Introduction

Patients with chronic low back pain (cLBP) are typically burdened with compromised physical function. As a result, patients adopt compensatory movement patterns which could be diagnostic of pain etiology and severity. The sit-to-stand (STS) maneuver is a discriminant functional test for cLBP, as it requires considerable postural control and places relatively high loads on the spine. While prior studies focused on discrete values such as peak joint angles, our study quantifies posture change across the entire STS maneuver using a point-of-care 3D motion analysis. Using principal component analysis (PCA) of 3D joint positions, we introduce a kinematic deviation index (KDI). We hypothesized that 3D kinematic time-series data reduced to a singular metric, KDI, could discriminate patients based on symptoms, underlying pathology, and persistence of pain, potentially representative of a biomechanical biomarker for cLBP.

METHODS

Following IRB approval, 29 cLBP patients and 22 controls were recorded performing STS during routine clinic visits by a markerless motion capture device. Filtered joint position data for the bilateral shoulders, hips, and knees were transformed in a generalized Procrustes analysis, and projected from shape space into Euclidean tangent space for statistical analysis. Subject motion during STS was represented as ordered sequences of postures in shape space over time. To assess the relationship between posture and disease state, controlled Procrustes linear models were generated using a residual randomization permutation procedure. PCA was performed on Procrustes shape coordinates in the tangent plane. Path distance, shape, and orientation were compared between disease states using Mantel tests. KDI, a novel metric representing overall postural control during the assessment, was calculated as the sum of squared distances between corresponding postures along the observed motion trajectory and a theoretical motion trajectory with the least amount of overall postural change. KDI was compared between groups using Wilcoxon tests, Kruskal-Wallis tests, and correlated with patient reported health measures using a Pearson’s correlation coefficient.

RESULTS

The overall shape of the posture change trajectory was significantly different between patients with cLBP and controls (Procrustes distance 0.27, p<0.05). Patients with cLBP were found to have significantly higher KDI than controls during both stand-to-sit (5.10 vs 3.40, p<0.001) and sit-to-stand (8.47 vs 7.22, p<0.05). Although there was no relationship between pain severity and treatment response (p = 0.15), there was significant variation in KDI among cLBP patients who responded to treatment within three months (4.79), those who did not (5.34), and controls (3.39) during stand-to-sit (p < 0.001), but not during sit-to-stand (p=0.07). There was no relationship between pain intensity scores and KDI (R=-0.03, p=0.88). During stand-to-sit, there were significant relationships between cLBP patients with disc pathology (5.31) versus other etiologies (5.02) and controls (3.39, p<0.001) as well as those with nerve compression (5.12) versus those without (4.96) and controls (3.39, p<0.01) but not during sit-to-stand.

Discussion

Dimensionality reduction techniques such as KDI may have clinical utility for cLBP patients for identifying underlying pathologies and predicting treatment response. This approach could distill large amounts of biomechanical data to a singular biomechanical biomarker for cLBP.