Research Article - IngentaConnect

Research Article Assessing Suicide Risk: What is Commonly Missed in the Emergency Room? Objective: Although risk assessment for suicide has been extensively studied, it is still an inexact process. The current study determined how busy emergency clinicians actually assessed and documented suicide risk, while also examining the differences between psychiatric and emergency medicine opinions on the importance of various suicide predictors. Method: Phase 1 of the study involved the administration of a survey on the relative importance of various suicide predictors for the specialties of psychiatry and emergency medicine. In phase 2 of the study, a chart review of psychiatric emergency room patients was conducted to determine the actual documentation rates of the suicide predictors. Results: Several predictors that were deemed to be important, including suicidal plan, intent for suicide, having means available for suicide, and practicing suicide (taking different steps leading up to suicide but not actually attempting suicide), had low documentation rates. Conclusions: Medical specialties have different opinions on the importance of various suicide predictors. Also, some predictors deemed important had low documentation rates. Educational interventions and simple assessment tools may help to increase documentation rates of several suicide predictors in busy clinical settings. (Journal of Psychiatric Practice 2017;23;82– 91) KEY WORDS: suicide risk, prevention, predictors, emergency room, mental health

Suicide is a global problem. It is estimated that there are approximately 10 million suicide attempts and nearly 1 million deaths from suicide worldwide each year.1 The problem of suicide is also a growing concern, as suicide rates have increased 60% over

82

March 2017

NAZANIN ALAVI, MD TARAS RESHETUKHA, MD ERIC PROST, MD KRISTEN ANTONIAK DIANNE GROLL, PhD

the last 45 years1; it is now the tenth leading cause of death worldwide and the second leading cause of death for youth 15 to 24 years of age in Canada.2 Contact with primary care providers before suicide is common, as 75% of suicide victims had contact with primary care providers within a year of their suicide, and 45% in the month before their suicide.3 The degree to which death by suicide can be predicted is one of the most researched topics in the field of suicidology; however, there is little evidence for accurate suicide prediction tools. The rarity of suicide and the fluctuation of the course of suicidality may explain why accurate suicide prediction seems to be impossible after decades of research.4 However, a number of factors predict suicide to varying degrees, including demographic characteristics (eg, older age, male sex, unemployment), psychiatric diagnoses (eg, depressive disorders, bipolar disorder, psychotic disorders, alcohol and substance use disorders), psychiatric history (eg, previous psychiatric treatment), psychological symptoms (eg, depression, hopelessness), and characteristics broadly related to suicidality (eg, suicide ideation, history of suicide attempts, history of abuse, history of bullying).5–7 Psychological autopsy studies show that 90% of all patients who die by suicide have a diagnosed mental illness,8,9 and that mental disorders such as substance use disorders, mood disorders, or personality disorders do indeed increase the risk of suicide.10 A history of suicide attempts is the strongest consistent predictor for future suicide attempts or suicide, as an individual with previous ALAVI, RESHETUKHA, PROST, ANTONIAK, and GROLL: Department of Psychiatry, Queen’s University, Kingston, ON, Canada Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved. Please send correspondence to: Nazanin Alavi, MD, Hotel Dieu Hospital, 166 Brock Street, Kingston, ON, Canada K7L 5G2 (e-mail: [email protected]). The study was partially funded through a Queen’s University Department of Psychiatry internal grant. The authors declare no conflicts of interest. DOI: 10.1097/PRA.0000000000000216

Journal of Psychiatric Practice Vol. 23, No. 2

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.

ASSESSING SUICIDE RISK IN THE ER suicide attempts is 48 times more likely to die by suicide compared with controls.11 Proximal warning signs such as thoughts of suicide, preparatory acts, stressful life events, and cognitive states seem to be important but they are of limited use in predicting future suicide attempts. Suicidal ideation itself is not a reliable predictor of suicide. One study found that 78% of individuals who died by suicide had denied suicidal ideation.12 Finally, not all factors involved in suicide risk assessment are negative, as protective factors such as religious affiliations, reasons for living, supportive social networks, marriage, and children may contribute to some degree of resiliency against suicidal behavior.13 Ultimately, the predictive value of any of these risk factors is low due to high false-positive prediction rates, and suicide prediction remains challenging. An integrated approach has been suggested to better meet this challenge.13 Given current high suicide rates and the uncertainties regarding prediction discussed above, the state of suicide risk assessment remains a cause for concern. In an attempt to advance this field, newer models for assessment have been proposed and studied. For example, models that use quantitative data from scales have been researched. Some strategies focus on suicidal ideation, hopelessness, and protective factors. Scales such as the Beck Hopelessness Scale,14 Beck Scale for Suicidal Ideation,15 and the Reasons for Living Inventory have been evaluated.16 The Beck Hopelessness Scale is a true/false 20-item scale that assesses hopelessness within the past week, the Beck Scale for Suicidal Ideation is a 21-item scale used to assess the severity of suicidal ideation within the past week, and the Reasons for Living Inventory is a 48-item scale used to assess the reasons for living that may serve a protective function for someone considering suicide.17 However, clinical scales cannot predict suicide at the individual patient level. Therefore, hospital admission for risk of suicide should not be based on cut-off scores on these scales.18 Decisions made by physicians on the basis of a single questionnaire may miss some at-risk patients and are not always best for the patient. Moreover, some suicide risk factors may not be included in suicide risk assessments that use scales. Factors such as a history of bullying in children and adolescents,19 a history of physical, sexual, and/or emotional abuse,20 and


substance abuse21 have been associated with an increased suicide risk and suicide ideation. In general, these factors are rarely integrated in questionnaires or asked about by physicians during assessments. In a series of articles describing a model for therapeutic risk management of the suicidal patient, Wortzel et al22 suggested developing a database to formulate levels of risk and provide appropriate interventions. Their model also involves augmenting clinical risk assessments with structured instruments and has the potential to improve the assessment and management of suicidal individuals by collectively yielding a suicide risk assessment and management process.23 Another approach is to identify who is at risk of attempting suicide or dying by suicide by studying which risk factors and warning signs are most useful in predicting these outcomes. Research by Rudd et al24 has yielded many specific risk factors and warning signs that signal concern regarding future suicidal behaviors. Although these variables are each valuable for suicide prevention efforts, they may be limited in their applicability to clinical practice.24 Thus, given these limitations, it is suggested that practitioners should not simply rely on risk factors or warning signs independently when assessing suicide risk, but that risk factors and warning signs be assessed together with the clinical interview to maximize the physician’s understanding of what is causing the patient’s suicidality.25 Study results suggest that the presence of risk factors can highlight whether someone is more likely to engage in suicidal behaviors, but the risk factors themselves do not provide information about an individual’s imminent risk for a suicide attempt. In addition, risk factors are not reliable in differentiating between those who have suicidal thoughts and those who will attempt to end their lives.26 Warning signs for suicide were proposed to overcome these limitations of risk factors, but research has also shown that most warning signs also lack the ability to identify who will most likely act on suicidal impulses.27 Thus, future research is needed to understand what variables can more successfully predict suicidal crises.28 On the basis of this overview, it is clear that (1) suicide prediction is difficult, and (2) the available tools are imperfect. Although our best practices are

March 2017


83

ASSESSING SUICIDE RISK IN THE ER flawed (and even because this is so), it is useful to study how busy clinicians actually assess suicide risk and document their efforts. Emergency department clinicians are exposed to suicidal patients on the verge of crisis, as attempted suicide and self-inflicted injuries account for 0.4% of all emergency room visits.29 Management, assessment, and treatment of suicidal patients are stressful tasks for clinicians: 28% of psychologists and 62% of psychiatrists have reported the loss of a patient due to suicide.30,31 During a potential suicide crisis, clinicians may be in a heightened state of anxiety and stress, which may result in inefficient assessment of suicide risk as well as a lack of clear and concise documentation of the patient’s risk level —and deficient risk assessment and questionable medication management are the most common clinical factors cited as causes of legal action against mental health professionals, with inadequate documentation representing 31% of all cases.32 A study by Malone et al32 showed that clinicians failed to document risk factors such as past suicide attempts in 24% of patients at admission and 28% of patients at discharge, highlighting the importance of proper documentation in a suicide assessment. The goals of this study were to determine how busy emergency clinicians actually assess and document suicide risk and to examine the differences between psychiatric and emergency medicine clinicians in opinions about the importance of various suicide predictors. Patients’ charts were reviewed to determine what suicide risk factors were assessed and missed in patients presenting at the emergency department for suicidal ideation during a 2-year period. The overarching aim of the study is to help us understand the suicidal predictors that might be missed when patients are assessed in the emergency department and improve suicide risk assessments conducted both by emergency medicine physicians and psychiatrists in emergency departments.

METHOD This study had 2 phases. First, a survey was sent to all psychiatry and emergency medicine faculty members and residents at Queen’s University explaining our study. If they consented to take part in the study, they were asked to complete an online questionnaire about their opinions on predictors of

84

March 2017

suicidal risk. A suicidal predictor questionnaire was developed for the purpose of this study that covered 45 suicidal risk predictors based on the literature (Table 1). The questionnaire asked specifically how each suicide predictor affected the physician’s decision to consult psychiatry or admit a patient with suicidal ideation. The physicians rated each predictor on a 4-point scale, coded as 0=not at all, 1=a little, 2=somewhat, and 3=very much. In the second phase of the study, charts of all patients who were seen by the emergency medicine and/or psychiatry team in all emergency departments in Kingston, ON (Kingston General Hospital and Hotel Dieu Hospital) from 2011 to 2013 were reviewed; if suicidal ideation or behavior was present, data on that patient were included in the study.

Chart Audit Process Two research assistants, who received training in reviewing patients’ electronic charts, reviewed all of the electronic patient charts in the sample population described above. The reviewers collected data from the electronic assessment forms that were completed by the emergency medicine team and/or psychiatry team at the emergency departments during each patient’s initial visit. All information recorded on the assessment forms was recorded by the physicians based on the patients’ self-reported answers to the physicians’ direct questions and the physicians’ observations. The team recorded the suicide predictors that were assessed, the clinical decision that was made, and the suicide predictors that were missed at the time of assessment. Following the data collection, the first author reviewed and compared the audit data with the electronic assessment forms as a quality check procedure and to assess reliability of the chart audit. The entire chart auditing process was completed in 6 months. Data for the variables under study were entered anonymously into a database separate from the clinical file. This study was approved by the Research and Ethics Board of Queen’s University, Canada.

Statistical Analysis First, a Mann-Whitney test was conducted to determine the differences in the rating of importance of



ASSESSING SUICIDE RISK IN THE ER TABLE 1. Survey Sent to Psychiatry and Emergency Medicine Physicians The survey was sent to the respondents with the following instructions: We would like to invite you to assist us with understanding the use of the common suicidal attempt predictors, in managing patients with suicidal ideation. You are asked to complete a series of survey questions that should take a few minutes. Responses to the survey will be collated for research purposes at Queen’s University and all your answers will be kept completely confidential. What term best describes you? Faculty member/resident (what year)? How much do the following suicide predictors affect your decision to admit a patient with suicidal ideation/to consult psychiatry? (depending on whom the survey was sent—psychiatrist versus emergency medicine physician—the question was different). Please rate each item using a 4-point scale, with 0=not at all, 1=a little, 2=somewhat, and 3=very much. 1. Age

24. Past suicide attempt

2. Sex(male)

25. Severity of previous suicide attempt

3. Not employed

26. History of bullying

4. Not educated

27. History of emotional abuse

5. Being professional/physician

28. History of sexual abuse

6. Lack of family support

29. History of verbal abuse

7. Lack of social support

30. History of physical abuse

8. No children at home

31. Current medical or physical disorder

9. Unmarried

32. Acute psychiatric symptoms

10. No religious background

33. Acute stressors

11. Ethnicity(not Caucasian)

34. Loss

12. Aboriginal

35. Brought by the police

13. Family history of suicide

36. On a form 1*

14. Recent immigration

37. Collateral information

15. Hopelessness

38. Agitation

16. Depression

39. Plan for suicide

17. Anxiety disorder

40. Intent for suicide

18. Psychotic disorder

41. Preparation for suicide

19. Alcohol abuse

42. Practicing for suicide†

20. Substance abuse

43. Suicide note

21. Bipolar disorder

44. Not contracted for safety (no suicide contract)

22. Personality disorder

45. Not future oriented

23. Psychiatric history

Are there any other factors that affect your management and follow-up decision when managing a patient with suicidal ideation or suicidal behavior? *Form 1: Canadian legal document signed by a physician requiring a patient to remain in a psychiatric hospital for up to 72 hours. †Practicing for suicide refers to taking steps that would lead up to suicide (eg, buying a rope, putting the rope around one’s neck, standing on a chair) but not actually attempting suicide.

suicide predictors between the different medical professionals and between residents and faculty members of both medical specialties on the basis of the survey conducted in phase 1 of the study. Another Mann-Whitney test was conducted to determine differences in the documentation rates of the 45 suicide


predictors by the different medical specialties on the basis of the data collected in phase 2 of the study. Additional χ2 tests were conducted to determine whether there were differences in the documentation rate of suicide predictors by emergency medicine physicians who consulted psychiatry compared with

March 2017


85

ASSESSING SUICIDE RISK IN THE ER those who discharged patients without involving mental health specialists. All statistical analyses were conducted using SPSS Statistics 2.2

RESULTS The survey was completed by 51 physicians in the psychiatry department (23 residents and 28 faculty members) and by 23 physicians in the department of emergency medicine (11 residents and 12 faculty members) at Queen’s University, Kingston, Canada. The survey asked about 45 suicide predictors to examine the differences in opinion between psychiatry and emergency medicine physicians on the importance of each in their clinical decision-making when a patient presented with suicidal ideation/ behavior. The response rate from the survey was 65% for the faculty and residents of the department of psychiatry and 56% for the faculty and residents of the department of emergency medicine. Opinions on Suicide Predictors Significant differences were found in ratings of the importance of several suicide predictors between the

2 medical specialties. Psychiatry faculty members and residents rated male sex, unemployment, being unmarried, no children at home, lack of religious background, not being white, aboriginal status, recent immigration, loss, being brought in by police, collateral information, plan for suicide, and a suicide note as more important predictors for a patient’s admission than the emergency medicine physicians and residents (Table 2). The emergency medicine physicians and residents rated the predictors psychiatric history, history of verbal abuse, and not contracted for safety as more important for a patient’s referral to psychiatry than the psychiatrists (Table 2). There were few significant differences in the rating of suicide predictors between faculty members and residents within the same medical specialty. Psychiatry faculty members rated both whether the patient was a professional or physician and agitation as more important suicide predictors when determining whether psychiatry should be consulted than psychiatry residents (U=215.0, P=0.036, r=0.29; U=218.5, P=0.025, r=0.31). Emergency medicine physicians rated depression and not being contracted for safety as more important suicide predictors for a patient’s admission compared with

TABLE 2. Ratings of Suicide Predictors by Psychiatrists and Emergency Physicians

Suicide Predictors Male sex Unemployment Unmarried No children at home Lack of religious background Not white Aboriginal Recent immigration Psychiatric history History of verbal abuse Loss Brought in by police Collateral information Plan for suicide Suicide note Not contracted for safety

Psychiatry Median*

Emergency Physician Median*

MannWhitney U

P

Effect Size (r)

2 2 2 3 2

1 1 1 2 1

401.5 379.5 409.0 357.5 227.5

0.017 0.011 0.026 0.003