Appendix.

<div><p>Background</p><p>Treadmill belt perturbations have high clinical feasibility for use in perturbation-based training in older people, but their kinematic validity is unclear. This study examined the kinematic validity of treadmill belt accelerations as a surrogate for overground walkway trips during gait in older people.</p><p>Methods</p><p>Thirty-eight community-dwelling older people were exposed to two unilateral belt accelerations (8 m s<sup>-2</sup>) whilst walking on a split-belt treadmill and two trips induced by a 14 cm trip-board whilst walking on a walkway with condition presentation randomised. Anteroposterior margin of stability (MoS), number of falls, and trunk and lower limb kinematics were quantified for the step prior and five recovery steps following the treadmill perturbations and the walkway trips which elicited elevating and lowering strategies.</p><p>Findings</p><p>Rates of falls following the treadmill accelerations and walkway trips were 0% and 13.1%, respectively. MoS was similar during the first recovery step (<i>P</i>>0.05) but less negative during subsequent recovery steps following treadmill belt accelerations than walkway trips (<i>P</i><0.01) regardless of recovery strategy. Excluding the first recovery step in the lowering strategy, recovery step lengths, toe clearance, maximum trunk, hip and knee angles (<i>P</i><0.05) were smaller during recovery on the treadmill compared to the walkway.</p><p>Interpretation</p><p>Destabilisation by treadmill belt accelerations quickly dissipated after only one recovery step but continued for multiple recovery steps following walkway trips. Smaller trunk displacement, step lengths, toe clearance and no falls on the treadmill indicate treadmill belt accelerations may not accurately simulate the biomechanical challenge of obstacle-induced trips in older people.</p></div>