Virtual reality to aid recovery from critical illness: a feasibility trial of a rehabilitation intervention co-developed with key stakeholders

Abstract

Background

Improved survival from critical illness has led to an increase in patients with long-term physical and psychological morbidity. Postintensive care rehabilitation services are variable and it is recognised that there is a need for inclusive and accessible interventions to support recovery after critical illness. We engaged with people with lived experience of critical care admission to co-produce a virtual reality mediated intervention (ViRtual REality to AiD recoverY post intensive care unit (ICU) (VR READY).

Objectives

Our primary objective was to determine the feasibility and acceptability of VR READY to improve the patient experience during recovery from critical illness. Secondary objectives included assessment of the feasibility and acceptability of selected outcome measures for use in future efficacy and cost-effectiveness studies for ICU recovery.

Design

The co-developed intervention (VR READY) was evaluated in a single arm, non-randomised feasibility trial with embedded process evaluation in patients leaving critical care.

Setting

District General Hospital in Wales.

Participants

Adults who had been admitted to an ICU requiring organ support for more than 48 hours during their current hospital admission. Patients were approached at the point of stepping down from the ICU to a general ward (±48 hours). 87 patients were screened for eligibility, 37 were approached and 22 participants were recruited.

Intervention

Participants were asked to use VR READY for at least 5 min per day for 14 days.

Primary and secondary outcome measures

The primary outcome was feasibility defined by prespecified criteria. Secondary outcomes prioritised by patients and pertaining to recovery and well-being were assessed at baseline and 14 days. A qualitative interview was conducted 4–6 weeks later.

Results

22 participants were recruited with 91% completion of baseline measures, 49.5% completion at 14-day follow-up and 27.3% completing the qualitative interview. Intervention adherence was poor (37.9%), however qualitative findings suggested that participants that engaged with the intervention considered VR READY beneficial to their recovery. While feasibility criteria were met for recruitment and data completion, criteria for retention and intervention adherence were not.

Conclusions

Critical illness is a traumatic experience. Our co-developed rehabilitation intervention VR READY was not feasible in its current format, potentially as a result of being offered too early in recovery when patients were struggling with fatigue, motivation and resilience. VR READY may be beneficial to recovery but further research is needed into optimal mode and timing.

Trial registration number

ISRCTN88854487.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Key stakeholders were involved throughout every stage of this study and continue to be involved in review and dissemination of the study findings.

• It is possible that those who volunteered to be involved in the intervention development process were more likely to be motivated and physically and psychologically well enough to be able to participate.

• The consequence of this is that voices of those still struggling with recovery, unable or unwilling to volunteer their time, were not heard.

• The feasibility trial design was constrained by time, budget and logistics potentially limiting the applicability of findings.

Introduction

Improvements in knowledge and new therapeutic interventions have increased survival rates from critical illness. However, survival is not the sole outcome of importance to patients.^{1 2} It is essential that critical illness research also focuses on survivors’ quality of life.

Post intensive care syndrome (PICS) is common following an intensive care unit (ICU) stay and can be significantly debilitating. Characterised by cognitive, psychological and physical impairments, PICS profoundly impacts the lives of patients and their families: a third of patients do not return to work, another third are unable to do the same work. A quarter of patients still require assistance in activities of daily living a year after ICU admission often affecting the working lives of family members.³ It has been suggested that ‘critical illness associated with ICU admission should be treated as a lifetime diagnosis with associated excess mortality, morbidity and the requirement for ongoing health support’.⁴ In terms of prudent healthcare, large investments made during intensive care are only sustained when continued support is in place following discharge.⁵

Although there are several interventions currently used in PICS prevention (eg, daily sedation breaks, delirium monitoring and early mobility),³ there is no standardised approach to post-ICU care. Despite published National Institute for Health and Care Excellence (NICE) guidance on rehabilitation after critical care in 2009, with quality standards in 2017,^{5 6} currently less than a third of ICU patients in the UK receive dedicated outpatient follow-up 2–3 months after ICU discharge.^7–10 An editorial in the British Medical Journal highlighted the lack of adequate post-ICU rehabilitation and recommended that research to determine the most clinically and cost-effective rehabilitation strategies should be a priority, with collaboration between patients, funders and researchers to identify and address evidence gaps.¹⁰ UK Guideline for the Provision of Intensive Care Services (GPICS2)¹¹ explains that leaving ICU is only the start of a long recovery process which may take months to years, and that there may be considerable residual impact on patient’s morbidity and longevity. It also emphasises the need for further research in this area and makes recommendations for monitoring outcomes and recovery, individual therapy, appropriate personnel, equipment and finance and use of specialist rehabilitation coordinators. Delivery of services to these standards, especially after leaving hospital, is rare, occurring in only 18% of sites in a recent survey,¹² mainly due to a lack of funding and staff.

The need for alternative strategies to deliver therapeutic interventions such as home-based services and virtual platforms to ‘ensure maximum inclusivity for patients into rehabilitation programmes’ must be considered in the context of limited availability of resources.⁸ GPICS2 further recommends a rehabilitation plan addressing all relevant domains for patients¹¹ while acknowledging that ‘the service-user voice is often missing in shaping research’ in this area.¹¹ The current literature highlights the importance of research into improving post-ICU recovery and strongly suggests that the development of accessible, inclusive interventions to aid recovery should be of central prominence.

One potential method of doing this is using immersive virtual reality (VR). VR has been used in a variety of healthcare settings and has demonstrated promise in a range of clinical areas.^13–16 Immersive VR has been shown to be beneficial in reducing pain and anxiety, helping with relaxation, and was found to be safe and acceptable to intensive care patients.¹⁷ ICU stress can be reduced significantly by a calm environment and relaxation techniques,¹⁸ which can be challenging to provide in a busy and noisy hospital ward for patients who are too unwell to leave their bed. ICU patients who trialled VR with immersive nature scenes valued the sense of escape from the hospital and the distraction and relaxation they felt when using VR.^{17 18} We believe that VR provides a unique opportunity to develop an effective, safe and acceptable intervention to aid recovery in ICU patients. However, there remain a number of uncertainties as to how this might be achieved in a real-world setting. Therefore, we proposed a programme of work centring service users in understanding what elements would be required to aid recovery and how we may best measure the effectiveness of any such intervention in future studies. Initial work¹⁹ led to the co-development of the ViRtual REality to AiD recoverY (VR READY) intervention. This was then tested in patients recovering from critical illness to examine the feasibility and acceptability of the intervention and its delivery. Here, we describe the results of the feasibility trial.

Aims and objectives

Our primary objective was to determine the feasibility and acceptability of an accessible, co-developed VR-mediated intervention (VR READY) to improve the patient experience during recovery from critical illness. Secondary objectives included assessment of the feasibility and acceptability of selected outcome measures for use in future efficacy and cost-effectiveness studies for ICU recovery.

Methods

Study design

The protocol for the VR READY study has been described previously.²⁰ Additionally, the full protocol, participant information sheet and consent form are available as supplemental information (online supplemental files 1 and 2) and can also be found within the trial registration page. We conducted a single arm, single-site non-randomised study to evaluate the feasibility and acceptability of the VR READY intervention in patients at the point of stepdown from critical care.

Participants

The target sample size was up to 25 participants, in line with accepted sample sizes for feasibility studies^{21 22} including those featuring qualitative evaluations.²³ Modelling has demonstrated that a sample of 20 users will identify 95% of usability problems with technology testing.²⁴

Participants were screened for eligibility and recruited within the critical care unit or a general ward of a district general hospital within 2 days before or after stepdown from critical care. Potential participants were identified and approached by a trained member of the research team who explained the trial, introduced the VR equipment, went through the participant information sheet and allowed the patients time to consider their involvement and the opportunity to ask questions. All participants provided written informed consent prior to inclusion in the study according to the following criteria:

Inclusion criteria: Adults with capacity to consent; current hospital admission involving a stay in critical care, requiring organ support for more than 48 hours; must have an established method of communication (to allow inclusion of patients with tracheostomies).

Exclusion criteria: History of severe motion sickness; photosensitive epilepsy; physical or anatomical contraindications to using VR headsets (eg, severe visual or hearing impairment, major skull or facial surgery); unable to communicate in English; presence of delirium (assessed daily using Confusion Assessment Method for ICU).

Intervention

VR READY was delivered via a VR headset. Following an initial training session with trial delivery staff, participants were asked to engage with the intervention for at least 5 min a day for a total of 14 days. Usage of VR READY was automatically recorded on the headset, and subsequently downloaded for analysis at the end of the intervention delivery period. Participants were able to choose which content they engaged with and for how long. Description of the VR READY content is detailed in table 1.

Table 1. Description of virtual reality experiences available in the intervention.

Content domain	Description
Exploration	Six different immersive environments with simple narration to introduce the user to the environment and invite them to explore the setting: ‘Underwater’, ‘Cities’, ‘Travel’, ‘Wild Hikes’, ‘Wildlife’, ‘Space’.
Mindfulness and Motivation	Six exercises to promote mindfulness and provide motivation for recovery. ‘Virtualisation for Motivation’—a bespoke motivational script written by the VR READY consultant clinical psychologist in conjunction with ICU survivors to provide reassurance and motivation to patients on their recovery journey ‘Sleep’—a guided relaxation experience to promote re-centring and relaxation. ‘Mindful Seeing’—a session on how to look mindfully at the world; ‘Body Scan’—a mindfulness session to reduce anxiety and stress through connecting with the body. ‘Calming Mind’—a session that involves watching the ripples of water in a lake to calm the mind and bring perspective. ‘Mindful Listening’—a session that supports listening to the world around us, learning to stop and be in the moment.
Breathing	The original breathing exercises featured in DR.VR modified in terms of their speed, tailored to the expected capabilities of the participant population. This entails three computer generated scenes (beach, snow and forest) with accompanying soundscape where the user is guided through a breathing exercise.
Information	This contains a number of information videos featuring key staff roles (psychologist, occupational therapist, dietician, speech and language therapist, physiotherapist) that work with critically ill patients. The videos simulate the specialist coming to see the viewer in a hospital environment to explain their role and why the patient might interact with them.

ICU, intensive care unit; VR READY, ViRtual REality to AiD recoverY.

Assessments

Outcomes were assessed at baseline and following completion of the intervention at day 14. A qualitative interview was conducted 6–8 weeks later. Interviews were conducted by two members of the research team from the Centre for Trials research, one a dedicated qualitative researcher with over 10 years experience and the other a senior research fellow with previous experience (3 years) of conducting qualitative interviews.

Primary outcome

The primary outcome of the study was feasibility as defined by recruitment, retention, data completeness and intervention adherence. Each feasibility criterion was assessed against the predefined thresholds outlined in table 2. For the trial to be deemed feasible and suitable to progress to an effectiveness trial without modification, all progression criteria must meet the green threshold. If one or more measures met the amber threshold, then modifications to the trial design should be considered prior to progression to an effectiveness trial. If one or more criteria fell within the red threshold then it would be deemed unfeasible to conduct an effectiveness trial.

Table 2. Primary feasibility outcome criteria and their feasibility thresholds.

Variable		Progression criteria
		Red	Amber	Green
Recruitment	Number screened actually recruited	<5%	5%–19%	20%
	Number approached willing to participate	<50%	50%–74%	75%–100%
	Number recruited	<6 in total	6–12	12–25
Retention	Number of partially active trial participants at primary end point	<50%	50%–79%	80%
Adherence	Completion of intervention sessions	<50%	50%–79%	80%
Data completeness	Completion of baseline measures	<50%	50%–79%	80%
	Completion of follow-up measures	<50%	50%–69%	70%

Secondary outcomes

Specific clinical outcome measures were selected in conjunction with the Public and Patient Involvement (PPI) steering group to assess the desired effects of the intervention (namely psychological well-being and quality of life). The PPI steering group identified sleep, mood and general well-being as key elements contributing to recovery. Final outcome measures were selected prioritising relevance, time to complete, relevant timeframes, complexity and total number of individual questions. These included: Brief Resilience Scale²⁵ designed to assess psychological resilience; EQ-5D,²⁶ a validated outcome measure for assessing health-related quality of life over five domains. Responses can be summarised to provide an index value which can be used to calculate quality-adjusted life years for use in cost–utility analysis²⁷; ICE Capability Assessment for Adults (ICECAP-A),²⁸ a validated well-being/capability instrument comprising five conceptual attributes covering attachment, stability, achievement, enjoyment and autonomy. The score can be used to calculate a cost per year of sufficient capability well-being²⁹; Richard Campbell Sleep Questionnaire (RCSQ), a 6-item questionnaire specifically designed to address sleep and sleep quality in critical care patients; Depression Anxiety Stress Scale (DASS-10),³⁰ a shortened version of a questionnaire designed for routine monitoring of outcomes in psychology practice to provide an overall measure of distress. These outcomes were largely self-report, but assistance with completion was provided by members of the research team.

An embedded process and technology evaluation included evaluation of technology acceptance³¹ (willingness and ability of participants to engage with VR as a technology), and acceptability of the intervention, outcome measures and trial processes.

Analysis

No formal hypothesis testing of outcomes was undertaken. The primary outcome of feasibility was assessed according to the predefined criteria specified in table 2. The percentage completion of baseline and follow-up measures was calculated as number of outcome measures completed against those expected and the number of data points completed across all measures against those expected. When looking at the completion of individual outcome measures at each timepoint, the percentage completion was calculated described above, but only for those attempting any measures at that timepoint, so non-completion from withdrawn participants was not incorporated. For analysis of adherence, usage data were downloaded from the VR headsets and total usage calculated for each day. The daily usage figure was determined as being adherent if this was more than 5 min in total. The number of days each participant met the adherence threshold was calculated and the percentage adherence across the 14-day intervention period was calculated as the number of days the participant was adherent (ie, used the intervention for 5 min or more) as a percentage of the total number of days (14) of the intervention delivery period. The analysis of acceptability was qualitative in nature. Recordings of qualitative follow-up interviews were transcribed verbatim. Transcripts were subject to analysis using an a priori framework based on the Unified Theory of Acceptance and Use of Technology³¹ and a mapping approach followed by a thematic analysis informed by Braun and Clarke³² and Adu³³ using NVivo software as a data management tool. Qualitative analysis was conducted by KS. Clinical outcome measure data were collated and analysed descriptively, by a member of the research team independent from those assisting data collection, using means and SDs and summarised in a tabular format. As no formal hypothesis testing was done, no methods for handling missing data were used beyond the reporting of data completeness. This was triangulated with the interview data using a mixed methods approach³⁴ to complement the qualitative evaluation. The cost of the VR Ready intervention was assessed in terms of the costs of equipment, staff training and time to train participants.

Results

Recruitment

A total of 87 patients were assessed for eligibility with 37 approached and 22 participants recruited into the trial between October 2024 and March 2025 (see figure 1, tables 3 and 4). Participants’ baseline and ICU characteristics were varied in terms of age, sex, ICU length of stay, invasive ventilation and severity scores but not in terms of ethnicity.

Figure 1. CONSORT diagram of patient screening, recruitment and assessment. CONSORT, Consolidated Standards of Reporting Trials; FU, follow-up; GCS: Glasgow Coma Scale; ICU, intensive care unit; VR, virtual reality.

Table 3. Participant characteristics.

Age	Sex	Ethnicity	Resilience
Range 21–85 (mean 59.05, median 61)	63.6% female	100% British white	43% low 52% normal 5% high

Table 4. Participant ICU characteristics.

Patient	ICU length of stay	Days of invasive ventilation	APACHE* score
1	4	0	20
2	10	0	25
3	5	0	27
4	5	5	8
5	9	6	26
6	49	24	25
7	17	13	21
8	3	0	17
9	19	13	21
10	4	0	18
11	35	22	31
12	4	9	17
13	16	14	20
14	6	0	15
15	2	2	13
16	18	11	13
17	14	12	19
18	32	33	18
19	2	0	26
20	2	0	21
21	4	0	8
22	3	0	19

^*Acute Physiology and Chronic Health Evaluation

ICU, intensive care unit.

Within the cohort, results of the Brief Resilience Scale indicated that 57% of the participants had normal or high resilience. Of the 6 participants that completed the entire trial including the follow-up interview, all but one had normal or high resilience, whereas of the 8 participants that withdrew from the trial after completing the baseline questionnaires, 4 had low resilience and 4 normal.

Primary outcome assessment

The predefined feasibility criteria were met in relation to the overall number of people recruited (22/25), the proportion of people screened subsequently recruited (25.3%) and the completion of baseline outcome measures (91.4%). However, criteria relating to the number of people approached willing to participate in the trial (59.5%) and overall retention (50%) met the amber threshold. Criteria relating to intervention adherence (37.9%) and data completeness from all recruited participants (49.5%) at follow-up were not met and fell in the ‘red’ category (table 5). Overall, the trial was deemed to be unfeasible in its current format and significant modifications would be needed to progress to a full trial.

Table 5. Primary feasibility outcome results.

Variable		Progression criteria
		Red	Amber	Green	Result
Recruitment	Number screened actually recruited	<5%	5%–19%	20%	25.3%
	Number approached willing to participate	<50%	50%–74%	75%–100%	59.5%
	Number recruited	<6	6–12	12–25	22
Retention	Number of partially active trial participants at primary end point	<50%	50%–79%	80%	50%
Adherence	Completion of intervention sessions	<50%	50%–79%	80%	37.9%
Data completeness	Completion of baseline measures (total data points)	<50%	50%–79%	80%	91.4%
	Completion of follow-up measures (total data points)	<50%	50%–69%	70%	49.5%

Intervention adherence

Participants were asked to use the VR READY intervention daily for at least 5 min for the 14-day intervention period. Analysis of the usage data downloaded from VR headsets demonstrated that largely, participants did not access the intervention as frequently as intended. The mean number of days used was 5.4 (SD=3.4) with the maximum used by any individual 12 days and the mode usage was 5 days. Where participants did use the VR intervention, this was predominantly for at least 5 min, with an average daily use of 11.4 min (SD=3.4). Participants interacted with the full range of content provided on the VR READY headset, with 128 individual content items accessed across the whole cohort. All content items were accessed at least once with the sole exception of one information video. The most popular content domain was ‘Exploration’ (n=80/128), followed by ‘Mindfulness and Motivation’ (n=23/128), ‘Breathing’ (n=16/128) and lastly ‘Information’ (n=9/128). The most popular individual content items were ‘Underwater’ (n=22), ‘Wildlife’ (n=18) and ‘Cities’ (n=15).

Secondary outcome measures

The mean scores for the secondary outcomes at baseline and the 14-day follow-up can be seen in table 6. The trial was not powered to detect differences between the baseline and follow-up assessments so no statistical analysis has been performed on these. All outcomes showed a high degree of variability between participants, but in general there was a trend towards improvement on the DASS-10 (higher scores are representative of greater distress), RCSQ (lower scores indicate more problems with sleep) and EQ-5D (higher scores reflect worse problems).

Table 6. Secondary outcome measures assessed.

Outcome measure	Baseline (n=21)		14-day follow-up (n=11)
	(Mean (SD))	Data completeness* (% data points)	(Mean (SD))	Data completeness* (% data points)
DASS-10	13.65 (7.1)	96.2	8.4 (6.5)	99.1
RCSQ	52.46 (34.21)	94.4	62.25 (24.82)	100
EQ-5D value	0.17 (0.35)	95.2	0.36 (0.36)	100
EQ-5D (VAS only)	52.0 (29.35)	95.2	49.55 (24.54)	100
ICECAP-A value	0.69 (0.16)	95.2	0.68 (0.24)	98.1

^*Data presented here represents data completeness for each measure according to the number of participants completing that measures. Results of overall data completeness are reported in table 4.

DASS-10, Depression Anxiety Stress Scale; ICECAP-A, ICE Capability Assessment for Adults; RCSQ, Richard Campbell Sleep Questionnaire; VAS, Visual Analogue Scale.

The EQ-5D Visual analogue scale (VAS) ranged from 0 to 100 at baseline and 0–85 at 14-day follow-up (with 100 the best health state). The dimensions on the EQ-5D-5L are scored from 1 to 5, with 1 being the best health state and 5 the worst. The EQ-5D-5L scores were converted to utilities using the EQ-5D crosswalk to the EQ-5D-3L as currently recommended by NICE.³⁵ Of the 10 participants with completed questionnaires at baseline and 14-day follow-up, 2 participants reported worsened health states from baseline to 14-day follow-up, 1 had no change and the rest reported improved health-related quality of life. Of the 10 participants with completed ICECAP-A questionnaires at baseline and 14-day follow-up, 4 reported worsening well-being/capability, 1 had no change, and the rest reported improved well-being/capability.³⁶

Process and technology evaluation

Interviews were completed with six participants who completed the trial. As a result of the relatively low proportion of participants interviewed, data saturation was not reached.

Recruitment and Baseline Assessments—“explained everything well”

The process of recruitment and how patients were approached to take part was acceptable.

It was clear and easy to understand. Participant 043

Allowing time for participants to try a headset before agreeing to take part was seen as an important part of the recruitment process for delivery staff to introduce patients to a technology that they may not have experienced and to demonstrate use, content and buy-in to the intervention.

Completion of questionnaires at baseline and after 2 weeks was not seen as burdensome.

Didn’t take long at all. I think it was all relevant. I don’t think there was anything additional that I would have asked I suppose or would have liked to have been asked. Participant 035

Reasons underlying decision to participate in the study “Why not?”

Participants described a variety of reasons underlying their decision to participate in the study. These included wanting to be ‘taken away’ from the hospital environment; being interested in what it could offer; hoping that it might help with breathing, sleep, stress management and general recovery and also feeling that taking part in research is a good thing and could help others.

I thought I’ll give it a go, I thought this can take you away from a lot of things. CTM049

Daily check-ins and timing of use—Very easy to use

Daily check-ins allowed staff to identify any problems and to assist those that needed help to use the headset. Participants interviewed appreciated the presence of a member of the research team to offer support with using the VR intervention. Although most people quickly became confident navigating the technology, some needed help lifting and positioning the headset due to physical weakness, some required assistance with charging the device or troubleshooting if it didn’t load up correctly straight away, and others just valued the reassurance. All interviewees felt the headset was comfortable once it was adjusted and positioned properly. Participants indicated use during the afternoons and evenings suited them best.

Because like in the morning, you were having to have the wash, having the doctors coming round all that kind of stuff. So, in the evening it was a bit more, there was a more of a gap. So, you had more chance to relax. Participant 040

VR was considered helpful—It was a bit of a lifeline

All of the participants interviewed said that they found the VR intervention helpful. The main ways it helped were: distraction; taking them out of their current environment; reducing anxiety, particularly around the transition to home; helping guide breathing; promoting relaxation; providing information and increasing confidence.

And I felt I kind of felt empowered in a way because I was able to feel like I had improved my breathing and it was something that I could keep trying to improve. Participant 040

Barriers to use—I might have liked more variety

The most commonly reported barrier to use related to the navigation and content selection and indicates that refinements need to be made to this aspect of the VR. When asked what they would recommend to improve the intervention, most participants suggested more content including: more outdoor experiences; more information about what to expect from recovery and the option of games and puzzles.

You could vary the categories. When you were looking for a distraction, it would have been nice to look at something new instead of what you’ve already seen. Participant 043

Particularly helpful for those without families –The times I found it really helpful was if I didn’t have any visitors come in.

Two of the participants interviewed had no family and lived alone. They appeared to find the VR intervention particularly helpful.

Like I say there were sort of quiet periods where ‘cause I didn’t have any visitors I was able to sort of kill up a lot of time because, let’s be honest hospital is not the best place to be at times. There are times where you just want to rest and recuperate so it’s all when it’s sort of alone and quiet at periods, it was helpful for that. Participant 053

Would recommend to others—I think everybody should at least give it a go.

Despite all of the participants interviewed reporting a benefit to themselves from the VR intervention, most of them recognised that it might not be for everyone and felt that older people in particular might not want to try it. However, they felt strongly that everyone should be offered the opportunity.

I would encourage people to use it in the future because I felt it benefit me, especially in terms of breathing and sort of control about my anxiety because I was sort of anxious about having sort of needles in my arm and stuff like that. Participant 53

Health economics

Patients were asked questions about use of health and social services after their hospital discharge and whether they had required help from family or friends. Most had required some form of support and none had returned to their full preillness function by the time of the interviews. One had medically retired from work; one had returned to work on light duties and others were unable to work. Wider societal perspective, including personal costs, should be considered for future economic evaluations. Key cost drivers of providing a VR mediated intervention are reported in table 6.

Discussion

There is much interest in improving the experience and outcomes of critical illness recovery but despite extensive research exploring a variety of interventions, very little has been shown to be beneficial,³⁷ with few studies reporting feasibility testing before conducting larger effectiveness trials.³⁸ Here we have reported on the feasibility and acceptability of a co-developed VR-based, psychological focused intervention to support recovery from critical illness.

We have demonstrated that integrating stakeholders’ views throughout can enhance the acceptability of novel interventions.

Evaluation of the VR READY intervention met 2/3 criteria relating to recruitment. The final criterion on consent rate met the amber threshold (59.5%) but this compares favourably with the median consent rate of 48% across 71 trials of postcritical care interventions,³⁸ suggesting that the prespecified threshold was ambitious in the context of critical care research and should be reconsidered for future studies.

Our data shows that the intervention was perceived positively by our patients and this was reflected in the qualitative feedback from participants indicating that they wanted to ‘give it a go’. However, only six participants (22%) took part in the qualitative interviews, introducing a survivor bias to the data. Importantly, and despite best efforts, we have not heard the views of the participants who withdrew from the trial and their experiences of the intervention. Patients interviewed after completing the intervention all found it acceptable, enjoyed the experiences, and felt it had been beneficial to them. They all felt it could benefit others. The satisfaction and acceptability reported may be reflective of the extensive PPI involvement and input. Criticisms were minor and mainly with regard to occasional technical issues with the hardware or requests for more varied content. Initial concerns about VR being more appealing to younger generations during the co-production phase were somewhat contradicted by the diverse age range of trial participants.

While data completion rates at baseline were good, there was still some missing data. We attribute this to the self-reported change in well-being in one participant and possible fatigue or poor concentration in others, which are well recognised issues in this population.^{3 39} However, participants interviewed reported that the questionnaires themselves were not burdensome to complete.

Retention of participants throughout the trial period was inadequate and adherence to the intervention as recommended was poor. Although these feasibility criteria did not meet prespecified thresholds, poor retention and intervention adherence is not an uncommon problem in research in patients recovering from critical illness. Reviews have reported widely variable retention rates of 18%–100%^{38 40} with personal and circumstantial factors influencing follow-up⁴¹. We know that these patients struggle with motivation to engage with rehabilitation and low mood also impacts further on this. Participants frequently volunteered their reason for withdrawal with the most common cause reported as feeling too unwell. One participant completed the intervention period but deteriorated and died before the follow-up data could be collected and several other patients were transferred to other hospitals necessitating withdrawal from the study. Patients discharged from critical care are generally still in poor health and dropout or loss to follow-up for these reasons is well recognised as one of the major challenges of research in this area.^{42 43} It has been recommended that loss to follow-up should be considered in sample size calculation for planned studies, as well as planned participant retention strategies.⁴⁰

The low level of adherence to the trial intervention seen in this study was likely multifactorial. Our recommendation to use the intervention for 5 min per day for 2 weeks was a pragmatic one. The manufacturer recommended ‘dose’ for the DR.VR device is 5–10 min per use and all of the experiences included in VR READY fell within that timeframe. There are no current clear guidelines or standards for VR research⁴⁴ and no published evidence on which we could base our decision on duration of the intervention. We opted for 2 weeks as this would cover the period from ICU stepdown to ward level care and the initial transfer home in some cases. Further, it seemed like a realistic period to deliver such an intervention while being long enough to perform a feasibility evaluation. There is scope to modify this or allow for greater flexibility in allowing patients discretion as to when and how they wish to engage with the intervention. We note a recent trial of VR in ICU where patients initiated the VR intervention when they felt ready, which was a median of 8 days after ICU discharge and most patients desired only one session,⁴⁵ so perhaps we offered our intervention too early and too frequently.

Reasons reported by participants for not using VR READY are resonant with factors identified in the literature associated with post-ICU patients’ engagement with rehabilitation activities. These include: fatigue,⁴⁶ motivation,³⁷ inner strength and willpower,⁴⁷ resilience,⁴⁸ self-determination⁴⁹ and therapeutic engagement.⁵⁰ Fatigue affects up to 81% of critical illness survivors^{39 46} and can reduce patients’ activity levels, including reduced engagement in pleasurable activities and health maintenance behaviours.⁴⁶ Self-motivation was identified in our early focus groups¹⁸ and another study⁵¹ as a key factor in recovery from critical illness and it has been proposed that future studies in this area should collect data on motivation and engagement which are crucial for maximising the benefits of rehabilitation.³⁷ The concept of willpower is related to motivation and describes ‘the ability to make a mental effort to maintain one’s prior resolutions by blocking reconsiderations’⁵² and the negative experiences of an ICU stay might exhaust patients’ willpower.⁵³ Intensive care withdrawal syndrome is a common (12%–100%) response to cessation of opioid or benzodiazepine medications used in critical illness and manifests as physical and psychological symptoms which can exacerbate the clinical course of critical illness and make it difficult to engage with therapeutic interventions.⁵⁴

Resilience describes an individual’s ability to adapt and adjust⁵⁵ and a previous study of resilience in ICU survivors found resilience to be normal or high in the majority of survivors.⁴⁸ Resilience was inversely correlated with anxiety, depression, post-traumatic stress, difficulty with self-care and pain.⁴⁹ Our cohort had a range of resilience scores, those with lower scores were less likely to complete the trial suggesting that resilience may have had an impact on participants’ decisions to continue with the intervention.

The study was not powered for formal hypothesis testing of outcome measures, rather we were trialling their use for feasibility and acceptability. However, there was a general trend towards improvement on all measures after the intervention (sleep quality, depression and anxiety, well-being/capability and health-related quality of life). It is not possible to say whether this was to do with the intervention or part of the expected process of recovery. Most participants required some support after discharge home and had engaged with healthcare professionals, home carers or help from family and friends. Although not sufficient to make any health economic conclusions, the assessment of key cost drivers in delivering the VR READY intervention have provided insights into considerations for cost-effectiveness evaluation in future effectiveness trials of this or similar interventions.

Conducting feasibility studies to identify uncertainties, test protocols and refine interventions as well as determining acceptability prevents wastage of valuable resources on poorly designed effectiveness trials that risk poor recruitment and retention.³⁸ The lack of feasibility demonstrated here is useful in preventing what may be a futile onward investigation.

In summary, we were able to effectively engage stakeholders to co-develop a VR intervention and evaluate that in a clinical setting. Despite high levels of acceptability and enjoyment of the intervention, the high dropout rate and low adherence to the intervention as set out in the protocol do not support proceeding to an effectiveness trial using the current trial design or format for intervention delivery.

The clear benefits described by those who did complete the study warrant further investigation with amended methods. Future studies could consider using the intervention at a different point in recovery when patients are less physically unwell and exhausted, and across a different time period, with less stringent adherence criteria.

Data availability statement

Data are available on reasonable request.