Suicide Risk Assessment, Management, and Mitigation in the Emergency Setting

Suicide is a leading cause of death worldwide. It is a serious and complex public health issue affecting families and communities (1). Effective suicide prevention requires a systems approach integrating both community and health care services—a systems approach focusing on interconnections across actors, factors, and sectors. By improving how these elements operate independently and together, suicide rates can be reduced (2).

Screening, assessment, and mitigation are the key clinical issues in suicide prevention. Screening aims to find the few people at risk within a large group (presumed to be at baseline risk). Assessment aims to determine whether a specific individual, suspected to be at greater risk, is at significant risk. Mitigation aims to reduce the risk of suicide or of a serious attempt for a person assessed to be at significant risk. There is no way to achieve these aims with perfect reliability, but there is evidence that some approaches are more effective than others. For a population, the Zero Suicide conceptual approach and practice (3, 4) is a systemwide approach to improve care and outcomes for individuals at risk of suicide and an aspirational goal for health care systems (5).

Suicide screening specifics are important. A positive screen for suicide may be the impetus for a visit to an emergency setting. Knowledge of the screening instrument, its setting, its strengths, and its weaknesses, may direct assessment. A worrisome screening with the Patient Health Questionnaire–9 (PHQ-9) in a private internist’s office (6) has different implications than does a yes response to a police officer’s question about suicidal thoughts as part of arrest paperwork.

Assessment is well understood by most practitioners. All training in psychiatry includes attention to signs and symptoms suggesting that a patient might die by suicide. However, changes in standards of care and documentation still merit review. It is expected that clinicians will ask about access to guns, not only because of the risk to the patient but also the potential for violence to the public. Indeed, given the high fatality rate of index attempts (first attempts) using a gun, routine questions about gun access are a useful addition to any assessment template. It can be argued that limiting gun access may be a particularly effective strategy to minimize violence to the community—not just because guns are more fatal for patients.

Mitigation of suicide risk is increasingly important to reduce emergency department (ED) boarding for patients awaiting psychiatric admission. Many patients do not need inpatient care when support and monitoring and contingency plans can be put in place—e.g., they may be safe at home pending a clinic appointment if supportive family are present, weapons are removed, and crisis services are available. Some patients may even be guided past a temporary crisis with brief psychotherapeutic intervention or resumption of medications. Homelessness is often the primary reason for suicidal thoughts; a bed in a shelter reduces this risk.

For an individual patient, a complicated mix of findings, risks, and interventions may be in play. Some factors are changeable, some are not. Suicidal thoughts may be related to transient circumstances, e.g., acute intoxication or an argument with significant others. Age and chronic illness are risk factors that are effectively constant. Social isolation and access to firearms are risk factors that may or may not be mitigated, depending on local resources and laws. No evidence-based screening or assessment tools exist that take all this complexity into account. Care of individual patients depends on good clinical assessment.

Suicide Screening

Routine screening is a key step in identifying individuals at risk of suicide to ensure assessment and appropriate care. Standardized screening can increase detection of suicidal ideation by 600% and reduce suicide attempts by up to 80% in high-risk populations (7). In 2007, The Joint Commission (TJC), issued an alert to medical centers of increased scrutiny on suicide and promulgated the National Patient Safety Goal 15.01.01 for suicide prevention (8, 9). Effective July 1, 2019, TJC required screening for suicidal ideation with a validated screening tool for all patients being evaluated or treated for behavioral health conditions as their primary problem. TJC did not require screening all patients. TJC added that it was important to assess medical-surgical patients whose prognosis or psychosocial issues put them at risk of suicidal ideation (10). Meanwhile, the United States Preventive Services Task Force has recommended depression and anxiety screening for children and adolescents, and it is working on recommendations for adults (11). The Centers for Medicare and Medicaid Services has made depression screening a quality measure for primary care federally qualified health centers (12). The accumulation of all these screening rules have led some organizations to introduce universal screening (13).

Some flexibility remains in determining screening protocols and selecting screening tools. Four screeners are on the TJC approval list (14–25) (Table 1). Different organizations make different recommendations about when to screen patients, depending on setting and expected level of risk (26, 27). The American Academy of Pediatrics has noted:

The Zero Suicide approach advocates that all patients are screened for suicide risk on their first contact with the organization and at every subsequent contact. All staff members should use the same tool and procedures to ensure that clients at risk of suicide are identified (28).

Primary care settings usually opt for single-item questions, and settings where patients are more likely to be at risk of suicide use a more comprehensive assessment of ideation, behavior, and intent to die (7, 13). The most widely used single-question item is item 9 of the PHQ-9, which asks, “Over the past 2 weeks, how often have you been bothered by . . . thoughts that you would be better off dead or of hurting yourself in some way?” Response options include not at all, several days, more than half the days, or nearly every day (15). The PHQ-9 was constructed as a symptom checklist corresponding to DSM criteria for major depressive disorder as part of the Primary Care Evaluation of Mental Disorders (PRIME-MD) (15). The question has been validated with clinical interviews and has demonstrated an ability to detect an increased risk of suicide among patients (17). There is also a single item on the Symptom Checklist-90. To achieve brevity, screeners exclude multiple factors associated with elevated suicide risk in favor of triggering a second level of screening or a more complete assessment. Simple, single-item measures of suicidal thoughts or behaviors are associated with misclassification and fail to distinguish between aborted, interrupted, and actual suicide attempts (29).

The Ask Suicide-Screening Questions (ASQ) (19, 20, 30) and the more extensive variation on which the screener is based, the Suicide Ideation Questionnaire (31, 32), are examples of rapid screening tools that gauge suicidal ideation and that are used by trained staff. The ASQ was initially developed to assess suicide risk in pediatric and young adult patients in the ED and has been validated for all ages (19, 20). It contains four questions: In the past few weeks, have you wished you were dead? In the past few weeks, have you felt that you or your family would be better off if you were dead? In the past week, have you been having thoughts about killing yourself? Have you ever tried to kill yourself?

Any positive response should lead to a suicide safety assessment by a trained mental health clinician to determine whether an evaluation is needed. However, the link between ideation and suicidal behavior has been inconsistent, and the association with death is understudied (7).

The Suicide Behaviors Questionnaire–Revised (SBQ-R) is a brief screener that assesses different dimensions of suicidality, including lifetime ideations and attempts, frequency of ideation in the past year, current threat, and the likelihood of suicidal behaviors (22). The SBQ-R grew out of work by Linehan and Nielsen (33) in the 1980s as they were exploring parasuicidal behavior and suicide attempts. Their work led to a series of four questions: Have you ever thought about or attempted to kill yourself? How often have you thought about killing yourself in the past year? Have you ever told someone that you were going to commit suicide or that you might, do it? How likely is it that you will attempt suicide one day?

This rating system was validated with a mix of adults and adolescents, inpatients, and nonpatients (22). The scores very reasonably separated the individuals who were suicidal (e.g., inpatients with appropriate diagnoses) from the individuals presumed not to be suicidal. The study did not follow the patients over time, so the predictive value is less certain (23).

The Columbia-Suicide Severity Rating Scale (C-SSRS) extended efforts to assess suicidality during drug trials (34, 35). C-SSRS was systematically validated with adolescents and adults (36). Later, it was adopted by the Department of Defense and TJC in 2016, leading to widespread use for all ages. Only three questions are always required: In the past week, have you wished you were dead or wished you could go to sleep and not wake up? In the past week, have you had any actual thoughts of killing yourself? Have you ever done anything, started to do anything, or prepared to do anything with any intent to die?

All patients with high-risk answers require patient safety precautions until a mental health professional can evaluate the patient. Although the C-SSRS has demonstrated predictability of short-term suicidal behavior, it does not classify intent and imminence (37). The C-SSRS screens for passive suicidality, followed up by questions of active suicidal ideation, plan, intent to act, and suicidal behavior (24). The C-SSRS is multidimensional, but it is not comprehensive and has shown inconsistent accuracy of risk classification, resulting in mounting concerns about its widespread adoption (38, 39).

Prevention of suicides on medical units and behavioral units for patients who are suicidal raises its own screening challenges. New tools are looking to integrate artificial intelligence techniques with traditional screening tools to improve detection (40). Evidence of the effectiveness and shortcomings of existing screeners, as well as evidence of promising new tools, continues to grow. Organizational needs may limit the options to just one across an entire health care system. It may be time to pause this advance to allow research and technological advances to weigh in on the most effective ways to affect lives, health care settings, and care systems.

The effort to handle screening false positives is another challenge, even assuming a health care organization accepts the cost of screening itself and arranging treatment for suicidal individuals. False positives may automatically trigger a consultation request for assessment and add the burden of managing these potential patients to an already busy clinic or ED setting. Neither staff nor patients appreciate the extra time involved. Wilson et al. (39) proposed an approach for EDs called ICAR²E as an evidence-based and practical way to manage potentially suicidal patients: identify suicide risk, communicate, assessment for life threats and ensure safety, risk assessment of suicide, reduce risk of suicide, and extend the care beyond the ED. Stanley and colleagues (41) evaluated the AIM model for outpatient mental health clinics: assess, intervene, and monitor for suicide prevention. No general solution is available based only on screening: screening is only one step in a process and cannot replace a comprehensive suicide risk assessment.

Assessment in Emergency Settings

Prelude to Assessment

Information from outside sources is often part of the prelude to a suicide assessment—e.g., weather and road conditions before a one-car accident and accessibility of the top of the cliff. Unfortunately, all sorts of useful information may be lost because of a general misconception that psychiatric clinicians can assess suicidal risk in the same way that internists assess cardiac risk: a few questions, a few physical findings, and a few lab tests, irrespective of anything else that happened earlier in the day.

Definite, deliberate, dangerous actions, with no obvious benefit to anyone else, lead to the presumption of suicide risk. This is especially true for hanging attempts with rope or a sturdy belt. Suicide assessment under these circumstances is reduced to determining whether the patient is likely to try again even within a treatment or controlled setting.

Indeterminate actions and ambiguous statements may lead to particularly challenging suicide assessments. The actions may be of uncertain intent—e.g., a patient who has insomnia and takes two dozen pills over the course of a night and is then found unconscious by family. The report may be of uncertain validity—e.g., a patient reports walking over to a bridge and then reconsidering, but there is no corroboration by friends or family. Clinicians for such cases are challenged to understand the realities of the patient and of the patient’s motivations, along with traditional risk factors.

The length of time from suicidal thought to act is often considered to be clinically significant; impulsive actions are less concerning than an attempt made after a period of deliberation and a chance for distraction. However, a study by Paashaus et al. (42) raised doubts about this clinical belief. In addition, there are the genuinely concerning patients who state, “I wish I’d died from my attempt.” Neither the time to reconsider nor the discomforts of the attempt and medical treatment have dissuaded them.

Suicidal statements and screening results are a bit simpler, because the context is clear or can be obtained from family, friends, or providers. This allows focus on mood, mental state, motivation, and traditional risk factors.

Interviewing in the Crisis Milieu

Most clinicians try to interview a patient once they understand the circumstances of the assessment. Ideally, this will yield a good understanding of a patient’s environmental and internal psychological experience. Interviewing can benefit from a bio-psycho-social-cultural-spiritual approach that allows the patient to trust the clinician, even in an ED, where neither clinician nor patient know each other and where stakes are high (death, hospitalization). Unfortunately, patients may have many reasons not to reveal their suicidal thoughts: concern about privacy, a reluctance to share thoughts with a stranger, shame, and fear of being involuntarily hospitalized or medicated. Standardized questions from a computer or a form are no panacea; they may be perceived as perfunctory and disrespectful (43).

An empathic interview with open-ended questions is crucial for patients who feel overwhelmed or have trouble putting their feelings into words. Such an approach may also help patients who are reticent or guarded or who flatly deny suicidal thinking yet leave a clinician with a strong suspicion that crucial information is being withheld. An empathic approach does require time, patience, and skill to establish rapport and to build an alliance—not just to obtain answers. Conducting such an interview may seem too time-consuming for an ED setting, but it may save time by allowing more nuanced conclusions and safe outpatient interventions. A patient who feels that someone listened and understood may be more willing to try to engage in outpatient treatment. Offering a glass of water or a blanket can be a concrete action that builds an alliance with the patient.

Patients occasionally refuse to participate in an interview or other assessment. Such refusal is commonplace in emergency settings when patients are brought against their will. Patients who give an incoherent or implausible story present a similar challenge. The clinician is left to assemble a reasonable account of the patient’s story from available sources: screening reports, ambulance staff or police reports, family reports, and such. Furthermore, a patient’s refusal to participate has meaning but is ambiguous and must be understood in context. A prisoner may try to delay returning to jail by hindering assessment. An incoherent patient may be implicitly confirming intoxication, ingestion of medications, delirium, or severe mental illness.

When patients dispute the report that brought them to assessment, additional collateral information becomes critical. There may have been some misunderstanding. Not everyone interprets screening questions the same way. Covering clinicians may not be familiar with a patient’s provocative style. A count of the remaining pills may confirm that the patient did not try to overdose.

Occasionally, friends and family lie to force patients into treatment. Sometimes families are desperate to arrange treatment even under circumstances in which local laws do not provide for involuntary treatment. Sometimes friends are vengeful. Luckily, these situations are rare; however, clinicians cannot expect that people will always be truthful and thus need to remain alert.

Even when patients and collateral sources cooperate to give an intelligible story, the clinician must judge whether the account is a legitimate story. “I’ve been depressed since I was born, and today I finally decided it was time to kill myself” might be accurate, but it is not a story, even by a young child’s standards. A slight variation yields a story: “I’ve been depressed for years, but I tried to tough it out until I had a job with insurance. Yesterday, I was supposed to have an intake appointment at the clinic, but my therapist developed appendicitis—they can’t reschedule me for two months—it’s a sign there is no hope for me.” A short call to the clinic could confirm this account.

Common Reasons for Suicide

Many different and unfortunate stories lead to suicidal thoughts, intent, and plans for many reasons. Finding the reason guides treatment and disposition. Below a way to group them is described.

Unbearable emotional pain.

Several authors (44–46) have described an affective “deluge” with an “urgent need for relief” and “escape” from intense feelings that wash over the patient. The feelings may be of shame, humiliation, self-hate, panic, anxiety, and rage. These may overlap with severe “depression,” the classic example of severe emotional pain. The feelings may be related to patient’s current living situation—e.g., shame of losing one’s home that is not ameliorated by having shelter.

Frightening psychosis.

Hallucinations, especially threatening or derogatory voices, and paranoia that convinces the patient that suicide is the only way out are common reasons for suicide (47). The precipitating event might be severe enough to break down an individual’s usual coping skills. This is where a psychodynamic understanding of the patient’s personality structure can assist in assessing the level of risk.

Severe loneliness or feeling utterly alone.

In loneliness, one remembers a connection to someone or something soothing but misses it. Alone is a desperate—existential at times—and unbearable feeling of being utterly without any connection to anyone or anything, even a deity or other meaningful social connection. It can drive a state of terror and suicidal despair. Loneliness is part of thwarted belongingness in the interpersonal theory of suicide (48, 49).

Revenge fantasies.

“They’ll see what they did to me.” The fantasy of being able to see, after death, the suffering of others who have wronged them (“They’ll suffer when I’m gone”) is another reason for suicide, as is the idea that suicide is a punishment (“If I die, then they’ll be sorry for what they did to me”) and that the person’s death can stimulate guilt and remorse on the part of the perceived or real perpetrator. For these patients, pointing out that their belief in their omniscience may not be accurate may help turn them around. “What if no one notices? What if no one comes to the funeral?” It may free up enough anger and righteous indignation to relieve a patient of thoughts of suicide, although the task is then to help the patient tolerate the rage that those thoughts were blocking and assess risk toward others or a quick relapse into another state of rage turned onto the self.

Other affective states.

Other affective states, such as dissociation, usually result from severe early life trauma, abuse, or neglect. A wish to escape danger or flee from these intense affects drives the suicidality.

Contingent suicidal threats.

Some patients make an attempt to control or manipulate with the threat of suicide. “I’ll kill myself if you don’t admit me to a detox program.” This is implicitly more interactive than suicide for revenge.

Nonsuicidal self-injury.

Patients cut themselves for emotional relief. This is not necessarily a stepping stone to suicidal activity. The dissociated or otherwise overwhelmed person may attempt to get relief from the “psychache” by a physical and concrete act (50), which may explain why the sight of blood relieves the patient. Some patients cannot perceive their body as part of themselves, and by extension they feel more “alive” and relieved of the mental distress by discharging it into a physical expression. This may explain why these patients often feel no pain from their self-injurious behavior until hours later. However, friends or family or others may interpret such injuries as suicide attempts. A challenging variation is when a nonsuicidal injury becomes life threatening.

Physical pain.

Physical pain or the fear that it will become worse and unremitting can lead a patient to believe that suicide is the only solution. This is more traditionally an issue for consultation-liaison psychiatry services working with patients with cancer.

Risk Perspective in Emergency Psychiatry

It is possible to systematically review an individual patient’s risks, even though there is no validated way to quantify a specific patient’s probability of suicide at a specific moment in time (51, 52). Statistical approaches exist for estimating the number of suicides over an interval for a set of patients with specified characteristics (53). Unfortunately, patients and families rarely present for statistical treatment.

People are more accustomed to reasoning by example or storyboard style—event-by-event review. Begin with the reason for suicide. Unbearable emotional pain from slowly worsening depression sets the stage for weeks of trouble. Mood states wax and wane, and the patient may feel better after talking with an empathic clinician and may be willing to attend a clinic for treatment. However, if depression worsens over a weekend, if the patient starts to wonder if feeling hopeful was foolish, what will prevent a suicide attempt? Patients and family can usually understand this sequence of events and plan accordingly. They may decide that the patient needs to be watched in some protective environment (a stay with family members or a stay in the hospital).

Storyboard review may also clarify minimal-risk scenarios. Consider a married patient with no past psychiatric history sent for assessment late Friday afternoon because of universal screening at a gastrointestinal clinic. An accompanying spouse will be with the patient throughout the weekend, and the spouse is sure that there are no weapons in the home and will take the patient to the family doctor on Monday if there is any hint of trouble. Otherwise, the spouse plans to discuss today’s events further with the patient and will ask the family doctor for a referral if there is any suggestion of depression and not just bureaucratic misunderstanding.

A structured review of empirically established risk factors augments an empathic interview. Risk factors should be pursued not only during interviews with patients but also during inquiries of family, friends, and treatment providers. Only one source might remember and mention a suicide attempt by the patient in the distant past or a relative who died by suicide. Likewise, not everyone may report firearms at home or know of their existence. No matter how empathic the interviewer, no matter how thoughtful the storyboard review, the clinician should not neglect any established risk factors. Both Gold (51) and Simon (52) have provided detailed lists of such risk factors. One should also be explicit about possible current precipitants, especially events leading to humiliation, shame, despair, loss of personal relationships, and family turmoil or chaos—any or all of which could lead to social isolation.

Proposed risk factors, such as being a veteran or being elderly, should be unpacked conceptually into their biopsychosocial implications. Life experiences can build resilience and bring wisdom. However, with age, a person’s friends may pass away, health may wane, and retirement savings erode. Some veterans keep weapons. The underlying, established risk factors are the same as for other patient populations.

Protective factors for suicide (such as forward thinking, religious beliefs, or familial responsibilities) have been less frequently examined than have risk factors, and thus less empirical evidence exists (54, 55). Social support, cultural and spiritual connection, emotional intelligence, the ability and opportunity to talk with family or friends about problems, and mental health treatment are associated with decreased suicidal behavior. Although no universal formula allows us to calculate the balance between risk factors and protective factors, studies among Native American youths have demonstrated that increasing one protective factor had an impact that was similar to decreasing three risk factors (55). Thus, protective factors are more usefully considered as part of risk mitigation plans.

The point of careful risk assessment is effective risk mitigation. Keep in mind that any patient may die in the next week without any suicide risk factors, most commonly in America because of accidental injury (56). Therefore, it makes sense to steer our patients away from any avoidable risks that become evident during an interview. Suicide assessment implies thoughtful inquiry, documentation, and reasoning about why death in the near term is or is not likely due to preventable actions by the patient. Even medium-risk patients may die if their underlying condition is not improved and there is no safety net. Elevated-risk cases include a wide range of patients. Some can be kept reasonably safe with a tight support network and immediate treatment; others are so despairing or psychotic that they are likely to attempt suicide even in a controlled hospital environment.

Mitigation of Suicide Risk

Mitigation of suicide risk or means reduction takes many forms—some will work for whole populations, and some are specific to individuals. Reducing publicly available quantities of acetaminophen by requiring blister strip packaging has reduced the number of poisonings (57). Blocking access to dangerous places or objects is effective, such as nets and restricted access to the Golden Gate Bridge in San Francisco (58). Means reduction on an individual scale might involve turning over pills to a trusted family member and having firearms removed from the home. Call-in crisis lines may help at the population and individual levels if the telephone numbers are known. Mann et al. (5) reviewed the evidence supporting these and other mitigation maneuvers. Sometimes, exploration and partial resolution of a crisis in an emergent session can mitigate risk.

Hospital admission is considered the default approach to risk mitigation. Psychiatric units are expected to be able to protect people from themselves; admission usually passes the responsibility for patient safety along to another treatment team. Unfortunately, admission is of little help to patients for whom the regressive pull of an inpatient unit is a problem. Admission is also of little help to patients who must board in an ED for days or weeks waiting for an opening in a psychiatric unit.

Pressures unrelated to any individual case may push for admission to reduce risk of malpractice liability, which worries administrators and staff more than wrongful imprisonment suits. There may be countervailing pressure for discharge: ED overcrowding or patients, friends, and family hoping to avoid the unpleasantness of ED boarding or time in a locked ward.

Preventing suicide or injury can be challenging even before an assessment is complete. Hospitals are expected to secure ED areas that hold suicidal patients (10). This requires attention to physical hazards, such as ligature points, and staff to observe patients (59). Other issues are less obvious. What belongings should be removed besides sharp ones? Should patients change into hospital gowns and be examined for wounds or marks? How should the period after an assessment and before admission be handled?

Approaches to keeping patients safe from self-harm within an emergency setting include enhanced observation, escorts to bathrooms, and removal of potential weapons (pens, nonplastic eating utensils, belts, shoelaces, etc.). Patient belongings may be searched and even locked away. Some EDs require suicidal patients to change into hospital garb—examination robes or Johnny coats—but this is controversial. There is a tension between patient safety and dignity. Depending on the potential degree of danger, constant one-on-one observation may be required. For less risky patients who are not acutely suicidal, one staff member might be able to keep a few patients in direct line of sight without using a video monitoring installation.

Intoxicated patients who are reportedly suicidal can present a difficult challenge. Busy EDs are poorly suited environments to hold such patients waiting for psychiatric evaluation. However, until the patient is clinically sober, evaluating risk properly is difficult (59).

Patients may ask that friends or family not be notified; however, clinicians may override that request if the clinical situation is deemed an emergency. During emergencies, safety takes precedence over privacy (60). Third parties are often crucial informants of a patient’s recent behavior and state of mind. When a disparity exists between what a collateral source reports and what a patient says, it is appropriate to explore the presenting problem.

Gun inquiries can be difficult with patients, family, and friends. As Rozel and colleagues (61) noted, “As with other culturally sensitive subjects (e.g., reproductive health, substance use, etc.), basic awareness of varying attitudes, behaviors, and beliefs surrounding gun ownership in the United States may help clinicians empathize and ‘meet the patient where they are.’ Firearm ownership in contemporary America is often a cultural issue with distinct political, social, and religious differences between owners and nonowners that impacts ownership, use, and storage. . . A gun-naive clinician may not even realize cultural differences at play in their interactions with a patient or family until they have inadvertently damaged their rapport.”

More concretely, these authors suggested reframing the typical question—“Do you own a gun?”—to something focused more on the patient: “You have an illness that can sometimes cause problems with emotions, decisions, and your sense of hope. Whenever this occurs, I am concerned about suicide and aggression as a risk, no matter how unlikely. You are too important to take chances with. May we talk about your access to guns?”

Clinicians can also take a cue from inquiries about alcohol consumption and ask, “How many guns do you have at home?” rather than “Do you own a gun?” The first implies that gun ownership is common and not in and of itself a cause for concern.

If discharge from the emergency setting is contemplated, common sense argues for means reduction pertinent to an individual’s circumstances. Guns, weapons, and poisons should be secured or removed from the home. Necessary medications can be dispensed by family, friends, or visiting nurses. An intoxicated patient can be accompanied home by family or friends. For patients being discharged, Betz and Boudreaux (59) recommended follow-up calls the next day; an appointment time with a clinic or provider for follow-up; a discussion about the safety of weapons, if present; and a safety plan with telephone numbers.

Safety plans are increasingly recommended to reduce suicide behavior by providing suicide warning signs, compiling coping strategies, and linking to supports in a written or app-based plan (62–64). From a psychodynamic perspective, a written or printed safety plan is a transitional object (65, 66). The more a clinician can engage the patient in preparing it, the more likely it will be effective. Some EDs use peer support workers to help patients formulate safety plans. Common lived experience adds credibility.

Safety plans have been described by the Department of Veterans Affairs and other authors (62, 63). A safety plan begins with a list of warning signs for the patient—i.e., a list of moods, thoughts, behaviors, situations, and so forth that are harbingers of a suicidal crisis for the specific patient.

To create the plan, the clinicians should sort out the following with the patient: Internal (personal) coping strategies, such as going for a walk or listening to music; social contacts for distraction, e.g., going with friends to a coffee shop or game; people who can be asked for help, e.g., family members and good friends; professionals who can help, such as the patient’s counselor or other treatment provider, a local crisis center, and even the National Suicide Hotline (988) and locally relevant numbers (211 in some states). The discussion might end with instructions to call 911 and to go to the nearest ED for uncontrollable impulses.

Immediate preparation, before the next hint of crisis includes means restriction, such as eliminating guns and lethal medications from the home.

Patient safety plans should be written out and reviewed, with role playing, to increase the odds that the patient will be able to use the plan in a looming crisis.

Careful readers may notice that a safety plan’s first steps implicitly depend on protective factors found during assessment. Strong spiritual beliefs suggest that reciting a koan or favorite passage from a holy text might be a useful coping strategy. Family and social supports figure directly into social coping plans, as do pets.

A recent meta-analysis by Nuij and colleagues (67) showed the effectiveness of safety plans in preventing suicide attempts, supporting the inclusion of such plans in clinical guidelines for suicide prevention. Safety planning is meant to be a conversation with the patient and family to develop meaningful plans and not mechanically ticking off checkboxes. It should be distinguished from a suicide contract, whereby patients promise to contact a crisis line or the patient’s mental health provider if feeling suicidal, because these such contracts have been shown to be ineffective (68). The evidence for safety plans includes Stanley and Brown’s (63) combination of telephone outreach with safety planning. Telephone outreach may carry most of the effect, because connecting with another person and the belief that one matters to another person may constitute the main protection from suicidal impulses.

The process of developing a safety plan may yield evidence to support the plan or may show that a different approach is needed. If the patient says there is no point in calling anyone because no one cares about him, perhaps the patient is more isolated than the clinician suspected. A clinician can inquire about a patient’s prior connections. Does the patient remember a time when he did feel he could call people who cared? A recollection of better times and a willingness to try to reconnect bode well for the plan. If the patient cannot summon any recollection of connection, this plan may not be sufficient mitigation.

Finally, documentation is a critical part of an evaluation: clear rationale for admission versus discharge; summary of acute and chronic risk factors, mitigating factors, and modifiable risk factors; description of risk assessment, creation of a safety plan, and efforts to minimize risk; and advice to the patient regarding follow-up. Documentation must be particularly robust for a patient who continues to report suicidal ideation but whom the clinician feels is not at imminent risk (suicidal statements made to police in route, threats made chronically without action, and a clear history of secondary gain) or admission for an individual who claims not to be suicidal.

Evidence has been building that caring contacts after psychiatric hospitalization or ED discharge reduce suicide attempts (69). This low-cost, high-yield intervention calls for eight periodic and personalized messages to patients expressing concern for their well-being without demands or expectations. They are personalized based on important dates to the patient. Initially, these took the form of letters. Postcards and text messages are now being used effectively. Examples of postcards are available from Zero Suicide (70). Potential mechanisms include bolstering a sense of connection and social support, improving emotion regulation function by mitigating feelings of worthlessness, and increasing suicide prevention literacy.

Summary

A death by suicide is not the inevitable result of suicidal ideation. Several factors can limit risk at the population level, as well as for individual patients. Screening is an important component to identify individuals at risk, some of whom may not report suicidal thoughts as a chief complaint. Assessment includes a careful patient interview, pursuit of collateral information, and assessment of risk. Treatment, including pharmacotherapy, psychotherapy, means restrictions, and follow-up (caring contacts), may sufficiently reduce the risk of suicide so that hospitalization is not inevitable for every patient with suicidal ideation or who arrives in an emergency setting after a suicide attempt. Suicidal ideation is often transient, and factors contributing to suicidal thoughts are modifiable. Finally, it is essential to document concisely the rationale for admission versus discharge.