A Randomized Pilot Study On The Effects Of A Socially Assistive Robot Intervention On Surgery Patients' Engagement, Perceived Quality of Care, And Quality Of Life

doi:10.21203/rs.3.rs-5277784/v1

Download PDF

Article

A Randomized Pilot Study On The Effects Of A Socially Assistive Robot Intervention On Surgery Patients' Engagement, Perceived Quality of Care, And Quality Of Life

https://doi.org/10.21203/rs.3.rs-5277784/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Aims and Objective: The study aimed to evaluate the feasibility and preliminary effects of a SAR intervention on patient engagement, perceived quality of care, and health-related quality of life in surgical wards.

Background: The healthcare sector faces significant challenges, including workforce shortages and increasing demands. Socially Assistive Robots (SARs) have emerged as a potential solution to enhance patient outcomes, such as engagement, perceived quality of care, and health-related quality of life. However, their implementation in hospital wards remains largely unexplored.

Methods: We conducted a randomized trial in the thoracic surgery and abdominal and general surgery wards in the University Medical Center Maribor. Patients (N = 229) were allocated to either the SAR intervention group (standard care + SAR) or the control group receiving only standard care. The SAR intervention involved daily interactions for patient education and support. Outcomes included patient engagement , perceived quality of care, and health-related quality of life. We additionally explored the moderators that may alter the effects of the intervention.

Results: The overall retention rate was 78.2% (90.1% in control and 66.9% in intervention), and the overall retention of patients from baseline to post-intervention was 86.9% (90.1% in control and 83.2% in intervention) demonstrating feasibility rather than attrition. While overall changes in patient engagement were not statistically significant, the intervention group showed a slight increase compared to a decrease in the control group. Perceived quality of care decreased significantly in both groups. The SAR intervention had a significant positive effect on pain/discomfort management. The department and acceptance of robots moderated the intervention's impact on patient engagement.

Conclusions: The study demonstrates the feasibility of implementing SAR interventions in surgical wards. While the observed overall effects on patient engagement and perceived care quality were limited, the positive impact on pain management and the moderating effects of contextual factors highlight the potential of SAR in specific domains. Furthermore, we did not detect any negative effects of the intervention. Future research should consider longer intervention periods, larger sample sizes, and more department-specific applications to fully realize the benefits of SAR in surgical care settings.

Health sciences/Health care

Health sciences/Medical research/Study design/Randomized controlled trials

Health sciences/Medical research/Pre clinical studies

Physical sciences/Mathematics and computing/Computational science

Scientific community and society/Social sciences/Psychology/Human behaviour

Socially Assistive Robots

Patient Engagement

Quality of Care

Digital Health Intervention

1.1 Background

The healthcare sector is facing significant challenges, including workforce shortages and increasing demands due to population aging ¹. Europe alone is faced with a shortage of approximately 2 million healthcare professionals ². These issues can lead to higher workloads for healthcare staff, potentially resulting in burnout, reduced job satisfaction, and suboptimal patient care ^3,4. Moreover, there has been a paradigm shift in healthcare objectives, with active patient engagement and self-management emerging as primary goals in the 21st century ^5,6. Active patient engagement refers to the process of involving patients in their own care, encouraging them to participate in decision-making and take responsibility for their health outcomes ⁷. Self-management, closely related to engagement, involves patients developing the skills and confidence to manage their health conditions effectively ⁸. These concepts have gained prominence due to their potential to improve health outcomes, enhance patient satisfaction, contribute to higher health-related quality of life, and reduce healthcare costs ^9–12.

However, achieving active patient engagement and promoting self-management can be challenging ^13,14, particularly in the context of workforce shortages and time constraints faced by healthcare providers ¹⁵. Innovative solutions are essential to complement traditional approaches and support both healthcare providers and patients in achieving better health outcomes. Socially assistive robots (SARs) have emerged as a promising technology that could potentially address these challenges. These robots are designed to provide assistance through social interaction, offering services such as education, feedback, support, coaching, and companionship ^16,17. SARs are equipped with multimodal sensing capabilities, allowing them to collect a wide array of data that can be used for personalizing healthcare and responding to individual patient needs ^18–23. The potential benefits of SARs in healthcare are multifaceted, particularly in promoting patient engagement and self-management. For patients, these robots can enhance their engagement, satisfaction, and overall well-being. They can provide continuous support and education without the time constraints faced by human staff, potentially improving health literacy and patient empowerment ⁵. This constant availability could be particularly beneficial in supporting patients' self-management efforts, providing reminders, encouragement, and information as needed ^24,25. A recent study has also demonstrated that SARs may exert a positive effect on health-related quality of life in some contexts, such as in the case of socially isolated older adults ²⁶. For healthcare providers, SARs may help alleviate workload by assisting with tasks such as providing information, reminding about medication, and offering basic care instructions ^27–29.

Despite these potential advantages, the large-scale implementation of SARs in healthcare settings is progressing slowly ³⁰. The main barriers include ethical and workload-related concerns, further driven by education and technology expectations ^31,32. This may be further extended, in part, to the limited number of systematic studies conducted in real-life healthcare environments ¹⁸. While existing research shows promise, many studies have focused on specific populations such as the elderly with dementia or children with neurodevelopmental disorders ^19,33–35. Moreover, in the context of surgery wards, where patient engagement and adherence to post-operative instructions are crucial for recovery, the potential of SARs remains largely unexplored. Finally, there has been limited integration of SAR interventions with existing in hospital care pathways in previous research ^36,37.

There is a clear need for more robust studies on the use of SARs in healthcare settings. Our study explores how a SAR can be used to deliver gamified nursing education and physiotherapy, complementing standard care practices in a surgical ward. The approach provides valuable insights into how SARs can be effectively integrated into existing healthcare workflows. By providing evidence on the feasibility and potential impacts of SAR implementation on patient outcomes, our study makes a significant contribution to the field and paves the way for future definitive trials.

1.2 Objectives

While in our complementary research we focus on examining the acceptability of the SAR intervention among patients and healthcare staff ³⁸, the primary aim of the present study is to assess the feasibility and potential effects of implementing a SAR intervention in surgery wards, with a focus on patient engagement, perceived quality of care, and health-related quality of life. Additionally, we will also explore the potential moderating variables that attenuate or accentuate the effects of the intervention. The specific objectives are as follows:

To evaluate the feasibility of implementing the SAR intervention in surgical wards by determining the retention rate and adherence to the SAR intervention, and compare it with standard care.

To assess the potential effect of the SAR intervention on patient engagement,, and compare it with standard care.

To assess the potential effect of the SAR intervention on patients' perceived quality of care,, and compare it with standard care.

To assess the potential effect of the SAR intervention on patients' health-related quality of life,, and compare it with standard care.

To explore the potential moderators that impact the effectiveness of the SAR intervention on patient-related outcomes.

To estimate the effect sizes for primary outcomes to inform sample size calculations for a future controlled trial.

2.1 Trial Design

This study was carried out as a randomized controlled trial design with two arms: an intervention group and a control group, with 1:1 allocation. The trial was registered under ISRCTN Registry (ISRCTN96689284). The trial was conducted in the thoracic surgery and abdominal and general surgery wards of the University Clinical Centre Maribor, Slovenia, between June 2023 and May 2024. Patients admitted for elective procedures were randomly allocated to either group. No follow-up was foreseen. The study design is outlined in Fig. 1.

While the control group received standard care, the intervention group was additionally exposed to interactions with a SAR Frida ³⁸ during various stages of their hospital stay. As outlined in Fig. 1, the Recruitment (phase 1) was carried out at the outpatient clinic. Patients were first screened against inclusion/exclusion criteria (see section Participants). If they were eligible, they were presented the study by a dedicated recruiter and asked to sign a written letter of consent. Participants in both groups completed baseline measures upon Recruitment (phase 1) and post-intervention measures just before discharge (phase 5).

During Admission (phase 2), control group filled in the regular admission questionnaire in paper format, and the SAR greeted patients in the intervention group, explained and presented the project and offered an interactive form of filling in the admission questionnaire, through speech enabled and SAR moderated interaction. In both phase 3 and phase 4, the key difference between the intervention and control groups was the use of the SAR to deliver information, conduct assessments, and provide guidance. For the intervention group, Pre-surgery visits (phase 3) and Post-surgery visits (phase 4) consisted of daily interactions with the socially assistive robot in a dedicated Intervention Room.

2.2 Changes to the Original Design

Several modifications were made to the study methods after trial commencement, reflecting the realities of implementing the study in a clinical setting. First of all, due to time and resource constraints, we have merged the present study with the study on clinical decision support systems (CDSS) and SARs in grand rounds, primarily focused on clinician-facing outcomes and registered under ISRCTN12048782 ³⁹. This merger allowed for a more comprehensive evaluation of how both the SAR interventions and the CDSS impact patient outcomes. The intervention was expanded to include not only the socially assistive robot for patient education and support but also the voluntary use of digital health monitors (1 to 3 times a day) and of CDSS during grand rounds (a separate report is being formulated). To facilitate this, the overall study design was modified to from within subject design to a randomized controlled trial (RCT). The targeted sample was adjusted accordingly, from original 90 subjects to 200 subjects (1:1 allocation ratio). Finally, in the initial protocol we have foreseen a 1-week interval followed by 1-week washout periods between iterations. Since we were able to allocate patients in the intervention into a dedicated room, the management of the spillover effect using washout periods was no longer necessary.

2.3 Participants

Those admitted to thoracic and abdominal/general surgery wards for an elective (non-emergency) procedure were screened for eligibility regarding the inclusion and exclusion criteria. Specifically, beyond the department they were admitted to, patients needed to be aged 18 years or above, willing to participate in the study, and willing and eligible to sign the informed consent. On the other hand, the exclusion criteria included emergency patients, patients already enrolled in other studies, patients with dementia, special needs, incapable of verbal communication or appointed guardians, and patients allocated to an intensive step-down unit or regimen.

The eligible patients were allocated to the control group using a simple coin toss method using the Googles Coin Flip software. Heads meant that the patient is assigned to the intervention group (SAR intervention) and tails that the patient was assigned to the control group (standard care). A research assistant not involved in patient care or outcome assessment performed the digital coin toss using at the time of each participant's enrollment, ensuring immediate and unbiased allocation. After randomization, the patients immediately entered the study.

The initial protocol aimed for a within-subject design with 90 subjects. However, modifications to a randomized controlled trial (RCT) design and the integration with a clinical decision support systems (CDSS) study necessitated an increase in the sample size to 200, aligning with the power analysis results. The targeted sample size of 200 patients was identified by an initial power analysis conducted using G*Power software. The primary analysis plan involved ANOVA repeated measures with within-between interaction (two groups, two measurements). Assuming a small to moderate effect sizes (d = 0.10 to d = 0.40), which is conservative and aligns with the novelty of SAR interventions in this context, and using conventional significance threshold and power (α = 0.05, 1-β = 0.80), the total sample size needed to detect the effects was calculated to be 200 participants.

2.4 Measures and Outcomes

All participants were asked to fill out the baseline measure (demographic variables, willingness to use the robot, patient engagement, perceived quality of medical care, and health-related quality of life) and the follow-up measure that occurred post-intervention (patient engagement, perceived quality of medical care, and health-related quality of life). All scales previously unavailable in Slovene were translated using the translation-backtranslation procedure. Internal reliabilities of all questionnaires used in this study are reported in the results section.

Demographic variables. At the beginning of the baseline measure, participants were asked about their gender (female, male, other), age (in years), highest education level (primary school or less, lower secondary education, higher secondary education, bachelor’s degree or equivalent, master’s degree or equivalent, doctoral degree or equivalent). We also recorded the surgery department to which they were admitted (i.e., either the thoracic surgery department or the abdominal and general surgery department).

Acceptance of socially assistive humanoid robots. Individuals’ acceptance of SARs in a hospital environment was measured with four questions (e.g., “Would you like to be treated in a hospital where a robot is part of the medical team?”), answered by using a dichotomous response format (“No” or “Yes”). Participants’ answers were then averaged, with higher values indicating a higher acceptance of SARs. In our previous research ³², the scale exhibited good internal consistency (α = .82).

Patient engagement. We used the Patient Health Engagement (PHE) scale to measure patient engagement ⁷. The scale consists of nine items (e.g., »Despite my illness, I know how to manage my life«), which were answered on a 5-point agreement scale by the patients in our study (1 – »Completely disagree«, 5 – »Completely agree«). Higher scores, calculated as the average of all items, reflect higher patient health engagement. In the appendix, we also report the results obtained using only five items of the scale that constitute the short version commonly used in previous research. In previous studies, both versions of the scale exhibited good internal consistency (α = .97; ⁷).

Perceived quality of care. We used the Perceived Quality of Medical Care (PQMC) scale to measure the perceived quality of care (Richmond et al., 1998). The scale has six items, answered using a 7-point semantic differential scale. An example item is “Medical care I have received was unsatisfactory/ satisfactory”. The scale has one dimension. Four of the six items (items 1, 2, 4, and 5) were recoded so that higher average scores indicate higher perceived quality of care. In the original validation study, the scale exhibited excellent internal consistency (α = .94; ⁴⁰).

Health-related quality of life. Health-related quality of life was assessed using the EQ-5D-3L instrument ⁴¹. The scale consists of six items. The first five items correspond with five dimensions of health-related quality of life: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each item has three response levels: no problems, some problems, and extreme problems. We recoded the items so that higher scores indicate higher health-related quality of life. The items can be analyzed separately but also averaged into an index, which exhibited accepted internal reliability in previous studies (e.g., Cardoso et al., 2016). Additionally, the sixth item of the scale is essentially a visual analog scale (VAS) that records patient’s self-rated health from 0 (the worst health one can imagine) to 100 (the best health one can imagine). The Slovenian version of the scale was validated by Prevolnik-Rupel and colleagues ⁴².

2.5 Interventions

The intervention utilized the Pepper robot (SoftBank Robotics), a 120 cm tall humanoid robot known for its affordability and popularity in human-robot interaction studies ⁴³. However, recognizing Pepper's limitations in advanced communicative capabilities and support for non-mainstream languages ⁴⁴, significant customizations were made to optimize its performance for this study. The details on the technical solution, outlined in Fig. 2, are described in detail in ³⁸. The cornerstone of these enhancements for the Pepper Model was the integration of an advanced Embodied Conversational System. The system comprised of several working in concert to deliver a seamless and effective interaction between the robot and patients. The RASA Chatbot served as the driver of discourse interactions, functioning as an API with its cognitive processing engine instantiated through RASA NLU (Natural Language Understanding). To overcome Pepper's limitations in speech production, especially in the Slovenian language, the PLATTOS speech synthesis system was implemented. This end-to-end multilingual and multimodal text-to-conversation synthesis system utilized a multi-stage architecture which included speech production and gesture synthesis system. To enable the robot to understand and respond to patients' verbal inputs, the ASR SPREAD (Automatic Speech Recognition) system was integrated. A critical aspect of the intervention's design was its integration with standardized healthcare data systems, specifically through the implementation of FHIR (Fast Healthcare Interoperability Resources). The study utilized a HAPI FHIR v2 based infrastructure, focusing on key FHIR resources such as Patient, QuestionnaireResponse, and Composition. This integration allowed for all patient responses and interactions to be stored in a standardized, interoperable format.

Throughout engagement the process, as outlined in Fig. 2, the robot delivered gamified nursing education, providing critical information on wound care, home care considerations, and daily care explanations through engaging, multimodal interactions. The physiotherapy guidance component utilized both verbal instructions and physical demonstrations, leveraging the robot's capability for movement and gesture synthesis. The system also offered ongoing patient assistance for non-urgent communication and support, with all interactions carefully designed to work within the enhanced capabilities of the Pepper platform. All the engagements were supervised by a experimenter whose role was to start the experiment by inserting a patient ID and initiating the dialog flow. During the process, the experimenter observed and intervened if errors in the discourse flow occurred.

2.6 Statistical analysis

We used the IBM SPSS Statistics 28 software for statistical analysis. We first cleaned the dataset and excluded all participants who did not provide us with their demographic data and at least partly filled out the baseline measure. We continued with other preparatory steps. First, we performed tests of initial equivalence and selective dropout. Second, we investigated the internal consistencies of all multiple-item scales using the coefficient α and calculated the factor scores in accordance with the scoring instructions. Third, we checked the assumptions of the chosen statistical tests (such as the normality of the distribution assumption).

Next, we calculated bivariate correlations between all variables of interest and conducted various inferential tests to investigate our research questions and hypotheses. Changes from baseline to post-intervention within each group (control, intervention) were investigated using Paired Samples t-tests. The main analyses, in which we investigated the relative effectiveness of the intervention (standard care with robot interactions) compared to the control group (standard care) on the selected outcomes, were performed using separate mixed-design (split-plot) analyses of variance (ANOVA), with the allocated group as the between-subjects factor and time as the within-subjects factor. Additional exploratory analyses, in which we investigated the potential moderators that attenuate or accentuate the effects of the intervention, were conducted using the PROCESS MACRO extension for SPSS ⁴⁵. Using Model 1, group allocation was used as the predictor variable, gender, age, education, department, and acceptance of robots as moderator variables, and the change (T2-T1) in patient engagement, perceived quality of care, and health-related quality of life as the outcomes. Additionally, the change in patient engagement and perceived quality of care were explored as moderators of the change in health-related quality of life as well. The results were interpreted as statistically significant if the p-value was below .05. All results were accompanied by effect sizes, such as Cohen’s d.

3.1 Sample characteristics

In total, 229 surgery patients were recruited for the study, an estimated 10% retention rate was used as stop condition; 111 patients (48.5%) were allocated to the control group, whereas the remaining 118 patients (51.5%) were allocated to the intervention group. Out of 229 recruited, only 206 patients provided their demographic data and partially completed the baseline measure (control group: n = 111, 53.9%; intervention group: n = 95; 46.1%). Of these, 133 (64.6%) were treated at the thoracic surgery department and 73 (35.4%) at the abdominal and general surgery department. Most (n = 117; 56.8%) were female, while the rest (n = 89; 43.2%) identified as male during the study. Their age ranged from 20 to 85 years (M = 61.20, SD = 13.80). In terms of education, the majority of patients had finished lower (n = 64; 31.3%) or higher (n = 58; 28.2%) secondary education, followed by a bachelor’s degree or equivalent (n = 41; 19.9%), primary school or less (n = 27; 13.1%); master’s degree or equivalent (n = 11; 5.3%), and doctoral degree or equivalent (n = 5; 2.4%). The control and intervention group did not differ based on the department they were admitted to (χ2(1) = .606, p = .467), gender (χ2(1) = .305, p = .672), education (U = 5124.50, Z = -0.14, p = .888), nor age (t(204) = 0.29, p = .770, d = .04), providing support for demographic equivalence of the two groups.

179 surgery patients (86.9% of those who completed the baseline measure) at least partly filled out the post-intervention measure, 100 from the control group (55.9%) and 79 from the intervention group (44.1%). Those who filled out the T2 measure and those who dropped out did not differ in terms of their group (χ2(1) = 2.16, p = .153), gender (χ2(1) = 0.08, p = .837), education (U = 2374.50, Z = -0.15, p = .881), and age (t(204) = -0.43, p = .669), indicating the absence of selective dropout based on individuals’ group allocation and demographic variables. However, those who dropped out and those who did not differed in terms of the department they were admitted to (χ2(1) = 5.50, p = .029); specifically, while only 9.0% (n = 12) of thoracic surgery patients discontinued their participation, this was the case for 20.5% (n = 15) of abdominal and general surgery patients. The flow chart of participants can be seen in Fig. 3 below.

3.2. Descriptive statistics and bivariate associations

We first calculated basic descriptive statistics and bivariate associations for the whole sample, regardless of individuals' group allocation. The results are displayed in Table 1 below.

Table 1

*Descriptive statistics and bivariate associations*
	N	M	SD	1	2	3	4	5	6	7	8	9	10	11	12
1. Gender	206	1.43	0.50	-
2. Age	206	61.20	13.80	.11	-
3. Education	206	2.81	1.20	− .11	− .26^***	-
4. Department	206	1.35	0.48	.11	− .17^*	.15^*	-
5. Acceptance of robots	172	0.72	0.39	.05	− .07	.13	.15	(.89)
6. Patient engagement (T1)	198	4.31	0.59	.14	− .05	.08	.19^**	.15	(.91)
7. Patient engagement (T2)	174	4.26	0.59	.15*	− .09	.03	− .08	.00	.30^***	(.93)
8. Perceived quality of care (T1)	195	6.30	0.95	.01	.06	.02	− .03	− .03	.25^***	.08	(.82)
9. Perceived quality of care (T2)	177	5.91	1.28	.03	.01	.01	− .29^***	− .06	− .03	.45^***	.14	(.90)
10. Health-related quality of life (T1)	206	2.67	0.30	.00	− .15*	.15^*	.06	− .01	.33^***	.29^***	.12	.07	(.67)
11. Health-related quality of life (T2)	172	2.67	0.33	− .01	− .06	.08	− .24^***	− .08	.11	.34^***	.11	.35^***	.48^***	(.76)
12. Self-rated health (T1)	206	68.50	17.95	.08	− .19^**	.11	.03	.01	.39^***	.34^***	.10	.15^*	.39^***	.23^**	-
13. Self-rated health (T2)	172	72.12	19.00	.09	− .07	.07	− .14	.07	.24^**	.39^***	− .02	.30^***	.31^***	.42^***	.38^***
Notes. Gender: 1 = Female, 2 = Male. Department: 1 = Thoracic surgery, 2 = Abdominal and general surgery. Cases were excluded pairwise. ^* p < .050, ^ p < .010, ^* p < .001.

3.3. Effects on patient engagement and perceived quality of care

Next, we investigated the changes in patient health engagement and perceived quality of medical care from baseline (T1) to post-intervention (T2) within each group and compared the changes between groups. The results of these analyses are displayed in Table 2.

Table 2

*Patient engagement and perceived quality of care: Changes from baseline to post-intervention in the control and intervention group*
	Baseline (T1)		Post-intervention (T2)		Comparison
	M	SD	M	SD
Patient engagement
Control group (N = 95)	4.31	0.63	4.20	0.67	t(94) = -1.39, p = .167, d = − .14
Intervention group (N = 79)	4.32	0.54	4.34	0.47	t(78) = 0.31, p = .759, d = .04
Perceived quality of care
Control group (N = 97)	6.33	0.92	5.89	1.30	t(96) = -2.86, p = .005^**, d = − .29
Intervention group (N = 77)	6.36	0.93	5.93	1.29	t(76) = -2.58, p = .012^*, d = − .29
Notes. ^* p < .050, ^** p < .010.

The results related to patient engagement showed a decrease in the control group, whereas the intervention group experienced a slight increase in patient engagement. However, the changes were not statistically significant in either of the two groups. Furthermore, we did not observe a statistically significant interaction, demonstrating that the changes in the intervention group were not statistically significantly different from changes observed in the control group (F(1, 172) = 1.55, p = .215, η_p² = .01). Moreover, our results showed that the perceived quality of care decreased statistically significantly from baseline to post-intervention in both control and intervention group. The trends of change were similar in both groups (F(1, 172) = .002, p = .966, η_p² = .00).

3.4. Effects on health-related quality of life

Next, we investigated the changes in health-related quality of life from baseline (T1) to post-intervention (T2). The results of these analyses are displayed in Table 3.

Table 3

*Health-related quality of life: Changes from baseline to post-intervention in the control and intervention group*
	Baseline (T1)		Post-intervention (T2)		Comparison
	M	SD	M	SD
Mobility
Control group (N = 95)	2.69	0.46	2.69	0.46	t(94) = 0.00, p = 1.000, d = .00
Intervention group (N = 77)	2.70	0.46	2.74	0.47	t(76) = 0.62, p = .535, d = .07
Self-care
Control group (N = 95)	2.91	0.29	2.76	0.43	t(94) = -3.73, p < .001^***, d = − .38
Intervention group (N = 77)	2.94	0.25	2.84	0.40	t(76) = -2.16, p = .034^*, d = − .25
Usual activities
Control group (N = 95)	2.64	0.50	2.58	0.50	t(94) = -1.10, p = .276, d = − .11
Intervention group (N = 78)	2.60	0.52	2.64	0.48	t(77) = 0.60, p = .552, d = .07
Pain/discomfort
Control group (N = 95)	2.44	0.52	2.40	0.49	t(94) = -0.65, p = .519, d = − .07
Intervention group (N = 77)	2.36	0.48	2.51	0.50	t(76) = 2.36, p = .021^*, d = .27
Anxiety/depression
Control group (N = 95)	2.61	0.55	2.78	0.42	t(94) = 3.62, p < .001^***, d = .37
Intervention group (N = 78)	2.63	0.51	2.76	0.43	t(77) = 2.00, p = .049^*, d = .23
Whole scale (excluding VAS)
Control group (N = 95)	2.66	0.31	2.64	0.34	t(94) = -0.51, p = .610, d = − .05
Intervention group (N = 77)	2.65	0.30	2.70	0.32	t(76) = 1.31, p = .195, d = .15
Self-rated health (VAS)
Control group (N = 95)	68.05	17.91	71.31	20.36	t(94) = 1.52, p = .131, d = .16
Intervention group (N = 77)	69.42	18.57	73.12	17.25	t(76) = 1.58, p = .118, d = .18

The results showed that mobility, as one aspect of health-related quality of life, slightly increased in the intervention group, whereas it remained about the same in the control group; however, the changes were not statistically significant in either of the two groups. Similarly, we did not find significantly different patterns of change between the two groups (F(1, 170) = 0.23, p = .629, η_p² = .00). Next, quality of life related to self-care decreased in both groups, but the change from T1 to T2 was significant only in the control group. The interaction was not statistically significant (F(1, 170) = 0.95, p = .332, η_p² = .01). Quality of life related to usual activities slightly decreased in the control group and slightly increased in the intervention group, but the changes were not statistically significant. The interaction was also not statistically significant (F(1, 171) = 1.39, p = .240, η_p² = .01), implying that the trends of change were relatively similar. Issues related to pain and discomfort increased in the control group, while they decreased in the intervention group, with the latter change being statistically significant. Moreover, the interaction was statistically significant as well, demonstrating different patterns of change between the two groups (F(1, 170) = 4.17, p = .043^*, η_p² = .02). Issues related to anxiety and depression decreased in both groups from baseline to post-intervention; the changes were statistically significant in both groups. However, the patterns of change were similar in both groups (F(1, 171) = 0.27, p = .605, η_p² = .00).

When we look at the results pertaining to the whole scale, we can notice that the score, interpreted as health-related quality of life, decreased in the control group, whereas it increased in the intervention group. None of the changes were statistically significant, and the interaction was not statistically significant as well (F(1, 170) = 1.76, p = .186, η_p² = .01). Lastly, self-rated health, measured with the VAS scale, slightly increased in both groups, but the changes were not statistically significant. Moreover, there was no interaction between groups (F(1, 170) = 0.02, p = .888, η_p² = .00).

3.5. Exploration of variables that moderate the effects of the intervention on the selected outcomes

We first explored the potential moderators of the effects of the intervention on patient health engagement. We found that gender (b = .21, SE = .21, p = .335), age (b = .01, SE = .01, p = .334), and education (b = .07, SE = .09, p = .423) were not significant moderators, whereas individuals' department (b = .64, SE = .22, p = .004^**) and willingness to use the robot (b = .57, SE = .27, p = .039*) were. Specifically, in the case of thoracic surgery departments, the change in patient health engagement from baseline to post-intervention was similar in both groups (control and intervention; b = − .06, SE = .12, p = .594). On the other hand, for abdominal and general surgery patients, the negative change from baseline to post-intervention was statistically significantly larger in the control group compared to the intervention group (b = .58, SE = .18, p = .002^**). This significant interaction is visualized in Fig. 4.

Moreover, patients with low acceptance of socially assistive robots (b = − .08, SE = .15, p = .595) exhibited a slightly higher change in patient engagement if they were in the control group than the intervention group, but the difference was not statistically significant. On the other hand, patients with high acceptance of socially assistive robots (b = .30, SE = .13, p = .022^*) exhibited a statistically significant higher change in patient engagement if they were in the intervention group, compared to the control group. This significant interaction is visualized in Fig. 3.

Similarly, we explored the potential moderators of the effects of the intervention on the perceived quality of medical care. We found that gender (b = .70, SE = .46, p = .129), age (b = .00, SE = .02, p = .827), education (b = .02, SE = .20, p = .915), department (b = .62, SE = .47, p = .184), and willingness to use the robot (b = .96, SE = .61, p = .114) were not significant moderators.

Lastly, we explored the potential moderators of the effects of the intervention on health-related quality of life (separately for the whole scale excluding VAS and the VAS score). For the whole scale, we found that gender (b = .04, SE = .10, p = .664), age (b = .00, SE = .00, p = .837), education (b = − .05, SE = .04, p = .285), department (b = .14, SE = .10, p = .172), willingness to use the robot (b = .09, SE = .13, p = .505), change in patient health engagement (b = .06, SE = .08, p = .425), and change in the perceived quality of medical care (b = − .02, SE = .03, p = .469) were not significant moderators. Similarly, gender (b = 1.91, SE = 6.43, p = .767), age (b = .13, SE = .24, p = .600), education (b = -3.35, SE = 2.76, p = .226), department (b = 1.82, SE = 6.81, p = .790), willingness to use the robot (b = -2.76, SE = 8.31, p = .741), change in patient health engagement (b = -1.12, SE = 4.91, p = .820), and change in the perceived quality of medical care (b = -2.31, SE = 2.17, p = .289) were not significant moderators of VAS scores. Simple bivariate correlations between changes in patient health engagement, perceived quality of care, and health-related quality of life may be found in the Appendix.

CDSS4.2 Feasibility of implementing a SAR intervention

The feasibility of implementing the SAR intervention in surgical wards was demonstrated by the high retention rate and successful integration into existing care pathways. The demonstrated successful integration of the SAR with existing hospital care pathways addresses a limitation noted in previous research ^36,37 and demonstrates and aligns the potential for SAR technology to complement standard care practices in surgical settings.

The study observed varying retention rates across different stages and groups. At baseline, all control group participants were retained, while the intervention group retained 80.5% of participants. This retention rate aligns favorably with typical dropout rates in clinical trials, which range from 5–40%, where web-based often report higher dropout rates, with a median of 23% (Crutzen et al., 2015).

When considering retention from baseline to post-intervention, the rates were 90.1% for the control group and 83.2% for the intervention group, with an overall rate of 86.9%. The retention rate from baseline to post-intervention is notably higher than the overall retention rate from initial recruitment, suggesting that most attrition occurred early in the study. While any attrition in a study can be problematic, early attrition is generally considered more favorable since, participants who drop out early typically have not contributed much data, minimizing the amount of missing data in the analysis ⁴⁶ Moreover, it is often easier to handle statistically ⁴⁷. Participants with no post-baseline data can be excluded from the primary analysis without introducing bias in many cases. The early attrition observed in the intervention group (19.5% between recruitment and baseline) is not ideal but is preferable to losing the same percentage later in the study. It also suggests that the initial introduction to or requirements of the SAR intervention may have deterred some participants, which is valuable information for refining future study protocols.

Even more importantly, we observed no selective dropout based on group allocation, gender, education, or age, which strengthens the validity of our between-group comparisons ⁴⁸. However, we did note a significant difference in dropout rates between departments, with abdominal and general surgery patients being more likely to discontinue participation compared to thoracic surgery patients. Several factors could explain this departmental difference in dropout rates. Abdominal and general surgery patients might experience different post-operative recovery trajectories or pain levels compared to thoracic surgery patients, potentially affecting their ability or willingness to complete follow-up assessments. This differential attrition warrants careful consideration in future studies and may indicate a need for tailored retention strategies across different surgical specialties.

4.3 Effect of the SAR intervention on Patient Engagement

The impact of the SAR intervention on patient engagementrevealed inconclusive results. While overall changes in patient engagement were not statistically significant for either group, we observed a slight decrease in the control group and a marginal increase in the intervention group. The descriptive results somewhat align with the potential benefits of SARs in promoting patient engagement and self-management, as suggested by Chita-Tegmark & Scheutz (2021) and Robinson et al. (2020). However, it is worth noting that the differences were not statistically significant.

Moreover, additional exploratory analyses uncovered significant moderating effects. Patients in the abdominal and general surgery department showed a significantly larger negative change in engagement in the control group compared to the intervention group. The finding aligns with the observed attrition rates, where 20.5% of abdominal and general surgery patients discontinued participation compared to only 9.0% of thoracic surgery patients. The consistency between lower engagement and higher dropout rates, in this subgroup aligns with the previous stipulation on different post-operative recovery trajectories, where abdominal and general surgery patients may face unique challenges that affect both their willingness to continue participation and their overall engagement with their healthcare. This finding further underscores the complexity of patient engagement, as noted by Graffigna et al. (2015). Moreover, patients with high acceptance of SARs exhibited a statistically significant higher change in patient engagement if they were in the intervention group. These findings align with the technology acceptance model proposed by Broadbent et al. (2009) ³⁶ and suggest that the effectiveness of SAR interventions may be influenced by both contextual factors (e.g., type of surgery) and individual attitudes towards technology.

4.4 Effect of the SAR intervention on Perceived Quality of Care

The results revealed statistically significant decrease in both the control and intervention groups from baseline to post-intervention. The similarity in trends between groups suggests that the SAR intervention did not significantly alter patients' perceptions of quality of care compared to standard care.

As suggested by Cvetanovska et al. (2023), achieving active patient engagement can be challenging, particularly in complex healthcare environments such as surgical wards. The immediate post-operative period may be associated with discomfort, pain, and temporary limitations, which could negatively impact patients' perceptions of quality of care regardless of the interventions in place. Moreover, the introduction of innovative technology like SAR might have raised patients' expectations about quality of care. If these heightened expectations were not fully met, it could result in lower perceived quality of care, despite objective improvements in care delivery. The discrepancy between anticipated and actual experiences with the SAR might have contributed to the observed decrease in perceived quality of care Mlakar et al. (2023). This aligns with the findings of Christoforou et al. (2020), who noted that the integration of new technologies in healthcare settings can sometimes lead to initial dissatisfaction if not carefully managed. However, as stated earlier, a similar decline also occurred in the control group.

4.5 Effect of the SAR intervention on Health-Related Quality of Life

The effect of the SAR intervention on patients' health-related quality of life varied across different dimensions. Notably, we observed a statistically significant interaction in the pain/discomfort dimension, with the intervention group showing a decrease in pain/discomfort issues while the control group experienced an increase. This aligns with the potential of SARs to provide continuous support and education, as suggested by Barello et al. (2012) and Chita-Tegmark & Scheutz (2021), as well as create pleasant and enjoyable interactions ⁴⁹.

For the overall health-related quality of life score, we observed a slight decrease in the control group and a minor increase in the intervention group, although these changes were not statistically significant. The lack of a significant interaction suggests that the SAR intervention did not have a substantial overall impact on health-related quality of life compared to standard care. This finding may be ascribed to health-related quality of life being a rather broad construct that is determined by a wide range of factors.

4.6 Effect Sizes and Implications for Future Research

The effect sizes need to be interpreted in the context of the existing literature on SAR interventions in healthcare. Pu et al. (2019), in their systematic review and meta-analysis of SARs for older adults, found small to moderate effect sizes for various outcomes, including quality of life and psychological well-being. Overall, our findings, to some extent, align with this trend, suggesting that the impact of SAR interventions may be subtle but potentially meaningful, especially when considering the complex nature of healthcare environments.

In particular, for patient engagement, the effect size (Cohen's d) was small for the intervention group). The small effect size for patient engagement may be noteworthy when considered alongside the work of Hibbard & Greene (2013)⁹, who emphasized the potential of patient activation to improve health outcomes and reduce healthcare costs. The moderate effect size for perceived quality of care), although negative, indicates that this outcome may be more sensitive to change and could be a primary focus in future investigations. This finding aligns with the observations of Christoforou et al. (2020)⁵⁰ and Mlakar et al. (2023)³² regarding the complex relationship between technology integration and perceived care quality. The small effect size for health-related quality of life is consistent with the findings of Kachouie et al. (2014)²⁰ in their mixed-method systematic literature review on socially assistive robots in elderly care.

4.7 Limitations

While our study provides valuable insights into the implementation of SAR interventions in surgical wards, several limitations should be considered when interpreting the results and planning future research.

Firstly, a significant limitation of our study stems from the modifications made to the original study design. The shift from a within-subject design to an RCT, while enhancing internal validity, introduced new variables and potential confounds. The integration of the CDSS study, while allowing for a more comprehensive evaluation of technological interventions, also increased the complexity of the intervention and made it more challenging to isolate the specific effects of the SAR component. These modifications, while enhancing the robustness of our research, also introduced new variables and potential confounds. The transition to a RCT design likely increased the internal validity of our study by reducing potential order effects and allowing for clearer between-group comparisons ⁵¹. However, the change also necessitated a larger sample size and potentially introduced new sources of variability between groups. Moreover, the integration of the CDSS component and additional measurement of health quality parameters created a more complex, multifaceted intervention. Although this allowed for a comprehensive evaluation of technological interventions, it is more challenging to isolate the specific effects of each component, a common issue in complex intervention research ⁵². Finally, while not part of the original protocol, a dedicated study room was ensured for patients in the interventions to minimize spillover effects. It is worth noting that this is well in line with the CONSORT extension for non-pharmacological treatments ⁵³.

Secondly, the relatively short duration of our study may not have allowed sufficient time for staff and patients to fully acclimate to and leverage the benefits of the SAR intervention. Future studies should consider longer intervention periods and follow-up periods to potential delayed effects of the intervention and evaluate potential (long-term) impacts that may emerge over time as users become more familiar with the technology. Namely, as suggested by Papadopoulos et al. (2022)²⁸, the similar trends observed in both intervention and control groups for perceived quality of care might indicate that a longer adaptation period is necessary.

Thirdly, our study may not have fully captured the complex, multifaceted nature of patient engagement and care quality, as highlighted by Krist et al. (2017)¹⁰ and Marzban et al. (2022)¹¹. While SAR intervention was innovative, it may not have addressed all aspects that contribute to patients' overall perception of quality of care. The lack of significant overall impact on health-related quality of life suggests that the benefits of SAR interventions may be more nuanced and domain-specific than initially anticipated. Our study found positive effects in specific areas, such as pain management, but these benefits did not translate to overall improvements in health-related quality of life. This limitation underscores the need for more targeted applications of SAR technology, focusing on areas where it can provide the most substantial benefits. Moreover, while our overall retention rate was high, the observed differential attrition by department highlights the complexity of conducting research across diverse surgical populations. Future studies should consider implementing targeted retention strategies for abdominal and general surgery patients and explore the factors contributing to higher dropout rates in this group.

Fourthly, the small to moderate effect sizes observed in our study, particularly for patient engagement and overall health-related quality of life, suggest that larger sample sizes may be needed to detect statistically significant effects. Linked to this, as highlighted in our discussion on perceived quality of care, the introduction of innovative technology like SAR might have raised patients' expectations, potentially influencing their perceptions of quality of care. Our study may not have adequately controlled for or measured these expectations. Future research should consider assessing and managing patient and staff expectations as part of the study design. The presence of researchers and the novelty of the SAR intervention may have created a Hawthorne Effect which influenced participant behavior, potentially affecting the study outcomes. Future studies should consider designs that minimize this potential bias, such as longer acclimatization periods or more naturalistic observation methods. Finally, our study was conducted in a single healthcare institution, which may limit the generalizability of our findings to other healthcare settings or cultural contexts. Future multi-center studies across diverse healthcare environments would enhance the external validity of findings related to SAR interventions.

The study reported in this article demonstrates the feasibility of implementing SAR interventions in surgical wards and provides insights into their potential effects on patient outcomes. While we did not observe significant overall effects on all measured outcomes, the identified moderating factors and the significant interaction in pain/discomfort management highlight promising avenues for future research. Our findings thus contribute to the growing body of literature on SAR applications in healthcare ^16,17 and address the gap in research on SAR use in surgical settings. They also align with the call for more robust, context-specific studies on SAR implementation in healthcare settings¹⁸.

Beyond the primary objectives, the potential implications to clinical practice are broader. The successful implementation and high retention rates demonstrate that SAR technology can be feasibly integrated into existing care pathways. This suggests that healthcare institutions can consider SAR as a viable option for augmenting care delivery, potentially addressing some of the challenges related to workforce shortages^1,2. However, the differential effects observed across surgical departments highlight the need for tailored approaches with department-specific factors, rather than adopting a one-size-fits-all approach. Furthermore, the potential of SAR to provide continuous support and education aligns with the growing emphasis on patient self-management ^5,6. Healthcare providers might leverage SAR for consistent delivery of patient education and support, particularly during periods when staff availability is limited. Moreover, the significant positive impact on pain/discomfort management suggests that SAR could be particularly useful in the aspect of post-operative care. Healthcare providers might explore targeted applications of SAR for pain management protocols. Finally, the study results suggest that SAR should be viewed as a complement to, rather than a replacement for, human care. Clinical practice should focus on integrating SAR in ways that enhance, rather than substitute, human-delivered care.

Acknowledgements

Not Applicable

Funding

This work has received funding by the European Union Horizon 2020 Research and Innovation Program project HosmartAI (grant number 101016834) and from Slovenian Research Agency (Research Core Funding) No. P2-0069, and the Young Researcher Funding 6316-3/2018-255, 603-1/2018-16. The content of this paper does not reflect the official opinion of the European Union or any other institution. Responsibility for the information and views expressed herein lies entirely with the authors.

Ethics Statement

This study was conducted according to the guidelines of the Declaration of Helsinki, and as part of protocol registered at ISRCTN: https://doi.org/10.1186/ISRCTN96689284. For the study a separate ethics approval was received from the Medical Ethics Commission of University Medical Center Maribor, Ljubljanska ulica 5, 2000 Maribor, Slovenia; (+386 (0)2 321 2489; [email protected]), on 15 December 2021, registration number: UKC-MB-KME-76/21.

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Consent for publication

Patients and staff in the images signed a written consent to participate in the study. The consent included publication the images.

Conflicts of Interests

The authors declare no conflict of interest.

Data Availability Statement

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Code availability

The underlying code for this study [and training/validation datasets] is not publicly available but may be made available to qualified researchers on reasonable request from the corresponding author.

Guarantor Statement

Izidor Mlakar (I.M.) and Vojko Flis (V.F.) are the principal researchers and the guarantors of this research. As the guarantor they take full responsibility for the article.

Author Contributions

conceptualization, I.M., U.S., I.R.R., N.P. data curation, V.S, U.S., N.P.; methodology, I.M., N.P, B.I., S.H., I.R.R.; software, I.M., V.S.; validation, N.P., U.S., B.I., S.H.; formal analysis, N.P., U.S., I.M., V.F.; investigation, I.M, U.S., N.P., B.I., S.H.; writing—original draft preparation, I.M., N.P., I.R.R.; writing—review and editing, , I.M., V.F., U.S., N.P., I.R.R.; V.S. B.I., S.H.; visualization, I.M.; N.P. supervision, I.M., V.F.; funding acquisition, I.M.,V.F.; All authors have read and agreed to the published version of the manuscript.

Boniol, M. et al. The global health workforce stock and distribution in 2020 and 2030: a threat to equity and ‘universal’ health coverage? BMJ Global Health 7, e009316 (2022).
Looi, M.-K. The European healthcare workforce crisis: how bad is it? BMJ 384, q8 (2024).
E, D. et al. The relationship between workload and burnout among nurses: The buffering role of personal, social and organisational resources. PloS one 16, (2021).
Pérez-Francisco, D. H. et al. Influence of Workload on Primary Care Nurses’ Health and Burnout, Patients’ Safety, and Quality of Care: Integrative Review. Healthcare (Basel) 8, 12 (2020).
Barello, S., Graffigna, G. & Vegni, E. Patient Engagement as an Emerging Challenge for Healthcare Services: Mapping the Literature. Nursing Research and Practice 2012, 905934 (2012).
Hickmann, E., Richter, P. & Schlieter, H. All together now - patient engagement, patient empowerment, and associated terms in personal healthcare. BMC Health Serv Res 22, 1116 (2022).
Graffigna, G., Barello, S., Bonanomi, A. & Lozza, E. Measuring patient engagement: development and psychometric properties of the Patient Health Engagement (PHE) Scale. Front Psychol 6, 274 (2015).
Lorig, K. R. & Holman, H. Self-management education: history, definition, outcomes, and mechanisms. Ann Behav Med 26, 1–7 (2003).
Hibbard, J. H. & Greene, J. What the evidence shows about patient activation: better health outcomes and care experiences; fewer data on costs. Health Aff (Millwood) 32, 207–214 (2013).
Krist, A. H., Tong, S. T., Aycock, R. A. & Longo, D. R. Engaging patients in decision-making and behavior change to promote prevention. Stud Health Technol Inform 240, 284–302 (2017).
Marzban, S., Najafi, M., Agolli, A. & Ashrafi, E. Impact of Patient Engagement on Healthcare Quality: A Scoping Review. Journal of Patient Experience 9, 23743735221125439 (2022).
Xu, R. H., Cheung, A. W. L. & Wong, E. L. Y. The relationship between patient engagement and health-related quality of life in patients: a cross-sectional study in general outpatient clinic in Hong Kong SAR, China. PPA Volume 13, 1451–1462 (2019).
Almutairi, N., Vlahu-Gjorgievska, E. & Win, K. T. Persuasive features for patient engagement through mHealth applications in managing chronic conditions: A systematic literature review and meta-analysis. Informatics for Health and Social Care 48, 267–291 (2023).
Woodward, E. N., Castillo, A. I. M., True, G., Willging, C. & Kirchner, J. E. Challenges and promising solutions to engaging patients in healthcare implementation in the United States: an environmental scan. BMC Health Services Research 24, 29 (2024).
Cvetanovska, N., Jessup, R. L., Wong Shee, A., Rogers, S. & Beauchamp, A. Patients’ perspectives of factors influencing active participation in healthcare interactions: A qualitative study. Patient Education and Counseling 114, 107808 (2023).
Nichol, B. et al. Exploring the impact of socially assistive robots on health and wellbeing across the lifespan: An umbrella review and meta-analysis. International Journal of Nursing Studies 153, 104730 (2024).
Scoglio, A. A., Reilly, E. D., Gorman, J. A. & Drebing, C. E. Use of Social Robots in Mental Health and Well-Being Research: Systematic Review. J Med Internet Res 21, e13322 (2019).
Guemghar, I. et al. Social Robot Interventions in Mental Health Care and Their Outcomes, Barriers, and Facilitators: Scoping Review. JMIR Mental Health 9, e36094 (2022).
Kabacińska, K., Prescott, T. J. & Robillard, J. M. Socially Assistive Robots as Mental Health Interventions for Children: A Scoping Review. Int J of Soc Robotics 13, 919–935 (2021).
Kachouie, R., Sedighadeli, S., Khosla, R. & Chu, M.-T. Socially Assistive Robots in Elderly Care: A Mixed-Method Systematic Literature Review. International Journal of Human–Computer Interaction 30, 369–393 (2014).
Pu, L., Moyle, W., Jones, C. & Todorovic, M. The Effectiveness of Social Robots for Older Adults: A Systematic Review and Meta-Analysis of Randomized Controlled Studies. Gerontologist 59, e37–e51 (2019).
Ragno, L., Borboni, A., Vannetti, F., Amici, C. & Cusano, N. Application of Social Robots in Healthcare: Review on Characteristics, Requirements, Technical Solutions. Sensors (Basel) 23, 6820 (2023).
Troccaz, J., Dagnino, G. & Yang, G.-Z. Frontiers of Medical Robotics: From Concept to Systems to Clinical Translation. Annu Rev Biomed Eng 21, 193–218 (2019).
Chita-Tegmark, M. & Scheutz, M. Assistive Robots for the Social Management of Health: A Framework for Robot Design and Human–Robot Interaction Research. Int J of Soc Robotics 13, 197–217 (2021).
Robinson, N. L., Connolly, J., Hides, L. & Kavanagh, D. J. Social robots as treatment agents: Pilot randomized controlled trial to deliver a behavior change intervention. Internet Interv 21, 100320 (2020).
Lee, O. E. K., Nam, I., Chon, Y., Park, A. & Choi, N. Socially Assistive Humanoid Robots: Effects on Depression and Health-Related Quality of Life among Low-Income, Socially Isolated Older Adults in South Korea. J Appl Gerontol 42, 367–375 (2023).
Christoforou, E. G., Avgousti, S., Ramdani, N., Novales, C. & Panayides, A. S. The Upcoming Role for Nursing and Assistive Robotics: Opportunities and Challenges Ahead. Front Digit Health 2, 585656 (2020).
Papadopoulos, I. et al. A qualitative exploration of care homes workers’ views and training needs in relation to the use of socially assistive humanoid robots in their workplace. Int J Older People Nurs 17, e12432 (2022).
Soriano, G. P. et al. Robots and Robotics in Nursing. Healthcare (Basel) 10, 1571 (2022).
Aymerich-Franch, L. Why it is time to stop ostracizing social robots. Nat Mach Intell 2, 364–364 (2020).
Borges do Nascimento, I. J. et al. Barriers and facilitators to utilizing digital health technologies by healthcare professionals. npj Digit. Med. 6, 1–28 (2023).
Mlakar, I. et al. Using Structural Equation Modeling to Explore Patients’ and Healthcare Professionals’ Expectations and Attitudes Towards Socially Assistive Humanoid Robots in Nursing and Care Routine. Int J of Soc Robotics (2023) doi:10.1007/s12369-023-01039-4.
Deutsch, I., Erel, H., Paz, M., Hoffman, G. & Zuckerman, O. Home robotic devices for older adults: Opportunities and concerns. Computers in Human Behavior 98, 122–133 (2019).
Latikka, R. et al. Older Adults’ Loneliness, Social Isolation, and Physical Information and Communication Technology in the Era of Ambient Assisted Living: A Systematic Literature Review. Journal of Medical Internet Research 23, e28022 (2021).
Rabbitt, S. M., Kazdin, A. E. & Scassellati, B. Integrating socially assistive robotics into mental healthcare interventions: applications and recommendations for expanded use. Clin Psychol Rev 35, 35–46 (2015).
Broadbent, E., Stafford, R. & MacDonald, B. Acceptance of Healthcare Robots for the Older Population: Review and Future Directions. Int J of Soc Robotics 1, 319–330 (2009).
Evans-Lacko, S., Jarrett, M., McCrone, P. & Thornicroft, G. Facilitators and barriers to implementing clinical care pathways. BMC Health Serv Res 10, 182 (2010).
Rojc, M. et al. End-to-End Evaluation of the Decentralized Symmetric Model of Interaction for Socially Assistive Robotic Nurses. SSRN Scholarly Paper at https://doi.org/10.2139/ssrn.4961578 (2024).
Mlakar, I. et al. A randomized controlled trial for evaluating the impact of integrating a computerized clinical decision support system and a socially assistive humanoid robot into grand rounds during pre/post-operative care. DIGITAL HEALTH 8, 20552076221129068 (2022).
Richmond, V. P., Heisel, A. M., Smith Jr., R. S. & McCroskey, J. C. The impact of communication apprehension and fear of talking with a physician on perceived medical outcomes. Communication Research Reports 15, 344–353 (1998).
Rabin, R. & Charro, F. de. EQ-SD: a measure of health status from the EuroQol Group. Annals of Medicine 33, 337–343 (2001).
Prevolnik Rupel, V. & Ogorevc, M. EQ-5D-5L Slovenian population norms. Health and Quality of Life Outcomes 18, 333 (2020).
Pandey, A. K. & Gelin, R. A Mass-Produced Sociable Humanoid Robot: Pepper: The First Machine of Its Kind. IEEE Robotics & Automation Magazine PP, 1–1 (2018).
Mishra, D. et al. An Exploration of the Pepper Robot’s Capabilities: Unveiling Its Potential. Applied Sciences 14, 110 (2023).
Hayes, A. F. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach. xvii, 507 (Guilford Press, New York, NY, US, 2013).
Bell, M. L., Kenward, M. G., Fairclough, D. L. & Horton, N. J. Differential dropout and bias in randomised controlled trials: when it matters and when it may not. BMJ 346, e8668 (2013).
Lachin, J. M. Fallacies of last observation carried forward analyses. Clinical Trials 13, 161–168 (2016).
Dumville, J. C., Torgerson, D. J. & Hewitt, C. E. Reporting attrition in randomised controlled trials. BMJ 332, 969–971 (2006).
Plohl, N., Mlakar, I., Musil, B. & Smrke, U. A short review of factors associated with acceptance of social robots in healthcare and lessons for their implementation in oncological settings. AIP Conference Proceedings 3030, 030004 (2024).
Christoforou, E. G., Avgousti, S., Ramdani, N., Novales, C. & Panayides, A. S. The Upcoming Role for Nursing and Assistive Robotics: Opportunities and Challenges Ahead. Front Digit Health 2, 585656 (2020).
Shadish, W. R., Cook, T. D. & Campbell, D. T. Experimental and Quasi-Experimental Designs for Generalized Causal Inference. xxi, 623 (Houghton, Mifflin and Company, Boston, MA, US, 2002).
Craig, P. et al. Developing and evaluating complex interventions: the new Medical Research Council guidance. BMJ a1655 (2008) doi:10.1136/bmj.a1655.
Boutron, I. et al. CONSORT Statement for Randomized Trials of Nonpharmacologic Treatments: A 2017 Update and a CONSORT Extension for Nonpharmacologic Trial Abstracts. Ann Intern Med 167, 40–47 (2017).

No competing interests reported.

Appendix.docx

Download PDF

Reviewers invited by journal
12 Nov, 2024
Editor assigned by journal
20 Oct, 2024
Submission checks completed at journal
18 Oct, 2024
First submitted to journal
16 Oct, 2024

You are reading this latest preprint version

A Randomized Pilot Study On The Effects Of A Socially Assistive Robot Intervention On Surgery Patients' Engagement, Perceived Quality of Care, And Quality Of Life

Status:

Version 1

Abstract

Figures

1. Introduction

1.1 Background

1.2 Objectives

2. Methods

2.1 Trial Design

2.2 Changes to the Original Design

2.3 Participants

2.4 Measures and Outcomes

2.5 Interventions

2.6 Statistical analysis

3. Results

3.1 Sample characteristics

3.2. Descriptive statistics and bivariate associations

3.3. Effects on patient engagement and perceived quality of care

3.4. Effects on health-related quality of life

3.5. Exploration of variables that moderate the effects of the intervention on the selected outcomes

4. Discussion

4.3 Effect of the SAR intervention on Patient Engagement

4.4 Effect of the SAR intervention on Perceived Quality of Care

4.5 Effect of the SAR intervention on Health-Related Quality of Life

4.6 Effect Sizes and Implications for Future Research

4.7 Limitations

5. Conclusion

6. Relevance to Clinical Practice

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1