Homelessness, defined as a situation in which an individual is without stable, safe, permanent and appropriate housing, or the immediate prospect, means or ability to acquire it(Reference Gaetz, Barr and Friesen1), is a concerning social phenomenon that affects an estimated 100 million people worldwide(2). Globally, there are also an estimated 1·6 billion people who lack adequate housing(3). The identified pathways into and out of homelessness are complex, but often include interrelated individual-level factors such as mental health conditions, substance abuse disorders or family violence, as well as system-level failures such as discharge from social or health services or institutions into homelessness(Reference Gaetz, Barr and Friesen1,Reference Hulchanski, Campsie and Chau4–Reference Fazel, Geddes and Kushel6) . Alongside the previous factors, structural factors such as poverty, discrimination and a lack of affordable housing contribute to the prevalence of homelessness(Reference Gaetz, Barr and Friesen1,Reference Hulchanski, Campsie and Chau4,Reference Gaetz, Donaldson and Richter5) .
Food security is defined by the Rome Declaration on World Food Security as an ideal state in which ‘all people, at all times, have physical, social and economic access to sufficient, safe, and nutritious food to meet their dietary needs and food preferences for an active and healthy life’(7). Globally, the number of food-insecure individuals is estimated to have reached 821 million, which is approximately one out of every nine people(8). Food insecurity (FS) is a complex phenomenon generally understood through an economic lens, as inadequate or insecure access to food due to financial constraints(Reference Tarasuk, Mitchell and Dachner9), poverty, urbanisation, environmental changes, the global trade market and agricultural policy(Reference Gibson10). FS has detrimental effects on individuals’ overall health and well-being. For example, FS results in increased negative physical health outcomes including chronic diseases, such as diabetes, pulmonary diseases, cardiac disease and obesity; a decrease in quality of life(Reference Fernandes, Rodrigues and Nunes11) and increased mortality rates(Reference Gundersen, Tarasuk and Cheng12). FS has also been found to be associated with poor mental health(Reference Davison, Marshall-Fabien and Tecson13–Reference Jones15) in the overall population. In particular, the risk of experiencing depression, anxiety disorders, mood disorders or suicidal thoughts increases with the severity of FS(Reference Jessiman-Perreault and McIntyre16).
Mental health conditions and FS are disproportionately prevalent in the homeless population(Reference Hulchanski, Campsie and Chau4,Reference Fazel, Geddes and Kushel6,Reference Fazel, Khosla and Doll17–Reference Weiser, Hatcher and Frongillo20) . In addition, the complex, interrelated factors that affect the pathways into and out of homelessness include mental health and food security(Reference Gaetz, Barr and Friesen1,Reference Hulchanski, Campsie and Chau4–Reference Fazel, Geddes and Kushel6) . Thus, it is important to understand the existing evidence on the nature and direction of the relationship between FS and mental health conditions. This information can also help inform research priorities and evidence-based social and public policies aimed at improving the mental health and food security of people experiencing homelessness.
We therefore conducted a scoping review to identify and synthesise the existing literature on the relationship between FS and mental health conditions among homeless adults. We particularly aimed to answer the following questions: (1) What is known about the relationships between FS and mental health conditions among homeless adults? (2) What are the nature and direction of these relationships? (3) What are the implications of the existing literature for future research?
Methods
We conducted this scoping review in accordance with the framework suggested by Colquhoun et al.(Reference Colquhoun, Levac and O’Brien21) and the recently published PRISMA extension checklist for Scoping Reviews (PRISMA-ScR)(Reference Tricco, Lillie and Zarin22).
Literature search strategy and study selection
The search strategy for the present review was developed in consultation with a research librarian and was designed to capture the most recent literature in an area that has only started to receive significant attention within the last decade. The search captured literature published between 1 January 2008 and 2 November 2018 (the date that the literature search was conducted) on the following databases: PubMed, Web of Science, Scopus, PsycINFO, Cochrane Library and CINAHL, using search terms associated with homeless adults, food security and FS, and mental health conditions (see online supplementary material, Appendix A and B).
We established the following inclusion criteria a priori: articles written in English and published in peer-reviewed journals that examined associations, relations or correlations between mental health-related conditions and FS among homeless adults (aged 18 years or more). We excluded articles where the population of interest was not homeless adults; where relationships between mental health and FS were not examined; and those published outside of the specified date range or in a language other than English.
The paper selection process is outlined in Fig. 1. The initial search yielded sixty-five articles after de-duplication. Then, the screening of titles and abstracts of retrieved records was conducted independently by at least two researchers. The results of the screenings were compared and discussed, after which thirteen articles were selected for full-text review. After full-text review (conducted independently by E.I.L., C.M.L. and J.L.), six articles were found to meet inclusion criteria. A manual review of the reference lists of these six articles yielded an additional three articles, for a total of nine papers included in the present scoping review (Fig. 1).
Overview of identification and screening records included in the present scoping review
The main characteristics and findings of the studies were abstracted independently by at least two researchers. Two researchers (C.M.L. and J.L.) conducted independent quality appraisal of the included studies, after which a consensus quality appraisal was reached. The AXIS tool(Reference Downes, Brennan and Williams23) was used for cross-sectional studies, and the Newcastle-Ottawa quality assessment scale(Reference Wells, Shea and O’Connell24) was used for longitudinal studies. The quality appraisal of the studies is presented in online supplementary material, Appendix C and D.
Results
Overview of included articles
The main characteristics of the nine studies included in this review are outlined in Table 1. Five articles were published from 2015 onwards, and the majority (n 8) had a cross-sectional design(Reference Hamelin and Hamel18,Reference Lee and Greif19,Reference Weiser, Bangsberg and Kegeles25–Reference Parpouchi, Moniruzzaman and Russolillo30) , while one had a longitudinal design(Reference Palar, Kushel and Frongillo31). As per The AXIS tool(Reference Downes, Brennan and Williams23), the majority of the cross-sectional studies were found to have a ‘good’ reporting quality and ‘good’ study design (see online supplementary material, Appendix C). However, the majority of the studies used secondary data, rather than collecting specific primary data, to respond to their studies’ objectives. Regarding the presence of potential bias, only four(Reference Hamelin and Hamel18,Reference Baggett, Singer and Rao26,Reference Martin-Fernandez, Lioret and Vuillermoz28,Reference Parpouchi, Moniruzzaman and Russolillo30) of the eight cross-sectional studies took steps to address the potential effect on the findings derived from the non-responders or missing data. Furthermore, the majority of the cross-sectional studies were based on self-reported data for the exposure or outcome variables, instead of using confirmed medical diagnosis criteria or administrative records. Moreover, some studies used single items rather than existing validated scales or criteria to assess food security(Reference Lee and Greif19,Reference Baggett, Singer and Rao26) or mental health problems(Reference Martin-Fernandez, Lioret and Vuillermoz28). Thus, the findings from the cross-sectional studies may be influenced by potential recalling and misreporting bias. The longitudinal study(Reference Palar, Kushel and Frongillo31) was appraised as fair quality using the Newcastle-Ottawa quality assessment scale(Reference Wells, Shea and O’Connell24), as it rated six stars over a maximum of nine stars available (see online supplementary material, Appendix D).
Extraction table of included studies
Most of the studies were conducted in the USA or Canada (n 8), and one study was carried out in Europe(Reference Martin-Fernandez, Lioret and Vuillermoz28). One study focused on homeless individuals aged 50 years or more(Reference Tong, Tieu and Lee29). One paper included homeless families, rather than individuals, as its study population(Reference Martin-Fernandez, Lioret and Vuillermoz28). Two of the papers analysed data from one study of individuals living with HIV/AIDS(Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) . All of the included studies considered other covariates in their analyses including demographics, indicators of socio-economic position (educational level and income), housing status, duration and frequency of episodes of homelessness, risk behaviours (alcohol and substances use), social support and physical health outcomes(Reference Hamelin and Hamel18,Reference Lee and Greif19,Reference Weiser, Bangsberg and Kegeles25–Reference Palar, Kushel and Frongillo31) . Only the longitudinal study(Reference Palar, Kushel and Frongillo31) specifically examined the relationship between food security and mental health conditions, while the remaining studies considered food security and mental health conditions as part of a larger set of predictor or outcome variables. Two of the studies were embedded within randomised clinical trials of Housing First, where food security was not the main outcome of the intervention evaluated(Reference O’Campo, Hwang and Gozdzik27,Reference Parpouchi, Moniruzzaman and Russolillo30) .
Identification and definition of homeless individuals
Definitions and identification of homeless individuals varied throughout the included studies. Four studies clearly defined their homeless populations as individuals with no permanent, adequate physical shelter and having little likelihood of obtaining it, or as living in places not intended as sleeping accommodation(Reference Lee and Greif19,Reference O’Campo, Hwang and Gozdzik27,Reference Tong, Tieu and Lee29,Reference Parpouchi, Moniruzzaman and Russolillo30) . The remaining studies used data from homelessness surveys like the 2003 Health Care for the Homeless User Survey(Reference Baggett, Singer and Rao26) or recruited homeless participants from a specific programme such as the Research on Access to Care in the Homeless Cohort, a cohort of HIV-infected homeless and marginally housed adults(Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) , or identified homeless individuals through their use of services for the homeless including clinical services(Reference Baggett, Singer and Rao26), temporary shelters(Reference Hamelin and Hamel18,Reference Weiser, Bangsberg and Kegeles25,Reference Martin-Fernandez, Lioret and Vuillermoz28,Reference Palar, Kushel and Frongillo31) , free-meal homeless programmes or services(Reference Hamelin and Hamel18,Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) , drop in centres(Reference Hamelin and Hamel18) and centres for asylum seekers(Reference Martin-Fernandez, Lioret and Vuillermoz28).
Food insecurity measures
FS was measured in a variety of ways. A modified version of the US Department of Agriculture’s Adult Food Security Survey Module was the most frequently used instrument, being used in four studies(Reference O’Campo, Hwang and Gozdzik27–Reference Parpouchi, Moniruzzaman and Russolillo30), followed by the modified version of the Household Food Insecurity Access Scale used in two studies(Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) . Two studies used a single item as a food security/food insufficiency indicator(Reference Hamelin and Hamel18,Reference Baggett, Singer and Rao26) , while in one study, authors constructed a ‘Hunger Scale’ composed of several items related to frequency, adequate amount and affordability of food, as well as on the prevalence of subsistence eating behaviours(Reference Lee and Greif19) (Table 1).
Mental health-related measures
Depression and depressive symptoms were the most common mental health conditions assessed, with associations examined in six articles(Reference Hamelin and Hamel18,Reference Weiser, Bangsberg and Kegeles25,Reference Martin-Fernandez, Lioret and Vuillermoz28–Reference Palar, Kushel and Frongillo31) . Depression and depressive symptoms were measured via the Mini International Neuropsychiatric Interview 6.0(Reference Parpouchi, Moniruzzaman and Russolillo30), 21-item Beck Depression Inventory version 2(Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) , the Center for Epidemiologic Studies Depression Scale(Reference Tong, Tieu and Lee29), the Diagnostic Interview Schedule(Reference Hamelin and Hamel18) and by a single-item question that asked whether a person felt sad, depressed or without hope in the 12 months prior to the interview(Reference Martin-Fernandez, Lioret and Vuillermoz28).
Other specific conditions assessed included alcohol and substance use problems(Reference Hamelin and Hamel18,Reference O’Campo, Hwang and Gozdzik27,Reference Parpouchi, Moniruzzaman and Russolillo30) . The Mini International Neuropsychiatric Interview 6.0 and Diagnostic Interview Schedule(Reference Hamelin and Hamel18,Reference Parpouchi, Moniruzzaman and Russolillo30) were used to assess drug or alcohol dependency, while the Global Appraisal of Individual Needs and the Colorado Symptoms Index were used to assess alcohol and substance use symptoms severity and mental health symptoms, respectively(Reference O’Campo, Hwang and Gozdzik27). Psychiatric hospitalisation was assessed using self-reported information on whether a participant has been hospitalised in the previous year to the interview(Reference Baggett, Singer and Rao26) (Table 1).
In some studies, authors used mental health-related composite measures. For example, Lee and Greif (2008) created an alcohol, drug or mental health index to capture difficulties that participants experienced during the year prior to the NSHAPC interview. Hamelin et al. (2009) grouped different mental conditions under diagnostic categories (axis I and axis II problems) based on the DSM-IV(Reference Hamelin and Hamel18). Parpouchi et al.(Reference Parpouchi, Moniruzzaman and Russolillo30) classified individuals as having two or more, and less severe v. more severe conditions(Reference Parpouchi, Moniruzzaman and Russolillo30). Furthermore, two studies assessed mental health-related functioning status using the twelve-item and thirty-six-item short forms of the health survey questionnaire (SF-12 and SF-36)(Reference Weiser, Bangsberg and Kegeles25,Reference Parpouchi, Moniruzzaman and Russolillo30) .
Overview of findings on food insecurity and mental health
Directionality of the associations
The study of the directionality of the associations between FS and mental health conditions varied within the reviewed literature. The longitudinal study considered food security as the exposure factor and mental health condition (depression symptoms and probability of depression) as the primary outcome(Reference Palar, Kushel and Frongillo31). Among the eight cross-sectional studies, six considered food security-related aspects as outcomes(Reference Lee and Greif19,Reference Weiser, Bangsberg and Kegeles25,Reference O’Campo, Hwang and Gozdzik27–Reference Parpouchi, Moniruzzaman and Russolillo30) , and two studies consider food security as the predictor factor for mental health-related problems(Reference Hamelin and Hamel18,Reference Baggett, Singer and Rao26) .
Main findings on the specific association between food security and mental health problems
Food security as exposure measure and mental health-related problems as outcome measure
The longitudinal study showed that severe FS was significantly associated with depressive symptom severity and probability of depression(Reference Palar, Kushel and Frongillo31). Two of the cross-sectional studies found that food insufficient individuals were more likely to have depression, emotional disorders(Reference Hamelin and Hamel18), axis 1 mental disorders and more frequent incidents of psychiatric hospitalisation(Reference Baggett, Singer and Rao26) (Table 1).
Food security as outcomes and mental health-related problems as main predictors
Based on the reviewed eight cross-sectional studies, most of the studies found that mental health problems were associated with food security, such as in the studies which found that homeless individuals with poor mental health scores(Reference Weiser, Bangsberg and Kegeles25,Reference Parpouchi, Moniruzzaman and Russolillo30) , more severe mental health problems(Reference Hamelin and Hamel18,Reference O’Campo, Hwang and Gozdzik27) and depressive symptoms(Reference Martin-Fernandez, Lioret and Vuillermoz28,Reference Tong, Tieu and Lee29) were more likely to be food insecure after considering other covariates in their models. However, one study found that there was no significant association between the severe cluster of mental health disorders(Reference Parpouchi, Moniruzzaman and Russolillo30) and food security.
The association of alcohol and drug problems with food security as an outcome remains inconclusive. Several studies found alcohol and drug symptom severity and dependence were not related to food security after adjustments, such as in the studies which found that drug and alcohol symptoms severity(Reference O’Campo, Hwang and Gozdzik27), and substance and alcohol dependence(Reference Parpouchi, Moniruzzaman and Russolillo30) were not significantly associated with FS. Another study found individuals with higher composite values of alcohol and drug problems were more likely to be food insecure when adjusted for other covariates(Reference Lee and Greif19). However, this study used a composite measure that includes alcohol, drug and mental health problems, which makes it impossible to differentiate the associations of FS with alcohol and drug problems v. the associations with mental health problems.
Discussion
In this scoping review, we analysed the relationship between food security and mental health conditions in homeless adults by reviewing peer-reviewed research over the past 10 years. We found limited research on this topic. The majority of studies used a cross-sectional design, where the association between food security and mental health conditions was not the primary goal of the analysis. Despite these limitations, the literature provides insight into the relationship between mental health and food security in homeless adults. More research in this area is needed, as food security, homelessness and poor mental health are conditions that have significant adverse health effects at the individual and population levels(Reference Jones15,Reference Vigo, Thornicroft and Atun32) .
Regarding the associations between food security and mental health-related conditions, the evidence from this review supports a relationship between FS and certain mental health conditions in the homeless adult population. This is particularly evident for depressive symptoms and the presence of depression, where food-insecure homeless individuals tend to have a greater probability of experiencing depressive symptoms or depression than food-secure individuals(Reference Hamelin and Hamel18,Reference Martin-Fernandez, Lioret and Vuillermoz28,Reference Palar, Kushel and Frongillo31) . In one study, homeless individuals with higher levels of depressive symptoms were more likely to be food insecure(Reference Tong, Tieu and Lee29). These findings are in line with what has been reported in studies carried out in the general population, where bidirectional associations between food security and depression have been observed(Reference Maynard, Andrade and Packull-McCormick33–Reference Abrahams, Lund and Field35). In interpreting these findings, it is essential to note that two of these studies(Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) examine patients with HIV/AIDS, who may have specific neurocognitive, immunity, comorbid and psychological profiles due to their HIV disease status. Thus, the associations of food security and mental health problems from these studies may not be generalisable to homeless adults without this condition.
We found that measures of mental health problems such as a composite measure of alcohol, drug and mental health problems(Reference Lee and Greif19), mental health symptom severity(Reference O’Campo, Hwang and Gozdzik27) and mental health status(Reference Weiser, Bangsberg and Kegeles25,Reference Parpouchi, Moniruzzaman and Russolillo30) were significant predictors of FS. In a recent large cross-sectional study on FS and mental health status performed in 149 countries, FS was related to poorer scores on several mental health indices(Reference Jones15). In our review, we also found that food insufficiency was a positive predictor of psychiatric hospitalisations(Reference Baggett, Singer and Rao26); this finding highlights the need for an improved understanding of the effects of food security on the mental health of the homeless population as well as health care systems(Reference Sullivan, Clark and Pallin36,Reference Rodriguez, Fortman and Chee37) .
There may be several potential pathways underlying the relationship between FS and mental health conditions in homeless people. Homelessness itself has negative impacts on physical and mental health, food security, nutrition and diet quality(Reference Lee and Greif19,Reference Parpouchi, Moniruzzaman and Russolillo30,Reference Whittle, Palar and Seligman38–Reference Bowen, Bowen and Barman-Adhikari40) . Homeless individuals live in extreme poverty and face resource-related barriers that prevent them from accessing nutritionally adequate foods(Reference Fazel, Geddes and Kushel6,Reference Hamelin and Hamel18) . Food programmes or food pantries that are available may offer limited food of inadequate nutritional quality and may be difficult to access for homeless individuals(Reference Tarasuk, Mitchell and Dachner9,Reference Li, Dachner and Tarasuk41,Reference Pettes, Dachner and Gaetz42) . Additionally, socioeconomically deprived individuals do not have access to important material sources and a stable living environment in which they can store and prepare foods, which contributes to low food security(Reference Hamelin and Hamel18,Reference Lee and Greif19,Reference Parpouchi, Moniruzzaman and Russolillo30,Reference Sprake, Russell and Barker43,Reference Kirkpatrick and Tarasuk44) .
Another potential pathway linking FS and some mental health conditions is biological in nature. Homeless individuals tend to consume a significant portion of their daily energetic intake in the form of highly processed foods(Reference Li, Dachner and Tarasuk41,Reference Sprake, Russell and Barker43,Reference Crawford, Yamazaki and Franke45–Reference Tarasuk, Dachner and Poland47) . As a result, this population group can experience poor nutrition and/or nutritional deficiencies more frequently than the general population(Reference Li, Dachner and Tarasuk41,Reference Sprake, Russell and Barker43,Reference Crawford, Yamazaki and Franke45,Reference Smith and Richards46) . Sustained nutritional deficiencies have adverse health impacts(Reference Palar, Kushel and Frongillo31,Reference Li, Dachner and Tarasuk41,Reference Crawford, Yamazaki and Franke45,Reference Smith and Richards46) , including the potential development of mental illnesses(Reference Palar, Kushel and Frongillo31).
Gaps in the existing literature and recommendations
We identified several gaps in the reviewed articles. First, the majority of studies were cross-sectional in design and thus cannot establish causality between FS and mental health conditions. Longitudinal and intervention studies are required to better understand the potential links between food security and mental health conditions in homeless adult and would serve to inform evidence-based interventions and support services for improving both mental health and food security in this population group. Second, there is a need to explore the relationship between FS and specific major mental health conditions such as psychotic disorder, major depressive disorder, post-traumatic stress disorder or mood disorders in homeless populations. Previous research on the general population has demonstrated that traumatic events(Reference Becker, Middlemass and Johnson48) or mood disorders are strongly associated with FS, and all these mental conditions have been identified as being disproportionately prevalent in studies on homeless populations(Reference Hwang, Stergiopoulos and O’Campo49).
Third, the existing literature focuses on the North American context, particularly Canada and the USA(Reference Gundersen50), and thus our understanding of this phenomenon in different socio-economic, cultural and political environments is limited. This highlights the need to conduct more research in this field in other geographical areas, such as low- and middle-income countries, where there may be a disproportionately high prevalence of homelessness(Reference Speak51) and FS(Reference Thome, Smith and Daugherty52), and where mental health problems may present in specific ways and rates(Reference Vigo, Thornicroft and Atun32,Reference Knipe, Williams and Hannam-Swain53) .
Fourth, few studies have focused on specific subgroups within the homeless population, with the exception of the studies included in this review that focus on homeless individuals living with HIV(Reference Weiser, Bangsberg and Kegeles25,Reference Palar, Kushel and Frongillo31) and homeless families(Reference Martin-Fernandez, Lioret and Vuillermoz28). Thus, further studies in overrepresented subgroups of homeless individuals such as LGBTQ+, Indigenous and other ethno-racial and cultural minorities, veterans and migrants are warranted to better understand the potential socio-cultural underlying pathways between food security and mental disorders in people experiencing homelessness.
Fifth, in the majority of the existing studies, the relationship between FS and mental health was not the primary focus of analysis and thus studies looking at these specific associations are needed. Primary research with longitudinal- or interventional-based study designs is particularly warranted.
Sixth, efforts are also needed to identify mechanisms underlying the relationship between FS and mental health among homeless individuals, which would require cohort and interventional studies.
Finally, the findings for the present scoping revision suggest a linkage between FS and mental health problems in adults experiencing homelessness. Hence, it is crucial to facilitate and support access to food sources and mental health care services for this socially excluded population.
The present scoping review has the following limitations. First, it only included peer-reviewed papers published in English. Second, this review only focused on the adult population, aged 18 years or older, and thus does not capture the experiences of homeless children or youth. Finally, due to the cross-sectional design and the limited contextual settings of the studies included in the present review, the findings may not be generalisable.
In conclusion, the findings of this scoping review suggest a potential association between FS and several mental health conditions among homeless adults. More longitudinal and interventional-based studies are needed to better understand the nature and directionality of the links between food security and mental health in this population. Validated tools to measure both food security and mental health problems should be used to improve the quality of evidence. Additional research in diverse populations and countries is needed. This study has implications for public policy, as it draws attention to the intertwinement of FS and mental health, which remain a significant issue in the lives of people experiencing homelessness. While more research is needed in this area, the existing evidence included in this review suggests that effective homelessness policy and interventions should include support and joint actions to reduce FS and address mental health issues of people experiencing homelessness.
Acknowledgements
Acknowledgements: None. Financial support: None. Conflict of interest: There are no conflicts of intrest. Authorship: C.M.L., E.I.L., J.L. and S.W.H. conceived the study. E.I.L. and C.M.L. performed the literature search. EIL performed the first screening of retrieved papers. C.M.L. and J.L. performed the second screening of retrieved papers and manually searched, reviewed and selected the final included manuscripts. C.M.L. and J.L. performed the quality appraisal of included papers. E.I.L., J.L. and C.M.L. extracted and summarised main information from included studies and drafted the manuscript. S.W.H. revised the manuscript. All the co-authors approved the final manuscript. Ethics of human subject participation: The manuscript is a scoping review done in existing published literature. Thus, ethical approval or participants’consents were not required.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980020001998
