Carpal). Lateral elbow XRs were assess using the Dimeglio method. Pelvis XRs were assessed for Risser stage and tri-radiate closure. Age was predicted using each hand method and correlated against chronological age (CA) using linear regression. This was repeated with the successive addition of the Dimeglio, Risser, and tri-radiate methods. Lastly linear regression was repeated with Dimeglio excluded.

Results: 109 patients met inclusion and exclusion criteria, 92 females (10.7-16.0 years) and 17 males (11.1-16.0 years). Inter-rater reliability for all methods ranged from ICC 0.83 for TW3Carpal, to 0.99 for Sanders staging. Due to limited data, female data was further analyzed. All hand methods showed a positive correlation, R2 0.27-0.49. Addition of Dimeglio only significantly increased positive regression for TW3Carpal, R2 0.27 to 0.40. Further evaluation of Risser stages negligibly increased correlation for all methods, R2 0.47-0.51. Adding tri-radiate closure did not improve correlation for any methods, R2 0.47-0.51. Correlation did not significantly change with Dimeglio exclusion, R2 0.44-0.50.

Conclusion: With the exception of the TW3Carpal method, using all 3 XRs, hand, elbow, and pelvis to determine SA did not significantly improve correlation with CA compared to using a hand XR alone. Therefore only a hand XR is needed to determine skeletal maturity with Sanders staging simple to clinically apply.

Study type: Cross-sectional retrospective

Level of evidence: III