Integrated approaches to health, housing, and social support for PEH is emphasized by health policies including recent NICE Guideline on care for PEH [21]. However, limited attempts have been made to develop and evaluate integrated approaches to care using a gold standard methodology. To our knowledge, this is the first community pharmacy-based multicenter randomized controlled trial that achieves this. The study was able to exceed progression criteria for a definitive trial in terms of recruitment, retention, intervention adherence, and collection of outcome data [26].

Our original rationale for conducting this RCT in community pharmacy—that it would be a positive place to engage patients—was supported by the success in participant recruitment. Pharmacy staff used their acquaintance, prescription records, and informal conversations based on their rapport to identify suitable persons to invite. We also used prescription records to identify people who were unable to collect prescribed medicines in person in pharmacy due to health problems, and approached them via the third sector worker ensuring equalities in opportunities for participation in the trial.

The number of participating community pharmacies was increased and the recruitment period was extended in one site to ensure recruitment targets were achieved. Prior experience of researchers and the PHOENIx delivery team in one site, due to experiences with a previous RCT, is likely to have led to a more rapid recruitment rate and higher numbers of contacts with participants in the PHOENIx arm. It may be prudent in a subsequent definitive trial, to schedule a run-in period where researchers and intervention staff become familiar with local venues and the local NHS organizational context.

Researchers, and intervention staff, were recruited to the study team on the strength of their demonstrable empathy, street sense, active listening, non-judgemental attitude, and relationship building skills. HVCSEs provide unconditional support to PEH and this ethos was also important for all staff (researchers and intervention team) involved in delivering PHOENIx. We suggest these skills are important factors in the success of recruitment and follow-up, given the past and ongoing traumatic life experiences of PEH that impact on emotional regulation, executive functioning, and forming secure relationships [10].

Three-month follow-up was impacted negatively due to the temporary absence of the researcher and third sector staff in one site. In addition, some delays in patient recruitment in one site led to recruitment and 3-month follow-up of some participants ongoing at the same time which added additional pressure to the field researchers. Assistance from the wider network of third sector organizations and agencies were received to enable 6-month follow-up which maximized response rate. When participants had moved pharmacies, temporary accommodation, moved to a different city, or were imprisoned, this brought about anticipated governance and data linkage barriers. However, healthcare utilization data was successfully obtained with a high success rate in both sites. In Scotland, these were facilitated through Community Health Index (CHI) search records from the NHS databases. In England, we approached individual GP practices, hospitals, and emergency departments—albeit this was labor intensive.

The intervention showed potential in preventing deterioration of emergency health needs as excess ED presentations were reduced in the intervention group. Both the utility scores and visual analog scale (VAS) showed potential improvement in quality of life. No remarkable trends were observed across secondary outcomes such as routine hospital admissions and frailty (meeting three frailty indicators). Signals of improvement across social outcomes such as rough sleeping and involvement with criminal justice system were encouraging. In particular, the third sector worker offered express referral to housing, furnishing, and tenancy maintenance support where needed to prevent rough sleeping.

The above findings corroborate with the evaluation of PHOENIx model of care in other settings and a pilot RCT of PHOENIx model of care for PEH with a history of overdose [17, 31]. Sparse literature using RCT evaluation of care models prohibits wider comparison of results.

The range of consultation frequencies across participants in both settings is noteworthy. Having a responsive intervention team with dedicated time and connections to meet participants’ fluctuating needs and engagement styles was seen as important. Widening the scope of practice for clinicians on outreach may offer even greater immediate help, if mobile teams are linked to General Practices [37]. Whether the PHOENIx model of care, provided by highly qualified independent prescribing pharmacists and HVCSE worker, is capable of being rolled out at scale and pace to the UK’s homeless population is unclear, due to workforce challenges and costs. A health economic analysis and definitive trial will help clarify whether there is merit in expansion of PHOENIx at scale, or whether alternative models, e.g., HVCSE with a GP linked Health Care Assistant conducting regular outreach to support diagnosis and treatment, might be a generalizable approach remains to be explored.

Community pharmacies have been referred to as “untapped” resource in terms of their potential to offer opportunistic intervention including advice, referral, and urgent treatment to mitigate health inequalities [38]. Examples of emerging roles include screening and treatment of blood-borne viruses including hepatitis C [39] and HIV [40] in marginalized communities. The results of this study provide positive signs that community pharmacy can be a suitable avenue to identify and offer support to people facing the worst health outcomes in society. The intervention was delivered predominantly in the participants’ choice of venue out with the community pharmacy, underscoring the importance of having a nimble team, ready to go to the participant. This, together with strong links to the local (specialist inclusion health) GP practices, the strength of the relationship between PHOENIx and participants, and the generalist skills of the pharmacist for managing multimorbidities, is likely to be an important feature to retain in an intervention in a definitive trial.

Given the pilot nature of the study, it was not set up to undertake hypothesis testing so signs of improved clinical outcomes need further exploration in a full RCT. There were missing data across different data fields owing to non-response in the face-to-face/telephone follow-up and data requests from other service providers. Some of the case record forms completed also had missing data fields as often the researchers could not complete the extensive data collection form in one session with the participant. Some clinical measures were missing as we had to utilize community researchers (third sector workers) to undertake clinical observations (after additional training) such as blood pressure and BMI checks. These were not always completed. Both settings were urban, and given the rise in numbers of PEH in remote and rural areas, and variations in presentations in coastal areas, there may be merit in including PEH from rural and coastal areas in the subsequent RCT.

Variations in the nature and extent of support given to participants between settings, more work is needed to understand the reasons, e.g., pharmacist training and experience. Our qualitative evaluation (under review at the time of the manuscript preparation) may shed light on these variations, from the perspective of participants and stakeholders and PHOENIx delivery teams. Glasgow is a smaller city than Birmingham with fewer places for researchers and the intervention staff to seek and follow-up participants. Along with prior experience of delivery of the PHOENIx feasibility study in Glasgow and an established local support network, it has likely led to higher follow-up rates in Glasgow compared to Birmingham. A “run in” period, as occurs in clinical trials, may be indicated, to ensure delivery teams are able to navigate local complex systems, prescribing protocols, housing referrals, and familiarize themselves with local HVCSE venues. A future trial will investigate the roles of geographical/center-level practice variations in any observed differences across outcomes.

VCSE workers’ rapport and relationship with participants was key to successful engagement and delivery of interventions in both settings. Some participants were imprisoned during the follow-up period and it was only possible to follow-up some of these participants in prison. The intervention team and the researchers also had difficulties in engaging with participants who moved out of the local geographical area. We realized that researchers found it more challenging to follow-up UG participants compared to IG due to the lack of intervention team engagement with the former group. Our intervention team realized that there could be multiple and often overlapping professionals attempting to assist PEH but functioning separately. Strengthening collaboration within local care services is key. Additional funding within the trial to support social wellbeing such as gym memberships, skills training, clothing, furniture, and subsistence could promote participant engagements.

All study researchers and intervention staff had street sense, were trauma-informed, and applied harm reduction principles. Community pharmacies who participated in the study also had acquaintance with the study population as they were providers of OST and needle exchange services, and were located in urban centers/nearby where PEH reside.

The intervention team aimed for weekly contacts for each participant which would equate to six appointments a day. However, intervention staff often had to spend half or the whole day with a participant who had high needs, for example, needed help with bank account opening or accompanying to a further healthcare appointment. Such longer contacts were key to promote engagement and develop rapport between the intervention team and the participants. In one of the study sites, there was a temporary absence of support worker which may have led to some participants disengaging. Future studies should ideally be better resourced to cover staff sickness.

While recruitment was primarily opportunistic based on who was presenting to the pharmacies on the recruitment days, a future trial could also consider identifying PEH who are not physically able to walk to the pharmacy to receive services.

PHOENIx teams supported some participants intensively, and others less intensively. Future studies can benefit from early identification and prioritization for provision of additional support including assertive outreach for PEH with high needs but least likely to engage with the intervention. Amendments in local governance including prescribing rights (e.g., for substance use disorder treatment) and better skill mix within intervention team (e.g., availability of prescribers on outreach with diverse clinical competencies) could minimize barriers to rapid provision of treatment and adaptation of intervention to PEH needs.

Given the limited space and availability of consultation rooms, some of the baseline and follow-up assessments by the researchers and intervention staff visits were conducted in community venues in close proximity to the pharmacy where both the researcher and the participant felt comfortable. Our experience also suggests that recruitment in wintertime can also be challenging due to low footfall in pharmacies.

Lessons were learned with regard to challenges in following up participants who were in prison, or moved out of the local geographical area. In Scotland, prison personnel helped the study team by accommodating visits by the study team for follow-up. However, in England, governance barriers did not allow visits to take place during the study period. Utilizing Rough Sleepers Outreach Team from local authorities to find participants would be key given their wider approach, access to participant accommodation data or “know how” of where a participant was known to bed down or beg. Pharmacy records were less useful in finding participants when they had moved to a different pharmacy as details of the new provider were not available. A future study should also gather information on participant whereabouts (including UC) more frequently during follow-up rather than at 3 and 6 months, and follow-up for longer, given the likely median duration of homelessness of over 20 years in some cohorts [17].

Future studies could also utilize the community pharmacy team to directly offer clinical assessment and prescribing for PEH as their roles in this study was limited to participant identification, referral, and provision of usual care. There is a potential to upskill pharmacy staff in research capacity such as in obtaining informed consent to facilitate recruitment of participants out of hours and to maximize efficiency in larger trials. This would require advanced agreement with clinical and information governance in local NHS areas, given that community pharmacists in the UK are contractors rather than NHS employees (PHOENIx pharmacists). PHOENIx teams also had support from Pharmacy leads in NHS Primary Care (Glasgow) and Acute (Birmingham) which supported information and clinical governance, both of which are pre-requisites to safe and effective clinical practice, and lacking in the organizational structure of community pharmacy in Glasgow and Birmingham. With prescribing rights being expanded to pharmacists in the UK [41], their roles in provision of generalist care for vulnerable populations have the potential to mitigate health inequalities.