Falling is common among individuals with incomplete spinal cord injury (iSCI), with most falls occurring while walking. Falls result in injuries (e.g., broken bones), hospital readmission, and reduced participation in work and recreation. In able-bodied people, falls can be prevented by taking one or more rapid, reactive steps. People with iSCI, however, have difficulty taking the reactive steps needed to prevent a fall. Research in the elderly and people with stroke has shown that repetitive training of reactive steps in a safe environment improves this balance reaction and prevents falls. The investigators will examine the feasibility and effectiveness of reactive step training in people with iSCI. The main objective is to determine if reactive balance training leads to greater improvements in balance reactions, scores on clinical scales, and fall rates compared with conventional walking training. A three year, pilot randomized clinical trial (RCT) will be completed. By improving balance and reducing falls, people with iSCI will experience fewer complications (e.g., injuries), and greater recovery of function and community participation.

The research aim is to determine if perturbation-based training (PBT) leads to greater improvements in balance reactions, scores on clinical scales, and fall rates compared with conventional gait training. The investigators hypothesize that individuals with iSCI who participate in PBT will show significantly greater gains in: 1) the ability to take a reactive step during the lean-and-release test, as evidenced by a reduction in both the time taken to initiate the reaction, and the number of steps taken to regain balance; and 2) clinical measures of gait, balance, and balance self-efficacy. The investigators also hypothesize that individuals who participate in PBT will experience fewer falls in the 6 months following training than the conventional training group.

All 24 participants with iSCI will complete 2 baseline assessments spaced 2 weeks apart. The participants will then be randomized into perturbation-based training and conventional gait training (12 participants/group) using a matched pairs design. Participants will be matched based on AIS level (C or D) and age (<60 years or >60 years). For both programs, the dosage will be 3 sessions/week for 8 weeks, with each session lasting 1 hour. After training, follow-up assessments, including the lean-and-release test and clinical measures, will be completed at 3 and 6 months post-training. PBT and conventional gait training will be delivered by a graduate student with a background in Physical Therapy or Kinesiology and supervised by a licensed physical therapist. PBT will consist of repetitive practice of reactive stepping for 1 hour. Participants will experience 60 perturbations per session during standing and walking activities, such as standing on a compliant surface or with eyes closed, stepping on targets on the ground, or stepping sideways. To create a perturbation, the researcher will apply unexpected pushes or pulls to a safety harness around the participant's trunk. The perturbation will be sufficient in magnitude to elicit a stepping response from the participant. Conventional training will consist of treadmill (0.5 hr) and over-ground (0.5 hr) gait training. On the treadmill, participants will walk as fast and as long as possible. Body weight-support may be used, if needed. During over-ground walking, participants will practice similar activities to the PBT group, but no perturbations will be applied. After the training period, participants will be followed for 6 months to track the occurrence and consequences of falls. If a participant experiences a fall, he/she will be asked to complete a short survey (online or on paper) within 24 hours of falling. A researcher will maintain monthly contact with participants to ensure falls are documented. Assessments will be performed twice during the baseline period, and after 4 and 8 weeks of training, and 3 and 6 months post-training for those participating in the RCT. The lean-and-release test will be administered by the research team due to the technical expertise required. All other measures will be administered by licensed PTs who are blind to study purpose and training allocation. The lean-and-release test will quantify the ability to take reactive steps. Participants will stand on a force plate and lean forward with 5-10% of their body weight supported by a horizontal cable. The cable is released at an unpredictable time, and participants must take a reactive step forward onto a second force plate. Ground reaction forces and surface electromyography of the legs (rectus femoris, vastus lateralis, biceps femoris, soleus, medial gastrocnemius, and tibialis anterior) will be recorded. The lean-and-release test has good test-retest reliability in older adults. The Mini-Balance Evaluation System's Test assesses four balance control systems, including postural responses to perturbations. This scale has been used in SCI, and has shown good validity and reliability in stroke and Parkinson's disease. The Activities-specific Balance Confidence Scale is a questionnaire on balance self-efficacy. Participants rate their confidence in their ability to maintain balance while performing 16 functional tasks. It has been used in individuals with SCI and has good psychometric properties in stroke. Spatio-temporal gait parameters, such as step width and double support time, will be collected with the Zeno Walkway (ProtoKinetics).

All electromyography (EMG), force plate and motion capture data will be sampled at 1 kHz. The EMG signals will amplified and bandpass filtered between 20-500Hz. EMG and force plate signals will be used to determine the timing of the reactive step, elicited by the lean-and-release test. The magnitude of activity in each muscle during the reactive step will evaluated using the root mean square of each muscle's EMG. The Shapiro-Wilk test will be used to test the assumption of normality. To compare the outcomes of the 2 training programs, a one-way repeated measures ANOVA or Friedman's Test will be used. The alpha level will be set at 0.05.

Inclusion Criteria:

• individuals with SCI rated as C or D on the American Spinal Injury Association Impairment Scale (AIS)
• chronic SCI (i.e., >1 year post-injury)
• a traumatic or non-traumatic and non-progressive cause of SCI
• ≥18 years old
• able to stand for 30s unsupported
• moderate level of trunk control as evidenced by the ability to reach forward >2 inches with the outstretched arm in standing (i.e., score of 2 on Berg Balance Scale Reaching Forward Task
• free of any other condition besides SCI that significantly affects walking or balance (e.g., no vestibular disorder, significant vision loss, stroke)
• no known, untreated orthostatic hypotension, hypertension, or atrial/ventricular arrhythmia

Exclusion Criteria:

• severe spasticity in the legs
• contractures in the lower extremity that prevent achieving a neutral hip and ankle position, or extended knee
• a pressure sore (>grade 2) on the pelvis or trunk where the harness will be applied
• a prior lower extremity fragility fracture.