Suicide is a sentinel indicator of population mental health and a major public health challenge in South Wales, which reported rates higher than the national average in 2023–2024. 1 Men accounted for 76% of suicides, with the highest rate among those aged 35–44 years. 1 Suicide is the leading cause of death for men <35 years of age in Wales, reflecting economic deprivation, mental health burden and fragmented service provision. 2,3 Wales reports higher suicide rates than England and approaches levels in Scotland, with both nations showing the greatest male:female disparity (Supplementary Table 1, available online at https://doi.org/10.1192/bjb.2026.10297). 1,3–Reference Hoeger, Gutierrez-Munoz, Tisi, Sadullah and Gudgeon6
South Wales Central is one of seven coronial regions in Wales, with a total population of approximately 1.07 million (Supplementary Table 2), and is sociodemographically diverse: Cardiff is young and multicultural, Merthyr Tydfil and Rhondda Cynon Taf retain post-industrial identities and Powys is predominantly rural. Age structures range from Cardiff’s young profile (18.2% <16 years old) to Powys’s older population (27.5% >65 years) (Supplementary Table 2). Cardiff and the Vale have higher earnings and employment than Merthyr Tydfil and Rhondda Cynon Taf (Supplementary Table 3).
Health indicators mirror these patterns (Supplementary Table 4). Cardiff reports the highest levels of good health and life expectancy, whereas Merthyr Tydfil reports the lowest. Women in South Wales live around 4 years longer than men. Mental health services vary, with Cardiff and the Vale offering the widest specialist services and Powys relying on community-based care.
Under the Coroners and Justice Act 2009, all suspected suicides undergo police and coronial enquiry. This study analyses suicides in South Wales Central between 2016 and 2023 to identify trigger factors and structural determinants, to inform regionally tailored prevention strategies and to assess the effects of lockdown and social isolation during the COVID-19 pandemic.
Methods
Data sources and literature review
To contextualise suicide incidence in South Wales, we reviewed publicly available data from the Office for National Statistics, Public Health Wales, Public Health Scotland and the National Centre for Violence Against Women and Girls. 1,4–Reference Hoeger, Gutierrez-Munoz, Tisi, Sadullah and Gudgeon6
A structured literature search was conducted using PubMed (MEDLINE), which indexes more than 36 million citations from biomedical journals, life science publications and online books. The primary search term was ‘suicide’. The initial 132 355 citations were further screened using combinations of the following terms: ‘substance misuse’, ‘drug misuse’, ‘alcohol misuse’, ‘gambling’, ‘debt’, ‘marital relationship’, ‘relationships’, ‘child abuse’, ‘health concerns’, ‘violence’, ‘criminality’, ‘cocaethylene’, ‘COVID’, ‘lockdown’ and ‘social isolation’. The search period was restricted to 1 January 2000 to 31 December 2025. After title and author screening, 1 657 citations were retained for contextual comparison.
Age-standardised suicide rates, gender-specific patterns (gender recorded at birth, consistent with coronial practice) and age-group trends identified in the literature were compared with those extracted from coronial inquest records. Qualitative insights were drawn to support interpretation.
Identification of suicides within the South Wales Central Coroner’s area
Existing evidence consistently shows that Wales has suicide rates similar to or higher than those in England, with a disproportionate burden among men in midlife and in socioeconomically deprived areas. Depression, substance misuse, relationship breakdown, bereavement, physical illness and social isolation are well-established risk factors, and several studies have examined suicide trends during the COVID-19 pandemic. However, few studies have used detailed coronial inquest data from a defined Welsh jurisdiction to characterise proximal trigger factors, toxicology findings, method-specific patterns and service contact across pre-pandemic, pandemic and post-pandemic periods.
Inquest files from the South Wales Central Coroner’s Service were identified using Civica, the electronic case-management system used by most coronial jurisdictions. A search for cases concluded to be suicide between 2016 and 2023 identified 757 individuals (586 males and 171 females), aged 14–97 years.
Four case files (two from 2016 and two from 2018) were unavailable. Although this did not affect annual suicide counts or gender distribution, it limited the availability of detailed information for those cases.
Three individuals recorded as female at birth had undergone gender reassignment during their lifetime; however, coronial records documented their gender as male. Consistent with coronial classification, these individuals were analysed as male.
Coronial investigation process
Under the Coroners and Justice Act 2009, coroners must investigate deaths that are violent, unnatural, of unknown cause or occur in state detention. All suspected suicides therefore undergo police investigation, autopsy and detailed coronial enquiry.
Inquests examine the circumstances leading to death, including method, presence of suicide notes (physical or digital), witness statements, medical and psychiatric history, and relevant social or relational factors. These enquiries provide comprehensive contextual information about the deceased’s well-being and stressors preceding death, supporting both public health learning and family understanding.
Toxicology
Toxicology reports were available for 60% of the suicide deaths between 2016 and 2018 and for 94% between 2019 and 2023. Toxicology may be unavailable when prolonged resuscitation or hospital admission reduces the reliability of post-mortem sampling. Toxicology findings were used to identify alcohol, drug and combined substance use (including cocaethylene) proximal to death.
Data extraction
Information from all available inquest files was extracted into Microsoft Excel for Windows 10 for analysis. Variables included: age at death; gender; method of death; location of death; history of previous suicide attempts; living circumstances; engagement with mental health services; 23 predefined trigger factors; occupation; summary of events preceding death; prescribed medication; and toxicology results. Data were analysed for trends in age, gender, method and trigger factors across the 8-year study period.
Ethical consent was not required, as the data source for this investigation was the records of deceased individuals.
Trigger factors
We identified 23 key trigger factors from the data extraction, with medical records, witness statements and suicide notes (including final text messages) being particularly insightful. Included in the study were factors that commonly occurred across the material reviewed, namely: depression/anxiety; relationship; alcohol dependency and/or misuse; drug dependency or misuse (including prescribed medications); cocaethylene; health concerns; psychotic illness; bereavement; suicide of a family member or close friend; debt; gambling; homelessness; financial concerns; employment issues including unemployment and disputes; violence predominantly within the family setting but also occurring within criminal gangs and prison settings; pending criminal prosecution including prosecutions under the Sexual Offences Act 2003; sexual assault; subject of child abuse; school bullying; examination stress; pointlessness; and gender change.
Less than 5% of cases had undefinable contributing or causative factors. Where more than one trigger factor was recorded, the weight of each factor was not attributed.
Statistical analysis
All data tables were collected, collated, sorted and organised using Microsoft Excel for Windows 10. All analyses were undertaken by direct computation from first principles using the standard test formulae within Excel.
Nominal and ordinal data organised in contingency tables (number of cases referred, inquested, mode of death by suicide, gender, incidence of trigger factors) were analysed using χ2 tests. Simple 2 × 2 contingency tables were corrected according to the Yates continuity correction.
Parametric data used standard parametric tests based on assumptions of underlying normal distributions, using the unpaired Student’s t-test for comparison of means in two sample cases (e.g. comparison of mean ages between males and females) or by one- and two-factor analyses of variance (e.g. comparisons between gender, survey years or alternative inquest conclusions), using formulae with appropriate corrections for unequal n and unequal sample variance.
Given the number of comparisons undertaken in the course of extensive exploratory analyses, no formal corrections are made for multiple comparisons, but caution is applied explicitly where small numbers of significant effects are noted among large numbers of comparisons. In such cases the significance levels are particularly important.
Results
Referrals and inquests across years
Across the 8-year study period, an average of 3753 cases per year were referred to the South Wales Central Coroner’s Service (range 3269–4613). The number of referrals differed very significantly between years (χ2 = 322.04; d.f. = 7; p < 0.001).
Of these referrals, an average of 525 cases per year proceeded to inquest (range 459–743), also showing a highly significant variation across years (χ2 = 118.90; d.f. = 7; p < 0.001).
The proportion of referrals accepted for inquest was 16.65% overall, varying significantly between 14.94 and 20.27% across years (χ2 = 62.12; d.f. = 14; p < 0.001; Supplementary Fig. 1(a) and (b)).
Coroners’ conclusions
A total of 4999 inquests were completed between 2016 and 2023. Twelve standard Home Office conclusions were recorded, 8 of which occurred >120 times (natural causes; accident/misadventure; narrative; alcohol/drug; suicide; industrial; open; road traffic accident); the remaining 5 occurred 4 times or fewer and were not analysed further (Supplementary Fig. 2).
Among the eight more frequent conclusions, the distribution varied significantly across years (χ2 = 407.51; d.f. = 49; p < 0.001). Inspection of annual patterns (Fig. 1) suggests that this heterogeneity was driven by:
-
• a disproportionately high number of deaths attributed to natural causes in 2016;
-
• fewer narrative conclusions in 2018–2019;
-
• a declining proportion of open conclusions;
-
• a progressive increase in alcohol- and drug-related deaths over time.
Inquest outcomes according to standard Home Office categories across the 8 years of the data sample (2016–2023). RTA, road traffic accident.
Analysis of suicide conclusions
A total of 757 deaths were concluded as ‘suicide’ during the study period, with detailed data available for 753 cases. Suicide was substantially more common in males (586 cases) than females (171 cases).
Method of suicide
Hanging was the most common method, accounting for 504 of 757 deaths (Fig. 2(a)).
Deaths by suicide (n = 757) recorded in coroners’ conclusions over the 8 years of the data sample (2016–2023). (a) Incidence of cases by cause of death for the total sample; (b) incidence of cases by gender; (c) percentage of cases by gender. The number of deaths for each conclusion is shown in parentheses above each data bar.
Gender differences were marked (Fig. 2(b) and (c)):
-
• hanging was significantly more common among males (418/594; 70.37%) than females (86/163; 52.8%), a difference in proportions that was very highly significant (χ2 = 17.83, d.f. = 1; p < 0.001);
-
• females were proportionally more likely than males to die by drowning (25.1% v. 8.25%; χ2 = 36.52; p < 0.05) and jumping/fall from heights (8.59% v. 4.21%, χ2 = 5.96; p < 0.05);
-
• all other gender differences occurred too infrequently to permit reliable contingency table analysis.
Suicides across years
Annual suicide counts are shown in Supplementary Fig. 3. Although the absolute number of male suicides was approximately four times higher than female suicides, the male:female ratio did not differ significantly across years (χ2 = 14.47; d.f. = 14; p not significant).
Trigger factors
Of the 757 suicides, 34 had no identifiable trigger factor. For the remaining 720 cases, between 1 and 8 trigger factors were recorded.
The mean number of trigger factors did not differ significantly between males (mean 2.31, s.d. = 1.30; n = 587) and females (mean 2.11, s.d. = 1.23; n = 169; t = 0.167; d.f. = 752; p not significant), although it was slightly higher in men.
Incidence of trigger factors
Twenty-three distinct trigger factors were identified (Fig. 3(a)–(c)). Their incidence varied considerably:
-
• depression was the most common trigger, present in approximately 40% of cases;
-
• this was followed by relationship problems, alcohol dependence or misuse, drug dependence or misuse, health concerns, psychotic illness and bereavement;
-
• the remaining 16 trigger factors each occurred in fewer than 10% of cases.
Occurrence of each suicide trigger factor across all years (2016–2023), subdivided by gender. (a) Incidence of cases for each trigger factor rank ordered by decreasing frequency; (b) incidence of cases of each trigger factor expressed as a percentage of the total number of cases for each gender; (c) mean age at death for each trigger factor, rank ordered by age. SO, sexual offences; §, no females exhibited this trigger factor; n.s., not significant.
Gender differences in trigger factors
Given the fourfold higher incidence of suicide among men than women, incidence of trigger factors was recalculated as the percentage of all male and all female cases (Fig. 3(b)). Most trigger factors showed broadly similar proportions between genders, with common factors such as depression occurring frequently in both groups and rare factors occurring infrequently in both.
Five trigger factors differed significantly by gender (Fig. 3(b)):
-
• higher in men: relationship problems, debt, employment difficulties, violence, and criminal conviction or pending trial
-
• higher in women: subject of child abuse (although rare)
-
• alcohol misuse and school bullying showed trends towards higher incidence among women (p < 0.1), but these did not reach conventional significance thresholds, partly owing to the smaller number of female cases.
Age at death by trigger factor
The mean age at death associated with each trigger factor is shown in Fig. 3(c), stratified by gender and ranked by combined mean age. Formal analysis of variance was not appropriate because individuals could exhibit multiple trigger factors, violating independence assumptions. Nonetheless, clear age-related patterns were observed.
Triggers predominantly affecting younger individuals included:
-
• examination stress
-
• school bullying
-
• cocaethylene toxicity
-
• gender change.
Triggers more common among older individuals included:
-
• health concerns
-
• feelings of pointlessness.
Although several youth-associated triggers appeared to affect young women at slightly younger ages than young men, the only statistically significant gender difference in unpaired t-tests was for cocaethylene toxicity (t = 1.85; d.f. = 37; p < 0.05). This relatively small difference must be further tempered by the absence of any correction for multiple comparisons.
Variation in trigger factors across years
To assess whether trigger-factor incidence varied across survey years, annual patterns were examined for the four most common triggers (Supplementary Fig. 4A–D). For several trigger factors, particularly among women, zero cases occurred in some years (e.g. relationship problems in 2019; drug misuse in 2016 and 2021). This issue became more pronounced for less common triggers and for both genders, limiting the feasibility of formal statistical testing.
The only clear temporal change was the emergence of COVID-19-related triggers from 2020 onwards (Supplementary Fig. 5).
Given the small gender differences and sparse data for some categories, further analyses were conducted with male and female cases combined. Figure 4 shows the mean age at death across years for the seven most common trigger factors (depression, relationship problems, alcohol misuse, drug misuse, mental-health problems, bereavement, and suicide of a friend or family member), as well as two youth-associated triggers (cocaethylene toxicity and violence) and two older-adult triggers (health concerns and pointlessness).
Mean age at death for suicides with the most common trigger factors across all years (2016–2023).
Across the study period, the mean age associated with each trigger factor remained relatively stable. The only notable deviation was a marked drop in the mean age for ‘pointlessness’ in 2000, driven by a single case involving a 15-year-old, which disproportionately influenced the mean within a small group of predominantly older individuals.
Discussion
This study demonstrates that patterns of referral to the Coroner’s Service, and the subsequent progression of cases to inquest, were shaped by legal, organisational and societal developments across the study period. The COVID-19 pandemic exerted a particularly marked influence, contributing to fluctuations in referral volumes and altering the profile of deaths investigated. Reference John, Pirkis, Gunnell, Appleby and Morrissey7 The increase in suicides observed in 2020 aligns with wider evidence of heightened psychological distress during periods of enforced isolation. Reference Pirkis, Gunnell, Shin, Del Pozo Banos, Arya and Aguilar8,Reference Gunnell, Appleby, Arensman, Hawton, John and Kapur9 The earlier decline in suicides among women compared with men may reflect gendered differences in coping strategies, social connectedness and access to informal support, including online networks. These findings highlight the importance of understanding how social isolation interacts with vulnerability during public health crises.
Consistent with national and international evidence, men accounted for the majority of suicides across all years. Hanging was the most common method for both genders, although proportionally more prevalent among men. 1,3,4 The finding that a substantial number of women also used highly lethal methods challenges the persistent assumption that women predominantly choose less violent means. This underscores the need for gender-informed prevention strategies that avoid reliance on outdated stereotypes and instead reflect contemporary patterns of distress and method choice. 10,Reference Oskarsson, Mehlum, Titelman, Isometsä, Erlangsen and Nordentoft11
The analysis of contributory factors illustrates the complex interplay of psychological, relational and social stressors preceding suicide. Reference O’Connor and Nock12,Reference Moran, Chandler, Dudgeon, Kirtley, Knipe and Pirkis13 Depression, often accompanied by anxiety, was the most frequently identified trigger, yet many individuals had been assessed as presenting insufficient acute risk to warrant intervention beyond pharmacotherapy. 10 This echoes long-standing concerns about the limitations of risk prediction and the potential for antidepressant medication to exacerbate suicidal ideation in some individuals. Reference DelPozo Banos, Rodway, Lee, Rouquette, Ibrahim and Lloyd14,Reference Durrani and Deahl15 Relationship breakdown, employment difficulties and substance misuse were also prominent, particularly among men. The role of alcohol in escalating domestic conflict, sometimes culminating in suicide, highlights the need for integrated approaches that address co-occurring mental health and substance-related harm. 10
Among younger adults, the combination of cocaine and alcohol resulting in cocaethylene toxicity was a notable feature, typically occurring alongside other stressors, such as depression or bereavement. Reference Hearn, Rose, Wagner, Ciarleglio and Mash16,Reference Pennings, Leccese and Wolff17 Importantly, there was no evidence of a suicide cluster comparable to the well-publicised events in Bridgend in 2007–2008, suggesting that the cases observed here were driven by individual rather than collective dynamics. Reference Jones, Gunnell, Platt, Scourfield, Lloyd and Huxley18,Reference Alho, Webb, Gutvilig, Niemi, Komulainen and Suokas19
Bereavement emerged as a significant risk factor, particularly following the suicide of a family member, although the gender difference observed did not reach statistical significance. Reference Pitman, Osborn, King and Erlangsen20–Reference Andriessen, Krysinska, Hill, Reifels, Robinson and Reavley23 The profound psychological impact of discovering the body of a loved one, or even learning of their suicide, was evident in several cases. The inclusion of pet bereavement as a meaningful stressor highlights the breadth of emotional losses that can precipitate crisis and the need for services to recognise the diversity of grief experiences.
In older adults, concerns about declining health, loss of dignity and the perceived futility of continued suffering were prominent motivators for suicide. Reference Balit, Erlangsen, Docherty, Turecki and Orri24,Reference Okolie, Dennis, Thomas and John25 These findings resonate with ongoing debates regarding assisted dying legislation, suggesting that some individuals may seek suicide in the absence of lawful alternatives. Reference Steck, Egger and Zwahlen26 The presence of existential distress in a 15-year-old, however, underscores that feelings of pointlessness are not confined to older populations and can emerge across the lifespan.
Taken together, these findings illustrate the multifaceted nature of suicide and the importance of situating individual cases within broader social, clinical and structural environments. They highlight the need for responsive mental health services, nuanced risk assessment and targeted prevention strategies that account for gender, age, bereavement experiences and substance-use patterns. Reference Hoeger, Gutierrez-Munoz, Tisi, Sadullah and Gudgeon6,Reference Niederkrotenthaler, Gunnell, Arensman, Pirkis, Appleby and Hawton27–29 They also demonstrate the need for coronial systems to remain adaptable in the face of evolving public health challenges, ensuring that emerging patterns – such as those associated with COVID-19 or combined substance use – are recognised and addressed promptly. Heightened awareness among coroners of research evidence (including drug company studies) that antidepressants can increase suicide on initial use or in response to an increase or decrease in dosage is necessary. Moreover, in 2003, the UK’s regulatory authority banned the prescription of most antidepressants for those under 18 years of age. Reference Ho30–Reference Boseley32
Strengths
This study benefits from the depth and breadth of coronial inquest data, which provide detailed contextual information rarely available in routine surveillance systems. The inclusion of autopsy reports, toxicology, witness statements, medical records and suicide notes allows for a nuanced understanding of proximal circumstances preceding death.
Limitations
This study acknowledges several limitations. First, coronial data are not collected for research purposes and may vary in completeness. Second, the greatest number of Home Office categories recorded were narrative conclusions, which mainly reflect the more complex cases in which several factors may be present, such as alcohol and drug misuse. Third, some trigger factors may be underreported if not disclosed to or by the family or professionals. Fourth, the observational nature of the study precludes causal inference. Fifth, several coroners undertake inquests within the jurisdiction, which inevitably introduces individual variability on determining the conclusion. Finally, small numbers for some triggers, particularly among females, limited statistical power for subgroup analyses.
Research and clinical implications
Suicide is a complex and multifactorial phenomenon arising from the interaction of individual vulnerability, social context and structural conditions. This study shows how legal and organisational changes, the COVID-19 pandemic and evolving patterns of mental health need shaped both referral pathways to the Coroner’s Service and the characteristics of deaths investigated. Depression, relational instability, substance misuse, the potential for antidepressant medication to cause or exacerbate suicidal thoughts and bereavement emerged as intersecting drivers of risk, alongside clear gender- and age-related differences in method and precipitating factors.
The persistence of highly lethal methods among both men and women challenges assumptions that have historically guided prevention strategies and underscores the need for interventions responsive to contemporary patterns of distress. The prominence of alcohol-related emotion dysregulation, the psychological impact of bereavement, including pet loss, and the existential concerns of older adults highlight gaps in current service provision, particularly in the integration of mental health, social care and substance-use responses.
These findings emphasise the limitations of risk prediction based solely on clinical presentation and the importance of sustained, relationally attuned support across the life course. Strengthening bereavement services, improving responses to relationship and employment crises, and addressing harm associated with combined substance use represent key opportunities for prevention. As medico-legal systems adapt to future public health challenges, ensuring that vulnerable individuals are identified and supported before reaching crisis point remains an urgent priority.
Together with existing literature, these findings reinforce that suicide arises from converging psychosocial and structural stressors rather than isolated risk factors. The prominence of depression, relational instability, substance misuse, bereavement and health-related concerns – combined with the widespread use of highly lethal methods – highlights the need for integrated prevention strategies spanning primary care, mental health services, substance use treatment and social care systems. Gender-specific and age-specific patterns suggest opportunities for targeted interventions, including support for men experiencing relationship or employment crises, trauma-informed responses for individuals with histories of abuse, and enhanced support for older adults facing health decline or existential distress. The emergence of cocaethylene-related deaths underscores the need for public health messaging addressing combined alcohol and cocaine use.
Inquest data, when systematically analysed, can serve as a powerful public health resource, informing real-time surveillance, refining risk assessment and guiding regionally tailored suicide prevention strategies in Wales and comparable settings.
About the authors
Sarah-Jane Richards, HM Coroner (retired), South Wales Central Coroner’s Office, Pontypridd, and Honorary Associate, School of Biosciences, Cardiff University, UK. Stephen B. Dunnett, Professor Emeritus, School of Biosciences, Cardiff University, UK. Graeme D. Hughes, Senior Coroner, South Wales Central Coroner’s Office, Pontypridd, UK.
Supplementary material
The supplementary material is available online at https://doi.org/10.1192/bjb.2026.10297
Data availability
Coronial inquest data may not be publicly shared owing to legal restrictions. Aggregated data may be available from the corresponding author on reasonable request and subject to approval by the Coroner’s Office.
Acknowledgements
We thank C. Everett and L. Roberts for their time in collating reference data.
Artificial intelligence use: the authors made use of the Microsoft 365 Copilot tool without modification to reduce the word count of the article’s introductory paragraphs during manuscript preparation.
Author contributions
S.-J.R.: conceptualisation, data extraction, analysis, writing – original draft; S.B.D.: statistical analysis, data interpretation, writing – review and editing; G.D.H.: methodology, supervision, writing – review and editing.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interest
None.
Open access
eLetters
No eLetters have been published for this article.