General Session: Pediatric Spine - Hall F

Presented by: J.F. Schulz

Author(s):

T. Joshi(1), R. Hanstein(2), J.A. Gomez(2), J.F. Schulz(2)

(1) Drexel University College of Medicine, Philadelphia, PA, United States
(2) Children's Hospital at Montefiore Medical Center, Orthopaedic Surgery, Bronx, NY, United States

Abstract

Background: Current protocol for classifying adolescent idiopathic scoliosis (AIS) via the Lenke classification requires full spine standing and bending radiographs to evaluate spinal flexibility. However, the nature of bending films creates inherent variability. Pre-operative assessment may also utilize full spine prone imaging that is noneffort and less variable. Few studies have addressed the correlation of different radiographic positions, specifically the prone position to the bending position, to assess spinal flexibility. The purpose of this study was to determine if prone position radiographs can obviate bend position radiographs to determine the degree of spinal flexibility.

Methods: A retrospective review of AIS patients from August 2006 to July 2016 was performed. AIS patients who underwent pre-operative full spine radiographic imaging were included. Cobb angle (CA) of the proximal thoracic, thoracic, and thoracolumbar/lumbar curves on standing, prone and bending radiographs were measured and compared using Wilcoxon rank sum test. Spearman correlation, Fisher's exact test and Chi-squared tests were used to assess the relationship between the measures.

Results: 226 AIS patients, 77% female, with a mean age of 14.99 years (range:10.3-23.2 years) were identified. A strong correlation existed between the prone and bending CA for the proximal thoracic (rs=0.736, p< 0.01) and thoracic curves (rs=0.707, p< 0.01). A moderate correlation existed between the prone and bending CA for the thoracolumbar/lumbar curve (rs = 0.553, p< 0.01). Current methods of care to determine the flexibility of a curve and consider it nonstructural is a bending CA≤25°. Fisher's exact test and Chi-squared tests were used in order to determine if measures of CA on prone radiographs could be used instead of CA measures on bending radiographs to assess curve flexibility/structurality. For a nonstructural proximal thoracic curve, a prone CA≤25° correctly identified a bending CA≤25° 95.7% of the time (p< 0.005). For a nonstructural thoracic curve, a prone CA≤35° correctly identified a bending CA≤25° 93.3% of the time (p< 0.005). For a nonstructural thoracolumbar/lumbar curve, a prone CA≤35° correctly identified a bending CA≤25° 95.8% of the time (p< 0.005).

Conclusion: Prone positioning radiographs demonstrated a moderate to strong correlation with bending radiographs for determining the degree of spinal flexibility. Prone position radiographs may be used as a proxy for determining spinal flexibility when the CA is less than a given standard. Bending radiographs may be indicated if the CA measures greater than the given standard.