O2131: Flex on ‘em: A Novel Flexibility Index Classification System for AIS

Background: Quantifying preoperative curve flexibility is an integral component in the evaluation and planning for surgical treatment of adolescent idiopathic scoliosis (AIS). Despite substantial research investigating determinants of scoliosis curve flexibility and its association with surgical outcomes, no standardized grading scale exists for the flexibility index metric. Previous literature has differentiated between stiff and flexible curves using arbitrary threshold values, which limits clinical consistency. The goal of this study is to develop a novel, data-driven grading scale for flexibility index values based on measures of surgical burden in AIS patients undergoing posterior spinal fusion (PSF) procedures.

Methods: Eligible patients from a single-center AIS surgical database were retrospectively reviewed. Patients aged 11-19 years who underwent PSF from 2011-2019 with at least two years of postoperative follow-up were included. Exclusions included any non-idiopathic curvatures and patients with history of prior spine surgery. Flexibility index was defined as the percentage of the major Cobb angle that was corrected in bending radiographs. Gaussian mixture modeling (GMM) with bootstrapping was applied to estimate flexibility index cut-offs. Surgical outcomes, including implant density, number of osteotomies, estimated blood loss (EBL), operative time, change in hematocrit level, and postoperative curve correction, were assessed for their association with the resulting cut-offs.

Results: The cohort included 525 patients, 84% were female with median age of 14 years with a median BMI of 21. GMMs identified three classifications of flexibility index: “stiff” (≤25%), “normal flexibility” (>25% to ≤70%), and “hyperflexible” (>70%), which was represented by 21%, 75% and 4% of the population, respectively (Figure 1). Overall surgical complexity was found to be highest in the stiff group for most intraoperative measures, followed by the normal flexibility and hyperflexible groups (Table 1). The total number of levels fused significantly decreased as flexibility increased, along with EBL, operative time, and number of implants (all p< 0.001). However, the number of osteotomies and occurrence of intraoperative complications did not statistically differ across flexibility index classifications. Postoperatively, a gradient was observed, with hyperflexible patients experiencing shorter hospital stays and greater curve correction compared to normal flexibility and stiff patients (both p< 0.05). Postoperative complication rates did not differ between groups (p=0.3).

Conclusion: This study introduces a model-based classification system for scoliosis curve flexibility based on flexibility index values. The proposed categories, “stiff”, “normal”, and “hyperflexible”, show clear clinical relevance, correlating with both surgical complexity and postsurgical outcomes for this operative AIS cohort. “Stiff” curves are associated with increased surgical burden, which may warrant enhanced preoperative planning and additional intraoperative considerations by physicians. These findings raise the potential for individualized clinical guidance, allowing surgeons to better anticipate recovery trajectories, including expected length of hospitalization and degree of postoperative curve correction, tailored to each patient’s curve flexibility classification.