Using financial incentives to support service engagement of adults experiencing homelessness and mental illness: A qualitative analysis of key stakeholder perspectives

Abstract Introduction Homelessness and mental illness are associated with poor service engagement, health and health service use outcomes. Existing literature suggests that financial incentives may effectively support service engagement of this population, but studies investigating key stakeholder perspectives are lacking. This study aimed to elicit, using qualitative methods, nuanced service user and provider experiences by using financial incentives to support service engagement among adults experiencing homelessness and mental illness. Methods This qualitative study is part of a larger mixed‐methods pragmatic trial of financial incentives (Coordinated Access to Care for the Homeless—Financial Incentives [CATCH‐FI]) within a community‐based brief case management programme (CATCH) in Toronto, Ontario. Twenty‐two CATCH‐FI participants were purposefully recruited to participate in in‐depth, semi‐structured interviews; five CATCH service providers participated in a focus group and seven key informants in individual interviews. Data collection occurred between April 2019 and December 2020. All interviews and the focus group were audio‐recorded and transcribed. Topic guides prompted participant perspectives on and experiences of using financial incentives to support engagement, health and well‐being. Grounded theory and inductive thematic analysis guided coding and interpretation of transcripts. Triangulation and member‐checking enhanced the analytical rigour and validity of findings. Results CATCH service providers, key informants and subgroup of CATCH‐FI participants perceived financial incentives to directly facilitate service engagement. The majority of CATCH‐FI participants however highlighted that intrinsic motivation and service quality may be relatively more important facilitators of engagement. Most study participants across stakeholder groups perceived that financial incentives have direct positive influences on health and well‐being in enabling access to basic needs and simple pleasures. Conclusions Our data suggest that for some adults experiencing homelessness and mental illness, financial incentives can directly support service engagement. In addition, financial incentives may positively impact health and well‐being by easing financial stress and enabling deeper attention to individual health needs. Further research on the effectiveness and acceptability of financial incentives is needed to improve understanding and uptake of a promising intervention to support health and health service use outcomes in an underserved population. Patient or Public Contribution Study participants provided input into the study research questions, study design, interview guides and interpretation of findings.

The majority of CATCH-FI participants however highlighted that intrinsic motivation and service quality may be relatively more important facilitators of engagement.
Most study participants across stakeholder groups perceived that financial incentives have direct positive influences on health and well-being in enabling access to basic needs and simple pleasures.
Conclusions: Our data suggest that for some adults experiencing homelessness and mental illness, financial incentives can directly support service engagement. In addition, financial incentives may positively impact health and well-being by easing financial stress and enabling deeper attention to individual health needs. Further research on the effectiveness and acceptability of financial incentives is needed to improve understanding and uptake of a promising intervention to support health and health service use outcomes in an underserved population.
Patient or Public Contribution: Study participants provided input into the study research questions, study design, interview guides and interpretation of findings.

K E Y W O R D S
financial incentives, health services research, homelessness, mental illness, qualitative, service engagement

| INTRODUCTION
Adults experiencing homelessness and mental illness face significantly worse health and health outcomes compared to the general population, including increased prevalence and severity of chronic health conditions, comorbid alcohol and substance use disorders, neurocognitive impairment and premature mortality. [1][2][3] Engaging this population in health services is challenging due to transiency in housing, complex health needs, financial barriers, limited availability of tailored and appropriate services and stigma and discrimination. [4][5][6][7][8] In addition, the literature on interventions to improve service engagement among adults experiencing homelessness and mental illness is limited. As low levels of engagement are associated with a range of poor outcomes, including greater illness severity, lower quality of life and higher rates of acute care use, 6,9,10 implementing strategies to improve service engagement of this population remains a priority across healthcare and social service settings.
A review of existing literature suggests that person-centred care and a strong therapeutic alliance within a recovery-oriented framework are helpful in supporting service engagement in people experiencing mental illness. 6 Studies also indicate that providing instrumental supports and services that attend to immediate needs, such as financial, housing and employment assistance, in addition to direct mental healthcare, are important for this population. 11,12 Financial incentives (FIs) in particular have been used successfully to influence health decisions in a variety of populations and healthcare settings. [13][14][15][16] It has been suggested that FIs may be particularly effective in facilitating service engagement of underserved populations in shorterterm interventions. [17][18][19] For people experiencing homelessness and/or mental illness, existing literature suggests that FIs have been successfully implemented to improve attendance in psychotherapy services 20,21 and to increase rates of medication adherence, 22 abstinence from substances 23,24 and smoking cessation. 25,26 In healthcare, behavioural economics suggests that one way in which FIs may effectively influence individuals' health decision-making is by appealing to our general tendency, as humans, to focus on the present and on immediate gratification versus future-oriented pay-outs. 17,27,28 Motivation theories, including self-determination theory, 29 also align with this notion and further suggest that incentive-based interventions may be relatively more effective among extrinsically motivated individuals. 30 While the literature has long-identified intrinsic motivation as central to sustained health behaviours, 29,31-33 little is known regarding the extent to which and how the experience and impact of FIs might differ within and across populations and settings.
Despite growing evidence of effectiveness, there is limited research on the acceptability of FIs, or a nuanced understanding of stakeholder perspectives on the impact of FIs on health service engagement in underserved populations. Moreover, significant ethical concerns have been raised, particularly regarding perceived coercion and its potential impact on autonomous decision-making, and the potential for unintended harms; service providers, researchers and planners have raised concerns that money might enable increased substance use, for example. [34][35][36][37][38] Given debated appropriateness of FIs to promote service engagement, more evidence is needed to better understand key stakeholder perspectives related to impact, utility and ethicality, specifically including perceived impact on autonomy and unintended consequences, which together with evidence of effectiveness, can help inform acceptable implementation in practice.
This study aimed to elicit, using qualitative methods, nuanced service user and provider experiences with financial incentives to REID ET AL. support service engagement among adults experiencing homelessness and mental illness posthospital discharge in Toronto, Canada.

| Intervention description
This study describes the qualitative component of a larger mixedmethods pragmatic randomized controlled trial (RCT), Coordinated Access to Care for the Homeless-Financial Incentives (CATCH-FI), described in-depth elsewhere. 39 This RCT aimed to evaluate the impact of FI on service engagement among adults experiencing homelessness and mental illness posthospital discharge in Toronto, Canada. The CATCH-FI study enroled and randomly assigned participants of a brief case management programme, Coordinated Access to Care for the Homeless (CATCH), to either: (i) an intervention arm, in which participants received a $20 FI for each week they contacted their CATCH case manager or another CATCH service provider (up to $80 per month) over a 6 months follow-up period or until they were discharged from the programme according to their care plan; or (ii) a control arm, in which participants received usual CATCH care without an FI to support engagement with the programme.
The CATCH programme itself has also been extensively described and evaluated. [40][41][42][43][44][45] In short, CATCH is a multidisciplinary brief case management intervention that bridges multiple organisations and sectors to provide comprehensive, short-term support to individuals experiencing homelessness and mental illness and transitioning from hospital to community services. The programme, informed by the critical time intervention model, 46  Thirty-four participants were recruited for this qualitative study, a sample size sufficient for achieving thematic saturation within each stakeholder group. Twenty-two CATCH-FI participants were purposefully selected based on the number of contacts with the CATCH team, sociodemographic representativeness, group assignment and study staff's assessment of participants' ability to provide in-depth reflections on their experiences. Eligibility criteria and recruitment efforts were adapted throughout to improve the representativeness of the study sample; for example, efforts were made to include participants with low levels of engagement and minority groups. This recruitment approach has been used by the research team in previous studies among people experiencing homelessness and mental illness 48,49 and within this particular programme. 40 In addition to CATCH-FI participants, 12 CATCH service providers and key informants were purposefully recruited based on their role in the CATCH programme or the local healthcare system.
CATCH service providers and key informants ranged in age, gender, experience and healthcare role. Focus group participants (n = 5) were CATCH programme staff in frontline (case management, nursing) and managerial roles. Key informants (n = 7) were healthcare providers in senior clinical and/or administrative leadership roles at leading local mental health and social service institutions that are partnered with the CATCH programme and serve the broader homeless population in community settings.

| Data collection
In-depth, semi-structured qualitative interviews averaging 40  were asked whether they believed FI would have, hypothetically, impacted their engagement, health and well-being. Topic guides were iteratively developed and refined by the study team to capture rich and diverse perspectives.
The research team is led by a primary investigator (PI) with extensive experience conducting qualitative research among people experiencing homelessness and mental illness; and the study is sponsored by the MAP Centre for Urban Health Solutions at Unity Health Toronto, who similarly have a history of successfully engaging this population locally. Together, this team has successfully conducted several previous studies with the study population. Data collection was conducted by trained and experienced interviewers who were known to participants by virtue of having conducted prior quantitative interviews as part of the broader RCT. Rigorous interviewer training, ongoing transcript review by the study PI and study staff and investigator triangulation were also employed to strengthen the quality, rigour and trustworthiness of results. Lastly, a memberchecking process was conducted in which results were reviewed, refined and confirmed with CATCH service providers in August 2021. All participants provided either written or verbal informed consent to participate; access to an interpreter was available to facilitate understanding and a capacity-to-consent questionnaire was available for use by study staff as needed.

| Data analysis
Grounded theory and inductive thematic analysis guided the interpretation of transcripts. [50][51][52] This approach allowed for analysis of both themes informed by existing literature, such as behavioural economics and self-determination theory, as well as themes that emerged independently, from the data itself. Coding was completed by three rigorously trained researchers, using an established methodology. [53][54][55] Two separate codebooks and databases were developed for CATCH-FI participants and for CATCH service providers and key informants, respectively. To develop the CATCH-FI participant codebook, three researchers independently coded six transcripts and collectively reviewed results to identify a set of key codes.
Once consensus was achieved, a codebook was developed and iteratively updated, guiding the coding of these remaining transcripts.
Similarly, to develop the CATCH service provider and key informant codebook, two researchers independently coded the focus group transcript and three researchers independently coded two key informant interviews to identify a set of key codes from which to build the codebook. In instances where consensus on ultimate code application was not reached after an initial meeting between researchers, the PI was consulted and the consensus was achieved in a subsequent meeting. Inter-rater reliability was assessed using Cohen's κ statistic, 56 for which scores were substantial for both service user data (κ = 0.79) and service provider data (κ = 0.77); percent agreement on all codes was 99%. All interviews coded for the purpose of the codebook development were later recoded using the developed codebook. As coding progressed, codes were grouped into higher-order themes. The PI and study staff met regularly to iteratively review and refine the coding framework and emerging themes.
Data saturation was achieved in which no new codes or themes emerged from later interviews. QSR International NVivo 9 qualitative analysis software was used to support data management, coding and analysis.

| RESULTS
In investigating the perceived impact of FI on service engagement, health and well-being, three primary themes emerged from participant narratives: Cash is king: The first theme describes how FI directly facilitated engagement in a subgroup of CATCH-FI participants. The majority of CATCH service providers and key informants similarly believed that FI did or would likely directly influence service engagement (n = 7/12). This was primarily due to the perceived universal utility of a financial gain and specific lack of financial capital among the study population. As one individual described, Addressing basic needs: Study participants, including over twothirds of CATCH-FI participants, described how FI afford more opportunities to meet basic needs that directly support health and wellbeing, such as satisfying hunger, taking prescribed medications and reducing stress. As one CATCH-FI participant in the intervention arm described, 'It had a mental health [impact], yes. And yeah physical health "cause I was able to eat… I was going like entire days without food"' (P4010). Other CATCH-FI participants enroled in the intervention arm described using the FI they received for food, medications, supplies, public transportation fare and to pay bills: basic needs that 'reduces stress because I'm able to get the little things I need… that money allowed me to pay my cell phone bill and it left me a small amount to get the groceries I needed for the week' (P4249). The experience was impactful for CATCH-FI participants, suggesting that the intervention 'really helps financially when you're on a very low fixed income… I use that money to be able to help me survive' (P4061). CATCH-FI participants assigned to the usual care arm similarly described that receiving FI would be useful in directly enabling individuals to meet basic needs that are essential to health and well- CATCH service providers and key informants similarly highlighted the perceived utility of FI in meeting basic needs to support health and well-being: 'I think that realistically, incentives do support people being able to maintain their wellness and engage in care' (HP5). For example, CATCH service providers described how incentives in this study were used practically by CATCH-FI participants 'to get from one appointment to the next and to accomplish some of these very important tasks that they need to do: to get their ID, or meet up with a doctor… get their medication' (FG5). Overall, this stakeholder group generally agreed that the use of FI can positively support the health and well-being of this population insofar as 'anything that may be able to help people stabilize those basic needs is critical to supporting their ability to engage in slightly less pressing health and social care' (HP6).
Allowing for simple pleasures: Similar to meeting basic needs, FI allowed CATCH-FI participants to experience simple pleasures, small gains or rewards that directly improved their mental health and well-being by relieving stress or giving them an enjoyable experience to anticipate.

| DISCUSSION
This qualitative study explored stakeholder perspectives on the use of FIs to support service engagement and improve health and wellbeing among adults experiencing homelessness and mental illness in a large urban centre in Canada. Consistent with prior research, our findings suggest that FI may successfully facilitate service engagement for some service users. A subset of service users in this study described the prospect of immediate financial gain as an externally motivating factor that encouraged continued contact with service providers. These findings are consistent with behavioural health economics principles, which suggest extrinsically motivated individuals and those particularly biased toward present and immediate rewards may be especially likely to respond to FI. 17,27,28 Our findings are unique, however, in identifying that the majority of service users perceived FI to be less impactful than other key facilitators of service engagement. In particular, service users highlighted the relatively greater value of intrinsic motivation, which renders FI extraneous.
This finding aligns with self-determination theory 29,30 and previous literature, suggesting intrinsic motivation is a key ingredient in initiating and especially sustaining health behaviours. [31][32][33]57,58 Our study findings also speak to the importance of quality of care, which emerged as another key facilitator of service engagement across all stakeholder groups. Many service users in this study de- Taken together, particularly among the subset of service users who might want or need an extrinsic motivator to support initial engagement, our findings suggest that FI may offer an effective opportunity to initially engage or 'hook' some service users; and that a complementary focus on the quality of care by service providers may further help sustain engagement. The need to enhance sustainability is consistent with findings from a recent qualitative study in which FI were associated with initial motivation to engage in low-barrier human immunodeficiency virus care but were less effective in facilitating sustained engagement 60 ; and reviews of existing quantitative evidence indicate FI is particularly effective for singular behaviours 61 and short-term engagement and behaviour change. 18 Beyond the direct influence on service engagement, participant narratives consistently described clear, positive impacts of FI on health and well-being. Primarily, participants described that FI enabled them to afford basic needs (e.g., food, transportation fare, bill payments) and to enjoy simple pleasures (e.g., coffee, entertainment) that were in turn perceived to directly support their physical and mental health and overall sense of well-being. That only one service user reported using their FI to purchase illegal drugs is consistent with previous research 62,63 and noteworthy, considering that a key criticism of this engagement strategy is the potential to exacerbate substance use. 18,36 As suggested by service providers in this study, it is possible that helping service users fulfil basic survival and immediate needs may additionally support service engagement. Given the limited literature on this topic, further research into the mechanisms by which FI may directly and indirectly support engagement is needed.
A final notable finding in this study is the divergence in stakeholder perspectives. Overall, service providers consistently expressed the view that FI would likely support service engagement of adults experiencing homelessness and mental health needs. This perspective reflected the assumption that the perceived utility of FI, specifically among this population who lack basic health, social and financial capital, was or would be high and sufficiently motivating to independently drive service engagement. This stakeholder perspective stands in significant contrast to the majority perspective among service user participants, two-thirds of whom articulated that FI alone were not or would not be a significant influence on their decision to engage in care. This divergence suggests stakeholders may differentially perceive utility and ascribe value to FI.
Further research and a better understanding of service user perspectives on the extent to which, how and why FI are useful will be essential to designing and implementing acceptable client-centred and contextspecific interventions to promote engagement.

| STRENGTHS AND LIMITATIONS
This study is part of a larger mixed-methods RCT and our findings are strengthened by a methodologically rigorous design. Our resulting rich, multistakeholder narrative data add high-quality evidence to an underdeveloped literature base and significantly improve our understanding of the perceived impact of using FI to support engagement, health and well-being in an underserved population. Our findings from a large urban centre are context-specific and reflect the experiences and potential biases of study participants. Nonetheless, our findings may be helpful to other programmes or jurisdictions considering alternative strategies to improve engagement among people experiencing homelessness and mental illness and other underserved populations.
Study limitations include the cross-sectional design and the narrow demographics of the qualitative sample, which was primarily male, Caucasian and at least high school-educated. Differences between this qualitative sample and the overall trial sample may reflect, in part, the purposive sampling strategy. The study team made several attempts to ensure a representative sample, including through the expansion of eligibility criteria, but recruitment remained challenged by reasons such as poor health status, nonresponse to invitations to participate and refusal to participate. Although sample characteristics were generally representative of the larger trial population, participants with different sociodemographic, health status and service engagement profiles may perceive and experience FI differently.

| Conclusions
Key stakeholders described that FIs may improve service engagement among some adults experiencing homelessness and mental illness. Stakeholders also described that FIs positively impact health and well-being, easing financial stress and enabling deeper attention to individual health needs. Findings from this study add to an underdeveloped literature base on stakeholder perspectives and experiences of using FIs to improve engagement, health and well-being of homeless adults with mental health needs.