A dementia-care review grades the evidence for Paro as low

The design

Published online 19 May 2023 and in print that September, 'The effectiveness of a therapeutic robot, Paro, on behavioural and psychological symptoms, medication use, total sleep time and sociability in older adults with dementia: A systematic review and meta-analysis' appeared in the International Journal of Nursing Studies, written by Nur Liyana Ashikin Rashid, Yijia Leow, Piyanee Klainin-Yobas, Sakiko Itoh and Vivien Xi Wu, based at Woodlands Health and the National University of Singapore, and Tokyo Medical and Dental University. The review's protocol was pre-registered on PROSPERO. Two independent reviewers searched nine databases from January 2003 to November 2022 for randomised, crossover and cluster trials testing Paro, the therapeutic seal robot developed by Japan's AIST, on people aged 60 and over with dementia in community, nursing-home or care settings; the search returned 12 trials covering 1,461 participants.

What the evidence says

The meta-analysis found a moderate effect on medication use (standardised mean difference -0.63) and small effects on anxiety (-0.17), agitation (-0.27) and depression (-0.40); the effect on total sleep time was negligible (-0.12). A separate narrative synthesis, not the pooled statistics, suggested Paro reduced apathy and increased sociability. Using the GRADE framework, the authors rated the overall quality of evidence for every outcome as low, stating this reflected 'methodological limitations, small sample size, and wide confidence intervals' across the 12 trials – the review's own words, not a stronger claim added elsewhere.

What it asks of people

A care setting adopting Paro on the strength of this review is being asked, by the review's own conclusion, to treat it as a plausible non-pharmacological supplement worth trying, not a proven treatment for a given resident; the authors explicitly say the variety of approaches across studies – group versus individual sessions, facilitated versus not – made it impossible to say which implementation actually produces the benefits measured.

Privacy and safeguards

Because Paro is a therapeutic robot without conversational AI, this review does not raise the data-privacy questions that apply to chat-based companions; it is silent on other safeguards, such as device hygiene between residents or guidance on appropriate use with a person who cannot consent to research. The authors call for 'more rigorous studies with a larger sample size... to fully understand the mechanism and effectiveness of Paro,' language that itself signals how much remains unaddressed.

• Does a moderate effect on medication use justify use even where evidence quality is rated low across the board?
• Which specific implementation – group or individual, facilitated or not – would a care setting need to test locally?
• What would a larger, higher-quality trial need to look like to move any of these ratings above 'low'?

The review is a careful, transparent accounting of a small and mixed evidence base; it supports cautious, documented trial use of Paro in dementia care, not a claim that its benefits are well established.