Validity of the rod and frame test and the romberg ratio as methods to assess visual dependence in human standing balance

Introduction: The dependence on visual information for balance control is potentially related to balance deficits (Lord & Menz, 2000). Despite its relevance, predominantly indirect methods to assess visual dependence are being used. As gold standard, we used small visual scene movements to perturb balance, where the evoked body sway provides direct insight into the visual contribution (Central Sensori-Motor Integration test; CSMI (Peterka, 2002)). In this study, we compared subjects’ performance in the rod and frame test (RFT), a perceptual measure of visual dependence, the Romberg ratio (RR), a ratio of spontaneous sway during eyes open and eyes closed, with the CSMI test in a within-subject design. We present correlations between test results and discuss the validity of the tests.Methods: 141 subjects (20-82 years; 74 w, 63 m, 0 d) conducted a virtual reality implementation of the RFT, four 60-s long spontaneous sway measures on a force plate, as well as the CSMI test at two stimulus amplitudes. The RR was calculated dividing mean center-of-pressure velocity during eyes open by that during eyes closed on a solid surface (RR_s) and on foam (RR_f). During CSMI tests, subjects wore a VR head-mounted display, viewing a high-contrast screen that was moving in a pseudo-random pattern at two very small amplitudes (peak-to-peak 0.5° and 1°). Body sway was measured and the sway component evoked by the stimulus was parameterized using a balance control model, providing an estimate of the relative visual contribution to standing balance (visual weight) at the two amplitudes Wv_0.5 and Wv_1 (Peterka, 2002). Parameters reflecting visual dependence in each of these tests were compared using Bonferroni-corrected Pearson’s Correlations, after removing outliers based on 1.5 inter-quartile ranges of individual parameters. Several data-points had to be excluded due to either missing data or values outside the inter-quartile ranges. However, each correlation was based on at least 101 samples.Results: No significant correlations between RFT and the other tests were found. The RR_s on solid surface showed a small correlation with Wv_1 (p=.04; r=0.27), but not with Wv_0.5. RR_f did not correlate with RR_s or with Wv_0.5 or Wv_1. A strong correlation was found between Wv_0.5 and Wv_1 (p<0.01; r=0.58).Conclusions: Our results question the validity of the rod-and-frame test to reflect the dependence on visual information during standing balance. RR_s was weakly correlated with Wv. Thus, it also seems to reflect visual dependence only poorly. A likely explanation is that RR reflects changes in spontaneous sway when closing the eyes, which may affect multiple aspects of balance control outside of visual dependence. In summary, we identified large differences between methods and recommend the CSMI methods for the most valid assessment of visual dependence or for testing sensory contributions in balance control in general.Disclosure: A similar version of this abstract has been submitted to the ECSS 2026 Congress.