Associations between housing and health can already be found in empirical literature (3, 7, 8). However, the literature indicates that the causal relationships going beyond the direct effects of structural deficiencies of the dwelling are poorly understood (9–11).

According to the WHO (3), healthy housing should provide “a feeling of home, including a sense of belonging, security and privacy.” (3), p. 2. A recent systematic review found significant associations with physical and mental health for a range of context-related housing attributes among ASR, most of which were material/physical in nature (e.g., mold and damp). This relation was intertwined with other attributes, such as discrimination, which is why the authors point to the need for better research tools to examine this more generally (12).

The way these attributes are linked is often referred to as psychosocial pathways, which are considered important mediators between housing and health (9–11, 13). We consider these attributes, which reflect on the social meanings associated with housing and create a sense of home, as psychosocial attributes in our research. Figure 1 provides an overview. They serve as the starting point for our study. A challenge is to better define the concept of “home” and to make its relevant attributes measurable (14).

To identify the psychosocial attributes of housing, we refer to a framework by Dunn (15). Dunn collates such attributes in the “meaningful dimension of housing,” one of three key housing dimensions through which social inequalities and health consequences are produced. The dimension is conceptualized around three psychosocial attributes derived from the Job-Demand Control Model (16): personal control, demand, and social support. Considering this, Dunn (15) assumes that these attributes that are most important for health at work will also apply to the place where people spend most of their time: the home. Dunn adds a specific attribute: satisfaction with housing. These psychosocial attributes are each defined by a set of indicators (15). Control over the dwelling includes indicators of feelings of comfort and security (space, privacy, safe neighborhood), while demand comprises the workload associated with the dwelling (gardening, housework, costs). Social meanings cover indicators such as the level of pride and self-reflection and the sense of belonging to the neighborhood; in this regard, Dunn speaks of “expression of status” (15), p.35 (Supplementary File S1).

Dunn (15) did not focus on particular population subgroups in this work. We argue that these attributes derived from the field of workplace organization also apply to the living situation of ASR. They often have little control over their housing situation due to asylum laws, for example, in Germany, where they are obliged to live in state-mandated initial reception facilities for a given time period (Asylum Act, Article 47, Section 1). As this population group already faces several postmigration stressors, we consider it useful to examine the demand associated with the housing environment. We further hypothesize that a lack of social support may negatively affect the health of ASR as they are exposed to social prejudice and loneliness, which in turn affects their integration (17).

We address the following research questions: 1) To what extent are psychosocial attributes of housing taken into account in empirical research assessing the relationship between housing and the health of ASR? and 2) What is the relationship between psychosocial attributes of housing and the health of ASR?

We applied the PECO (Population - Exposure - Comparator - Outcome) framework (19). The population of interest included adult ASR who resettled to a high-income country (HIC). We set no limitation to language, gender, ethnicity, or time since migration. The exposure was related to the housing context and assessed in two stages. First, we considered any material/physical attributes of housing assessed in the studies and their link to health to identify categories (Supplementary File S4). We understood material/physical attributes as attributes that we identify in the literature on the general relationship between housing and health, i.e., all attributes that have not been considered from a psychosocial perspective and their immediate consequences. Second, we assessed which of these attributes falls under psychosocial attributes according to Dunn (15). In order not to exclude potentially relevant studies due to a too narrow concept of health, we included studies assessing any health-relevant outcome referring to physical and mental health, morbidity, and mortality, corresponding to the WHO definition of health (20). We searched PubMed, Web of Science, PsycINFO and PsycINDEX, SocINDEX, CINAHL and the Cochrane Library. We built the search terms (Supplementary File S2) by drawing on the PECO framework. We derived the specific terms from previous housing and/or forced migration research and conducted a pilot search in PubMed.

Eligibility criteria are displayed in Table 1. The mixed-methods approach was considered as useful since psychosocial attributes (especially unexpected or previously unknown ones) may be more frequently identified in qualitative research due to the inductive nature of this research method. However, we did not want to overlook quantitative research that may have assessed these attributes, therefore considering both study types eligible for the review. We considered the year 1995 as a time point by which the consequences of the wars that ended in Bosnia and Croatia were increasingly thematized, as preliminary analyses in PubMed showed. Since attributes of housing are measured in different ways in empirical research, we kept eligibility criteria broad and included all studies that may point to any relationship between housing and health in the abstract or title by comprising the term “housing” or synonyms (Supplementary File S2) and “health” or related terms (Supplementary File S2). We excluded studies that focused on homelessness only or defined the housing context only broadly, which would not provide insights into which housing-related attributes are potentially related to health. We did not include studies conducted on internally displaced persons (IDPs) or the general population of (im-)migrants unless the study provided disaggregated data for ASR, as we explicitly wanted to capture experiences with the housing situation of ASR in HICs.

Two reviewers (TB, VD) independently screened titles and abstracts of all matches for eligibility. Next, we retrieved texts of potentially eligible studies and again assessed them independently. We discussed inconsistencies during these processes; if necessary, with the help of a third reviewer from the study team.

For data extraction, we used a predesigned form (Supplementary File S3). We piloted it by having both reviewers extract five (approx. 16%) of the studies. After discussing inconsistencies, both reviewers extracted half of the remaining studies.

To identify psychosocial attributes of housing as objectively as possible, we used indicators derived from Dunn (15) (Supplementary File S1). We identified these in three ways: i) direct quantitative elicitation, ii) direct description by study participants or iii) indirect inference from the results. This was the case when study participants referred to attributes without explicitly naming them or when quantitative research measured variables which were similar to the indicators of attributes.

We used the Mixed-Methods Appraisal Tool (MMAT) for the assessment of the study quality (21, 22). The MMAT provides different sets of criteria for different study designs: qualitative, quantitative, and mixed methods. Two reviewers (TB, VD) independently applied the MMAT on half of the studies after a pilot phase in which both assessed five studies (approx. 16%) and discussed inconsistencies. To ensure high internal validity, low-quality studies were excluded.

We conducted narrative synthesis since we expected a high heterogeneity of results and a diverse range of methods applied. We followed the Guidance on the Conduct of Narrative Synthesis for Systematic Reviews (23). Accordingly, we conducted a thematic analysis (23). As housing was assessed as a two-level exposure, we first looked for the most important material/physical attributes and listed them in a table. We assigned thematically similar attributes to superordinate housing categories and listed the associated health effects (Supplementary File S4). Second, we listed the psychosocial attributes. We categorized the individual aspects by color (Supplementary File S4) to allow for better classification and overview. Using vote-counting, we investigated the frequency and, thus, the relevance of the attributes. To explore relationships within and between studies, we drew a conceptual map depicting relationships between housing-related attributes and health, also including findings of psychosocial attributes (Supplementary File S5). To assess the robustness of the synthesis, we critically reflected on the process: We discussed the limitations of the method, quality and generalizability, and discrepancies identified in the findings and how we handled them (23).

The largest proportion of studies was conducted in the UK (n = 8) (24–31), followed by Germany (n = 6) (32–37) and Australia (n = 5) (38–42). One study each was conducted in Canada (43), the Netherlands (44), Austria (45), Ireland (46), Switzerland (47), Norway (48), and Sweden (49). Most studies (n = 25) included adult ASR (24–34, 36–38, 40–43, 46, 47, 49–53), and were published between 2015 and 2022 (n = 24) (24, 26, 27, 30, 32–38, 40–43, 45–51, 53, 54). The number of participants in qualitative studies ranged from 6 to 106, and in quantitative studies, from 105 to 5,678 persons. Nineteen studies focused on the outcome of mental health (25, 26, 28, 30, 31, 36–38, 41, 43–46, 48–53). We found housing as a main topic in seven studies (25, 31, 33, 40, 42, 44, 50); in the remaining 25 it was a side topic. The most common countries of origin of the study participants were Syria, Afghanistan, Iraq, Iran, Somalia, and Pakistan.

After quality assessment (Supplementary File S7), we classified 27 studies (84.4%) (25–38, 40–47, 49–52, 55) as high and five (15.6%) (24, 39, 48, 53, 54) as moderate regarding their quality. In the subset of quantitative-descriptive studies, the nonresponse rate, reasons and statistical compensation for nonresponse were not stated in ten (25, 26, 29, 32, 38, 39, 43, 44, 53, 54) of 16 studies. The target population and sample were not adequately described in two studies (39, 53). In one study (54), we awarded zero points for the sampling strategy. Three (4, 53, 54) out of the 16 quantitative-descriptive studies were of moderate quality, the others of high quality (25, 26, 32, 33, 35–38, 43, 44, 49, 50, 55). One quantitative non-randomized study (24) was rated as of moderate quality because it did not provide appropriate outcome measures and data. The evaluation of the qualitative studies showed different limitations. The criterion most frequently violated was the link to the research question, which was missing (28, 42, 46, 48). On the other hand, all qualitative studies were of high quality, except one of moderate quality (48).

We identified seven categories of material/physical attributes and their immediate consequences: 1) crowding, space, and privacy, 2) housing instability (including security of tenancy, affordability, and residential mobility), 3) safety, 4) physical conditions of/in the dwelling, 5) housing satisfaction, 6) neighborhood and location, and 7) institutional practices in the dwelling or accommodation center (Supplementary File S4).

We found crowding, space, and privacy in 17 (53.1%) (24, 27, 29, 30, 32–35, 37, 40, 42, 44, 45, 47, 51–53) and housing instability in 16 (50%) (25, 28–31, 34, 38, 39, 41, 42, 45, 49–53) of the 32 studies. We further identified the category physical conditions in nine (28.1%) (27, 29, 34, 40, 42, 45, 47, 54, 55) and safety in eight studies (25.0%) (27, 29, 30, 33, 34, 40, 42, 51). We identified fewer housing-related attributes in quantitative studies. Only in terms of housing satisfaction quantitative approaches dominated (23.5% quantitative vs. 6.7% qualitative; number referred to: 17 quantitative and 15 qualitative studies in total).

Psychosocial attributes were directly assessed quantitatively in 12.5% (n = 4) (26, 29, 36, 43) of the 32 studies, reported by the study participants in eight studies (25%) (27, 30, 31, 34, 40, 42, 46, 52), and indirectly derived from the results by the reviewers in ten studies (31,3%) (24, 29, 30, 33, 41, 45, 47, 48, 50, 51). Psychosocial attributes were present in 20 (62.5%) (24, 26, 27, 29–31, 33, 34, 36, 40–43, 45–48, 50–52) of the 32 studies.

Looking at the thematic analysis (Supplementary File S4), we found some overlap between the material/physical attributes and the psychosocial attributes according to Dunn (15).

We classified the material/physical attributes lack of privacy and space as attributes of the category crowding, space, and privacy under the psychosocial attribute control.

The material/physical housing-related category housing instability included attributes such as residential mobility and housing costs/affordability. As participants perceived these as burdens, we assigned affordability to the psychosocial attribute demand, specifically to its indicator strain of meeting costs, and residential mobility to the psychosocial indicators worry of frequent move and worry of forced move as part of the psychosocial attribute control.

The category safety—within the shelter due to (aggressive) strangers and in the neighborhood—was often linked to a fear of crime and victimization. Thus, we assigned it to the corresponding psychosocial indicator fear of crime/victimization.

Regarding neighborhood, good social relations were also associated with a sense of belonging as an indicator of the psychosocial attribute expressing status. In one study (40), participants perceived access to green spaces, as an aspect of location, as positive for health. We associated this, in turn, with the feeling of wellbeing at home and thus assigned it to the psychosocial indicator place of refuge.

The fourth housing-related category, physical conditions of/in the dwelling, included attributes such as gardens and outside spaces which participants saw as a burden. We thus subsumed it under the indicator housework strain, as an indicator of the psychosocial attribute demand.

In addition, satisfaction emerged as a material/physical attribute, which is also defined as a psychosocial attribute according to Dunn (15).

The seventh category was institutional practices in the dwelling: when it came to low control regarding institutional practices, participants did not feel as comfortable as they would in a “real” home, which we, in turn, attributed to the psychosocial control indicator place of refuge. Moreover, participants felt ashamed of these practices and thus their home did not represent who they really were (46). This was consistent with the definition of the psychosocial attribute expressing status, specifically its indicators pride and self-reflection.

The attributes we identified most frequently related to control. Of its four indicators, the indicator place of refuge was most prevalent (n = 13, 40.6%) (24, 27, 30, 31, 33, 34, 40, 42, 46–48, 51, 52), followed by fear of crime and/or victimization (n = 7, 21.9%) (27, 33, 34, 40, 42, 50, 51), and the indicators concern about forced (30, 31, 40, 42) and frequent moves (31, 41, 45, 51) (n = 4, 12.5%). Expressing status was the second most common. While we found its indicators pride (46) and self-reflection (46) once each (3.1%) in the 32 studies, we identified the sense of belonging to the neighborhood six times (18.8%) (29, 33, 40, 42, 48, 50). This was followed by the attribute demand. Its indicator strain of meeting costs was addressed in 12.5% (n = 5) (40, 42, 51, 52) and the indicator housework strain in 3.1% (n = 1) (33) of the studies. Finally, we identified the indicator satisfaction in five studies (15.6%) (26, 29, 33, 36, 43). We identified all indicators except satisfaction more frequently in qualitative than in quantitative studies.

Starting with limitations, we did not conduct a “grey” literature search, screen the references for more potential studies and include other than peer-reviewed articles. This may have incurred publication bias (67) as non-significant results may not have been published (68). If so, we would be overestimating the strength of association between psychosocial attributes and the health of ASR. Because one of the three methods we used to identify psychosocial attributes was to derive them indirectly from study results, we may have been subjective in our assessment which could affect the reproducibility of our approach. This could also have been the case with classifying the material/physical attributes into the seven categories. Contrary to recommendations (21, 69), due to time constraints as one reviewer left the working group, the two reviewers only extracted and assessed five studies independently. The remaining studies were split up, extracted and assessed by one person each. This approach potentially increased the risk of error and the possibility that a single person’s bias influenced data selection. We tried to minimize this problem by frequently communicating about our procedures and discussing them intensively.

We included studies featuring a variety of methods and with heterogeneous results. So, we could only conduct narrative synthesis, including thematic analysis, instead of meta-analysis, which may have reduced transparency and methodological clarity (70, 71). The heterogeneity of studies in terms of design, sample size, methodical approach, and research focus limits their comparability, interpretation, and generalizability.

However, to our knowledge, our review is the first to consider the psychosocial perspective as a pathway between housing and the health of ASR in HICs. By conducting a systematic review, we minimized the risk of ad hoc search and bias (72). Adherence to the PRISMA checklist improved the reporting process and ensured a systematic, transparent approach (18). The broad range of databases searched can also be seen as a strength. By not specifying a particular housing concept, we avoided bias that might have resulted from the rejection of potentially relevant studies. For the same reason, we included qualitative studies since we assumed that psychosocial attributes of housing might be likely to emerge from perceptions and experiences. To counteract selection bias, we did not restrict the search in terms of language or full-text availability. Additionally, our study focused only on resettlement in HICs which might reduce the contextual heterogeneity of studies. Finally, a clear definition of psychosocial attributes according to Dunn (15) contributed to a systematic search and strengthens our findings’ reproducibility.

Our study provides evidence of the influence of psychosocial attributes of housing on ASR health. To conclude, housing in the context of health needs to be considered beyond its purely material/physical structures. Such structures are closely associated with psychosocial attributes which make homes meaningful, in particular, the perceived degree of control over the home.

On this basis, we recommend that housing for ASR offers sufficient space and privacy, as well as opportunities to develop. Affordable housing is an important mental health issue and should hence be given high priority. Mass shelters, in which ASR often spend long periods of time (73), are associated with perceived insecurity and a lack of opportunities for privacy, negatively affecting mental health and wellbeing. Therefore, as a short-term response, opportunities for more privacy must be established until private accommodation becomes available, and social support services need to be strengthened. Furthermore, ASR are often placed in deprived areas, which may be related to increased crime rates in the neighborhood and thus increase the fear of being victimized and having to move. The sense of belonging to the neighborhood is often weak, so its role as an important mediator between housing and health that can promote integration is lost. Accordingly, attention must be paid to neighborhood conditions when allocating housing to ensure healthy, anxiety-free living, and successful integration. As a long-term solution to addressing the broader structural inequalities that constrain the lives of ASR, ASR-led organizations are calling for the abolition/dismantling of mass shelters (74).

Further research should examine the associations between psychosocial attributes and the health of ASR more closely, especially regarding physical health outcomes. The interaction of the attributes with one another should be investigated. In addition, quantitative studies need to cover psychosocial attributes more comprehensively as qualitative findings demonstrate their relevance. However, due to the strong interconnectedness with the material/physical attributes of housing, psychosocial attributes should not be investigated detached from material/physical attributes. Since Dunn’s assumptions are based on the reality of Western societies, particularly Canada, further research is needed specifically on our population of interest, which is diverse and comes from different socio-cultural contexts.