Abstract
Background People experiencing homelessness (PEH) are disproportionately affected by frailty, yet few interventions have targeted this syndrome in this population.
Aim To assess the feasibility, and potential impact of a combined exercise and nutritional intervention for PEH living with, or at risk of, frailty.
Design & setting A single-arm feasibility trial in a GP clinic for PEH in the Republic of Ireland.
Method A two-month exercise and nutritional intervention tailored for PEH was offered to potential participants. Individuals with, or at risk of, frailty attending a GP clinic for PEH in the Republic of Ireland were invited to participate. The primary outcome was feasibility, which was assessed using Bowen’s framework. Secondary outcomes were based on potential impact and included frailty scores (Clinical Frailty Scale [CFS] and Survey of Health, Ageing and Retirement in Europe-Frailty Instrument [SHARE-FI]) and weight. A process evaluation explored participant experience.
Results Of 124 eligible individuals, 108 (87.1%) enrolled, and 75 (69.4%) completed follow-up. Among those followed up, 70 (93.3%) engaged with at least one component of the intervention, with the majority finding the intervention easy to follow. CFS and SHARE-FI scores improved following the intervention in those followed-up.
Conclusion This study supports the feasibility of a primary care-based exercise and nutritional intervention for PEH living with frailty. The intervention appeared to be safe and to improve the frailty status of participants. It can also be used to inform the design of a definitive trial.
How this fits in
Frailty is common among people experiencing homelessness (PEH) and is associated with poor health outcomes, yet evidence-based interventions remain limited. Multicomponent programmes, including exercise and nutrition support, are effective in older adults but have rarely been tested in this population. Our feasibility trial shows that such an intervention can be delivered in a primary care setting for PEH. These findings provide early evidence to guide clinicians on the potential benefits and practicalities of addressing frailty in this underserved group.
Introduction
Homelessness encompasses a spectrum of living circumstances, defined by European Typology of Homelessness and Housing Exclusion (ETHOS) as rooflessness (sleeping rough), houselessness (emergency or temporary accommodation), insecure accommodation (for example, sofa surfing, eviction risk), and inadequate accommodation (for example, severely substandard or overcrowded housing).1
Research illustrates that people experiencing homelessness (PEH) encounter significantly higher levels of morbidity compared with the general population.2,3 Mortality among PEH is stark. In the Republic of Ireland, coroner data from 2020 report median ages at death of 41 years for men and 36 years for women, although homelessness may be under-recorded.4 This likely reflects accelerated ageing, where biological age exceeds chronological age.5 Additionally, the age of people entering homelessness is rising,6 meaning many already have age-related vulnerabilities.
Correspondingly, a high burden of geriatric conditions among PEH, including falls, cognitive impairment, and incontinence has been identified.7–9 Frailty, a state of diminished physiological reserve,10 is associated with increased risks of illness, dependency, and death.11–13 At the other end of this health spectrum, resilience reflects the ability to recover from external stressors.14,15
Among adults aged ≥50 years in the general population, frailty affects 12% and pre-frailty 46%.16 In PEH, a systematic review reported frailty prevalence of 16%–70% and pre-frailty 18%–60%,17 while a cross-sectional survey of 2288 PEH found 26.6% of those aged 18–29 years were frail.18
Given the high morbidity, accelerated ageing, and mortality in PEH, frailty can lead to earlier functional decline, increased acute care use, and reduced ability to exit homelessness, making support critical to prevent further health deterioration and social exclusion.
National Institute for Health and Care Excellence (NICE) guidelines recommend that PEH with frailty require tailored long-term care strategies,19 however, they offer limited direction on how such approaches should be implemented.
In community-dwelling older adults, resistance exercise and nutritional interventions have been shown to be both feasible and effective for reducing frailty and enhancing resilience.20–23 In contrast, very little research has assessed frailty interventions for PEH.
A feasibility trial of 31 participants, combining supervised exercise with post-exercise nutritional supplementation, showed good initial uptake but high attrition (>50%) and no significant physical improvements.24 Similarly, a randomised controlled trial (RCT) of a community-based intervention for pre-frailty and frailty in women experiencing homelessness found no significant changes in frailty or substance use.25 Evidence for interventions targeting other geriatric conditions in PEH is also limited.
Furthermore, existing evidence has illustrated substantial challenges for PEH to participate in research. This includes issues such as unstable accommodation, competing priorities, mistrust of services, high mobility, and inconsistent access to health care.26–28
Accordingly, further work is warranted to develop and evaluate interventions for frailty in this population.
Method
This single-arm feasibility trial was conducted in a walk-in GP clinic in Dublin, which provides primary care and addiction services for PEH; a setting chosen to maximise participation. Recruitment occurred from August 2024 to January 2025, with follow-up completed in June 2025. The study follows the Consolidated Standards of Reporting Trials (CONSORT 2010) extension for pilot and feasibility trials.29 Methods deviated slightly from the protocol by omitting the FRAIL scale, as alternative measures were deemed sufficient and more practical. Sample size was based on a co-author’s feasibility study,22 providing a more applicable estimate.
Participants
Exclusion criteria (applied before inclusion for efficiency) were: acutely unwell; Clinical Frailty Score (CFS)<3 or>6;30 current malignancy; chronic kidney disease stage 3–4; or severe dementia. Inclusion criteria were: aged ≥18 years; CFS 3–6 (‘pre-frail’ to ‘moderately frail’);31 living in ‘roofless’, ‘houseless’, or ‘insecure’ accommodation per ETHOS;1 and capacity to consent.
Procedure
During recruitment, all adults attending a morning clinic with one of the study GPs (TC or JT) were screened. Each GP saw approximately 20 patients per clinic, with roughly 20% meeting eligibility after applying exclusion criteria.
Eligible individuals were offered verbal and written study information, and advised that participation was voluntary. Frailty was explained in simple terms (for example, feeling slower, exhausted, weight loss, slower recovery), with further clarification as needed. Owing to the highly transient nature of the clinic population and the risk of losing contact, most participants provided written consent during the same consultation. Time was provided to answer any questions and check understanding to support informed consent.
After informed consent, baseline data were collected, including demographics, lifestyle factors, and chronic conditions. Body mass index (BMI), CFS score, and the Survey of Health, Ageing and Retirement in Europe-Frailty Instrument (SHARE-FI) were recorded,32 covering exhaustion, appetite loss, handgrip strength, slowness, and low physical activity. Handgrip strength was measured using a Constant dynamometer.
Intervention
A two-month exercise and nutrition intervention, previously developed in a general practice RCT for older adults,23 was adapted and delivered for PEH. Study authors, informed by discussions with clinic staff and patients, adapted the intervention to suit a younger, mobile population experiencing homelessness. Practical challenges, such as unstable housing, limited cooking and/or storage, and competing priorities, informed changes including using younger participants, outdoor exercises without specialist equipment, and adding an oral nutritional supplement (ONS) to the nutrition component.
Participants received a leaflet with photographic instructions for 10 self-directed exercises focused on strength, balance, and resistance (Supplementary Material). Based on feedback from a previous public and patient involvement study,31 gender-specific exercise sheets were used.
To support post-exercise muscle protein synthesis,33 participants were prescribed a daily compact ONS, providing 18 g of protein, to be taken within 1 hour of exercising, with weekly dispensing via the participant’s pharmacy. They also received a pictorial guide to protein-rich foods (Supplementary Material Box 3) and were encouraged to consume 1.2 g protein/kg/day.34
A 1-month follow-up call assessed concordance and provided support, which has been previously shown to increase adherence by 20%.22 Participants were invited for reassessment at 2 months.
Primary outcomes
The primary outcome was feasibility, assessed using Bowen’s eight-domain framework35 (Table 1), using data on participant engagement, follow-up completion rates, self-reported ease of use, and perceived health impact.
Secondary outcomes (exploratory)
Secondary outcomes included changes in frailty scores at 2 months (CFS and SHARE-FI), and change in weight. These exploratory outcomes were intended to assess potential intervention effects and adverse outcomes, and to inform future trial design.
Sample size
Our target sample size was informed by a previous feasibility study of 94 participants,22 inflated by 15% for expected attrition and population differences, giving a target of 108. Although formal calculations are not usually required for feasibility studies, this size was chosen to support exploratory analyses and future trial planning. A prior RCT in older adults23 reported a 0.7 SHARE-FI difference (standard deviation [SD] = 1.2), for which 33 participants would provide 90% power, showing that our target exceeded this.
Data analysis
Feasibility was assessed across Bowen’s domains using descriptive statistics. Adherence was examined with binomial logistic regression including age, sex, opioid agonist therapy (OAT) use, baseline CFS, and receipt of the 1-month follow-up call. Changes in CFS, SHARE-FI, and weight were analysed using paired tests in a complete case analysis (paired t-tests for normally distributed variables; Wilcoxon signed-rank tests for non-parametric or ordinal variables; McNemar’s exact test for binary outcomes). Analyses were conducted in Stata (version 18).
Process evaluation
Followed-up participants were asked about any difficulties with the exercises or nutritional component, and their open-ended responses were recorded verbatim in a data tool. They were also invited to an audio-recorded interview to explore contextual factors and additional feedback. Interview data were analysed using descriptive content analysis.
Results
Participant characteristics and follow-up
Of 124 eligible individuals, 108 (87.1%) participated, completed baseline assessments, and received the intervention (Figure 1). Mean age was 42.2 years (SD 8.7 years; range 24–62 years), and 26.9% were women (Supplementary Table S1).
Figure 1 is a participant flow diagram for a trial. The diagram is arranged vertically with labelled blue stage markers on the left and white boxes connected by black arrows. At recruitment, 108 participants consented to participate. An arrow leads downward to the 1-month call stage, where 50 out of 108 participants, or 46.3%, completed the one-month check-in call. From this stage, arrows lead to three 2-month follow-up outcome boxes. At 2 months, phone follow-up was completed for 21 participants, representing 19.4% of the total sample. In-person follow-up was completed for 54 participants, representing 50.0%. No follow-up was completed for 33 participants, representing 30.6%.
Most had not completed secondary education (74.1%) and lived in homeless accommodation (69.4%). Chronic conditions were common, particularly respiratory (31.5%) and mental health disorders (29.6%), with a median of one condition (interquartile range [IQR] 1–2).
Smoking was nearly universal (97.2%), 62.0% were on OAT, and recent illicit drug use was frequent: benzodiazepines (53.7%), cocaine (52.8%), and cannabis (52.8%); 32.4% reported alcohol use.
On CFS, 66.7% were pre-frail, with the remainder very mildly to moderately frail. On SHARE-FI, 61.1% were pre-frail, 32.4% frail, and 6.5% non-frail. Baseline resistance exercise was low (11.1%), although most (90.7%) walked ≥20 minutes at least weekly.
Follow-up and retention
At 1 month, 50 participants (46.3%) completed a support phone call, with up to three contact attempts; nine (8.3%) had no personal phone. Overall, 75 participants (69.4%) completed the 2-month follow-up: 54 in person and 21 by phone. Phone follow-ups could not include weight, grip strength, or CFS assessments, limiting frailty scoring.
Acceptability
Potential participants responded positively to discussions of frailty and the combined exercise and ONS intervention, with 108 of 124 eligible individuals (87.1%) consenting, indicating high acceptability in a primary care setting. At follow-up, most found the intervention easy to use: 84.2% of exercisers (n = 48/57) and 100% of ONS users (65/65) rated it very or somewhat easy (Table 2).
Demand
Of 75 participants followed up at 2 months, 70 (93.3%) engaged with at least one intervention component: 52 (69.3%) with both exercise and ONS, 5 (6.7%) with exercise only, and 13 (17.3%) with ONS only. Median exercise frequency was 3 times/week (range 1–7, IQR 3–7, n = 57), and median ONS intake was 7 times/week (range 3–7, IQR 7–7, n = 65).
Binomial logistic regression (Table 3) showed participants living with frailty had lower odds of exercise adherence than those at risk of frailty (odds ratio [OR] 0.29, 95% confidence interval [CI] = 0.09 to 0.98, P = 0.046); no other predictors were significant. For ONS adherence, no predictors reached significance, although frailty trended toward lower adherence (OR 0.28, 95% CI = 0.06 to 1.32, P = 0.108) and OAT use toward higher adherence (OR 2.97, 95% CI = 0.68 to 12.89, P = 0.146).
Implementation
The intervention was delivered within routine primary care without extra staffing or structural changes. Introducing it during a consultation took 3–5 minutes, including exercise demonstration and ONS explanation, and follow-up was integrated into existing GP workloads. Implementation was therefore feasible within current primary care infrastructure at minimal additional cost.
Practicality
Delivering the intervention in a primary care setting for people with complex needs required balancing acute issues during brief consultations. Unpredictable attendance, phone changes, and housing instability complicated follow-up. Despite this, the low-intensity, flexible design allowed opportunistic delivery without disrupting routine workflows.
Adaptation
The core materials, including the exercise leaflet and ONS, remained unchanged, although participants sometimes made informal adaptations to suit physical limitations or routines. This flexibility highlights the intervention’s simplicity and potential for further tailoring in wider implementation.
Integration
No formal system changes were needed to integrate the intervention into routine primary care. Materials were delivered during consultations, and ONS was prescribed via existing pharmacy arrangements. Delivered by two GPs without specialist staff or equipment, the intervention could be implemented in similar settings for PEH without structural changes, given clinician support within existing resources.
Expansion
Originally developed in general practice, the intervention was successfully implemented in a GP clinic for PEH. Its low-cost, non-specialist format and brief delivery allowed adaptation without extra infrastructure. Despite follow-up challenges from walk-in attendance, the intervention fit well with primary care workflows, supporting potential real-world implementation in similar clinics.
Limited efficacy
As part of feasibility, participants reported perceived benefit over 2 months (Table 2): 32.9% felt ‘much better’, 44.3% ‘slightly better’, and 22.9% ‘about the same’, with no one feeling worse. Although based on a short-term convenience sample, these findings suggest a positive impact on general health. Limited efficacy, per Bowen’s model, is reflected in secondary outcomes such as frailty scores and weight.
Secondary outcomes
Among 54 participants completing in-person follow-up, frailty improved over 2 months (Table 4). The distribution of CFS scores shifted towards lower values, and SHARE-FI scores fell from 2.19 to 1.09 (P<0.001), with participants without frailty increasing from 9.3% to 48.1% and participants living with frailty decreasing from 29.6% to 14.8% (P<0.001). Weight (+0.3 kg) and handgrip strength showed no significant change (P = 0.56 and P = 0.96, respectively).
Across all 75 follow-up participants, significant improvements were seen in SHARE-FI components including exhaustion (P<0.001), appetite loss (P<0.001), and slowness (P<0.05).
Process evaluation
Sixty-six participants reported on their experience with the intervention, and 26 took part in audio-recorded interviews. Most described positive effects, including increased energy, improved mood, and better appetite. One participant noted: ‘Feeling like it’s really improved things, it’s given me motivation. I'm doing it first thing in the morning … it gives me a great start to the day.’ Several participants adapted the exercises to suit physical limitations or developed complementary routines, with one reporting: ‘I have adapted the exercises to meet my needs.’
Barriers to engagement were also discussed. This included pain (‘Some of the exercises did cause some pain, but overall found helpful’), lack of motivation, hospitalisation, injuries, or difficulty accessing the ONS owing to pharmacy issues.
Homelessness-related factors were also noted, with participants losing exercise materials during hostel moves, competing survival priorities, and limited capacity to focus on health activities. Comments such as ‘Being homeless, it’s not on my mind’, ‘A lot going on for me, have barely been able to get a sandwich’, and ‘Lost the exercise sheet between hostel moves’, reflected the impact of instability on day-to-day engagement.
These findings suggest that while the intervention was experienced positively for some, engagement was sensitive to the broader realities of experiencing homelessness. No serious adverse events were reported.
Discussion
Summary
This study showed that a combined exercise and nutritional intervention for adults experiencing homelessness who are living with, or at risk of, frailty can be feasibly delivered in primary care. Recruitment and retention were high (n = 108/124 enrolled; 75 followed-up), although 21 follow-ups occurred by phone, limiting physical measurements, and more than half did not complete the 1-month call, reducing insight into sustained engagement.
Most participants engaged with at least one intervention component, with higher adherence among individuals at risk of frailty. The process evaluation highlighted how participants’ living situations and the practical realities of experiencing homelessness influenced engagement with the intervention. The intervention was implemented without requiring additional staffing or infrastructure, was generally easy for participants to follow, and there were no intervention-related adverse events.
Exploratory outcomes showed statistically significant improvements in frailty scores (P<0.001), while weight and grip strength did not change significantly, potentially relating to the short follow-up period.
Strengths and limitations
A key strength was successful recruitment and follow-up of a relatively large sample of PEH, a group often described as ‘hard to reach’.27 Embedding the study in a familiar primary care setting enabled engagement with participants who might otherwise face barriers to research participation. However, total patients screened were not recorded, reasons for declining were not explored, and recruitment by treating GPs may have introduced perceived obligation.
Limitations affecting generalisability include exclusion of participants with severe frailty (CFS ≥7), single-centre recruitment, and an open-label design that may have introduced bias, particularly as self-reported improvements occurred without changes in grip strength, although secondary outcomes were exploratory, intended to inform future trials.
Pragmatic considerations included no prespecified continuation criteria, lack of formal cost evaluation, and use of a non-validated self-report for perceived health benefits to reduce burden. These highlight areas for enhancement in future studies, including economic evaluation and more rigorous outcome measures.
Comparison with existing literature
Frailty interventions are generally more effective in older adults at risk of, rather than living with, frailty,36 a pattern reflected in our study. This may be amplified among PEH owing to instability and complex health needs, highlighting the importance of intervening at the pre-frail stage.37
Despite potential barriers, engagement was generally high, potentially owing to delivery by treating clinicians within a dedicated primary care service for PEH, which has been suggested to support recruitment with research in this population.38 Evidence also shows that familiar settings, continuity of care, and trusting relationships enhance healthcare uptake among PEH,39,40 suggesting that incorporating frailty interventions in such services may facilitate engagement.
Implications for research and practice
While housing remains a key determinant of health for PEH, tailored clinical interventions are also needed. Our findings support frailty-focused interventions delivered within primary care to improve outcomes.41
Specialist primary care can address health inequities,38 and provide a platform to engage PEH in research. Future implementation of frailty interventions could benefit from multidisciplinary input, particularly from inclusion health teams and addiction specialists, alongside holistic assessment of mental health and social needs. Individuals transitioning to supported housing may be key targets, as housing stability could facilitate frailty reversal.
Participant experiences highlighted that engagement is shaped by homelessness-related instability, with insecure accommodation, illness, and pharmacy access disrupting participation. Future scale-up may require flexible delivery, simplified materials, and research informed by the practical realities of experiencing homelessness.
A definitive RCT should include longer follow-up, multidomain frailty measures, blinded outcomes, flexible follow-up, and staffing continuity.27 Comprehensive process evaluation, fidelity assessment, and qualitative exploration could clarify why individuals with greater frailty engaged less. Alternative approaches, such as realist evaluation or ‘living labs’, may better accommodate practical challenges and contextual factors affecting implementation and outcomes.
Notes
Funding
TC was supported by an ICGP / Aspire Post-CSCST Fellowship award.
Ethical approval
Ethical approval was obtained from the Irish College of General Practitioners Research Ethics Committee (application number: ICGP_REC_2024_2082).
Provenance
Freely submitted; externally peer reviewed.
Data
The dataset relied on in this article is available from the corresponding author on reasonable request.
Acknowledgements
The authors would like to sincerely thank all participants for their time, engagement, and invaluable contributions to this study. We also extend our gratitude to the staff at the Granby Clinic for their support and facilitation throughout the project. Additionally, thanks to Professor Noel McCarthy. Finally, TC gratefully acknowledges the support of the Irish College of General Practitioners (ICGP) in the conduct of this research.
Competing interests
The authors declare that no competing interests exist.
- Received December 8, 2025.
- Accepted December 17, 2025.
- Copyright © 2026, The Authors
This article is Open Access: CC BY license (https://creativecommons.org/licenses/by/4.0/)
References
LinkedIn
Jump to section
Keywords
More in this TOC Section
Related Articles
Cited By...
