Radiographic predictors of functional and pain outcomes in Scheuermann's kyphosis: an ROC-based minimal clinically important difference (MCID) analysis.

OBJECTIVES: This study uses receiver operating characteristic (ROC)-based minimal clinically important difference (MCID) analysis to quantify which radiographic changes best predict meaningful functional and pain improvements following posterior spinal fusion for Scheuermann's kyphosis (SK). SK is a rigid spinal deformity resulting in significant back pain and functional impairment, commonly affecting adolescents and young adults. Surgical correction aims to restore sagittal balance, but the relationship between radiographic correction and patient-reported outcomes is not well-defined.

METHODS: A retrospective cohort of 40 patients aged 15-35 with SK who underwent posterior spinal fusion and osteotomies was analyzed. Pre- and postoperative radiographs were reviewed for thoracic kyphosis (TK), lumbar lordosis (LL), pelvic incidence-lumbar lordosis mismatch (PI-LL), and T1 pelvic angle (TPA). Functional and pain outcomes were measured via Oswestry Disability Index (ODI) and Visual Analog Scale (VAS) at final follow-up (minimum 24 months for all patients), with MCID thresholds used to dichotomize outcomes; multivariable regression identified the strongest radiographic predictors.

RESULTS: Surgical correction yielded significant improvements in alignment (mean TK reduced from 85.6° to 46.3°, PI-LL mismatch from 27.5 to 10.1; p < 0.001). VAS scores improved by 1.39 points (p = 0.010) and ODI by 6.6 (p = 0.011, Wilcoxon). Clinically meaningful pain relief was achieved in the majority, with about half exceeding MCID for function. ROC analysis revealed that final PI-LL mismatch was the most predictive of functional improvement (AUC up to 0.81 for correction magnitude), with thresholds favoring near-neutral or slightly negative alignment. For pain relief, a greater degree of TK correction was the best predictor (AUC 0.60). LL and TPA showed limited predictive power for outcomes. Multivariable models demonstrated high apparent discrimination (AUC up to 0.97) but tended to overestimate mid-range probabilities due to small sample size.

CONCLUSIONS: Achieving near-neutral PI-LL mismatch and maximally reducing thoracic kyphosis are key for optimizing functional and pain outcomes in SK surgery. The study supports targeting individualized radiographic corrections over universal alignment thresholds for best clinical improvement.

STUDY DESIGN: Retrospective study.

LEVEL OF EVIDENCE: IV.