Introduction
According to the World Health Organization (WHO), nearly 800,000 people die from suicide each year [
1]. Among them, approximately 90% suffer from a mental illness, with 40–70% suffering from major depressive disorder (MDD) [
2]. In China, its lifetime prevalence is 6.8% [
3]. A comprehensive meta-analysis on Chinese patients with MDD revealed that the lifetime prevalence rates of suicidal ideation, suicide plans, and suicide attempts were 53.1%, 17.5%, and 23.7%, respectively [
4], which were consistent with rates reported in international studies. Compared to patients without suicidal ideation (SI) or suicidal behaviour (SB), patients with MDD with SI and/or SB were at a higher risk of comorbid mental and medical conditions [
5]. They consumed more antidepressants, experienced severe side effects, had poorer drug efficacy and quality of life, and required additional medical resources and expenses [
5]. Meanwhile, their caregivers also bore enormous burdens, including financial and psychological [
6]. Therefore, suicide in patients with MDD represents an extremely massive burden for society and their families. Early identification of suicide risk in patients with MDD and timely and efficacious intervention is an important public health issue.
Suicide in patients with MDD is a highly complex behaviour with individual differences, which can be challenging to predict and identify [
7]. Studies have attempted to find associated risk factors from multiple aspects, such as demographic characteristics [
8,
9] (younger age, male, smoking, unemployment, family history of mental illness and history of suicide attempts), psychosocial factors [
10] (low social support), life stressors [
11] (childhood and adult life adversities), comorbidities [
12,
13] (borderline personality disorder and coronary heart disease), and clinical features [
9,
14] (severity of depressive symptoms, presence of psychotic symptoms and melancholic depression). Although these factors are associated with a higher risk of suicide in MDD, they only provide limited predictive value. Therefore, further research aims to identify objective biological markers related to suicide in depression. Studies identified potential biological markers associated with depression-related suicide, such as interleukin-6 [
15], C-reactive protein [
16], the serotonin transporter gene [
17], brain-derived neurotrophic factor gene [
18] and abnormal DNA methylation levels [
19]. However, these studies are still in the early stages and face certain challenges and limitations. The consistency and replicability of their findings require further verification. Additionally, the discovery of biomarkers necessitates larger-scale investigations.
Managing patients with MDD with suicide ideation and/or behaviour is particularly challenging owing to the difficulty in predicting and detecting suicide, urgency of the condition, need for immediate life-sustaining intervention and paucity of available treatments [
20]. For high-risk patients, conventional antidepressants usually have a slow onset and often taking more than two weeks to achieve efficacy [
21]. In clinical practice, some physical therapy methods, especially electroconvulsive therapy (ECT), are often used to eliminate patients’ suicidal ideation and behaviour. Studies found that ECT effectively reduced the risk of suicide [
22] and might be superior to antidepressants in preventing recurrence of suicide in depressed patients [
23], pending further empirical evidence. However, when considering its impact on cognition[24], some patients and their families had concerns. Ketamine was a blocker of the N-methyl-D-aspartate (NMDA, glutamate-gated ion channel) receptors. Ketamine and its isomer, es-ketamine, reduced the risk of suicide in patients in a short time, as confirmed by relevant studies [
25,
26]. As the primary modalities of long-term intervention, the use of antidepressants [
27], lithium [
28,
29], second general antipsychotics [
30] and psychotherapy (including cognitive behavioural therapy [
31], dialectical behavioural therapy [
32], etc.) alleviated depressive symptoms and reduce the risk of suicide death among patients. However, there is ongoing debate and controversy regarding the effectiveness of the aforementioned treatment modalities in suicide prevention for individuals with depression [
32‐
35].
In summary, suicide is the most serious consequence of depression and requires early identification and prompt intervention. Both domestic and foreign guidelines recommend close monitoring and treatment of patients assessed for suicidal ideation and behaviour and admission to hospital for drug treatment and psychotherapy [
36]. The current treatment strategies for suicidal ideation and behaviour in patients with MDD are unclear. Most randomized controlled trials (RCTs) on drugs and physical therapy for MDD excluded patients with suicidal tendencies for safety reasons [
37], which made the evidence obtained not generalizable to a clinic setting. Starting from disease databases and real-world data, exploring the characteristics of individuals with SI and SB, understanding their disease burden, and seeking effective treatment strategies is crucial. This study aimed to understand and compare the characteristics and disease burden of individuals with MDD with or without suicidal ideation or behaviours in psychiatric hospitals through real-world data. Furthermore, by tracking medication adjustments during the hospitalization period, we hoped to identify the treatment pattern transitions for individuals with suicide to optimize clinical treatment pathways.
Methods
Data source
This descriptive, retrospective cohort study used electronic health records based on the Hospital Information System. The study investigated the clinical characteristics, treatment features and disease burden of patients with major depressive disorder with suicidal ideation or behaviour. The cohort was selected from inpatients at Beijing Anding Hospital Capital Medical University between January 2013 to December 2020. All patients were observed for a one-year follow up period after discharge to explore recurrence (rehospitalization). The inclusion criteria were: (1) patients diagnosed with major depressive disorder (by an experienced attending physician and another senior physician) according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) diagnostic criteria and (2) availability of reliable medical records, without any missing information. Exclusion criteria were: (1) those diagnosed with other mental disorders, such as schizophrenia and bipolar disorder, (2) severe physical diseases, (3) long-term hospitalized patients (>6 months), (4) patients with drug or alcohol dependence or abuse issues and (5) women who were pregnant and/or lactating.
Patient identification
Among the included data, we screened a portion of samples that included complete discharge records with a hospitalization duration of at least > 2 weeks. Based on the presence of suicidal ideation and/or behaviour at admission (extracted from specific suicide-related fields recorded on the day of admission), the enrolled patients were divided into two groups: major depressive disorder with suicidal ideation or behaviour (MDS) and with no suicidal ideation or behaviour (MDNS). Demographic and clinical, detailed treatment and re-hospitalization information within one year were obtained from the database. Furthermore, the clinical characteristics, treatment features and disease burden were analysed between the two groups. In addition, relapse and recurrence of depression were evaluated via the one-year follow-up data after discharge.
Treatment patterns
We included the main psychiatric medications during hospitalization, which included antidepressants (AD), antipsychotics (AP) and mood stabilizers (MS). AD was further divided into SSRI, Serotonin Norepinephrine Reuptake Inhibitor (SNRI), Norepinephrine Dopamine Reuptake Inhibitor (NDRI), Noradrenergic and Specific Serotonergic Antidepressant (NaSSA) and others based on pharmacological mechanisms. Given the severity of patients’ conditions and close monitoring in the hospital setting, there is a high probability of actively modifying medications. Opting for a relatively narrow timeframe can facilitate the identification of modifications in the therapeutic regimen, we defined the observation period as seven days to observe treatment regimens (patterns). If drug A was used continuously in combination with drug B and drugs A and B were used in combination for more than seven days, the treatment regimen was defined as drug A combined with drug B (A + B). If drugs B and A overlapped for less than seven days and the patient mainly used drug B in the next observation period (seven days), it was defined as a drug switch. Both drug combination and switch were considered as a new treatment regimen. Any change to the class of medications was reflected as a new treatment regimen, which included (1) the main medication composition changed and in-class switching was captured as a new regimen, such as an individual on escitalopram who switched to sertraline, or (2) addition of a new medication to an existing regimen, (combination therapy), such as AP and MS added to AD.
Common data model
Data were mapped to standard concepts according to the Observational Medical Outcomes Partnership Common Data Model [
38]. Furthermore, the treatment sequence analysis was performed within the Observational Health Data Sciences and Informatics framework.
Statistical analysis
Statistical Product and Service Solutions version 26.0 (Chicago, IL, USA) was used for statistical analyses. Continuous variables did not conform to the normal distribution and homogeneity of variance. Data were presented as median and quartile ranges. A Kruskal-Wallis H-test was used to compare the differences in numerical variables between the two groups. All categorical variables were expressed as numbers and percentages (%) and compared using a chi-squared test. Two-tailed P values were used for all statistical analyses, and p < 0.05 was considered as statistically significant.
Discussion
Suicide is the most serious adverse outcome for patients with MDD. It requires early identification and timely intervention. However, the treatment of patients with depression who experience suicidal ideation and behaviour remains highly challenging. This study used real-world data to understand the treatment and clinical characteristics as well as disease burden in patients who were depressed and suicidal ideation and behaviour. Patients with major depressive disorder with suicidal ideation and behaviour had unique demographic and clinical characteristics. They required more frequent physical treatments and had a higher relapse rate within six months. Regarding treatment, approximately 20% required three or more treatment courses. Furthermore, over 90% used antidepressant medications alone or in combination, and SSRIs were most commonly used. Regardless of the initial treatment phase or throughout the entire hospitalization period, the combination of antidepressants with antipsychotics or mood stabilizers was frequently prescribed. ECT was considered an effective treatment for suicide treatment. Our study found a higher proportion of patients with suicidal tendencies underwent ECT during treatment period. Regarding treatment pathways, switching to antipsychotics or mood stabilizers in combination with antidepressants was the preferred option, followed by changing the type of antidepressant medication.
Suicide, a complex behaviour, was influenced by multiple interacting factors [
7]. Studies have not comprehensively examined various characteristics among patients, which could result in inconsistent findings [
9]. We found that younger age (≤ 29 years older), being female, and a history of alcohol consumption were associated with a higher likelihood of suicidal ideation or behaviour among patients with depression. According to data from the WHO, suicide was the second leading cause of death among those aged 15–29 years. These people often faced high levels of social stress, experienced emotional volatility, and had limited understanding and coping skills when it came to depression [
39]. Due to a lack of consistent definitions, the correlation between gender and suicide related to depression remains unclear. Studies also found that male individuals with depression had a higher risk of suicide [
9] in the general population. Furthermore, females tended to exhibit suicidal ideation and suicide attempts, whereas males were more likely to die by suicide [
7].
The stress-diathesis model of suicide suggests that childhood abuse experiences and stressful events in adulthood serve as initial and precipitating factors for suicide respectively. Additionally, both are risk factors for the occurrence of depression and subsequent suicidal behaviour [
40]. Increased risk of suicide associated with stress may be related to dysregulation of the hypothalamic–pituitary–adrenal axis [
41], elevated inflammatory levels [
42], enhanced glutamate system function [
43] and other factors [
44]. We analysed the precipitating factors of patients’ current onset and found that patients with depression who experienced stressful life events were more likely to have suicidal ideation or behaviour. Although we did not separate the timing of stress occurrence, tracking patients’ stress history through medical records and analysing laboratory indicators, especially those related to stress, such as inflammatory factors and hormones, would provide additional evidence for identifying depression-related suicide.
Due to the presence of suicide, the treatment for these patients requires more urgent and proactive intervention measures. ECT is a first-line treatment for rapidly relieving suicidal ideation and behaviour in hospitalized patients with MDD [
45]. Multiple studies reported significant improvements in reduction of suicidal ideation, tendencies and behaviour in patients with depression who received ECT [
46,
47]. We also found a higher proportion of patients with suicidal ideation or behaviour received ECT treatment compared to the positive control group with depression. Further analysis revealed that these patients initiated ECT treatment earlier and received it for a longer duration. This suggested that the treatment of depression with suicide was more challenging. Since the cost of ECT treatment was higher compared to medication, these patients bore a heavier disease burden. However, the effectiveness of ECT was not sustained. Some studies suggested an increased risk of suicide after ECT treatment [
48]. Hence, managing patients with suicide risk may require longer and further persistent treatment strategies. Therefore, providing appropriate and effective treatment to these patients is beneficial for improving their mental health condition and reducing risk of suicide.
We observed changes in medication prescription strategies across different regimens. The proportion of patients who used antidepressant monotherapy was low in both groups. More patients received a combination of antipsychotics or mood stabilizers, especially with the AD + AP treatment program, which showed a higher utilization rate across each regimen. The combination of various treatment plans could reflect patients’ clinical characteristics (comorbid psychotic symptoms or paranoid personality), comorbid conditions (anxiety or sleep disorders) and severity and refractoriness of suicidal ideation or behaviour [
20]. Further analysis that combined medication dosage or blood drug concentration with clinical symptoms could provide a better explanation for the rationality of different combination strategies.
In subsequent analyses, we attempted to identify a subgroup of patients who required multiple medication changes and tracked their medication switching patterns to observe changes in the treatment pathway. When the treatment outcomes were unsatisfactory after one round, most patients preferred combination with antipsychotics. This could be attributed to their clinical characteristics and comorbid conditions and also the synergistic effect of combining antipsychotics [
49,
50]. Additionally, for the MDS group, the proportion of combining mood stabilizers over the existing treatment plan was higher than that in the MDNS group. This may be due to the relatively clear efficacy of mood stabilizers (especially lithium) in preventing suicide [
28]. Nevertheless, the proportion of patients who combined mood stabilizers was much lower than those who combined antipsychotics, possibly due to concerns among clinicians regarding the risks of lithium overdose and other safety events [
51]. Furthermore, increasing evidence has suggested that antipsychotics (chlorpromazine, quetiapine, etc.) could reduce the risk of suicide in patients with depression [
30].
Most current treatment guidelines suggest that SSRIs should be the first choice and, in the case of non-response, an antidepressant from a different pharmacological class should be used. There are relatively few data to inform the choice of which antidepressant from which class should be used in non-responders. Some studies have suggested that SNRIs should be the second step choice [
52]. Upon discharge, it is conceivable that patients may not have been commenced on antidepressants, a decision potentially underpinned by considerations of an associated escalation in suicide risk attributable to the utilization of such pharmacological interventions. This is particularly salient in individuals with a predisposition to high impulsivity, where the prescription of antidepressants necessitates an augmented level of clinical vigilance. Additionally, given the observational nature of this real-world study, it cannot be discounted that a subset of patients may have been discharged prior to the initiation of antidepressant therapy.
In the one-year follow-up after discharge, we found that within the first six months after discharge, patients with depression who had suicidal thoughts or behaviours were more likely to experience a relapse. Recent suicide attempts were identified as a high-risk factor for recurrent suicidal ideation or behaviour in patients with depression [
53]. Research demonstrated that approximately 14% of patients had another suicide attempt within six months after the previous attempt [
54], which was similar to the findings of other longitudinal studies that lasted one year or longer [
55]. The logistic regression analysis identified only one variable predicting a potential recurrence of suicide, which was receiving relevant psychological or medication treatment at admission [
56]. Additionally, patients who were hospitalized in a psychiatric specialty hospital were more likely to attempt suicide again after discharge [
56], which suggested a critical period for suicide intervention following discharge.
Our findings have implications for clinical practice. MDD patients with suicidal ideation or behavior display distinct clinical features, such as an earlier age of onset and often presenting with stress events, indicating that they may experience heightened psychological stress and social challenges. MDS patients require more intricate medication regimens, incur higher hospitalization costs, and exhibit higher relapse rates, underscoring the necessity for more comprehensive and long-term treatment approaches. When adjusting treatment plans, the addition of antipsychotic medications or mood stabilizers appears to be the preferred course of action, followed by changing the type of antidepressant prescribed. In essence, this study offers a detailed insight into the clinical profile and treatment requirements of individuals with MDD and suicidal ideation or behavior, providing valuable guidance for enhancing diagnostic and treatment pathways, implementing early interventions, and developing effective treatment strategies in the future.
This study has some limitations. First, due to the early stage of construction and operation of the large database platform at Beijing Anding Hospital, Capital Medical University, data collection was not comprehensive enough to include indicators related to disease severity, previous suicide attempts and cognitive function, which were relevant to the risk of suicide in patients with depression. Further validation can be conducted based on these findings. Second, this study was a single-centre study, which may not have covered all patients’ medical and treatment data, which could have potentially led to research bias when exploring relapse and recurrence. Third, this study did not include the treatment plans used by patients in outpatient and other healthcare institutions after discharge. Therefore, we cannot determine the long-term effectiveness of the treatment plan at discharge, as previous studies reported that nearly 50% of patients with depression with suicidal ideation or behaviour received at least three different treatment plans within one year [
20].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.