Abstract
Background
This meta-analysis aimed to identify risk factors for suicide attempts (SA) in patients with depression to inform clinical practice.
Material/Methods
We searched multiple databases up to January 1, 2025, including MEDLINE and Embase. Case-control and cohort studies reporting risk factors for SA in patients with depression were included. Study quality was assessed using the Newcastle-Ottawa Scale (NOS), and meta-analyses were performed using Rev Man 5.4 software. Results are expressed as odds ratios (OR) and 95% confidence intervals (CI). Heterogeneity was assessed using I2 and P values, and publication bias was evaluated using funnel plots and Egger’s test. The registration information was deposited in the International Register of Systematic Reviews and Meta-Analyses (PROSPERO) trial registry (CRD420251061401; Retrospective registration).
Results
Out of a total of 3792 records, 22 case-control studies were included. The overall heterogeneity (I2) ranged from 0% to 91.9%. Significant risk factors for SA included suicidal ideation (OR=4.98, 95% CI 3.21–7.22), previous hospitalizations (OR=1.38, 95% CI 1.18–1.61), family history of suicide (OR=2.59, 95% CI 1.89–3.57), psychotic symptoms (OR=2.77, 95% CI 1.98–3.88), frequent depressive episodes (OR=2.58, 95% CI 1.58–4.22), self-blame (OR=2.43, 95% CI 1.78–3.31), negative life events (OR=3.77, 95% CI 2.85–5.51), and delusion (OR=3.14, 95% CI 1.99–4.96). Publication bias was detected for family history of suicide and suicidal ideation, but OR values remained significant after correction.
Conclusions
Our findings highlight the need for comprehensive risk assessments and targeted interventions in clinical practice to prevent suicide attempts in patients with depression. Future research should explore the mechanisms and interactions of these risk factors to refine prevention strategies.
Keywords: Depression; Meta-Analysis; Observational Study; Psychotropic Drugs; Receptors, Purinergic P2X7; Risk Factors
Introduction
Depression is a prevalent mental health disorder that significantly impacts individuals’ quality of life and is associated with a high risk of suicide attempts (SA). Major depressive disorder (MDD) is characterized by persistent low mood, loss of interest in activities, and significant impairment in daily functioning [1]. One of the most severe consequences of MDD is the increased risk of SA, which poses an important public health challenge [2]. Identifying and understanding the risk factors associated with SA in patients with MDD is crucial for developing effective prevention strategies and improving patient outcomes. Previous studies have identified various risk factors, including demographic, clinical, and psychosocial factors, which can help in the early identification and management of patients at risk [3,4].
Several studies have highlighted the importance of specific risk factors for SA in patients with MDD. For instance, childhood maltreatment has been identified as a significant risk factor for SA in individuals with major depression [1]. Additionally, clinical factors such as the presence of comorbid psychiatric disorders, severe depressive symptoms, and a history of previous SA have been shown to increase the risk of SA [5,6]. Psychosocial factors, including recent stressful life events and social isolation, also contribute to the risk of SA [7,8]. These findings underscore the complexity of the relationship between depression and SA and highlight the need for a comprehensive approach to risk assessment and management.
Despite the extensive research on risk factors for SA in MDD, there remains a need for a systematic synthesis of the evidence to inform clinical practice and guide future research. Meta-analyses provide a powerful tool for aggregating data from multiple studies, thereby increasing statistical power and providing more robust estimates of risk factors. This meta-analysis aims to systematically review and synthesize the evidence on risk factors for SA in patients with MDD, with a focus on identifying modifiable risk factors that can be targeted in clinical interventions. By providing a comprehensive overview of the current evidence, this study aims to contribute to the development of more effective prevention strategies and improve the management of patients at risk of SA.
While prior reviews identify isolated risk factors, no meta-analysis has quantified the pooled effect sizes of both clinical and psychosocial predictors exclusively in MDD using PRISMA standards, nor addressed publication bias via trim-and-fill methods. Specifically, this review advances beyond existing syntheses by: (1) comprehensively evaluating 11 modifiable and static risk factors within Chinese MDD cohorts – contrasting with reviews focused on single domains (eg., on childhood trauma [1]; on demographic factors [2]); (2) implementing trim-and-fill correction to adjust pooled estimates for publication bias; and (3) providing GRADE evidence profiles to quantify clinical applicability. We hypothesize that modifiable factors (eg, suicidal ideation, untreated psychosis) will demonstrate stronger associations with SA than static risks (eg, family history). Our findings aim to inform targeted risk assessment protocols for MDD populations globally.
Material and Methods
Literature Search Strategy
A comprehensive literature search was conducted in 6 English-language databases (MEDLINE via PubMed, Embase, PsycINFO, Web of Science, and Cochrane CENTRAL) and 3 Chinese-language databases (CNKI, WanFang, VIP), plus 2 trial registries (ClinicalTrials.gov and WHO-ICTRP). Each source was last searched on 1 January 2025. The complete Boolean string for MEDLINE is provided in Supplementary Material 1; analogous strategies were applied to all other databases. We did not perform backward or forward citation chaining and did not hand-search journals, conference proceedings, or other grey-literature sources beyond the 2 registries listed above. No language filters other than English and Chinese were imposed; grey literature and unpublished abstracts were included. No automation tools were used; all records were de-duplicated and screened in EndNote X9.1. We did not employ backward/forward citation searching or hand-searching of conference abstracts or additional grey literature beyond the 2 clinical-trial registries mentioned above.
Literature Inclusion and Exclusion Criteria
The inclusion criteria were: (1) study subjects were patients of any age who had clear diagnostic criteria for depression; (2) risk factors of SA in patients with depression; (3) the type of study was a case-control study or a cohort study. Exclusion criteria were: (1) studies on bipolar depression; (2) duplicate published studies; (3) articles without full text, incomplete information, or data extraction; (4) low-quality literature; (5) non-English/Chinese articles; (6) reviews/case reports; (7) studies with overlapping cohorts. Rationale for language restriction: Because this review targets Chinese MDD populations, we limited languages to English and Chinese to ensure accurate translation and cultural relevance.
Literature Screening and Data Extraction
The extracted information included: study characteristics (first author, publication year, research design, sample size, quality assessment score), participant characteristics (age range, diagnostic criteria used), detailed information related to risk factors for suicide attempts (such as suicidal ideation, previous hospitalization history, suicide family history, psychiatric symptoms, frequent depressive episodes, self-blame, negative life events, delusions, feelings of hopelessness, years of education, anxiety symptoms), and the reported statistical indicators (such as odds ratio, confidence interval). The 2 reviewers compared the extracted data; if the reported information was unclear or incomplete, attempts were made to contact the original study authors for clarification.
Quality Evaluation of Literature
Two reviewers (Shanhong Wang and Yudiao Liang) independently assessed risk of bias using the Newcastle-Ottawa Scale (NOS) across 3 prespecified domains: Selection (representativeness of cases, ascertainment of exposure), Comparability (control for confounders), and Outcome (assessment and follow-up). Each study was rated 0–9; scores 0–3 were considered low quality, 4–6 moderate, and 7–9 high quality. Discrepancies were resolved by discussion with a third reviewer (Youguo Tan) [9,10]. Detailed domain-level risk-of-bias judgments for each study are provided in the Supplementary Table.
Statistical Methods
Rev Man 5.4 software was used for meta-analysis, and the effect size is expressed as an OR value and 95% CI. To determine eligible studies for each synthesis, we categorized them based on the specific risk factors they investigated, such as suicidal ideation, previous hospitalizations, and family history of suicide. We then matched these categories to the planned analyses. Studies were grouped by their focus on particular risk factors to ensure alignment with the research objectives. This method ensured that only relevant studies were included in each synthesis, maintaining the validity and focus of the results. In this meta-analysis, we also converted various statistical measures into a consistent format, such as presenting all effect sizes as odds ratios (ORs) and 95% confidence intervals (CIs). If studies reported multiple measures for a single risk factor, we selected the most relevant and complete data for analysis to ensure comparability and consistency of the results. We present individual study and synthesis results using tables and forest plots. Tables list study characteristics, risk factors, and outcomes. Forest plots show effect sizes (ORs) and 95% CIs for each study and overall, aiding in visual assessment of significance and heterogeneity. If P≥0.1 and I2 ≤50%, the heterogeneity was small, and the fixed-effect model was used for combined analysis. Otherwise, it indicated large heterogeneity and was pooled using the random-effects model. Sensitivity analysis was performed by deleting articles one by one by stata14. P<0.05 was considered statistically significant. We relied on visual inspection of forest plots and the I2 statistic to assess heterogeneity. No specific subgroup analyses or meta-regressions were performed in this meta-analysis. Publication bias was assessed by the funnel plot and Egger’s test. Publication bias existed when P<0.05 was indicated by an asymmetric funnel plot or Egger’s test. If significant publication bias was detected, the clipping and filling method was used to explore the impact of possible missing studies on the pooled estimates. Certainty of evidence was evaluated with GRADEpro GDT following GRADE guidance for prognostic factors[GRADE Working Group]. Evidence started at ‘low’ for case-control designs and could be downgraded for risk of bias, inconsistency, indirectness, imprecision or publication bias; upgrading was not applicable. Summary-of-findings tables were generated for each outcome.
Results
Results of Literature Screening
A total of 3792 articles were retrieved. After removing duplicate articles, 887 articles remained. Title and abstract screening: By reading titles and abstracts, articles that clearly did not meet the inclusion criteria were excluded, resulting in 88 articles remaining. Full text reading: The 88 articles were read in full text, and the articles that did not meet the inclusion criteria were further excluded. Finally, 22 studies were included for both qualitative synthesis and quantitative meta-analysis. Figure 1 provides a detailed description of the entire literature screening process.
Figure 1. PRISMA flow diagram of literature search and study selection.
This diagram details the 4-stage identification process for studies included in the meta-analysis. Initial searches yielded 3792 records (including 1 supplementary source). After duplicate removal (n=2905), 887 records underwent title/abstract screening. Of 81 full-text articles assessed, 59 were excluded with specific methodological reasons documented. Final inclusion comprised 22 case-control studies for both qualitative and quantitative synthesis. Exclusion criteria are categorized per PRISMA standards. PICOS – population, intervention, comparator, outcome, study design; CNKI – China National Knowledge Infrastructure; VIP – China Science and Technology Journal Database. Created with GraphPad Prism 10 (Version 10.1.2, GraphPad Software Inc., San Diego, CA, USA).
Basic Characteristics and Quality Evaluation Results of the Included Literature
A total of 22 articles were included in this study, all of which were case-control studies. The NOS scores of the included literature ranged from 6 to 8, indicating high overall quality. The diagnostic criteria of each study mainly relied on DSM-IV, DSM-5, ICD-10, and some studies used the Chinese version of MINI 5.0 or CCMD-3. The risk factors involved in the study included delusions, increased years of education, comorbid anxiety symptoms, comorbid psychotic symptoms, frequent depressive episodes, self-blame, negative life events, family history of suicide, hopelessness, number of previous hospitalizations, and suicidal ideation, as shown in Table 1.
Table 1. Basic characteristics and quality scores of the included studies.
This table outlines the basic characteristics and quality scores of the included studies. Columns represent the author and year of publication, study type, patient age range, diagnostic criteria used, number of participants in the case and control groups, risk factors assessed, and the Newcastle-Ottawa Scale (NOS) quality score.
| Author (Year) | Study type | Age (range)/criteria | Case group | Control group | Risk factors | NOS score |
|---|---|---|---|---|---|---|
| Wang Yongxue 2005 | Case-control | 16-75/CCMD-3 | 68 | 134 | 1, 6, 7, 8, 9 | 7 |
| Gan Jingli 1995 | Case-control | Not specified/CCMD-2 | 51 | 152 | 1, 6, 7, 8, 9 | 7 |
| Liu Qunfang 2004 | Case-control | 16-75/CCMD-2R | 71 | 150 | 1, 6, 7, 8, 9 | 8 |
| Liao Zhenhua 2014 | Case-control | 19-83/DSM-IV | 51 | 54 | 2 | 6 |
| Wei S 2017 | Case-control | 36±15/DSM-IV | 54 | 73 | 2, 11 | 8 |
| Li Gengliang 2024 | Case-control | 12-22/DSM-IV | 51 | 54 | 2 | 7 |
| Lu Huali 2019 | Case-control | ≥14/DSM-IV | 142 | 2090 | 3, 4, 5, 11 | 6 |
| Wang Hong 2021 | Case-control | 18-65/DSM-IV | 67 | 290 | 3, 11 | 7 |
| Li Xiudi 2021 | Case-control | 18-65/DSM-IV | 82 | 265 | 3, 4, 10, 11 | 7 |
| Shi Yajie 2022 | Case-control | 18-75/DSM-IV | 192 | 166 | 3, 4, 8 | 7 |
| Pan Yue 2022 | Case-control | 41(31,47)/DSM-IV | 95 | 237 | 4, 9, 10 | 8 |
| Xin Limin 2015 | Case-control | 39±12/DSM-IV | 135 | 593 | 5, 10, 11 | 8 |
| Xin Limin 2019 | Case-control | 39±12/DSM-IV | 120 | 506 | 5, 10, 11 | 7 |
| Chen Lin 2013 | Case-control | 16-65/DSM-IV | 989 | 183 | 5, 10 | 8 |
| Zhang Xu 2022 | Case-control | ≥18/DSM-IV | 39 | 400 | 6 | 6 |
| Zhang Yuqi 2007 | Case-control | ≥14/DSM-IV | 215 | 215 | 7, 8, 9 | 8 |
| Wei Bei 2016 | Case-control | 16-80/DSM-IV | 50 | 50 | 9 | 7 |
| Chen Lin 2017 | Case-control | 18-65/DSM-IV | 42 | 135 | 10, 11 | 7 |
| Zhou Xueying 2019 | Case-control | 16-60/DSM-IV | 36 | 114 | 10, 11 | 6 |
| Gao Yuanyuan 2024 | Case-control | 12-18/DSM-IV | 49 | 168 | 11 | 8 |
| Li Zhenyang 2022 | Case-control | 11-25/DSM-IV | 113 | 195 | 11 | 6 |
| Dong Jianie 2019 | Case-control | 11-59/DSM-IV | 28 | 134 | 11 | 6 |
Risk factors are abbreviated as follows: 1. Delusion; 2. Increased years of education; 3. With anxiety symptoms; 4. With psychotic symptoms; 5. Frequent depressive episodes; 6. Self-blame; 7. Negative life events; 8. Family history of suicide; 9. Despair; 10. Previous number of hospitalizations; 11. Suicidal ideation.
Results of Meta-Analysis
Table 2 and Figures 2–12 present the results of the meta-analysis of risk factors for suicide attempts in depressed patients [11–32].
Table 2. Preliminary results of risk factors for suicide attempts in patients with depression.
This table presents the preliminary results of the meta-analysis for risk factors associated with suicide attempts in patients with depression. It includes the number of studies included for each risk factor, results of heterogeneity tests (I2 and P value), the model used (fixed-effects or random-effects), and the meta-analysis results expressed as odds ratios (OR) with 95% confidence intervals (CI), along with the corresponding Z and P values. Risk factors are labeled with their respective numbers for clarity.
| Risk factor | Number of literature | Heterogeneity test | Model | Meta-analysis results | ||||
|---|---|---|---|---|---|---|---|---|
| I2 | P | OR | 95% CI | Z | P | |||
| Delusion | 3 [11–13] | 0 | 0.74 | FE | 3.14 | (1.99–4.96) | 4.91 | 0.00 |
| Increased years of education | 3 [14–16] | 92% | 0.000 | RE | 0.74 | (0.38–1.43) | 0.90 | 0.37 |
| Presence of anxiety symptoms | 4 [17–20] | 86% | 0.000 | RE | 1.42 | (0.56–3.61) | 0.74 | 0.46 |
| Presence of psychotic symptoms | 4 [17,19–21] | 0 | 0.840 | FE | 2.77 | (1.98–3.88) | 5.97 | 0.00 |
| Frequent depressive episodes | 4 [17,22–24] | 81.0% | 0.001 | RE | 2.58 | (1.58–4.22) | 3.79 | 0.00 |
| Self-blame | 4 [11–13,25] | 0 | 0.900 | FE | 2.43 | (1.78–3.31) | 5.64 | 0.00 |
| Negative life events | 4 [11–13,26] | 29% | 0.240 | FE | 3.77 | (2.85–5.51) | 6.84 | 0.00 |
| Family history of suicide | 5 [11–13,20,26] | 0 | 0.590 | FE | 2.59 | (1.89–3.57) | 5.86 | 0.00 |
| Despair | 6 [11–13,21,26–27] | 96% | 0.000 | RE | 3.17 | (1.37–7.35) | 2.69 | 0.00 |
| Previous number of hospitalizations | 7 [19,21–24,28–29] | 65% | 0.009 | RE | 1.38 | (1.18–1.61) | 3.98 | 0.00 |
| Suicidal ideation | 11 [15,17–19,22–23 28–29,30–32] | 69% | 0.00 | RE | 4.98 | (3.21–7.22) | 7.18 | 0.00 |
RE random-effects model; FE fixed-effects model. Fixed-effects models were used when I2 ≤50% and P≥0.1; otherwise, random-effects models were applied.
Figure 2. Forest plot for delusion.
Forest plot for delusion as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Fixed-effects model; no heterogeneity detected (I2=0%).
Figure 3. Forest plot for increased years of education.
Forest plot for increased years of education as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Random-effects model; substantial heterogeneity observed (I2 = 92%).
Figure 4. Forest plot for presence of anxiety symptoms.
Forest plot for presence of anxiety symptoms as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Random-effects model; substantial heterogeneity observed (I2=86%).
Figure 5. Forest plot for presence of psychotic symptoms.
Forest plot for presence of psychotic symptoms as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Fixed-effects model; no heterogeneity detected (I2=0%).
Figure 6. Forest plot for frequent depressive episodes.
Forest plot for frequent depressive episodes as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Random-effects model; substantial heterogeneity observed (I2=81%).
Figure 7. Forest plot for self-blame.
Forest plot for self-blame as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Fixed-effects model; no heterogeneity detected (I2=0%).
Figure 8. Forest plot for negative life events.
Forest plot for negative life events as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Fixed-effects model; low heterogeneity (I2=29%).
Figure 9. Forest plot for family history of suicide.
Forest plot for family history of suicide as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Fixed-effects model; no heterogeneity detected (I2=0%).
Figure 10. Forest plot for despair.
Forest plot for hopelessness (despair) as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Random-effects model; substantial heterogeneity observed (I2=96%).
Figure 11. Forest plot for previous number of hospitalizations.
Forest plot for previous number of hospitalizations as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Random-effects model; moderate heterogeneity observed (I2=65%).
Figure 12. Forest plot for suicidal ideation.
Forest plot for suicidal ideation as a risk factor for suicide attempts in major depressive disorder. The meta-analysis was performed using Review Manager (RevMan) version 5.3 (Cochrane Collaboration, 2020). Random-effects model; moderate heterogeneity observed (I2=69%).
Clinical Features
Suicidal ideation emerged as the strongest modifiable predictor (11 studies, OR=4.98, 95% CI 3.21–7.22, I2=69%). Psychotic symptoms and delusions were similarly associated (OR=2.77 and 3.14, respectively), underscoring the need for early antipsychotic evaluation in severe depression.
Illness Course Indicators
A history of frequent depressive episodes (OR=2.58, 95% CI 1.58–4.22, I2=81%) and previous hospitalizations (OR=1.38, 95% CI 1.18–1.61, I2=65%) marked chronicity and severity as key drivers.
Psychosocial Stressors
Negative life events (OR=3.77, 95% CI 2.85–5.51) and family history of suicide (OR=2.59, 95% CI 1.89–3.57) elevated risk, highlighting the cumulative impact of external stress and genetic vulnerability.
Non-Significant Correlation
Neither the increase in years of education (OR=0.74, 95% CI 0.38–1.43, P=0.37) nor the anxiety symptoms (OR=1.42, 95% CI 0.56–3.61, P=0.46) reached a significant level. High heterogeneity (with I2 values of 92% and 86% respectively) and the narrow educational range in the clinical sample are likely to have masked the true impact; the definition of anxiety varies in different studies (eg, generalized anxiety disorder vs anxiety disorder), which makes a clear comprehensive analysis difficult. Due to the relatively small number of studies included in these 2 risk factors, it is impossible to determine their source of heterogeneity through meta-regression or subgroup analysis. Individual study data and crude effect estimates for each risk factor are shown in Table 3.
Table 3. Individual study data and crude effect estimates for each risk factor.
This table provides the raw 2×2 data and unadjusted effect estimates for every study included in the meta-analysis. For each risk factor, the following are presented: study (author and year), number of cases (suicide attempters) and controls (non-attempters) in whom the risk factor was present, and the corresponding crude odds ratio (OR) with 95% confidence interval (CI). All values are as reported in the original publications; no statistical corrections were applied.
| Risk factor | Study (Author, Year) | Cases | Controls | Crude OR | 95% CI |
|---|---|---|---|---|---|
| Delusion | Wang Yong-xue 2005 | 68 | 134 | 3.78 | 1.51–8.42 |
| Gan Jing-li 1995 | 51 | 152 | 2.56 | 1.49–5.95 | |
| Liu Qun-fang 2004 | 71 | 150 | 3.57 | 1.51–8.41 | |
| Increased years of education | Liao Zhen-hua 2014 | 51 | 54 | 0.39 | 0.16–0.95 |
| Wei S 2017 | 54 | 73 | 1.30 | 1.07–1.57 | |
| Li Geng-liang 2024 | 51 | 54 | 0.61 | 0.48–0.79 | |
| Presence of anxiety symptoms | Lu He-li 2019 | 142 | 2090 | 2.33 | 1.20–4.52 |
| Wang Hong 2021 | 67 | 290 | 0.38 | 0.20–0.71 | |
| Li Xiu-di 2021 | 82 | 265 | 2.20 | 1.02–4.74 | |
| Shi Ya-jie 2022 | 192 | 166 | 2.20 | 1.02–4.74 | |
| Presence of psychotic symptoms | Lu He-li 2019 | 142 | 2090 | 2.49 | 1.45–4.29 |
| Li Xiu-di 2021 | 82 | 265 | 3.74 | 1.46–9.56 | |
| Shi Ya-jie 2022 | 192 | 166 | 2.63 | 1.57–4.42 | |
| Pan Yue 2022 | 95 | 237 | 3.85 | 1.13–13.05 | |
| Frequent depressive episodes | Chen Lin 2013 | 989 | 183 | 2.07 | 1.22–3.49 |
| Lu He-li 2019 | 142 | 2090 | 4.31 | 3.55–5.23 | |
| Xin Li-min 2015 | 135 | 593 | 2.05 | 1.14–3.68 | |
| Xin Li-min 2019 | 120 | 506 | 2.06 | 1.23–3.46 | |
| Self-blame | Wang Yong-xue 2005 | 68 | 134 | 2.99 | 1.31–6.83 |
| Gan Jing-li 1995 | 51 | 152 | 2.15 | 1.28–4.34 | |
| Liu Qun-fang 2004 | 71 | 150 | 2.98 | 1.13–8.64 | |
| Zhang Xu 2022 | 39 | 400 | 2.35 | 1.53–3.62 | |
| Negative life events | Zhang Yu-qi 2007 | 215 | 215 | 6.64 | 2.48–11.04 |
| Wang Yong-xue 2005 | 68 | 134 | 3.84 | 1.67–8.85 | |
| Gan Jing-li 1995 | 51 | 152 | 2.31 | 1.26–4.85 | |
| Liu Qun-fang 2004 | 71 | 150 | 3.84 | 1.67–8.62 | |
| Family history of suicide | Zhang Yu-qi 2007 | 215 | 215 | 2.98 | 1.17–5.04 |
| Wang Yong-xue 2005 | 68 | 134 | 3.63 | 1.54–8.76 | |
| Gan Jing-li 1995 | 51 | 152 | 2.54 | 1.36–5.15 | |
| Liu Qun-fang 2004 | 71 | 150 | 3.63 | 1.54–8.76 | |
| Shi Ya-jie 2022 | 192 | 166 | 1.81 | 1.02–3.21 | |
| Despair | Zhang Yu-qi 2007 | 215 | 215 | 7.68 | 5.75–13.74 |
| Wang Yong-xue 2005 | 68 | 134 | 9.31 | 3.67–23.79 | |
| Gan Jing-li 1995 | 51 | 152 | 2.01 | 1.21–5.02 | |
| Liu Qun-fang 2004 | 71 | 150 | 9.31 | 3.67–23.82 | |
| Wei Bei 2016 | 50 | 50 | 1.38 | 1.06–1.81 | |
| Pan Yue 2022 | 95 | 237 | 0.79 | 0.65–0.96 | |
| Previous hospitalizations | Chen Lin 2017 | 42 | 135 | 1.73 | 1.09–2.74 |
| Xin Li-min 2015 | 135 | 593 | 1.18 | 1.02–1.37 | |
| Xin Li-min 2019 | 120 | 506 | 1.21 | 1.04–1.42 | |
| Li Xiu-di 2021 | 82 | 265 | 1.52 | 1.11–2.07 | |
| Zhou Xue-ying 2019 | 36 | 114 | 3.71 | 1.43–9.67 | |
| Chen Lin 2013 | 989 | 183 | 1.17 | 1.03–1.34 | |
| Pan Yue 2022 | 95 | 237 | 2.23 | 1.45–3.41 | |
| Suicidal ideation | Gao Yuan-yuan 2024 | 49 | 168 | 10.21 | 1.91–54.56 |
| Chen Lin 2017 | 42 | 135 | 3.90 | 1.51–10.09 | |
| Xin Li-min 2015 | 135 | 593 | 3.55 | 2.28–5.54 | |
| Xin Li-min 2019 | 120 | 506 | 2.34 | 1.44–3.79 | |
| Li Zhen-yang 2022 | 113 | 195 | 2.60 | 1.39–4.87 | |
| Lu He-li 2019 | 142 | 2090 | 4.04 | 2.31–7.10 | |
| Wang Hong 2021 | 67 | 290 | 26.66 | 6.30–112.91 | |
| Li Xiu-di 2021 | 82 | 265 | 30.03 | 8.00–112.77 | |
| Zhou Xue-ying 2019 | 36 | 114 | 8.35 | 2.34–29.79 | |
| Wei S 2017 | 54 | 73 | 53.11 | 5.87–480.45 | |
| Dong Jia-ni 2019 | 28 | 134 | 2.50 | 1.42–4.39 |
Sensitivity Analysis
To test the robustness of our findings, we performed a leave-one-study-out sensitivity analysis. When any single included study was removed, the pooled effect estimate remained essentially unchanged, indicating that our results are stable and reliable (Supplementary Material 2).
Publication Bias Analysis
In this study, the publication bias of each risk factor was tested. The results showed that the funnel plot of suicidal ideation was asymmetric, and the Egger test showed P=0.001, indicating the existence of publication bias. After 5 iterations of clipping and supplementing 4 articles, the adjusted odds ratio (OR) was 26.52, which was still significant (P<0.001). Number of previous hospitalizations: funnel plot was asymmetric, Egger test P=0.000, there was publication bias. After 2 iterations of clipping and supplementing 2 articles, the adjusted OR was 3.36, which was still significant (P<0.001). Family history of suicide: funnel plot asymmetry, Egger test P=0.005, there was publication bias. After 4 iterations of clipping and supplementing 2 articles, the adjusted OR was 9.90, which was still significant (P<0.001). The Egger test P values of other factors were all >0.05, and no significant publication bias was shown. The specific results are shown in Table 4, and the funnel plots of each risk factor and the cut-and-filled funnel plots are shown in Supplementary Material 3.
Table 4. Publication bias test results.
This table summarizes the publication bias test results for each risk factor assessed in the meta-analysis. It includes the P value and 95% confidence interval (CI) for each risk factor. A P value <0.05 suggests potential publication bias. The “trim-and-fill” method was used to adjust for potential missing studies, and the adjusted odds ratio (OR) is provided where applicable. GRADE summary of findings-risk factors for suicide attempts in major depressive disorder.
| Risk factor | Pooled OR (95% CI) | Study design | Risk of bias | Inconsistency | Indirectness | Imprecision | Publication bias | Overall certainty (GRADE) |
|---|---|---|---|---|---|---|---|---|
| Suicidal Ideation | 2.84 (2.15–3.52) | Case-control | Serious1 | Not serious2 | Not serious | Not serious | Detected3 | Moderate |
| Previous hospitalizations | 1.22 (1.12–1.31) | Case-control | Serious1 | Not serious2 | Not serious | Serious4 | Detected3 | Low |
| Family history of suicide | 2.33 (1.52–3.13) | Case-control | Serious1 | Not serious2 | Not serious | Serious4 | Detected3 | Low |
| Psychotic symptoms | 2.66 (1.79–3.63) | Case-control | Serious1 | Not serious2 | Not serious | Serious4 | Not detected | Low |
| Frequent Depressive | 2.67 (1.40–3.94) | Case-control | Serious1 | Serious5 | Not serious | Serious4 | Not detected | Very low |
| Negative Life Events | 3.23 (1.85–4.60) | Case-control | Serious1 | Not serious2 | Not serious | Serious4 | Not detected | Low |
| Self-blame | 2.38 (1.57–3.18) | Case-control | Serious1 | Not serious2 | Not serious | Serious4 | Not detected | Low |
| Despair | 1.58 (0.74–2.42) | Case-control | Serious1 | Serious5 | Not serious | Serious4 | Not detected | Very low |
| Delusion | 3.07 (1.42–4.71) | Case-control | Serious1 | Not serious2 | Not serious | Serious4 | Not detected | Low |
| Years of education | 0.78 (0.27–1.29) | Case-control | Serious1 | Serious5 | Not serious | Serious4 | Not detected | Very low |
Downgraded 1 level (serious risk of bias) – all studies are non-randomised case-control.
Not downgraded for inconsistency when I2 ≤50% or when heterogeneity was explained by clinical context.
Downgraded 1 level (publication bias) – Egger’s P<0.05 and asymmetric funnel plot (trim-and-fill applied).
Downgraded 1 level (serious imprecision) – total events <300 or 95% CI crosses OR=1.00.
Downgraded 1 level (serious inconsistency) – I2 >75% and no clear clinical explanation.
GRADE Evidence Profile and Summary Table of Findings
The certainty of evidence for each risk factor is summarized in Table 5. All estimates were downgraded from “low” (case-control design) to “moderate,” “low,” or “very low” owing to risk of bias, imprecision, inconsistency, or publication bias. Table 6 shows the summary of the survey results. The risk-of-bias assessment using the Newcastle-Ottawa Scale for the 22 included case-control studies is presented in Table 7. Overall, the studies demonstrated varying methodological quality, with total scores ranging from 0 to 9 stars across the three core domains of Selection, Comparability, and Outcome.
Table 5. GRADE summary of findings-risk factors for suicide attempts in major depressive disorder.
This table presents the GRADE assessment for each risk factor synthesized in the meta-analysis. For every outcome we report the pooled odds ratio (OR) with 95% confidence interval, study design, and judgments on risk of bias, inconsistency, indirectness, imprecision, and publication bias. The overall certainty level (high, moderate, low, or very low) is derived from the GRADE framework starting with “low” for case–control evidence and applying predefined downgrades.
| Risk factor | Study (Author, Year) | Cases | Controls | Crude OR | 95% CI |
|---|---|---|---|---|---|
| Delusion | Wang Yong-xue 2005 | 68 | 134 | 3.78 | 1.51–8.42 |
| Gan Jing-li 1995 | 51 | 152 | 2.56 | 1.49–5.95 | |
| Liu Qun-fang 2004 | 71 | 150 | 3.57 | 1.51–8.41 | |
| Increased years of education | Liao Zhen-hua 2014 | 51 | 54 | 0.39 | 0.16–0.95 |
| Wei S 2017 | 54 | 73 | 1.30 | 1.07–1.57 | |
| Li Geng-liang 2024 | 51 | 54 | 0.61 | 0.48–0.79 | |
| Presence of anxiety symptoms | Lu He-li 2019 | 142 | 2090 | 2.33 | 1.20–4.52 |
| Wang Hong 2021 | 67 | 290 | 0.38 | 0.20–0.71 | |
| Li Xiu-di 2021 | 82 | 265 | 2.20 | 1.02–4.74 | |
| Shi Ya-jie 2022 | 192 | 166 | 2.20 | 1.02–4.74 | |
| Presence of psychotic symptoms | Lu He-li 2019 | 142 | 2090 | 2.49 | 1.45–4.29 |
| Li Xiu-di 2021 | 82 | 265 | 3.74 | 1.46–9.56 | |
| Shi Ya-jie 2022 | 192 | 166 | 2.63 | 1.57–4.42 | |
| Pan Yue 2022 | 95 | 237 | 3.85 | 1.13–13.05 | |
| Frequent depressive episodes | Chen Lin 2013 | 989 | 183 | 2.07 | 1.22–3.49 |
| Lu He-li 2019 | 142 | 2090 | 4.31 | 3.55–5.23 | |
| Xin Li-min 2015 | 135 | 593 | 2.05 | 1.14–3.68 | |
| Xin Li-min 2019 | 120 | 506 | 2.06 | 1.23–3.46 | |
| Self-blame | Wang Yong-xue 2005 | 68 | 134 | 2.99 | 1.31–6.83 |
| Gan Jing-li 1995 | 51 | 152 | 2.15 | 1.28–4.34 | |
| Liu Qun-fang 2004 | 71 | 150 | 2.98 | 1.13–8.64 | |
| Zhang Xu 2022 | 39 | 400 | 2.35 | 1.53–3.62 | |
| Negative life events | Zhang Yu-qi 2007 | 215 | 215 | 6.64 | 2.48–11.04 |
| Wang Yong-xue 2005 | 68 | 134 | 3.84 | 1.67–8.85 | |
| Gan Jing-li 1995 | 51 | 152 | 2.31 | 1.26–4.85 | |
| Liu Qun-fang 2004 | 71 | 150 | 3.84 | 1.67–8.62 | |
| Family history of suicide | Zhang Yu-qi 2007 | 215 | 215 | 2.98 | 1.17–5.04 |
| Wang Yong-xue 2005 | 68 | 134 | 3.63 | 1.54–8.76 | |
| Gan Jing-li 1995 | 51 | 152 | 2.54 | 1.36–5.15 | |
| Liu Qun-fang 2004 | 71 | 150 | 3.63 | 1.54–8.76 | |
| Shi Ya-jie 2022 | 192 | 166 | 1.81 | 1.02–3.21 | |
| Despair | Zhang Yu-qi 2007 | 215 | 215 | 7.68 | 5.75–13.74 |
| Wang Yong-xue 2005 | 68 | 134 | 9.31 | 3.67–23.79 | |
| Gan Jing-li 1995 | 51 | 152 | 2.01 | 1.21–5.02 | |
| Liu Qun-fang 2004 | 71 | 150 | 9.31 | 3.67–23.82 | |
| Wei Bei 2016 | 50 | 50 | 1.38 | 1.06–1.81 | |
| Pan Yue 2022 | 95 | 237 | 0.79 | 0.65–0.96 | |
| Previous hospitalizations | Chen Lin 2017 | 42 | 135 | 1.73 | 1.09–2.74 |
| Xin Li-min 2015 | 135 | 593 | 1.18 | 1.02–1.37 | |
| Xin Li-min 2019 | 120 | 506 | 1.21 | 1.04–1.42 | |
| Li Xiu-di 2021 | 82 | 265 | 1.52 | 1.11–2.07 | |
| Zhou Xue-ying 2019 | 36 | 114 | 3.71 | 1.43–9.67 | |
| Chen Lin 2013 | 989 | 183 | 1.17 | 1.03–1.34 | |
| Pan Yue 2022 | 95 | 237 | 2.23 | 1.45–3.41 | |
| Suicidal ideation | Gao Yuan-yuan 2024 | 49 | 168 | 10.21 | 1.91–54.56 |
| Chen Lin 2017 | 42 | 135 | 3.90 | 1.51–10.09 | |
| Xin Li-min 2015 | 135 | 593 | 3.55 | 2.28–5.54 | |
| Xin Li-min 2019 | 120 | 506 | 2.34 | 1.44–3.79 | |
| Li Zhen-yang 2022 | 113 | 195 | 2.60 | 1.39–4.87 | |
| Lu He-li 2019 | 142 | 2090 | 4.04 | 2.31–7.10 | |
| Wang Hong 2021 | 67 | 290 | 26.66 | 6.30–112.91 | |
| Li Xiu-di 2021 | 82 | 265 | 30.03 | 8.00–112.77 | |
| Zhou Xue-ying 2019 | 36 | 114 | 8.35 | 2.34–29.79 | |
| Wei S 2017 | 54 | 73 | 53.11 | 5.87–480.45 | |
| Dong Jia-ni 2019 | 28 | 134 | 2.50 | 1.42–4.39 |
Table 6. Summary of findings.
This table summarizes the certainty of evidence and clinical implications for each risk factor identified in the meta-analysis. For every outcome we report the number of contributing studies, study design (CA – case-control), pooled odds ratio (OR) with 95% confidence interval, overall GRADE certainty rating (high, moderate, low, or very low), and a concise recommendation for clinical practice derived from the evidence.
| Risk factor | No. of studies | Study design | Pooled OR (95% CI) | Certainty (GRADE) | Implications |
|---|---|---|---|---|---|
| Suicidal ideation | 11 | CA | 2.84 (2.15–3.52) | ⊗⊗⊗○ Moderate | Screen routinely; strongest modifiable predictor |
| Previous hospitalizations | 7 | CA | 1.22 (1.12–1.31) | ⊗⊗○○ Low | Ensure post-discharge follow-up |
| Family history of suicide | 5 | CA | 2.33 (1.52–3.13) | ⊗⊗○○ Low | Offer family psycho-education |
| Psychotic symptoms | 4 | CA | 2.66 (1.79–3.63) | ⊗⊗○○ Low | Consider combined antidepressant-antipsychotic therapy |
| Frequent depressive episodes | 4 | CA | 2.67 (1.40–3.94) | ⊗⊗○○ Very low | Monitor closely; relapse prevention strategies |
| Negative life events | 4 | CA | 3.23 (1.85–4.60) | ⊗⊗○○ Low | Provide targeted psychosocial interventions |
| Self-blame | 4 | CA | 2.38 (1.57–3.18) | ⊗⊗○○ Low | Integrate cognitive restructuring in therapy |
| Delusion | 3 | CA | 3.07 (1.42–4.71) | ⊗⊗○○ Low | Screen for psychotic features; prompt antipsychotic evaluation |
| Despair (hopelessness) | 6 | CA | 1.58 (0.74–2.42) | ⊗○○○ Very low | Evidence uncertain; further validation needed |
| Years of education (higher) | 3 | CA | 0.78 (0.27–1.29) | ⊗○○○ Very low | No clear association; not actionable for risk stratification |
CA – case-control. GRADE Working Group grades of evidence high certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
Table 7. Risk-of-bias assessment for each included study (Newcastle-Ottawa Scale).
This table details the Newcastle-Ottawa Scale ratings for all 22 case-control studies across the 3 core domains: Selection (cases, controls, exposure), Comparability (confounder control), and Outcome (assessment, follow-up). Stars (⋆) denote low risk of bias; dashes (—) indicate high or unclear risk. The total score (0–9 stars) classifies each study as high (7–9), moderate (4–6), or low (0–3) quality.
| Study (Author, Year) | Selection domain | Comparability domain | Outcome domain | Overall score | |||
|---|---|---|---|---|---|---|---|
| 1. Representativeness of cases | 2. Selection of controls | 3. Ascertainment of exposure | 4. Control for key confounders | 5. Assessment of outcome | 6. Adequacy of follow-up | (/9 stars) | |
| Wang Yongxue 2005 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Gan Jingli 1995 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Liu Qunfang 2004 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆⋆ | 8 |
| Liao Zhenhua 2014 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 6 |
| Wei S 2017 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 8 |
| Li Gengliang 2024 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 7 |
| Lu Huali 2019 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 6 |
| Wang Hong 2021 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Li Xiudi 2021 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Shi Yajie 2022 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Pan Yue 2022 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆⋆ | 8 |
| Xin Limin 2015 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 8 |
| Xin Limin 2019 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Chen Lin 2013 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 8 |
| Zhang Xu 2022 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 6 |
| Zhang Yuqi 2007 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 8 |
| Wei Bei 2016 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Chen Lin 2017 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 7 |
| Zhou Xueying 2019 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 6 |
| Gao Yuanyuan 2024 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | 8 |
| Li Zhenyang 2022 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 6 |
| Dong Jianie 2019 | ⋆ | ⋆ | ⋆ | ⋆ | ⋆ | — | 6 |
‘⋆’ – low risk of bias (star awarded); ‘—’ – no star/unclear or high risk. Overall score: 0–3=low quality; 4–6=moderate quality; 7–9=high quality.
Discussion
Suicidal ideation, previous hospitalization history, family history of suicide, psychotic symptoms, and frequent depressive episodes were found to be significantly associated with the risk of SA. These findings reveal the complexity of suicide risk in patients with MDD and point to key directions for clinical intervention and future research.
Suicidal ideation was confirmed to be a strong predictor of SA. Several studies have shown that MDD patients who experience suicidal ideation are more likely to attempt suicide than those who do not. This finding is consistent with the study of Pawlak et al [4], who pointed out that suicidal ideation is a precursor of suicidal behavior and reflects the severe psychological distress and hopelessness of patients. Similarly, Sokero et al [10] found that almost all MDD patients who attempted suicide had suicidal ideation, and its severity was directly related to suicidal behavior. Clinicians should prioritize the assessment and management of suicidal ideation in patients with MDD to prevent it from progressing to attempted suicide. A history of previous hospitalization was associated with an increased risk for suicide. Shang et al showed that MDD patients hospitalized for mental health problems had a higher risk of SA [3]. This may be related to the severity of depressive episodes requiring hospitalization or may reflect inadequate continuity of care after discharge. Therefore, providers should ensure that a comprehensive care plan is in place for patients with a history of hospitalization, including follow-up support and relapse prevention strategies. Family history of suicide is also an important risk factor for SA [25,27]. MDD patients with a family history of suicide may be more susceptible due to genetic, environmental, and psychological factors [7]. Studies have shown that a family history of suicide may increase risk through a genetic predisposition to mental disorders, exposure to suicide-related trauma within the family, and potential “normalization” of suicidal behavior, and prevention strategies targeting this population should consider the interplay of these complex factors. The presence of psychotic symptoms significantly increases the risk of suicide in MDD patients [3,33]. The study by Seo et al showed that these symptoms can exacerbate the severity of depressive symptoms and functional impairment, thereby increasing the likelihood of suicide [34]. Early identification and treatment of psychotic symptoms are essential to reduce the risk of suicide in this population, and antipsychotic medications may be required in combination with antidepressant medications and in combination with psychotherapy that targets psychotic experiences. Frequent depressive episodes are associated with an increased risk of suicide, highlighting the impact of chronicity and severity of the illness. Liu et al [35] showed that patients who experience multiple depressive episodes can have long-term distress and functional impairment, which increases suicide vulnerability. Studies have shown that the cumulative burden of multiple depressive episodes can lead to feelings of being trapped and hopeless, which are strongly associated with suicidal behavior. Clinicians should closely monitor patients with a history of multiple episodes and consider strategies to stabilize mood, prevent relapses, and enhance resilience. The same number of studies [12,36] found that hopelessness had a high value in predicting depression suicide, and the more severe the degree of hopelessness, the greater the probability of suicide.
This study found that the effect size of negative life events was relatively high (OR=3.23), while Goldberg et al [1] emphasized that childhood abuse was the core factor of long-term suicide risk. Years of education and anxiety symptoms: increased years of education and anxiety symptoms did not show a significant correlation after being combined in this study (P>0.05), which was partially contradictory to the conclusion of Wei et al [14], who found that low education level was related to SA [19]. It is possible that the sample included in this study was mainly clinical patients, with little difference in education, and the heterogeneity of anxiety symptoms (such as generalized anxiety disorder and panic disorder) was not sufficiently distinguished.
This meta-analysis has key limitations. (1) Exclusive case-control design prevents establishing causality. (2) Retrospective registration of the protocol (PROSPERO CRD420251061401) may have introduced post-hoc analytical flexibility. (3) English/Chinese language restriction may have caused bias and excluded relevant studies. (3) No backward/forward citation searching or hand-searching was performed, which could have missed additional studies. (4) Insufficient studies (<4 per factor) precluded exploring heterogeneity via meta-regression/subgroup analyses. (5) Self-reported outcomes are susceptible to recall/social desirability bias. (6) Significant heterogeneity likely stems from population/methodological differences. (7) Publication bias exists for some factors. (8) Variable study quality (lower NOS scores) may affect reliability. (9) Generalizability is limited by specific populations and unmeasured confounding. Given the observational nature of the evidence and the noted limitations, these findings should inform, but not dictate, clinical decision-making until confirmed by prospective cohorts. (10) GRADE certainty levels (Table 5) ranged from moderate to very low, reinforcing that these observational findings should inform but not dictate clinical decisions until confirmed by prospective cohorts.
Based on moderate-certainty evidence, we recommend routine screening for suicidal ideation and psychotic symptoms, combined antidepressant-antipsychotic treatment when indicated, and targeted psychosocial interventions for recent negative life events, and family psycho-education in patients with a family history of suicide. To advance the field, prospective multi-center cohorts with standardized measures are required to clarify temporal relationships, explore subgroup effects (sex, age, treatment-resistance), and validate culturally adapted prediction models. Future improvements at the review level, such as prospective protocol registration, broader language coverage, and meta-regression, will further strengthen the certainty of our conclusions.
Conclusions
While this meta-analysis of 22 case-control studies identified suicidal ideation, previous hospitalizations, and psychotic symptoms as potential markers of suicide attempts in people with major depressive disorder, the evidence is observational, heterogeneous, and limited by retrospective registration and publication bias. Consequently, these findings should inform, not dictate, clinical risk assessment. Prospective cohorts and individual-patient data analyses are needed to validate and refine suicide prevention strategies.
Footnotes
Conflict of interest: None declared
Registration and Protocol: This systematic review was registered retrospectively in the International Prospective Register of Systematic Reviews (PROSPERO: CRD420251061401). The protocol can be accessed at https://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD420251061401.
Declaration of Figures’ Authenticity: All figures submitted have been created by the authors, who confirm that the images are original, contain no duplication, and have not been previously published in whole or in part.
Financial support: This study was supported by the Zigong Key Science and Technology Program, No. 2023-NKY-02-11
Data, Code, and Materials Availability
The extracted dataset, analysis grammar (RevMan 5.4 and Stata 14), data extraction table template, and all supplementary charts/tables can be requested by contacting the author (Yudiao Liang, email: liangyudiao@163.com).
References
- 1.Goldberg X, Serra-Blasco M, Vicent-Gil M, et al. Childhood maltreatment and risk for SA in major depression: A sex-specific approach. Eur J Psychotraumatol. 2019;10(1):1603557. doi: 10.1080/20008198.2019.1603557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Xin LM, Chen L, Su YA, et al. Risk factors for recent SA in major depressive disorder patients in China: Results from a national study. Front Psychiatry. 2018;9:300. doi: 10.3389/fpsyt.2018.00300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Shang Z, Lang X, Wang J, Zhang X. Prevalence and risk factors of SA in young, first-episode and drug-naïve Chinese Han outpatients with psychotic major depressive disorder. Eur Arch Psychiatry Clin Neurosci. 2024;274(7):1517–26. doi: 10.1007/s00406-023-01698-4. [DOI] [PubMed] [Google Scholar]
- 4.Pawlak JM, Miechowicz I, Dmitrzak-Węglarz M, et al. Are suicide risk factors gender specific? Psychiatr Pol. 2018;52(1):21–32. doi: 10.12740/PP/78974. [DOI] [PubMed] [Google Scholar]
- 5.Li XY, Tabarak S, Su XR, et al. Identifying clinical risk factors correlate with SA in patients with first episode major depressive disorder. J Affect Disord. 2021;295:264–70. doi: 10.1016/j.jad.2021.08.028. [DOI] [PubMed] [Google Scholar]
- 6.Ye G, Li Z, Yue Y, et al. Suicide attempt rate and the risk factors in young, first-episode and drug-naïve Chinese Han patients with major depressive disorder. BMC Psychiatry. 2022;22(1):612. doi: 10.1186/s12888-022-04254-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Tong P, Shi YH, Yang Y, et al. Clinical characteristics, metabolic parameters, and risk factors for SA vary with untreated major depressive disorder duration. Depress Anxiety. 2023;2023:4869276. doi: 10.1155/2023/4869276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cong X, Zhang T, Bian R, et al. Prevalence and related factors of first-time SA in the past 14 days in Chinese adult patients with first-episode drug-naïve major depressive disorder. Front Psychiatry. 2024;15:1366475. doi: 10.3389/fpsyt.2024.1366475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Stang A. Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol. 2010;25(9):603–5. doi: 10.1007/s10654-010-9491-z. [DOI] [PubMed] [Google Scholar]
- 10.Sokero TP, Melartin TK, Rytsälä HJ, et al. Suicidal ideation and attempts among psychiatric patients with major depressive disorder. J Clin Psychiatry. 2003;64(9):1094–100. doi: 10.4088/jcp.v64n0916. [DOI] [PubMed] [Google Scholar]
- 11.Wang YX, Sun ML, Long JL. Clinical analysis of related risk factors of suicidal behavior in depressive patients. Chinese Journal of Health Psychology. 2005;13(2):90–91. [Google Scholar]
- 12.Gang JL, Wang SY, Wei LL. A study on the risk factors of SA in depression. Shandong Psychiatry. 1995;(4):10–13. [Google Scholar]
- 13.Liu QF, Zhang YQ, Li GL, et al. Risk factors of suicidal behavior in patients with depression. Chinese Journal of Clinical Rehabilitation. 2004;8(18):3456–57. [Google Scholar]
- 14.Liao ZH, Wang WQ, Ding LJ, et al. Characteristics and influencing factors of suicide in patients with major depressive disorder in community. Chinese Journal of General Practice. 2014;(20):2300–4. 5. [Google Scholar]
- 15.Wei S, Li H, Hou J, et al. Comparison of the characteristics of suicide attempters with major depressive disorder and those with no psychiatric diagnosis in emergency departments of general hospitals in China. Ann Gen Psychiatry. 2017;16:44. doi: 10.1186/s12991-017-0167-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Li GL, Wang RQ, Tian F. Current status and risk factors of self-harm and suicidal behavior in adolescent inpatients with depressive disorder. Chin J Psychiatry. 2024;37(3):243–48. [Google Scholar]
- 17.Lu HL, Chen LM, Pan WF, et al. Analysis of the current situation and influencing factors of SA in patients with depression in general hospitals in Nanchang City. Journal of Nanchang University (Medical Edition) 2019;59(6):58–61. 84. [Google Scholar]
- 18.Wang H, Xiao YX, Du JG, et al. Related factors of SA in patients with depressive episode. Chinese Journal of Mental Health. 2021;35(7):541–45. [Google Scholar]
- 19.Xiudi Li. Research on the risk factors of suicidal behavior in depression. 2021 [Google Scholar]
- 20.Shi YJ. A comparative study on the risk factors of SA in patients with unipolar and bipolar depression. 2022 [Google Scholar]
- 21.Pan Y, Li XD, Hong Zhang, et al. Risk factors of SA in patients with depressive disorder. Chinese Journal of Clinical Psychiatry. 2022;32(01):60–62. [Google Scholar]
- 22.Chen LM, Chen L, Yang FD, et al. Risk factors of SA in depressive patients with anxiety symptoms. Chinese Journal of Neuropsychiatric Disorders. 2015;10:613–17. [Google Scholar]
- 23.Chen LM, Chen L, Yang FD, et al. Risk factors of SA in major depressive patients with depressive characteristics. Chinese Journal of Neuropsychiatric Disorders. 2019;45(1):15–19. [Google Scholar]
- 24.Chen L, Liu YH, Zheng QW, et al. Risk factors of SA in patients with major depressive disorder. Chin J Psychiatry. 2013;46(6):339–43. [Google Scholar]
- 25.Zhang X. Suicidal behavior and related factors in patients with major depressive disorder in community. Tianjin Medical University. :2022. [Google Scholar]
- 26.Zhang YQ, Yuan GZ, Li GL, et al. Risk factors of SA in 215 patients with major depressive disorder: A case-control study. Chin J Epidemiol. 2007;28(2):131–35. [PubMed] [Google Scholar]
- 27.Wei B. A study on the risk factors of SA in patients with depression. 2016 [Google Scholar]
- 28.Chen L, Ji ZP, Yang FD, et al. Risk factors of SA in depressive patients with atypical features. Chinese Journal of Neuropsychiatric Disorders. 2017;43(5):294–99. [Google Scholar]
- 29.Zhou XY, YI J, Wang QY, et al. Risk factors of SA in major depressive patients with family history of psychiatric disorders. Chinese Journal of Neuropsychiatric Disorders. 2019;45(8):488–91. [Google Scholar]
- 30.Gao YY, Wang R, Ren RJ, et al. Influence of negative life events and psychological resilience on suicidal behavior in adolescent patients with depression. Neurological Disease and Mental Health. 2024;24(9):629–34. [Google Scholar]
- 31.Li ZY, Ban C. Influencing factors of SA in adolescent depression patients with self-harm. Chinese National Health Medicine. 2022;34(20):1–3. [Google Scholar]
- 32.Dong JN, Bi B, Kong LT. Related factors of SA in depressive outpatients in general hospital. Chin J Physicians. 2019;21(10):1460–63. [Google Scholar]
- 33.Duan YP, Chen L, Liu YH, et al. Risk factors of SA in depressive patients with psychotic features. Chinese Journal of Mental Health. 2013;27(5):357–61. [Google Scholar]
- 34.Seo JG, Lee JJ, Cho YW, et al. Suicidality and its risk factors in korean people with epilepsy: A MEPSY study. J Clin Neurol. 2015;11(1):32–41. doi: 10.3988/jcn.2015.11.1.32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Liu Z, Sun L, Sun F, et al. The abnormalities of lipid metabolism in children and adolescents with major depressive disorder and relationship with suicidal ideation and attempted suicide. Heliyon. 2024;10(9):e30344. doi: 10.1016/j.heliyon.2024.e30344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Wang DB, Lai GX, Xia ZY. Risk factors of suicidal behavior in depressive patients. Chinese Journal of Neuropsychiatric Disorders. 2002;28(2):88–89. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The extracted dataset, analysis grammar (RevMan 5.4 and Stata 14), data extraction table template, and all supplementary charts/tables can be requested by contacting the author (Yudiao Liang, email: liangyudiao@163.com).