Psychiatric Annals

The articles prior to January 2012 are part of the back file collection and are not available with a current paid subscription. To access the article, you may purchase it or purchase the complete back file collection here

Structured Assessment Methods May Improve Suicide Prevention

David Mays, MD, PhD


Standard patient interview processes can mislead clinicians about acute risk factors.


Standard patient interview processes can mislead clinicians about acute risk factors.

It is not surprising that having a suicidal patient raises anxiety levels in clinicians. Most malpractice cases against psychiatrists arise from the failure to prevent suicide. These cases also account for the highest percentage of settlements and verdicts paid by malpractice carriers.1 Although an enormous amount of space in the professional literature has been devoted to the topic, uncertainty reigns as to what constitutes a legally and clinically adequate approach to the suicidal patient.

By contrast, risk assessment of aggressive behavior is a field that has generated some specific models of evaluation during the past decade. These developments have arisen largely through the recognition that actuarial tools permit a meaningful degree of prediction of behavior in certain populations.2 The Violence Risk Appraisal Guide (VRAG), the Hare Psychopathy Checklist, and other tools have provided the groundwork for progress in this area.

Attempts to construct similar actuarial instruments for suicide prediction have not been successful for several reasons. Suicide is a much rarer event than aggression, and prospective studies with controlled designs have not been possible for various ethical and practical reasons. Precise definitions of what constitutes suicidal behavior also have been problematic. In addition, relationships and interactions of risk factors are not clearly understood in spite of quite sophisticated statistical analysis. The bottom line is that no actuarial instrument can be constructed to help the clinician predict the occurrence of suicide.3-5

Legally, the courts have acknowledged that suicide cannot be predicted but have held that a clinician may be liable for damages if the suicidal act was "foreseeable." Foreseeability is defined as the reasonable anticipation that harm or injury is a likely result from certain acts or omissions. In general, experts have interpreted this to mean that the risk of suicide is foreseeable, but the act of suicide is not.6 The fine distinction here may or may not be intuitively obvious, but the legal conclusion is that an "adequate assessment" for the risk of suicide must be performed. Although the court has not specified what constitutes an adequate assessment, at least one court has determined that a simple risk assessment checklist does not suffice.6,7


It is a relevant question to ask how well clinicians perform suicide risk assessments. We have no way of knowing how many suicides are prevented each year by skillful clinical assessment and intervention. Undoubtedly, there are many. However, suicide rates during the past century show remarkable stability (Figure), despite two epidemics of suicide, one during the major influx of immigrants at the beginning of the 20th century, and the other during the Great Depression. This stability is quite surprising, given that the 20th century has seen the rise of modern psychiatry, with the American Psychiatric Association founded in 1921, the development of crisis intervention as a field of study in the 1950s, the discovery of modem psychotropic medications (chlorpromazine in 1953, Imipramine in 1956, and fluoxetine in 1987), and the adoption of Assertive Community Treatment models in the 1960s. The suicide rate in 1900 was 10.2 per 100,000 people. In 2000, it was 10.7 per 100,000 people.8 In short, there is no obvious evidence that the psychiatric profession has been particularly effective in preventing suicide.

Another way of approaching this question is to perform retrospective reviews of the treatment records of completed suicides and follow the thinking of the clinician involved. In the state of Wisconsin, a 1989 state law (Wisconsin Act 336) has enabled state officials to track many Wisconsin suicides, and review records of care. We reviewed all such suicides for the year 1996 (n - 60). Charts were obtained for 41 cases. The progress notes in these charts were examined for the year before the last appointment. Cases were then classified into one of three categories. Cases in which the clinician was negligent showed no evidence of any suicide assessment at any time. Those deemed unavoidable showed evidence that an assessment had been performed and the clinician concluded that the patient was suicidal. Some steps were taken by the clinician to prevent the patient's suicide. The remaining cases were categorized as unexpected. These cases provided evidence that a suicide assessment had been performed and the clinician did not believe the patient was suicidal. Nonetheless, suicide occurred.

In our review, seven charts (17%) were placed in the negligent category, 11 (27%) charts were placed in the unavoidable category, and 23 (56%) were placed in the unexpected category. That is, most of the time, the last assessment of a patient who would soon complete suicide indicated that there was no suicide risk.

This is not an isolated finding. The Suicide Data Bank of the American Foundation for Suicide Prevention is an organization designed to gather and analyze information from therapists who have had a patient complete suicide. In 26 investigated cases, they found that the therapist did not recognize the patient as suicidal in 14 cases, or 54%.9 In a different study of charts and clinician and patient interviews of 53 caregivers who had a patient attempt suicide, me vast majority of clinicians did not note suicide risk as a significant problem.10

The unavoidable conclusion is that in the final assessment of patients who complete suicide, clinicians usually do not believe suicide is an issue. The next reasonable question would then be concerned with what clinicians actually do to assess risk.

In an effort to find an answer, 128 crisis-intervention workers in Wisconsin were surveyed by questionnaire in 1997. They were asked:

* How often do you assess patients for suicidality? Every contact, every month, every 6 months, every year, when indicated?

* What factors are most important Ln your assessment? Demographics, patient mental state in the interview, "gut feeling" from the interview, collateral information from family and friends, history of attempts?

* How confident are you in your ability to assess for suicidality?

* Have you ever had a patient complete suicide?

US Suicide Rate per 100,000

US Suicide Rate per 100,000

Ninety-eight percent of respondents were quite confident of their ability to assess for suicide. Sixty-two percent assessed for suicide whenever they felt it was indicated, and their assessments overwhelmingly were based on the immediate interview - both the mental state of the patient and their own "gut feelings." Thirty-six percent of respondents had experienced a patient suicide.

In a follow-up survey, five ease studies were presented to 124 crisis-intervention workers at their annual state meeting. These case studies were constructed to reflect differing degrees of suicide risk in five areas: demographics, history of previous violence or suicide attempts, reports from family and friends, mental status, and feeling from the interview. For example, a patient might be demographically at high risk for suicide but be convincing in the interview that he or she did not have suicidal ideation. The respondent was then asked whether he or she would take some sort of suicide precaution with the patient and why.

The results of this study confirmed clinician feeling about the interview and the degree of symptomatology shown by the patient during the interview have primacy over all other data, including suicidal history, concerns from family and friends, or highrisk demographics. An interview that leaves good feelings in the clinician will always overshadow any other data.

A review of the literature is consistent with these results. Mental health professionals are usually quite confident in their suicide risk-assessment skills. In a study previously mentioned, 82% of mental health professionals who had recently had a patient attempt suicide felt a high degree of confidence in meir assessment ability.10

Additionally, a review of die meager literature dealing with how clinicians actually assess for suicide confirms that clinicians prefer to rely on interviewbased data, usually directly about suicidal behavior. In a survey of 81 psychiatrists from Ontario, Canada who were asked to rank the relative importance of factors in their determination of suicide risk, the most important factor was a feeling of hopelessness in die patient, followed by suicidal ideation or plan.11 In a separate survey of 94 mental health professionals selected randomly from the American Association of Suicidology, the American Psychiatric Association, me American Psychological Association, and the National Association of Social Workers, the most important clinical questions in a suicide assessment concerned the presence of a plan with an available method, suicidal ideation, substance abuse, and a previous attempt.12 There is some evidence, however, to suggest this may not be the most useful way of approaching a suicide risk assessment.


It is important to clarify what we know about suicidal behaviors. Consider the four main types: ideation, planning, communication, and previous attempts. Data gathered from the National Comorbidity Study13 indicate each of these behaviors is associated with a different risk of suicide. While the lifetime prevalence of suicidal ideation is relatively common (13.5%), the risk of completed suicide among ideatore is only 0.5%. The cumulative probability that a "thinker" will become a "planner" is 34%. Only 10% of planners and 10% of attemptors will go on to complete suicide. This risk is low, but substantially higher than if a person only has suicidal ideation.13

Communication of suicidal intent apparently occurs before most suicides, primarily to relatives, not caregivers. In one study of completed suicides, 60% had communicated suicidal behavior to a spouse, 50% to relatives, and 18% to a healthcare professionaL14 This underlines the importance of collateral information in suicide assessment.

In other words, of all suicidal behaviors, suicidal ideation is the least likely to be associated with completed suicide. It is also the least reliable. In a nationwide chart review of 76 patients who completed suicide while inpatients or immediately after discharge, 77% denied suicidal ideation at their last clinical interview.15 A narrower analysis of these charts is detailed in the article in this issue by Fawcett and Busch (see page 357).

A prospective study looking at 25 suicides among 954 psychiatric patients in clinics of five US universities found that suicidal ideation was not correlated with completed suicide within I year of the assessment.16 In a prospective study of 813 depressed patients in the Mood Disorders Unit at Prince Henry Hospital in Sydney, Australia, the 31 suicide completers were found to have less suicidal ideation at their initial interview than those who did not complete suicide.17 In a retrospective review of 571 completed suicides in Finland, only 22% were found to have mentioned suicidal intent during their final contact with a healthcare professional.18


This lack of correlation of suicidal ideation with suicide completion seems at first to be counterintuitive, but upon some reflection, it is not. It is important to remember that the clinical interview represents a patient's mental state during a very brief period of time. It is a snapshoL not a movie. It does not and cannot provide a comprehensive view of a patient's Ufe outside the interview room. The clinical interview can be misleading for a number of reasons related to clinician and patient communication. Problems with communication have previously been identified in reviewed cases of suicide.9 There are additional problems as well.

Intentional Deception by the Patient

In one study, a number of patients who completed suicide deliberately concealed suicidal feelings and intention from their caregivers.9 There are many reasons patients might want to deceive their caregivers about suicidal intention. They might want to avoid hospitalization or other complications in their lives. They may be afraid of the effect of such an admission on the confidence of their therapist, or they may be afraid to admit these feelings to themselves. They also may want to keep all of their options open. Regardless of the reason, in all these cases, the clinician's role may change from that of an ally to that of an adversary.19

Empathy Failure in the Clinician

In general, clinicians do fairly well in understanding how their patients feel. However, in a 1994 study, 52 suicidal patients and their therapists were given a post-session questionnaire in which both were asked to rate the feelings of the patient during the session. It was found that while clinicians were accurate in assessing the patient's psychological pain, stressors, self-regard, and hopelessness, they were not accurate in evaluating the patient's degree of agitation.20 This is an extremely disturbing observation in view of the consistent finding that of ail the noted features of suicidal patients, psychic pain and agitation appear to be most closely linked to an immediate suicide crisis.16-17,21

Clinician Denial

No one wants to have a patient who is suicidal. Questions easily may be framed, consciously or otherwise, to minimize the likelihood of a positive suicidal response. For example, "You're not suicidal, are you?"


Alexithymia is an inability to describe one's feelings. In a retrospective study of non-serious and serious suicide attemptors, the lack of willingness or ability for self-disclosure significantly differentiated the serious attemptors from the ideators and mild attemptors. This lack of capacity for self-disclosure was not mediated by depression, anxiety, or hopelessness.22 One small study found a relationship between panic disorder, alexithymia, and suicidality.23

Anxiolytic, Affirming Effects of the Interview

Clinicians function in a dual role when conducting assessment interviews. They are both gathering information and inspiring trust Patients generally feel better when they are talking to a good listener. They are less anxious, more hopeful, and more connected. Good clinicians who have gone thorough assessments have concluded that their patients were not suicidal, only to have suicide result a short time later. It is certainly possible, if not likely, that the patient was being honest and the clinician was being thorough in many cases. During the course of the interview, the patient felt progressively more hopeful and less anxious. Suicide became less of an issue. The problem occurred when the patient left the office and found nothing substantial in his or her Life had changed. The suicidal impulse was only temporarily suppressed.


If the standard clinical interview with an emphasis on suicidal ideation is not a trustworthy basis for performing a suicide assessment, what are the alternatives? One alternative is to improve the accuracy of the interview. For example, Shea24 has developed a scheme for suicide risk assessment that guides the interviewer through a systematic exploration of suicidal behavior in the at-risk patient. The Chronological Assessment of Suicide Events approach is chronologically based and prescribes wellknown but often overlooked interview techniques designed to elicit relevant information. For more information on this approach, see me article by Shea on page 385.

A different, complementary alternative is to use the model emerging from the violence-risk-management literature; specifically, a compromise between a strict actuarial scheme and pure clinical judgment. Actuarial approaches have great statistical wer when applied to groups of people. Their weakness appears when a predicted outcome with a specific individual is required. This, of course, is precisely the case in a clinical setting. In an effort to "individualize" an actuarial approach, a hybrid model has been proposed - structured professional judgment or structured discretion. This model attempts to combine the best of both approaches - the power of actuarial prediction individualized by specific circumstances.25 In this model, actuarial tools are constructed to combine static, demographic predictors of violence, such as age and violent history, with individual modifying factors, such as treatment responsiveness and presence of personal support.

Awareness-Assessment-Action Model

The approach of structured professional judgment also may be used in suicide risk assessment (Sidebar, see page xxx). In the first step, awareness, die clinician is aware of die demographic factors that suggest greater risk in completed suicides - certain populations are more at risk than others. These risk factors are frequently listed in the literature and are present in all of the readily available suicide assessment tools. They include presence of mental illness, age, gender, race, history of suicide attempts, suicidal behaviors, marital status, family history of suicide, history of abuse, and so on. These risk factors establish the context for the assessment. While they are not predictive, they cannot be ignored, even in the face of denial of suicidal thinking. Imagine an emergency department physician who fails to consider the possibility of myocardial infarction in a 60-year-old man with chest pain simply because the patient denies that he has heart problems.

In the next step, assessment, a patient is interviewed for the immediate presentation of the complaint - specifically, how vulnerable this person is and what the mental state is. Individual vulnerabilities would include active substance abuse, a precipitating event, poor treatment alliance and compliance, help negation, poor coping ability, positive beliefs about death, social isolation, presence of lethal means, medical problems, impulsivity, and so on. Evaluation of the mental state would include panic and agitation, psychic anxiety, insomnia, poor concentration, and an acute, intense affective state that would include feelings of desperation, self-hatred, humiliation, abandonment, and rage. These factors individualize the context, particularly the immediacy of the suicidal threat.

The final step, action, should follow logically from the first two. The goal is suicide prevention and the process is to minimize risk. Those risk factors that can be modified, such as untreated mental illness and substance abuse and symptoms of agitation, are addressed in a treatment plan. Potentially lethal means are removed from the home, and significant others are enlisted in support. Hospitalization or crisis housing is used if necessary.


Mr. Brown is seen in the clinic because his son has brought him in to .seek help for his depression. He is a white, recently divorced, 65-year-old with a history of depression and substance abuse for which he has not been treated. He has never made a suicide attempt. He does not know anyone who has completed suicide. He denies suicidal ideation and says he has not considered a plan. Although Mr. Brown denies being suicidal, he is in a demographic group of people who are at the highest risk for suicide. He needs to be assessed more moroughly.

Assessment uncovers patients' various vulnerabilities and overall mental state. For example, Mr. Brown has been drinking nightly to help him sleep; he complains of insomnia. There was a recent divorce. He does not believe in psychiatry because he doesn't think it does any good to talk about problems. He would prefer not to take medication, and copes with stress by staying busy. He does not have anyone he can talk to about this. He is a hunter, but didn't go hunting last year because of worsening hip pain. He appears impatient and fidgety. He acts uncomfortable and restless, possibly because he has not been sleeping well. He admits to feeling "at the end of his rope." He becomes desperate when hospitalization is suggested. He says he cannot afford to miss any more work, and he wants to go home and be left alone. The individual assessment puts Mr. Brown in a category of people who are at high, even immediate, risk for suicide, even though he denies suicide is an issue with him.

Although Mr. Brown denies any suicidal behaviors, such as ideation, planning, communicating, or attempting, he is demographically at risk. His individual assessment is ominous. Some sort of action to reduce the risk of suicide needs to be taken. This action might include hospitalization, which should be explained to Mr. Brown as a way to relieve his insomnia and depression more quickly, rather than as a means to keep him safe. This disposition would certainly be more comfortable for the clinician. In many cases, however, a suitable outpatient plan can be devised. In any case, risk factors that should be immediately addressed are treatment for the depression and anxiety, involvement of significant others, substance abuse referral, removal of guns from the home, and effective follow-up assessment and care.

This example of suicide-risk management transparently demonstrates the thinking of the clinician. The factors considered in the assessment are documented, and the risk-management plan follows directly from the assessment. The clinician was not overly reliant on the presence of suicidal behavior in making the decision. In this case, if there is a bad outcome, either in the hospital or as an outpatient, the clinician clearly will have met the standard of documenting a thorough risk assessment and acting to reduce that risk.


All mental healthcare professionals are expected to assess for the risk of suicide in their patients and take action to prevent suicide from occurring. By relying exclusively on the results of a clinical interview, and specifically on the presence of suicidal ideation, clinicians may be falsely assured that suicide is not an issue with their patients. A model of "structured professional judgment" may help clinicians remain more alert to the possibility of suicide in a given patient, and may ultimately improve suicide prevention outcomes. A sample scheme, Awareness-Assessment-Action, provides one model of structured professional judgment, as well as a logical template for documenting risk assessment and the subsequent suicide riskreduction plan.


1. Gutheil TG. Liability issues and liability prevention in suicide. In: Jacobs DG, ed. The Harvard Medical School Guide to Suicide Assessment and Intervention. San Francisco, CA: Jossey-Bass; 1999:561-578.

2. Grove WM, ZaId DH. Lebow BS, Snitz BE, Nelson C. Clinical versus mechanical prediction: a meta-analysis. Psychol Assess. 2000;12(I):I9-30.

3. Hughes DH. Can the clinician predict suicide? Psychiatr Serv. 1995;46(3):449-451.

4. Oxley SL, Meter SV. The assessment and management of the suicidal patient. Journal of Practical Psychiatry and Behavioral Health. 1996;2(6):327-335.

5. Goldstein RB, Black DW, Nasrallah A. Winokur G. The prediction of suicide. Sensitivity, specificity, and predictive value of a multivariate model applied to suicide among 1906 patients with affective disorders. Arch Gen Psychiatry. Ì991;48(5):418-422.

6. Simon RI. Suicide risk assessment: what is the standard of care? J Am Acad Psychiatry Law. 2002;3(K3):340-344.

7. Williamson v Liptzin, 548 SE2d 734 (NC 2001).

8. National Center for Health Statistics. Available at: Accessed March 16, 2004.

9. Hendin H, Maltsberger JT, Lipschitz A, Haas AP1 Kyle J. Recognizing and responding to a suicide crisis. Suicide Life Threat Behav. 2001:3 1(2): 115-128.

10. Coombs DW, Miller HL, Alarcon R, et al. Presuicide attempt communications between parasuicides and consulted caregivers. Suicide Life Threat Behav. 1992;22(3):289-302.

11 Truant GS, O'Reilly R, Donaldson L. How psychiatrists weigh risk factors when assessing suicide risk. Suicide Life Threat Behav. 1991;21(2):106-114.

12. Jobes DA, Eyman JR, Yufit RI. How clinicians assess suicide risk in adolescents and adults. Crisis Intervention. 1995;2:1-12.

13. Kessler RC, Borges G, Walters EE. Prevalence of and risk factors for lifetime suicide attempts in the National Comorbidity Survey. Arch Gen Psychiatry. 1999;56(7):617-626.

14. Robins E, Gassner S, Kayes J, Wilkinson RH Jr, Murphy GE. The communication of suicidal intent: a study of 134 consecutive cases of successful (completed) suicide. Am J Psychiatry. l959;115(8):724-733.

15. Busch KA, Fawcett J. Jacobs DG. Clinica) correlates of inpatient suicide. J Clin P sychiatry. 2003:64(1): 14-19.

16. Fawcett J, Scheftner WA, Fogg L, et al. Timerelated predictors of suicide in major affective disorder. Am J Psychiatry. 1990; 147(9): 1189-1194.

17. Gladstone GL, Mitchell PB, Parker G. et al. Indicators of suicide over 10 years in a specialist mood disorders unit sample. J Clin Psychiatry. 2001 ;62(12): 945-951.

18. Isometsa ET, Heikkinen ME, Marttunen MJ, et a). The last appointment before suicide: is suicide intent communicated? Am J Psychiatry. 1995;152(6):919-922.

19. Resnick PJ. Recognizing that the suicidal patient views you as an "adversary." Current Psychiatry. 2002:1(1):8.

20. Eddins CL, Jobes DA. Do you see what I see? Patient and clinician perceptions of underlying dimensions of suicidality. Suicide Life Threat Behav. 1994:24(2):170-173.

21. Busch KA, Clark DC, Fawcett J. et al. Clinical features of inpatient suicide. Psychiatr Ann. 1993;23(5):256-262.

22. Apter A. Horesh N. Gothelf D, Graffi H, Lepkifker E. Relationship between self-disclosure and serious suicidal behavior. Compr Psychiatry. 200l;42(l):70-75.

23. Iancu I, Dannon PN, Poreh A, et al. Alexithymia and suicidality in panie disorder. Compr Psychiatry. 200l;42(6):477-481.

24. Shea S. The Practical Art of Suicide Assessment. New York, NY: John Wiley & Sons Inc.; 1999.

25. Douglas KS, Kfopp PR. A prevention-based paradigm for violence risk assessment. Criminal Justice and Behavior. 2002;29(5):617-658.


Sign up to receive

Journal E-contents