Social services utilisation and referrals after seeking help from health services for self-harm: a systematic review and narrative synthesis

Search strategy

PubMed, PsycINFO, AMED, Social Policy and Practice, EMBASE, Medline, Web of Science and Social Care Online (from SCIE databases) were searched for articles published between 1 January 2000 and end of February 2023 (figure 1). Searches were conducted in Febuary 2023. Separate searches were conducted for MeSH terms and titles/abstracts (see online supplemental file 1 for full lists of search terms). Examples of search terms included: ‘self-harm’, ‘suicidal’, ‘self-injur*’, ‘self-poison*’, ‘self-cut*’, ‘parasuicid*’ and ‘overdos*’ to capture studies reporting self-harm. The automated searches were limited to years 2000 onwards for pragmatic reasons, as recommended when several thousand returns are generated.20 No language restrcitions were imposed. Grey literature was searched by Google search engine and the authors’ hard copies of reports were checked. Citation lists of included articles and narrower search terms were used to search for relevant evidence published before 2000.

Request reuse permission

Inter-rater agreement, data extraction and synthesis

Information relating to healthcare setting, patient group, study design and study findings (proportion referred to/in contact with social services) was extracted from the included studies. Critical appraisal tools (selected according to study design) from the Joanna Briggs Institute collection were used to assess the quality and risks of bias of each study.

SS screened all of the 2625 titles and abstracts and a second reviewer (FM) independently screened 10% (261). Differences in screening outcomes between the two reviewers were examined using percentage agreement and Cohen’s kappa to measure inter-rater agreement. Independent reviewing of 261 abstracts by FM led to three records recommended for retrieval (all also recommended by SS, with SS recommending an additional three records). The kappa score for these ratings was 0.66 and the level of agreement was 98.9% (258/261). Two reviewers (SS and FM) independently conducted quality assessments of all included studies. Following independent assessment, SS and FM discussed overall conclusions of quality, after which the assessment of one study was downgraded from high moderate to moderate.

Narrative synthesis was conducted guided by the Popay et al’s framework.21 This approach was used as we were interested in understanding the nature of social services utilisation after self-harm and exploring the health settings in which the evidence related to. In addition, pooling of results and meta-analysis would be unsuitable for this research question due to heterogeneity in study periods and study designs. Initially, we examined how the included studies might be organised to best describe patterns in their findings. Findings between studies, including across types of study designs, patient populations and outcome measures were then explored to analyse how these factors might influence study findings. The robustness of the evidence was then assessed by considering the strength of the overall evidence and in which populations and contexts in which it was representative.

Patient and public involvement

Four service users and carers with lived experience of health services for self-harm worked with the research team to design the study and interpret the findings. Specifically, the public contributors worked on the study inclusion criteria and the conceptualisation of social services care following self-harm. The group (named Mutual Support for Mental Health-Research, MS4MH-R) is linked with the National Institute for Health and Care Research Greater Manchester Patient Safety Research Collaboration.

Description of included studies

All seven studies were based in EDs and one also included a specialist self-harm unit.24 Five out of the seven included studies were conducted solely in the UK, one study was based in an Irish ED,26 and one study was based in two sites; in Oxford, UK and in Newcastle, Australia22 (table 1). Six studies derived the study outcome solely from information recorded in patient assessments or hospital records. One study24 used data from other health and social care agencies as well as the patient assessments in ED. Five studies included presentations of intentional self-harm involving self-poisoning and self-injury regardless of intent, one included self-poisoning episodes only22 and another included ‘suicidal behaviour’.26

Findings of included studies

In five out of six of the studies using ED patient data solely, between 1% and 4% of self-harm presentations were referred to social care services. Two of these studies reported the proportion referred among those receiving a psychosocial assessment and both found referral rates of around 4%. In the study comparing management between patients assessed by psychiatric staff and those assessed jointly by psychiatric staff and social workers, the proportions were 18% (9/50) and 44% (22/50), respectively.27 In a study using ED data with linked data from other health and social care agencies, 15% (62/427) of people attending a self-harm unit were referred to social services and 21% (466/2205) of all hospital self-harm patients were recorded as attending social services over a 3-year period.24

The years the studies were conducted ranged from 1983 to 2011. There was no clear evidence for a relationship between study findings and year of study. The highest proportion of patients referred to social care services was reported in Bateson et al’s study using data from 1983 and 1984, which may reflect availability of services at that time or the involvement of social workers in self-harm psychosocial assessments in that study.27 The quality of studies was generally high to moderate (table 1). One study was assessed as low to moderate due to poor definition of the exposure ‘suicidal behaviour’ and limited details regarding data extraction.26

Main findings

All studies included in this review were based in EDs and were mainly UK based. The quality of studies was generally high to moderate. All but one study derived the outcome data solely from information recorded in patient assessments or hospital records. Rates of referrals were generally relatively low—around 1%–4%. When actual service use data were captured, around one in five used social care services following self-harm—though evidence of this was limited to a single study. We found some evidence that when social workers were involved in conducting a psychosocial assessment, social services referral rates were higher.

Implications and comparison with existing evidence

A systematic review of resource utilisation in the year following a hospital presentation for self-harm found that social services costs comprised 13% of the total health and social services costs,32 the second highest cost after inpatient psychiatry. While the relatively low proportions referred to social services found in our review may appear discrepant with Sinclair et al’s findings, there are several factors to consider. Most of the studies in our review relied on routinely recorded clinical data from a single service, which may have underestimated the use of social care services following self-harm. When linked data from various health and social care agencies were considered,24 the proportion of people in contact with social services was considerably higher, suggesting people who have harmed themselves were already in contact with social services or subsequently began receiving care. In addition, most of the studies in our review were based on all patients identified as attending the ED for self-harm, regardless of whether or not they received a psychosocial assessment. Without a psychosocial assessment, there is unlikely to be an opportunity to arrange appropriate follow-up care. Furthermore, the true needs of patients are likely to be underestimated in these studies. Previous research in 31 hospitals in England found that the proportion of patients receiving a psychosocial assessment following self-harm varied widely, from 22% to 88%.13 Recent evidence suggests there are multiple significant barriers to psychosocial assessment faced by people who have harmed themselves.33 It is possible that the low referral rates found in our review reflect greater barriers faced by people who have social services needs.

Rates of referrals to social services are likely to be influenced by the professional background of the assessor. A previous study found that the professional background of clinicians conducting self-harm assessments influenced patients’ subsequent clinical management.34 The presence of multidisciplinary teams, including social workers, was thought to improve quality of aftercare for people presenting to ED following self-harm.35 In one study included in this review, assessments conducted jointly with psychiatrists and social workers had higher rates of referrals to social services.27 A study conducted in the 1970s found that social workers conducting assessments following self-harm placed greater emphasis on relationship and family problems and were more likely to identify physical illness compared with junior doctors.36 However, more recent studies indicate that referrals for follow-up care do not always lead to offers of care after being treated in hospital for self-harm.37 For example, significant clinician and patient barriers to the recommended psychological therapies following self-harm have been reported.35 38

There was an absence of studies from primary care settings in our study. A recent review found that there was limited information relating to social services needs, and social needs more broadly, recorded in UK primary care data; this gap may partly explain why no primary care-based were identified in our review. There are recognised gaps in self-harm clinical guidelines and training for general practitioners (GPs)39 40 and research into social care needs and referrals to social services among patients seeking help for self-harm from their GP is needed.

It is also important to understand how referrals made to social services following ED-presenting self-harm relate to future patient outcomes such as further health and social care services use and risks of further self-harm and death by suicide and other external causes. Currently, evidence relating to such outcomes is limited. One challenge is that observational evidence regarding outcomes among people referred to social services is subject to strong limitations of confounding. For example, a study of individuals in Sweden who had received welfare interventions during childhood, such as foster care, was at higher risk of suicide attempt in adulthood, even after adjusting for important measured confounders.18 Reviews of social work approaches to suicide prevention have found an absence of high-quality intervention research and advocate for more qualitative evidence to guide the development of interventions.41 42 In one of the few studies addressing this gap, Petrakis and Joubert evaluated a social work intervention comprising assertive brief psychotherapeutic intervention alongside support linking to community services, with individuals presenting to an ED after suicide attempt.43 While this was not a controlled study and there was no comparison group, individuals receiving the intervention reported improvements in several domains including work, finance, relationships and living circumstances after 3 months.

Given the lack of robust evidence for healthcare services-based psychosocial interventions following self-harm, integrated approaches involving social services are an important future direction for suicide prevention. Few interventions for self-harm have involved social services, though some social work-based and integrated interventions have been associated with improvements in mental health and social circumstances.43 A service for men with suicidal feelings addressed financial, housing and employment problems alongside providing emotional support.44 The service was associated with reductions in suicidal ideation and was valued by service users.

Strengths and limitations

This is the first review of social service referral and utilisation among people seeking help from health services following self-harm. The systematic review methodology with narrative synthesis enabled us to explore factors potentially influencing the findings reported in each study. Our research question aimed to confirm current practice and identify variation in practice, and we judged systematic review with narrative synthesis to be the most appropriate approach. However, we acknowledge that there is some overlap in the aims of systematic review and scoping review methodology, and that alternative approaches may have also been appropriate.19 All but one study examined referrals to social services only, with one24 also measuring utilisation of social services up to 3 years after a self-harm episode. While we are unable to draw conclusions based on one study, it is possible that examining referrals following a healthcare presentation for self-harm underestimates the level of social services utilisation among this patient group. The findings should be interpreted in the context of the small number of studies included in the review. In addition, the studies in this review spanned a range of time periods from 1983 to 2011 and no studies included years past 2011.

Emergency healthcare and social services in the UK faced numerous changes during and since that period. For example, in 2010/2011, a greater number of English EDs had formal arrangements with social services to provide assessments for self-harm patients than in 2001/2001.13 In addition, the introduction of integrated care systems (ICSs) in England from 2022 is aimed at linking National Health Service (NHS), local authority and community organisations to deliver health and care services. One of the aims of ICSs is to improve access to health and care services. It is possible that such partnerships will affect patterns of referrals following self-harm. Therefore, the findings cannot necessarily be generalised to the entire period of study nor to more recent years. The majority of studies were conducted in the UK, with one each in Ireland and Australia. The findings are unlikely to reflect international practice due to variations in service provision and the availability of health and social care services. We defined social services as care provided by social workers situated in health services, or social services provided by local authorities. However, studies generally did not define this study outcome in detail, so it was not possible to understand exactly what service people were referred to. We did not include studies of people seeking help for suicidal ideation; future research should investigate the clinical management of people presenting to services with suicidal thoughts. Finally, there was no evidence from primary care settings.