Clinical profile of patients with eating disorders attending emergency services after suicide attempt

Abstract

Background

Suicidal behaviour is a major public health issue, particularly among individuals with eating disorders (EDs), who exhibit elevated risks of suicidal ideation and attempts. EDs, including anorexia nervosa and bulimia nervosa, are associated with significant psychological distress and psychiatric comorbidities, contributing to higher suicide rates.

Objective

This study aimed to explore the clinical characteristics and suicidal behaviour in patients with EDs following a suicide attempt. Specifically, we sought to analyse the influence of sociodemographic factors, psychiatric comorbidities, impulsivity, and childhood trauma on suicide-related outcomes.

Methods

A total of 1441 adults were included, of whom 131 (9.1%) had an ED diagnosis. Participants were categorized into ED and no-ED groups. Group comparisons were conducted using χ² and ANOVA tests. Linear and logistic regression models explored the influence of sociodemographic and clinical variables (e.g., comorbidities, impulsivity, trauma) on suicidal ideation intensity, number of suicide behaviours, and medical damage from the index attempt.

Results

The ED group was younger, predominantly female, and had higher educational attainment. Clinically, they showed greater psychopathology, higher rates of non-suicidal self-injury (NSSI), emotional and sexual abuse, and more previous suicide attempts. They also exhibited a higher number of suicidal behaviours (d = 0.70), but only slightly greater ideation intensity (d = 0.28). Regression models identified psychiatric comorbidities, depression, acquired capability, and past attempts as key predictors of ideation; suicide behaviours were associated with ED (binging-purging subtype), comorbidities, psychotic symptoms, NSSI, and prior attempts.

Conclusions

EDs, particularly those with binging-purging patterns, are linked to repeated suicidal behaviour. Suicide risk assessments in ED patients should consider comorbidities, trauma, NSSI, and impulsivity to guide targeted interventions.

Introduction

Suicidal behavior is a complex and multifactorial health concern influenced by various clinical, biological, personality, and social factors. It remains one of the leading causes of suffering, disability, and non-natural death globally, with suicide mortality rates rising in recent years.¹ While suicide attempts are more frequent than completed suicides, the relationship between the two significantly increases the risk of future suicide completion. Importantly, a history of prior suicide attempts is the most critical risk factor for death by suicide.²

This growing burden has made suicide prevention a global health priority. Recognizing the severity of the issue, the World Health Organization (WHO)³ has incorporated suicide prevention into its Comprehensive Mental Health Action Plan 2013–2030, urging member states to implement national strategies aimed at reducing suicide rates. These strategies focus on mental health promotion, early intervention, and establishing support systems for those at risk. This initiative aligns with WHO's General Programme of Work, which advocates for stronger international collaboration in addressing mental health as a crucial aspect of public health. As suicide rates continue to rise, there is an urgent need for preventive strategies that address both suicide attempts and the broader spectrum of suicidal behaviors to mitigate this growing public health crisis.

The peak age for suicide is estimated to be between 15 and 25 years old, which coincides with the age group most at risk for developing eating disorders (EDs).^4,5 EDs are serious and prevalent mental health conditions characterized by abnormal eating behaviors that negatively impact an individual's health, emotions, and overall functioning. These disorders are often marked by an extreme preoccupation with food, weight, and body image, leading to harmful patterns such as severe food restriction, binge eating – where large quantities of food are consumed in a short period with a sense of loss of control – purging behaviors such as self-induced vomiting or the excessive use of diuretics or laxatives, and food rituals such as eating in a specific order or avoiding certain “forbidden” foods. While not all behaviors occur in every individual or across all ED subtypes, these examples highlight the diverse presentations of these disorders.⁶

EDs have the highest mortality rate among mental disorders, driven by both medical complications and deaths by suicide. In addition, they are associated with substantial morbidity related to non-fatal suicidal and self-injurious behaviors, including suicidal ideation, suicide attempts, and non-suicidal self-injury.⁷ Studies have shown that individuals with EDs are at an increased risk of suicide, with a twofold increase in suicidal risk and a 1.8 times higher likelihood of suicide attempts, even when controlling for common comorbidities like depression, anxiety, and substance abuse. According to the literature, 9–36% of individuals with EDs have a history of suicidal behavior.^8,9 However, most research has focused on anorexia nervosa, with fewer studies examining bulimia nervosa and other EDs.¹⁰ This highlights the need for more comprehensive research across the broader spectrum of EDs.

The mortality rate among individuals with anorexia nervosa is 4–14 times higher than in the general population, with most deaths resulting from medical complications related to malnutrition. A meta-analysis concluded that 1 in 5 patients with anorexia who died did so due to suicide, accounting for 25% of total deaths in this disorder.^1,11,12 The literature suggests that anorexia nervosa and suicide attempts are related to factors such as low body mass index, a greater number of past treatments, increased impulsive behaviors, substance abuse, and major depressive disorder. In bulimia nervosa, some studies have found greater psychopathology, a higher number of treatments, and increased impulsivity.^13,14

The relationship between EDs and suicidality is complex and warrants further investigation. Some authors propose a possible predisposition for suicide in EDs, while others suggest that suicidality may precede the onset of EDs. Both EDs and suicide could share mechanisms (biological, psychological) and mediators (such as impulsivity, trauma, comorbidities, or other risk factors) (Smith et al., 2018). Other factors associated with EDs, identified in various meta-analyses and systematic reviews, include sociodemographic aspects, genetic factors, family history, diagnoses and psychopathological features such as ED type, depression and other comorbidities, pain tolerance, interpersonal traits, and childhood abuse or neglect.^1,10

Suicidal behavior is a major concern in individuals with EDs, with suicide attempts being notably prevalent in this population. The relationship between EDs and suicide is multifaceted, with various risk factors influencing the severity of suicidal outcomes. Research indicates that individuals with EDs are at an elevated risk of suicide attempts and ideation, often exacerbated by psychiatric comorbidities like depression, anxiety, and substance abuse.^7,15 Patients with EDs who have attempted suicide require focused attention, as their risk of suicide is heightened compared to the general population.

Outcomes related to suicidal behavior in ED patients, such as the intensity of suicidal ideation, the frequency of suicidal acts, and the degree of medical harm associated with suicide attempts, warrant further exploration. The medical severity of these attempts, often requiring hospitalization, is commonly linked to co-occurring psychiatric conditions, underscoring the need for targeted interventions in this vulnerable group. Additionally, understanding how ED specific factors contribute to suicidal outcomes is crucial for tailoring prevention and treatment strategies.

Potential risk factors for suicidal behavior in ED patients include impulsivity, poor emotion regulation, low self-esteem, and a history of trauma, including sexual, physical, and psychological abuse during childhood.¹ These factors may explain the heightened suicidal ideation and repeated suicide attempts often observed in this population. Furthermore, interpersonal difficulties – such as perceived burdensomeness and thwarted belongingness, as posited by Joiner's Interpersonal Theory of Suicide – may increase suicide risk in ED patients, particularly through habituation to pain caused by repetitive harmful behaviors like purging, bingeing, and self-harm.^1,10

The interaction between ED pathology and suicidal outcomes is critical to understanding the expression of suicidal behavior in this population. EDs, especially those involving self-harming behaviors like purging or excessive exercise, can reduce the fear of death and increase the acquired capability for suicide, leading to more lethal suicide attempts and a greater likelihood of fatal outcomes.^1,16,17 Moreover, EDs often exacerbate underlying psychiatric conditions, such as depression and impulsivity, further contributing to the risk of suicidal behavior.^1,18

Given the high prevalence of suicidal behavior among ED patients, particularly those who have attempted suicide, a thorough assessment of suicidal risk factors is essential in clinical settings. Future research should focus on elucidating how ED-specific characteristics influence the severity of suicidal outcomes and on developing strategies to address these risks more effectively. Longitudinal studies are needed to establish causality between ED pathology and suicidal behavior, considering the complexity of this relationship.¹⁰ Expanding research beyond cross-sectional and retrospective designs will help optimize suicide prevention and treatment for ED patients.

This study aims to (1) compare sociodemographic and clinical profiles between ED and non-ED suicide attempters and (2) identify key predictors of suicidal ideation, behaviors, and medical lethality in ED patients.

Methods

Sample

The sample consisted of 1441 adults (69.9% women; M = 40.98 years, SD = 15.7) drawn from the SURVIVE study.¹⁹ All participants were recruited from the psychiatric emergency departments of eight public, general, university hospitals across Spain: Hospital Clínic de Barcelona, Corporació Sanitària Parc Taulí, Hospital del Mar, Hospital Clínico San Carlos, Hospital Universitario La Paz, Hospital Universitario Araba-Santiago, Hospital Universitario Virgen del Rocío, and Hospital Universitario Central de Asturias. To be included in the study, participants were required to have made a suicide attempt within the last 15 days. Additionally, all participants needed to be able to understand the study procedures and provide informed consent. The exclusion criteria were: (1) inability to give informed consent, (2) lack of fluency in Spanish, and (3) participation in another clinical study that, in the investigator's opinion, might interfere with the objectives of the current study. The study was conducted in accordance with the Declaration of Helsinki and adhered to good clinical practice guidelines. Ethical approval for the study was obtained from the Human Research Ethics Committee at each participating site.

Measures

The participants were assessed by a trained mental health professional (i.e., clinical psychologists or psychiatrists) using a wide battery of clinical tools, within 15 days after the emergency department admission. First, a sociodemographic interview was administered to assess sociodemographic factors (i.e., sex, age, employment status, marital status). Psychiatric diagnoses were established using the Mini-International Neuropsychiatric Interview (MINI),²⁰ version 7.0.2, a structured clinical interview based on DSM-5 criteria.²¹

For the purposes of the analyses, participants with eating disorders (ED) were classified into two subgroups: anorexia nervosa (AN) and binging-purging eating disorders (BP-ED). The BP-ED category included participants meeting DSM-5 criteria for bulimia nervosa and other ED presentations characterised by recurrent binge-eating and/or purging behaviours identified during the MINI interview. These subgroups were defined to differentiate restrictive ED presentations from those involving binge-eating or purging behaviours.

The primary variable was suicidal behavior in patients with ED. The secondary variables were age, depression, psychopathology, symptom type, impulsivity, psychiatric comorbidity, family history of suicide acquired capability and child trauma and abuse.

Some additional tools were delivered: The Brief Symptom Inventory (BSI)²² to assess the psychopathology symptoms according to nine domains (i.e., anxiety, depression, hostility, OCD, paranoid, phobic, psychoticism, sensitivity and somatization); the Barratt Impulsiveness Scale (BIS-11)²³ to measure impulsivity levels.

The Acquired Capability for Suicide Scale-Fearlessness about Death (ACSS-FAD)²⁴ was also carried out to assess levels of the acquired capability for suicide. ACSS-FAD is a 7-item self-report measure which uses a 5-point Likert scale, from 0 (not at all like me) to 4 (very much like me) to measure fearlessness about death and pain tolerance.

The assessment of suicidal ideation and behavior was conducted using the Columbia Suicide Rating Scale (C-SSRS). The C-SSRS is a clinical scale that assesses and classifies suicidal ideation and behavior, including the presence, severity, and intensity of suicidal ideation as well as different types of suicidal behavior and family history of suicide. The outcomes analysed in the present study were: (1) lethality of the suicide attempt, measured as “Actual Lethality/Medical Damage”; (2) number of suicide attempts (first actual attempt vs. more than one attempt); and (3) intensity of suicidal ideation (most severe type of ideation reported by the participants). Medical damage was rated by the clinician according to the physical consequences of the attempt and the level of medical intervention required, following the C-SSRS classification ranging from no or minor medical damage to moderate or severe medical damage requiring medical treatment or hospitalization. The presence of traumatic events and child abuse has been evaluated using the short form of the Childhood Trauma Questionnaire (CTQ-SF).²⁵ It can be divided into five main sub-scores: Emotional abuse, Physical abuse, Emotional neglect, Physical neglect and Sexual abuse.

Data analysis

Participants were classified into two groups: those diagnosed with an eating disorder (ED group, n = 131) and those without an eating disorder (no-ED group, n = 1310). Within the ED group, participants were further categorized into anorexia nervosa (AN, n = 44) and binging-purging eating disorders (BP-ED, n = 87). Descriptive statistics were presented for sociodemographic and clinical factors across groups. Continuous data was summarized using means and standard deviations, and categorical data were expressed as percentages. Group comparisons were conducted using χ² tests for categorical variables and ANOVA for continuous variables, including suicide-related outcomes such as suicidal ideation intensity, number of suicide behaviors, and medical damage from the index attempt. Effect sizes were estimated using Cramer's V for categorical variables and η² for continuous ones. In line with previous recommendations,²⁶ and considering the large sample size, only results with at least medium effect sizes were considered meaningful.

To investigate the contribution of sociodemographic and clinical risk factors, multivariate regression analyses were conducted using generalized linear models (GLM). For the intensity of suicidal ideation and the number of suicidal behaviours, GLMs were estimated and results are presented as odds ratios (OR) with 95% confidence intervals to facilitate interpretation. The medical damage of the index attempt was analysed using logistic regression. Continuous predictors (e.g., age, psychopathology severity, impulsivity) were entered in their original scale. Therefore, odds ratios represent the expected change in the outcome associated with a one-unit increase in the predictor. To assess potential multicollinearity among predictors, variance inflation factors (VIF) were examined. All VIF values were within acceptable ranges, indicating no problematic multicollinearity among the included variables.

To assess the added value of including risk factors, we compared a full model (including all covariates) with an unconstrained model (including only sociodemographic variables). Model fit was evaluated using the Akaike Information Criterion (AIC), with lower values indicating better model performance. Adjusted R² values were reported for continuous outcomes as measures of explained variance, and odds ratios (ORs) with 95% confidence intervals were used to interpret effect sizes and the direction of associations.

All statistical analyses were performed using R software (packages: psych, lavaan, lme4). Prior to regression analyses, the assumptions of normality, homoscedasticity, and independence of residuals were verified. All regression models were adjusted for age and sex.

Results

Data on sociodemographic and clinical factors are displayed in Table 1. Approximately 9% of participants (n = 131) showed a diagnosis of an ED, with the binging-purging subtype being more common (66.4% of attempters with an ED). A higher proportion of female cases with an ED (91.6%) was found in comparison to those without an ED (no-ED group; 67.7% female), χ² (1) = 32.77, p < .01, V = 0.15. Meaningful differences were also found in terms of age, with ED patients being younger (AN: M = 26.95, SD = 8.75; BP-ED: M = 33.84, SD = 12.84) than those from the no-ED group (M = 41.78, SD = 15.6), F = 39.47, p < .01, η² = 0.03. Educational attainment also differed between groups, χ² (1) = 14.47, p < .05, V = 0.07, with a higher proportion of participants from the ED group reporting secondary and tertiary education. Regarding work status, χ² (1) = 27.88, p < .01, V = 0.10, more participants in the ED group were students, whereas more retired participants were found in the no-ED group.

Table 1.

Sociodemographic and clinical features of sample according to the study groups.

	Study groups			Contrast test	ES
	AN	BP-ED	no-ED
n	44	87	1310
Sex at birth (%female)	95.5	89.7	67.7	32.77**	0.15
Age (years)	26.95 (8.75)	33.84 (12.84)	41.78 (15.6)	39.47**	0.03
Education attainment				14.47*	0.07
Less than primary	–	–	3.4
Primary	4.5	11.5	16.5
Secondary	56.8	49.4	51.9
Tertiary	38.6	39.1	28.2
Work status				27.88**	0.10
Unemployed	34.1	30.6	27.6
Active (employed)	34.1	36.5	40.7
Student	25	23.5	10.8
Retired	6.8	9.4	20.9
Number of diagnosed disorders^a	4.43 (1.85)	4.31 (2.13)	1.7 (1.51)	309.9**	0.18
Number of prescribed drugs^b	1.89 (1.26)	2.3 (1.39)	1.83 (1.22)	11.06**	0.01
Psychopathology symptoms^c
Overall severity	2.23 (0.7)	2.16 (0.71)	1.83 (0.81)	19.41**	0.01
Anxiety	2.24 (0.8)	2.19 (0.89)	1.85 (1)	22.12**	0.02
Depression	3.02 (0.96)	3.16 (0.91)	2.65 (1.08)	13.7**	0.01
Somatization	1.84 (0.87)	1.69 (0.9)	1.49 (0.99)	5.89*	0.01
OCD	2.44 (1.04)	2.5 (0.94)	2.13 (1.04)	12.71**	0.01
Sensitivity	2.69 (1.17)	2.41 (1.05)	1.92 (1.13)	25.02**	0.02
Hostility	1.49 (1)	1.63 (1.06)	1.34 (1.06)	6.45*	0.01
Phobic	1.84 (1.01)	1.65 (1.1)	1.33 (1.13)	10.61**	0.01
Paranoid	1.85 (1.03)	1.8 (0.92)	1.63 (1.01)	3.31	0.01
Psychoticism	2.19 (0.9)	2.12 (0.93)	1.71 (0.95)	21.58**	0.02
Impulsivity^d	55.36 (15.18)	58.8 (18.3)	54.27 (17.53)	5.4*	0.01
Reflective function^e
Hypermentalization	0.67 (0.68)	0.5 (0.66)	0.71 (0.73)	6.49*	0.01
Hypomentalization	1.55 (0.77)	1.62 (0.89)	1.27 (0.81)	18.01**	0.01
Acquired capability for suicide^f	20.5 (7.05)	18.98 (5.85)	18.23 (6.58)	2.72	0.01
Traumatic experiences^g
Emotional abuse	15.7 (6.62)	15.76 (6.37)	12.52 (6.5)	25.37**	0.02
Physical abuse	9.63 (5.35)	10.24 (5.76)	9.34 (5.77)	1.92	0.01
Sexual abuse	12.16 (7.62)	11.11 (7.4)	9.15 (6.41)	11.3**	0.01
Emotional neglect	14.98 (4.56)	13.89 (5.48)	13.17 (5.8)	2.7	0.01
Physical neglect	8.07 (3.25)	8.7 (3.45)	8.68 (4.04)	0.07	0.01
Non-suicidal self-injury (%yes)	84.1	62.1	37.8	55.12**	0.14
Number of previous attempts^b	4.74 (4.32)	7.69 (16.04)	3.06 (3.88)	60.77**	0.04

Note. For continuous variables, means and standard deviation (between brackets) are displayed across the groups; the F test is used as a contrast test, and the η² estimate is used as an effect size statistic. For categorical variables, proportion of cases is displayed across the groups; the Pearson's χ² test was performed as a contrast test, and the Cramer's V is used as an effect size statistic. AN: anorexia nervosa; BP-ED: binging-purging disorders; no-ED: group whose patients did not show the diagnosis of an eating disorder; OCD: obsessive–compulsive disorder symptoms; ES: effect size estimate.

a

Diagnoses according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; APA, 2013).

b

This variable was loglinearly transformed to conduct contrast test.

c

Assessed using the Brief Symptom Inventory (BSI).

d

Assessed using the Barrat Impulsiveness Scale (BIS-11).

e

Assessed using the Acquired Capability for Suicide Scale (ACSS).

f

Assessed using the Childhood Trauma Questionnaire (CTQ).

*

p < .05.

**

p < .01.

In terms of clinical factors (see Table 1), we found meaningful differences between groups in overall severity of psychopathology symptoms (F = 19.41, p < .01, η² = 0.01), depression symptoms (F = 13.7, p < .01, η² = 0.01), sensitivity (F = 25.02, p < .01, η² = 0.02), and psychoticism (F = 21.58, p < .01, η² = 0.02). Emotional abuse (F = 25.37, p < .01, η² = 0.02) and sexual abuse (F = 11.3, p < .01, η² = 0.01) were also significantly higher in the ED group. Moreover, the presence of non-suicidal self-injury (χ² (1) = 55.12, p < .01, V = 0.14), and the number of previous attempts (F = 60.77, p < .01, η² = 0.04), were significantly more frequent in the ED groups.

Regarding suicidal behavior outcomes, intensity of ideation showed significant differences between participants, with a higher number of behaviors in the ED group, although the magnitude of these differences was small (t = 2.62, p < .01, d = 0.28). We did find meaningful differences in terms of the number of suicide behaviours, with higher number of behaviours in the ED group (t = 7.77, p < .01, d = 0.70). Data on the intensity of suicidal ideation and the number of behaviours by group are displayed in Fig. 1. No between-group differences were found in terms of the medical damage of the index attempt, χ² (1) = 0.21, p > .05, V = 0.02.

Fig. 1.

Intensity of ideation after a suicide attempt and number of suicide behaviors, according to clinical groups. Note. AN: anorexia nervosa; BP-ED: binging-purging subtype, eating disorders; no-ED: group whose patients did not show the diagnosis of an eating disorder. The bar whiskers represent the standard error of the mean.

No evidence of problematic multicollinearity was observed among the predictors included in the regression models. Model comparison was conducted by exploring AIC values across the generalized regression models. We found lower AIC values for the full model (i.e., model with covariates) across the outcomes, suggesting that including all the aforementioned sociodemographic and clinical risk factors enhances the regression model's ability to explain the outcome. Data on model fit and factor loadings are displayed in Table 2.

Table 2.

Association coefficients of risk factors with suicide-related outcomes.

	Intensity of ideation		Number of suicide behaviors		Medical damage of index attempt
	OR (CI95)	Z	OR (CI95)	Z	OR (CI95)	Z
Sex at birth (ref.: female)	0.73 (0.28, 1.89)	−0.65	0.99 (0.93, 1.04)	−0.46	1.49 (1.04, 2.11)	2.2*
Education attainment (ref.: up to primary)
Secondary	0.88 (0.28, 2.78)	−0.22	0.99 (0.92, 1.05)	−0.43	0.99 (0.64, 1.56)	−0.03
Tertiary	0.7 (0.19, 2.52)	−0.54	1.01 (0.93, 1.09)	0.17	1.16 (0.71, 1.9)	0.58
Employment status (ref.: student/employed)^a	1.45 (0.58, 3.62)	0.8	0.99 (0.94, 1.05)	−0.33	1.58 (1.11, 2.28)	2.5*
Age	1.13 (0.69, 1.87)	0.49	0.97 (0.94, 1)	−2.19*	1.21 (1, 1.48)	1.97*
ED (ref.: no-ED)
AN	1.23 (0.09, 16.47)	0.16	0.98 (0.84, 1.15)	−0.22	1.71 (0.6, 4.23)	1.09
BP-ED	1.43 (0.21, 10.02)	0.36	1.15 (1.02, 1.29)	2.37*	0.97 (0.41, 2.08)	−0.07
Number of psychiatric diagnoses	3.36 (2.03, 5.56)	4.7**	1.03 (1, 1.07)	2.19*	0.97 (0.79, 1.18)	−0.32
Impulsivity^b	0.85 (0.49, 1.49)	−0.57	0.97 (0.93, 1)	−2.03*	1.1 (0.88, 1.36)	0.83
Psychopathology symptom^c	6.08 (3.65, 10.13)	6.95**	1.12 (1.09, 1.16)	7.51**	0.77 (0.63, 0.94)	−2.58**
Traumatic experience^d	1.26 (0.79, 1.99)	0.97	0.98 (0.96, 1.01)	−1.13	1.11 (0.92, 1.34)	1.08
Acquired capability for suicide^e	4.51 (2.95, 6.91)	6.93**	1.02 (1, 1.05)	1.73	1.04 (0.88, 1.23)	0.48
Non-suicidal self-injury (ref.: no)	0.61 (0.23, 1.59)	−1.02	1.16 (1.1, 1.23)	5.08**	1.03 (0.7, 1.51)	0.16
History of previous attempts	2.06 (1.32, 3.22)	3.18**	1.05 (1.03, 1.08)	3.89**	1.13 (0.98, 1.3)	1.84
Model fit index
AIC
Unconstrained	10180.8		2112.61		1032.81
Full model	9658.77		1862.78		1011.41
Radj2	.13		.14		.03

Note. no-ED: group whose patients did not show the diagnosis of an eating disorder; AN: anorexia nervosa group; BP-ED: binging-purging subtype, eating disorder group. AIC: Akaike Information Criterion (the unconstrained model includes only sociodemographic factors, while the full model includes all the covariates). Radj2: adjusted R² (only for the full model). OR: odds ratio. CI₉₅: confidence interval of the OR at 95%. Z: statistic derived from the Wald's test.

For continuous predictors (e.g., age, psychopathology severity, impulsivity), odds ratios represent the expected change in the outcome associated with a one-unit increase in the predictor.

a

This variable was dichotomized to reduce the number of categories (student/employed vs. unemployed/retired).

b

Assessed using the Barrat Impulsiveness Scale (BIS-11).

c

Assessed using the Brief Symptom Inventory (BSI), global severity index.

d

Assessed using the Childhood Trauma Questionnaire (CTQ) total score.

e

Assessed using the Acquired Capability for Suicide Scale (ACSS).

*

p < .05.

**

p < .01.

First, the intensity of ideation outcome was positively associated with the number of psychiatric diagnoses (OR = 3.36, Z = 4.7), the severity of psychopathology symptoms (OR = 6.08, Z = 6.95), the acquired capability for suicide (OR = 4.51, Z = 6.93), and the history of previous attempts (OR = 2.06, Z = 3.18). No significant association was found with ED subtypes.

The number of suicidal behaviors was positively associated with presenting a binging-purging ED (OR = 1.15, Z = 2.37). Additionally, this outcome was associated with the number of psychiatric diagnoses (OR = 1.03, Z = 2.19), psychopathology symptoms (OR = 1.12, Z = 7.51), non-suicidal self-injury (OR = 1.16, Z = 5.08), and the history of previous attempts (OR = 1.05, Z = 3.89). Conversely, the outcome was negatively associated with impulsivity (OR = 0.97, Z = −2.03) and sensitivity symptoms (OR = 0.94, Z = −2.12).

Finally, the medical damage of the index attempt was positively associated with the history of previous attempts (OR = 1.13, Z = 1.84), age (OR = 1.21, Z = 1.97), and work status (unemployed/retired; OR = 1.58, Z = 2.5), and negatively associated with psychopathology symptoms (OR = 0.77, Z = −2.58). The distribution of medical damage categories indicated that most suicide attempts resulted in no or minor medical damage, with fewer cases classified as moderate or severe medical damage. All the association coefficients are displayed in Table 2.

Discussion

The current study investigated suicidal behavior outcomes among individuals with eating disorders (EDs) admitted to psychiatric emergency services following a suicide attempt. The results revealed clear sociodemographic and clinical differences between patients with and without EDs, emphasizing the complex interplay between eating pathology and suicidality. These findings expand the existing literature by identifying how specific factors – such as impulsivity, psychopathological severity, and childhood trauma – may contribute to suicidal vulnerability in this population.

Participants with EDs were significantly younger, predominantly women, and showed higher educational attainment. This is consistent with previous research indicating that EDs predominantly affect young women, often beginning during adolescence and early adulthood, a developmental period characterized by increased vulnerability to both EDs and suicidality.^4,5 Additionally, participants in the ED group were more likely to have attained secondary or tertiary education, suggesting a potential link between higher education and EDs, possibly due to increased societal pressures regarding body image and achievement in these populations.⁶

Clinically, individuals with EDs displayed greater psychopathology, including elevated depression, sensitivity, and psychoticism, as well as higher rates of emotional and sexual abuse. These adverse experiences likely contribute to maladaptive coping strategies such as non-suicidal self-injury (NSSI), which was markedly more prevalent in the ED group and has been consistently associated with greater suicide risk.^1,7 Together, these results highlight how comorbid psychopathology and early trauma may interact to heighten vulnerability to self-destructive behaviors in ED populations.

When examining suicide-related outcomes, we found that participants with EDs exhibited a higher number of suicide behaviors, but not a significantly greater intensity of suicidal ideation compared to those without EDs. This suggests that individuals with EDs may have a higher propensity for engaging in repeated suicidal behavior, possibly due to a greater acquired capability for suicide, which reduces the fear of death and increases the likelihood of lethal attempts.²⁴ However, the intensity of suicidal ideation, although higher in the ED group, did not reach a clinically meaningful effect size, indicating that the frequency of attempts may not always correspond to the severity of suicidal thoughts. The marked prevalence of obsessive–compulsive symptoms in individuals with eating disorders may contribute to explaining this association.²⁷

Contrary to what was expected, there were no significant differences between the ED and non-ED groups in terms of the medical damage from the index of suicide attempt. This finding may be explained by the complex interaction of psychiatric comorbidities, impulsivity, and emotional dysregulation in ED populations, which could influence the lethality of suicide attempts. For instance, impulsivity – commonly elevated in individuals with bulimic or binging-purging behaviors – has been associated with more frequent but less medically severe suicide attempts, as impulsive individuals may act on suicidal urges without premeditation or a strong intent to die.^28,29

Our regression analyses revealed several key risk factors associated with suicidal behavior in ED populations. The intensity of suicidal ideation was positively associated with the number of psychiatric diagnoses, psychopathology severity, acquired capability for suicide, and the history of previous suicide attempts. These findings are consistent with previous research showing that psychiatric comorbidity, particularly depression, is a significant predictor of suicidal behavior in ED patients.^10,30 Moreover, the acquired capability for suicide – reflecting fearlessness about death and higher pain tolerance – was a strong predictor of ideation intensity, supporting the idea that repeated exposure to self-harming behaviors in EDs (e.g., purging, bingeing, or NSSI) may desensitize individuals to the fear of death and increase their risk of suicide.^1,31

Interestingly, impulsivity was negatively associated with suicidal ideation intensity, suggesting that while impulsivity may contribute to a higher frequency of suicide attempts, it may not necessarily increase the severity of suicidal thoughts. This finding aligns with research indicating that impulsive individuals may engage in suicide attempts without experiencing prolonged or intense suicidal ideation, highlighting the importance of assessing impulsivity in suicide risk evaluations.²⁹

The number of suicide behaviors was positively associated with presenting a binging-purging ED, as well as with psychiatric comorbidities, psychopathological severity, psychotic symptoms, NSSI, and the history of previous suicide attempts. These results underscore the multifactorial nature of suicidal behavior in ED populations, with multiple overlapping risk factors contributing to the likelihood of repeated suicide attempts.¹ In contrast, sensitivity symptoms were negatively associated with the number of suicide behaviors, suggesting that individuals with high sensitivity may experience heightened emotional distress without necessarily translating this into suicidal behavior, possibly due to protective factors such as emotional regulation or interpersonal support.

The findings of this study have important clinical implications for the assessment and treatment of suicidal behavior in individuals with EDs. Given the high prevalence of repeated suicide attempts and the significant role of psychiatric comorbidities, clinicians should adopt a comprehensive, multidimensional approach to suicide risk assessment in ED patients. This includes evaluating not only the severity of suicidal ideation but also the number of past attempts, impulsivity levels, and the presence of trauma or NSSI.³² Tailored interventions addressing these specific risk factors, such as emotion regulation therapies for impulsivity and trauma-focused treatments for childhood abuse, may help mitigate the risk of future suicide attempts in this vulnerable population.

Moreover, our study highlights the need for further research on the interaction between ED-specific characteristics (e.g., bingeing, purging, restrictive behaviors) and suicidal outcomes. Longitudinal studies are particularly necessary to establish the temporal relationship between ED pathology and suicidal behavior, as well as to identify potential mediating mechanisms, such as the acquired capability for suicide and emotional dysregulation. Expanding research beyond anorexia nervosa to include other ED subtypes (e.g., bulimia nervosa, binge-eating disorder) will also provide a more comprehensive understanding of the link between EDs and suicidality.¹⁰

This study has several limitations that must be considered. First, the cross-sectional design precludes causal inference regarding the directionality between ED pathology and suicidal behavior. Second, data were collected within 15 days of an emergency admission, which may have influenced participants’ recall and emotional state. Third, the ED sample, although clinically meaningful, represented a smaller subgroup (9%) compared to the full cohort, which may limit statistical power for some comparisons. Finally, while all regression assumptions were verified, unmeasured factors, such as treatment history, personality traits, or medication adherence, might have influenced the associations observed.

Conclusion

In conclusion, this study provides valuable insights into the complex relationship between EDs and suicidal behavior. Individuals with EDs, particularly those with a history of impulsivity, NSSI, and childhood trauma, are at an elevated risk for repeated suicide attempts, underscoring the need for targeted interventions aimed at reducing suicide risk in this population. Future research should continue to explore the underlying mechanisms driving this association, with the ultimate goal of developing more effective prevention and treatment strategies for individuals with EDs who are at risk of suicide.

Funding

This study was supported by the Instituto de Salud Carlos III-FIS research grants (PI19/00236, PI19/00569, PI19/00685, PI19/00941, PI19/00954, PI19/01027, PI19/01256, PI19/01484, PI20/00229 and PI23/00147, PI23/01483, PI23/00822, PI23/00614, PI23/00085, PI23/00707, PI23/01469, PI23/01277, PI23/01367, PI23/01066), co-funded by the European Regional Development Fund (ERDF) “A Way to Build Europe”. IG thanks the support of; the CIBER of Mental Health (CIBERSAM); the Spanish Ministry of Science and Innovation (MCIN) (PI19/00954) integrated into the Plan Nacional de I+D+I and cofinanced by the ISCIII-Subdirección General de Evaluación y Fondos Europeos de la Unión Europea (FEDER, FSE, Next Generation EU/Plan de Recuperación Transformación y Resiliencia_PRTR); the Instituto de Salud Carlos III and the Secretaria d’Universitats i Recerca del Departament d’Economia I Coneixement (2017 SGR 1365), CERCA Programme/Generalitat de Catalunya as well as the Fundació Clínic per la Recerca Biomèdica (Pons Bartran 2022-FRCB_PB1_2022) and a FPU grant (FPU22/04382) from the Spanish Ministry of Universities. The Government of the Principality of Asturias (Ref. IDE/2024/774), “la Caixa” Foundation (Ref. HR23-00421 CaixaResearch Health 2023).

Conflict of interest

Marina Diaz-Marsa reports financial support was provided by Carlos III Health Institute. Victor Perez reports financial support was provided by Carlos III Health Institute. Annabel Cebria reports financial support was provided by Carlos III Health Institute. Iria Grande reports financial support was provided by Carlos III Health Institute. Alejandro de la Torre-Luque reports financial support was provided by Carlos III Health Institute. Angela Palao reports financial support was provided by Carlos III Health Institute. Pilar Alejandra Saiz reports financial support was provided by Carlos III Health Institute. Miguel Ruiz Veguilla reports financial support was provided by Carlos III Health Institute. Pilar Alejandra Saiz reports a relationship with Adamed Pharma S.A. that includes: consulting or advisory. Pilar Alejandra Saiz reports a relationship with Angelini Pharma Spain that includes: consulting or advisory. Pilar Alejandra Saiz reports a relationship with la Caixa Foundation that includes: consulting or advisory. Pilar Alejandra Saiz reports a relationship with Center for Biomedical Research in Mental Health Network that includes: consulting or advisory. Pilar Alejandra Saiz reports a relationship with European Commission that includes: consulting or advisory. Pilar Alejandra Saiz reports a relationship with Government of the Principality of Asturias that includes: consulting or advisory. Pilar Alejandra Saiz reports a relationship with Carlos III Health Institute that includes: consulting or advisory. PAS has been a consultant to and/or has received honoraria or grants from Adamed, Alter Medica, Angelini Pharma, “la Caixa Foundation”, CIBERSAM, Ethypharm Digital Therapy, European Commission, Government of the Principality of Asturias, Instituto de Salud Carlos III, Johnson & Johnson, Lundbeck, Otsuka, Pfizer, Plan Nacional Sobre Drogas, Rovi, Servier, and Viatris España outside the submitted work. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The following are members of SURVIVE: Iñigo Alberdi-Páramo, Natalia Angarita, María Fe Bravo-Ortiz, Guillermo Cano-Escalera, Irene Canosa-García, Alvaro Carrasco-Diaz Fernando Corbalán, Manuel Couce-Sánchez, Javier Curto Ramos, Veronica Fernandez-Rodrigues, Ana García Carpintero, Ainoa García-Fernández, Noelia Iglesias, Elvira Lara, Itziar Leal-Leturia, Maria Purificación Lopez-Peña, Vicenzo Oliva, Luis Olivares, Beatriz Orgaz, Ivan Perez-Diez, Michele de Prisco, Ares Ramos, Pablo Reguera-Pozuelo, Julia Rodríguez Revuelta, Elisa Seijo-Zazo, Lara Suárez-López, Elizabeth Suarez-Soto, Carlos Schmidt, Alba Toll, Mireia Vázquez, Luis Vicente Valor, Eduard Vieta, Iñaki Zorrilla.

The authors would like to acknowledge the support of the research participants, who helped to make this work possible.