Treadmills have many advantages over typical overground labs for gait training. For one,they allow for continuous collection of data within a small capture volume. They can also be integrated with virtual reality systems to provide visual cues, including optic flow and real-time feedback. Treadmills can be useful for both assessment of gait and fall risk and potentially for gait retraining . For treadmill based training to improve real-world function, treadmill and overground gait must have similar underlying processes such that practice on the treadmill can effectively transfer to performance overground . While the task of walking should, in theory, be mechanically equivalent on treadmills and overground , some data suggests that there are differences, including altered kinematics [4, 5], kinetics [5, 6], and energy costs [7, 8]. This is particularly common in populations other than young healthy adults, such as patients with hemiplegia [9, 10] and healthy elderly [11–13].
Previous researchers have speculated that differences might be caused by various factors including: differences in compliance of the walking surface , subtle intra-stride variations in treadmill belt speed , the constraint of treadmill (narrow belts and railings), and unfamiliarity of walking on a treadmill . Others suggested that differences may arise from the altered sensory feedback encountered on a treadmill. In particular, visual feedback is incongruous since a visual sense of movement caused by the relative motion between an observer and the environment (optic flow) is lacking [3, 17]. Experimental variations in visual flow in healthy adults caused modulations in walking speed [18–20], stride length [19, 21], and cadence [20, 21]. Sheik-Nainar and Kaber studied the gait pattern of individuals during treadmill walking with and without a virtual reality (VR) system providing optic flow . In their study, treadmill walking lead to a flatter foot contact angle and decreased knee flexion compared to overground walking. When optic flow was added during treadmill walking, the knee flexion angle was indistinguishable from overground walking .
One population that may benefit from treadmill training in a virtual environment is individuals with amputation . The assumption of equivalence between treadmill and overground walking may, however, be questionable in this population. Persons with transtibial amputation (TTA) may have difficulty adjusting to the altered constraints of treadmill locomotion [22, 23]. They may also be more reliant on visual information than able-bodied adults due to the loss of proprioception from their involved limb [24, 25]. For patients with transtibial or transfemoral amputations, walking on the treadmill was about two and a half times more energetically costly than walking overground . Only one case-series has examined the kinematic differences between overground and treadmill walking without the use of virtual reality in persons with TTA . For two of the three participants in that study, the asymmetry of stride and stance times were reduced on the treadmill compared to overground . Small differences in peak angles at the hip and knee between overground and treadmill walking (< 2.8°) were also reported but not compared statistically, due to the small sample size.
Overall, the changes that have been reported between the two walking conditions have been small. As such, even the differences that reach statistical significance may not be physically or functionally meaningful. Riley et al. measured the coefficient of repeatability for healthy, able-bodied individuals walking overground to determine if differences observed between overground and treadmill walking were greater than expected measurement variability . The coefficient of repeatability is a measure of precision which is conceptually similar to the minimal detectable change value  that indicates the minimum level of change required to have 95% confidence that a real change occurred between conditions. Riley et al. found that, in all cases, the mean kinematic differences between overground and treadmill were less than this coefficient. The only difference in kinetics that exceeded it were knee extension moment, and anterior-posterior maximum and medial-lateral minimum ground reaction forces .
The purpose of this study was to determine if there were kinematic and/or temporal-spatial differences between overground walking and treadmill walking in a computer assisted rehabilitation environment (CAREN) in healthy control subjects or in individuals with TTA. This system has the advantage of complete immersion in an environment where the visual scene moves at the speed the subject is walking, providing appropriate optic flow. The system also consists of a wide treadmill belt to minimize changes that might occur due to any width constraint of the treadmill. We hypothesized that there would be no temporal-spatial or kinematic differences between overground walking and treadmill walking in a virtual environment for either group.