Identifying Social Skills That Support Housing Attainment and Retention Among Homeless Persons With Serious Mental Illness

The set of social skills was developed by using a literature review, key informant interviews, a national consensus panel of experts in psychosocial rehabilitation and homelessness, and focus groups with homeless-experienced persons with serious mental illness. Data were collected from February 2017 to January 2018. The U.S. Department of Veterans Affairs (VA) Greater Los Angeles’ Institutional Review Board approved all procedures. Because key informants and expert panelists were not considered human subjects, informed consent was not obtained from these groups. Informed consent was obtained from focus group participants.

The literature review included published journal articles, clinical manuals, and unpublished documents on social skills training for persons with serious mental illness. The review began with meta-analyses and systematic reviews published in English from 1990 to 2015, identified using the following keywords in PsycINFO and MEDLINE: social skills training, skills training, serious mental illness, severe mental illness, and schizophrenia. Six articles were identified. Next, one author (S.G.) spoke with authors of these articles and clinical leaders in this area. These individuals were asked for additional salient literature on this subject (regardless of publication status) and for the names of other knowledgeable persons in this area. After several iterations, this process led to the review of articles cited in the introduction of this article and other key references (29–34). Twelve articles, three clinical manuals, and three unpublished documents were included.

To complement this review with the clinical experiences and practice evidence of national leaders in this area, key informant interviews were conducted (in person or by phone) with 12 experts, including clinicians, implementation leaders, researchers, and academicians. Respondents were selected for their national or regional reputation as leaders in this field; most had authored articles from the literature review and others were recommended as content experts by these authors. Efforts were made to ensure that respondents represented a breadth of stakeholders in mental health treatment. Informants were asked to discuss social skills that they believed were most important to improve community functioning among persons with serious mental illness; they were also asked to identify and describe social skills that were most pertinent to homeless-experienced persons with serious mental illness during the transition into independent apartments. Interviews also discussed social skills training content of limited or no relevance to the population of interest.

The interviewer took detailed notes during each interview; two authors (S.G. and E.K.) used these notes to systematically create templated summaries of each interview, matching domains in the interview guide (35). Matrix analysis methods were used to synthesize key themes by domain and develop a detailed list of relevant social skills (36). This list was reviewed to identify categories that were conceptually consistent with the literature review and with established social skills training interventions that are effective for persons with serious mental illness (12, 13). Two authors (S.G. and A.Y.) reviewed these skills, identified redundancies and similar themes, and reworded skills as needed.

A national panel of 11 individuals was subsequently identified, including prominent clinicians, academic experts, and administrators with expertise in homelessness or psychosocial rehabilitation for persons with serious mental illness. Reputation-based snowball sampling—beginning with a list of experts identified by the authors (37)—was used to recruit a panel with demonstrated experience in the rehabilitation of homeless-experienced persons or persons with serious mental illness. Four panelists were clinical and research psychologists focused on psychosocial rehabilitation, five were primary care physician-investigators with expertise in homelessness, and two were licensed clinical social workers with doctoral degrees who were experienced in homelessness and serious mental illness. The approach used was modified from the RAND/University of California, Los Angeles, appropriateness method (38), which has been employed in numerous studies (39–41) to reach consensus on mental health practices.

Each panelist was e-mailed a document presenting an overview of the proposed intervention and its goals, an executive summary of findings from the literature review and key informant interviews, and salient literature on social skills training for persons with serious mental illness. Panelists were also e-mailed a link to a Web-based preconference survey that included a list of 39 social skills and two questions regarding each skills. Skills were identified from core themes in the literature review and key informant interviews; themes were matched with skills from the literature review’s clinical manuals (12, 13, 42). Some skills were adapted to reflect interview narratives. Asking panelists to think of homeless-experienced consumers with serious mental illness engaged in scattered-site PSH, survey questions assessed feasibility and impact, respectively: “How likely is it that a consumer could apply this skill in his/her everyday life?” and “How likely is this skill to substantially improve consumers’ rates of housing attainment or retention?” These domains have precedent in mental health consensus panels (39, 41). Responses to both questions were made on a 9-point Likert scale (1, extremely likely, to 9, extremely unlikely. Panelists were also given the opportunity to add social skills (in free response fields), rating them 1 to 3 in feasibility and impact.

Means, standard deviation, and the number of responses with ratings of >3 (responses of 3 anchored as moderately feasible or impactful) were calculated for each skill (in total and by feasibility and impact separately). These summary statistics were fact-checked and reviewed with several key informants who were not panelists; eight skills with rating dispersion (mean score of <3 but >25% of responses of >3) were identified for further discussion.

The national panel was virtually convened (by telephone or Web-based modality) in August 2017 to reach consensus on the final set of social skills. The goal of the conference was to refine the set of social skills while increasing consensus for the final product. Preconference ratings were summarized, and the research evidence supporting the aforementioned eight skills was reviewed. Panelists could reword, split, combine, or discard skills. They were encouraged to negotiate areas of disagreement to increase consensus on the final skills. After discussion, panelists confidentially rated each skill again using the questions from the preconference survey.

Last, in two focus groups (about 1 hour each) moderated by two authors (E.K. and S.G.), the set of social skills identified by the panel was discussed with persons with serious mental illness receiving PSH at VA Greater Los Angeles (N=17). Maximum variation purposive sampling (43, 44) was used to select heterogeneous participants across age and gender, two variables associated with differential risk for experiencing homelessness (45–47). Five were women and 12 were men; their mean±SD age was 46.4±14.1 years. Most (N=10) were African American; others were white (N=5), Hispanic (N=1), or Asian (N=1).

The focus groups reviewed the entire set of social skills derived from the expert panel. Participants were given an opportunity to reword, add, or remove skills. They brainstormed behavioral instruction (e.g., pertinent role-plays) for each skill. Both moderators took detailed field notes during the groups; these notes were combined. Parallel to analyses of the key informant interviews, two authors (E.K. and S.G.) used the combined notes to create templated summaries of each focus group that aligned with domains discussed in the focus group (35). Matrix analysis methods (36) identified key themes by domain.

The literature review and key informant interviews were used to identify 39 skills within seven skill domains: finding and renting an apartment, using one’s time well, getting closer to people, managing finances, avoiding problems with drugs and alcohol, solving interpersonal problems, and managing one’s health. This list of skills was presented to the national panel for preconference ratings.

Skills (N=19) with a mean preconference rating of >3 in feasibility or impact were reviewed with selected nonpanelist key informants. Noting the preconference ratings, these skills were discussed iteratively and reworded if needed; four were kept, seven were removed, and eight were marked for panel discussion. Other skills (with mean preconference ratings of ≤3 in feasibility and impact) were also reviewed with these nonpanelist key informants; two additional skills were removed per these discussions. The virtual conference increased consensus on the eight discussed skills. Rewording was suggested for several skills. After each skill was discussed, panelists rated the skill’s feasibility and impact again; two were kept (mean feasibility and impact ≤3) and the other six were removed.

The refined set of 24 skills was shared in the focus groups with homeless-experienced persons with serious mental illness (N=17). These groups enhanced the language of several skills to increase their consumer-centeredness. Although participants could add or remove skills, they opted not to do so; rather, the groups focused on techniques to effectively train the population of interest in these skills (e.g., role-play scenarios relevant to their everyday lives).

Table 1 presents the final set of social skills and their rating by the national panel; when applicable, this table reflects the rewording recommended by the focus groups. The skill rated as most feasible was developing strategies for remembering appointments and medications. The skills rated as having the strongest impact were interviewing for an apartment with a landlord or property manager, finding productive things to do, identifying personal financial patterns, and developing a budget. The mean of all skills was 2.6 for feasibility and 2.4 for impact.

Despite PSH services, homeless-experienced persons with serious mental illness often struggle to acquire and retain housing (6–8). Although improved social skills can help these individuals exit homelessness (10), there is a dearth of knowledge surrounding specific social skills that affect this population’s housing outcomes. Using literature review, key informant interviews, a national consensus panel, and consumer focus groups, this project identified 24 social skills that enable housing attainment and retention among persons with serious mental illness engaged in scattered-site PSH. Feasibility of training and impact on housing outcomes were concurrent foci of this project. A highly feasible skill with low impact is unlikely to bolster housing outcomes in this vulnerable population. Conversely, persons with serious mental illness are unlikely to see improvement in skills with low feasibility of training, regardless of potential impact.

Prior to this project, there was no known research or consensus development that could inform social skills training interventions for homeless-experienced persons with serious mental illness. The diverse stakeholders who informed this work had considerable experience in psychosocial rehabilitation for persons with serious mental illness or a history of homelessness. There was strong consensus that it was feasible to train the population of interest in an identified set of social skills and that these skills can strongly affect housing attainment and retention among homeless-experienced persons with serious mental illness. In addition, consumers showed strong support for this skill set, changing nomenclature as opposed to content and brainstorming ways to effectively train these skills using behavioral instruction that parallels challenges in their daily lives.

Beyond the content modifications undertaken in this project, there are a number of important issues regarding the implementation of social skills training interventions for persons with serious mental illness engaged in scattered-site PSH. Stirman’s (28) classification system refers to these issues as context modifications (relevant to an intervention’s format, setting, personnel, or population). To guide context modifications, qualitative data collection with PSH staff and participants is a critical next step in this research. For example, homeless persons often have fundamental needs—for food, clothing, or shelter—that compete with their adherence to services (48). As such, incentivizing participation in this intervention and offering flexibility (e.g., open groups allowing walk-in participation) in the training’s structure may be important for this population. Moreover, to improve functioning, social skills training interventions are anchored in role-plays and other strategies that facilitate skill practice in real-world settings (9); the context and setting of PSH settings are distinct from those of psychosocial rehabilitation programs (e.g., day treatment) in which social skills training is frequently delivered. Iterative testing and tailoring of a housing-focused social skills training program will be needed to adapt to the needs of homeless-experienced persons with serious mental illness engaged in scattered-site PSH. In addition, populations with significant comorbidities (e.g., recalcitrant substance abuse) may require disparate sets of social skills and/or treatment delivery modalities.

This set of social skills has limitations. First, the derived set of social skills is not completely inclusive; dependent on a consumer’s strengths and deficits, additional skills may be important for housing attainment and retention. Second, the set of social skills reflects the experiences and perspectives of the specific stakeholders engaged in this project; involvement of other experts or consumers might have resulted in a different skill set. Third, the set of skills will need to be revised as gain a better understanding of specific social skills that predict housing outcomes. Overall, this set of social skills should be viewed with caution until future studies determine whether improving these skills enhances housing attainment and retention among homeless-experienced persons with serious mental illness.

This project established consensus on a set of social skills that facilitate housing attainment and retention for homeless-experienced persons with serious mental illness. Though PSH effectively addresses homelessness, many persons with serious mental illness struggle to obtain and sustain housing in these programs. Little is known about social skills that predict these individual’s housing outcomes. To date, although PSH case managers often perform informal social coaching, there is no literature on the systematic development, implementation, or testing of social skills training interventions for persons with serious mental illness in PSH programs. Moreover, although PSH programs often boast multidisciplinary teams with a breadth of expertise, real-world implementation of PSH often suffers from poor fidelity to supportive services, instead prioritizing speed of housing attainment and low barriers to program entry (49). As PSH programs bolster their supportive services to improve care for the most vulnerable, content modification of social skills training is critically needed. This set of social skills is a first step toward content modifications. Future directions include research on the effectiveness and implementation of a social skills training intervention that is modified to this set of skills. Iterative adaptation and context modifications of this intervention are anticipated to establish an effective intervention with strong potential for systematic uptake in the PSH context.

The authors acknowledge Brian Mittman and JoAnn Kirchner for their invaluable contributions to this work.