Authors: Zelfa Hamadieh, David G. Goldfarb, Narelle Ellendon
Categories: Systematic Review
Source: BMC Psychiatry
Authors: Zelfa Hamadieh, David G. Goldfarb, Narelle Ellendon
Benzodiazepines are among the most prescribed psychotropic drugs worldwide, yet their potential association with suicide remains poorly understood and often overlooked, raising questions that current clinical practice and safety guidelines have yet to address. This study aimed to conduct a meta-analysis to quantitatively synthesize existing evidence on the association between prescribed benzodiazepine use and suicidal behaviors, explore potential mechanisms underlying this relationship, and identify gaps in the current literature.
Following PRISMA guidelines, we conducted a systematic review and meta-analysis of studies published in PubMed between 2016 and 2025, supplemented with earlier studies identified in a prior systematic review. Eligible studies examined the association between prescribed benzodiazepines and suicidal behaviors. Pooled estimates were calculated using a random-effects model to assess heterogeneity, and publication bias was evaluated using Egger’s regression test.
Twenty-one studies met inclusion criteria. Benzodiazepine use was associated with a significantly elevated risk of suicide (pooled effect size = 2.74; 95% CI: 2.06–3.63).
Prescribed benzodiazepines are associated with a more than twofold higher likelihood of suicide outcomes (attempted or completed), representing a public health concern that warrants further investigation. This consistent association underscores the need to investigate underlying biological and behavioral mechanisms, refine and standardize prescribing guidelines, and implement rigorous suicide risk assessment and monitoring protocols, particularly for high-risk patient groups.
not applicable.
This systematic review was conducted in accordance with a registered protocol (PROSPERO ID: CRD420251074013) and adheres to PRISMA 2020 guidelines.
Benzodiazepines (BZDs) are central nervous system depressants that enhance gamma-aminobutyric acid (GABA)-mediated inhibitory effects by binding to benzodiazepine receptors, leading to sedation, anxiolysis, and relaxation [1]. BZDs are widely prescribed for multiple conditions including epilepsy, insomnia, and anxiety [2]. In the United States, 12.5% of adults used BZDs in 2015–2016 [3].
The need to focus on BZDs arises not from their high prevalence, but rather from safety concerns related to their use, including drug overdoses [4], abuse [5], Benzodiazepine Use Disorder (BUD) [6], and suicide [5, 7, 8]. Recently, multiple studies have revealed an alarming trend that suggests a potential link between BZDs and overdose or suicide. According to the Centers for Disease Control and Prevention (CDC), BZD-involved overdose deaths increased by 519.6% from April–June 2019 to April–June 2020 following a rise in their illicit supply [9]. Moreover, multiple postmortem toxicological studies detected BZDs in a significant portion of the individuals who died by suicide [10, 11]. Notably, BZDs have been reported as the most frequently identified drug in both poisoning [12–14] and non-poisoning suicide death cases [15].
However, the association between suicide and BZDs remains a complex and intricate subject. Yet the underlying mechanisms behind it are not well understood in both clinical [16, 17] and epidemiological fields [18], with existing literature presenting conflicting theories. One perspective suggests that benzodiazepines may have a protective effect against suicide by alleviating anxiety and reducing insomnia, both known risk factors for suicide [19]^−^ [20]. This theory is supported by the assumption that the association between BZDs and suicide reported in observational studies is attributed to indication bias [21, 22], where underlying conditions influence both prescription and suicide risk, rather than a causal relationship. On the other hand, several studies have challenged this theory by controlling for confounding by indication bias and still reported a positive association [23, 24].
Yet, this question remains unanswered, highlighting a notable gap in the literature. To date, only two systematic reviews have specifically investigated the relationship between prescribed BZDs and suicide, and no meta-analysis has been conducted. The first systematic review, published in 2017, is now outdated [25]. The second, conducted in 2025 [26], provided a qualitative synthesis without protocol registration or quantitative methods. Building on this earlier work, the present study advances the field through a prospectively registered PROSPERO protocol, adherence to PRISMA 2020 guidelines [27], clearly defined exposure and outcome criteria, and a comprehensive quantitative meta-analysis incorporating random-effects modeling, heterogeneity assessment, and publication bias evaluation. Together, these methodological elements establish the current review as a structured and transparent quantitative evaluation of the association between prescribed benzodiazepine use and attempted or completed suicide.
To our knowledge, this is the first meta-analysis examining suicidal behaviors among BZD users across diverse populations. This study seeks to provide valuable insights into how prescription patterns and patient characteristics may relate to the association between BZD use and suicidal behaviors. Furthermore, the findings may serve as a foundation for future investigations into the mechanisms underlying this association, ultimately informing clinical practices and public health policies for safer prescription guidelines and protocols.
We conducted a systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines [27]. The review protocol was prospectively registered with PROSPERO (ID: CRD420241078432) on June 21, 2025 [28].
A literature search was conducted in PubMed to identify studies published between October 12, 2016, and January 30, 2025. Searches were performed using predefined keyword-based queries entered directly into the PubMed search interface, consistent with the registered protocol. The following exact keyword combinations were “Benzodiazepine AND Suicide,” “Alprazolam AND Suicide,” “Diazepam AND Suicide,” “Clonazepam AND Suicide,” “Temazepam AND Suicide,” and “Lorazepam AND Suicide.” All searches applied identical PubMed filters (English, Humans, Full text, and the specified date October 12, 2016–January 30, 2025). Articles were eligible for inclusion only if they examined prescribed benzodiazepine use in relation to suicide outcomes (attempts, reattempts, or suicide deaths) and were available as peer-reviewed full-text publications.
For studies published prior to October 12, 2016, the literature review titled “Prescribed Benzodiazepines and Suicide Risk: A Review of the Literature” [25] was considered to identify additional eligible articles. This earlier review applied broader inclusion criteria than those defined in the current study. Therefore, we applied our new set of inclusion criteria to exclude any articles that did not meet these revised standards. Consequently, studies published before and after October 12, 2016 were subjected to identical inclusion criteria.
Studies were included if they examined the relationship between prescribed BZD use, either as a class or as individual agents, and suicide outcomes (attempts, reattempts, or suicide deaths). Only prescribed BZD use for medical purposes was eligible, excluding non-medical, recreational, illicit use, misuse or abuse, and polysubstance use. Additional exclusions included bivariate models; studies focusing on withdrawal, discontinuation, or tapering effects; polytherapy with other psychotropic medications; concurrent psychotherapy; suicidal ideation or self-harm; broader benzodiazepine medication groups encompassing BZD-like drugs (e.g., Z-drugs); or studies lacking 95% confidence intervals and p-values for effect estimates.
Comparative studies between different benzodiazepine drugs were only considered if they used a non-user reference group to maintain comparability across studies. Additionally, because most overdose studies did not differentiate between intentional and unintentional overdoses, all single-method suicide articles (such as overdose or hanging) were excluded to avoid bias.
For studies reporting both suicide completion and attempt outcomes, only suicide completions (the more severe outcome) were considered. All BZDs exposures were included regardless of duration (short-term or long-term), prescription patterns, or dosage regimens.
Title and abstract screening and full-text eligibility assessment were conducted by a primary reviewer, while a second reviewer independently verified the screening decisions. Any disagreements were resolved through discussion to reach consensus. Data extraction was performed by the primary reviewer using a standardized Excel sheet, and all extracted information was independently checked by the second reviewer for accuracy and completeness.
The Excel extraction sheet captured study identifiers (author and year of publication), study design, population characteristics (including age and sex where reported), exposure definitions and details related to prescribed benzodiazepine use, outcome definitions, data sources or datasets used, study location, and reported measures of association, including effect estimates and corresponding confidence intervals or p-values where available. Table 1 summarizes the key study characteristics extracted and reported for the purposes of this meta-analysis.
Table 1Summary of studies examining the relationship between prescribed benzodiazepine exposure and suicide outcomesStudyExposureOutcomePopulationMeasure of Association(95% CI, p-value*)Boggs et al., 2020 [29]1–2 BZD fills (within 200 days)Suicide deathPatients with anxiety or sleep disorders, aged ≤ 89 yearsOR = 1.66(1.21–2.27, < 0.01)3–5 BZD fills (within 200 days)Suicide deathPatients with anxiety or sleep disorders, aged ≤ 89 yearsOR = 3.17(2.28–4.42, < 0.01)15 + BZD fills (within 200 days)Suicide deathPatients with anxiety or sleep disorders, aged ≤ 89 yearsOR = 5.64(3.01–10.54, < 0.01)Guina et al., 2017 [30]BZD prescription (current or past)Suicide attemptTrauma survivors aged ≥ 18 yearsAOR = 1.72Not significantTubbs et al., 2021 [31]Past-year use of sedative BZD (defined as triazolam, temazepam, and flurazepam)Suicide attemptAdults aged ≥ 18 yearsOR = 1.75(1.06–2.87)Cato et al., 2019 [32]BZD treatment within 2 years prior to suicideSuicide (died by suicide or uncertain cause of death)Psychiatric patients aged 13–96 yearsOR = 1.83(1.06–3.14)Demesmaeker et al., 2021 [33]BZD prescriptionSuicide reattempt (within 14 months, adjusted)Patients of legal age (≥ 18 years) with a suicide attempt within 7 days prior to inclusionHR = 1.87(1.25–2.81, < 0.01)Deka et al., 2018 [34]At least one 30-day prescription for BZDSuicide deathVeterans with PTSD (all ages included)HR = 2.74(2.40–3.13)Donovan et al., 2019 [35]Long-term BZD use (≥ 90 d)Suicide deathVeterans aged ≥ 40 years with COPD and PTSDHR = 2.33(1.14–4.79)Strømme et al., 2024 [36]Use of specific BZD (alprazolam, diazepam, flunitrazepam, etc.)Suicide death or attemptPatients with schizophrenia (mean age at index 41.1 years)HR = 1.15(0.39–3.36)Gibson et al., 2024 [37]Long-term BZD use (90 days out of 180 days or more)Suicide deathVeteran women aged ≥ 50 yearsHR = 2.99(1.82–4.91)Xie et al., 2025 [38]AlprazolamSuicide deathAges 18–85 yearsROR = 21.04(19.99–22.14)Smith et al., 2024 [39]BZD use within 90 days before deathSuicide deathVeterans with and without HIV (98.9% male; mean age 54.2 ± 10.7 years)IRR = 2.12(1.58–2.84, < 0.001)Tiihonen et al., 2015 [40]BZD high exposureSuicide deathIndividuals aged 16–65 years with a schizophrenia diagnosisHR = 2.16(1.29–3.64)BZD low exposureSuicide deathIndividuals aged 16–65 years with a schizophrenia diagnosisHR = 1.30(0.85–2.00)BZD moderate exposureSuicide deathIndividuals aged 16–65 years with a schizophrenia diagnosisHR = 1.43(0.88–2.34)Narindrarangkura et al., 2023 [41]BZD prescriptionSuicide attemptIndividuals with diabetes (age 8–88 years)AOR = 2.26(2.24–2.28, < 0.001)Lee et al., 2018 [42]BZD prescriptionSuicide deathAll age groupsHR = 1.314(1.098–1.571, < 0.003)Høier et al., 2025 [43]BZD -class hypnotic prescriptionSuicide death, malesNationwide sample of individuals living in Denmark aged ≥ 15 yearsAIRR = 2.1(1.9–2.4)Suicide attempt, malesNationwide sample of individuals living in Denmark aged ≥ 15 yearsAIRR = 3.4(3.1–3.7)Suicide death, femalesNationwide sample of individuals living in Denmark aged ≥ 15 yearsAIRR = 2.6(2.3–3.0)Suicide attempt, femaleNationwide sample of individuals living in Denmark aged ≥ 15 yearsAIRR = 3.6(3.4–3.9)Voaklander et al., 2008 [8]BZD prescription 30 days preceding suicide deathSuicide deathSeniors aged ≥ 66 yearsOR = 4.46(3.25–6.11)Tiihonen et al., 2012 [44]BZD useSuicide deathPatients with schizophrenia aged 16–65 yearsHR = 3.83(1.45–10.12)Huang et al., 2024 [45]AlprazolamSuicide deathAll age groupsROR = 12.23(11.58–12.91)Su et al., 2024 [5]LorazepamSuicide attemptMainly aged > 20 years oldROR = 3.13(3.01–3.25)Abel et al., 2025 [46]BZD prescriptionSuicide deathVeterans with comorbid bipolar disorder and musculoskeletal disorder diagnoses aged ≥ 16 yearsARR = 1.54(1.48–1.60)Neutel & Patten, 1997 [47]BZD use (triazolam, flurazepam, lorazepam, diazepam, oxazepam) within 60 daysSuicide attemptNon-antidepressant users aged ≥ 20 yearsOR = 6.20(2.6–15.4)* p-values are based on reported statistical tests from the original studies where applicable. Characteristics and findings of the 21 studies included in the systematic review and meta-analysis, evaluating the association between prescribed benzodiazepine use and attempted or completed suicide. Columns present study citation, exposure definition, outcome, population characteristics, measure of association, and reported effect estimates with 95% confidence intervals and p-values (where available)
Quantitative synthesis combined different relative effect measures into a pooled relative effect estimate. Under a rare-outcome approximation, hazard ratios, incidence rate ratios, risk ratios, and odds ratios converge to a relative association. To evaluate the robustness of this approach, sensitivity analyses stratified by effect measure were conducted.
The random-effects model developed by DerSimonian and Laird was employed for the meta-analysis [48]. This approach was selected over the fixed-effects model because it accounts for both within-study and between-study variability [49], which is appropriate given the observational design and expected heterogeneity of the included studies. The inconsistency index (I²) was used to assess heterogeneity, with values indicating the proportion of variability due to true heterogeneity rather than random 0%-30%, 31%-60%, 61%-75%, and 76%-100% representing low, moderate, substantial, and considerable heterogeneity, respectively [50].
The Q statistics were applied to further evaluate the degree of heterogeneity, with a p-value < 0.10 indicating statistically significant heterogeneity [51]. However, it is important to note that the Q statistic has some limitations, including its ability to determine whether heterogeneity is present or absent without reporting the extent of it [52]. Additionally, It is also sensitive to the number of studies, more studies can increase the power of the test, which might lead to detecting small differences that aren’t practically meaningful [53]. Forest plots were used to visualize study estimates.
Publication bias was assessed using visual inspection of funnel plot asymmetry and statistically tested using Egger’s regression test [54]. Egger’s test evaluated the association between effect sizes and their standard errors, with a non-significant p-value indicating no evidence of small-study effects.
Risk of bias was assessed qualitatively at the study level, focusing on study design, clarity of exposure and outcome definitions, and confounder adjustment. This qualitative assessment informed a sensitivity analysis restricted to studies with higher reporting strength, which was conducted to evaluate the robustness of the pooled estimate. Studies were classified as having higher reporting strength if they were population-based or healthcare-based observational studies with clearly defined BZD exposure and suicide outcome definitions and reported multivariable-adjusted effect estimates. Studies that did not meet these criteria were excluded from the sensitivity analysis.
Subgroup analyses were conducted to explore potential sources of heterogeneity by patient group. Separate analyses were performed for studies focusing on schizophrenia and for studies including broader clinically defined psychiatric populations. These analyses were exploratory and intended to assess patterns of effect estimates and heterogeneity rather than to formally compare subgroups.
The certainty of the evidence was assessed using a multi-criteria approach, considering study design, qualitative risk of bias assessment, consistency of findings across studies, heterogeneity, precision of effect estimates as indicated by the reporting and width of confidence intervals, assessment of publication bias, and robustness of findings as evaluated through sensitivity analysis.
All statistical analyses and visualization were performed using RStudio (Mountain Hydrangea, version 2023.06.0 + 421, Posit Software, PBC). The following packages were “metafor,” “meta,” and “dplyr”.
Our systematic search identified 460 records (443 from PubMed, 2016–2025, and 17 from prior reviews). After removing duplicates (n = 28) and screening titles/abstracts, 83 full-text articles were assessed for eligibility. Twenty-one studies met inclusion criteria (17 from PubMed, 4 from prior reviews) (Fig. 1). Table 1 summarizes the 21 studies included in the final review along with the key characteristics of their study populations.
Fig. 1PRISMA 2020 flow diagram of study selection, illustrating the systematic search and selection process for studies examining the association between prescribed benzodiazepine use and suicide-related outcomes (completed or attempted), 1990–2025. A total of 443 records were identified from PubMed after October 12, 2016, and 17 additional records from a prior literature review. After duplicate removal and screening, 21 studies met inclusion criteria
Using a random-effects model, the pooled relative effect estimate was 2.74 (95% CI: 2.06–3.63), indicating an elevated association between BZD use and suicide outcomes. Figure 2 presents a forest plot summarizing effect estimates from the 21 included studies assessing the association between benzodiazepine (BZD) use and suicide-related outcomes.
Fig. 2Forest plot of benzodiazepine use and suicide risk of the random-effects meta-analysis evaluating the association between prescribed benzodiazepine use and suicide-related outcomes (completed or attempted) across the included studies
Substantial heterogeneity was observed across studies (Q = 11386, df = 24, p < 0.001), with an I² value of 99.7%, indicating that most observed variability reflected between-study differences rather than chance. The τ² estimate (0.469) further indicated considerable between-study heterogeneity.
There was no evidence of publication bias. Visual inspection of the funnel plot showed symmetry (Fig. 3), and Egger’s regression test did not indicate small-study effects (z = − 0.85, p = 0.397), suggesting that the meta-analysis results are unlikely to be influenced by small-study effects or selective reporting.
Fig. 3Funnel plot assessing publication bias in the meta-analysis among included studies. Visual inspection suggested symmetry, and Egger’s regression test (z = − 0.85, p = 0.397) indicated no significant evidence of publication bias
In subgroup analyses of studies including broader clinically defined psychiatric populations (k = 8), the association remained elevated (2.34, 95% CI: 1.66–3.30), with substantial heterogeneity (I² = 92.0%). In contrast, subgroup analyses restricted to studies focusing on schizophrenia (k = 3) showed a statistically significant association (2.18, 95% CI: 1.29–3.70), with markedly lower heterogeneity (I² = 24.5%).
In sensitivity analyses restricted to studies with higher reporting strength, conducted as part of the risk-of-bias assessment, the association between BZD use and suicidal behaviors remained statistically significant (2.45, 95% CI: 2.13–2.81). Heterogeneity remained high (I² = 97.2%).
In sensitivity analyses restricted to studies reporting odds ratios (ORs) only (k = 6), the pooled estimate was 2.76 (95% CI: 2.10–3.63; I² = 82.6%). Similarly, analyses restricted to studies reporting hazard ratios (HRs) only (k = 9) yielded a pooled estimate of 1.99 (95% CI: 1.47–2.68; I² = 84.8%). The consistency in effect direction and magnitude across effect measures supports the robustness of the primary analysis.
Based on a multi-criteria assessment, the overall certainty of the evidence was rated as moderate, with limitations related to the observational nature and substantial heterogeneity.
This study aims to systematically evaluate the association between BZD use and suicidal behaviors (both attempted and completed). To our knowledge, this is the first meta-analysis conducted to further investigate suicide risk among BZD users. Using a random-effects model, the meta-analysis revealed a significant association between BZD exposure and suicide risk. Specifically, BZD users had a 2.74-fold higher estimated likelihood of attempting or completing suicide compared to non-users, indicating a consistent association between BZD use and suicidal behaviors.
Several biological mechanisms have been proposed to explain the observed association between BZD use and suicide. Postmortem neurobiological studies have consistently identified alterations in the GABAergic system among individuals who died by suicide [55, 56]. Notably, BZDs were found to alter and modulate this system through various pathways, including downregulation of GABA-A receptors [57], receptor desensitization [58], and rebound upregulation following withdrawal [59]. These neuroadaptations could potentially disrupt the brain’s inhibitory balance and may contribute to increased vulnerability to suicidal behavior. In addition to these changes, BZDs have been shown to increase cerebrospinal fluid (CSF) GABA levels [60], which have been directly associated with heightened impulsivity [61], a key factor implicated in suicidal acts. However, the exact biological mechanism by which BZDs contribute to suicide is unconfirmed yet.
Consistent with this disinhibitory effect, BZDs were reported to be associated with “paradoxical reactions,” where, rather than exerting their intended calming effects, they lead to increased aggression, impulsivity, and altered mental status in a 1–20% of individuals [18, 62, 63]. In some cases, the aggression reached a level of an assault or a rape [64]. In addition to these behavioral side effects, the use of hypnotics, including BZDs, has been reported to be associated with a higher incidence of depression, a well-established risk factor for suicide [65, 66].
Apart from the induced neurobiological and behavioral effects, BZD use has been linked to the development of Benzodiazepine Use Disorder (BUD) [1, 67–69]. In a clinical sample from Lebanon, 63.1% of BZD users developed BUD [6]. In comparison, approximately 17.2% of adults who use BZDs engaged in misuse in the United States [3]. BUD is characterized by both physical dependence and psychological symptoms [70], including withdrawal effects that resemble those observed in alcohol withdrawal [1], which may contribute to suicide risk.
Despite these numbers, the potential mediating role of BUD in the relationship between BZD use and suicide has not been adequately explored in the literature. Challenges in assessing this relationship stem from difficulties in reporting, mainly due to the absence of BUD as a distinct separate diagnostic category in both the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) and the International Classification of Diseases, 10th Revision (ICD-10). Instead, it is included under broader classification “Sedative, Hypnotic, or Anxiolytic Use Disorder” [70, 71], which includes multiple substances beyond BZDs, potentially leading to inaccurate prevalence estimates and misclassification in reporting. Similarly, in the literature, BUD is often misreported, with many articles misidentifying it as dependence [18], prolonged use [72], or high-dose consumption. Terms like dependence and addiction are often used interchangeably [68, 73, 74], leading to inconsistency and further complicating efforts to assess its potential role in suicide risk.
Another interpretation of the association between BZDs and suicide lies in the inconsistency of prescribing guidelines and poor adherence among healthcare providers. Concordance with BZD prescribing guidelines was found to be associated with a lower likelihood of suicide-related outcomes [29]. However, contrary to these recommendations, most long-term benzodiazepine users in the USA receive prescriptions from non-psychiatric providers [75]. Furthermore, BZDs are often prescribed for older adults in long-term [75], despite the significant risks associated with their use in older ages [76]. Notably, one in four older adults met the definition of long-term use after one year of being prescribed BZDs [77].
Given these safety concerns and suicide or dependence risk, there have been calls to assess the risks and benefits through a “risk-benefits” ratios to weigh out the benefits of using such drugs [78]. While many argue that benzodiazepines are safe for short-term use, a major challenge in evaluating their long-term safety lies in the lack of agreement and the inconsistent definition of “long-term use” across the literature and guidelines. For example, Olfson et al. (2015) defined long-term use as a minimum of 120 days’ supply within a year [75], while other studies have used continuous use of 90 days or more as the threshold [35, 79]. Lader et al. (2011) proposed that the risk–benefit ratio for BZDs remains favorable in the short term (2–4 weeks), but becomes uncertain beyond this period [18]. In terms of suicide risk, Boggs et al. (2020) suggested that three fills of benzodiazepines could indicate long-term use, with an average fill duration ranging from 7 to 14 days, equating to 3 to 6 weeks of use [80]. In contrast to that, USA FDA guidelines recommend the use of Alprazolam for generalized anxiety disorder as no longer than 4 months [81]. These inconsistencies complicate suicide risk assessment and underscore the need for a safe drug-specific threshold of long-term use, particularly in light of the observed association between benzodiazepine use and suicidal behaviors identified in this study.
When considering this association, it is important to acknowledge the potential for indication bias. Most of the studies included in our review addressed indication bias to varying degrees through study design and statistical adjustments. Some studies employed matched case-control or cohort designs and accounted for related confounding variables, demonstrating strong control over indication bias [32–34, 37, 39]. While other studies applied moderate to partial controls. Yet, the association between BZD use and suicide risk persisted, challenging the notion that indication bias may fully explains the link. Notably, only one study reported that the association was eliminated after conducting a post hoc analysis and adjusting for Post-Traumatic Stress Disorder symptoms’ severity. However, the author cautioned that BZDs might be contributing to the severity of the disorder rather than merely indicating it [30].
While the primary meta-analysis demonstrated substantial heterogeneity, which is not unexpected in observational research, subgroup analyses showed that heterogeneity was markedly reduced when analyses were restricted to studies focusing on schizophrenia, whereas heterogeneity persisted in analyses of broader clinically defined psychiatric populations. This pattern suggests that a meaningful proportion of the observed heterogeneity is likely driven by differences in underlying population characteristics rather than random variability. Other potential sources of variability may include broad exposure definitions and differences in study population characteristics. Such variability likely reflects features of the existing literature rather than methodological shortcomings of the present analysis. Accordingly, the pooled estimate should be interpreted as an average association across highly variable contexts, and its generalizability to specific patient subgroups may be limited. Subgroup analyses findings should therefore be considered exploratory and not used to support definitive group-specific conclusions.
We emphasize that this study does not advocate for forced discontinuation of long-term BZD therapy. While the study identifies an association between BZD use and suicide risk, its purpose is not to promote policy changes that pressure providers to abruptly withdraw treatment, particularly in clinically complex cases or where the safety of discontinuation remains uncertain. These findings must be interpreted with caution, drawing lessons from the unintended harms of the 2016 CDC opioid guidelines [82]: when opioid therapy was abruptly tapered, patients faced devastating consequences, including increased overdoses [83], uncontrolled pain, psychological distress, suicide [84], and illicit drug use [85]. Similarly, forced BZD discontinuation in long-term users may carry significant risks, such as heightened mortality or suicidality [86]. Rather than endorsing aggressive or mandatory prescribing changes, this study advocates for the development of evidence-based prescribing guidelines, strengthened monitoring and risk assessment protocols, cautious initiation of BZD treatment, and the adoption of multidisciplinary care models that minimize reliance on BZDs while maintaining patient safety.
Due to the limited availability of data, several important areas could not be assessed or included in this literature the use of BZDs as an illicit drug, add-on treatment, or in poly-substance use, as intentional abusers often use benzodiazepines to amplify the effects or mitigate the adverse effects of other drugs [68].
All included studies were observational, which inherently limits causal inference and leaves residual confounding unaddressed, preventing definitive conclusions about the causal relationship between BZD use and suicidal behaviors. High heterogeneity complicates the interpretation of the single pooled estimate (2.74), which represents an average across diverse study populations, exposure definitions, and clinical contexts rather than a uniform effect. Accordingly, the pooled estimate should be interpreted with caution and viewed as an overall summary of association rather than a setting-specific effect size. Furthermore, by excluding illicit and multiple substance use contexts, our findings may underestimate the suicide risk in populations prone to misuse, it is expected that these groups may have higher suicidal risk due to the compounding effects of multiple substance dependencies and associated psychosocial factors [69].
Potential publication bias, stemming from language restrictions (English only), reliance on peer-reviewed journals, and the use of a single bibliographic database may have further constrained the evidence base. The lack of differentiation between reported suicide methods and the frequent absence of clear temporal sequencing between BZD exposure and suicidal behavior complicated risk interpretation. Finally, due to data constraints, this study was unable to conduct an in-depth analysis of specific high-risk groups, evaluate patterns of use, or compare long-term versus short-term or long-acting versus short-acting BZD use. Also, since many studies focused on specific subpopulations (such as veterans or older adults), the generalizability of our conclusions to broader patient groups remains limited.
Our literature review and meta-analysis identified an association between BZD use and suicidal behaviors, highlighting an important public health concern that warrants careful consideration. This includes further research into the underlying biological mechanisms and investigation of the epidemiological pathways, as well as a thorough review of current clinical guidelines and prescribing practices. Such efforts should aim to minimize suicide risk by updating, standardizing, and ensuring adherence to prescribing protocols while safeguarding the well-being of current users, particularly those on long-term therapy.
This study also highlights several critical gaps in the literature. First, more research is needed to clarify the association between BZD use and suicidal behaviors, particularly through experimental or longitudinal designs that extend beyond observational data. Second, the literature lacks clarity regarding the potential role of BZD use disorder (BUD) in contributing to the observed suicide risk. The ambiguity and inconsistency in the standardized reporting of BUD make it challenging to develop a comprehensive understanding of the underlying mechanisms by which BZDs may be linked to suicide. Third, it is essential to review, update, and standardize BZD prescribing practices to ensure consistency and adherence to evidence-based guidelines among healthcare practitioners, thereby reducing potential suicide risk while maintaining patient safety.