What the Evidence Says

Walking practice with electromechanical devices

In its 2020 update, a Cochrane review led by Jan Mehrholz pooled 62 trials with 2,440 participants comparing electromechanical or robot-assisted gait training plus physiotherapy against usual care after stroke. People who trained with the devices were more likely to walk independently by the end of treatment, with an odds ratio of 2.14 across 38 trials and 1,567 participants. Walking speed rose by about 0.06 metres per second compared with controls. The authors rated the certainty of the main findings as moderate to low, partly because several included trials carried a high risk of bias. Read those figures as an average across many devices, settings and patient groups, not as a promise about any one system.

Who gained most in the walking trials

The same review found larger effects in people treated within roughly the first three months after stroke than in those treated later, and in people who could not walk independently at the start compared with those who already could. That pattern matters when you plan a service. A gait robot used mainly with people who already walk, months or years after their stroke, may show less benefit than the trial averages suggest. Effects at the end of treatment were also clearer than those at later follow-up. If you are writing a business case, match your expected caseload to the groups that drove the results, and plan to measure walking independence in your own patients rather than assuming the published odds will carry over.

Robot-assisted arm training after stroke

A separate Cochrane review, published in 2018 by Mehrholz and colleagues, included 45 trials with 1,619 participants. It found that electromechanical and robot-assisted arm training improved activities of daily living, arm function and arm muscle strength, with standardised mean differences of 0.31, 0.32 and 0.46, and it rated the quality of evidence as high. Training did not increase the risk of dropping out, and adverse events were rare. Effects of that size are modest, so robot sessions are best treated as a way to add structured, repeatable practice to a programme rather than as a substitute for therapist-led work. The authors still urged caution in reading the results, because trials varied in intensity, duration, type of treatment and the people enrolled.

What the reviews leave unanswered

Neither review identifies a best device type or an ideal dose. The arm review says plainly that it cannot tell which device or training frequency works best, because the trials differed in so many ways. Trial participants were also selected, supervised and assessed in ways a busy clinic may not reproduce. Before buying, ask makers which published trials used the exact model and software version on offer, how many sessions a week those trials delivered and what staffing supported them. Then plan a local audit using outcome measures your team already records, so that after an agreed review period you can judge whether the system is adding practice time or simply replacing sessions that would have happened anyway.

Sources and further reading

• Electromechanical-assisted training for walking after stroke (Cochrane review, 2020) — Cochrane Database of Systematic Reviews.
  Cochrane review of 62 trials on robot and electromechanical gait training after stroke, with certainty ratings.
• Electromechanical-assisted training for improving arm function and disability after stroke — Cochrane.
  Cochrane summary of 45 trials on robot-assisted arm training after stroke, with effects on daily activities.

Common questions

Does this evidence apply to exoskeletons for spinal cord injury?

Not directly. Both Cochrane reviews discussed here concern people after stroke. For spinal cord injury, look for studies in that population specifically, and ask makers which of them used their device, in what setting and with how much supervision.

Are these trial results relevant to Indian rehabilitation centres?

Use them as a starting point. Trial settings, staffing levels and patient groups may differ from your own centre, so compare them carefully, then collect local outcome data during an initial pilot before you commit to wider use or a second device.