The main trial is a 2-group quasi-experimental trial of comparing the outcome indicators between joining and not joining the 10-week personal assistive robot (PARO) training program in residents with dementia or mild cognitive impairment (MCI). The study questions are as follows:

Primary study questions:

i. Can participation of the 10-week PARO training program reduce neuropsychiatric symptoms at week 10?

ii. Can participation of the 10-week PARO training program reduce loneliness at week 10?

Secondary study questions:

iii. Can participation of PARO training session reduce pulse rate (as an indicator of stress and anxiety)?

iv. Can participation of PARO training session improve oxygen saturation (as an indicator of stress and anxiety)?

v. What are the participants' emotions and engagement in PARO training sessions?

Auxiliary study questions:

vi. Can participation of the 10-week PARO training program reduce neuropsychiatric symptoms at week 14?

vii. Can participation of the 10-week PARO training program reduce loneliness at week 14?

viii. Can participation of the 10-week PARO training program improve quality of life at week 10?

ix. Can participation of the 10-week PARO training program improve quality of life at week 14?

x. Can participation of the 10-week PARO training program improve sleep quality at week 10?

xi. Can participation of the 10-week PARO training program improve sleep quality at week 14?

xii. Can participation of the 10-week PARO training program improve cognitive functioning at week 10?

xiii. Can participation of the 10-week PARO training program improve cognitive functioning at week 14?

xiv. What are the feelings, satisfaction, and acceptance of PARO of the participants and PARO facilitators?

Study design

This is a 2-group quasi-experimental trial of comparing the outcome indicators between joining and not joining the 10-week PARO training program in residents with dementia or mild cognitive impairment (MCI) in Haven of Hope Woo Ping Care & Attention Home. Residents from the Haven of Hope Hang Hau Care and Attention Home for Severely Disabled will also be recruited and join the PARO training program, but their data will be analysed independently. The feelings, satisfaction, and acceptance of PARO of the participants and PARO facilitators.

Procedures

Training of PARO facilitators

Nurses, social workers, occupational therapists, health care assistants and program workers of the service units will be recruited and trained as PARO facilitators. The formal PARO facilitator training lasts for about 3 hours. An intervention protocol will be provided for the facilitators in running the training sessions.

Participants' recruitment and consent

In the residential care homes under Haven of Hope Christian Service (HOHCS), the care staff and nurses will check the eligibility of the elderly and invite eligible participants to join the trial. Residents scoring above the cutoff for dementia in Montreal Cognitive Assessment (MoCA) and not receiving a medical diagnosis of dementia or other mentally incapacitating disease will be classified as competent in giving consent. For residents scoring below the cutoff in MoCA or receiving a diagnosis of mentally incapacitating disease, we will follow the criteria stated in the Alzheimer Europe Report (2011) to ensure that they are provided chances to give consent. In particular, four criteria will be evaluated, including 1) the person must have sufficient capacity to understand the information, 2) the person is able to retain, use and weight up such information for making a decision, 3) the person will understand the benefit, risk, and convenience, 4) the person must have the ability to communicate the decision. The University of Hong Kong (HKU) research staff and the care staff will first commit them in a causal chat to assess their ability to communicate and understand the dialogues.

If residents are competent in giving consent, both residents and their family members, will be approached for consent. In cases residents are not competent in giving consents, only their family members will be approached for consent. They will then be referred to join a 1-to-1 study introduction session. In this session, HKU research staff and service unit staff will introduce the study to the participant, and invite them to sign a consent form.

After the consent is obtained, the HKU research staff will assist the participant to complete the UCLA 3-Item Loneliness Scale (UCLA 3-item), Mini-Mental State Examination (MMSE), the 5-level EQ-5D version (EQ-5D-5L), and Insomnia Severity Index (ISI), and assist the care staff to complete Neuropsychiatric Inventory Questionnaire (NPI-Q). Participants will be allocated to the experimental group or the control group according to their floor of the residential care homes. The participants in the experimental group will be arranged to join the PARO intervention in Week 1 to 10. The control group will be arranged to attend 2 PARO training sessions after all assessments are completed. As a voluntary participation, the participants have the right to withdraw from the study and intervention any time without consequences.

Interventions

Training sessions for experimental group

Each session aims to facilitate engagement, communication and interaction with the robots and other participants. Each session will take place for 30-45 minutes, twice a week over a 10-week period. Each session will be attended by a cluster of 4 to 6 participants. In each session, several activities with PARO will be supervised by the facilitator including baby-sitting PARO, feeding PARO, grooming PARO, etc. The first session will be used as a pilot run to revise our test protocol. The first three group sessions will be primarily focused on enhancing the intimacy between the group participants and their therapeutic relationship with the PARO companion robot. Participants will be introduced with the PARO training program and each other, followed by naming PARO. Fun activities such as informing each other of one's Chinese zodiac and animal's name matching game will be used to develop their friendliness and rapport towards PARO. In the third session, the facilitators will show seal photos and videos (*subject to the availability of audio-visual (AV) equipment in the care homes) so that participants will be able to recognise PARO.

Facilitators will facilitate participants' interaction with PARO for 7 main activities, namely: (1) hugging PARO: (1a) touching and hugging PARO; (1b) comparing PARO with a pet; (2) expressing emotions to PARO: (2a) selecting seven emotional stickers and talking about emotions; (2b) observing positive/negative reactions of PARO; (3) remembering PARO: (3a) talking about PARO's name and seal; (3b) becoming accustomed to PARO's movements, noises, etc; (4) feeding PARO: (4a) searching animal's feed, (4b) sharing food and snacks, (4c) feeding PARO; (5) sleeping PARO: (5a) sing favourite song or singing and clapping song, (5b) putting PARO to sleep with a cradle song; (6) bathing PARO: (6a) cleaning hands, (6b) cleaning PARO, (6c) combing PARO's hair; (7) decorating PARO: (7a) decorating PARO with ribbon, scarf, cap, etc., (7b) decorating oneself with favourite things.

Free-play sessions

For the experimental group, during the 10-week training program, to increase the exposure of participants in experimental group to PARO, two 15-30 minutes free-play sessions in each week are conducted on days when there are no training sessions. In these sessions, every participant can interact with one PARO without instructions but are under the safety supervision of a PARO facilitator.

Intervention for control group

Control group will receive no PARO-related intervention from week 1 to week 10. They will attend 2 training sessions after all assessments are completed. Activity content in these training sessions will be similar to those for the experimental group.

Data collection

HKU research staff who are blinded to the group allocation will administer all the questionnaires at baseline, 10-week and 14-week. For experimental group, after each training session and free-play session, PARO facilitators will complete Engagement of a Person with Dementia Scale (EPDS) for every participant. After each training session, they will complete two items about compliance and non-compliance with instructions for every participant. There were three selected training sessions in every cluster in the beginning phase, middle phase and ending phase that biomarkers will be collected and process observed by HKU senior research assistant. The biomarkers of pulse rate and oxygen saturation of the participants (as an indicator of feelings of stress and anxiety) will be measured using pulse oximeter before and after each of the three selected training sessions. The emotions of every participant and any special events during the three selected training sessions will be documented by HKU senior research assistant using Observed Emotion Rating Scale (OERS). To reduce the possible impacts brought by the presence of research team, the observer will introduce himself before and after the residents move to the activity room. Also, the observer will not interfere into the training activities.

In case of events (e.g. COVID-19) that prohibit HKU research staff to conduct assessment on site, assessment will be conducted via online communication apps (e.g. Zoom), and processes of training sessions will be video-recorded,

Qualitative interview

To collect feedback towards the PARO companion robot intervention, we will use purposive sampling to select 10 participants and at most 6 facilitators to conduct semi-structured qualitative interviews to collect opinion on their satisfaction and perceived acceptance on PARO features. An interview guide with open-ended and iterative questions will be used to probe for more experiences from the interviewees. Each interview will be conducted by a trained research assistant and will last about 30 minutes.

Blinding

Participants, training session moderators and observers of PARO sessions cannot and will not be blinded to the intervention. Assessors of the follow-up outcomes and the research analysts will not be involved in the recruitment and intervention delivery and will be blinded to the group allocation (single blindness).

Sample size determination

The sample size is estimated by the current number of dementia/MCI residents in the test sites. In the Woo Ping Care & Attention Home, 52 residents have been diagnosed dementia/MCI. In the Care and Attention Home for Severely Disabled, 5 have been diagnosed. Therefore, the study will include 57 residents in total.

Data analyses

Main analysis

Linear mixed model analysis will be used to test the interaction between group and time. A partial-eta square will be used to estimate the effect size of the outcome indicators in comparing the intervention group with the control group.

Process evaluations

For the biomarkers, linear mixed model analysis will be used to test the interaction between group and time. A partial-eta square will be used to estimate the effect size of the outcome indicators in comparing the intervention group with the control group. For the rated scales, descriptive statistics will be used to show the changes of emotions and incidence of negative emotions.

Qualitative interview

The interview content will be transcribed verbatim in Chinese for further analysis. We will analyze the qualitative interview transcripts using framework analysis to construct a coherent and logical structure from the classification of many opinions and perceptions of the PARO training program. The results will then be discussed and consolidated in the panel meetings with the co-authors.

Inclusion Criteria for residents in main trial:

• Is diagnosed with mild to moderate dementia by doctors
• Score at least 5 (Pan et al., 2020) and lower than the norm-derived age and education adjusted cutoff scores 7th percentile in HK-MoCA (Wong et al., 2015), non-dementia patients

Exclusion Criteria for residents in main trial:

• Have unstable physical or psychological conditions as advised by doctors or nurse in charge
• Have severe medical conditions which limit the ability to participate in the PARO sessions
• Have visual or hearing impairment that interfere with cognitive intervention
• Fear or reluctance in using robots
• Unable to communicate

Inclusion Criteria for residents in qualitative interviews:

• Have participated in the main trial
• Able to verbally communicate in Cantonese as perceived by the staff of the related home

Exclusion Criteria for residents in qualitative interviews:

• None

Inclusion Criteria of PARO facilitator for qualitative interviews:

• Have participated in the main trial

Exclusion Criteria for PARO facilitator in qualitative interviews:

• None

Research data and documentation will be available upon request.

Data will be available for 10 years.

Participants, training session moderators and observers of PARO sessions cannot and will not be blinded to the intervention. Assessors of the follow-up outcomes and the research analysts will not be involved in the recruitment and intervention delivery and will be blinded to the group allocation (single blindness).