Caring for people experiencing homelessness in times of crisis: Realities of essential service providers during the COVID-19 pandemic in Copenhagen, Denmark

Objectives The COVID-19 pandemic shed light on rooted social inequalities and on housing as a crucial social determinant of health. Little is known on current practices and new opportunities to support people experiencing homelessness in a situation of a global health crisis. This study explores frontline workers’ experiences of providing essential services to people experiencing homelessness in Copenhagen, Denmark, during the first COVID-19 lockdown, and highlights best practices of care in future crises. Methods Semi-structured interviews were conducted in August–September 2020 with nine service providers working in six organizations offering essential services to people experiencing homelessness during the spring 2020 lockdown in Copenhagen. The data analysis, following inductive coding, drew upon the concept of disaster resilience. Results Several initiatives were undertaken by the municipality and local organizations to ensure the continuation and adaptation of essential services to people experiencing homelessness during the COVID-19 crisis. These included collaborations with and financial support from businesses, the municipality, and other service providers; a mobile test unit, temporary shelters, and isolation sites; and an increased availability of opioid substitution treatment. Several improvements are to be made, particularly regarding sustainability and long-term benefits of the initiatives, facility-level risk preparedness, universal access to essential services for undocumented migrants, as well as collaboration between service providers. Conclusions The lockdown offered the opportunity to experiment with innovative ways of working, of which many had a protective effect on people experiencing homelessness. This knowledge can be used to improve services and reduce the long-term vulnerability of people experiencing homelessness.

determinant of health [1,2]. People experiencing homelessness become more vulnerable during crises due to poor housing conditions, increased risks of disease spread, a generally poor health status increasing risks of health complications, and limited access to health information [1,3]. Closing indoor public spaces such as libraries and public bathrooms reduces access to hygiene facilities (i.e., showers, toilets, sinks), the cancellation of public events impedes the money earning potential from e.g. collecting bottles and cans, and increases social isolation among people experiencing homelessness [1][2][3]. To support future pandemic preparedness, it is critical to develop an evidence base of successes and failures on how service providers have responded to people experiencing homelessness in a situation of national health crisis.
People experiencing homelessness (including rough sleepers, shelter users, homeless individuals admitted in hospitals or institutions) in high-income countries have a lower life expectancy of up to 15 years difference compared to the general population [4][5][6]. They experience reduced access to healthcare systems [5,7,8], and use emergency services more than the general population, partly explained by high exposure to several risk factors (alcohol, tobacco, and drug use, lack of support system, violence or accidents, high rates of suicides, mental health conditions, etc.), low use of primary healthcare services, as well as a higher tendency to postpone seeking care [5,7,8].
Addressing disease outbreak risks in homeless services is particularly challenging. Research in the United States showed that service providers in the homeless sector face several structural challenges, such as lack of financial and human resources, and complex networks of providers that fail to coordinate and collaborate efficiently [9]. Literature has also demonstrated that individuals working in homeless services experience high levels of stress, burnout, depression, anxiety and secondary traumatic stress, which can affect the quality of services [10]. Recent studies on the impact of COVID-19 on homeless services discussed the importance of trauma-informed response, and trust-building between service providers and the service users to ensure their safety and adherence to guidelines [2]. Based on COVID-19, and on the SARS experience in 2008, outbreak control recommendations in homeless services include: a) direct channels between local authorities and service providers to communicate guidelines and restrictions; b) continuation of essential services to people experiencing homelessness despite infection control efforts; c) basic staff training on infectious diseases and standard screening procedures; d) site-specific plans for outbreak management (including personal protective equipment and social distancing measures), and the creation of city-wide contingency plan for homeless services; e) identification of sick users and staff, and availability quarantine and isolation sites [11,12], Furthermore, resilient shelter systems in "normal times" to cope with emergencies [13] and the integration of long-term homelessness reduction efforts into outbreak response and planning have been recommended [5,12]. As a result of these recommendations, this study assesses the impacts of the COVID-19 pandemic on essential services to people experiencing homelessness in Copenhagen, Denmark, and reflects on how the resilience of these structures can be enhanced for future crises.

Study participants and data collection
Using a phenomenological qualitative study design, in-depth semi-structured interviews were conducted with participants selected purposefully due to their common experience of providing essential services to the homeless population of Copenhagen, the capital city of Denmark, during the COVID-19 lockdown of spring 2020.
Organizations providing homeless services in Copenhagen were contacted via e-mail or telephone, resulting in the recruitment of five participants. Four additional participants were recruited through snowballing. The nine participants were employed at six different facilities (See Table 1) during the first lockdown and included three managers, two counselors, two social workers, as well as two former volunteers promoted to a paid position during the lockdown. One participant was no longer employed at the organization at the time of the interview. Most participants had a background in social sciences or social work and their experience in the field varied from a few months to 14 years. Three of the interviewees were males and six were females. The interviews took place in August and September 2020, in a location chosen by the participants. Interviews were held in English, lasted between 40 and 65 min and were audio-recorded. The interview guide covered topics such as the response of local and national authorities and how they supported the work of local organizations and the livelihood of homeless people, the participants' organizations' experience of the lockdown and adaptations that were made, future preparedness and planning, as well as participant's perceptions of how people experiencing homelessness had been impacted during the lockdown.

Data analysis and theoretical framework
The interviews were transcribed and anonymized. Transcripts, interview notes, and illustrative documents provided by the participants, were analyzed by systematic thematic coding using the qualitative data analysis software NVivo 12. No predefined category of themes was created prior to the start of the coding exercise.
The phenomenological qualitative approach was chosen, which allowed for an aggregated individual insider perspective of providing essential services during the lockdown, and can be used for policy evaluation and policymaking [14]. The approach fitted the field of the study, which consists of a relatively small and local network of service providers sharing similar experiences [15].
We drew on the concept of disaster resilience to understand how communities get organized and mobilize resources and social networks to cope with, and manage future risks [16]. Providers of essential services to people experiencing homelessness contribute to reducing their beneficiaries' vulnerability on many levels: provision of housing, food, and basic healthcare, counselling, administrative support, bridge with systems of healthcare and welfare, and other harm reduction initiatives. In this study, the concept of resilience was used to understand how organizations offering essential services to people experiencing homelessness organized themselves internally, with stakeholders in the network, as well as with their users, and developed coping strategies during the first wave of the COVID-19 pandemic. Paidakaki (2012) describes vulnerability as "a person's or group's incapacity to anticipate, cope with, resist and recover from the impact of a natural hazard" (19, p.139). The vulnerability analysis of a certain group depends on the hazards faced and on pre-existent push and pull factors that increase vulnerability and decrease capacity to cope [17]. According to Wisner, Gaillard & Kelman [17], these factors are divided in three categories: the root causes of vulnerability (e.g., historical and cultural context, ideologies), the dynamic pressures at play in the system (e.g., strength of institutions, economic crises, political structure, etc.), and the individual characteristics contributing to fragile livelihood (e.g., housing, revenue, social network, access to food, educational level, etc.).
To achieve resilience and safety, factors of vulnerability to hazards must be tackled at all levels of the progression of vulnerability [17]. In a crisis, the most imminent needs to fulfil are the ones that constitute a "fragile livelihood" in order to achieve immediate physical safety [17]. In our context, the work of local organizations mainly focuses on addressing immediate needs, and sometimes dynamic pressures as well (e.g., research projects, awareness campaigns or political pressure), as presented in Table 1. To build resilience in the current crisis and reduce vulnerability for the future, dynamic pressures and root causes of vulnerability must be addressed [17]. The model of the progression of safety, in Fig. 1, demonstrates this theory in the context of the COVID-19 response in Copenhagen.
Achieving immediate safety may be enough to cope with the ongoing crisis but does not address the structural conditions that make a social group vulnerable [17]. Even when primary needs are met during a crisis (i.e., white box in the model), dynamic pressures and root causes must be addressed to ensure long-lasting safety [17]. If this fails to be achieved, safety is likely temporary, and the factors of vulnerability will remain. The realization of a resilient society necessitates long-term initiatives tackling systemic deficiencies and root causes [17].

Integrating theoretical perspectives
Phenomenology focuses on the shared lived experiences of individuals, while the theories of resilience and vulnerability heighten analytical attention to their sense-making regarding coping with hazards at the societal level. Integrating both perspectives enables to explore opportunities to reduce vulnerability and strengthen resilience for people and structures that provide services to people experiencing homelessness, and thus also for the people experiencing homelessness themselves. The analysis of first-hand experiences and the account of recommendations of people working at the forefront of homelessness offers a bottom-up, community-based approach to policymaking. By bringing forward these voices, we identify strategies and policies that proved to be coherent and efficient on the ground, and the ones that were not feasible or challenging. This scope of analysis for crisis management offers a unique opportunity to bring forward innovative coping strategies developed by individuals and small NGOs, as well as positive initiatives that deserve to be considered for scale-up.

Ethical considerations
Ethical approval by the Research Ethics Committee for Science and Health of the University of Copenhagen was granted on June 24, 2020. The research project respects general data protection regulations (GDPR). A consent form regarding data processing was orally explained to and signed by participants prior to the interviews. Participation in the study was voluntary and not compensated.

Rapid adaptations in schedules and staffing
During the pandemic, essential service providers in Copenhagen were rapidly considered a critical function of the state, and organizations received crisis funding from the municipality, which allowed to extend the opening season of emergency winter shelters. It allowed organizations that received it to organize themselves and adapt to maintain the provision of shelter and the fulfilment of primary needs. All six facilities where the participants worked remained open during the lockdown. Several participants described a situation of chaos and confusion at the announcement of the national lockdown. They expressed a strong sentiment of urgency and the feeling of being needed.
"All those places where the homeless women are used to spend the whole day, they closed down. So, from the first day, we felt forced to keep open, and I guess everybody just stood together in some way. It was a really special time, I guess everybody could sense that. So, we were all just working overtime until we had a schedule and were covering in, so that we could stay open, so we didn't have to send [the users] out." (staff, O3) Several participants used strong vocabulary to describe their work environment and the organizational changes that they experienced, such as: "I pretty much hit the ground running when I got there" (staff, O4), "for our sanity" (staff, O3), "it was such a crazy 24 h service" (staff, O4), "the pressure was on" (staff, O6) or "sometimes, we were just looking at each other like our head is exploding" (staff, O3). The intensity of the words chosen by several informants to describe changes in their work environment (extension of opening hours, poor planning, longer shifts, busier environments, and frequent disruptions of workflows) indicated higher stress levels than usual.
On the other hand, service providers stressed that the users had benefited from this new schedule: "During the lockdown, we were only closed for two hours [a day]. So, the [users] who stayed in [our unit] were really happy, because some of them have been homeless for a long time and they've been leading a life where they have to get up and go out for so many hours every day. Suddenly, they could, like I do in the weekend, just sleep in, walk around in your pajamas all day and eat in bed. All the things that don't really seem like a luxury when you just have it. But when you don't have it, it's something you miss." (staff, O3) Circling back to the framework of the progression of safety ( Fig. 1), essential service providers extending their opening hours secured the immediate needs of their users: access to food, hygiene, and sanitation, and to warm and safe accommodation. This more stable routine, paradoxically, offered a safer environment to some users than before the pandemic.

Communication channels
Service providers identified different communication channels: between the authorities and the organizations, between the management and staff, and between staff and users. Regarding the communication of public health measures and restrictions between the authorities and service providers, most organizations did not have direct channels established with the authorities but learned about the measures from the media, and relied heavily on COVID-19 governmental hotlines when in need of clarifications.
Regarding communication and information-sharing with users, participants noted that many users were amused or annoyed by restrictions regarding social distancing and hygiene. Nonetheless, several of them mentioned a positive attitude from the users and a general respect of safety guidelines. Participants from a facility specialized with homeless migrants expressed disappointment towards the lack of access to official information for non-Danish-speakers and mentioned spending considerable time translating relevant guidelines and restrictions in different languages to ensure access to information for users. When planning emergency response, it is crucial to ensure that crisis communication and guidelines reach the whole population at risk with minimum delays and barriers [18]. Therefore, ensuring that non-Danish-speaking populations have timely access to information plays a great role in the efficacy of the response. Following criticism and high spread among migrant communities [19], the Danish government later on started providing information regarding COVID-19 and regulations in twelve languages [20].
Two interviewees mentioned communication issues between staff and users of their services. One of them, working with homeless migrants, pointed out a general mistrust in the authorities and fear of being deported as a challenge in obtaining good cooperation: "I know there were problems with people not believing that it was real, coronavirus, and getting very angry, and saying: this is an excuse for you to kick us out of the country. And they just didn't believe that they had it, especially because they didn't have symptoms […]. It's like this natural distrust of anyone kind of working for them, I guess […]. That was a big issue, actually, because some of them refused to quarantine because of that. So, they refused, they said < No, I don't believe this. You're just trying to kick me out!> So, they left the shelter and didn't quarantine." (staff, O4) As exemplified, the lack of cooperation and adherence to guidelines were sometimes linked to beliefs regarding the pandemic being a hoax. Refusal to get tested or to isolate is a potential effect of the fear of immigration services and of mistrust in the authorities, which increases risks of spreading the disease and might have severe health consequences for the affected person who avoids healthcare [21].
Olagnier & Mogensen (2020) argued that the efficient control of COVID-19 in Denmark during the spring of 2020 was partially explained by the general trust in the authorities and institutions, resulting in high compliance of the population [22]. During emergency response, it is crucial that the authorities gain and maintain the trust of the population, including people without legal residency status, to enhance adherence to guidelines and limit outbreak risks in marginalized groups.
Lack of social skills and/or severe mental health issues, were also considered a challenge in communicating and informing about the outbreak situation, as the following quote describes: "We have people who, still now, don't know that there is a virus because they are mentally ill and they are so into drugs and alcohol… So, we have people who don't know or don't care because the voices in their heads are so big. So, it doesn't matter if it's COVID-19, or it's christmas or it's cold, because they have bigger issues." (Manager, O5) Several participants shared the idea that many users did not particularly worry about the COVID-19 pandemic because "they have bigger issues". They stressed that many homeless individuals experience a constant situation of unmet needs. Thus, while the pandemic constituted a major disruption for the majority of society, people experiencing homelessness are used to perpetual crisis and the focus for many of them remains the fulfilment of everyday essential needs such as shelter, food, showering, earning money, etc. [1,23] To facilitate information dissemination in future crises, clear communication channels from the government shall be designed, ideally with an appointed person in the municipal administration for organizations to contact directly when in need of clarification or assistance. Furthermore, inclusive information and communication channels for non-Danish speakers and other groups with communication barriers shall be available as soon as possible. Finally, to prevent misinformation and build trust with marginalized populations, only clear and transparent crisis response strategies, based on up-to-date science, shall be implemented, and misleading or false information shared by government officials should be publicly addressed.

Implementing public health measures
All facilities followed similar social distancing and hygiene measures concerning the number of people allowed in indoor spaces and a minimum distance of 2 meters between the beds. In organizations providing sleep-in services, restrictions translated into fewer beds available and fewer users. All organizations implemented similar guidelines regarding availability and use of hand sanitizer. Food, beverages, and dishes were only handled by staff. They implemented more rigorous cleaning plans than usual (e.g., disinfection of shared areas, surfaces, and door handles).
In an infection risks management effort, one of the organizations separated their staff into two teams that worked alternatively each week. This measure was meant to reduce contact between staff members and lower risks of contagion in case staff contracted SARS-CoV-2.

Collaborations with businesses and solidarity effort
To ensure safe housing during the pandemic, two organizations concluded agreements with hostels and hotels that could not receive guests because of the lockdown. One organization negotiated opportunities for people experiencing homelessness to move to a hotel of their choice for a limited period. The purpose was to provide safe housing while ensuring that users maintained regular contact with social workers and advanced their caseworks, as to avoid simply moving a situation of isolation from the street to a hotel room.
"The most vulnerable people are not coming inside because they don't want to be in shelters, they don't want to be with other people, and we saw that all the hotels were empty. So, we started to put these people in the hotels, but there were two criteria: the first one was that they choose their own hotel, because we always hear from the homeless that they have to do something that other people decide. So, they could choose their own hotel, and nobody chose the most expensive thing. Almost all of them chose a hotel near the area where they normally sleep on the street. And the other thing that we said was [that] you can get this hotel, but you have to meet with a social worker every day." (manager, O5) The participant noted positive results from this Housing First experiment that was made possible because of the lockdown. The beneficiaries of the offer seemed to quickly stabilize once they were able to stay in their chosen hotel, presented signs of better physical and mental health, and were more inclined to meet with their social workers. The Housing First approach is increasingly adopted across Europe and North America, and has shown numerous benefits for its beneficiaries in terms of long-term housing stability, and reduction of use of emergency health services [24].
Another organization moved their entire service in a hostel that was closed to the public. By the end of the lockdown, this hostel housed about 70 users in shared rooms. One participant explained that the agreement was not simply a solidarity gesture from the hostel, but became possible because of additional funding the organization received, and because the hostel could not accept clients at the time. The hostel staff was responsible for cleaning the users' rooms, which was a source of occasional tension. All personal contact with users was strictly handled by the staff and volunteers of the organization and not by the hostel staff. The participant assured that users would follow the rules of the hostel most of the time, but admitted feeling insufficiently trained to handle conflictual situations.
Other participants recalled expressions of solidary emanating from neighbors, individuals, and businesses. Some organizations received high-quality food donations, which was highly appreciated by users. A participant mentioned neighbor organizations allowing the shelter to access restricted parts of the building to use as isolation spaces, and a general propensity to offer help and negotiate physical space and property in a flexible and caring manner. The beginning of the COVID-19 crisis enhanced solidarity and tightened local social links in several contexts. Similarly, a study in Barcelona showed that micro-level solidarity and neighborhood ties had strengthened during the lockdown [25,26].
The closure of hotels, hostels, and indoor public spaces, allowed to rethink possibilities of offering shelter. It made possible pilotprojects that would likely not have happened if the lockdown had not precipitated such initiatives. One participant working in an outreach organization that provided hotel options to rough sleepers stressed the possibility to pursue the project sustainably on a larger scale and described the potential for users' stabilization that the project bears, contributing to the progression of safety for individuals experiencing homelessness. As discussed, vulnerability reduction efforts must embed long-term housing insecurity reduction and social reinsertion strategies [5,12,27].

Relationships in the network in Copenhagen
Essential services available to people experiencing homelessness in Copenhagen tend to work in silos and this was no different during the pandemic. Overall, participants described weak collaboration and coordination within the network of organizations. Most participants mentioned cooperating with other organizations regarding specific services to users, such as sharing lists of contact and services offered by other organizations to guide users to different shelters when capacity was full. Organizations sharing similar user groups (e.g., migrants, women) tended to cooperate more and have regular contacts, particularly during the lockdown. Nonetheless, participants felt that overall collaboration in the network should be improved: "I think the cooperation could be better, especially in Copenhagen […]. Because a lot of the organizations are fighting about the same money, you know? […] And some of the organizations, I would say, have difficulties working together because they feel it's a fight. I think it's getting better, it has been worse before. I think we have a good cooperation… But I think also that the lockdown showed… It was pretty easy to see who is used to working together and who wants to work together." (manager, O5) This quote describes funding uncertainty and instability faced by organizations providing crucial social services to society. In a commonly admitted extensive welfare state, community-based social work is subject to uncertainty of budget allocation, pushing organizations to compete instead of working towards a strong collaborative network. This type of inefficiencies in the homelessness sector was reported in other contexts as well [9]. The participant mentioned that in the event of future crises, information and transparency regarding allocation of funding from the municipality were crucial factors for organizations to plan activities and organize their workforce. It could be useful to consider transitioning to a different funding system that would increase financial stability for organizations, and potentially decrease competition for funding.

Collaboration with the authorities during the pandemic
Two participants from different facilities described a fruitful collaboration with the municipality during the lockdown. One mentioned the early creation of a working group involving the municipality and three organizations, including theirs, to address the needs of people who use drugs during the COVID-19 crisis.
"We thought [the municipality] had a very good and quick response. We made a group very quickly with [the municipality] and us, NGOs down here, it was called "samarbejdsforum" […]. So, we sat down, and we did that every week: we talked about our urgencies and what needed to be done, and they listened and they did it. So, things went very quickly, which normally doesn't, you know… If you have to change something, it's slow processes. But here, all the most urgent things, they actually helped us with." (manager, O6) The collaboration was so fruitful for all actors that the working group was maintained on a permanent basis. Another participant explained having weekly contacts with the municipal authorities and was met with a supportive and proactive attitude. A multistakeholder working group is a positive example of community engagement in disaster response and recovery planning, as recommended by the Sendai framework for disaster risk reduction [16,28,29]. Participatory approaches allow to identify and address specific risks, enhances network ties, and plan for sustainable vulnerability-reduction strategies targeting users of the organizations involved [28,30]. Although this working group is a step in the right direction, it would likely be more sustainable to create a larger forum including more organizations in the network, which would strengthen collaboration and sectoral capacity-building.
The intense collaboration with specific actors was not scaled up to the entire network, and many organizations had little to no direct contact with the municipality during the lockdown. Organizations in frequent contacts with the authorities were those that already received full funding before the lockdown. These were providing specific services that no other organization or institution provided (street outreach work, risk reduction for people who use drugs), which granted them a status of expert.
Another element of collaboration during the lockdown was user information-sharing between emergency shelters that share similar user groups. Emergency shelters usually work on a first-come, first-served basis. Two organizations mentioned collaborating to avoid users traveling too much between facilities and sharing information with one another regarding contact-tracing. They tried convincing users to stay in the same place for several days instead of traveling between shelters, in an effort to reduce risks of virus transmission. This initiative was particularly difficult to achieve, since staff had no means to enforce that users would remain in the same location.

Provision of financial support and protective equipment
Participants were overall satisfied with the response of the authorities towards people experiencing homelessness during the lockdown. Most participants considered the additional funding an appropriate measure which allowed to pursue activities and to try out innovative ways of offering services. Several participants working in organizations that had a mandate agreement with the municipality (because they fulfil the municipality's obligation to offer housing to homeless people with complex needs) received personal protective equipment (e.g., masks, disinfectant, gloves, and sterile suits) in addition to funding. The crisis funding was offered from March to May 2020. One participant mentioned challenges with planning, as the funding was released two weeks by two weeks. The funding could have been distributed in a more stable manner to allow facilities to organize themselves optimally and provide clear information to users and staff.

Mobile test unit, temporary shelters, and isolation sites
A positive measure discussed by all participants was the mobile test unit launched by the Capital Region, which drove to shelters and anonymously tested users. This initiative suited the realities of people experiencing homelessness, who are likely to face difficulties navigating the official system for COVID-19 testing (which required online booking, and the possession of a civil registration number, called CPR number). Anonymity was an important aspect of the project, given that many homeless individuals, particularly migrants, were nervous at the idea of subjecting themselves to testing in a non-anonymous manner.
"It was made only for the homeless […], so they can go where the homeless are and get them checked and have a quick answer. And that is pretty good because that is how the homeless population is living. They cannot go somewhere and wait for three hours at a hospital and be clean and then wait two days for an answer, because nobody knows where they are." (manager, O5) This initiative was a good example of a tailor-made response strategy that recognizes the special needs of people experiencing homelessness and adapts to the reality of the target group [18]. While recognizing the positive efforts to make testing available to people experiencing homelessness, one participant mentioned that it was surprisingly difficult for staff to access testing at the beginning of the lockdown. This difficulty should be interpreted in light of the shortage of tests at the beginning of the pandemic and the necessity to prioritize people in risk groups and health workers [20]. This issue was later resolved, when testing became widely available to the general population [20].
Several participants were impressed by the municipality's concern for people experiencing homelessness and the relatively rapid measures to ensure adequate care. They noticed a shift from the usual management of homelessness issues towards a solution-oriented strategy, demonstrating willingness to take rapid action and mitigate risks. Participants listed the opening of additional temporary emergency shelters, and temporary quarantine and isolation facilities as initiatives showing the municipality's effort to ensure sufficient housing during this critical period. The opening of isolation centers and additional shelter spaces reduced pressure on NGO service providers and ensured that all people experiencing homelessness could find shelter despite the reduction of beds in emergency shelters. Nonetheless, opportunities for the authorities to improve their response strategy were discussed, and one important criticism was the matter of sustainability and long-term planning. One participant was critical of the quality of the municipalityś temporary housing offers: "They weren't even appropriate. They weren't like proper housing. I saw [one] and it was just beds and there was nothing else. I think they didn't even have enough sheets for instance, you know? There wasn't enough equipment for people. So, they just provided a place to stay and that's it. They could have provided a proper place or whatever, and more structured, I don't know." (staff, O3) Aside from improper equipment, the participant also expressed doubts on the management of these temporary shelters, particularly regarding sustainability. They worried about the emergency housing capacity throughout the city and expressed frustration about a lack of long-term perspective, particularly amid a crisis that could evolve in many uncertain ways: "Now all those places are closed again, you know? So, I'm wondering where all these people are, because those places were also packed. I wonder where these people are sleeping now […]. They could have invested in more appropriate housing, you know, invest in a place that these people [could use]… Because we don't even know if there will be a vaccine or anything. Maybe this virus will just evolve or whatever, you know?" (staff, O3) As demonstrated in Fig. 1, structural changes and long-term planning are needed to reduce vulnerability and build resilience. Although additional shelter spaces responded to immediate needs during the pandemic, they were not designed for the long term, and therefore, were not likely to address dynamic pressures that put people in a situation of vulnerability. In the context of the COVID-19 crisis, future waves of infections and virus mutations are hazards to prepare for and integrate when responding to one outbreak. The response to the first wave of infections should have integrated these uncertainties and prioritized strategies that both respond to the current crisis and reduce vulnerability to the next [27,31,32]. The lack of sustainability of the additional shelter initiatives of the municipality demonstrated a missed opportunity to apply a long-term thinking into the disaster response strategy.
Other participants were also preoccupied by the sustainability of the municipality initiatives and expressed concerns about the proactivity of the authorities to help the homeless population being only temporary. One participant described the COVID-19 pandemic as a potential turning point for collaboration. Although the pandemic led to a global crisis causing severe public health, economic, and social damages, it created a momentum for restructuring political dynamics, re-evaluating social systems, and creating opportunities for collaboration and re-organization of civil society actors [33]. The participant insisted on the importance to cultivate momentum, and maintain the positive outcomes of the crisis.

Availability of opioid substitution treatment (OST)
In organizations working with people who use drugs, participants mentioned a positive attitude of the municipality towards facilitating access to OST during the spring 2020 lockdown. Direct access to OST in the organizations showed very positive results for the users, including increased chances of adherence to treatment, as well as in the context of isolation and quarantines stays. Inflexible attitudes generally observed in the healthcare system regarding the needs and realities of people who use drugs were challenged during the COVID-19 crisis in order to ensure adherence to public health measures and isolation stays. This initiative had positive impacts beyond the immediate public health needs brought by the pandemic and can therefore decrease dynamic pressures of vulnerability for people who use drugs by offering them an easier and safer access to addiction treatment and better chances of adherence.

Access to services for homeless migrants
The response deployed by the authorities regarding homeless migrants was discussed with all participants and raised the most criticism from all interviewees, including the ones for which migrants were not the primary user group.
People who are not registered in Denmark do not have a civil registration number (called CPR), which makes their access to health services very difficult [34]. Among unregistered migrants, a distinction is made between unregistered people with residency rights, such as European Union citizens, and people who do not have residency rights, whose presence on Danish territory is illegal [34]. In Denmark, free healthcare cannot be provided to people without a CPR number, except in medical emergencies [34]. In practice, unregistered homeless migrants face much more restrictive access to healthcare, shelters, and services than Danish and registered homeless individuals [1].
During the pandemic, the difference of treatment between people with a CPR number and those without was particularly tangible. One participant denounced a purposeful discrimination in funding allocation for organizations that work with homeless migrants, due to the authorities being generally unwilling to support more than the bare minimum of essential services for this group. Such organizations typically only receive partial funding to open emergency winter shelters. Other participants who worked in organizations that did not accept unregistered migrants also admitted that the main reason for this choice was access to municipal funding.
Regarding the management of the COVID-19 crisis with homeless migrants that needed testing, isolation or quarantine services, a participant stressed the difficulty for the authorities to respect laws regulating provision of state services while ensuring a coherent epidemic response in the homeless population. The lack of flexibility shown by the municipality regarding the provision of basic services to homeless migrants is engrained in a national system of rights recognition based on the possession of a CPR number [34]. The attribution of a CPR number is what makes a person recognized as a member of society and eligible to services in Denmark. The free Danish healthcare system was heavily challenged by the COVID-19 crisis, during which the health of people without a CPR number had the potential to put at risk the health of the entire population. One of the organizations specialized with migrants had an outbreak of SARS-CoV-2 in April 2020, which forced the municipality to intervene. The participant described a tense communication with the municipality regarding procedures for testing and isolation of users, as the organization insisted on maintaining anonymous services. Another participant from the same facility expressed strong disappointment regarding the timing of the municipality to act on the outbreak. It took three days for the municipality to admit the COVID-19-positive users in an isolation center. During this time, the organization had to isolate the 16 positive users in the basement of the facility, with little resources to enforce their isolation. Another participant mentioned an argument with the staff of a municipality-owned isolation center, who refused to keep a migrant person who tested positive to SARS-CoV-2 after discovering that the person had no CPR number. Nonetheless, it appears that the outbreak among homeless migrants pushed the municipality to revise its offer and turn towards a more inclusive and adapted strategy afterwards [3].
Another element of poor crisis management with homeless migrants was negative effects of anti-immigration discourses during the lockdown, resulting in misinformation and refusal of certain migrant individuals to cooperate with the authorities [1,3]. One participant expressed frustration with the crisis response measures and discourses based on anti-immigration politics instead of scientific evidence: "Nobody really wants to help them. And the only help they could get was if they get into isolation, and then afterwards get sent [to their home country]. Nobody took that offer because nobody wants to be sent home afterwards. So, I think that all the things that we did with the Danish homeless were good, and we got them pretty quickly inside […], we got them checked for the virus and everything. But for the migrants, the mobile EU citizens who live on the streets, I think that was the wrong way to do. And it was only political. If you talk to a doctor or a social worker or people who work with them, there was no reason why they did it like that. It was only because they had the wrong passport.
[…] And I think that the municipality of Copenhagen was better to fix it than the Danish government. But again, it was slow, and it was not as good as it could be. Because, when we have a big crisis like this, they shouldn't look at nationality." (manager, O5) Circling back to the progression of safety (Fig. 1), the deportation risks mentioned by the participant (based on a political speech by the immigration minister, Mattias Tesfaye, in the spring 2020) are examples of how dynamic pressures in society, such as stigma, institutional racism, and anti-immigration discourses, contribute to fragilize an already vulnerable group. Several participants mentioned rumors circling amongst users about the immigration authorities using SARS-CoV-2 testing to identify people without residency rights and deport them. Although none of the participants knew anyone being deported for this reason, rumors themselves had enough impact to potentially deter homeless migrants from seeking healthcare or using isolation and quarantine arrangements. These dynamic pressures had direct impacts on the immediate safety of migrants experiencing homelessness, among which several were reluctant to seek healthcare and testing, as they feared the immigration authorities more than the virus. Fears of deportations following a contact with the healthcare system were not observed in Copenhagen only. The World Health Organization report on social impacts of the COVID-19 crisis on refugees and migrants worldwide showed that 22% of people who did not seek healthcare (including testing), did so because of fears of deportation [21].
Mobile EU citizens are the largest group of unregistered migrants in Copenhagen [1]. Although they are not illegally in the country, they do not have access to services, including healthcare [34]. Many were concerned with risks of deportation, although they most likely would not have been forcefully sent to another country. The fact that many unregistered migrants appeared unaware of their legal status and rights regarding deportation is also symptomatic of their social exclusion and of the power of misinformation.

Preparedness to future risks
In the local service providers' network, preparedness includes the identification of risks and adequate preparation for their potential occurrence [12]. Such planning can consist of training staff on infectious diseases and symptom screening, clear and transparent communication channels known by all staff (including direct internal channels, lists of emergency contacts and helplines), contingency plans, including spatial separation plans in case of outbreaks, hygiene and food handling procedures, educative sessions and materials for users, sectoral strategic plans for outbreak management across organizations in the network, etc. [12].
In terms of outbreak preparedness, most participants admitted that infectious diseases risks were not a priority in their facility before the pandemic. Most organizations did not have contingency plans or special procedures for infectious diseases management and would usually rely on the governmental non-acute health hotline or emergency numbers if an event requiring health professionals occurred. The only exception was food handling and hygiene procedures in organizations operating a kitchen, which are required to follow food safety and hygiene regulations. Consequently, all participants but two described a relatively chaotic situation at the beginning of the lockdown and denoted issues regarding organization, schedules, and implementation of hygiene and social distancing measures. Despite the will to follow guidelines and restrictions, small organizations with no preparedness and little resources struggled to operate quick and effective changes and to comply with all COVID-19 regulations.
Two participants argued that their experience handling emergencies in their daily work had been useful in dealing with the COVID-19 crisis, which was not perceived as a particularly difficult situation compared to their usual work environment. One participant, although recognizing that their facility did not have any specific procedures for infectious diseases, insisted that staff had proven resilient and proactive dealing with the crisis, and would have been able to rely on existing emergency procedures if the facility had been affected. The second participant explained that their staff providing street outreach services was trained to handle emergency situations, to recognize, and react to infectious diseases signs. Therefore, their staff reported being sufficiently prepared and did not have to adapt their ways of providing services.
While the spring 2020 presented important challenges to service providers, it also offered a unique opportunity to reconsider the structures in place and revise priorities. Several participants highlighted that the COVID-19 crisis made them more aware of the importance of hygiene and disease transmission risks. They mentioned that cleaning and disinfection, and measures to avoid crowding, were likely to stay in place even when not requested by the authorities. Such precautions and the implementation of stricter hygiene protocols can be categorized as risk mitigation measures [31,35].

Discussion
The lockdown shed light on the extreme vulnerability and social isolation of people experiencing homelessness in Copenhagen, Denmark. Paradoxically, while part of the vulnerability of people experiencing homelessness is due to social exclusion, the lack of contacts with other citizens also limited their exposure to the virus and had a somewhat protective effect, as very few people experiencing homelessness contracted the virus during the first wave in Denmark [3,36]. Nonetheless, the health consequences of outbreaks in this population group would have likely been far more serious than for the general population, given the poorer health and poorer access to the health system [37,38].
Investment in long-term housing strategies for people experiencing homelessness coupled with integrated social and psychological support [5-7,12] is essential. As discussed by Benjaminsen [4], extensive homelessness reduction initiatives have a high implementation cost, but become more cost-effective in the long-term, as they reduce public expenses for emergency shelters, institutional placement of people experiencing homelessness (hospitals, rehabilitation centers, prisons, etc.), and improves the chances for social and professional rehabilitation [4,5]. Extensive and sustainable homelessness reduction strategies would benefit people experiencing homelessness, people at risk of homelessness, as well as society as a whole in terms of risk mitigation and vulnerability reduction [16,18].
This study provides new insights into the daily realities of workers in the homeless sector, but must be considered in the context of its setting. Copenhagen is an interesting setting because of Denmark's extensive welfare system, despite several weaknesses in services to marginalized people [39]. In addition, Denmark was relatively spared by the COVID-19 crisis during the first wave of the spring 2020 (22). These elements predisposed Copenhagen to have a potential for good crisis responses for people experiencing homelessness. Opportunities for change and lessons learnt can serve as a source of inspiration for policymaking in similar high-income urban contexts. The COVID-19 crisis stressed the importance for European authorities and for NGOs to plan and mitigate risks at all levels [12]. The generalization of access to healthcare for vulnerable migrants, at least in times of crisis, is recommended. Tailor-made initiatives for marginalized groups implemented during the COVID-19 pandemic should be maintained to improve the efficacy of health interventions for these groups (e.g., availability of OST, health screening mobile center, 24h housing options). Likewise, services developed specifically in response to the crisis (test units, quarantine and isolation facilities, shelters) should remain anonymous and ensure amnesty with the immigration authorities for users in situation of irregular stay. Furthermore, extremely vulnerable groups such as unregistered migrants and people experiencing homelessness shall be included in vaccination plans, not only because they are at higher risk to severe complications, but also because it would undermine public health efforts to reach herd immunity if this population group is neglected [37,38,40]. Decriminalization of citizen solidarity towards people in situations of illegal stay, particularly for organizations specialized in providing services for marginalized populations, should be introduced [34].
Moreover, and particularly during crises, it is important to improve communication and trust between the health system and people experiencing homelessness [7,12,34]. Miscommunication from the authorities, rumors and poor knowledge of legal rights can result in reluctance to accept testing or to use isolation and quarantine sites, due to fears of deportation. Recommendations to improve this situation include clear communication, in all appropriate languages, about the rights of unregistered and registered people, in collaboration with organizations providing services to these groups to ensure efficient dissemination. If unregistered migrants are protected against deportation during major crises or when seeking healthcare, they should be informed. Clear communication and transparency would increase trust and reduce risks of potential carriers of infectious diseases and people with critical health conditions not seeking timely healthcare. Additionally, healthcare and social workers who may be in contact with extremely marginalized people, such as people experiencing homelessness who are unregistered, or who have addiction or mental health issues, should be competent to handle interactions respectfully, and ensure high quality and dignified services through trauma-informed care [2,7,34]. Participants in the present study raised the necessity to adapt to the daily realities of people experiencing homelessness, including difficulties to follow-up with appointments, to be punctual, lack of social skills, etc. One participant stressed a need for improvement in the municipal social services to enhance flexibility and adapt better to users. Indeed, experiences of homeless individuals within the healthcare system include feelings of inadequacy, the fear of being a burden, and a sentiment of invisibility [7]. These experiences can likely be generalized to interactions with the social system as well. Stigma, lack of communication skills, difficulty understanding bureaucracy, mental health conditions, and substance addiction are barriers that people experiencing homelessness face when interacting with the public system [7,8] For social work to be as efficient as possible, it is important to understand the barriers that the target group faces and to plan adaptive interventions that reduce these barriers, particularly during crises but also in normal times [18].
Regarding risk and crisis management within the providers' network, a crisis consultation group that includes municipalities and representatives of all interested civil society actors in the network should be created and actionable in case of future outbreaks [28]. This multi-stakeholder initiative shall ensure the dissemination of information and good practices among the network and allow all organizations to raise concerns and communicate their priorities. This could, for example, facilitate the creation of additional shelters to increase housing capacity during the crisis, as well as the quarantine and isolation facilities. Such collaboration, including collaborations for Housing First initiatives and long-term rehousing, is encouraged also in normal times.
This research has limitations. The results are based on self-reported data and although many of the information was confirmed by several participants, government or media sources, social desirability bias, or confusion regarding chronology cannot be excluded. In addition, the study was conducted at a local level and included a small number of participants, which limits the generalizability of their experiences to other contexts. Another limitation is that the study team was not able to conduct follow-up interviews later during and after the pandemic.

Conclusions
This study has uncovered the capacities and challenges of essential services in a European urban setting to care for people experiencing homelessness during the COVID-19 health crisis. Good practices, such as implementing a strategic working group with relevant actors, and early targeted initiatives for special-needs groups, such as the anonymous mobile test unit, were found to enable care and service provision. Other initiatives helping to safeguard the health and well-being of people experiencing homelessness were: special funding to prevent discontinuation of services, the opening of isolation and quarantine facilities, additional shelters to ensure sufficient housing spaces during the lockdown, as well as enhanced access to OST. Innovation arising from the pandemic response must be used to enhance the resilience of services and reduce the long-term vulnerability of people experiencing homelessness. A number of challenges were also identified. The situation of unregistered migrants in Denmark is extremely precarious and immigration policies during the public health crisis were not transparent, something other settings may also experience. The study findings offer an opportunity to advance reflection and research about legal rights and national identity in the context of global pandemics that challenge the notions of borders and citizenship.

Authorś contributions
CZ and SSJ conceptualized the study. CZ collected and analyzed the data, and SSJ, MS and IMG reviewed it. CZ prepared the manuscript, which was critically reviewed by all authors. All authors have read and approved the final manuscript.

Funding
The study was carried out without external funding.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Anonymized data are available by the first author upon request.