Abstract Virtual reality (VR) has emerged as a promising tool for assessing upper-limb motor function, yet its validity remains unclear, particularly regarding the influence of human–machine interface (HMI) design and age. This observational cross-sectional study examined whether performance patterns observed in conventional assessments are preserved across VR implementations of the Box and Blocks Test (BBT) and Nine-Hole Peg Test (9HPT), and investigated how display modality (head-mounted display vs. 2D monitor), interaction modality (haptic device vs. optical hand tracking), and age affect motor performance. Twenty-six young and twenty-six older adults completed both physical and VR assessments under four interface configurations using a within-subject Williams design. Performance differences, agreement metrics, generalized estimating equations (GEE), and survival analysis were applied. Performance in VR was consistently worse than in physical conditions (all p values < 0.001 ) with large systematic biases, wide limits of agreement, and near-zero reliability, indicating a lack of quantitative equivalence. GEE analyses revealed that performance depended on the interaction between display and interaction modalities (BBT: p = 0.026; 9HPT: p < 0.001), rather than on their isolated effects, and that these interactions were further modulated by age (BBT: p = 0.040; 9HPT: p = 0.010). These findings demonstrate that VR-based dexterity assessment is inherently interface-dependent and should not be interpreted as a direct surrogate of conventional tests. Instead, VR provides a distinct measurement environment capable of capturing multidimensional kinematic and interaction data, whose outcomes depend on both technological configuration and user characteristics.
Cisnal et al. (Fri,) studied this question.