Randomised controlled feasibility trial of an intervention to increase activity and reduce sedentary behaviour in people with severe mental illness: Walking fOR Health (WORtH) study Authors: Suzanne M McDonough, Maurice Dillon, Catherine M McDonough, Judith McAuley, Mary Clarke, Claire Murphy, Mike Clarke, Duana McArdle, Ailsa Niven, Julie Williams, Iseult Wilson, Mark A Tully, Marie H Murphy, Sarah C Howes Published: 2026 ABSTRACT AbstractObjectivePeople with severe mental illness (SMI) engage in less physical activity (PA) and more sedentary behaviour (SB) than the general population, contributing to poorer physical health outcomes in this population. Therefore, the aim of this study was to evaluate the feasibility of a multi-component behaviour change intervention called Walking fOR Health (WORtH), delivered by community mental health teams, aimed at increasing PA and reducing SB compared with a one-off education session in people with SMI.DesignFeasibility randomised controlled trial (RCT).SettingStudy recruitment and intervention delivery took place within four community mental health teams in the UK and Ireland.ParticipantsEligible participants had a diagnosis of a SMI and no contraindications to participating in physical activity. Fifty-four participants (25 male, 29 female; mean age 51.6 years) were recruited.InterventionsParticipants were randomised to the 13-week WORtH intervention, comprising education, activity tracking and health coaching or an education-only control.Primary and secondary outcome measuresFeasibility outcomes included recruitment, retention, adherence and acceptability. Clinical outcomes included device-measured (Axivity AX3) and self-reported PA and SB, body anthropometry, physical function and mental well-being.ResultsThis feasibility study met 90% target recruitment and 94% of participants provided follow-up data. Adherence to allocated intervention and engagement with all core intervention components was >80%. Qualitative feedback indicated high levels of satisfaction. Valid device-measured moderate-vigorous PA (MVPA), the intended primary outcome for a definitive trial, was obtained from 90% of participants at baseline and 75% of participants at post-intervention. Point estimates indicated a mean increase of 8.6 min/day of MVPA in the intervention group (baseline 54.7 min/day (95% CI 39.5 to 70.0); follow-up 63.3 min/day (95% CI 50.1 to 76.4)) and of 1.0 min/day in the control group (baseline 42.1 min/day (95% CI 24.6 to 59.6); follow-up 43.1 min/day (95% CI 29.6 to 56.5)).ConclusionsThe results of this study support the feasibility of the WORtH intervention in adults with SMI, and findings will be used to optimise the design of a definitive RCT.Trial registration numberNCT04134871.Keywords: MENTAL HEALTH, Chronic Disease, Clinical Trial, PUBLIC HEALTH, Wearable Devices STRENGTHS AND LIMITATIONS OF THIS STUDY The study used a pragmatic feasibility design embedded within routine community mental health services, enhancing methodological relevance and applicability to real-world settings. The study used a pragmatic feasibility design embedded within routine community mental health services, enhancing methodological relevance and applicability to real-world settings. The intervention and study procedures were successfully adapted for remote recruitment, training and delivery in response to COVID-19-related restrictions. The intervention and study procedures were successfully adapted for remote recruitment, training and delivery in response to COVID-19-related restrictions. A structured and standardised training approach was employed, including a detailed intervention manual with flow diagrams and scripted guidance, and will be used to support consistent delivery across sites and clinicians in a future definitive trial. A structured and standardised training approach was employed, including a detailed intervention manual with flow diagrams and scripted guidance, and will be used to support consistent delivery across sites and clinicians in a future definitive trial. The intervention was implemented across multiple mental health teams in two regions and jurisdictions, strengthening the methodological generalisability and scalability of the trial procedures. The intervention was implemented across multiple mental health teams in two regions and jurisdictions, strengthening the methodological generalisability and scalability of the trial procedures. Methodological limitations included the absence of blinded outcome assessment and missing wearable device data>20%; both issues will be addressed in a future definitive trial through blinded assessments and enhanced strategies to support and incentivise device wear. Methodological limitations included the absence of blinded outcome assessment and missing wearable device data>20%; both issues will be addressed in a future definitive trial through blinded assessments and enhanced strategies to support and incentivise device wear. Background Severe mental illness (SMI), encompassing conditions such as schizophrenia, psychosis, bipolar disorder and major depression, significantly impacts individuals’ daily functioning and overall quality of life.1 Patients with SMI often experience higher rates of physical inactivity and sedentary behaviour (SB)2 compared with the general population, resulting in additional health complications,3 including obesity,4 cardiovascular disease5 and decreased life expectancy.6 Physical activity (PA) has been consistently linked to numerous benefits for individuals with SMI in improving physical health outcomes3 7 and offsetting the physical health risks related to both the illness itself and potential side effects of psychiatric medications. In spite of these benefits, individuals with SMI may struggle with motivation, cognitive barriers and external determinants impacting their engagement.8 9 It is therefore imperative to explore effective interventions designed to promote PA and reduce SB among people living with SMI. There is growing evidence to support health coaching as a valuable intervention for supporting PA behaviour change10 and emerging evidence in people with SMI. Recent feasibility evidence by members of our team11 has shown that health coaching, with an aim to increase PA and reduce SB in adults with SMI, was feasible in an urban setting and supported people to increase their PA levels. These findings may not be transferable into more rural environments where there may be more barriers to being active independently.12 Remote delivery of health coaching interventions in people with SMI has also not been tested. Although the original study protocol did not plan for remote delivery, the COVID-19 pandemic necessitated the switch of many interventions to remote delivery. This provided an unanticipated opportunity to explore this mode of delivery for the Walking fOR Health (WORtH) intervention. Remotely delivered interventions may reduce clinician and patient burden, leveraging the accessibility and convenience of technology. However, the digital divide presents a significant challenge for individuals with SMI, who may lack digital literacy skills and access to reliable technology required to participate in digital health interventions.13 14 Furthermore, cognitive impairments associated with SMI may exacerbate disparities in digital literacy, impacting ability to recruit and retain people in remote interventions. There is a need to understand the requirements of individuals with SMI to engage with remotely delivered interventions that integrate digital tools, to ensure inclusivity and their feasibility and effectiveness. Given the complex, multi-level determinants of PA in this population, including individual, social, environmental and digital challenges, an evidence-based and theoretically informed approach is required to ensure the intervention targets the mechanisms most likely to influence PA in this population. The Capability, Opportunity, Motivation-Behaviour (COM-B) model14 was selected as the theoretical foundation because it offers a comprehensive lens through which to understand and address the interacting capability, opportunity and motivation components that shape PA behaviour in people with SMI. This model has been used previously by our team to develop PA interventions, including interventions for people with SMI.1115,17 Objective To investigate the feasibility and acceptability of a multicomponent behaviour change intervention aimed at improving PA and SB compared with a one-off education session in adults with SMI living in rural and semi-rural locations. Methods Study population and recruitment Participants were screened according to study eligibility criteria. Eligible for inclusion were males and females aged ≥18 years with a diagnosis of an SMI (ICD-10 clinical diagnosis of a schizophrenia spectrum disorder (F20-29), bipolar affective disorder (F31) or serious depression (F32.3, F33.2, F33.3). Individuals who were already active per General Practitioners’ Physical Activity Questionnaire or unable to participate in a PA intervention due to a physical or mental health condition per revised Physical Activity Readiness-Questionnaire, general practitioner and/or mental health practitioner were excluded. Recruitment took place between July 2019 and March 2022, including a 16-month break due to COVID-19. There were four recruitment sites (two in Northern Ireland and two in Ireland), including one additional site identified to increase recruitment due to COVID-19 related challenges. Each site was open to recruitment for 3–11 months in total, with one site completing a recruitment wave pre-pandemic. Consultant psychiatrists were asked to identify and refer eligible service users according to the eligibility criteria. In two sites, recruitment was by an occupational therapist known to the participants, while in two sites recruitment was by a member of the mental health team (mental health nurse, support worker, research nurse) not known to the participants. Sample size The recruitment target was 60 participants, in keeping with guidance on the sample size of feasibility trials which was available at the time of the trial’s design.18 This sample size is similar to those for UK feasibility trials more generally during the 2010s.19 It was deemed sufficient to provide evidence that would inform the feasibility of delivering the intervention in a future definitive randomised trial and the conduct and size of such a trial, including providing an estimate for the standard deviation (SD) for the primary outcome. A sample size of 60 would allow a 90% probability that this measure of outcome measure variability in our feasibility study would represent the SD in a future main trial.20 Randomisation The protocol outlined 1:1 cluster randomisation,12 which was implemented with the first wave of participants. When recruitment resumed during the pandemic, participants underwent individual 2:1 randomisation (intervention:control). Individual randomisation was used to minimise delays and to allow recruitment of multiple participants from the same centres at different times within the recruitment period without having to restrict recruitment to a single allocation per centre. The unequal allocation ratio was implemented to ensure sufficient intervention exposure for optimal information on intervention delivery. Allocations were undertaken using a permuted block design with unequal block sizes (of 3, 6, 9). Block sizes varied randomly and the recruiting researcher was unaware of the sequence of block sizes. A researcher independent of the study, not involved in any other aspect of the study including screening, recruitment or baseline assessments, generated the randomisation sequence and allocated participants after baseline data collection. The randomisation sequence was concealed from members of the research team involved in determining participant eligibility to prevent biased allocation. Once consent and baseline assessments were complete the recruiting researcher contacted the independent researcher via email to request the allocation assignment. Interventions WORtH intervention The WORtH intervention was a 13-week health coaching intervention (details in online supplemental appendix 1), with embedded behaviour change techniques, designed to increase participants’ capability, opportunity and motivation to increase PA and reduce SB.12 21 22 The intervention included an education session followed by six coaching sessions, delivered every 2 weeks and each lasting approximately 30 min. Intervention group participants received a MiBand (Xiaomi MiBand 3 Fitness Tracker, Xiaomi Inc., Beijing, China), a commercially available wrist-worn activity tracker that connected to a smartphone app worn daily throughout the intervention period, and a paper-based activity diary to record their daily steps. A summary of the content, structure and behaviour change techniques included in the WORtH intervention sessions is outlined in figure 1. The intervention was designed to target one or more COM-B components; for example, the education session focused on building psychological capability while the coaching sessions targeted motivation through goal setting with social support from the coach and self-monitoring using the activity tracker and diary supported capability and motivation. The intervention also included strategies to enhance opportunity and motivation for behaviour change through coaching conversations (online supplemental appendix 1). The intervention was delivered by community mental health team clinicians (occupational therapy and physiotherapy) trained in REACH Coaching Conversations23 and the WORtH intervention, as described in online supplemental appendix 1. The content and structure of the WORtH intervention sessions, including summary of behaviour change techniques included in the intervention. BCTs, behaviour change techniques; PA, physical activity; SB, sedentary behaviour; WORtH, Walking fOR Health. In the first wave all intervention sessions were delivered in-person and per-protocol,12 including a group education session and group walk. For subsequent waves, the group component was omitted due to the pandemic, and study visits (one education session and six coaching sessions) were delivered remotely to individuals via telephone or video call depending on their access and preference, or using a hybrid approach according to current COVID-19 guidelines and participant preference. Intervention training Clinicians received 2 days of training in REACH Coaching Conversations23 delivered by NN, and two half-days training on intervention delivery by the research team (SMM/SCH). Training content included the use of coaching-style conversations to facilitate goal setting, review and progression of goals, action coping planning, problem solving and addressing barriers. The WORtH intervention manual (available on request) outlined session structure and provided in-depth information on intervention component delivery at each session including scripted sample conversations. Ongoing group supervision was provided every 2 weeks by the research team (SMM/SCH) and every 2 months by NN. Clinicians completed coaching record forms which were returned to the research team to ensure fidelity of delivery. Intervention components related to goal setting (setting and reviewing behavioural goals, problem solving, action planning) and feedback and monitoring (feedback from coaches on behaviour, checking participants’ self-monitoring) were considered core intervention components and were the focus of fidelity checking. Strategies to support participants to engage with remotely delivered intervention Preparation for remote delivery reflected on the support required by participants in the first wave to engage with the in-person intervention alongside the digital divide impacting this population group.13 14 To enable access to the remotely delivered intervention, participants were provided with tablet devices, support for set-up and ongoing technical support as required. Longer study visits were broken into shorter sessions to reduce cognitive load. Participants received study visit reminders and additional or amended instructional materials to support use of the device and navigation and completion of the study materials. Participants were also encouraged to seek support from a family member, friend or support worker. Control The control intervention included an information leaflet and a one-off education session on the importance of improving PA and SB and ways to do so, delivered by a member of the research team. Outcome measures Outcome assessment was unblinded; however, all baseline assessments were completed prior to randomisation. Demographics, medical and social details were collected at the baseline assessment. Participants also completed the Blind version of the Montreal Cognitive Assessment (MoCA-Blind) which is validated for use by telephone, and scored out of 22 with a cut-off of ≥19indicating normal cognitive function.24 Timepoints for collection of clinical outcomes were baseline and post-intervention, and a subset of participants completed later follow-up timepoints as outlined in the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) Schedule in the study protocol.12 Feasibility Criteria for progression to a main trial is detailed in the protocol.12 The feasibility outcomes were recruitment, retention, adherence, signal of efficacy and acceptability. Evaluation of recruitment, retention, adherence, signal of efficacy is summarised in table 1. Acceptability of the intervention was based on participant and clinician views collected through semi-structured interviews. Following the first wave of the study, we introduced a number of strategies to improve the quality of the qualitative data collected. This included conducting in-person interviews when possible, giving participants a summary of interview questions in advance of the interview and reminders just prior to allow them to think ahead and prepare their answers if they wished, providing visual cues of the study components during the interview, and altering questioning and use of member checking to encourage open conversations. Table 1Feasibility outcomes and the traffic light system for trial progressionProgression criteriaIndicator: feasibility outcomesWORtH outcomesTraffic light classificationRecruitmentThe number of individuals who consented to participate as a percentage of target recruitment within the recruitment period. The number of participants screened and invited to participate was also recorded to inform a future trial.54/60 (90%)Amber (70–99%); amendRetentionThe percentage of participants with follow-up data. This was recorded for each outcome measure individually, but participants were considered to have follow-up data if they returned any post-intervention data.51/54 (94%)Green (85–100%); proceedAdherenceThe percentage of participants who received and adhered to their allocated programme. This was recorded at intervention component level and as number of sessions attended, and participants were considered non-adherers if they discontinued their allocated intervention. Attendance at study visits and engagement with intervention components was also recorded for both groups.47/54 (87%)Green (85–100%); proceedSignal of efficacy and safetyBased on the CI for the effect estimate for the primary outcome.Evidence of potential harm, in terms of adverse events.Greater improvement of MVPA in intervention group (8.6 min/day) vs control group (1 min/day)*CI for effect estimate for primary outcome that is mostly positive and no evidence of potential harm*Intervention group MVPA at baseline 54.7 min/day (95% CI 39.5 to 70.0) and at follow-up 63.3 min/day (95% CI 50.1 to 76.4); control group MVPA at baseline 42.1 min/day (95% CI 24.6 to 59.6) and at follow-up 43.1 min/day (95% CI 29.6 to 56.5).MVPA, measured moderate-vigorous physical activity; WORtH, Walking fOR Health. Feasibility outcomes and the traffic light system for trial progression Intervention group MVPA at baseline 54.7 min/day (95% CI 39.5 to 70.0) and at follow-up 63.3 min/day (95% CI 50.1 to 76.4); control group MVPA at baseline 42.1 min/day (95% CI 24.6 to 59.6) and at follow-up 43.1 min/day (95% CI 29.6 to 56.5). MVPA, measured moderate-vigorous physical activity; WORtH, Walking fOR Health. Clinical outcome measures Device-measured PA (intended primary outcome for a future definitive trial) At each timepoint, participants were asked to wear the Axivity AX3 wrist-worn accelerometer on their non-dominant wrist for seven consecutive days for 24 hours per day. The Axivity was initialised in OmGui software (V.1.0.0.43, Open Movement, Newcastle, UK) to record tri-axial acceleration data at a frequency of 100 Hz with a dynamic range of ±8 g. During remote delivery, Axivity devices were mailed to participants with a letter on when and how to wear the device. Devices were set up to start recording for 7 days from midnight 3 days after posting. Participants received reminder calls at the start and end of the recording period prompting them to start wearing and to mail the device back in the pre-paid envelope provided, respectively. Other clinical outcome measures (detail including psychometric properties is reported in the protocol12), included body mass index and waist circumference, Simple Activity Questionnaire, Sedentary Behaviour Questionnaire, 30 s sit-to-stand test, 5-sit-to-stand test and Warwick-Edinburgh Mental well-being Scale (WEMWBS). Process outcomes were the Behavioural Regulation in Exercise Questionnaire-2 (BREQ-2) and the Psychological Needs Satisfaction and Exercise Scale (PNSE). Data analysis Data were entered into the Statistical Package for Social Sciences (SPSS) V.26.0 (SPSS, Chicago, USA) and analysed as described in the protocol. Descriptive statistics, presented as means and SD or frequency and percentages, were used to summarise the recruitment, retention, adherence and population demographics. As feasibility studies are not designed to investigate effectiveness, no significance testing was conducted.18 Pre-intervention and post-intervention data for clinical endpoints were presented as mean and 95% CIs. Accelerometer data were downloaded in .cwa format using OmGui and processed using R package GGIR using default package arguments.25 26 Thresholds to separate inactivity and light, moderate and vigorous activity were based on cut points from Esliger and Philips.25 26 Participants were included in this analysis if they had ≥3 valid days of data (ie, days with ≥16 hours of valid wear-time).27 Values were averaged over all valid days and reported as minutes per day spent in each intensity of activity/mg. Minutes of MVPA is intended to be the primary outcome measure for a definitive trial. Qualitative data from participant and clinician interviews were transcribed verbatim. A template with the intervention components and characteristics was used to capture feedback on participants’ experiences with each component. Factors influencing engagement were inductively coded independently by two researchers. Codes were discussed and agreed by the researchers and then grouped into categories, then subthemes and themes. As this paper focuses on feasibility outcomes, only brief qualitative findings relating to acceptability are reported here; a full qualitative analysis will be reported in a separate manuscript. Patient and public involvement In the design of this study, we consulted with two mental health charities (Shine and Hearing Voices) and a stakeholder panel of individuals with lived experience of SMI and their family members, who supported the importance of the research question and influenced the final research design. A member of the research team (DM, the Area Lead for Mental Health Engagement), actively worked with the stakeholder panel on a monthly basis. Once the study started the stakeholder panel reviewed proposed content and structure of the study sessions and materials and provided feedback and suggestions which were implemented in delivery of the intervention. For example, their feedback led to the restructuring of the education session and the development of a Participant Information Leaflet that summarised the Participant Information Sheet in a reader-friendly way for potential study participants. A sub-group of people with lived experience reviewed the study findings including proposed changes for a future study and provided feedback to support development of recruitment and intervention delivery for a main trial. Some recommendations that will be considered for a main trial include engaging with peer support workers to support trial recruitment and building capacity for implementation through a buddy system, with training for those who have completed the WORtH programme to support others participating in the programme. Findings Participants’ flow through the study is shown in figure 2. Of n=179 individuals screened, n=54 participants (mean age 50.81±12.29 years; 25 male:29 female; 68.5% with schizophrenia-spectrum disorders) were eligible and recruited. After baseline assessment they were randomised to the WORtH intervention group (n=34) or the control group (n=20). Participants’ characteristics are summarised in online supplemental appendix 2. Over half (58%) of the participants demonstrated cognitive impairment according to their MoCA-Blind score, and almost 30% lived in a supported living setting. CONSORT flow diagram. *Missing data for individual outcomes are recorded in online supplemental appendix 2. CONSORT, Consolidated Standards of Reporting Trials. Feasibility outcomes Feasibility results (table 1) were summarised using the traffic light system,28 as defined in the protocol.12 Recruitment Figure 2 summarises participant flow including recruitment. Screening of n=179 individuals excluded n=45 ineligible individuals. Of n=134 eligible individuals, n=54 were recruited. This represented 90% of the target recruitment and was deemed Amber, indicating that amendments would be required in a future trial. Recruitment was open at each site (n=4) for an average of 6.75 (range 3–11) months within the recruitment period. It is estimated that 4.2% (range 2.2–8%) of the caseload were screened at each site, and the conversion rate from screening to consent was 30% (range 20–45%). Review of recruitment data at site-level identified that three out of four sites reached their recruitment target (100%), with one site not meeting their target recruitment (67%). This site faced particular challenges, including the transfer of the referring consultant to another service due to COVID-19. Although it was recruiting for ten months and screened the highest % of total caseload, it had the lowest recruitment and conversion rates. Reasons for declining to participate included lack of interest in increasing activity, believing they were already sufficiently active, or concerns about physical health limiting participation. Only one individual stated they had concerns about wearing an accelerometer which they considered a tracking device. Some individuals screened as eligible were not contactable. Conversion from approach to enrolment appeared lower when recruitment was conducted by individuals not previously known to participants rather than by a familiar member of the mental health team. Retention Retention scored green on the traffic light system, that is, between 85% and 100% (see table 1) and is detailed below. Baseline The return rates for each outcome are listed in the tables in online supplemental appendix 2. For the majority of participants (53/54, 98%) all paper-based questionnaires were completed at baseline. Between 94% to 96% completed the process outcomes (BREQ-2, n=2 missing and PNSE, n=2–3 missing). Due to COVID-19, the measure of physical function was changed from Timed-Up-and-Go (TUG) to Sit-To-Stand (STS), explaining the STS missing data for the first nine participants recruited. All participants consented to wearing the Axivity at baseline, however, one participant started wearing the device after the measurement period had ended due to a postal issue, and n=4 participants provided fewer than 3 days valid data and were not included in the analysis. 13 weeks The return rates for the majority of outcomes were 94%, as listed in the tables in online supplemental appendix 2. From a researcher and participant perspective, the most challenging aspect of this data collection was the measurement of physical function remotely, and at follow-up more individuals refused to complete these measures. In order to minimise data loss, we identified a minimal dataset that excluded physical function and the two process measures, in order to encourage completion of the other outcomes. Nine participants did not wear the Axivity at post-intervention (n=3 participants were lost to follow-up, n=5 declined, n=1 was not asked due to their mental health symptoms causing them to experience concerns over the MiBand tracking). Three participants provided fewer than 3 days valid data and were not included in the analysis. Adherence Adherence also scored green on the traffic light system (see table 1). All participants received their allocated intervention. Four participants (n=3 intervention group, n=1 control group) withdrew from their allocated intervention before completing any intervention sessions. Intervention engagement The level of engagement with intervention and control components is summarised in table 2. Table 2Engagement with intervention and control components in the trialIntervention componentEngagementControl group—education session95%(19/20; n=1 received leaflet but declined session)Intervention group—education session91%(31/34; n=1 withdrew due to health, n=2 lost to follow-up (LTF) after receiving pack)Coaching sessions80%(164/204; n=26 attended all, n=4 attended 1–3, n=4 attended no sessions)MiBand activity tracker88%(30/34; n=3 withdrew/unknown as LTF; n=1 returned)App (smartphone)50%(17/34; n=4 not offered/in-person; n=10 declined; n=3 withdrew/unknown as LTF)Diary (paper-based)82%(28/34; n=3 withdrew/unknown as LTF; n=3 coach reviewed app at in-person visits pre-pandemic) Engagement with intervention and control components in the trial Average daily step data collected at the first and last coaching sessions was available for 22/34 intervention group participants. Of these, n=9 increased their average daily steps by >2000 steps, n=5 increased by 1000–2000 steps, n=5 increased by <1000 steps and n=3 reduced their average daily steps (see online supplemental appendix 2). Signal of efficacy In terms of signal of efficacy, the data were green on the traffic light system, that is, CI for effect estimate for primary outcome is mostly positive with no evidence of potential harm, see online supplemental appendix 2. The data on acceptability confirm the lack of harm (see below). Clinical outcomes Adverse events Three participants were admitted to hospital for management of a deterioration of their mental health unrelated to the study. There were n=2 mild adverse events likely related to increasing PA, due to hip pain and skin irritation on groin due to chafing of clothing. Six participants experienced skin irritation due to wearing the Axivity wristband. Acceptability A sample of n=23 participants providing qualitative data was representative of the study sample according to age, sex, mental health diagnosis and cognitive function. All nine clinicians (n=8 occupational therapists and n=1 physiotherapist) who had delivered the intervention completed interviews. Participants’ interview responses indicated that walking was an acceptable form of PA and that the MiBand and the Coach were the most important aspects of the intervention, providing motivation, support and accountability. They described goal-setting as motivating and leading to behaviour change, the emotions associated with goal attainment, as well as the importance of action planning in facilitating goal attainment. Participants described improved awareness of their PA levels, ability to create strategies to meet their goals, confidence as they progressed through the programme. Clinicians reported high acceptability of the intervention and expressed satisfaction with the training and ongoing support provided. They provided positive feedback on the intervention components, although they noted that delivery required more time than anticipated. Clinicians perceived that participants engaged well and viewed the intervention components positively, with the MiBand activity tracker, the coach and goal setting identified as the most important components. Online supplemental appendix 3 contains example quotations. Discussion The main finding of this study was that the a priori feasibility thresholds were met, suggesting it would be feasible to conduct a definitive trial evaluating the effects of the WORtH intervention. It was feasible to recruit, via community mental health teams across this island of Ireland, and retain participants in a trial investigating a health coaching intervention compared with a single education/advice session. Healthcare staff were satisfied with the training received to deliver the intervention which included health coaching approaches in combination with wrist-worn activity trackers to support participants’ to change their PA and SB. Despite almost 60% of participants experiencing at least mild cognitive impairment, there was excellent participant engagement with the intervention components, despite its remote delivery, and positive qualitative feedback emphasising the importance of the coach and the wearable device. The discussion will reflect on observations made during this study that should be explored further in a main trial. Overall, the study met 90% target recruitment with a 30% conversion rate from screening to consent. This represented amber feasibility overall, but three of the four centres were green (100% recruitment target met). Of note, two of these centres met their recruitment target within 3 months. At both these sites recruitment was by a clinician known to the participants. This may have facilitated recruitment due to existing rapport and previous experience of engaging in activities offered by the clinician. Our conversion rate compared favourably to our previous studies of similar interventions across clinical populations; for example, the conversion rates were 20% in SMI11 and low back pain29 and 11% in chronic obstructive pulmonary disease (COPD).16 While offering a promising recruitment strategy, this relies on clinicians’ positive attitude towards research, recognition of it as part of their role rather than additional work, and protected time and organisational support to be involved in research.30 31 Recruitment planning for a future trial would account for expected variation in recruitment across sites and implement tailored, site-specific recruitment strategies and support to optimise enrolment. We were able to explore components of intervention engagement for both WORtH and the control group. In both groups over 90% engaged with the educational sessions. Engagement with other WORtH intervention components varied; over 80% of participants attended the coaching sessions, completed the diary, and used the wearable device (MiBand), whereas only 50% used the app linked to the wearable device. This lower engagement rate was not unexpected as we anticipated cognitive challenges to present in our population and that synching the phone app to the device may be too technically complex for some participants. We therefore planned a priori that if a participant, or their clinical team, thought this was the case, that the clinician would sync the wearable device to their phone in order to review step counts. In the switch to remote delivery this clinician support was no longer possible. Therefore, in order to support engagement tablet devices were purchased for those who did not have access to a smart phone or tablet, and training how to use was offered. In an effort to make the device and app as user friendly as possible, the devices did not require pass codes and all other apps were removed from the home screen. None of this influenced engagement; unless a participant already owned a smartphone, they did not use the mobile phone provided and preferred to use paper-based recording in the diary. In a future trial, we would continue with remote delivery and the existing intervention components, and continue to offer monitoring options that are non-digital or do not rely on smartphone apps to align with participant preferences. The intended primary outcome for a main trial is device-measured change in daily minutes of MVPA. Exploratory analysis of this outcome was included in this feasibility study to assess the suitability of this outcome for a larger trial rather than to evaluate intervention effectiveness. Descriptive trends indicated a greater improvement in MVPA in the intervention group compared with the control group post-intervention, however the study was not powered to detect between-group differences and these findings should not be interpreted as evidence of effectiveness. Valid feasibility data were available from 90% participants at baseline and 76% participants at post-intervention. Prior to a main trial, it is critical that strategies are put in place to minimise data loss which are mainly explained by having insufficient valid data, or having no data as the device was not worn. Some strategies we would implement to reduce missing device data in a future trial are described below in the strengths and limitations section. In terms of refusal to wear the device, at baseline all participants consented to wearing the Axivity whereas five participants declined to wear it at post-intervention. We initially explored if this could be explained by skin irritation experienced by some participants when wearing the Axivity wristband. This was not the case, and of interest there were no reports of skin irritation with same wrist-worn device from over 100 000 participants in the UK Biobank Study.27 Wrist-worn devices have been associated with greater compliance due to lower participant burden and lower risk of irritation associated with skin-taping of devices.32 However, wrist-worn devices can overestimate PA due to arm movements,33 which may have contributed to higher than expected baseline MVPA for both groups when compared with previous research in this population.2 It is also possible that, of this underrepresented population, the most motivated were recruited to this study contributing to higher activity levels at baseline. In a main trial, prioritising strategies to effectively recruit those hard-to-reach participants will ensure broader and more inclusive representation, and minimising loss to follow-up post intervention will be critical to enhancing generalisability of study findings and ensuring the study has sufficient power. Despite potentially recruiting participants who were more active than expected, there was disparity between participants’ baseline clinical outcomes when compared with data for the general population. On average, WORtH study participants were aged in their late 40s and early 50s, yet their muscle strength and functional ability at baseline suggested very poor physical function, closer to that of community-dwelling older adults at least one34 to three35 decades older. This emphasises the need for annual physical health checks, to include physical function. Baseline data for WORtH participants’ mental well-being and cognitive function followed a similar trend. Baseline WEMWBS values were lower (46.5) than the mean score for the general population in the UK (51.0), while Blind-MoCA scores indicated cognitive impairment for 58% of WORtH participants. During the first in-person wave, cognitive challenges influenced participants’ engagement with the WORtH study. A number of modifications were necessary to support participants to engage with outcome assessment and intervention delivery, particularly when the study moved to remote delivery. Completing a cognitive screening tool in future trials may enhance understanding of the cognitive difficulties experienced by participants with SMI. Additionally, implementing strategies to support participation may improve engagement with and maintenance of PA to improve health outcomes. Strengths and limitations The strengths of this study are the excellent recruitment and retention rates, despite the requirement to switch the study to remote recruitment and intervention delivery due to the COVID-19 pandemic. Moreover, this switch did not adversely affect the participants’ high engagement with key WORtH intervention components such as the use of the wearable device to track and monitor their daily step counts. The high engagement levels were particularly notable given the cognitive impairments among our participants. Finally, to support fidelity of intervention delivery, we had a very robust ongoing training programme, adapted when we switched from in person to remote delivery. We developed a detailed training manual with summary flow diagrams and step-by-step scripted text as guidance for each element of the coaching session. Our findings are generalisable to existing staff working within publicly funded community mental health teams who are trained to deliver a health coaching intervention focused on increasing physical activity and reducing SB. Based on our positive feasibility findings across four mental health teams in two distinct regions across different jurisdictions, we believe it will be possible to scale up our study for a main trial with multi-site recruitment by mental health teams across the Island of Ireland. There are some limitations to our study in terms of lack of blinding of outcome measures and high rates of missing device data that we plan to address in a main trial. The lack of blinding for outcome assessment, necessitated by the resource constraints of this feasibility study, represents a limitation and may have introduced bias. However, the primary outcomes focus on feasibility rather than the unblinded clinical outcomes, and blinding would be implemented in a future definitive trial. In order to reduce our missing device data to less than 10% we will address the two main reasons for data loss (insufficient wear days; refusal to wear the device). Participants with insufficient days will be asked to wear the device again. For refusal to wear, we will use a minimal dataset, if required, in the main trial and emphasise with participants the importance of wearing the Axivity watch, and we will incentivise its wear. Conclusion The findings of this study suggest it is feasible to progress to a definitive trial evaluating the effectiveness of the WORtH intervention. Insights from this feasibility study will be used to optimise the design and delivery of a future definitive trial, particularly to enhance recruitment and retention. The study contributes important knowledge on supporting people with SMI, including those with associated cognitive impairment, to engage with remotely delivered interventions and participate in physical activity interventions. These findings may inform strategies to improve engagement in and maintenance of physical activity among people with SMI. Supplementary material Acknowledgements The WORtH project would like to thank the participants who took part and clinicians involved in recruitment and delivery of the WORtH intervention at the four study sites. The authors wish to acknowledge Kathleen Bennett in the RCSI Data Science Centre (DSC) for providing statistical advice and support. The WORtH trial steering committee, comprising Prof Fiona Gaughran, Dr David Shiers and Prof Brendon Stubbs, has provided valuable contribution throughout the delivery of the WORtH intervention. Footnotes Funding: The trial was registered on clinicaltrials.gov (NCT04134871) on 22 October 2019. European Union’s INTERREG VA Programme, managed by the Special EU Programmes Body, awarded to the Health and Social Care Research & Development Division Cross-border Healthcare Intervention Trials in Ireland Network (CHITIN) programme, grant number CHI/5429/18. Northern Ireland Public Health Agency HSC R&D division COVID-related extension, grant number COM/5628/20. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-107567). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study involves human participants and was approved by Office for Research and Ethics Committees Northern Ireland (IRAS ID:250401) and the Health and Social Care Executive Dublin North East Research Ethics Committee in Republic of Ireland. Participants gave informed consent to participate in the study before taking part. Local research governance approval was obtained from the relevant healthcare trusts. The full trial protocol has been published elsewhere to reduce the details here.12 The reporting of this study follows the Consort extension guidelines for pilot and feasibility trials.18 Data availability free text: Anonymised data will be provided upon reasonable request. Patient and public involvement: Patients and/or the public were involved in the design, or conduct, or reporting or dissemination plans of this research. Refer to the Methods section for further details. Data availability statement Data are available upon reasonable request. References Supplementary Materials Data Availability Statement Data are available upon reasonable request.