Background

Homelessness is a serious global problem [1] that affects adults of all ages [2]. National assessments of homelessness vary in the definition of homelessness that is applied, making direct comparisons between countries difficult. In the European Union, it is estimated that 4.1 million people have a homeless episode in a year [3] while, based on shelter use, an estimated 150,000–300,000 people experience homelessness in Canada in a year [4].

The definitions of homelessness from the Australian Bureau of Statistics’ (ABS) [5] used in this review were;

  • a person living in streets or without a shelter that would fall within the scope of living quarters, [if a person has] no place of usual residence who move frequently between various types of accommodation (including dwellings, shelters or other living quarters) or [if a person is] usually resident in long-term shelters or similar arrangements for the homeles (pp. 5); OR

  • a person living “in a dwelling that is inadequate; or has no tenure, or if their initial tenure is short and not extendable; or does not allow them to have control of, and access to space for social relations” (pp. 11), or

These are broad definitions that include people experiencing homelessness in different ways, such as a sofa surfer or someone with insecure housing tenure. The definitions thus reflect the variability and changing circumstances of people who experience homelessness. Applying this definition, the Australian census found 116,000 people were homeless on Census night in 2016, representing 50 homeless people per 10,000 [5].

The recent Rough Sleeping Statistics (Autumn 2017) for England [6] counts:

  • People sleeping, about to bed down (sitting on/in or standing next to their bedding) or actually bedded down in the open air (such as on the streets, in tents, doorways, parks, bus shelters or encampments), and

  • People in buildings or other places not designed for habitation (such as stairwells, barns, sheds, car parks, cars, derelict boats, stations, or “bashes” which are makeshift shelters, often comprised of cardboard boxes).

These definitions do not include people in hostels or shelters, people in campsites or other sites used for recreational purposes or organised protest, squatters or travellers [6]. This may explain why the single night snapshot for England was 4751 people or 20 per 10,000 [6], was so much lower than Australian Figs. [5]. What is concerning is that irrespective of the definition used, the rate of homelessness in Australia had increased by 4.6% between 2011 and 2016 [5], whilst in England there was an 18% increase in London and a 14% increase in the rest of England between 2016 and 2017, of people sleeping rough [6].

The state of being homeless impacts significantly on health [7], resulting in higher rates of premature mortality [4], with the average age of death for men and women experiencing homelessness in England being 47 and 43 years respectively [7]. In Canada and the USA, the current mean age of people experiencing homelessness is approximately 50 years [8,9,10].

Consequently, many people experiencing homelessness also present with age-related declining physical and cognitive functions [4, 9, 10]. Compared to the general population, people experiencing homelessness have higher levels of mental illness [8], drug and alcohol use [4], infectious diseases, including influenza, tuberculosis, human immunodeficiency virus (HIV), hepatitis, and sexually transmitted infections [11], oral cancer and other dental problems [7, 8], injury and assault and skin problems (often related to cold exposure) [11]. Approximately 40% adults experiencing homelessness report having at least one chronic health condition, which is often poorly controlled [7, 12]. Compounding poor health is the transient nature of homelessness, which often precludes sustained management of these health conditions [7]. For instance, people experiencing homelessness typically attend acute care services, such as hospital emergency departments, when they experience health crises, rather than accessing planned, preventative models of care [13].

Appropriate assessment tools are needed to identify the health needs of adults experiencing homelessness. Appropriate assessments can inform the development of policies and practices to provide effective prevention of health problems, and interventions that improve health. Appropriate in this context means that the assessment should use language, items, and constructs which are relevant to homelessness. For instance, many validated tools are available to evaluate sleep quality, nutrition and hygiene in the general population. These tools often have questions worded in a manner that assume that respondents sleep in a bed, and have access to food preparation and storage services, as well as bathroom and toilet facilities. As this is may not be the case for many people experiencing homelessness, it is important that health screening tools are properly validated and appropriate for specific circumstances of homelessness.

The usefulness of health screening and assessment tools to collect relevant information on particular individuals, and/or a specific target population is typically expressed in terms of validity, reliability and utility of application. Testing for validity establishes that the health screening or assessment tool captures all the constructs (or elements) that it purports to measure [14], and that the scores compare with other similar measures [14]. Reliability is about understanding variability and error in repeated measurement [14]. Health, as a construct for people experiencing homelessness, has been overlooked in a hard-to-reach group who do not easily engage with traditional services until there is a crisis. The identification of valid and reliable tools is the first step towards developing proactive health assessments that will be relevant and acceptable to people experiencing homelessness.

This overall aim of this review was to identify health screening and assessment tools that have been developed for, or used with adults experiencing homelessness. The review:

  • critiqued papers for methodological quality

  • determined which health screening and assessment tools were validated or specifically developed for adults experiencing homelessness, and

  • described the psychometric properties and utility of those tools.

Methods

The review proposal was registered with PROSPERO (CRD42017068769). The conduct and reporting of the review findings follow the PRISMA guidelines [15].

Search strategy

A preliminary database search identified common key words and MeSH terms that were then applied to the main search. The keywords included: homeless, homelessness, homeless persons, vagrancy, health status, health, health issues, health assessment and health screening. Where required, truncations (*) and MeSH terms were used in individual databases.

Library databases were searched in May 2017 and updated in September 2018. They included PubMed (and Medline), PsychInfo, Scopus, CINAHL and ERIC. The search was restricted to human studies published in English, and was not limited by date or type of article. Potentially-relevant articles were also identified through hand-searching of references cited in the reviewed articles. These were then screened according to the review method. The search strategy is provided in Additional file 1.

Review method

Articles identified through the library database search were exported into EndNote v7 [16]. Duplicate articles were removed, and remaining articles were independently screened by title and abstract (AB and TD). Potentially-relevant articles were then screened in full text. Comparisons were made between reviewers at each screening phase and a third reviewer (SJG) resolved discrepancies.

Exclusion criteria

This review did not include research about children and adolescents (younger than 18 years) experiencing homelessness, or studies that assessed health services for people experiencing homelessness, or studies that described tools that aimed to diagnose health conditions.

Inclusion criteria

A two-step inclusion criteria was applied. Firstly, articles were included when they described the application of health screening or assessment tools for health issues related to adults experiencing homelessness (aged 18 years and over). Included articles also explicitly described the study sample as experiencing homelessness as per those proposed by the Australian Bureau of Statistics’ (ABS) definition of homelessness and provided in the introduction of this paper [5].

Secondly, the assessment and screening tools reported in these articles were aggregated to identify a sub-set, for which claims had been made about being developed for or with, and/or validated with people suffering homelessness. Where necessary, additional developmental literature for each tool was sourced for information on whether, and how the tool had been validated for the target population. This was completed by initially checking the reference lists and/or undertaking an independent literature search for additional material on the tool itself. Studies that used tools which had not been validated for the target population were noted, but no further analysis of methodological quality was undertaken.

Data extraction

Data were extracted to detail all the articles by author, publication year, study population, setting, purpose, tool descriptions, condition/s assessed, primary outcomes, and tool psychometric properties (if applicable). Authors were contacted for further information when there was a lack of detail regarding the tool.

Hierarchy of evidence

The hierarchy of evidence of the articles reporting on validated tools was assessed based on the National Health and Medical Research Council of Australia (NHMRC) recommendation [17]. The NHMRC hierarchy of evidence is internationally used in evidence reviews. It is based on the CEBM hierarchy model in which the hierarchy of evidence for diverse research questions contains different types of research designs (fit for purpose). It was appropriate for this review because of the likely heterogeneity of research designs.

Critical appraisal

This took two forms.

  1. 1.

    Two reviewers independently appraised the methodological quality of the articles using validated tools relevant to the study design. These were 1) McMaster University Quantitative Critical Review Form [18], 2) McMaster University Qualitative Critical Review Form [19] or 3) the Leeds Evaluative Tool for Mixed Method Studies [20]. A third reviewer (SJG) arbitrated on discrepancies between reviewers, when required. Seven individuals conducted paired appraisals (AB, TD, SG, NB, MK, TB, CH). The McMaster critical appraisal instruments (designed for quantitative and qualitative designs) provided a ubiquitous approach to scoring methodological quality, without the need to use different tools for different designs (for instance the CASP suite). The McMaster critical appraisal tool provided a standardised way of comparing methodological quality across diverse research designs.

  2. 2.

    To assess the psychometric properties of the tools for which claims had been made regarding validation in the target population, the International Centre for Allied Health Evidence (iCAHE) Ready Reckoner was used. This is a structured checklist of psychometric properties and utility of application, developed for assessment and outcome measures [21]. The Ready Reckoner allows structured standardised comparison of psychometric properties of assessment instruments. It requires reviewers to identify the developmental literature on assessment instruments, and extract information on what psychometric properties were tested (and how), and what information is provided on utility. To gain a point for each category in the Ready Reckoner, instruments needed to have tested for, and reported on, the values for each psychometric test. The values themselves are not reported in the Reckoner, as the Reckoner is simply a way of summarising the amount and type of testing to which an instrument has been subjected during developmental phases. Moreover, as values are often dependent on sampling strategies, sample size and comparison instruments, it generally requires a great deal more information to interpret a value, thus reporting a value on its own can be misinterpreted. As completion of the Ready Reckoner requires access to the developmental references, these are cited, and readers interested in exploring more about an instrument can access further information efficiently to make their own judgements.

Data synthesis

Tools were categorized to; 1) health screening or 2) health assessment.

Results

Search results

The database search and screening process is summarized in the PRISMA flow diagram (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram detailing the search and selection process

Full size image

Potentially-relevant studies and the tools they reported on

Of 2122 identified studies, 37 were identified as potentially-relevant (Table 1). These studies reported on 73 health screening or health assessment tools, across 11 domains: oral health [22,23,24,25,26,27,28,29,30,31,32]; health related quality of life, and health status [9, 33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48]; nutrition [49,50,51,52,53,54,55,56]; psychological and cognitive function [9, 22, 34, 35, 41, 42, 44, 56,57,58,59]; substance use [28, 33,34,35,36, 51, 57]; injury [29, 30, 36, 58]; chronic conditions [33, 44, 45, 60, 57]; demography and anthropometry [34,35,36,37,38,39,40, 44, 49, 51,52,53,54,55, 60]; functional decline and frailty [9, 37, 38, 54]; hearing and vision [9, 36, 61]; and pelvic floor health [9].

Table 1 Assessment instruments reported in the included literature, developmental reference(s) for the instrument (DRI) if relevant to homeless populations, and population(s) on which the instrument had been developed or validated. Shaded tools are validated or developed with people experiencing homelessness
Full size table

Description of tools validated for people experiencing homelessness

Eleven assessment tools (reported in 13 articles) had been developed for, and/or validated in people experiencing homelessness. These tools addressed domains of health-related quality of life and health status, substance use, injury, mental health and psychological and cognitive function. They comprised:

  • Addiction Severity Index (ASI) [33]

  • Beck Depression Index (BDI II) [35, 45]

  • Brain Injury Screening Questionnaire (BISQ) [58]

  • Brief Instrumental Functioning Scale [9]

  • Colorado Coalition for Homelessness Consumer Outcome Scale [34]

  • Delighted-Terrible Faces Scale (DTFS) [40]

  • Rural Homelessness Interview Schedule [35]

  • Life Fulfilment Scale (LFS) [44]

  • Nottingham Health Profile [44]

  • Short Form Survey-12 (SF-12) and/or Short Form Survey-36 (SF-36) [39, 60, 41]

  • World Health Organization Quality of Life 100 (WHOQoL 100) and/or World Health Organization Quality of Life BREF (WHOQoL-BREF) [26, 42, 43]

Fifty-five other assessment and screening tools were potentially-relevant to persons experiencing homelessness, but no evidence was found that validity had been established with this target population. Potential relevance was assessed in terms of generalizability or transferrable constructs between the population with which they had been developed and tested, and people experiencing homelessness. This was defensible for measures which are common to the general adult population, such as anthropometry and demography; oral health; continence; vision and hearing but not risk factors. Other potentially-relevant tools were found for conditions which could well be experienced by people who were homeless, including substance use, mental health issues, inadequate diet, poverty, injury and chronic illness. There were a further five tools for which little could be found in terms of psychometric properties or utility.

Table 1 lists the tools, the studies in which they were reported, the populations with which the tools had been developed or validated (where available), and references for additional literature providing evidence that the tool was valid for people experiencing homelessness. These studies are highlighted for ease of identification. The characteristics table for all included studies is provided in Additional file 2.

Hierarchy of evidence

Of the 13 studies that used validated tools for the target population, three were classified as NHMRC III-1 [34, 35, 44] (prospective observational cohort studies). Two studies involved the development and validation of new assessment tools [34, 35] and the third [44] tested previously validated outcome measures in a pre-post study of the impact of case management for people experiencing homelessness. The remaining studies were classified as NHMRC III-2 as they were all cross-sectional studies reporting on surveys or measurements taken at a single time point.

Methodological quality of included studies

Methodological quality of the included studies ranged from 57.1% [38] to 90.9% [9, 34, 39]. For the NHMRC III-1 studies the median quality score was 72.7% (25th % 64.9-75th % 81.8), whilst for the NHMRC III-2 studies the median quality score was 81.8% (25th % 72.7- 75th % 90.9). The individual quality appraisal items for each potentially-relevant article are provided in Additional file 3.

Ready reckoner scores

The psychometric properties and utility of the validated tools are summarized in Table 2. As all tools had been specifically validated and developed for people experiencing homelessness, the question on relevance was deleted. The best performing tool for the target population was the WHOQoL 100 or WHOQoL BREF (scoring 100%), followed by BDI II, SF-12 and SF-36, and then the ASI (all scoring 94%). The Colorado Coalition for Homeless Consumer Outcomes scale ranked best of the purpose-built tools (82%), with the Rural Homelessness Interview Schedule (RHI) scoring poorest (56%).

Table 2 Ready Reckoner psychometric properties and utility
Full size table

Discussion

This systematic review is the first that we know of, that has collated information on the validity of health screening and assessment tools used for people experiencing homelessness. Accurate information on health needs can only be obtained by the use of psychometrically-sound tools applicable to this population. While 73 health screening and assessment tools were reported as having been used to collect information on a range of issues in people experiencing homelessness, only 11 had published evidence of psychometric testing applicable to this target population. These tools captured information regarding quality of life, health status, substance use, injury, and psychological and cognitive function.

No validated tools were identified that assessed oral health, chronic conditions, anthropometry, demography, nutrition, continence, functional decline and frailty, or vision and hearing. However, assessments of physical constructs (such as oral health, anthropometry, vision and hearing) could be applied to homeless people on a presumption of validity, because the constructs would be measured clinically in the same manner as for people living in dwellings.

On the basis of the findings of this review, information reported on chronic health conditions, demography, functional decline and frailty, nutrition and pelvic floor health in people experiencing homelessness, may not be valid indicators of their health. For instance, it cannot be assumed that someone ‘living rough’ is a ‘community dweller’, which was a common description of the population for which the non-validated tools for people experiencing homelessness had been developed or tested [9] (See Table 1).

The iCAHE Ready Reckoner summarises common interpretations of validity and reliability as extracted from the developmental literature, as a way of demonstrating how well and thoroughly, a tool has been tested [21]. However, it is important to view testing in terms of the real world experienced by the target population, as findings may be constrained by the capacity of individuals to interpret and complete assessments. For example, issues concerning literacy and memory recall (due to substance abuse, illness or dementia) may affect acceptability of the tool and the quality of data collected with it. Thus, tool utility for the target population also needs to be considered in terms of its language and literacy levels, how long it takes to complete the items, and how many items the tool includes. Other aspects which assessors may need to consider are whether the tool is freely available and whether there are population norms for comparison.

Human health research must align with the ethical principles of beneficence, non-maleficence, health maximization, efficiency, and respect for autonomy, justice and proportionality [72]. People experiencing homelessness are a particularly vulnerable population, therefore assessment tools should enable efficient collection of representative data, relevant to the individual, and be delivered with minimum intrusion and maximum likelihood of doing good [1, 7, 72].

The health screening and assessment tools identified in this review contained 60 individual measures, of which 55 had available developmental material suggesting that with further research, the tool might be made relevant to people experiencing homelessness (See Table 1). Assessments of physical constructs (such as oral health, anthropometry, muscle strength, balance) appear to have been applied to the study population on a presumption of validity, because the constructs would be measured in the same manner as people living in permanent dwellings. The importance of oral health assessment of the target population was highlighted by the 11 different oral health tools identified in this review (of which nine had evidence of development and application). Whilst none had been validated with people experiencing homelessness, two warrant further development in this population; the Shortened Oral Health Impact Profile (OHIP-14 tool incorporating oral health-related quality of life interview) [22] and the Geriatric Oral Health Assessment Index (GOHAI) [27]. Both had been developed for efficient application with older people, needing minimum physical assessment time and including questions on oral health quality of life. Similarly, the validity of objective measures such as anthropometry, vision and hearing could be readily applied in homeless populations, as testing would be presumed to be similar for any adult. Heterogeneity of people experiencing homelessness supports the need for validated tools appropriate to specific subgroups, for example; different sex-age groups, or people with varying language and reading capabilities.

Without evidence of validity, self-report tools of most concern regarding applicability to homeless populations include those measuring health domains such as chronic conditions, functional decline and frailty, cognitive function, continence and nutrition. Not only could concerns be raised about the capacity and willingness of homeless people to complete written self-reports (because of literacy, vision or trust) but also question relevance to their circumstances. Recalling information over a set time period is a common element in most self-report tools, and this assumes that events of interest have occurred within that time period, and that an individual’s memory allows them to be accurately recalled and expressed. This is particularly problematic in dietary intake tools where the types of foods used to prompt memory may be inappropriate to circumstances of homelessness, and frequency and location of eating may be variable [50, 56, 60].

Two tools were identified which had been developed directly for, and with input from, homeless people [34, 35]. These tools had moderate to good study methodological quality, but poor to moderate tool property quality. The latter reflects the specific nature of the study population, and includes additional testing for reliability and external generalizability. Both tools captured information leading to homelessness, employment options, education, children, housing options, and access to benefits. These constructs had been identified during interviews with people experiencing homelessness, and were included in the fit-for-purpose tools, rather than existing tools that had been adapted for homeless populations. Consulting with homeless adults appears to be essential in designing sensitive, comprehensive, and multidimensional tools that capture the range of health concerns of this specific population group [14, 72, 73].

The WHO global healthy ageing initiative promotes good health for all adults, irrespective of country or circumstance [73, 74]. People ‘living rough’ have been shown to have symptoms of aging much earlier than the general population [4, 7, 9, 10]. Therefore, a lens might be placed on comprehensive test batteries developed for older people, to adapt them for people experiencing homelessness. Concerns in the elderly such as poor balance, falls, vision loss, incontinence, arthritis, chronic pain and poor skin health may also be relevant to younger homeless people [2, 10]. However, it will be necessary to engage homeless people in determine which criteria are important to them. Improving the quality and comprehensiveness of screening and assessment tools to quantify the prevalence and impact of such conditions currently associated with ageing, but which may also be found in homeless populations will provide important new information on the true impacts of homelessness. Additionally, this will provide an improved understanding of accelerated aging associated with poverty and sleeping rough.

Limitations

This review was conducted as comprehensively as possible, using an in-depth search approach, that focused on English-language health screening and assessment tools developed for, used in, or potentially applicable to, homeless populations. It did not include articles about assessment of the specific diseases which may occur in people experiencing homelessness (for instance cardiac or respiratory diseases, mental illnesses or infectious diseases). The findings have not been differentiated by age group, sex, location or geography (which limits comparative insights into variability by sex, and among homeless populations globally). Future research might address these knowledge gaps to better establish health screens and assessments specific to different types of homelessness, and different types of people experiencing it [6, 7].

Conclusions

The findings of this review confirm the complexities of assessing, accurately representing and addressing the potentially poor health of homeless populations. Few studies have reported on validated and reliable health screening and assessment tools, informed by input from people experiencing homelessness. Improving harm reduction strategies and early identification of health conditions in adults experiencing homelessness is vital, recursively flowing to and from empirical evidence, government policies and targeted services to address these significant and long-standing health inequities. That said, if research is asking, and by extension reporting on, the ‘wrong’ items and measures, there is significant potential for discord between what is espoused as needed, versus what should be triaged, to advance the health of homeless people. Based on the findings of the current review we recommend the development and consistent use of a suite of measures informed by and validated for people experiencing homelessness. This is a vital first step to accurately capturing, reporting and addressing the complex health needs of this vulnerable population.