Abstract
Introduction:
Homelessness has risen across high-income countries in the last decade, and in the United Kingdom, there has been a drastic increase in people living on the streets. Due to these increases, policy responses from public services are required to address the needs of this group. The risk factors for homelessness and conditions that this group live in mean they are at elevated risk of both mental health problems and contact with the criminal justice system. Despite this, there is little previous research on the homeless in police custody.
Methods:
Our study used a matched sample of homeless (n = 77) and non-homeless (n = 77) individuals to examine whether there were different needs across this group and whether the responses of a criminal justice mental health service differ for this group. This study is a secondary data analysis of a more extensive study.
Results:
Homeless and non-homeless detainees referred to the mental health service were broadly similar. However, differences in some variables show that homeless detainees had higher rates and frequency of substance misuse as well as some suggestion of more acute immediate need. Onward referrals were lower for homeless people, and it is not clear why this is the case. In addition, for those referred contact with services over time was reduced compared to the non-homeless group.
Discussion:
Our findings indicate that mental health services in police custody may need adaptations to ensure homeless individuals’ higher level of need is addressed and that they receive appropriate care both during and after detention. Further quantitative and qualitative research is needed to confirm why responses differ and to assess what can be done to address this issue.
Introduction
There has been a rise in homelessness across many European countries over the last decade and in the United Kingdom, with recent estimates suggesting almost 5,000 people are officially recorded as homeless (Fransham & Dorling, 2018). Charities estimate this number to be far higher, but nonetheless, official figures have documented an increase in rough sleeping of more than 200% since 2010 (Fransham & Dorling, 2018). In London, trends similar to national figures have been found with a doubling of rough sleepers between 2010 and 2017 (Fitzpatrick et al., 2017), and these figures do not include ‘the hidden homeless’ who are in precarious or insecure housing, whose numbers are thought to be far greater still. This rise in homelessness occurred within the context of a difficult economic climate and has been exacerbated by shortages of social or affordable private housing and changes to the benefit system, with the introduction of universal credit. This has occurred at the same time as cuts to local authority social services that would usually support this group (Fitzpatrick et al., 2017; Perry & Craig, 2015). Increases in homelessness and street living warrant more focus on addressing their needs and on the public services providing support to those who are homeless. This is particularly true for health services and the criminal justice system, which often encounter this group.
Homeless populations have common risk factors for both higher rates of morbidity and mortality and involvement with the criminal justice system, as both perpetrators and victims of crime (Aldridge et al., 2018; Beijer et al., 2012; Gentil et al., 2019; Hill, 2016; McNamara et al., 2013; Nilsson et al., 2019; Tsai & Rosenheck, 2012). These risk factors are exacerbated by the social conditions in which homeless people live during periods of homelessness, but also when they are accommodated with lack of shelter, poor social support and problems accessing medical care. This situation becomes ‘mutually enhancing’, with existing vulnerabilities and homelessness combining to heighten the risks of poor health, criminality and victimisation (Busch-Geertsema et al., 2010), leading to contact with the criminal justice system.
People who are homeless are more likely to come into contact with the criminal justice system for several reasons. There is some evidence of increased rates of offending, but it has also been acknowledged that those living on the streets are disproportionately targeted by police with heavy handed responses for minor offences, such as public urination and begging (McNamara et al., 2013). Once arrested, homeless people are often treated differently by police and courts and are less likely to be granted immediate bail, as community alternatives are thought to be less likely to succeed (Department of Health, 2009). In addition, this group is likely to have histories of offending and to have already spent time in prison (McNamara et al., 2013; Tsai & Rosenheck, 2012). In addition to this higher risk of becoming involved in the criminal justice system and receiving harsher responses, people who are homeless are also more likely to have poor health. Rates of morbidity are high for a range of physical health problems (Busch-Geertsema et al., 2010; Fransham & Dorling, 2018), and poor mental health and mental disorders are widespread. Estimates of prevalence for this group are difficult to quantify due to the difficulties of including this group in research, but rates of serious mental illness are estimated to be 25% to 30% among people who are street homeless and in direct-access hostels (Busch-Geertsema et al., 2010; Fazel et al., 2008; Perry & Craig, 2015; Saddichha et al., 2014) and over half are reported to have problems with alcohol and drug dependence (Fazel et al., 2008). Because of the co-occurrence of these issues, comorbidity and dual diagnosis of severe mental illness of substance misuse and/or personality disorder are also common in this population.
There is also an additional dimension of ethnicity in both homelessness and contact with the criminal justice system which needs to be considered. While people from minority ethnic groups make up 13% of the general population (Office for National Statistics, 2012), they are vastly overrepresented in the homeless population where they are estimated to be a third of cases (Perry & Craig, 2015). In addition, recent increases in homelessness have not been evenly distributed and rates of homelessness from these groups have increased at a faster pace than for White British people (Shelter, 2018). In addition to higher homelessness, overall arrest rates for people from minority ethnic groups, particularly for those from Black ethnicities, are higher than those for other ethnic groups. This issue is highlighted most prominently in the Lammy (2017) report which emphasised the creation of misrepresentative contact with the criminal justice system based on race and not frequency or severity of offending.
As homeless population at are higher risk for both contact with the criminal justice system and mental health problems, it follows that this group will be overrepresented within police custody and in referrals to mental health teams working in this setting. It is known that many detainees have mental health problems (McKinnon et al., 2013), and this notion is supported by recent research examining detainees in police custody where it was found that 8% of mental health referrals are homeless (Forrester et al., 2017) meaning they are vastly overrepresented compared to the overall proportion of people who are homeless. Despite the high number of homeless people with mental health problems in police custody, there has been a lack of research on the characteristics of this group or the response from health services to this population while in custody. Focusing on this issue is important to ensure that people who are homeless receive appropriate and high-quality care in custody and that this contact with public services is used as an opportunity to identify people in this marginalised group who are in need and may not contact services in the community.
This study aimed to examine these issues by using a matched cohort of people referred to mental health services in police custody to investigate differences in characteristics between homeless and non-homeless people within this population. Furthermore, the study aimed to examine whether health service responses to homeless and non-homeless people were different at assessment and during follow-up in the community.
Method
Setting and sample
The participants in this study were retrieved from a larger sample examining 1,092 referrals to a criminal justice mental health service operating in two police stations in one South London borough (Forrester et al., 2017). The service operated 7 days a week between the hours of 8:00 a.m. and 8:00 p.m. and used an ‘open referral system’ allowing referrals from clinicians, non-clinical staff (e.g. police officers) and detainees themselves. Assessments were designed to take place within 4 h and were undertaken by community psychiatric nurses after a triage process identified the most urgent cases. Community psychiatric nurses had access to telephone advice from a consultant forensic psychiatrist if required, and in practice, this was generally only used where support for diversion to hospital from the police station was needed in the most acute cases.
Data collection
Data were collected as part of routine clinical practice of the criminal justice mental health service, and information was entered by clinical staff into a standardised record that was maintained in the mental health trusts’ electronic records. Information on the following variables was obtained and used in this analysis from a combination of self-report and review of electronic clinical records: age, gender, ethnicity, diagnosis, frequency of substance abuse, history of and current self-harm and health service utilisation. Information on intellectual disability was confirmed using the Learning Disability Screening Questionnaire (McKenzie et al., 2012) if this was self-reported or suspected by clinical staff. Onward referral by the criminal justice mental health service was recorded at the time of assessment, and records were reviewed after 2 and 4 weeks and 3 and 6 months to audit appropriate follow-ups after contact with the service.
A more detailed account of the setting, sample and procedure can be found in the initial evaluation (Forrester et al., 2017).
Matching approach and analysis
Participants who reported being homeless (n = 77) were identified from the larger database, and each of these participants was randomly matched with a participant of the same age and gender who did not report being homeless. Differences between these two groups were examined on a series of variables relating to ethnicity; prevalence of mental disorder, substance misuse and risk of harm to self; previous health service utilisation; onward referral by the criminal justice mental health service; and contact with services at a series of follow-up time points after referral. Data were categorical, and chi-square tests for multinomial outcomes and bivariate logistic regressions for binary outcomes were used to analyse differences between the homeless and matched non-homeless groups. Data analysis was conducted using SPSS version 22 (Figure 1).
Flow diagram.
Ethics and governance
Local National Health Service (NHS) Trust governance approval was received for the study as a service evaluation project and relevant approvals from agencies within the criminal justice system were obtained.
Results
Background and demographic characteristics
On the two matching variables, the gender of participants in the homeless and non-homeless group was equal with 65 males (84%) and 12 females (16%) in both groups. The average age was 37.50 (SD = 11.43) years for participants in the homeless group and 37.61 (SD = 10.8) years for participants in the non-homeless group.
The most frequent self-reported ethnicity for participants in both the homeless and non-homeless group was White British (n = 58, 37.7%) with large groups of participants from White Other (n = 26, 16.9%) and Black (n = 54, 35.1%) ethnicities and fewer reporting Mixed (n = 16, 10.4%) or Asian ethnicity (n = 3, 1.9%). Details for the homeless and non-homeless group are shown in Table 1, and no significant difference in ethnicity was found between the groups.
Self-reported ethnicity, prevalence of mental disorder, previous and current risk of harm to self, health service utilisation and onward referral and engagement with services.
GP: general practitioner.
Of those referred.
Prevalence of mental disorder, substance use and risk of harm to self
The prevalence of mental disorder, frequency of substance use and previous and current risk of harm to self are reported in Table 1. Overall rates of mental disorders were similar across the two groups with 85.7% (n = 66) of the homeless group and 80.5% (n = 62) of the non-homeless group having a mental disorder noted in clinical records or from self-report. The most common disorders were depression, schizophrenia/psychosis, personality disorder and substance misuse. Rates of these disorders were similar across groups apart from a larger proportion of the homeless group with substance misuse (35.1%, n = 27) compared to non-homeless (14.3%, n = 11). This difference was significant (odds ratio (OR): 8.48, p = .004). A further analysis demonstrated that of those who misused substances, participants in the homeless group did so more frequently than those in the non-homeless group, with daily use most common in those who were homeless (72.7%, n = 32) and weekly or monthly use most common in the non-homeless group (68.9%, n = 31).
A large number of both groups had a history of both intentional self-harm and attempted suicide and a substantial minority reported current ideas of suicide, were judged to present a current suicide risk and had self-harmed prior to entry to custody or in the recent past. Of these variables, a significantly higher proportion of those in the homeless group had history of suicide attempts than in the non-homeless group (53.7% vs. 33.8%; p = .019) and had significantly higher rates of recent self-harm (16.7% vs. 4.2%; p = .023).
Previous health service utilisation and health services response after custody
There were large differences across the homeless and non-homeless group on previous and current health service utilisation. Fewer of the homeless group were currently registered with a general practitioner (GP) than the non-homeless group (58.9% vs. 92.0%; p < .001) or were currently a client of a community mental health team (43.6% vs. 63.0%; p = .045). There were not significant differences on either current use of psychiatric medication or previous inpatient care. Information on these variables is shown in Table 1.
After assessment by the criminal justice mental health service, significantly fewer of the homeless group had onward referral to other services than the non-homeless group (36.4% vs. 64.9%; p < .001). Of those with an onward referral, fewer made any contact with services over a 6-month follow-up period (50.0% vs. 74.0%; p = .001) and differences in contact were maintained at a series of time points within this follow-up. Each of these differences was significant in statistical tests.
Discussion
This study addresses a gap in research by examining the differences between homeless and non-homeless populations referred to a mental health service in police custody. Our findings suggest that homeless and non-homeless detainees referred to a criminal justice mental health service within police custody have broad similarities but there appears to be some important differences in their clinical characteristics, current risk of harm and health service utilisation. In addition, this study aimed to determine whether a mental health service within police custody had differing responses to the homeless and non-homeless group. It appears that there are differences in response with fewer onward referrals for those who are homeless and different levels of follow-up by mental health services after custody.
Homeless and non-homeless people had similar ethnicities and similar overall rates of mental disorder, but on substance misuse, in particular, homeless people had both higher rates of use and greater frequency of use than non-homeless individuals. Among the homeless participants, daily use of substances was most common and weekly or monthly use was most common among non-homeless participants. A large number of both groups had a history of self-harm, and the high level of present suicide ideation and current suicide risk in this population is concerning but was not more present in homeless groups. However, there did seem to be notable differences between the groups. For homeless people, there was evidence of more recent and current self-harm, and a higher proportion had a history of suicide attempts compared to the non-homeless group. In addition, many fewer homeless people were currently registered with a GP or were in contact with a community mental health team. These findings suggest that homeless people referred to mental health teams within police custody share many characteristics with non-homeless people but may be in crisis manifested by recent self-harm and may face more unmet needs due to a lack of contact with other health services in the period preceding arrest.
It is concerning that despite high levels of mental health need shown in this study and expected physical health needs in this group, registration with GPs was low and appears to be lower than general samples of homeless people in general (Elwell-Sutton et al., 2016). GPs act as gatekeepers to primary and secondary care services in the United Kingdom, and without this point of contact, it is difficult to access services and there is a lack of ongoing monitoring of health concerns and needs (Loudon, 2008). Mapping surveys show that some primary care services for homeless people are available in the South London area from which this sample was recruited (Crane et al., 2018), but they are limited in number and resource, and our results suggest more intervention is needed for this group. To achieve this, an additional focus may be needed from specialised services and local commissioning groups to ensure that services that recognise the marginalisation and vulnerabilities of homeless people are implemented.
In this study, homeless people had a lower level of onward referral compared to non-homeless people, despite having poorer mental health at assessment in police custody. It is unclear why this is the case and there may be several reasons for this. Low levels of GP registration and contact with community mental health teams may present difficulties for onward referral as the lack of a fixed address or health professionals acting in a care coordination role means other health services have no way of acting on already limited referrals. Another reason may be linked to the management of people with dual diagnoses of mental illness and substance misuse, which can be challenging, and it is possible that higher levels of substance misuse in this group is driving this response (Priester et al., 2016). In time-pressured assessments in police custody, substance misuse may be more apparent and salient than diagnoses or symptoms of mental illness, and this may lead to overshadowing and lead to a lesser focus on mental health. Alternatively, there could be a lack of clarity from staff about where to refer dual diagnosis cases or these services may be limited. For the group of homeless people who were referred, there were lower levels of contact over a period of 6 months with only a small number of those referred making further contact with mental health services. This is an issue which has been documented previously in the wider homeless population (O’Brien et al., 2009), and as assessment within police custody presents an opportunity for mental health services to identify people within this group with unmet needs, consideration is needed of how continuity of care after referral can be effectively achieved. Mental health assessments in police custody are an opportunity for contact with this underserved group, but it should be recognised that these teams have limited scopes and their ability to coordinate care is limited by the short time detainees spend in police custody and restricted resources. Each of the issues mentioned above may be best addressed by interventions within criminal justice services that ensure GP registration is pursued after release from detention and these services, especially if designed for homeless people, should be more equipped to ensure other appropriate primary and secondary follow-up care is accessed (Gray et al., 2017; Leclair et al., 2019; Reingle Gonzalez et al., 2018).
The study has several limitations that should be acknowledged. First, the sample was recruited from a single police mental health team in London, and the results may not be applicable to other settings within the United Kingdom or internationally. Second, the distinction between the homeless and non-homeless people used in the sample was derived from self-reporting of housing status at assessment in police custody, and there is a possibility that the group classed as non-homeless here has histories of homelessness. This snapshot approach may not fully capture the complexity of this issue (Brown et al., 2017). Third, the mixed approach of collecting information from both self-report and clinical records ensured that data were collected on all referrals to the mental health service, but there may have been inaccurate reporting on some self-report items, and the quality of information recorded in clinical records is variable. Fourth, the approach used for statistical analysis has limitations that should be acknowledged. The matching approach included only age and sex and was able to create a similar group on these outcomes, but it is possible that unbalanced characteristics on both observed and unobserved variables led to differences in response from the mental health service to homeless people. Despite this possibility, the groups did seem balanced on key characteristics in bivariate tests, and if any direct effect was expected, it would be towards a greater impact of homelessness on service response as some variables indicated this group had more severe mental health problems.
Future studies would be helpful to determine whether the results seen in this study are replicated in other settings. Larger and nationally representative samples could employ a more comprehensive propensity score matching approach using a wider range of variables, and this could reinforce the notion that differences in characteristics and service response across the groups seen here are as a result of homelessness and not of other related variables. In addition, quantitative and qualitative studies exploring response to homeless populations within police custody may help to confirm any reasons for difference and identify where interventions could be targeted to ensure that mental health service responses to homeless people are appropriate and represent high quality of care. The results of this study suggest that health service utilisation, both before and after assessment in police custody, is poor for this group, and adaptations to models of community care should be evaluated to ensure that this group, who are at risk of poor health outcomes, are appropriately followed up by services. Interventions to increase GP registration after this initial contact with services may prove an important step in this process.
Conclusion
Rates of homelessness have risen across high-income countries over the last decade, and in the United Kingdom, there has been over a 200% increase since 2010. This has led to increased attention on the needs of this group and focus on policy responses. Those who are homeless have common risk factors for both poor mental health and contact with the criminal justice system and the conditions in which people who are homeless live further exacerbate this risk. Despite this, little research has focused on homeless groups with mental health vulnerabilities in police custody and whether they have distinct needs and receive appropriate responses from mental health services in these settings. Using a matched sample of homeless and non-homeless people from a larger dataset, our study suggests that these groups have broadly similar characteristics; however, homeless people appeared to have greater current distress and more problems with substance misuse. Homeless people were also less likely to be registered with a GP or in contact with community mental health services and, despite seeming to have acute needs, were less likely to receive onward referral after initial assessment and had lower levels of follow-up after time. This study addresses an issue that has not received focus previously and suggests that focus is needed on homeless groups in custody to ensure that their needs are being met and appropriate onward referrals are made. Future research will be needed to add certainty to these conclusions and to evaluate how high-quality care can be ensured for this group.
Footnotes
Acknowledgements
Thank you to Guy’s and St Thomas’s Charity, whose funding made this project possible. Thank you also to team members who supported data collection and to our partner organisations, who helped facilitate this work.
Conflict of interest
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This service received funding from Guy’s and St Thomas’ Charity as part of the provision of a mental health service in the criminal justice system.
