Dr. Whitley and Dr. Drake are affiliated with the Department of Psychiatry, Dartmouth Psychiatric Research Center, 2 Whipple Pl., Suite 202, Lebanon, NH 03766 (e-mail: rob.whitley@dartmouth.edu). Dr. Harris is with Community Connections in Washington, D.C.

Research suggests that upwards of 30% of the homeless population in urban areas have a severe mental illness and that most individuals in this subgroup have co-occurring substance use disorders and a history of trauma (1,2,3,4). It has frequently been noted that many are constantly revolving through an "institutional circuit" of hospital, jail, homelessness, and various housing programs (5,6). Providing suitable living quarters and supports for this population should be an integral component of psychiatric rehabilitation (7). Similarly, ongoing policies should ensure that the utmost is done to keep people housed once access has been gained.

Unfortunately, findings are equivocal regarding the relative merits of independent scatter-site housing with off-site support (commonly known as "supported housing") and traditional congregate housing with on-site support (8,9,10,11). However, these studies clearly suggest that people with severe mental illness who enter housing with any substantial support experience significantly better outcomes than those who enter housing with no support. This is especially the case during the transitional period to housing among formerly homeless or hospitalized persons. This is a critical time for adjustment and stability in this population, and intensive supports seem necessary to ease the transition (12,13).

Given that the evidence base does not imply a "one size fits all" solution, it has been argued that a range of housing options should be provided so that individual needs and preferences can be matched to available settings (14). This emphasis on consumer choice is also consonant with key tenets of the recovery paradigm and with wider shifts in psychiatry away from paternalism toward shared decision making (15). With all this in mind, a mental health services provider in Washington, D.C.—Community Connections—set up small-scale recovery-oriented housing in a single building in January 2005. The housing attempted to favorably blend some of the salutary factors associated with aspects of both supported housing and traditional congregate housing. This hybrid we call "recovery housing."

The building consists of 18 separate self-contained apartments, with Community Connections acting as landlord. Each of the apartments is leased to a Community Connections consumer who expressed a desire to live in a small-scale building with fellow consumers. Staff members do not live on site. The configuration of this recovery housing is based on various aspects of the literature. Its small-scale nature is in line with research suggesting that functioning and symptom reduction are facilitated in settings with fewer occupants and where a larger proportion of other tenants live with mental illness (16,17). It was also hoped that this configuration would overcome the isolation and prejudice sometimes reported by people living in supported housing, simultaneously giving consumers the desired independence and privacy associated with having a self-contained apartment.

Consumers who chose to reside in the recovery housing are all living with severe mental illness. The vast majority of the consumers in our sample had a history of physical or sexual abuse (or both), and most also had a co-occurring substance use disorder. As part of the recovery orientation of this housing, residents and staff developed a set of core directing principles from the outset. These stated that the building aimed to be a place where a sense of community could develop—one that aspired to be inclusive, safe, collaborative, empowering, permanent, self-reflexive, and self-correcting.

We used qualitative research methods to examine the progress of life in this building over a two-year period, from the lived-quotidian perspective of residents. The aim was to identify and understand factors influencing adjustment and stability in the recovery housing, especially during the crucial transitional period. This involved examining the role of the core recovery principles adopted by residents and staff and allowing consumers the latitude to identify factors of importance to them.

Focus groups with residents and concomitant participant observation were conducted over a two-year period between 2005 and 2007. Four focus groups, following guidelines set down by Krueger (18), were conducted at regular intervals (Table 1). Each meeting lasted approximately one hour. All residents living in the building at the time of the focus group were invited to participate by open invitation. A total of 17 residents participated in the groups; some came to them all, and others attended a few.

Fourteen of the residents who attended the focus groups were African American, and three were Caucasian. Fifteen were women, and two were men. Most residents were in their 30s or 40s. A well-trained graduate student and the first author jointly facilitated the focus groups. Focus groups followed a topic guide, the aim of which was to give voice to the views and ongoing experience of residents. Open questions probed which factors were facilitating positive experience in the housing and which were contributing to negative experience. For example, facilitators asked, "How has it been going here recently?" and "What could be improved here?" Issues arising spontaneously from earlier focus groups were followed up in later focus groups for elaboration. Focus groups were audiotaped.

Participant observation was also conducted around the same time as each focus group. This followed the guidelines set down by Spradley (19). The first author spent time casually mingling with residents, eating with them and regularly engaging in informal conversation. Similarly, the first author engaged in detailed conversation with case managers and supervisory staff. Notes based on these observations became an important component of the overall data set. After complete description of the study to the participants, written informed consent was obtained. Participants were given $15 for each focus group attended. The Dartmouth Medical School Institutional Review Board reviewed and approved the study.

Data analysis was conducted inductively. It was propelled by guidelines of qualitative content analysis, which involves the progressive abstraction of themes from raw data (20). Specifically, the first author listened to the recordings and distilled key themes from each group. The third author conducted a similar analysis independent of the first author. Subsequently, the authors discussed their independent findings and came to agreement over the most prominent themes. Consensus was rapidly reached; both analysts noted that in addition to numerous subthemes, the raw data pointed at a singularly dominant emergent theme. This was corroborated by the second author who was also acquainted with the raw data.

The pervasive theme to emerge from the data was that of safety and security. This issue was forcefully raised over and over again in every focus group and during participant observation. Even in the focus groups that were held toward the end of the study period, the issue of safety and security was raised spontaneously in answer to generic questions at the beginning of the group without prompting by the facilitator. At times, residents lauded the perceived safety and security inside the residence. This was especially the case during the first focus group, which was conducted just after participants moved in. There was a chorus of agreement when one resident answered the facilitator's question "What is the single best thing here?" with "Security! Your privacy! Not being violated! Everybody has their own key." The perceived safety was often actively compared with previous negative experiences on the streets or in homeless shelters. One resident declared, "Look where we come from! We were homeless, mental abuse, physical abuse, sexual abuse, spousal abuse." Positive feelings of safety and security were strongly prominent in this context of transition.

The first focus group was littered with examples of residents extolling the safety and security in the building. For example, "We look out for one another, my neighbors. They know me; they're there for me; it's secure. One of my associates came back and knocked on one of the other tenant's window, and I found out later, they didn't let him in, and that shows me the security part. They took the initiative and knocked on my door to see if I was home. That shows me I am safe. I feel real secure."

It is of interest that the resident in this example linked feelings of security with feelings of community. These feelings of community were commonly expressed in the first focus group and were also a prominent subtheme in later focus groups. Residents expressed a desire for meaningful connections with other residents. Such connections were often achieved—many friendships and alliances formed in the building. Some residents referred to fellow tenants as "sisters" (almost all residents were women), and there were many examples of mutual support. For example, one resident noted how her neighbor signed for her medication when she wasn't there, and another said that she sometimes shared child care with a neighbor.

However, by the time of the second focus group and throughout the later focus groups, residents stated that feelings of safety and security had considerably diminished and were under serious threat from the actions of a minority of fellow residents. Strong concerns were especially focused on danger associated with substance abuse relapse among fellow residents. Participants alleged that a small minority of residents would appear in the building "consumed by chemicals or alcohol," including crack cocaine. This led to a large degree of fear, anger, and frustration among those who were attempting to maintain sobriety. One resident noted that this situation made her "real on the defense," and another noted that she had "a verbal, almost a physical, altercation" with someone who was allegedly bringing drugs into the building.

A quote from a resident in the second focus group illuminates the forceful impact of other tenants' activities on safety issues: "I refuse to let anyone to use in the building. It's a recovery building. Look what you have done; you are in violation of your lease. When you do these things illegally, you put us in danger, because there are kids in the building. The drug crowd are gonna come around. We can't have it, no drugs coming into the building. I'll be damned if I'm gonna be exposed to this. You know my stomach's gonna do somersaults and flip. We all know this is a recovery building. Don't put us at risk!"

As illustrated in the above quote, residents were not only worried about the drug-related actions of fellow tenants, they were equally worried about their associates—"the drug crowd"—disrupting the equanimity and equilibrium of the building. In fact, this was a prominent theme throughout the focus groups.

Many residents expressed a strong fear of unknown outsiders entering the building, even people who may ostensibly appear unthreatening. For example, some residents stated that they were wary of anyone who claimed to be on "official business," such as locksmiths and couriers. One resident said, "I have a problem with uniformed people, like UPS and FedEx, somehow getting inside the building…. I like the secure building." Even agency staff were singled out in this regard. One resident said, "The staff come unannounced. I don't think they should do that, because if someone knock on my door, I ain't opening. It's my door!" In fact, residents frequently mentioned the importance of doors, locks, and windows, and the need to ensure that these were securely fastened to prevent entry by strangers. One resident told a new tenant that it was his duty to get to know other residents in order to discriminate tenants from strangers, saying to him, "You don't know anyone here. I could have been a killer in the hallway!"

In fact, the focus groups later in the study period were dominated by this theme, with participants giving numerous admonitions to newer residents to beware of outsiders and to pay very close attention to ensuring that the building was secure. "Some people will open the door and let anyone in. That's unhealthy. You are putting us at risk! And yourself at risk! When they knock at the door and you don't know them, don't open it!"

The results indicate that the threat to safety and security was deemed to emanate from three primary sources: first, from other residents in the process of relapse; second, from their associates also involved in substance abuse; and third, from unknown outsiders. A final threat to safety and security was the threat from oneself. Many residents were acutely aware that they were at risk of substance abuse relapse, especially when others were bringing drugs and alcohol back into the building or coming back smelling of drugs and alcohol. Residents tended to agree with the tenant who declared, "We all make mistakes, but we are there to help one another." This spirit of mutual support was often invoked as a helpful factor in continued abstinence.

However, tenants seemed aware of the precarious nature of their situation. Indeed the comment "I could have been a killer!" is telling in this regard. It might reflect not only a theoretical abstraction about strangers in the hallway but also a strong fear of what one might be capable of when engaging in substance abuse. As noted, this risk to self appeared to be amplified when other residents were abusing substances. For example, one resident said, "I know that people in that building care about me, but I still have a problem reaching out to me [sic], because drugs have beat me up. I want to show these drugs that I can work against them, but it's not working right now." Another resident reported, "You know I am scared. I am scared to shit. I get real nervous when people are using, because crack cocaine was my drug of choice."

Residents identified safety and security as the key quotidian factor influencing their experience in the recovery housing. Safety trumped such factors as the inclusion, collaboration, and empowerment principles that were identified at the outset by staff and consumers as potentially important in the progress of the recovery-oriented setting. The theme of safety and security was also much more dominant than other potentially important variables suggested by previous research, such as intensity of case management or loneliness (8,9). Nevertheless, our finding accords somewhat with results in the wider literature stretching back to Maslow (21). It suggests that the issue of safety and security is prominent in maintaining ontological security within individuals.

Recent research indicates that urban residents with mental health problems are strongly fearful of crime, especially when danger is perceived to be located in the three-dimensional space bordering their own dwelling (22,23). This fear often derives from regular experience of violence; some research indicates that more than 25% of people with severe mental illness have been victims of a violent crime in the past year, a rate more than 11 times higher than in the general population (24). Luhrmann (6) found pervasive violence to be a feature of life among women with severe mental illness in North Chicago. In addition, Desjarlais (25) noted that violence is common in sheltered, transitional housing.

The results raise the question of whether participants' concerns are based mostly on an "objective" appraisal of surrounding danger or on "subjective" paranoid-like anxiety colored by previous experience and fear of relapse. During the course of the project, there were no reports of burglary in the building or of violence directed by outsiders toward residents. In fact, the building is located in a part of the city that has traditionally been considered one of the safer quarters. However, about a third of apartments were vacated during the two years of the project, mostly when the landlord evicted tenants for serious relapses to substance abuse. This outcome suggests some level of reality in residents' claims that other residents and their associates posed threats to safety and security. Nevertheless, this fear and associated hypervigilance may have been amplified by previous experience.

In addition to homelessness, mental illness, and substance use disorders, many residents also had a history of physical and sexual abuse, of trading sex for drugs, and of prostitution. One interpretation of the concerns expressed in regard to "violation" of the building's entry and exit systems and strangers "entering" the building is that these are semiotic representations of fear of rape, physical abuse, and sexual exploitation. Concern with unwanted and aggressive violation of the boundaries of the buildings may be a metonym for ongoing fears about violation of the boundaries of the individual body.

The results also indicate an ongoing tension among residents between achieving meaningful connections to fellow residents (desire for community) and wanting more privacy and independence (desire for autonomy). It has been argued that this struggle is pervasive among individuals in postmodern societies (26). However, this issue may be especially pertinent for people with severe mental illness. Their housing options commonly fall into two categories—congregate housing, where there is community at the expense of autonomy, and supported housing, where there is autonomy at the expense of community. Indeed, one aim of the recovery housing described here was to provide a housing option in which autonomy and community would be appropriately balanced. However, even in this setting residents had specific difficulties navigating an appropriate middle road between community and autonomy. This could be seen in ambivalence about staff activities and peer support in the building. The development of mutual trust was hampered by regular relapses among a minority of residents, which led to suspicion, hostility, recriminations, and watchful withdrawal among other residents. Again, these feelings may be magnified among people with severe mental illness, who may struggle with issues of basic trust because of their previous experience of victimization and exploitation.

The results imply some simple interventions that may meliorate a sense of safety and security in small-scale recovery housing as well as in analogous congregate housing. Ensuring physical integrity in the building, with secure doors and locks, may be a simple but effective first step. Other measures showing that the agency takes safety and security seriously—for example, by providing 24-hour on-call assistance—may reassure residents. Intentional strategies that increase social interactions and esprit de corps among residents may help them attain a desired balance between autonomy and community. Appropriate internal "policing" and correctional mechanisms should also be agreed upon.

This case study of a single recovery community suggests that safety and security is a dominant issue for people with severe mental illness living in a downtown metropolitan area. Threats in this regard are deemed to emanate from fellow residents who are experiencing relapse to substance abuse and from their associates—unknown outsiders—and from actions of the self. Future research might directly compare the safety and security of consumers living in recovery buildings and those in supported housing and traditional congregate settings.

The authors report no competing interests.