A Paired Mixed-Effects Analysis of Lumbar Shear and Compression Forces in Manual Lifting

The Revised NIOSH Lifting Equation and the screening practices built around it interpret lumbar injury risk primarily through axial compression, yet anterior–posterior shear is associated with injury at loads below established compressive thresholds. L4 and L5 as are henceforth defined as the 4th and 5th lumbar vertebrae Using a harmonized database of 106 peak-loading lifting postures reconstructed through digital human modeling across five studies, we developed parallel families of linear mixed-effects regression models for L4/L5 compression and shear that shared identical predictor sets and matched random-effects structures, isolating biomechanical differences from differences in statistical specification. Load magnitude and horizontal hand reach acted as universal drivers of both outcomes—positive, highly significant, and stable across model reduction—although compression was several times more sensitive in absolute terms. The two components diverged in their posture sensitivities: knee flexion was significant only for shear, whereas trunk flexion, shoulder flexion, and vertical hand position acted predominantly on compression. Critically, no predictor significant in both families reversed sign, indicating that posture selectively modulates two independently sourced quantities—the moment arm of the supported mass, which governs compression, and the orientation of the lumbar segment, which governs shear—rather than redistributing a single conserved spinal reaction. Between-subject variability consistently exceeded between-task variability and persisted across model reduction, identifying individual lifting strategy as a structural determinant of spinal exposure. Among the 102 observations below the 3.4 kN compression criterion, 11 (10.8%) exceeded the lower 700 N shear reference, although none exceeded the upper 1000 N reference. Together these findings give a quantitative rationale for regarding compression-only assessment as structurally incomplete and support a multidimensional, posture-aware approach to ergonomic risk evaluation.