Introduction
The management of suicide risk is often integral to the care of people with treatment-resistant depression (TRD). Suicide is a major public health concern worldwide, and yet it is a problem that remains poorly understood. Suicide occurs across a wide range of clinical presentations, including TRD, but the mechanisms accounting for suicide risk in TRD are not fully understood. Nevertheless, treatment guidelines for suicide risk in TRD are beginning to emerge, although more evidence is needed. In the sections that follow, we provide a brief overview of the phenomenology, risk, and treatment of suicidal thoughts and behaviors (STBs), and then we make recommendations for best practices for managing suicide risk for those with TRD.
Terms and definitions
Prior to considering management of suicide risk in TRD, it is worthwhile to define and clarify several key constructs related to suicide. Suicide, also referred to as suicide death, refers to the intentional ending of one’s own life. A discussion of STBs can be structured according to a recently developed theoretical model describing the pathway to suicide. This model begins with thoughts about suicide and identifies the sequence of thoughts and behaviors that culminate in a suicide attempt ( ). Suicidal ideation refers to thoughts about one’s own death or suicide. Suicidal ideation is often considered either passive (i.e., the wish to no longer be alive) or active (i.e., the wish to take one’s own life). Although active suicidal ideation is typically considered the more risky of the two, recent research suggests that passive and active ideation may be somewhat equivalent in terms of risk for suicide attempts, as well as other outcomes ( ). A suicide plan refers to the selection of a method and place to attempt suicide ( ). Suicide attempt refers to an intentional act undertaken with some intent to die as a result, regardless of whether death results. Nonsuicidal self-injury (NSSI; often referred to as “self-injury” or “self-harm”) refers to self-directed, harmful acts without intent to die as a result, most often performed for the benefit of reducing distressing psychological states, such as negative emotions ( ). NSSI is different than suicide, and is not the focus of this chapter, but NSSI is common among those with treatment-resistant depression ( ), and is a top risk factor for a future suicide attempts ( ).
Suicide prevalence
Suicide is a public health concern of enormous magnitude. In the United States, suicide is the 2nd leading cause of death among those ages 10–35, and the 10th leading cause of death among all age groups . In 2018, over 48,000 Americans died by suicide. Around the world, approximately 800,000 people die by suicide each year, according to the . Suicide accounts for 1.4% of all deaths worldwide, making it the 18th leading cause of death per year globally ( ). To put these numbers into perspective, suicide accounts for more global deaths each year than war, automobile accidents, AIDS, and homicide combined ( ). In addition to suicide death, there are also an estimated 25 million suicide attempts, and 140 million people engaging in suicidal ideation annualy around the world ( ).
Suicide prevalence rates vary between countries, and some data have suggested that suicide is more common in developed versus developing countries ( ). However, attempts to better understand this variation have not provided additional insights. Perhaps more illuminating is that several demographic characteristics seem to be commonly associated with suicide rates internationally. For example, males are more likely to die by suicide, whereas females are more likely to attempt and think about suicide ( ; ; ). In addition, younger individuals engage in STBs at a higher rate than older individuals ( ).
Some recent epidemiological data suggest that suicide may be on the rise, with rates in the United States increasing for both males and females between 2000 and 2016 ( ). This alarming statistic has drawn warranted attention from the media. However, considering a broader span of time adds nuance to the conclusion that suicide rates are increasing. During the 20 years prior to 2000, suicide rates in the US actually decreased ( Fig. 39.1 ).
Unlike many of the other leading causes of death worldwide, including cancer, heart disease, Alzheimer’s disease, stroke, and violence, we have not been able to effect a meaningful reduction in suicide rates. Our ability to generate knowledge about suicide and suicide prevention has been hampered in at least three ways. First, suicide is an extremely low base-rate phenomenon, with fewer than 11 out of every 100,000 people dying by suicide annually ( ). From a statistical perspective, as discussed in greater detail below, predicting the occurrence of such a rare phenomenon requires very large samples of research participants. Second, those who die by suicide cannot be directly studied using the tools most common in mental health research. Although we can examine morphological or biochemical abnormalities in the brains of suicide victims, these approaches have not revealed major insights about the phenomenology of suicide, or how we may prevent it. Third, suicide has not been studied systematically for very long, and research efforts have largely focused on a small number of constructs theorized to cause suicide ( ). A primary goal of this chapter is to highlight what we do know about suicide generally and among those with treatment-resistant depression, and to discuss future research priorities to better understand and prevent this enigmatic tragedy.
Suicide risk factors
Identifying risk factors for suicide is one approach toward better understanding the phenomenology of suicide, as well as more effectively predicting suicide and intervening before it occurs. A risk factor refers to a prospective predictor of an outcome of interest. Risk factors precede the outcome in time, and they predict whether the outcome is likely to occur by dividing people into high- and low-risk categories ( ). This is different than a correlate, which is associated with the outcome of interest, but the nature of that association is unclear. For example, if a cross-sectional study identified sleep loss as a correlate of suicidal ideation, it would be possible that suicidal ideation caused sleep loss—or vice-versa. Longitudinal data collection methods are therefore necessary to identify risk factors by establishing temporal precedence. It is important to note that identifying a risk factor does not necessarily indicate that the risk factor is causal. Causal risk factors are a specific type of risk factor in which manipulation of the risk factor creates an observable change in the likelihood of the outcome ( ). Very few studies have attempted to influence risk factors to determine causal relationships with suicide, and thus our current knowledge of suicide risk is largely informed by risk factors not yet known to be causal.
A recent meta-analysis of the past five decades of suicide research, which included data only from longitudinal studies of suicide risk factors, has provided some of the most compelling evidence to date about risk for suicide ( ). This meta-analysis examined the prospective relationship between risk factors and STBs (suicidal ideation, suicide attempt, and suicide) across 365 separate studies. The relationship between each risk factor and STB outcome was quantified in this study with a weighted odds ratio. This weighted odds ratio provided an estimate of how much more likely the outcome is to occur for those who are high in the risk factor, while taking into account the size of the sample used for the original analysis. Odds ratios above one indicate that outcome is more likely to occur for those higher in the risk factor, whereas the opposite is true for odds ratios less than one. An odds ratio of one indicates that the outcome is just as likely to occur for those who are higher vs lower in the risk factor (i.e., the outcome is predicted at chance level by the risk factor).
Depression has long been considered a risk factor for suicide, and indeed, countless studies have demonstrated an association between depression and STBs. However, found that depression was one of the five strongest risk factors only for suicidal ideation. For suicide attempt and suicide death, depression (categorized together with other forms of internalizing psychopathology) was rated sixth and ninth among the strongest risk factors, respectively. Data from other studies (some of which were included in the meta-analysis) have suggested that the effect of depression on increased suicide risk may not result from depression itself, but rather from the conditions that are commonly comorbid with depression. For example, two studies that included large, representative samples (one from the United States, one international) found that depression was one of the strongest risk factors for suicidal ideation. However, among those with suicidal ideation, depression predicted future suicide attempt and suicide death no better than chance, whereas other common mental disorders, especially those characterized by anxiety and agitation (panic disorder, posttraumatic stress disorder) or by poor impulse control (substance use disorders), were far better predictors than depression of whether ideators would go on to make a suicide attempt ( ). Moreover, the likelihood of making a suicide attempt was shown in each of these studies to be higher for those with greater numbers of comorbid mental disorders ( ). In other words, individuals with one mental disorder were less likely to attempt suicide than those with two mental disorders, who were less likely than those with three, and so on. Together, this body of evidence suggests that depression alone often leads to thoughts about death and suicide, but other conditions lead depressed individuals to act on these thoughts. This may be particularly important for TRD, wherein comorbid mental disorders are particularly common ( ).
According to the meta-analysis, prior self-injurious thoughts and behaviors (including suicidal ideation, suicide attempt, and NSSI) were among the strongest risk factors for all STBs outcomes, as shown in Table 39.1 .
Rank | Risk factor | wOR | CI |
---|---|---|---|
Top 5 risk factors for suicidal ideation | |||
1 | Prior suicide ideation | 3.5 | (2.6, 4.8) |
2 | Hopelessness | 3.3 | (1.5, 7.2) |
3 | Depression | 2.5 | (1.4, 4.3) |
4 | Abuse history | 1.9 | (1.6, 2.3) |
5 | Anxiety | 1.8 | (1.3, 2.4) |
Top 5 risk factors for suicide attempt | |||
1 | Prior NSSI | 4.2 | (2.9, 6.9) |
2 | Prior suicide attempt | 3.4 | (2.7, 4.3) |
3 | Suicide risk screening | 2.5 | (1.8, 4.4) |
4 | Axis II diagnosis | 2.4 | (1.9, 2.9) |
5 | Prior psychiatric hospitalization | 2.3 | (1.6, 3.4) |
Top 5 risk factors for suicide death | |||
1 | Prior psychiatric hospitalization | 3.6 | (2.8, 4.5) |
2 | Prior suicide attempt | 2.2 | (1.7, 3.0) |
3 | Prior suicide ideation | 2.2 | (1.5, 3.4) |
4 | Low socioeconomic status | 2.2 | (1.3, 3.7) |
5 | Stressful life events | 2.2 | (1.6, 2.9) |
Prior psychiatric treatment, including inpatient hospitalization, is another top risk factor for suicide ( ). The week following inpatient hospitalization is one of the highest risk periods for suicide attempts ( ). Although the factors that contribute to the significantly elevated risk of suicide in those recently discharged have not been adequately delineated, the nature of the reasons for inpatient hospitalization (e.g., STBs, emotional distress), and other challenges posthospitalization (e.g., persistent symptoms, disappointment with lack of greater improvement during hospitalization) are plausible contributors to risk.
Additional categories of risk factors, including physical illness and demographic characteristics were shown by to be poor predictors of suicide. In fact, their effects on suicidal ideation were close to chance. In practical terms, this means that by flipping a coin, one could predict a person’s suicidal ideation almost as accurately as one could by knowing that person’s ethnicity or their history of asthma.
Alternative methods for suicide prediction
A fundamental goal of understating suicide risk is to be able to predict suicide so that we can ultimately prevent it. This is indeed a tall order. To predict the real-life occurrence of suicide requires identifying a person who is at risk and/or a context that puts an at-risk person in imminent danger of attempting suicide. Until recently, predicting suicide largely relied on self- or clinician-reported risk factors, like those discussed above. For example, in hospital emergency departments across the globe, clinical staff assess suicide risk by asking patients to self-report about risk factors like STB history, current suicide plans, and intent to die by suicide. Relying on self-report in this way is a method of suicide prediction with some limitations. First, those who are at risk are often motivated to conceal or deny their true intentions to act on suicidal thoughts. Revealing suicidal intentions can be viewed as shameful, and for those with strong intentions, communicating these can lead to unwanted intervention. One study found that 78% of people who died by suicide denied suicidal ideation immediately prior to their death ( ). Second, suicidal thoughts are now known to be transient, varying in presence and intensity throughout a single day ( ; ; ), and so self-reported current suicide risk may not be an accurate indicator of risk a week or even a day later. Third, assessment of suicide risk through self-report often relies on patients’ STB histories (i.e., assessing past STBs, which are a top risk factor, as discussed above), but these self-reports are based on retrospective recall, which can be biased and inaccurate ( ). Nonetheless, among suicide attempters presenting at a psychiatric emergency department, self-reported prediction of future suicide attempts (i.e., the patient responded to the question “what is the likelihood that you will make a suicide attempt in the future?) is a significant predictor of future suicide attempt within a 6-month follow-up period, even when controlling for key risk factors, such as depression and history of prior suicide attempts ( ). As such, self-reported suicide risk is likely the most useful and easily available indicator of suicide risk within a 6-month time frame.
Clinical care decisions for suicidal patients also typically rely on the subjective judgment of a clinician. This judgment is often based on intuition, but also the experience of seasoned clinicians. However, clinician prediction of future suicide attempt does not outperform actuarial prediction of suicide based on known risk factors, and neither of these predict suicide better than chance ( ). Although this finding casts doubt on the ability of clinicians to accurately predict suicide risk using their own subjective clinical judgment, it accords with prior research showing that clinical judgment often has limited utility for predicting important mental health outcomes ( ).
A more recent method for assessing suicide risk that has gained attention is the assessment of implicit cognitions about death and suicide. Implicit cognition refers to the influence of memories or perceptions on thoughts and behaviors that is outside of conscious awareness. In the same study by referenced above, the research team compared a measure of implicit suicide-related cognition against self-reported, clinician-reported, and actuarial prediction of suicide attempts. They found that, even after controlling for the predictive effects of each of these alternative methods, a measure of implicit suicide-related cognition identified those who would go on to make suicide attempts 6 months into the future. Subsequent studies have supported the predictive validity of suicide-related implicit cognitions ( ; ). Implicit cognition has two major benefits over more traditional suicide risk prediction methods. First, implicit cognitions, which are measured using responses to stimuli calculated down to the millisecond, are considered a more objective measure of suicide risk since they are resistant to explicit attempts at concealing stigmatized thoughts ( ). Second, implicit cognitions are shown to have exceptional specificity, in contrast to the vast majority of other prospective risk factors, which have low specificity for suicide. For example, two well-known risk factors for suicide, depression and sleep disturbance, each respectively predict anxiety and accidental falls better than they predict suicide ( ; ). However, optimizing the specificity of methods for suicide prediction is only half the battle.
Specificity, as exemplified above, refers to the ability to correctly identify the absence of the outcome of interest (i.e., true negatives; correctly classified nonattempters), but highly specific tests are prone to failing to identify those with the outcome of interest (i.e., false negatives; misclassified nonattempters). Sensitivity refers to the ability to correctly identify those with the outcome of interest (i.e., true positives, correctly-classified attempters), but highly sensitive tests are prone to misclassifying the absence of the outcome of interest (i.e., false positives, misclassified attempters). A major limitation of these previously-discussed approaches is finding the “optimal” balance between sensitivity and specificity. Ideally, we want to identify and intervene for a small number of people (high sensitivity) who are almost definitely going to attempt suicide (high specificity). Failing to intervene for people who actually attempt suicide because of low sensitivity is obviously problematic, but intervening for lots of people who aren’t truly at risk because of low specificity unduly consumes valuable clinical resources and can lead to unnecessary involuntary hospitalizations. Self-report and behavioral measures of risk tend to be highly sensitive, but not highly specific. This is because suicide is an extremely low base rate behavior, and so the odds ratios often used to indicate risk for suicide can be misleading, increasing the difficulty of clinical decision-making. Artificial intelligence, otherwise known as machine learning (ML), approaches are beginning to show promise by combining a broad array of risk predictor types while offering the ability to determine the tradeoff between sensitivity and specificity.
Although ML approaches can enable researchers to measure a broad range of risk factors simultaneously (from electronic medical records and other rich sources of data) from extremely large and representative samples, ML has found only limited practical benefit beyond that of the more commonly used suicide risk prediction approaches already discussed. For example, surveyed outpatient mental health clinics’ data and used an ML approach to classify individuals’ level of risk based on key information obtained from their health records. Their approach revealed extremely high odds ratios for the risk metrics included, but misclassified many patients as high risk at the highest levels of sensitivity (only 5%–10% of those classified as highest risk by the algorithm actually attempted suicide). used a similar ML approach to classify individuals’ risk for a first suicide attempt based on five health care systems’ electronic medical records. This study also found high odds ratios for the risk metrics they surveyed, but their approach misclassified even more patients as high risk at the highest levels of sensitivity (only 1%–6% of those classified as highest risk by the algorithm actually attempted suicide). These results indicate that ML approaches can be highly sensitive, and so they are capable of correctly identifying nearly all truly high-risk people, but at the cost of erroneously identifying truly low-risk people as high risk. The clinical utility of ML approaches for suicide risk prediction essentially depends on the intervention that is offered. If, for example, a hospital system using ML to detect suicide risk among its patients planned to utilize a low-cost intervention (i.e., interventions requiring little human labor and low risk to the rights of patients), such as a weekly phone check-in for all high-risk people, then the ML approach can provide practical benefits. However, if the hospital system planned to use a higher-cost intervention, such as involuntarily hospitalization, for patients classified as high risk, then ML approaches to suicide risk prediction would be unsuitable in their current state. However, researchers are working to improve ML methods, and so this promising area of research may soon hold tremendous practical value to clinicians and hospital administrators.
Suicide risk in TRD
Although it may seem intuitive that TRD is associated with increased risk for STBs, very few studies have examined the association between suicide risk and TRD. In part, the dearth of studies in this area may stem from the lack of definitional consistency of TRD, or the possibility that TRD, though a notable clinical challenge, may not be a categorically distinct syndrome from other forms of depression (e.g., Major Depressive Disorder in an individual who responds to the first several treatments or persistent depressive symptoms that fail to meet diagnostic criteria for a major mood disorder and do not respond to multiple treatments). Nonetheless, in this section, we summarize what is currently known about suicide risk in TRD, and then provide theory- and evidence-based speculation about why suicide risk might be particularly high in TRD.
Approximately 30% of patients with TRD attempt suicide at least once in their lives ( ). This figure is staggering in its own right, but when considered in relation to nonresistant depression, the problem of suicide in TRD is even more concerning. Compared with depressed patients whose symptoms either responded to intervention or had not yet failed to respond to two or more interventions, those with TRD experience a twofold increased rate of suicide attempts, and a 10-fold increase rate of suicide death ( ). There are several possible reasons that STBs occur in such high numbers among those with TRD, and these can be organized under the theoretical framework of one the most well-cited theories in suicide research: the Interpersonal Theory of Suicide ( ; ). According to the Interpersonal Theory, three interconnected factors cause STBs: thwarted belongingness, perceived burdensomeness, and hopelessness. Meta-analytic results support first-order and interacting associations between each of the theory’s proposed risk factors and suicide ( ). Below, we argue that each of these risk factors are likely to be high among those with TRD, and may be at least partially responsible for the high rates of STBs observed in this population.
The need to belong is a fundamental part of human nature. Thwarted belongingness refers to the unmet need to belong, resulting in loneliness. Social isolation, conflict with others, having a small social network, and social withdrawal are each components of thwarted belongness, and are common in depression ( ; ; ; ; ). Some of these aspects of thwarted belongingness are also elevated in TRD ( ). A considerable body of evidence indicates that thwarted belongingness predicts STBs, and interacts with both perceived burdensomeness and hopelessness in its association with STBs ( ). Additionally, the association between thwarted belongingness and STBs is especially strong in late adulthood ( ), when risk for suicide in TRD is particularly high ( ). Taken together, these findings indicate that thwarted belongingness may be one mechanism linking TRD with STBs.
Suicide risk among those with TRD may also increase because these individuals construe themselves to be a burden on others (i.e., perceived burdensomeness), such as significant others and treaters. The construct of perceived burdensomeness encompasses the perception that one’s life is less valuable to others than one’s death. Perceived burdensomeness is thought to be driven by the eusocial nature of human beings, who, like other eusocial animals, depend on finite, communal resources and cooperation among individuals to ensure a group’s survival ( ). Tenets of inclusive fitness suggest that under certain environmental scenarios, such as threat to the health or resources of a group, self-sacrifice can aid in the group’s survival by reducing the threat to the group as a whole. This form of self-sacrifice has been observed in a wide array of nonhuman animals, such as bees, naked mole rats, and shrimp ( ). Human suicide can be considered a self-sacrifice similarly intended for social benefit when an individual perceives that their continued life prevents others from consuming needed resources, as mentioned above. A large number of studies have shown that self-reported perceived burdensomeness predicts suicide ( ), and is associated with depression ( ). Because impairment in functioning among those with TRD often necessitates the provision of resources from others, perceived burdensomeness in TRD may increase suicide risk, though no studies to date have investigated this hypothesis.
The final and perhaps most obvious reason that STBs are so common in TRD is that TRD and STBs are each linked with hopelessness. Hopelessness, broadly defined as negative expectations about the future and often encompasses the belief that problems are inescapable, is a central component in multiple suicide theories ( ; ). Hopelessness has been shown by some studies to be a relatively strong predictor of future STBs ( ; ). With such a history of failed attempts at symptom remediation, it is intuitive that those with TRD are prone to experience hopelessness ( ). It is likely that those with TRD begin to feel hopeless because their attempts to find symptom remediation have failed to yield sustained positive changes and, instead, have been met with repeated disappointment. They may also feel hopeless because of their belief that concerns related to thwarted belongingness and perceived burdensomeness, as described above, are intractable ( ). These individuals may see the future as increasingly bleak, and may estimate their chances of recovery to be very low. It is also possible that the relationship between hopelessness and TRD may be bidirectional, such that hopelessness results from repeated treatment nonresponse, but also predicts persistence of depression symptoms after treatment ( ). If so, this deleterious cycle may erode the motivation to seek treatment, and increase the appeal of suicide as a means of escape.
Treatment of suicide in TRD
To date, no treatments have been developed to prevent STBs specifically in TRD. However, several therapies, ranging from pharmacological, to neurostimulation, to behavioral, have shown some promise for reducing STBs in depressed patients, and in some cases, TRD. Below we outline many of these promising approaches, and provide speculative commentary on their utility for treating STBs in those with TRD.
Pharmacological treatments
Intravenous administration of ketamine has recently received considerable scientific and clinical attention as a treatment for depression, culminating in the approval by the Food and Drug Administration (FDA) of its isomer, esketamine, administered intranasally for resistant depression. Multiple studies have shown that a single dose of ketamine reduces depressive symptoms in as many as 70% of depressed patients, and that these antidepressant effects are apparent in as little as 24 h after administration ( ; ). Accordingly, the rapid pace and potential efficacy of ketamine has become an exciting area of investigation for STB remediation, as well. In a sample of TRD patients with histories of STBs, a single-dose administration of ketamine was associated with a significant decrease in the intensity of suicidal ideation 24 h after administration, though this difference may have been accounted for by reductions in depressive symptoms, which were also clinically significant ( ). In addition, this study found that ketamine administration was associated with a reduction in implicit escape-related cognitions. A similar follow-up study found that implicit suicide-related cognitions, as well as explicit suicidal ideation also decreased following administration of ketamine, relative to a psychoactive placebo ( ). It is important to note that one study found no association between ketamine and implicit death-related cognitions, though this study did find that repeated ketamine administration was associated with reductions in explicit suicidal ideation ( ). Nonetheless, these findings are encouraging for two reasons. First, ketamine and esketamine may be effective treatments for those facing acute suicidal emergencies, such as those presenting to an emergency department following a suicide attempt, because ketamine and esketamine work rapidly, relative to more traditional antidepressant medications. Second, findings that ketamine may impact implicit cognition related to suicide provides an indication as to one mechanism by which this medication may exert its antisuicidal effect—namely, by reducing automatic thoughts about escape and death. Although more research is needed to determine whether ketamine or esketamine should be administered as a first-line treatment for those with TRD and STBs, the results reviewed above suggest that these approaches may be promising.
Lithium is another pharmacotherapy shown to be effective for reducing STBs in depressed patients, and may be a fruitful approach for this purpose in TRD. In fact, one recent meta-analysis found that lithium had a larger effect for STB reduction than for mood symptom reduction, suggesting potential specificity of lithium for STBs ( ). Lithium is thought to exert antisuicidal effects through two possible mechanisms. First, and most relevant to TRD, lithium may help prevent relapse of depressive symptoms, thereby limiting ongoing psychological distress, such as hopelessness. Second, lithium is shown to reduce agitation and impulsivity ( ; ), which are associated with STBs, as mentioned above ( ). It is noteworthy, however, that the efficacy of lithium for reducing STBs has primarily been documented relative to placebo, and lithium’s possible antisuicide efficacy relative to other pharmacotherapies in individuals with unipolar and bipolar mood disorders is an area of continued investigation.
Some research indicates that clozapine, an atypical antipsychotic usually prescribed for symptoms of schizophrenia or other psychotic disorders, may be effective for preventing suicide in high risk patients. In a study comparing clozapine with olanzapine, researchers recruiting psychotic patients (schizophrenia or schizoaffective disorder diagnoses) from 11 different countries found that those treated with clozapine had a lower rate of suicide attempts over a 2-year follow-up period ( ). The patients treated with clozapine also required fewer hospitalizations following suicide attempts, as well as fewer acute rescue interventions, suggesting that clozapine may have implications for reducing the lethality of suicide attempts. Of note, over one quarter of the patients in both treatment groups in this study were considered treatment-resistant at the time of enrollment ( ). One study has since corroborated the efficacy of clozapine for the treatment of STBs in schizophrenia, finding that clozapine was associated with reductions in both suicidal ideation and suicide attempts, based on retrospective medical record reviews ( ). Although intriguing, it is unclear whether these results would generalize to patients with TRD, since neither study examined the efficacy of clozapine for reducing STBs in depressed patients without a psychotic disorder, and the only treatment compared to clozapine was olanzapine, which is now infrequently prescribed in unipolar depression. Regardless, the putative mechanisms by which clozapine protects against STBs may be relevant to TRD. Clozapine is shown to reduce substance use in schizophrenia patients, likely by altering dopamine-mediated reward processing ( ). This impact on substance use is particularly noteworthy, given that depressed patients with comorbid substance abuse are at elevated risk for suicide ( ).
Although the use of opioid medications to treat depressive symptoms fell out of common clinical practice in the 1950s (due to increasing concerns about abuse and overdose, as well as the discovery of tricyclic antidepressants and monoamine oxidase inhibitors), one opioid medication, buprenorphine, a mu opioid agonist and kappa opioid antagonist, could help treat STBs in patients with TRD. A study showed that patients treated with buprenorphine experienced a significantly larger reduction in suicidal ideation than patients who received placebo ( ). Reductions in suicidal ideation among the group treated with buprenorphine were apparent within a week of beginning treatment, and these reductions persisted through the conclusion of the 4-week trial. Critically, the sublingual dose of buprenorphine used in this study was very low (mean daily dose of 0.44 mg) and posed virtually no risk for overdose, and so outpatients were able to fill prescriptions of buprenorphine outside of the hospital study sites. This presents a major advantage over many of the other pharmacotherapies described above, such as ketamine, which must be administered repeatedly under direct medical supervision in order to achieve antisuicidal effects. Given that a recent meta-analysis indicated that buprenorphine is effective for reducing depressive symptoms in TRD patients ( ), buprenorphine may be a promising treatment for STBs in this population as well, although more research is needed.
Neurostimulation treatments
In addition to the pharmacological approaches discussed above, three neurostimulation approaches have also been evaluated as suicide prevention measures in those with depression: electroconvulsive therapy, deep brain stimulation, and vagus nerve stimulation. Electroconvulsive therapy is the most well-studied of these three neurostimulation approaches. Electroconvulsive therapy is shown to be highly effective for remediating depressive symptoms in TRD ( ), but historically was considered a last-line suicide prevention among chronically suicidal depressed patients, due to potentially harmful side effects, including memory loss and seizures. However, published guidelines by the American Psychiatric Association in 2001 promoted the use ( ) and continued refinement of ECT, leading to widespread utilization of electroconvulsive therapy as a treatment for TRD ( ). Since then, several studies have demonstrated that electroconvulsive therapy may be effective for reducing in STBs among depressed patients ( ; ; ).
Though less well-established as a treatment for TRD than electroconvulsive therapy, studies suggest that vagus nerve stimulation is associated with reduced incidence of STBs in depressed patients. For example, one study of TRD patients showed that the group treated with vagus nerve stimulation experienced significantly fewer STBs than patients in the “treatment as usual” group ( ). Additional studies have also reported suicide attempt rates in samples of TRD patients treated with vagus nerve stimulation ( ; ; ). Although the number of STBs in these samples is promisingly low, it is impossible to conclude from these studies whether vagus nerve stimulation had a discernible impact on STBs in TRD, and so more work is needed in this area.
Similar to electroconvulsive therapy, little empirical research has investigated deep brain stimulation for reducing STBs. Two anecdotal reports of deep brain stimulation in depressed patients suggest the possibility that deep brain stimulation may be associated with increased suicide risk ( ; ). However, other studies have concluded that the small number of STBs in TRD patients treated with deep brain stimulation during clinical trials was unrelated to this procedure, though no statistical inferences were provided to support this this lack of association ( ; ). Thus, the efficacy of deep brain stimulation for reducing STBs in TRD remains an open question.
Finally, repetitive transcranial magnetic stimulation is another neurostimulation approach that may help reduce STBs among depressed patients. Repetitive Transcranial Magnetic Stimulation was approved by the Food and Drug Administration in 2008 as a treatment for TRD. Since then, many studies have demonstrated its efficacy in remediating depressive symptoms ( ). A recent retrospective study of patients who had been treated with repetitive transcranial magnetic stimulation showed that patients’ self-reported suicidal ideation reduced significantly during the course of their treatment ( ). Because this study was retrospective in nature, and suicide attempts were not measured, these results should be considered preliminary evidence for the effectiveness of repetitive transcranial magnetic stimulation for treating STBs.
Behavioral treatments
Cognitive Behavioral Therapy is one of the most commonly utilized forms of psychotherapy worldwide. Cognitive Behavioral Therapy aims to provide relief from the challenges of psychopathology and emotional suffering through a combination of strategies for emotion identification/regulation and in vivo exposure to distressing emotional states. A large body of evidence indicates that Cognitive Behavioral Therapy is effective in reducing STBs ( ; ; ). Some evidence also suggests that Cognitive Behavioral Therapy is effective for remediating depressive symptoms in TRD ( ); however, no studies have evaluated changes in STBs after Cognitive Behavioral Therapy for TRD patients.
Dialectical Behavior Therapy combines many aspects of Cognitive Behavioral Therapy with mindfulness- and acceptance-based skills. Dialectical Behavior Therapy was created for self-injurious (those with STBs as well as NSSI) Borderline personality disorder patients who struggled with the putatively less flexible aspects of traditional cognitive behavioral therapy ( ). Dialectical Behavior Therapy has been evaluated as a treatment for STBs against treatment as usual ( ; ), waitlist ( ), community-based psychotherapy ( ), and more direct suicide risk management approaches ( ). Although some of these studies have provided evidence for the efficacy of Dialectical Behavior Therapy in reducing STBs ( ; ; ), results are mixed ( ; ). Though no studies to date have examined whether Dialectical Behavior Therapy is shown to reduce STBs in TRD, Dialectical Behavior Therapy is shown to reduce depressive symptoms in TRD ( ).
Additional psychotherapeutic interventions, such as Interpersonal Psychotherapy, have shown some promise for reducing STBs ( ), and may be effective for TRD ( ). However, more research regarding other therapy interventions are needed in order to identify them as candidates for reducing STBs in TRD.
Additional best practices for risk management
Patients experiencing a suicide crisis (i.e., an increase in the intensity of suicidal ideation that may include a plan and/or intent to act on these thoughts) require immediate intervention. Pharmacological, neurostimulation, and behavioral therapies are often insufficient for managing acute suicide risk without additional targeted interventions. Several clinical best practices, as well as novel crisis interventions for managing suicide risk are important to discuss in the context of this chapter on managing suicide risk among those with TRD.
Regular assessment of suicide risk is of paramount importance for clinicians treating any population at elevated risk for suicide, such as TRD. Because suicidal thoughts and intentions can vary from moment to moment ( ), clinicians cannot rely on a patient’s recent history as a determinant of current risk, and should seek to evaluate current suicidal ideation, plans, intentions, and access to methods, as well as any recent preparatory behaviors for suicide (e.g., stockpiling medications, purchasing weapons, etc.), recent nonsuicidal self-injury, and recent suicide attempts. These thoughts and behaviors, which represent the most concerning group of suicide risk factors ( ), should be assessed very regularly with a patient with TRD. Contrary to the somewhat popular misconception, asking questions about STBs is not associated with an increase in suicidal thoughts for the patient ( ), and so clinicians should not feel deterred from asking about suicide for fear of negatively impacting the patient’s risk. On the contrary, direct risk assessment using available information (e.g., patient interview) is perhaps the most accurate and readily available tool to identify the need for a clinician’s support or a higher level of care. Similarly, thorough documentation of regular risk assessment, and of crisis interventions administered, is important for clinicians. By documenting the information obtained through risk assessment interviews, as well as all interventions offered, clinicians can create a record of a patient’s changes in risk throughout therapy, and protect themselves from potential liability.
For those deemed to be at suicide risk through self-report of STBs or other risk factors, clinicians and patients should collaboratively agree upon a safety plan. A safety plan is a list of strategies an at-risk patient can take to keep themselves safe from suicide, NSSI, substance abuse, or other harmful behaviors. The steps of a safety plan are sequential, such that patients should start the first strategy listed, and move on to subsequent strategies if the last strategy attempted was ineffective. A template of a safety plan is shown in Fig. 39.2 .