BackgroundSpinal cord injury (SCI) is a lesion of neural elements of spinal cord. Frequently, there is motor incomplete injury and locomotor disability. Robot assisted locomotor training, functional electrical stimulation (FES) devices, and repetitive transcranial magnetic stimulation (rTMS) devices are commonly used to improve motor control and gait of patients with SCI. The efficacies of different interventions for SCI have not been reported comprehensively, and a full comparison of these strategies may benefit clinicians who apply these techniques in the rehabilitation of patients with SCI. This study aimed to comprehensively evaluate the efficacies of different interventions for SCI. MethodsWe did a systematic review and meta-analysis by searching for publications using PubMed, Embase, the Cochrane Database, the database of the US National Institutes of Health, and the WHO International Clinical Trials Registry Platform on Dec 30, 2018. Randomised controlled trials and controlled trials that compared interventions with usual care or a no exercise control, and that were published in English, were selected for our meta-analysis. The primary outcomes included functional independence (assessed using the functional independence measure, SCI independence measure, American Spinal Injury Association [ASIA] upper extremity motor score [UEMS] or lower extremity motor score [LEMS], and other hand function tests), walking speed (10 m walking test [10MWT]) and walking distance (6 min walking test [6MWT]), walking capacity (walking index for spinal cord injuries version II [WISCI II]), spasticity (modified Ashworth scale [MAS]), and quality of life of patients with SCI. The secondary outcomes included patient dropout and adverse events during intervention. Weighted mean differences and 95% CIs were used to measure intervention efficacy, and the OR was used for analysis of adverse events. Physiotherapy Evidence Database (PEDro) scale scores were used for quality assessment of the included studies. Meta-analysis was performed using Revman 5.0 software. FindingsOur initial search yielded 1410, 1643, and 452 English articles reporting the effectiveness of rTMS or FES, exercise or training, and robot assisted locomotor treadmill training on functional recovery after SCI. Of these articles, 27 randomised controlled trials and three controlled trials, including 1020 patients with SCI, were included for meta-analysis. Four articles were available for evaluating rTMS versus control (81 participants), six for FES (174 participants), nine for massed practice or treadmill training (218 participants), and 11 for robot assisted treadmill training (547 participants). rTMS improved walking speed (mean absolute difference [MD] 0·09, 95% CI 0·01–0·16, p=0·03, 10MWT) and lower extremity motor function (4·41, 1·55–7·27, p=0·003, ASIA LEMS) compared with control (sham rTMS plus gait training), but not spasticity and walking index. FES significantly increased upper extremity independence (2·92, 0·37–5·48, p=0·03) compared with control (conventional occupational therapy [COT] or resistance and aerobic training), but not upper extremity or lower extremity function and quality of life. Activity-based therapy (massed practice and treadmill training) showed no benefit to nerve injury and functional rehabilitation compared with control (no intervention, overground mobility therapy, self-regulated exercises, or conventional rehabilitation program), whereas robot assisted treadmill training significantly improved ASIA LEMS scores (5·00, 3·44–6·56, p<0·00001) and lower extremity independence (3·73, 2·53–4·92, p<0·00001) compared with control interventions (no intervention, cycling and walking), but not walking capacity (10MWT and 6MWT). No difference was seen in the incidence of adverse events in pairwise comparison of robot assisted treadmill training with control and activity-based therapy with control. InterpretationrTMS, FES, and robot assisted treadmill training are effective in rehabilitating patients with SCI. The efficacies of rTMS and robot assisted treadmill training were relatively consistent among articles. The combination of rTMS and robot assisted treadmill training was effective in the rehabilitation of patients with SCI. FundingNone.