A comparison of the COVID-19 response for urban underserved patients experiencing healthcare transitions in three Canadian cities

Objectives The COVID-19 pandemic and response has highlighted existing strengths within the system of care for urban underserved populations, but also many fault lines, in particular during care transitions. The objectives of this study were to describe COVID-19 response policies for urban underserved populations in three Canadian cities; examine how these policies impact continuity of care for urban underserved populations; determine whether and how urban underserved community members were engaged in policy processes; and develop policy and operational recommendations for optimizing continuity of care for urban underserved populations during public health crises. Methods Using Walt & Gilson’s Policy Triangle framework as a conceptual guide, 237 policy and media documents were retrieved. Five complementary virtual group interview sessions were held with 22 front-line and lived-experience key informants to capture less well-documented policy responses and experiences. Documents and interview transcripts were analyzed inductively for policy content, context, actors, and processes involved in the pandemic response. Results Available documents suggest little focus on care continuity for urban underserved populations during the pandemic, despite public health measures having disproportionately negative impacts on their care. Policy responses were largely reactive and temporary, and community members were rarely involved. However, a number of community-based initiatives were developed in response to policy gaps. Promising practices emerged, including examples of new multi-level and multi-sector collaboration. Conclusion The pandemic response has exposed inequities for urban underserved populations experiencing care transitions; however, it has also exposed system strengths and opportunities for improvement to inform future policy direction.


Introduction
The COVID-19 pandemic and subsequent public health response have compounded pre-existing health and social inequities. Urban structurally vulnerable populations encompass those individuals who face additional risk factors or inequalities related to the social determinants of health, and perpetuated through underlying systemic or structural factors (Clark & Preto, 2018). When these vulnerabilities affect access to quality healthcare, an individual is considered underserved. There are many ways that people can be vulnerable to poor health outcomes and experience inadequate access to health and social care as compared with the general population (Patra et al., 2007;Pottie et al., 2020). For instance, communities living with poverty, unstable housing, and more frequent physical and mental illness and substance use also experience challenges accessing effective care. These proximal determinants of health intersect with more distal influences, such as the impact of racism and Canada's colonial legacy, to compound medical underservice for urban Indigenous peoples (McLane et al., 2022;Browne et al., 2010). Despite the potential for high-continuity care environments to reduce the morbidity related to social determinants of health, provide safer and higher-quality care, improve trust and satisfaction with care, and lower system costs (Starfield & Shi, 2004), urban structurally vulnerable populations report high unmet need for care (Bhui et al., 2006), often seek acute care as their main access point into the healthcare system (Hwang et al., 2011), face obstacles in accessing primary care (Greysen et al., 2014), and describe being treated poorly when accessing care (Martins, 2008). Predisposing adverse circumstances, fewer enabling supports, and intersecting complex needs create barriers to successful transitions from one care space (e.g. hospital) to another (e.g. shelter-based care) (Gelberg et al., 2000;Martins, 2008;Virapongse & Misky, 2018). Further, the traditional organization of services exposes structurally vulnerable patients to a "candidacy"-oriented negotiation process, in which patients must prove their worth when seeking services (Dixon-Woods et al., 2006). Thus, structurally vulnerable populations are not well managed by the formal health system, and medically underserved.
The arrival of the COVID-19 pandemic to Canada has introduced additional challenges to this medically underserved group. Urban underserved populations are at higher risk of COVID-19 infection due to social factors precluding the ability to effectively engage in physical distancing and/or handwashing (Green et al., 2021;Okonkwo et al., 2021;Tsai & Wilson, 2020). They are more likely to experience severe COVID-19 illness due to comorbidities, and experience exacerbation of baseline poor health due to constrained access to harm reduction and treatment options for conditions such as substance use disorders and/or mental illnesses (Green et al., 2020;Henderson et al., 2021;Karamouzian et al., 2020). Moreover, countermeasures to control COVID-19 transmission have required modified approaches to care, including replacing in-person visits with virtual healthcare, restricting visitor and outreach activities in healthcare facilities, and temporarily closing community clinics and municipal amenities. The unintended consequence of these measures is disruption in the continuum of supports typically available to urban underserved populations. Since they may not have sufficient resources to adapt, the urban underserved may experience disproportionate deterioration in continuity of care despite their higher risk of illness and illness severity (Douglas et al., 2020;Smith & Judd, 2020;Tsai & Wilson, 2020).
Around the globe, healthcare providers, outreach workers, community advocates, and other stakeholders tending to the health of urban underserved populations were called upon to mobilize supports in response to pandemic-related challenges. Research has exposed the increased risks and challenges experienced by individuals dealing with substance use and homelessness during COVID-19 (Aronowitz et al., 2021;Parkes et al., 2021;Pixley et al., 2021). The response to COVID-19 drew in part from experience with epidemics, natural disasters, and other emergencies (Edgington, 2009;Leung et al., 2008), and also from communities' history of mobilization due to government inaction (Friedman et al., 2007). However, these earlier experiences provide little specific policy guidance on how to optimize care continuity and reduce care fragmentation during a viral pandemic response, thus leading to challenges in addressing unique population needs during the COVID-19 pandemic.

Objectives
The COVID-19 pandemic and response has highlighted existing strengths within the system of care for urban underserved populations, but also many fault lines. When care transitions go poorly during a pandemic, the implications for population health include the risk of higher community transmission, worsening of poverty and other social determinants of health, and increased morbidity and mortality. We sought to examine COVID-19-relevant policies involving governments, health authorities, non-governmental organizations, advocacy groups, and community members at the local, provincial, and national levels that impact continuity of care, especially during transitions between care spaces arising from illness. The specific objectives of this study were to (1) describe health and social COVID-19 response policies for urban underserved populations in three cities (Edmonton, Winnipeg, Toronto); (2) examine how these policies impact continuity of care for urban underserved populations; (3) determine whether and how urban underserved community members were engaged in policy processes; and (4) co-develop policy and operational recommendations for optimizing continuity of care for urban underserved populations during public health crises.

Design
Case studies are appropriate for in-depth investigation of complex real-world phenomena with numerous intersecting influences; as such, the COVID-19 response is ideal for a comparative case study approach to policy analysis. Health and social policies are developed in the complex interactions between the content of policy, the actors involved, context, and processes. Walt and Gilson's Policy Triangle framework (Walt & Gilson, 1994) was used to explore the interrelationship and interaction among four main components of policy-making which include actors (individuals, groups, and organizations involved, and their interactions with one another), processes (how policies are formulated and implemented), context (socio-political, cultural, economic, and health and social care setting), and content (the policy's substance and details such as objectives, decisions made, and implementation plans) within the policy documents from different cities. The policy triangle framework was used to organize and systematically examine how these four components might affect policy decisions on continuity of care for urban underserved populations. Furthermore, the policy triangle is consistent with a definition of policy inclusive of both formal decision-maker directives and broader public interests, ideas, and actions. Whereas most policy analyses focus on policy content, this framework also assesses the context, actors, and processes involved in policy development and implementation. Emphasis on these factors can help promote more effective, informed, and pragmatic policy (Walt & Gilson, 1994) (Fig. 1), which is important in the context of COVID-19 and future emergencies.

Setting, participants, and partnerships
The case study compared the COVID-19 response from January to December 2020 in urban underserved settings in Edmonton, Winnipeg, and Toronto. These cities provide complementary yet distinct populations, settings, and experiences during the first year of the COVID-19 pandemic (Table 1). Toronto's urban underserved population is very densely concentrated and multicultural; in contrast, Edmonton has a comparatively large Indigenous population and medium population density, and Winnipeg also has a large Indigenous population (including many northern Manitobans displaced to the urban setting both prior to and during the pandemic) but with a lower population density. With respect to COVID-19 cases during the study period, Toronto saw a higher number of cases earlier in the pandemic, Edmonton witnessed an intermediate but rising number of cases, and Winnipeg experienced its first cases later with fewer first-year cases overall. Public health is primarily administered at the provincial level in Alberta and Manitoba, whereas the municipal level largely directs public health activities in Ontario. All three provinces were governed at the federal level by a centrist party and at the provincial level by a conservative party, but with an ideological mix represented on municipal councils. Last, all three jurisdictions were engaged in active work long term to address continuity of care and care transitions for the general population (Peckham et al., 2018). Policy actors of interest ranged from senior policy makers to front-line service providers, and included the government, health sector (public health, acute care, primary care), and social sector (housing, non-profit organizations).
Supporting urban underserved populations during a pandemic requires attention to patient and community engagement, given that the integration of people with lived experience in service planning is associated with increased trust, strengthened relationships between team members, sustained collaboration, and systemic transformation (Jagosh et al., 2012). For traditionally excluded patient populations, this participation can reduce inequity, confer agency, and increase public awareness of issues affecting them (Jürgens, 2005). The COVID-19 crisis has presented significant potential tensions between patient-centred care, patient safety, and Fig. 1 Adapted policy triangle framework. Source: Walt and Gilson (1994)

Document review
In the first phase of the case study, publicly available written policy documents that address the COVID-19 response specific to urban underserved populations were retrieved and reviewed using a targeted search strategy. Primary data sources included websites of relevant local, regional, and national stakeholders involved in COVID-19 policy and/or health or social policy specific to urban underserved populations. This included Hansard, multi-level government ministries and departments, health authorities, councils, and relevant community, health, and social support agencies. With assistance from a university librarian experienced in searching grey literature and team members familiar with local stakeholders and data sources, internet and database search strings were created and tested using pre-specified COVID-19, urban underserved population, source, and setting terms and modifiers. Using the same search strings, primary source data were complemented by internet searches for media articles to corroborate the timing and content of policy documents and identify relevant policy directions not yet identified in primary data. January 2020 was set as the start date for collection of Canadian COVID-19 policy content, and active searching continued until December 2020 to capture any policy developments as the pandemic evolved. Policy documents and media articles were considered for inclusion if they addressed the COVID-19 response specific to urban underserved populations. This included policies and media articles for unique settings (e.g. acute care, isolation hotels, addiction treatment facilities) and subpopulations (e.g. women, older adults, Indigenous peoples). Policy documents and media articles were excluded if they were specific to youth and/or children, or if they were not publicly accessible. Policy documents were also excluded if newer, updated versions were retrieved. Two reviewers tested and refined document inclusion and exclusion criteria on initial search results until > 80% agreement was reached. A four-reviewer team subsequently screened and selected documents, ensuring two reviewers screened each document and at least one reviewer for each document was familiar with the city being referenced.
Selected documents were imported into qualitative analysis software (ATLAS.ti). Text sources were analyzed inductively, paying specific attention to document elements that address continuity of care and/or lived experience involvement. Using inductive latent content analysis (Mayan, 2009), two team members reviewed the first 10 primary documents from each study city (for a total of 30 documents) to develop a codebook guided by the four elements of the Policy Triangle framework (context, content, actor, and process) (Walt & Gilson, 1994). Once consensus was reached on the codebook, it was then applied to all remaining documents and refined iteratively, with at least one coding team member familiar with each city being referenced. Codes were then clustered into categories and eventually themes based on the Policy Triangle framework elements. A policy timeline of events was created in tandem with in-depth coding.

Key informant group interviews
In the next phase, document review findings were triangulated with key informant group interviews. Front-line and livedexperience policy actors from each city with a variety of identities, roles, and perspectives were recruited from each city with assistance from front-line partner intermediaries, focusing on these actors due to their relative under-representation in preliminary review of available policy documents. Front-line partners approached potential participants via existing networks based on their involvement in the planning and/or delivery of COVID-19 response services for the urban underserved, seeking a multidisciplinary and sociodemographically diverse range of informant perspectives. Partners then connected interested individuals to the research team to review study information electronically and provide consent and demographic information prior to focus group sessions. Sessions involving 3-7 participants each were moderated by primarily female-identifying, academic and clinician team members, held using the Zoom virtual platform for public health reasons, and lasted approximately 2 h. Front-line service providers gathered for one session in Edmonton, one in Toronto, and two in Winnipeg, and a final multi-site lived-experience group was also convened. Consenting participants were offered an honorarium of $50 if the session was conducted during personal time to acknowledge their contributions.
Building on the emerging themes from the document review, the semi-structured interview guide assessed how and why policies were developed (or not developed) and whether they were carried out as planned. Because these themes were less prominent in the document analysis, specific questions explored how well policies worked, what would have worked better, how continuity of care was addressed, and how urban underserved populations themselves were engaged. Interviews were audio-recorded and transcribed verbatim, and researchers created field notes.
Again, using an inductive latent content analysis approach and guided by the Policy Triangle framework, research team members applied the document review codebook to interview transcripts. The coding framework was refined for any responses that were not captured within the earlier examined textual data sources. This resulted in the refinement of existing codes but also the development of a modest number of newly arising codes. Following this descriptive analysis, qualitative analysis software was used to perform cross-case comparisons. Major findings from the document review and interviews from each case were compared and synthesized.

Results
A total of 237 public-facing local, regional, and national-level policy documents and media articles that spoke to policy decisions published between January and December 2020 were retrieved and analyzed, including 53 government, 51 nongovernment, and 133 media texts. Five group interview sessions were then held with 22 participants, including 16 frontline practitioners and six lived-experience individuals. Further breakdown of document and key informant sample characteristics are presented in Tables 2 and 3 respectively, and a full list of policy documents by type and jurisdiction is presented in Table 4. A high-level policy event timeline is also presented in Fig. 2. The findings which follow provide a synthesis of the triangulated policy document and key informant interview themes organized according to the four Policy Triangle framework elements, followed by a fifth section on promising practices that were highlighted by participants for their positive impact during the pandemic and potential future benefit. Policy content and context were similar across all three cities, whereas the actors and processes were more varied. With the exception of digital literacy barriers (a theme unique to interview data), presented themes were present in both document and interview data sources, with the interviews lending additional context and nuance to document-derived themes.

Policy content
Across all three jurisdictions, the COVID-19 policy response occasionally addressed continuity of care. However, it was not a key consideration for most policy documents-emphasizing instead the control of COVID-19 spread-and a number of policies were developed that could have unintended negative care impacts for urban underserved populations. Strategies recommended to maintain health and social service continuity included using technology to bridge continuity gaps and adjusting service delivery locations and formats (e.g. group to individual). For example, as described in a policy document from the Ontario Ministry of Health: It is recognized that much of the support and care that is provided by community-based Mental Health and Addiction service providers may not be deferred.
Organizations are encouraged to customize and prioritize service. 8 These written policy recommendations did not consider the interpersonal nature of support for underserved populations and the importance of face-to-face engagement to promote continuity, which was highlighted by one key informant: I would say because more people came for physical follow-up visits in the past, in general I find I have a higher success rate among that population with inperson visits than I do with getting them on the phone. EP2 For drug poisonings in particular, this loss of contact from new isolation requirements could have devastating consequences: People who are used to living in congregate spaces now have own rooms, closed doors, even when you had overdose prevention sites in the hotels people would still understandably want to use in their own rooms and with a toxic drug supply that's led to, as we've seen from coroner's reports the catastrophic losses that are more than double previous years. TP1 Moreover, many early government-level policies reflected a lack of familiarity with resource availability on the front lines (e.g. lacking hardware to participate in virtual care) or community members' personal resources (e.g. lack of family support, internet/phone access, ability to shelter in place) and required subsequent adjustments later in the pandemic. Adjustments included Toronto offering Wi-Fi and cellular supports, Edmonton offering free transportation to COVID-19-specific care spaces, and Winnipeg opening additional winter shelter spaces.
Policies also focused primarily on people experiencing visible homelessness, with minimal attention to the pandemicrelated continuity needs of other structurally disadvantaged groups such as precariously or unsafely housed individuals, people experiencing violence, or people who use drugs. Notably absent in any formal policy was specific guidance for women and gender non-binary groups, apart from shelter bed allocations for women and reminders to service providers to collaborate as needed: System leaders in homeless services should ensure they are informed of the response being taken in the Violence Against Women sector, and effectively, and as necessary confidentially, communicate that information to services. 89 Indigenous peoples were more frequently mentioned, with the provision of specific cultural supports in Housing First bridge housing facilities in Edmonton and isolation shelters in Winnipeg, and the creation of a parallel support strategy for Indigenous peoples facing homelessness in Toronto.
Implemented supports were also often temporary despite potential advantages to longer-term care models. In all three cities, for example, documents and key informants verified that live-in isolation facilities with on-site primary care-style supports were implemented for COVID-19-symptomatic people experiencing homelessness:       But now they might be there for a couple weeks because they're recovering so it gives people a lot more time to get to know that person and get maybe thinking about it might be good to call their family doctor while they're here and set up an appointment for the day after they leave and hey, they're not on [income support], maybe we should get them set up with income. W2P3 However, most documents stated that these supports would be retracted upon resolution of the pandemic. Many group interview participants in all three jurisdictions worried about the impact of these losses: I worry about the outcome of that when that is pulled away. 'Sorry, we don't have those services for you anymore and there's not a pandemic', right. W1P1 In summary, policy content was primarily focused on infection control, provided guidance and resources for specific groups (e.g. visible homeless) but not others, offered temporary support without clear post-pandemic transition planning, and infrequently addressed front-line resource scarcity or how to mitigate the risks of social isolation.

Policy context
The common policy agenda across all jurisdictions' documents was the need to protect the health of the general population, with the aim of protecting the health of urban underserved populations (homeless populations in particular) being secondary and less immediate in overall agenda setting. Key informants also explained that policies were developed within a context of multiple pre-existing system constraints as well as emerging system strain, and that cities seemed unprepared: It's like my city does not really have a backup plan when something disaster like happens. So, everything is like chaos in the beginning, city just tried to scramble, scramble anything temporary to put all these people, homeless people, people who need a warm place to stay overnight you know. LEP6 With few pre-existing emergency planning documents addressing emergency-associated health system overload, staff shortages, and socioeconomic changes, strategies were developed de novo to mitigate impending crises: Plan for employee absences and prepare by crosstraining staff. Resource operations as needed so that the focus can be on essential services. Be creative and flexible in service delivery. 44 Some policies were also developed to address challenges with pre-existing capacity and accessibility issues such as communal shelter designs. Many social and community services that facilitate access to and continuity of care, such as in-person income support services and dropin spaces, were markedly restricted. Other services required major expansion, relocation, and/or co-location to adhere to infection control measures. These sudden and sometimes disruptive changes to usual support pathways created intersecting challenges with COVID-19. Group interview participants alluded to even more difficult system navigation than usual: A lot of our participants ended up collecting CERB and being cut off assistance and then losing housing so, that was a real difficult thing to navigate and having meetings and advocacy with employment and income assistance to try and continue that relationship. It's hard enough for our clients to get on assistance, it was just making it more difficult as the whole welfare system shut down and had only one contact number too. W1P2 Participants also lamented the markedly reduced access to trusted care providers: Every few months I have to see my family doctor but since COVID hit, I haven' t seen her since March 2020 and then because of COVID, you don't see the doctor in person, most of them are virtual you know and at that time I don't have a phone line Fig. 2 Timeline of key policy events to call to talk and so, everything is broke down. LEP6 In tandem, participants witnessed the loss of access to safer common spaces in which to connect, exacerbating care fragmentation even further through social disconnection and the loss of direct system navigation support: With physical distancing measures, we haven't been able to let people just hang out in our waiting room. For obvious reasons, but I do think that has maybe interfered with being able to locate patients because traditionally you could just put a note on someone's chart and say okay, they hang out here all the time, they don't have a phone, but we know they're going to be just hanging out. EP1 Thus, the COVID-19 response seemed to occur within an already-strained system of care that was under-prepared for a new public health crisis, compounding pre-existing barriers to access and continuity of care and reducing the available opportunities to provide system navigation and social connection.

Policy actors
Governments and health system decision-makers were the most common actors in policy development. However, nongovernmental organizations (NGOs) and front-line teams often stepped in to develop a response where policy gaps existed. Government documents were primarily guidance documents, whereas NGO documents described fully implemented responses (e.g. Winnipeg Main Street Project's ninepoint priority actions (48); Edmonton coalition of NGOs delineating hours, partners, and services on site at temporary support facility (6)), reflecting the role of these actors in policy development.
Documents also highlighted that relative involvement of different actors varied by jurisdiction. In Edmonton and Toronto, governments and health system decision-makers figured prominently, with municipal governments most directly involved in policy. Community-based NGOs were actively involved in population-specific planning groups, and corporate groups (e.g., hotels for emergency isolation shelter in both settings, Toronto telecommunications involvement to facilitate virtual care) were brought into discussions to facilitate policy implementation. In contrast, the Winnipeg response saw private groups and NGOs leading the policy response and pushing decision-makers for funding and a comprehensive policy framework.
Though there were examples of engagement, very little policy in any of the three cities was informed by or developed in collaboration with community members, or even front-line workers: Honestly, I didn't see shit. Oh, excuse my language. But I didn't see any of our community members getting consultation about COVID at all. Just all of a sudden COVID hit and then all of a sudden boom, okay here's this temporary shelter and boom, here's what's going on, boom wear your masks, boom get tested just nobody asked. LEP3 Instead, urban underserved community members and advocates were vocal in the media, where they expressed concerns about such intersecting issues as drug poisonings, income support, and housing, and pushed for policy change. In Edmonton, for example, community members developed an encampment which put pressure on changes to local policy: Camp Pekiwewin has issued a new set of demands as organizers partner with inner-city agencies while continuing to pressure governments into lasting changes for Edmonton's homeless community. 106 In summary, the relative involvement of different policy actors varied considerably by city. Government and health system actors provided more proactive direction in some cases and more reactive direction in others. Front-line and NGO actors provided the comprehensive details necessary for policy implementation. Community members were seldom directly engaged and instead informed policy through informal means such as the media.

Policy process
Jurisdictions adopted slightly different approaches to policy development and implementation. In Winnipeg, bottom-up grassroots initiatives began at the NGO level without significant government-level collaboration, as explained by one key informant: So, we also came together as different programs, outreach networks, did a lot of stuff by email, online, sharing resources, Facebook was huge, being able to post resources and find foodbanks and getting the word out. W2P2 These non-governmental actors publicly called for increased government funding and action (e.g. COVID-19 outreach van, alternative isolation accommodations), but community members were not directly engaged.
In Edmonton, the approach was primarily top-down with multilateral collaboration between NGOs, municipal government, and other groups; NGOs adapted rapidly and collaborated to support care continuity. As verified by key informants, urban underserved community members were not involved, but made calls for increased funding and action on intersecting issues-in some cases supporting encampments which prompted action. In Toronto, top-down directives were also common but with similar multilateral collaboration, especially around implementation: We [front line clinicians] had given the forewarning [about emerging shelter outbreaks] and said as soon as the first signals of that, they went from being interested in what we were doing in funding some isolation support to 'Please come to the tables, you have to be in these spaces to actually help coordinate.' So, I think that would be very unusual, it certainly was for us to be invited to formal ministerial meetings and tables around our emergency systems planning. Usually, homelessness is not part of the acute care spaces. TP1 Most policies did not demonstrate lived-experience collaboration, with the occasional exception of Toronto outreach teams that produced specific outreach and overdose response guidance: This [consultation] process resulted in a series of learnings for future  recovery sites, and insights from those on the ground about how to best provide respectful and dignified care for those in need. 26 Community members and advocates in Toronto made similar calls for policy change, including filing a lawsuit against the city to uphold COVID-19 mitigation strategies in shelters, sparking major system expansion: A fairly large legal challenge... resulted in the legally mandated social distancing in shelters, such as the city had to lease over 20 hotels. So not just that they were doing it for isolation and protection, but they were actually mandated to do that by court injunction. TP1 Thus, different jurisdictions adopted varying degrees of top-down vs bottom-up policy development approaches. Though collaboration between service providers and higherlevel decision-makers eventually emerged, few instances of direct community member consultation occurred, contributing to indirect measures taken to influence policy.

Promising practices
Across all three jurisdictions, the urgent policy window created by the pandemic required equally quick collaboration and coordination, bringing together a variety of stakeholders more efficiently and effectively than in non-pandemic times, through such structures as emergency advisory committees: I see the spirit of collaboration continuing not maybe in the same, gusto perhaps… but I do truly believe for the first time I see hope that that collaboration will continue, and something changed during COVID and I'm really happy to see that. LEP4 On occasion, efforts were also made to include underrepresented groups. For example, Winnipeg developed a COVID-19 committee that prioritized membership for Indigenous peoples, newcomers, and individuals with disabilities.
Supporting virtual connectivity proved feasible in Toronto despite multiple structural barriers, where a partnership between telecommunication networks, charitable groups, and the municipal government supported the provision of shelters and affordable housing units with free wireless access and cellphones. However, this was a temporary intervention only, and did not address digital literacy or other barriers to virtual care: The individuals who didn't have phones and who didn't have internet, who didn't have computers, didn't have case workers to come over and provide their phones for these virtual appointments or coordinate them, because you obviously need someone to call in and book these appointments and if you don't have a phone, you can't do that. TP2 Edmonton chose to co-locate multiple health and social supports (e.g. housing intake, nursing care, isolation beds, supervised drug consumption) within expanded physical spaces: That's what happened here in Edmonton once COVID hit. We had a temporary shelter at the Edmonton Convention Centre, and we had all supports there. LEP3 Though intended to reduce the risk of viral transmission in smaller agency spaces, this also led to improved coordination between different services. This was paired with free public transportation to promote uptake as new supports were not within walking distance of the most underserved neighbourhoods.

Discussion
Although some promising strategies have been described to help maintain continuity during care transitions, the results of this policy case study analysis underscore a fragmented care system for urban underserved groups in Canada, echoing the emerging literature on inequities during COVID-19 for structurally vulnerable populations (Wojtak et al., 2020). Documents and key informant interviews especially highlighted pre-existing system capacity and resource challenges. However, the findings also suggest that COVID-19 has presented a unique policy window with opportunity to improve how the health systems support transitions in care for urban underserved populations. Though there were variable policy actor roles and response processes in each city, government and health system decision-makers typically provided highlevel guidance and funding, whereas front-line NGOs and other groups collaborated to operationalize and implement a number of innovative support strategies. Ideally, both highlevel and front-line policy actors should partner a priori to inform each other's roles and actions, and adapt to a rapidly evolving context more effectively. These partnerships would benefit from sustained post-pandemic collaboration to review promising innovations emerging from the COVID-19 response and determine how to maintain these supports to improve urban underserved care transitions over the longer term (Heimer et al., 2020;Wenger et al., 2021). Public health must play an active role in this partnered work given the inequities faced by urban underserved populations, the impact on community health, and the potential for upstream prevention.
With few exceptions, people from urban underserved communities were not included in developing policy around the COVID-19 response. This finding is echoed by other COVID-19 research demonstrating a relative lack of patient engagement (Wojtak et al., 2020). A lack of meaningful engagement risks compounding the inequities faced by urban underserved populations such as more frequent COVID-19 acquisition, higher severity, exacerbation of intersecting social issues, and less access to related supports. The sudden imposition of service restrictions, paired with the temporary nature of pandemic supports with no clear plan forward to sustain them (Kaur et al., 2021), may further erode urban underserved communities' trust in health and social systems. Conversely, as evidenced in this study and elsewhere in the literature, a collaborative and community-centred policy environment can mitigate many of these concerns (Morgan et al., 2021;Heimer et al., 2020). There is a clear need for an equity and justice lens in future emergency responses and policy development around care transitions, which would benefit greatly from a co-design approach with people with lived experience (Sayani et al., 2021).
The COVID-19 response demonstrated key areas of opportunity for policy and decision-makers to support transitions in care for urban underserved populations over the longer term. First, connectivity solutions could support virtual care for under-resourced individuals and organizations (Kaur et al., 2021;Ghidei et al., 2022). As these communication tools may be new for some service recipients, their utility will depend on ongoing front-line support during adoption of virtual options. Second, connecting people to bridge-and long-term housing supports continuity of care, in particular when interdisciplinary teams are integrated into bridge facilities and able to support transitions to more permanent housing. Third, the pandemic has underscored an urgent need to address Canada's ongoing drug poisoning syndemic and has forced service providers to reconsider traditional approaches to substance use management (Heimer et al., 2020;Wenger et al., 2021). More widely available harm reduction interventions, especially when paired with other supports such as shelters and healthcare settings, may encourage connection to essential health and social services while reducing the risk of death. Last, the co-location of multiple health and social services and removal of structural barriers to access (e.g. transportation) can promote more integrated care overall (Kaur et al., 2021).

Limitations
This comparative case study comes with several limitations. First, it examined a narrow policy window; not all policy decisions were public-facing or easily retrievable during the early waves of the pandemic. The key informant group interviews were in part designed to capture less well-documented policy and its implementation and reduce the likelihood that major decisions were missed. Another limitation is the use of virtual group interview methods, in which some participants may have experienced tacit digital divides to their full participation despite the offer of connectivity support from the research team. Further, the relatively conservative political landscapes in each province studied influenced the policy context and may potentially limit the applicability of findings to other conservative-leaning jurisdictions rather than more broadly. Last, a full intersectional analysis was beyond the scope of the study; policy documents contained little to no information specific to women, gender-diverse, and Indigenous groups, which restricted the scope of the analysis. The findings herein should be interpreted with caution for specific urban underserved subpopulations, and highlight the need for an intersectional approach to related policy development in the future.

Conclusion
The COVID-19 pandemic has had a disproportionately negative impact on urban underserved populations. Service constraints, time-delimited supports, intersecting crises, and minimal lived-experience consultation amplified the pre-existing system inequities faced by this population. However, the COVID-19 response has also demonstrated the feasibility of multi-stakeholder collaboration and support. Ongoing partnership in the form of government and decision-maker resources, front-line innovation, and lived-experience involvement is needed for urban underserved populations experiencing care transitions during COVID-19 and beyond.

Contributions to knowledge
What does this study add to existing knowledge?
• COVID-19 policy responses for Canadian urban underserved populations were largely reactive and temporary, compounding an already inequitable system of care. • Community members were rarely involved; however, a number of community-based initiatives were developed in response to policy gaps. • Promising practices emerged that should be considered for longer-term use, including examples of new multi-level and multi-sector collaboration, virtual connectivity supports, and collocation of services.
What are the key implications to public health interventions, practice, or policy?
• Given the inequities faced by urban underserved populations, public health professionals should apply an equity and justice lens in future emergency responses, in direct and timely partnership with people with lived experience and other policy stakeholders. Consent to participate All human participants provided informed consent.

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .