The characteristics, life problems and clinical management of older adults who self‐harm: Findings from the multicentre study of self‐harm in England

Older adults have a high risk of suicide following self‐harm. Contemporary information on self‐harm in this population is needed to inform care provision.


| INTRODUCTION
Mental health conditions in later life are common, with suicide rates globally being highest amongst those aged 65 years and over. 1,2 Selfharm has been identified as a significant risk factor for future suicide, and a number of studies have found this association to be especially strong in older adults. [3][4][5][6][7][8] A UK multicentre study conducted on Emergency Department (ED) presentations in England between 2000 and 2007 showed that adults aged 60 years and over who selfharmed were 67 times more likely to die by suicide than older adults in the general population. 6 With the world's population ageing, and around 20% of UK residents now aged 65 years and over, contemporary insight into the incidence, preceding problems, and outcomes following hospital presentation for older adults who selfharm is needed to inform care provision and the prevention of selfharm and suicide in this population. 9 Globally reported rates of self-harm in older adults which involved a hospital presentation have ranged from 19.3 to 65.0 per 100,000. 8,10 The high female-to-male ratio of self-harm seen in younger adults reverses in older age, with a higher rate of self-harm seen in men who are 75 years and over than in women of the same age. 6,11 Motivations and life problems preceding self-harm and suicide in older adults differ from those in other age groups. Older adults are less likely than younger people to report self-harm as a coping strategy, more often having difficulties with physical health, an isolated lifestyle, with also a high prevalence of interpersonal concerns as preceding life problems. 7,[12][13][14][15][16][17] Gender differences in the life problems experienced prior to self-harm have also been reported in older adults. 5 Previous studies have found the prevalence of self-harm repetition in older adults is lower than in other age groups. 18,19 In one UK centre, 8.2% of adults aged 60 years and over re-presented to hospital with self-harm within 12 months 5 while a repetition prevalence of 12.8% was reported in a subsequent study from multiple centres. 6 Identified risk factors for self-harm repetition in one study of older adults included being 60-74 years of age, previous self-harm, problematic alcohol and substance misuse, and previous psychiatric treatment. 6 Similarly, in another study, contact with psychiatric services was associated with repetition but in contrast with the first study, drug and alcohol misuse were less commonly associated with repetition. 4 National Institute for Health and Care Excellence Guidance on the management of self-harm in older adults recommends assessment by a mental health specialist to consider the potential presence of physical illness, cognitive impairment, and depression, the home and social situation and the higher risk of suicide associated with self-harm for individuals over the age of 65 years 20 A systematic review exploring the characteristics of self-harm in older adults identified that in the past only 52.4% received a psychiatric assessment following self-harm. 8 Morgan and colleagues investigated clinical management of older adults following self-harm in a primary care cohort and found 59.3% were prescribed an antidepressant (including 11.8% who were prescribed a tricyclic antidepressant). 21 Only 11.7% were referred to mental health services, with women more likely than men to be referred and prescribed psychotropic medication. 21 Older adults are an often-underrepresented group in self-harm research, which may be attributed to self-harm rates being lower in this population relative to other age groups. 17 Despite the lower prevalence, the risk of future suicide is highest for those aged 55 years and over. 22 Historically, studies have compared self-harm in older individuals to younger groups, with minimal attention to within-group differences in this population which encompasses a large age range and likely occurrence of several important life events. A key aim of this study was therefore to understand trends in self-harm rates in this population to explore potential subgroup differences. This is important given increases in self-harm rates in all age groups in England 23 and the higher associated risk of suicide following self-harm in the older population. 22 Secondly, given the likely difference in life events across this broad age span, we aimed to identify potential key differences in the clinical characteristics, preceding life problems and clinical management related to subgroups, given the potential impact this may have on treatment needs and outcomes. 5,8,20,21 Thirdly, we aimed to explore the incidence and factors associated with self-harm repetition by subgroups, as previously this has only been studied in the older adult population as a whole.

| Design
The data used were from the Multicentre Study of Self-Harm in England. 24 Data were collected on individuals who presented to the ED at five general hospitals following self-harm in three study centres: Oxford (one hospital), Manchester (three hospitals) and Derby (one hospital; previously two). Each centre has an established monitoring system to collect data on all presentations of self-harm to the ED at the study hospitals. Data from 1 January 2003 to 31 December 2016 were included.
There are two main routes for data collection: (1) mental health clinicians complete study data capture forms following a psychosocial assessment of self-harm presentations (and ED staff in Manchester complete study data capture forms), (2) members of the research team scrutinise ED record systems to identify all non-assessed selfharm presentations.
The definition of self-harm used within the study is any non-fatal act of intentional self-poisoning and/or self-injury regardless of motivation or degree of suicidal intent. 25 This definition is widely used in UK research, clinical practice and policy. 20 Individuals who present with non-fatal self-harm are included in these data. Those who subsequently died following self-harm while in hospital are also included, however, those who died prior to their arrival at hospital following self-harm are excluded. An act of self-harm which leads to an immediate fatality is identified as suicide and is beyond the immediate definition of self-harm used in the study.
Key study variables include the method of self-harm, which includes self-poisoning (which is the intentional ingestion of any drug in an amount that is more than that prescribed or recommended 26,27 ), self-injury (any injury that is intentionally self-inflicted and may involve self-cutting, hanging or asphyxiation, traffic-related selfinjury and other methods (e.g., drowning)) and both self-poisoning and self-injury. Clinical management variables relate to clinical intervention following the self-harm presentation and is captured by whether a psychosocial assessment was provided, admission to a general hospital bed for physical healthcare and outcomes from the hospital presentation, which includes referral for psychiatric outpatient care (which can be a new referral, or re-referral to community treatment teams individuals are already under), psychiatric inpatient care and General Practitioner (GP) care. Clinical characteristic variables include a history of self-harm, which is identified during the psychosocial assessment, from a previous record of self-harm in the Multicentre Study database or from existing hospital records. 26,27 Psychiatric treatment involves historical or current care, also identified using the same methods as for previous self-harm from, the psychosocial assessment or hospital records and can include psychiatric outpatient care or psychiatric inpatient care. As part of the psychosocial assessment, single or multiple preceding life problems are identified. These may include, for example, problems with relationships, physical health, mental health, housing and finances. 26,27

| Older adult age groups
There are no standard criteria for the classification of individuals as older adults. The lower age threshold in previous research has ranged from 55 to 65 years 16,28,29 In keeping with previous studies from the Multicentre Study of Self-harm in England, and other comparable studies, older adults were defined as individuals aged 60 years and over. 5,6 Two age groups were used: 60-74 years and 75 years and over.

| Rates of self-harm
Rates of self-harm were calculated using an analytic sample where an individual's postcode was recorded as being within the city catch-

| Repetition of self-harm
Repetition was defined as any further presentation for self-harm to a hospital in the same study centre within 12 months of the first recorded presentation in the study period. In this analysis, data up to 31 December 2015 were used (to allow 12-month follow-up until 31 December 2016). A 12-month time period was used for identification of factors associated with repetition as around one in six people repeat self-harm within one year and identifying people who are at risk of early repetition is a key objective of assessment. 31

| Statistical analyses
Descriptive analyses are based on an individual's first recorded presentation ('index presentation') for self-harm to one of the study hospitals within the study period (2003-2016) whilst aged 60 years and over. Basic demographic data (gender, age) and the method of self-harm were available for almost all presentations. Analyses involving clinical characteristics, preceding life problems and clinical management included only presentations that resulted in a psychosocial assessment (or an ED assessment in Manchester), as this information was collected through assessment.
Self-harm incidence rates were calculated using the number of presentations to hospital (numerator) and the mid-year population estimate (denominator) for each calendar year for individuals aged 60 years and over. Incidence rates were calculated by gender and age group. 95% confidence intervals were derived using the Poisson exact method. Poisson regression models were used to estimate changes in self-harm rates over time, with 95% confidence intervals.
Pearson's chi-square tests were used to compare categorical variables (clinical characteristics, preceding life problems and clinical management) by age groups: (60-74 and 75 years and over), and gender. Where tests included more than two categories, standardised residual scores were used to assess which categories were statistically different. All drugs that featured in at least five percent of selfpoisoning presentations are included in Table 3. Missing data were excluded from analyses. The percentage of missing data are presented for each variable and drugs that featured in atleast five percent of self-poisoning presentations are provided in the Tables.
Factors associated with the risk of self-harm repetition within 12 months were investigated using univariable logistic regression.
Four models were generated (all males 60 years and over, all females 60 years and over, all individuals 60-74 years of age and all individuals 75 years and over), with key variables known to be associated with repetition (method of self-harm, alcohol use, current psychiatric treatment, previous psychiatric treatment, previous selfharm, ethnicity, and preceding problems with a relationship, bereavement, mental health and physical health) entered as covariates. 4,6,8 In these models, because many co-variates were available only for those who were assessed, only individuals who received an assessment were included.

| Self-harm methods
Self-poisoning was the most common method of self-harm (81.4%; n = 2185) ( Table 1). This was followed by self-injury (15.5%; n = 416), while 3.1% (n = 83) of presentations involved both self-injury and self-poisoning. Men were more likely to self-injure, while women were more likely to self-poison. Of all presentations that involved self-injury (18.6%; n = 499), cut or stab (58.1%) was the most frequently used method. This was followed by asphyxiation/hanging (13.1%), and other self-injury methods (12.1%), for example, head banging or scratching. There were no differences in self-harm method between the two age groups ( Table 2). In all presentations that involved self-poisoning (84.5%; n = 2268), pure paracetamol was the single most frequently used drug, and its use was similar for men and women (30.4% vs. 28.7%) ( Table 3). Antidepressants (including tricyclic antidepressants) were used less often by men than women (15.0% vs. 18.3%) and were also used less often by those 75 years and over than by 60-74-year-olds (9.7% vs. 19.7%).

| Clinical characteristics
A history of prior self-harm was recorded for 40.9% of individuals (n = 614) (Table 1), with more women than men having previously T A B L E 1 Characteristics of individuals aged 60 years and over who presented to hospital for self-harm, by gender.  T A B L E 2 Characteristics of individuals aged 60 years and over who presented to hospital for self-harm, by age group.   T A B L E 3 Self-injury methods and drugs used in self-poisoning by individuals aged 60 years and over at first presentation to hospital by age group and gender. Method of self-harm     (Table 1). Problems with physical health were more frequent for those aged 75 years and over than in 60-74-year-olds (55.5% vs. 31.9%; χ 2 (1) = 77.77, p < 0.001). Those aged 60-74 years were more likely than those 75 years and over to have problems with alcohol, finances, employment and in their relationships with partners or friends ( Table 2).

| Clinical management
Two-thirds of individuals (65.9%; n = 1768) received a psychosocial assessment following hospital presentation and 76.8% (n = 1261) were admitted to a general hospital bed for medical treatment (Table 1).

| Repetition of self-harm
Overall, 17.6% (n = 436) of individuals had at least one further presentation to hospital with self-harm during the study period, and 10.8% (n = 267) re-presented within 12 months. Of those who were assessed, 18.8% (n = 304) re-presented in the study period, including 11.5% (n = 187) who re-presented within 12 months.
Those aged 60-74 years were more likely to re-present to the ED with self-harm compared with those who were 75 years and   6 Rates in our study increased over time in 60-74-year-old men. Problems with mental and physical health were the main concerns, but problems also differed by gender and age group. More women reported problems with relationships, and problems with finance and alcohol were reported by more men.
For the older age group, problems with physical health often preceded self-harm, whereas multiple life problems were reported by those in the younger age group. Self-harm repetition within 12 months was strongly associated with previous self-harm and receipt of psychiatric treatment at the time of the self-harm presentation. Repetition of self-harm was also associated with problematic alcohol use in men over 60 years, and with physical health problems in both genders aged 75 years and over.
The increase in the self-harm rate from 2008 to 2016 compared to the rate in 2000-2007 in our earlier study, 6 and specifically the increased rate over time for men aged 60-74 years in the present study, may suggest a cohort effect, as an increase in self-harm rate was observed in a previous multicentre study in 40-59-year-old men, particularly after 2008. 32 In the current study, some individuals presenting to hospital with self-harm will have moved from midlife to the 60-74 age group, which has perhaps led to the increase in rates observed from 2012 to 2015 for 60-74-year-old men. Furthermore, between 2003 and 2016, an increase in suicide rates was observed in men aged 60-74 years. 33 The increase in self-harm in 60-74-yearold men may mean that this subgroup is not accessing support which may lead to a potential increase in demand for health services following self-harm. Also, the gender ratio of self-harm rates usually seen in younger age groups was not observed, with the rate higher in men aged 75 years and over than in women of the same age which aligns with previous findings. 6,11 This gender reversal and the increase in rates in 60-74 years olds might be explained by men, including those in older age groups, having lower levels of helpseeking behaviour and consequently being less likely to access health services than do women. 17,34 This is supported by our findings that men over 60 years were less likely to have been receiving psychiatric treatment at the time of self-harm or have received such treatment in the past, despite a similar proportion of men and women having a history of self-harm over the 12 months preceding their selfharm presentation.
Over three-quarters of individuals were admitted to a general hospital bed for medical treatment following self-harm and 22.3% subsequently to a psychiatric inpatient bed, this being more common in those who were 75 years and over. This may be linked to caution being exercised by clinicians because those who are older have an increased associated risk of suicide as well as of repetition of selfharm and also greater mental and physical health needs, requiring more acute care. 20 Other reasons for psychiatric admission could be gaps in provision of psychosocial or home-based support services.
Our recommendations for considering other preventive measures for older adults due to a high rate of referral to primary care following hospital admission for self-harm, 21 including greater care in prescribing psychotropic medication and avoidance of tricyclic antidepressants, given their toxicity when taken in overdose. 21 The proportion of presentations followed by repeat self-harm within 12 months was lower than reported in other studies. 6,21 However, repetition across the whole study period (17.6%) was higher, which may partly be attributed to the availability of longitudinal data. 5,6,8 The current study builds on previous findings by identifying subgroup differences in risk indicators for repetition.
Problematic alcohol use was previously reported as being associated with repeat self-harm in all older adults, 6 but the current findings highlight this as an important indicator for all men in the present population.

| Strengths and limitations
The data covers a 14-year period and although only up to 2016, data are collected across three geographically spread centres in England,

of 17
with differing socio-economic profiles. 32 The hospital catchment areas are largely urban, so the findings may not be fully representative of all areas in the country. Self-harm episodes occurring in the community or other care settings that did not result in an ED presentation are not captured in these data.
The definition of an older adult used was broad as there are no standard criteria. Individuals were identified as belonging to one of two age groups. Whilst there was a clear rationale for use of age groups, the age range used within each subgroup may have influenced the distribution of results.
The findings are largely based on individuals who received a psychosocial assessment which may have biased the findings, but is unlikely to have done so markedly. More assessed individuals represented to hospital with further self-harm, a finding which contrasts with other studies. 35 One explanation may be that psychosocial assessment was more likely in those with more severe and chronic mental and physical health problems. It could also be that those who re-presented had more help-seeking characteristics. It is also possible that having an assessment provides an understanding of support that is available via assessment and treatment, so individuals may re-present to access this support.
A limitation of the data is that individuals who died in hospital following their index presentation are not removed from the database, which may introduce a small bias with regard to the figures for early repetition of self-harm. Longer-term outcomes, including mortality are important to evaluate in this population, although this was the focus of an earlier investigation from the Multicentre Study. 6 A further limitation is that we could not identify self-harm repetition which did not result in a hospital presentation.

| Implications for practice and future research
Appropriate aftercare for older adults who self-harm should involve a comprehensive psychosocial assessment by a mental health specialist. 20 The finding that a third of older adults did not receive the recommended assessment is of concern, especially given the increased risk of suicide associated with self-harm in older age. 35,36 Recently there has been more financial investment in liaison psychiatry services in the UK, which may mean that the provision of assessments will increase. 37 Self-harm related ED presentations appear to be increasing in older females but at a less rapid rate than presentations by men aged 60-74 years. Common problems preceding self-harm in the latter group include difficulties with employment, finance and alcohol misuse. Public health strategies should be designed to enable communities to improve access to psychiatric services as well as to address the multiple socio-environmental factors which are relevant in older age, which can also be important in preventing both selfharm and suicide, 38 but action requires an assertive approach accompanied by systematic change. 39 Referral to voluntary agencies and social prescribing may help reduce loneliness and improve physical and mental wellbeing, 40 which became particularly important in the context of the COVID-19 pandemic. Whilst our study data do not cover this period, it is important to acknowledge how the pandemic disproportionately affected this population both in terms of potential physical health consequences 41 and reduced social and emotional support 42 due to physical distancing and lockdowns.
The strong relationship between self-harm and both physical and mental health problems found in the present study echoes the need for further in-depth exploration of this issue. 43,44 Primary care services can play an important role by improving assessment and treatment of both mental and physical health problems (such as chronic illnesses and pain). Future research exploring the longer-term outcomes (including suicide and other causes of mortality) for older adults who present to hospital following self-harm is also needed.

| CONCLUSIONS
Self-harm in older adults may be increasing, with a rise seen in men aged 60-74 years. The findings indicate that older adults should not be considered as a homogenous group, with key subgroup differences found in this population, including the problems that precede selfharm. Prevention and clinical management should address the range of commonly reported concerns and may require alternatives to prescribed medication, including relationship counselling, help with finances, and guidance on safer alcohol use. All older adults presenting to hospital following self-harm should receive a thorough psychosocial assessment. 20 With the observed increase in self-harm rates, and as the world's population ages, preventive health strategies and clinical management centred on early intervention are recommended, with self-harm in this group likely to require greater focus in clinical policy, practice and research.