Robot-assisted training offers little improvement in severe arm weakness and function after stroke

Robot-assisted training can help stroke patients with moderate or severe upper limb impairment to perform repetitive tasks.


Why was this study needed?

Around 80 per cent of people who have a stroke experience some loss of arm function. This has a distressing and often lasting impact on their quality of life, with everyday tasks such as washing and dressing becoming difficult.

Robot-assisted training can help stroke patients with moderate or severe upper limb impairment to perform repetitive tasks. A 2018 Cochrane Review of 45 trials with 1,619 participants found significantly improved function at the end of training but the studies were generally small and the quality of evidence generally low.

Enhanced therapy programmes focus on repetitive practice of everyday tasks and have been associated with improved arm function in previous smaller studies.

Known as ‘RATULS’ (robot assisted training for the upper limb after stroke), this trial is the largest study of its kind. Its findings should be more reliable because of its size and the harmonisation of treatment protocols, devices and outcomes.

What did this study do?

RATULS was a randomised controlled trial carried out at four NHS centres in the UK between 2014 and 2018. A total of 770 adult stroke patients (average age 61) with moderate or severe upper limb functional limitation, were recruited between one and five years after their first stroke. They received either robot-assisted training using the MIT-Manus robotic gym (257 patients), enhanced therapy (259 patients), or usual NHS care (254 patients).

Robot-assisted training and enhanced therapy were provided for 45 minutes, three times per week for 12 weeks, in addition to usual post-stroke care.

After three months, improvement in participants’ arm function was measured, using the Action Research Arm Test (ARAT), which assesses the ability to perform basic tasks. This was a robust trial design, and the results should be reliable.

What did it find?

  • Significant improvement on the ARAT scale was achieved by 103 (44 per cent) of 232 patients in the robot-assisted training group, 118 (50 per cent) of 234 in the enhanced therapy group, and 85 (42 per cent) of 203 in the usual care group.
  • Compared with usual care, robot-assisted training did not improve upper limb function (adjusted odds ratio [aOR] 1.17, 98.3 per cent confidence interval [CI] 0.70 to 1.96) nor did enhanced therapy (aOR 1.51, 98.3 per cent CI 0.90 to 2.51).
  • The effects of robot-assisted training did not differ from enhanced therapy (aOR 0.78, 98.3 per cent CI 0.48 to 1.27).
  • In the trial, it was estimated that robot-assisted training cost an average of £5,387 per participant; enhanced therapy cost £4,451 per participant; usual care cost £3,785 per participant. The incremental cost per null at six months for the enhanced therapy group compared with the usual care group was £74,100, with a 19 per cent chance of being cost-effective at the usual NHS £20,000 willingness-to-pay threshold.

What does current guidance say on this issue?

NICE recommends that stroke patients are offered a minimum of 45 minutes of an appropriate therapy for a minimum of five days per week.

If a patient is unable to participate for 45 minutes of therapy, it should still be offered on five days per week, for a shorter time. If more rehabilitation is needed at a later stage, then this should be tailored to the patient’s individual needs.

What are the implications?

The RATULS trial suggests that neither robot-assisted training nor enhanced therapy would be effective if routinely used for stroke patients with moderate or severe weakness of the arm after a stroke. Although robot-assisted training can give measurable improvements in upper limb impairment, translating these into real-life changes in function and improvements in daily activities remains a challenge.

This picture may change if we discover how to target these therapies more effectively to different groups, perhaps those with milder weakness, and when best to use them in the recovery phase of stroke.

These results may help inform guidelines in the rehabilitation of people who have suffered a stroke and may help in funding decisions by healthcare commissioners. 

Citation and funding

Rodgers H, Bosomworth H, Krebs H et al. Robot Assisted Training for the Upper Limb after Stroke (RATULS): a multi-centre randomised controlled trial comparing robot-assisted training; an enhanced upper limb therapy programme; and usual care. Lancet. 2019; May 22. doi:10.1016/S0140-6736(19)31055-4. [Epub ahead of print]. This project was funded by the NIHR Health Technology Assessment Programme (project number 11/26/05).


Mehrholz J, Pohl M, Platz T et al. Electromechanical and robot-assisted arm training for improving activities of daily living, arm function, and arm muscle strength after stroke. Cochrane Database Syst Rev. 2018;(9):CD006876. NICE. Stroke rehabilitation in adults. CG162. London: National Institute for Health and Care Excellence; 2013.

The full NIHR Signal and additional expert commentary was published on 6 August 2019.

Expert view

Jane Burridge, professor of restorative neuroscience, School of Health Sciences, University of Southampton and President of the Association of Chartered and Physiotherapists in Neurology (ACPIN) and Ann-Marie Hughes, associate professor, School of Health Science, University of Southampton.

This NIHR-funded trial was the largest robot therapy trial ever completed. The findings support the Royal College of Physicians (2016) current guidelines that robots should only be offered as an adjunct to conventional therapy or in the context of a clinical trial.

This trial matters because it contributes to data sets of stroke rehabilitation. Why did this trial not show a more positive result? Firstly, robot therapy may not be effective for everyone in the chosen population. Secondly, the robot used did not allow participants to practice functional tasks, and thirdly, participants with severe arm weakness and those over one-year post-stroke may not have the potential for recovery. It may also illustrate why large randomised controlled trials that provide standardised therapy for all participants may not generate meaningful clinical outcomes. In clinical practice patients do not receive such rigid therapy because their problems are varied and variable.

We need to consider what alternative research methodologies might be more appropriate in neurological rehabilitation. Data mining of large datasets may identify what factors predict response to which therapy enabling testing of personalised rehabilitation.

Number of subscribers: 2

Log in to comment and read comments that have been added