Multijoint loading patterns across supermarket manual-handling tasks: a reproducible descriptive reanalysis

Background Regional spinal, shoulder, and knee forces in supermarket handling have been reported separately. We examined whether published task-level estimates yield a shared loading pattern and illustrate region-specific trade-offs. Methods We reconstructed 26 two-handed handling conditions from published tables based on 1,479 modeled lifts in 15 workers. The outcomes were mixed-model-adjusted task estimates of peak forces, not new force measurements. We used five-indicator and equal-region principal component analyses (PCA), descriptive standardized regressions, task-design diagnostics, and leave-one-product-family-out analyses. No second-stage population confidence intervals were estimated. Results The original five-indicator PC1 explained 62.0% of variance and loaded strongly on three lumbar indicators and knee force, but weakly on shoulder force (0.284). An equal-region PCA retained a shared PC1 explaining 60.8%, with a larger shoulder correlation loading (0.632); the two PC1 scores correlated at r = 0.898. Product-specific mass aligned with lumbar compression (standardized coefficient 0.812) and knee force (0.935), whereas destination height aligned with shoulder force (0.887). Model influence and product-family exclusions identified design sensitivity but generally preserved coefficient directions. Conclusion The incremental result is a reproducible, task-level joint presentation of published force estimates and their regional trade-offs. Component interpretation depends partly on anatomical weighting. These descriptive patterns do not establish causal effects, worker-level uncertainty, injury risk, or intervention effectiveness.