Résumé
There is inconsistent evidence suggesting that people with chronic low back pain may differ in variability of repeated trunk movements compared to people without chronic low back pain. These inconsistencies may be due to low reliability and task dependence of movement variability measures, which can be addressed using multiple movement tasks and summary measures.
Participants with and without chronic low back pain were recruited. Measurement sessions involved 30 repetitions of movements in the sagittal, transverse, and combined planes. Kinematics of the pelvis, thorax, and lumbar spine were estimated using inertial sensors placed on the sacrum and thorax. Magnitude of movement variability was quantified as the mean standard deviation of Euler angles for the thorax, pelvis, and lumbar spine across normalized cycles for each movement axes, resulting in 27 variables. Additionally, structure of variability was assessed using Lyapunov exponents for local dynamic stability, yielding 9 additional variables. Principal Component Analysis reduced the dimensionality of each variability measure (magnitude and structure). Stepwise logistic regression with principal component scores tested for differences between groups.
In the magnitude of variability analysis, four principal components were retained. The first two principal components significantly differentiated between people with low back pain and controls, accounting for 32.5 % and 14 % of the total variance, respectively. In the structure of variability analysis, no principal components were found to significantly contribute to differentiating between the two groups.
Summary measures of the magnitude, but not the structure, of trunk movement variability differentiated between people with and without chronic low back pain.
Clinical trial: NCT02059317, CPP: 2013.11.09bis Sud Méditerranée III, N° RCB: 2013-A01379-36.
•Variability measures can differentiate between people with and without chronic low back pain.•Principal Component Analysis was used to reduce dimensionality of variability data.•Stability measures did not significantly differentiate between patient and control groups.