Can seizure therapies and noninvasive brain stimulations prevent suicidality? A systematic review

Abstract Background Suicide is a major public health issue and the majority of those who attempt suicide suffer from mental disorders. Beyond psychopharmacotherapy, seizure therapies and noninvasive brain stimulation interventions have been used to treat such patients. However, the effect of these nonpharmacological treatments on the suicidal ideation and incidence of suicidality remains unclear. Here, we aimed to provide an update on the effects of seizure therapies and noninvasive brain stimulation on suicidality. Methods We conducted a systematic review of the literature in the PubMed, EMBASE, Cochrane Central Register of Controlled Trials, Elsevier ScienceDirect, and Wiley Online Library databases using the MeSH terms “Electroconvulsive Therapy”, “Magnetic Seizure Stimulation”, “repetitive Transcranial Magnetic Stimulation”, “transcranial Direct Current Stimulation”, “Cranial Electrostimulation” and “suicide”. We included studies using seizure therapies and noninvasive brain stimulation as a main intervention that evaluated suicidality, regardless of diagnosis. Results Among 1,019 records screened, 26 studies met the inclusion criteria using either electroconvulsive therapy (n = 14), magnetic seizure therapy (n = 2), repetitive transcranial magnetic stimulation (n = 9), or transcranial direct current stimulation (n = 1). We observed that studies reported significant results, suggesting these techniques can be effective on the suicidal dimension of mental health pathologies, but a general statement regarding their efficacy is premature due to limitations. Conclusions Future enquiry is necessary to address methodological limitations and evaluate the long‐term efficacy of these methods both alone and in combination with pharmacotherapy and/or psychotherapy.


| INTRODUC TI ON
According to the World Health Organization, suicide leads to approximately 800,000 deaths every year (WHO, 2019). The worldwide number of suicide attempts is even higher with a rate of 16 per 100,000 individuals, and based on data provided by the European Psychiatric Association, the rates of attempted suicides rate among males and females were 18 and 11 per 100,000, respectively (Plans et al., 2018).
Every suicide is a tragedy for their families, communities, and entire countries for a long period. Suicides may occur throughout the lifespan and were the second leading cause of death among 15-to 29-year-olds in 2016. Suicide not only occurs in high-income countries, but it is also a global phenomenon in all regions of the world, with over 79% of the completed suicides occurring in low-and middle-income countries (WHO, 2019). Mental disorders, especially depression, are frequently found in the psychological reports of suicides in Asian as well as Western countries (Cheng et al., 2000;Phillips et al., 2002).
Given the high suicide risk of patients diagnosed with mental illnesses, it is important to track whether specific treatment strategies have the potential to rapidly and effectively reduce suicide risk.
The common treatment approaches-such as pharmacotherapy and psychotherapy-have over years emerged as essential interventions reducing suicidality. However, lithium and ketamine, while effective, have various limitations such as drowsiness, dizziness, poor coordination, toxicity, and risk for dependency (Andrade, 2017;Zalsman et al., 2016). Regarding psychotherapy, a recent meta-analysis concluded that dialectical behavior therapy reduced self-directed violence and resulted in less psychiatric crisis interventions; however, it did not reduce suicidal ideation (DeCou et al., 2018). Therefore, to decrease suicidal ideation and prevent suicide attempts other alternative or adjunct methods are warranted. Among them, seizure therapies and noninvasive brain stimulation methods have been proposed as possible interventional tools.
Seizure therapies such as electroconvulsive therapy (ECT), magnetic seizure therapy (MST), and noninvasive brain stimulation (NIBS) methods such as repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), and cranial electrostimulation (CES) are under investigation (Desmyter et al., 2011). They may alleviate symptoms in several psychiatric conditions depending on the targeted psychiatric condition and application (Lefaucheur et al., 2014(Lefaucheur et al., , 2017Milev et al., 2016;Rosa & Lisanby, 2011;UK ECT Review Group, 2003). However, their efficacy on suicidality is still unclear. The aim of the present systematic review was to evaluate the effects of these applications on suicidality, that is, suicidal ideation, suicide attempt, and/or completed suicide.

| Search strategy and selection criteria
We performed a systematic review of the literature according to PRISMA guidelines (Liberati et al., 2009) including full-length original articles published in English and peer-reviewed journals until 1 June 2019. This systematic literature search was conducted in the PubMed, Cochrane Central Register of Controlled Trials, Elsevier ScienceDirect, and Wiley Online Library (STM) databases using a term list including words relative to ECT, MST, rTMS, tDCS, CES, and suicidality. We used words with combination of thesaurus [MeSH Terms] related to medical subject heading description, and [Other Term] related to keywords used in studies: (1) "Electroconvulsive Therapy" AND "suicide"; (2) "Magnetic seizure stimulation" AND "suicide"; (3) "repetitive Transcranial Magnetic Stimulation" AND "suicide"; (4) "Transcranial Direct Current Stimulation" AND "suicide"; (5) "Cranial Electrostimulation" AND "suicide".
Two independent investigators (YC and CM) screened the results according to the eligibility criteria, first on titles and abstracts and then on full-text articles. Eligibility criteria were as follows: (1) interventions designed for patients with a mental health disorder with available suicidality assessment as a main or secondary outcome; (2) seizure therapies (ECT and/or MST) and/or NIBS methods (rTMS and/or tDCS and/ or CES) specified as the main interventions in acute or continuative phase; and (3) quality of the design corresponding to level 1, 2, or 3 on the Sackett scale. Duplicate articles and double search results were identified and removed throughout the search process.

| Data extraction and study quality assessment
Methods and results were extracted for each included article. We paid special attention to participant characteristics (sample size, diagnosis, treatment), study design, statistical analysis, and reported findings (immediately after treatment and long-term); next, we analyzed similarities and differences between articles with the aim of verifying which parameters were effective and for which mental disease.
Each study design was evaluated according to the modified Sackett Scale which is based on physiotherapy evidence database (PEDro) scores for its level of evidence grade. The Sackett Scale involves 5 levels of evidence: level 1-high-quality randomized controlled trial (RCTs) (PEDro ≥6), divided into level 1a and 1b, based on the number of RCTs supporting the evidence; level 2-RCTs (PEDro score <6), cohort studies, and prospective controlled trials; level 3-case controls such as retrospective studies comparing conditions; levels 4 and 5-case reports, case series, uncontrolled pre-post-tests, and observational studies.
In addition, we referred to the grading of recommendations, assessment, development, and evaluations (GRADE) method to rate the quality of evidence. GRADE presents four grades as levels of evidence: high, moderate, low, and very low, considering different factors such as bias related to imprecision, inconsistency, publication, and indirectness of the study (Guyatt et al., 2011).

| RE SULTS
The literature search retrieved 1,019 articles. After duplicates and double search results were removed and titles and abstracts were assessed, 58 articles were selected for further evaluation. Among them, 26 studies met all inclusion criteria following full-text evaluation and were systematically analyzed. Due to the search results, George et al. (2014) included suicidal patients with mild traumatic brain injury was also accepted in our review. Berlim et al. (2014), Desmyter et al. (2016), andHadley et al. (2011), were accepted as additional record identified through other sources by reading references of Weissman et al. (2018). Figure 1 shows the results of the search and selection process.

| Diagnostic tools
All the studies but two studies (Hadley et al., 2011;Hunt et al., 2011) used a diagnostic tool to select participants at baseline such as

| Stimulation protocol
Only half of the ECT studies described the treatment parameters.
The parameters included features such as current, duration, frequency, and pulse width. Keshtkar et al. (2011) andCiapparelli et al. (2001)  Technical characteristics of ECT are presented in Table 2.
MST intervention of these two studies was delivered with a MagPro MST machine (MagVenture) and a twin coil symmetrically placed over the frontal cortex with the maximal electric field over the Fz electrode position according to the international 10-20 system (DaSilva et al., 2011). Details are provided in Table 3.
Intensity (expressed as % of the resting motor threshold (RMT)), frequency of stimulation, frequency of sessions, train duration and intertrain intervals, type of coil, and coil location are important parameters that influence the effects of rTMS.
A large variability of these parameters was observed between studies, while a large majority of studies used high-frequency  Table 4. Table 5, the only available tDCS study used a current density of 0.8 A/m 2 (2 mA/25 cm 2 ) per 30 min/d using

| Associated therapies
Potential add-on treatments such as medication that can be delivered during the stimulation protocol were not specified in twelve studies (Berlim et al., 2014;George et al., 2014;Hadley et al., 2011;Hunt et al., 2011;Keshtkar et al., 2011;Liang et al., 2017;Munk-Olsen et al., 2007;Nordenskjöld et al., 2013;Popiolek et al., 2018;Sun et al., 2016Sun et al., , 2018Weissman et al., 2018). In the studies from Baeken et al. (2019) and Desmyter et al. (2014Desmyter et al. ( , 2016 and, rTMS was given as a monotherapy. In other seven studies, participants were demanded to use the same type psychotropic medication to eliminate its medical effects on the outcomes. For instance, ECT was delivered together with antidepressant therapy as an experimental group in two studies (Brådvik & Berglund, 2006;Kellner et al., 2005).

| Sham stimulation
No sham-stimulation groups were included in the ECT or MST studies. Five rTMS studies (Baeken et al., 2019;Desmyter et al., 2014Desmyter et al., , 2016George et al., 2014;Weissman et al., 2018) and the tDCS study (Brunoni, Júnior, et al., 2013;Brunoni, Valiengo, et al., 2013) were double-blind sham-controlled whereby both patients and investigators were blinded to the treatment allocations. In rTMS studies, the sham stimulation was administrated with the coil angled 90° away from the scalp in a single-wing tilt position producing some cutaneous sensation and similar sound intensity to that of active stimulation (Blumberger et al., 2011, or a specially designed commercial sham coil, exactly the same as the active coil in appearance, was placed on exactly the same target area of the scalp but without any active stimulation (Baeken et al., 2019;Desmyter et al., 2014Desmyter et al., , 2016George et al., 2014). For the tDCS experiment, the sham method consisted of a brief (<30-60 s) period of active stimulation to mimic skin side effects such as tingling and itching before the simulated procedure.

| Duration
In the ECT experimental studies, the treatment lasted from 2 weeks (four sessions) for the shortest (Veltman et al., 2018) to 4 months (thirteen sessions) (Kellner et al., 2005) for the longest. The two MST studies did not indicate the accurate duration of experiment.
In the rTMS studies, the duration varied from 3 days to 8 weeks and stimulation was delivered one session per day during 10 days in the tDCS protocol. Apart from four ECT studies (Hunt et al., 2011;Liang et al., 2017;Munk-Olsen et al., 2007;Patel et al., 2006) and two rTMS studies (Berlim et al., 2014;Croarkin et al., 2018), the other eight ECT studies, seven rTMS studies, and the only available tDCS study evaluated participants during follow-up ranging from weeks to years.

Frequency of treatment
Inventory of Depressive Symptoms-Adolescent version (QIDS-A17) and the 16-item Quick Inventory of Depressive Symptomatology-Self-Report (16-QIDS-SR) were used as the combination of several of these scales or specific scale for adolescents (Berlim et al., 2014;Wall et al., 2011).

| Results of ECT studies
Ten out of the fourteen ECT studies reported significant efficacy of ECT on suicidality. Five studies highlighted the beneficial effect of ECT on suicidal ideation (Ciapparelli et al., 2001;Kellner et al., 2005;Keshtkar et al., 2011;Patel et al., 2006;Veltman et al., 2018). Another retrospective study observed that ECT can reduce the occurrence of suicide attempts (Brådvik & Berglund, 2006), and suicide attempts were seen significantly less in the ECT group than in the pharmacotherapy-alone group (Avery & Winokur, 1978;Nordenskjöld et al., 2013). Two other papers reported the long-term effect of ECT on completed suicides (Ahmadi et al., 2015;Liang et al., 2017). Three ECT studies did not observe any beneficial effects of ECT on suicidality (Black et al., 1989;Hunt et al., 2011;Popiolek et al., 2018). The ECT study published by Munk-Olsen et al. (2007) described an increase in suicide rates in patients receiving ECT. They believed that the bias (confounding by indication) is a likely explanation for the moderately increased suicide risk among ECT patients in this study and the more marked increase shortly after treatment.
Within them, Kellner et al. (2005) randomly assigned participants whose condition remitted and remained remitted for one week without treatment, after an acute course of three bilateral ECT sessions for a 6-month continuative phase, to receive either continual ECT or continual nortriptyline+lithium treatment. The results showed that 38.2% of the patients had a score decrease from 3 or 4 to 0 at item 3 of the 24-HDRS after 1 week (3 ECT sessions), 61.1% after 2 weeks (6 ECT sessions), and 80.9% at the end of the course of treatment (9 ECT sessions). In addition, comparing the rapid antisuicidal effects with other NIBS like rTMS, Keshtkar et al. (2011) randomized 73 patients with MDD in two groups: one rTMS and one ECT group. After ten sessions, ECT led to greater depressive symptom reductions and suicidal behavior than ten sessions of rTMS, as evaluated by the BDI and HDRS.
There were some studies that failed to demonstrate that ECT had a preventative effect on suicide in the long term. According to a national clinical survey in England, Hunt et al. (2011) concluded that the reduction in ECT use did not influence suicide rates in patients who received this treatment by collecting an 8-year sample of suicide cases (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006). Black et al. (1989) Sun et al. (2016) and (2018) applied the same stimulation MST protocol and assessed the SSI to evaluate the suicidal ideation after 24 MST sessions, or until remission of depressive symptoms. Both trials reported significant reduction in suicidal ideation.
Two studies from the same team highlighted the efficacy of left unilateral (10Hz) or bilateral (left: 1Hz; right: 10Hz) DLPFC rTMS on reducing depression in patients with TRD (Blumberger et al., 2011. Two years later, they integrated data from these two trials and determined that bilateral rTMS was superior to sham rTMS in reducing suicidal ideation in patients with TRD with a significant odds ratio (OR) = 3.03 after fifteen sessions over 3 weeks (Weissman et al., 2018). Similar to the results observed with ECT, inconsistencies in rTMS studies were also reported. Despite some studies reporting beneficial effects, some others failed to report any superiority of active rTMS over sham (Baeken et al., 2019;Croarkin et al., 2018;Desmyter et al., 2014;George et al., 2014).

| Results of tDCS study
This single tDCS RCT showed significant decreases in suicidal ideation. A significant decrease in suicidal thoughts measured by MADRS 10th item was observed after 10 daily tDCS sessions over sham (p = .04) (Brunoni, Júnior, et al., 2013;Brunoni, Valiengo, et al., 2013). This study suggests that patients with depression and suicidal ideation could benefit from tDCS.

| D ISCUSS I ON
In the present review, we have examined the current status of ECT, MST, rTMS, and tDCS in the treatment of suicidal ideation, suicide attempts, and completed suicides.

| ECT studies
Among the fourteen retrieved studies, five ECT experimental studies for MDD reported that patients who had received ECT showed significant faster and larger effects on suicidality than control patients groups (pharmacotherapy (Avery & Winokur, 1978;Nordenskjöld et al., 2013;Kellner et al., 2005) or rTMS (Keshtkar et al., 2011) or compared with baseline (Veltman et al., 2018). whereby ECT considered a first-line treatment for acute suicidal ideation (Milev et al., 2016). A review (Sharma, 2001) of the role of ECT in suicide prevention also recommended the consideration of ECT in treatment algorithms to rapidly reduce suicide rates. The outcomes of these studies seem to be encouraging, and their level of evidence was relatively high (superior to PEDro level 2).
The eight other ECT studies are retrospective or observational studies investigating the occurrence of suicidal behavior in cohorts of patients who received ECT. With a PEDro level 3, their conclusions were less convincing than those from two RCTs and two prospective trials. Among them, three studies did not observe any influence of ECT on completed suicides and suicide attempts (Black et al., 1989;Hunt et al., 2011;Popiolek et al., 2018). discrepancies observed between studies regarding the long-term effect of ECT in their meta-analyses. Difference may be due to the assumption that there is a relapse in depressive episodes, rather than "incomplete" recovery and psychomotor retardation after ECT, despite an expected good short-term effect on these symptoms (Brådvik & Berglund, 2004). Sharma, 2001(Sharma, 2001

| MST studies
The aims of Sun et al. (2016) and (2018) were to explore indicators or neuroplasticity of remission of suicidal ideation following MST. With the similar schedule as ECT, usually two to three times per week, an index course of twelve sessions (Milev et al., 2016), we could assume that these two studies evaluated the suicidal ideation at the baseline, and 2 or 3 months after the intervention. The results on suicidal ideation of participants with TRD were reasonable and logical because of its similar antidepressive effect compared with ECT (Kayser et al., 2011).
According to CANMAT, its acute efficacy on MDD was recommended with evidence level of 3 (Milev et al., 2016) but its overall recommendation, especially in suicidality, needs to be further explored.

| rTMS studies
As observed with ECT studies, our review revealed inconstancies across rTMS studies. Hence, four study effects with PEDro level over 2 concluded that rTMS may have an antisuicidal effect (Berlim et al., 2014;Hadley et al., 2011;Wall et al., 2011;Weissman et al., 2018) whereas four others ( However, iTBS is thought to create more robust neuroplasticity effects that likely become evident later following treatment due to the commended neurophysiology (Chung et al., 2014). The evaluation was likely done too soon to find a significant difference between these two groups. Regrettably, Desmyter et al. (2016) designed a crossover trial to evaluate the difference between these two groups after the first week of stimulation to avoid carryover effects that could be expected during the second week. They applied iTBS as an intervention; however, they failed to estimate its profound influence on suicidal ideation after the intervention. To confirm the placebo response, Baeken et al. (2019) reported a significant frontopolar prefrontal perfusion decrease after 4 days of sham iTBS which was related to the attenuation of suicidal ideation. The assumption was that the placebo effects interfered with the final results, but they would only emerge in terms of the suicidal symptoms, but not the depression. Furthermore, due to a lack of meta-analysis specifically exploring the antisuicidal effects of rTMS, we have not successfully sought out evidence of higher levels to appraise these findings.
Considering that this intervention was recommended by the U.S. Food and Drug Administration (FDA) for the treatment of resistant depression since 2014, accelerated DLPFC-rTMS is clinically acceptable and could be one of the therapeutic options in addition to its antidepressant effects. Thus, further research is necessary to extend to more naturalistic conditions, and larger sample to confirm its superior antisuicidal effect over sham.

| tDCS study
According to the one available study, tDCS could be a choice to decrease suicidality. Another study found that tDCS combined with sertraline was more effective than tDCS, or sertraline alone, on depression. Following this trend, probably the best treatment for suicide might be to combine different interventions, such as pharmacotherapy, to decrease suicidality of patients with depression (Brunoni, Júnior, et al., 2013;Brunoni, Valiengo, et al., 2013).

| Safety
In reference to safety, while both rTMS and tDCS are generally considered safe and acceptable if the appropriate guidelines and recommendations are followed, more research is needed to verify the safety parameters of rTMS and tDCS for the purpose of preventing suicidality. The adverse events of ECT potentially could be prevented. It is possible that the routine application of an ECT checklist could diminish the rates of adverse events.

| Limitations
Our aim was to evaluate the efficacy between five physical treatments (ECT, MST, rTMS, tDCS, and CES) on suicidality. Due to the limited number of studies and the high heterogeneity between protocols, we were not able to complete a quantitative synthesis and to undertake a meta-analysis. The efficacy of seizure therapies and NIBS as a continuous intervention also should be further explored as Nordenskjöld et al. (2013) and Popiolek et al. (2018) were the only two studies discussing the long-term effect of ECT as continuation treatment. None of the MST, rTMS, or tDCS studies investigated MST and/or rTMS and/or tDCS as a continuous intervention to assess its long-term influence after the acute course of intervention sessions. With a PEDro level of 3 in eight of them, a lack of control group, weak design, and possible cohort effects were obvious methodological shortcomings. Additionally, another obvious limitation of the present review constitutes selection bias. We restricted our search strategy to full-text articles in English. For example, a systematic review of Norwegian papers revealed doubt upon the efficacy of ECT in the prevention of suicide. We could not critically evaluate those articles or challenge the reported data. Moreover, as mentioned earlier, we had to exclude six studies whose Sackett Scale was equal or inferior to 4 (Brunoni, Júnior, et al., 2013;Brunoni, Valiengo, et al., 2013;Fligelman et al., 2016;Gambill & McLean, 1983 (George et al., 2014). These symptoms were lack of further evaluations and discussions as related phenotypes.

| CON CLUS ION
Taking into consideration all the current evidence, we support the effect of ECT for acute suicidal ideation, but we cannot recommend the same regarding MST, rTMS, or tDCS. However, further research is necessary to determine whether there is any clear and persuadable evidence for the long-term efficacy of ECT on suicidality and to investigate the potentiation of ECT-elicited neuroplasticity changes targeting suicide theory. The antisuicidal effect of rTMS with a standard protocol needs to be confirmed. There is also a need for high-quality tDCS and CES trials designed for suicide. Combination therapy, both in short term and long term, could be a promising asset toward a faster and more sustained amelioration of suicide. We are deeply grateful to all participants.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare. All authors contributed to and have approved the final submitted version of the manuscript.

AUTH O R CO NTR I B UTI O N
Authors YC and CM were involved in the conception, design of the review, and screened the articles; Author YC analyzed the data and drafted the manuscript, figures, or tables under the guidance of CM and EP. Authors CM, EP, EL, and JB provided writing assistance and proof reading the article. All authors contributed to and have approved the final submitted version of the manuscript.

E TH I C A L A PPROVA L
All analyses were based on previous published studies; therefore, no ethical approval and patient consent are required.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/brb3.2144.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed for supporting the findings of this study are available from the corresponding author upon reasonable request.