Comparison of reliability between the PUMC and Lenke classification systems for classifying adolescent idiopathic scoliosis.

Study Design: Comparison of 2 radiographic scoliosis classification systems by multiple surgeons.

Objective: Compare the reliability of Peking Union Medical College (PUMC) and Lenke scoliosis classification systems and analyze their differences.

Summary Of Background Data: The PUMC classification is a newly reported system based on radiographic measurements with recent popularity, while the Lenke classification is widely accepted worldwide in surgical design. Both these classification systems have their own individual characteristics, hence it is necessary to compare their reliability.

Methods: Five scoliosis surgeons independently evaluated and classified presurgical radiographs of 62 adolescent idiopathic scoliosis patients based on the PUMC and Lenke classification systems on 2 separate occasions. Radiographs were cleaned before each evaluation. Inter- and intraobserver reliabilities were quantified using Kappa statistics. Data were compared using chi2 analysis.

Results: The PUMC classification's inter- and intraobserver percentage of agreement averaged to 91.0% (Kappa coefficient 0.896) and 90.2% (Kappa coefficient 0.892), respectively. While those of the Lenke curve type classification were 86.5% (Kappa coefficient 0.808) and 87.4% (Kappa coefficient 0.826). The PUMC classification from 10 individual measurements had 17 cases (27.4%) of disagreements, while in the Lenke curve type classification, 24 cases (38.7%) had disagreements. PUMC classification normally has discrepancies between type IIb, IIc, and IId, while Lenke classification has discrepancies in curve types 1 and 2. Out of 17 inconsistent PUMC curve type cases, 7 did not affect surgical fusion levels, while in the Lenke's only 2 out of 24 cases with discrepancies did not affect fusion range selection, with an obvious statistical difference.

Conclusion: The reliability of both PUMC classification and Lenke curve type classification were categorized as good-to-excellent. PUMC classification is relatively simple, with less confusion among inter- and intraobservers, with corresponding surgical fusion guidance and planning. The mismatch of curve classification had less influence on PUMC's fusion range selection than Lenke's.