Walking reveals a forward-shifted trunk–pelvis operating point that standing alignment conceals: trunk–pelvis coupling is preserved across radiographic balance–compensation groups in degenerative spinal disease — a cross-sectional instrumented gait analysis

Standing radiographs define sagittal imbalance but cannot capture how the trunk and pelvis behave during walking. We asked whether instrumented gait analysis separates radiographic balance–compensation groups in which standing measures collapse into two tiers and whether worsening imbalance reflects a breakdown of trunk–pelvis coupling or a preserved coupling relationship operating around a forward-shifted posture. Forty-eight adults with degenerative spinal disease underwent standing full-spine radiography and optoelectronic gait analysis during self-paced walking. Marker-based trunk inclination (TL) and pelvic tilt (PT) were quantified in both conditions; within-pass trunk–pelvis coupling was the Spearman correlation between TL and PT (negative = antiphase), benchmarked against a rigid-trunk geometric null. Radiographic classification used two axes — decompensation (SVA > 50 mm) and activated pelvic compensation (PT > 20°) — yielding three groups: balanced without compensation ( n = 15), balanced with compensation ( n = 18), and decompensated with insufficient compensation ( n = 12). Diagnostic performance was summarized by ROC analysis with Benjamini–Hochberg control of multiplicity, and correlation strength was interpreted on a published scale. Walking TL increased monotonically across the groups (7.7°, 10.3°, 16.1°; p < 0.001), whereas standing TL separated only the end point: the compensated group was indistinguishable from balanced-uncompensated patients while standing (4.3° vs. 4.7°; q = 0.79). Antiphase coupling was present in 45 of 45 patients, far exceeding the geometric null, and did not differ across groups ( p = 0.261): worsening imbalance reflected a forward shift of the operating point of a preserved coupling relationship. Each degree of standing TL increased the odds of decompensation 1.49-fold (95% CI 1.17–1.89; AUC 0.904; threshold 5.8°; sensitivity 0.92, specificity 0.73). Nurse-assessed gait instability correlated with standing SVA and TL (q = 0.008, 0.038) but not with gait parameters. Worsening sagittal imbalance during walking was a translation of an intact, antiphase trunk–pelvis pattern to a forward-shifted operating point rather than a loss of coupling. Walking, unlike standing, ordered the groups along a graded trunk-inclination gradient, although the compensated group was separated from the balanced tier only at a borderline level, and a brief standing trunk-inclination capture (≈ 5.8°) triaged decompensation. Gait kinematics discriminated radiographic compensation groups with preserved trunk–pelvis coupling in degenerative spinal disease, positioning instrumented gait analysis as a functional stress test of sagittal balance.