Assessing the characteristics of suicidal ideation and self-harm in a national older adult population attending emergency departments across Ireland: cohort study protocol

Abstract

Introduction

Older people (people aged 65 years and older) have high rates of death by suicide, and self-harm is a major risk factor for suicide. While rates of self-harm decrease with age, rates of suicide increase among this age group. The overall aim of this research project is to identify real-life evidence of the characteristics associated with older people who present with self-harm and suicidal ideation to emergency departments in Ireland. In examining the variables associated with self-harm, we may be better able to identify the characteristics of older adults who are at highest risk, including those presenting with high lethality attempts.

Methods and analysis

Our data are a cohort study of older people in Ireland involving two workstreams. The first will use a 5-year cohort of data from the National Clinical Programme for Self-Harm and Suicide-related Ideation (NCPSHI) which comprises over 70 000 presentations. The second workstream will use a 15-year cohort of electronic patient records from the Mater Misericordiae University Hospital (MMUH) comprising over 30 491 presentations (900 aged 65 years and older) to collect more detailed information on characteristics of older people presenting with self-harm and suicidal ideation.

Ethics and dissemination

This study has received full ethical approval. The Clinical Research Ethics Committee of the MMUH approved the MMUH workstream—Reference number: 1/378/2327 TMR. Ethical approval for the NCPSHI workstream has been granted by the University College Dublin’s Office of Research Ethics.

Our findings will be disseminated via peer-review publications and presentations to the scientific community, along with reports for clinicians and policymakers.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study will examine the factors associated with self-harm in older adults, a group associated with higher suicide rates, and will provide data on high-lethality self-harm.

• This study comprises two large datasets, a national dataset comprising over 70 000 presentations of self-harm and suicidal ideation and a local registry of patients attending one emergency department with mental health presentation (n=30 941).

• This study will use existing datasets to identify any differences in self-harm between presentations of older people and those aged under 65.

• The study is limited by the retrospective nature of the evaluation; however, the populations studied are complete clinical samples.

• The national data source is presentation-based and not individual-based and therefore repeat presentations of individuals are not captured, but these data are available from the local registry.

Introduction

NICE (National Institute for Health and Care Excellence) guidelines define self-harm as, “intentional self-poisoning or injury, irrespective of the apparent purpose.”¹ This may include acts with and without suicidal intent, along with those where there may be a degree of ambivalence. Self-harm may serve a variety of functions for the person; however, it is a significant risk factor for future death by suicide.^{2 3} Suicidal ideation and behaviours may be a feature of a number of psychiatric disorders and are an important symptom, indeed a diagnostic criterion of depression.⁴ Self-harm and suicide are associated with mental health problems, with depression being the most prevalent diagnosis among those dying by suicide.⁵

Mental illness, self-harm and suicidal ideation

Suicidal ideation and self-harm are associated with a range of other conditions including depression, adjustment disorder, bipolar disorder, personality disorders and substance use disorders across hospital and community settings.^{6 7} Suicidal ideation is common in depressive disorders, and this is widely iterated in the literature with a prevalence of between 10% and 48%.⁸

Depression is a common mental health disorder in later life. The Irish Longitudinal Study on Ageing (TILDA) showed a point prevalence of clinically significant depressive symptoms of 10% in an Irish community-dwelling population aged 50 years and over.⁹ The prevalence among people in hospital and long-term care greatly exceeds this and may range from 17% to 35%.¹⁰

Older people suicide and self-harm

Although self-harm is more frequent among adolescents and young adults, older adults have the highest suicide rate in almost all countries. This rate is as high as 48.7/100 000 for older men in the USA (over four times the national age-adjusted rate of 11.1/100 000) and 140/100 000 for older men in rural China (older people are defined as aged 65 years and over in the USA and 60 years and over in China).11,13 It has been reported that the suicide rate steadily increases with age among people over the age of 60 years.

There is evidence that rates of suicide attempts, on the other hand, steadily reduce with age. In their study of suicide over the lifespan, Hawton and Harriss found that the ratio of self-harm to suicide varied from 200:1 among teenagers to 10:1 in individuals aged 60 years and older.¹⁴ Even among older adults who died by suicide, the prevalence of self-harm decreased with age. The lifetime prevalence of self-harm was higher in those older adults aged 65–74 years compared with those aged 75–84 years. This study reported that the prevalence of treatment of any psychiatric disorder and social stressors such as legal and financial stressors and relationship problems all decreased with age. In contrast, physical health conditions and bereavement were found to increase with age. This finding was supported by research conducted by Fassberg et al: they reported that a range of physical illnesses, which along with physical disability were significantly associated with suicidality in older people.¹⁵ This study reported that social isolation or reduced connectedness is associated with suicidal ideation, self-harm and death by suicide among older adults.

In addition to the risk factors for self-harm, the additional risk factors for suicidal death in older adults are being widowed, having other mental illnesses, physical illnesses and bereavement.¹⁶ Sinyor et al identified three clusters among ‘the oldest old’ older adults aged 80 years and older who died by suicide: the first cluster comprised those who were married or widowed, the second comprised individuals who were living alone or socially isolated, and the third one comprised those with a diagnosis of dementia or depression.¹⁷ In their 2019 systematic review, Troya et al identified that comorbid physical problems are common, as is increased loneliness, and perceived burdensome ageing along with physical health problems.¹⁸ Likewise, increased risk of repetition has been reported among older adults with a history of self-harm and with previous and current psychiatric treatment.¹⁹

Alcohol use is associated with an increased risk of self-harm and suicide. In their 2016 study on alcohol consumption as a risk factor for suicidal behaviour, Norstrom et al found that alcohol abuse and intoxication are often present in suicidal behaviour.²⁰

In their study on self-harm in older adults, Patel et al²¹ found that alcohol use was associated with self-harm repetition in men.²¹ In Ireland, 52–60% of presentations with self-harm or suicidal ideation were associated with alcohol across all age-groups.²² Troya et al’s 2024 paper on self-harm presentations to emergency departments (EDs) of older adults in Ireland found that alcohol was involved in 30.3% of presentations.²³

COVID-19 and suicide and self-harm

A range of studies have examined the impact of the pandemic on presentations with self-harm.²⁴ To date, there is divergent and relatively sparse data pertaining to the impact of the COVID-19 pandemic on rates of self-harm, and little which pertains to older people in Ireland. Previous pandemics have provided inconsistent data regarding suicide rates; with initial evidence relating to this current pandemic demonstrating no definitive association with suicide.^{25 26} However, establishing an association between the COVID-19 pandemic and these constructs including rates of self-harm has not been clear.²⁷

One early study from Ireland reported an initial decrease in the numbers of presentations with self-harm to the ED.²⁸ This study noted that the presentations of people being seen in hospital were presenting with higher lethality self-harm, and the authors speculate that perhaps people with lower lethality self-harm were less likely to seek medical care in the context of the early stages of the pandemic.

Steeg et al’s 2022 systematic review supported this interpretation, identifying several potential reasons for reduction in presentations during COVID-19 including a change in thresholds for help-seeking, and the increased use of non-statutory supports.²⁹

Studies conducted in France and India also reported a decrease or marginal increase in rates of self-harm, associated with higher lethality of method.^{30 31} In Brazil, Orellana et al reported that although overall suicide rates reduced by 13% in 2020, the rates of suicide in older people increased during the pandemic—especially in certain geographical areas and in older women.³²

Rationale in the Irish context and by this group; knowledge gap

In the Irish context, TILDA has provided valuable context for the prevalence of depressive symptoms older adults in Ireland.⁹ Corcoran et al (2010) was the first major study of self-harm in older adults in Ireland and reported a slightly different pattern to international studies with lower rates of suicide in older people than those reported in international studies.³³ In their recent paper, Troya et al reported important differences in rates of self-harm between older males and females in Ireland.²³ Regarding the impact of the COVID-19 pandemic, one of our studies reported an initial decrease in the numbers of presentations with self-harm and increased lethality.²⁸ This raised the question of whether people with lower lethality self-harm were less likely to seek medical care. Maguire et al reported on data from the National Clinical Programme for Self-harm and Suicide-related Ideation (NCPSHI) in the context of the COVID-19 pandemic, comparing data from 2020 with two pre-COVID years. This study focused on the role of substance use, which was not a prominent factor in those patients aged over 60 years.²²

Potential impact: closing the gap

Given the high risk of suicide in older adults, it is a clinical priority to identify the characteristics of self-harm in older adults, and to understand what clinical, social, geographical and service-related factors are associated with risk of suicidal ideations and behaviours to allow us to develop treatments and services to meet their needs.

Objective and endpoints

The overall aim of the research project is to identify the clinical and sociodemographic variables associated with presentations of older people with self-harm and suicidal ideation to EDs in Ireland. We know that older people have higher rates of death by suicide, and that self-harm is a major risk factor for suicide. In examining the variables associated with self-harm, we may be better able to identify the characteristics of older adults who are at highest risk, including those presenting with high lethality attempts. This will help inform service development and identify gaps where needs are not being met.

Primary objective

The primary objective of this study was to characterise the demographic and clinical features of self-harm presentations among older adults in Ireland.

Secondary objectives

The secondary objectives of this study were as follows: (1) to identify the sociodemographic characteristics of presentations of older people who present to EDs with self-harm and suicidal ideation; (2) to identify the clinical factors associated with ED presentations for self-harm and suicidal ideation in older people; and (3) to identify any differences in self-harm between presentations of older people and those aged under 65.

Methods and analysis

This study will examine the prevalence of self-harm in older people in Ireland and will explore the sociodemographic and clinical covariates of suicidal ideations and behaviours in this older adult cohort. Our study involves two workstreams. We will first use data from NCPSHI for ED presentations due to self-harm and suicidal ideation for a national overview. The National Clinical Programmes for Mental Health was established in 2010 as a joint initiative between HSE Clinical Strategy and Programmes Division and the College of Psychiatrists of Ireland. A working group established in 2010 produced the NCPSHI to ensure that all patients who self-harm or with suicidal ideation and present to the ED are offered a timely, expert biopsychosocial assessment, including a written Emergency Care Plan (ECP), that next of kin or supportive adults are involved in assessment and management, and all patients are followed up and linked to the next appropriate care. Data on all presentations and clinical activities within the Clinical Programme are collected and collated centrally.³⁴

Second, we will collect clinical data from the Mater Misericordiae University Hospital (MMUH: a large Dublin teaching hospital) to allow a more granular assessment of the characteristics of a smaller cohort of older people over a 15-year period who present with suicidal ideation and self-harm, which will include a greater level of detail around clinical, demographic and treatment characteristics than the NCPSHI data.

Design

The first workstream is a cohort study of older people in Ireland using data by the NCPSHI on self-harm and suicidal ideation. The NCPSHI oversees the assessment and treatment of self-harm and suicidal ideation in the acute hospital ED setting. Since 2018, a database including age demographic of presentations has been maintained. This will be a national dataset to provide a perspective for the years 2018–2022, a 5-year cohort, and comprises over 70 000 presentations.

The second workstream will collect clinical data from the MMUH. We will use electronic patient records (EPR) to identify all older adults who presented with self-harm or suicidal ideation from 2008 to 2022 (a 15-year cohort).

Participants

The national anonymised data have been sought from the NCPSHI. This will include the anonymised presentation data from all the participating adult EDs, which comprises 24 of the 26 EDs in Ireland.

No participants can be identified from the dataset transferred from the Data Management Office of the NCPSHI, as the presentation data collected do not include identifiable information of individuals presented.

The second workstream will use EPRs from the MMUH, a 15-year cohort, and comprising over 30 000 referrals for psychiatric review in the ED (both adults and older adults). We will include all patients aged 65 years who present to the ED with suicidal ideation and behaviours. We will exclude those individuals with indirect self-harm (eg, refusal to eat or drink, or self-neglect).

The dataset

Workstream 1

The clinical nurse specialists (CNSs) of the NCPSHI, in addition to their clinical role in conducting assessment both in the ED, have a lead role in coordinating data collection through which the programme outcomes are monitored and evaluated. The NCPSHI ensures regular training for CNSs and their line managers on data collection.³⁴

The Commission on Patient Safety and Quality Assurance (2008) identified clinical audit as an essential component of clinical governance, stating that it, “constitutes the single most important method which any healthcare organisation can use to understand and assure the quality of the service that it provides”.³⁵ In line with these principles, the NCPSHI maintains a standardised database system, which measures a range of key variables and key performance indicators (KPIs) to measure the clinical service provided. The recommended metrics outlined in the Model of Care enable clinicians, teams and the NCPSHI to track progress and inform the effectiveness, quality and efficiency of service provision. These KPIs are collected nationally to assist oversight and governance at the national level. They facilitate the provision of audit and evaluation at local, regional and national levels, and provide prompt feedback to the clinical services. A key to data collection is the availability of IT systems and provision of administrative support and training to collate data in a timely manner at both local and national levels.

The NCPSHI has established an Audit and Research Group for Assessment and Management of Self-Harm in the ED. This group includes representatives from academia and clinical practice. The purpose of the group is to support the development of local, regional and national audit practice within the National Clinical Programme. It also supports the development of local, national and regional research within the NCPSHI.

The NCPSHI supports practice-related research and is mindful of its obligations to ensure that any research conducted is both ethical and respectful of service users and providers. Research may inform and contribute to the improvement of the programme. Research will include service evaluation. The NCPSHI works closely with universities, clinicians and the National Suicide Research Foundation in developing research projects relevant to the NCPSHI.

Family members who have been bereaved by suicide provide regular feedback to the NCPSHI through the Implementation Advisory Group and the Research and Audit Committee. Individual services have completed service user evaluation studies, the results of which have been presented at annual training days for clinical leads and CNSs.

The NCPSHI collects the following data on each presentation attending EDs. The data are wholly anonymised at point of input into the NCPSHI prespecified electronic templates. All people presenting who are seen by the liaison psychiatry or on-call psychiatry service are included in the dataset, with no possibility in identifying individuals, rather their presentations:

1. Sociodemographics.

   • Sex (male, female, non-binary).

   • Age groups.

   • Ethnic background (following the Central Statistics Office ethnicity grouping, and including the Irish Traveller community as a separate ethnicity).

   • Employment status (collected since beginning of 2020).

   • Hospital/hospital group.

2. Type of self-harm act and type of ideation (suicidal or self-harm ideation).

3. Lethality of attempt (collected since beginning of 2020).

With regard to the assessment of lethality, the ‘lethality measure’ in the dataset will be used for lethality. This is a value assigned by the CNS who is entering the data into the database (low/medium/ high), and as such, it is a measure that it is difficult to standardise. In addition, the need for medical/surgical admission or admission to critical care will be used as an additional proxy measure for lethality. The dataset does not include any information on lethality as perceived by the patient.

• 4 1Substances contributing to the presentation (including alcohol or/and drugs).
• 4 2Currently attending mental health services (indicates whether the person is attending a statutory, voluntary or private services).
• 4 3Referred by:

  • Child and adolescent mental health services; adult mental health services, police, general practitioner (GP); self/family/supportive friend; voluntary organisation; other.

• 4 4Follow-up calls for those assessed within 24 hours of assessment.
• 4 5Next referrals:

  • Child and adolescent mental health services, adult mental health services, dialectical behavioural therapy in mental health services, forensic mental health services, GP, home-based treatment team, statutory addiction services, mental health of intellectual disability services, psychiatry of old age, rehabilitation mental health team, social work (other than mental health services), voluntary addiction services, voluntary counselling agency, private counselling agency, multiple referrals, other.

• 4 6Suicide interventions of the programme.

All of these are KPIs of the NCPSHI:

• Biopsychosocial assessment.

• Letter to GP sent within 24 hours.

• Patient given emergency care plan.

• Collateral history obtained.

• Next of kin involvement.

Workstream 2

Data will be extracted from MMUH EPRs from 2008–2022 (15-year cohort). Data will be completely anonymised. A Data Protection Impact Assessment has been completed under the guidance of the Data Protection Officer.

The following data will be extracted from patients identified as presenting with self-harm or suicidal ideation:

1. Sociodemographic factors including age, sex, employment status, educational level, marital status, housing status, forensic history, social isolation, medical history.

2. Clinical.

   1. Diagnosis.

   2. Past psychiatric history—Diagnoses, treatment (psychological, medications, other), hospitalisations.

   3. Self-harm—Number of times, means, treatment required.

   4. Family history of self-harm and suicide.

   5. Substance misuse—Current and past.

   6. Medical treatment required—ED assessment/medical team review/medical admission/critical care admission.

   7. Psychiatric treatment—Assessment/Community Mental Health Team (CMHT) referral/psychiatric admission.

3. Service use.

   1. Number of ED attendances in the past year.

   2. Active to a CMHT or other mental health intervention.

Power calculation

The proposed study is a retrospective cohort analysis of all presentations to the national (workstream 1 n=30 000 over 5 years) and local (workstream 2 n~30 000 over 15 years) services; therefore, a sample size is dictated by the size of these datasets. The numbers above are an approximation of the whole numbers of presentations: older people will be a smaller subgroup, the size of which is yet to be determined.

Statistical analysis

This study will assess the characteristics of suicidality in a national and local sample, using descriptive analysis, such as χ² tests. We will explore any variables (clinical and demographic) associated with suicidal ideation and behaviours using multivariable analysis, which will allow the study to control for confounding variables. SPSS software will be used, and significance level will be set at 0.05, with reporting of effect sizes and CIs to avoid type 2 error.

Similar approaches will be taken in relation to the modelling of datasets 1 and 2. Dataset 1 provides is a near population-level dataset and thus provides a more comprehensive picture of the population, but over a shorter time period and in less depth. Dataset 2 includes a wider variety of variables to explore and spans a longer time period.

In dataset 1, monthly rates of national ED presentations for self-harm and suicidal ideation will be calculated relative to total population of older adults (65+), with adjustment of the denominator for two national centres who did not participate in data collection for some of the timepoints using their ambulance catchment area. In dataset 2, rate of ED presentations for self-harm and suicidal ideation will be calculated relative to total number of ED presentations. Interrupted time series models will be fitted to these data, for example, using Poisson regression models to evaluate the changes over time in the rates of ED presentations in the population for self-harm and suicidal ideation in the Irish population, while accounting for significant interruptions to these rates due to the COVID-19 pandemic. Relevant literature in this field will be consulted to determine optimal approaches for modelling data, taking into consideration possible autocorrelation (correlation in prevalence between present and prior observations) and seasonality. We will perform sensitivity analysis using a range of alternative methods (eg, restricted maximum likelihood, autoregressive integrated moving average, Bayesian approaches) to evaluate robustness of model results. Modelled rates will be displayed graphically, for example, using line plots with model estimated rates and CIs, and numerically in tabular format.

Supervised learning methods (such as logistic regression) with feature selection will be used to select distinguishing features of those who self-harm, high-lethality self-harm and those who engage in suicidal ideations. For example, distinguishing features between those who self-harm and high-lethality self-harm can be explored using logistic regression models. Covariates of interest would include gender, age, clinical treatments. Location variables (eg, urban/rural) can be included as covariates. Random effects will be included for study sites to account for within-site correlation and potentially relevant interaction effects (to be predefined by the investigator) will also be considered (including interaction effects). An appropriate feature selection method (eg, recursive feature elimination, LASSO regression) will be used to select the most distinguishing sociodemographic and/or clinical features between subjects who self-harm and high-lethality self-harm. Robustness of findings will be evaluated through the use of cross-validation or bootstrapping.

Subgroup analyses

The databases will allow us to conduct the following analyses:

1. The overall trends in self-harm and suicidal ideation in older adults nationally from 2018 to 2022 and at a local level, from 2008 to 2022.

2. The role of sex in self-harm and suicidal ideation in older adults: Are the risk factors for women different from those of men?

3. We will examine if there any particular regional differences in patterns of self-harm and suicidal ideation, conducting analyses of urban versus rural settings.

4. We will examine what service-related variables may be associated with patterns of presentation and engagement with services. We will compare the difference among community healthcare organisation (regional) areas. We will compare model 3 and model 4 hospitals, hospitals where the NCPSHI is delivered as part of a consultation-liaison psychiatry service, and control for the level of staffing recommended by the NCPSHI (this recommends 1 CNS/200 presentations per annum, but smaller centres are staffed better for their activity compared with many of the larger inner-city hospitals). A model 4 hospital provides 24/7 ED, acute surgery, acute medicine, critical care, tertiary care and, in many instances, supra-regional care. A model 3 hospital has a 24/7 ED and provides acute medical and surgical care and is equivalent to a district general hospital.³⁶

Patient and public involvement

We have identified two patient and public involvement (PPI) representatives for this research, through the REFOCUS group of the College of Psychiatrists of Ireland, one of whom (MN) is a coauthor to this protocol. The analysis, as a key component of this work, will be mainly conducted by the research team with statistical advice from University College Dublin Centre for Support & Training in Analysis & Research (CSTAR). PPI representatives will input into the interpretation of the findings and to help us consider what the implications will be for patients and carers. When the project reaches the point of write-up and dissemination, the input of the PPI representatives will be especially valuable, and it will be essential to have the patient and carer voice heavily involved at this juncture: for the academic outputs, and especially for the reports for the public and policymakers.

Strengths and limitations

This study has several strengths, the first of which is the large sample size, giving a national overview. The second is the dearth of research into this population and the potential for this study to bridge that gap.

A significant limitation of this study is that data from the NCPSHI are presentation-related rather than individual based and there are therefore individuals who have presented more than once in this cohort. A further limitation is that this study captures only self-harm and suicidal ideation that results in a presentation to an ED and omits those at community level. Another limitation is that this study is retrospective and thus at greater risk of bias.

Ethics and dissemination

The findings will be important at a national and international level. There is very little formal research in Ireland on the characteristics of older people who present to EDs with self-harm and suicidal ideation, despite the high rates of depressive illnesses and the elevated risk of death by suicide in this age group. Given the prevalence of mental disorders in older people and association with self-harm and suicide, this is important on an individual and population level. Improving our understanding of mental illness and associated self-harm in this population will help inform service development and identify gaps where needs are not being met.

It is expected that the findings will be highly relevant to the development of policy for older people. The NCPSHI has very clear standards for the assessment of people of all ages who present to EDs following self-harm or with suicidal ideation, but there are no differences in the standards for older people compared with younger people, although their higher risk profile is discussed in the Model of Care.³⁴ This research has the potential to inform the NCPSHI for older populations, including whether a different assessment or treatment approach may be indicated for this group. Despite the high-risk profile of older people who present following self-harm, they receive little attention in the National Clinical Programme for Older People Part 2: Specialist Mental Healthcare for Older People.³⁷ Indeed, in clinical practice, their care is delivered in acute hospitals by liaison psychiatry teams, which are usually staffed by specialists in working age rather than old age psychiatry.³⁶ We hope that this research will help to inform the relevant Clinical Programmes and to provide a robust evidence base.