Journal of Rehabilitation Medicine 51-2 | Page 7

82 J. Schröder et al. footplates and therefore there is no foot clearance during the swing phase. • The exoskeleton Lokomat (38, 45, 49) is a robotic- driven orthosis consisting of actuators applying motion to the hip and knee joints of the patient to guide locomotion in a pre-programmed kinematic trajectory based on characteristics of a healthy sym- metrical gait pattern. There is an exception to this division. Ochi et al. (46) investigated a treadmill-based system including robotic arms, which guide the thighs and legs to reproduce a physiological gait pattern. As this system resembles the characteristics of before-mentioned exoskeletons (i.e. precise control of kinematics in the hip and knee joints), it will be accounted as such in the analysis. These training modalities are compared with a control group, which is provided with conventional physiotherapy. This includes pre-gait exercises aiming at paretic leg strengthening and sitting balance. If pos- sible, manual-assisted over-ground balance and gait training was provided. However, the exact content of the control intervention throughout the included studies is poorly described. Few studies provided detailed information on the therapy dose. Ada et al. (36) documented that partici- pants were able to walk 129 m during the first session of BWSTT compared with only 26 m in the control group. Tong et al. (39) documented that participants performed 500–1,000 steps during a session using an end-effector robot and during conventional therapy 50–100 steps only. Pohl et al. (35) found that partici- pants walked with the same device 851 to 1,076 steps, similar to the results of Morone et al. (43). In addition, Peurala et al. (48) found that, with robot assistance, participants were able to initially walk 20 min without needing resting breaks, while none in the control group were able to. A similar documentation on exoskeletons is lacking, but authors declared that the exoskeleton allowed patients to practice walking at much higher doses compared with the control condition (45, 49). Overall therapy dose, as measured by the total augmen- ted time spent walking in the experimental group, is found to vary between 300 and 960 min. Most studies provided additional 400 min of walking practice in 20 sessions over 4 weeks, meaning that 5 training sessions were provided weekly (see Table II). Descriptive analysis of the feasibility In total, 53 patients dropped out of the experimental group, while 55 dropped out of the control group (see Table III). The great majority of drop-outs were unre- lated to the intervention (e.g. scheduling interference). In a single study, adverse events were reported without any difference between experimental and control group (43). In addition, few studies reported minor events caused by training, such as discomfort due to the harness (47), hypotension (43), pain (36, 43) or pres- sure sores (45), which led to a temporal discontinuity of the intervention. However, no study documented a significant difference between groups in the occurrence of such events (see Table III). Quantitative analysis of the effectiveness The following outcome measures on the comparative effectiveness of repetitive gait training were detected and classified according to the ICF. Table III. Extracted data from included studies on feasibility and effectiveness on gait-specific outcomes, as documented in the published article Experimental/Control group Activity a (i.e. walking ability) Body function a ID Adverse events Drop-outs Walking Independence post- intervention Walking Independence Walking Walking follow-up Speed Endurance Motor control Peurala et al. (48) 2009 Tong et al. (39) 2006 0 5/3 × × × 0 0/4 □ □ □ 4/3 12/9 □ * □ * × × Chua et al. (40) 2016 0 7/13 Pohl et al. (35) 2007 Chang et al. (38) 2012 0 0 5/6 1/3 Han et al. (49) 2016 0 0/4 Schwartz et al. (45) 2009 Ng et al. (42) 2008 Morone et al. (43) 2011 Morone et al. (44) 2012 0 4/2 Ochi et al. (46) 2015 0 Franceschini et al. (47) 2009 0 Ada et al. (36) 2010 0 0/0 9/3 4/2 Dean et al. (37) 2010 Nilsson et al. (41) 2001 4/3 0 × □ × □ × × □ × × □ * × × × × × × □ × □ □ Muscle strength □ × × × × × × × × × × × *Indicates the low motricity group, i.e. the group with more motor impairments at baseline. The studies Tong et al. 2006 and Ng et al 2008, Morone et al. 2011 and Morone et al. 2012, and Ada et al. 2010 and Dean et al. 2010 are dependent as they investigated the same dataset. a Results for gait-specific outcome are reported as stated in the original article. These might differ from results of the meta-analysis due to differences in statistical methodology: (x), neutral or uncertain effect; ( □ ), beneficial effect or likely to be beneficial. www.medicaljournals.se/jrm