Forensic psychology: Part 4: Chapter 12: Assessing offender populations

by Lois Rosine, Ph.D., C.Psych.^{Footnote 1}

Objectives

1. To identify theoretical and practical issues in the assessment of suicidal offenders.
2. To offer guidelines for referral agents.
3. To offer guidelines for clinical practice.
4. To identify professional needs.

Suicides in jails are tragic phenomena that cause a great deal of pain and grief to offenders, their families and the staff who are responsible for their care. Such suicides generate much public concern as the state is seen as responsible for the well being of those in its care.^{Footnote 2}

Psychologists in forensic settings need to be aware of both their corporate and their professional responsibility. Corporate responsibility within the Correctional Service of Canada (CSC) is clearly defined in its policies regarding suicide^{Footnote 3} — all employees are required to take the actions necessary to prevent the suicide of an offender.

Psychologists working in correctional settings are also bound by a code of ethics^{Footnote 4} and the standards of practice of provincial licensing bodies, as well as various pieces federal and provincial legislation. These varying responsibilities, the repercussions of false negatives, ^{Footnote 5} and the limits of current scientific knowledge regarding the prediction of suicide serve to make suicide assessment a difficult task.

The difficulty of assessing suicide potential is further exacerbated by society's expectation about its prevention. Through the 19th and 20th centuries, suicide was seen as a manifestation of mental illness, meaning the individual was not deemed responsible for his or her behaviour. The idea of suicide as a form of mental illness shifted the responsibility for its prevention into the domain of the mental health professional. Concurrently with this evolution in thinking, an attitude developed in society that mental health professionals are not only able to predict suicidal behaviour, but are also able to prevent suicides. ^{Footnote 6} In reality, the prediction of the future behaviour of an individual is fraught with a number of difficulties. When attempting to predict suicide, it is more realistic to talk about the prediction of risk rather than prediction of actual outcome. ^{Footnote 7}

In correctional settings, it is often front line personnel who are the first to become aware of a potentially suicidal offender. Identifying an individual as at-risk represents a judgment about the behaviour of a specific individual at a given point in time. This is a difficult task given that suicidal behaviour can incorporate a wide constellation of behaviours that include suicidal ideation, gestures, self-injury, parasuicide, suicide attempts and completed suicides. Further complicating the issue is that each of the sub-behaviours of suicide are not dichotomous variables, but rather represent classes of behaviours that fall along a continuum of seriousness and intent. These factors necessitate that front line personnel be sensitive to person specific clues.

The following provide examples of some clues that warrant exploration regarding suicide potential:

• feelings (sad, despondent, hopeless, helpless, worthless, guilt);
• thoughts (wish I were dead, my problems will soon end, no one can help me now, can't take it any more, everyone would be better off without me, etc.);
• behaviours (giving away possessions, loss of interest in life, withdrawal, extreme behaviour change, reckless behaviour, self-injury);
• physical symptoms (lack of interest in appearance, physical health complaints, disturbed sleep, change or loss of appetite/ weight/interest in sex); and
• verbal behaviour (statements about suicide — jokes, talking a lot about death, direct statements about wanting to die, etc.).

The next step is to determine if there is a potential for suicide. Gathering information in the following areas will enable the officer to form a judgment.

1. Stress: Identify the stressors in the offender's life. How are the stressors perceived by the offender? Are the stressors seen as overwhelming or unmanageable? How are the stressors contributing to the individual's current stress?
2. Symptoms: How is the offender responding to the stressors?
   1. Emotional — are there mood fluctuations, crying spells, etc.?
   2. Cognitive — is thinking slowed, diffused, constricted?
   3. Behavioural — is there fatigue, withdrawal, agitation?
   4. Physical — what has happened to eating, sleeping patterns?
   5. Coping — are there ongoing maladaptive coping strategies (such as drug/alcohol abuse)?
3. Is there a current suicide plan: Is the person at-risk for considering suicide? Is there a plan? How lethal is the plan? It is important to ask directly about suicide. Contrary to public myth, asking directly about suicide does not cause suicides, but rather relieves the person-at-risk's anxiety and facilitates disclosure.
4. Prior suicide behaviour: Is there a personal history of suicide gestures, attempts, or a history of suicide in the family?
5. Resources: What physical and emotional support systems does the person-at-risk have?

If, in the officer's opinion there is a suicide risk, a referral should be made to psychology. When making the referral, it is important that the information gathered above be included on the referral form. Referrals need to provide information about the specific factors which led the officer to be concerned. Personnel should be encouraged to identify all the factors, including the intuitive feelings that caused them concern. This will provide the psychologist with the necessary data for the evaluation and a treatment plan. When evaluating suicidal behaviour, caution is the operative word. If an offender denies any suicidal behaviour but, as mentioned above, the officer's judgment suggests otherwise, a referral is in order.

Front line personnel should receive basic training in crisis and suicide intervention skills. Most suicide intervention training programs focus on identification of suicide risk factors, basic assessment of level of risk and basic intervention strategies that are matched to the risk level.

Understanding and predicting suicide is replete with methodological problems. These include definitional problems, under-reporting, difficulty studying the target population, studying a low frequency behaviour, and problems in statistical versus clinical prediction.

i) Classification issues
Classifying any complex human behaviour presents a major difficulty as there is no single criterion on which to base the classification. In the case of suicide, the outcome represents the end result of a process, not the process itself. A number of questions arise. What role does intent to commit suicide play? What constitutes suicidal behaviour? Do all selfdestructive behaviours fall along a continuum with suicide at one extreme pole?

The term suicide is used to define a broad spectrum of behaviours. Behaviours such as suicidal ideation, suicide attempts, suicidal gestures, completed suicides, destructive lifestyles^{Footnote 8} and high risk behaviours ^{Footnote 9} are often lumped under the heading of suicide. The suicide literature exemplifies semantic confusion, lack of clarity regarding definitions and the use of terms, and even contradictory use of terms within the same study (Pokorny, 1974). Generally, suicidal behaviour has been conceptualized as falling along a continuum of increasing lethality (Weisman, 1971).

The implication is that zero suicidal behaviour is one extreme pole with suicidal ideation and ever increasing life-threatening behaviours moving to the alternate pole of completed suicides Simon & Murphyy, 1985; Zubin, 1974). The premise that increasing lethality of life-threatening behaviours represents a linear progression to suicidal outcomes may not be accurate.

In contrast, Heney (1990) offers an alternative conceptualization based on her work with women who are the victims of childhood sexual abuse. Many of these women display a number of serious, life-threatening, self-destructive behaviours. She proposes that these behaviours are not suicidal behaviours but rather "health seeking" behaviours.

These "health seeking" behaviours can be conceptualized using a classical conditioning paradigm in that, as children, these women experienced a great deal of anxiety prior to each assault. When the assault occurred, the child experienced significant psychological and/or physical pain. During the period immediately following the assault, there were intense feelings of relief, as the child knew she would experience a temporary respite from further assault. Through classical conditioning, the pain became an unconditioned stimuli for anxiety reduction. Both the feelings of pain and relief are extremely salient stimuli, and through repeated exposure the child learns that pain reduces anxiety.^{Footnote 10}

More importantly, the child has learned that intense pain is a method of reducing anxiety. This learning is so strong that, as adults, when exposed to the appropriate cues for anxiety, self-destructive behaviours become the preferred method for reducing anxiety. In these adult women, the life-threatening behaviours are used as a method of anxiety reduction, not to end life.

Heney indicates that this health seeking behaviour can co-exist in an individual who is also suicidal, but the assumption cannot be made that life-threatening behaviours equal potential suicide.

The intent of the behaviour must be included in the definitional equation.

One of the obvious difficulties in studying suicide is that the population of suicide completers cannot be identified until the time of their self-inflicted death. To deal with this problem, the most common method of studying suicidal behaviour has been to study the behaviour of those who attempt suicide (Neuringer, 1962).

However, even the designation of a suicide attempter is problematic. For example, does a cut to the wrist with minimal bleeding represent the same class of behaviour as an attempted strangulation that is only interrupted by a spouse who returns home early because of a cancelled meeting? Shneidman (1985) proposes that the term suicide attempt be reserved for those individuals who tried to end their lives and only fortuitously survived.

He recommends the use of the English term "parasuicide" for other self-injurious behaviours. This concept of suicide includes the idea of both intent and lethality.

Using substitute subjects (i.e., those who attempt suicide) as a method of studying suicide has been challenged by a number of authors who suggest the method is invalid because attempted suicide and completed suicide represent different classes of behaviour (Davis, 1967; Dorpat & Ripley, 1967; Shneidman, 1985; Wilkins, 1967).

Lester (1970) recommended a solution to this dispute. He suggested that when some quality, such as the behaviour of mildly serious attempters to very serious attempters, increases in a linear fashion then it can be predicted that this quality would occur more frequently in samples of completed suicides.

Heney's thesis, discussed above, seriously challenges this assumption as she suggests that in some cases even life-threatening behaviour does not represent suicidal behaviour.

In an effort to address the classification problems, a number of authors have presented methods for classifying suicidal behaviour. One of these authors, Pokorny (1974), proposed a model with broad categories of behaviour: suicidal ideas, suicide attempts, and completed suicides. Suicidal ideas represent those behaviours that are observed directly or by inference and that represent a "move in the direction of a possible threat to the individual's life, but ... the potentially lethal act has not actually been performed" (p. 36). Suicide attempts are those circumstances in which the person performs an actual or apparent life-threatening behaviour with the intention of endangering his/her life but the act does not result in death. Completed suicides include those behaviours where there was a wilful, life-threatening, self-inflicted act that ended in the person's death.

Additionally, Pokorny recommended that each suicidal behaviour be rated on five other dimensions: lethality, intent, mitigating circumstances, method and degree of certainty.

Certainty would be the degree, in percentages, to which that particular behaviour was deemed to be suicidal. Lethality represents the degree of objective medical danger (i.e. lack of reversibility and/or medical rescue due to the method chosen). Intent is the measure of seriousness and determination in terms of the subject's desire to end his/her own life. Mitigating circumstances represents variables such as mental illness, alcoholism, age, intelligence, etc., that may temporarily exacerbate a propensity to selfdestruction. The method of injury is seen as important as it may correlate with intent, lethality and/or mitigating circumstances.

Farber (1968) proposed an alternate method of classifying suicidal behaviours that incorporates the interaction of "intent to die" and the actual "outcome" of the behaviour in a 2 x 2 matrix.

In this model, when the intent to die is high and the outcome of the behaviour is death, then the presenting behaviour was a suicide. When the intent to die is high and the outcome is life, an attempted suicide has occurred.

Similarly, when the intent to die is low and the outcome is death, an accidental or an undetermined death has occurred. When the intent to die is low and the outcome behaviour is life, then a parasuicide has occurred. Shneidman (1985) also submits that any definition of suicide must include a recognition of lethality and intentionality.

ii) Under-reporting
Under-reporting is a major problem in the literature on suicide (Brugha & Walsh, 1978; Liberalds & Hoenig, 1978; McCarthy & Walsh, 1975; Ovenstone, 1973). Under-reporting occurs because of such factors as attitudes of certifying officials (Farberow, MacKinnon, & Nelson, 1977), the type of certification system, and the qualifications of the officials (Pokorny, 1974; Shneidman, 1985). Such underreporting has the potential to invalidate crossjurisdictional comparisons of suicide rates (Atkinson, Kessel, (Sz. Dalgaard, 1975; Douglas, 1967).

Under-reporting is considered to be a major problem when specific populations such as adolescents, the elderly, and aboriginals are studied (Report of the National Task Force on Suicide in Canada, 1987).

iii) Prediction of suicide
Present models and methodologies allow reasonable accuracy in the prediction of group behaviour within certain parameters.^{Footnote 11} Unfortunately, the accuracy of the prediction of specific individual behaviour is not nearly as sophisticated. Further complicating research in this area is that prediction and prevention are interwoven. If accurate prediction of each suicide were possible, then ethically, there would be an obligation to ensure every effort was made to prevent the fulfilment of this prediction. Successful prevention would then negate the prediction (Murphy, 1974). This interaction, while not circular, is certainly problematic in terms of predicting outcome.

There are inherent problems in the prediction of low frequency behaviours (Rosen, 1954). In a retrospective study of 40 women from the Terman Genetic Studies of Genius, Tomlinson-Keasey, Warren & Elliott (1986), using seven variables in a discriminant function analysis, were able to correctly classify 83% of the subjects into one of three groups: suicides, living controls, and deceased controls. Twenty-five percent of the suicides were missed. Pallis, Gibbons and Pierce (1984), using a suicide intent scale, were able to correctly identify 83% of their subjects retrospectively. In this study, the authors indicated that for every true suicide, 33 non-suicidal clients would be identified as suicidal. Smith (1982-83), in a retrospective assessment using a battery of tests to predict suicides, was successful in identifying 85% of the cases. Subjects were classified as no suicide, mild attempt, serious attempt or completed suicide. In the case of completed suicides, 21% of the cases were miss-classified. Farberow and MacKinnon (1974), using an 11-item scale, were able to correctly identify 79% of the suicides and incorrectly identify 25% of the non-suicides. While these methods offer promise, hit rates that miss 17% to 25% of the suicide completers have low clinical acceptability.

An additional issue in hit rates for the prediction of suicide is the number of false positives. In settings where there are few suicidal clients, the number of false positives does not pose major logistical problems. In a forensic environment, too many false positives have the potential to incapacitate prevention and treatment resources. For example, there are a limited number of observation cells as well as personnel to monitor at-risk offenders. Similar limitations surround mental health treatment resources in terms of post-identification assessment and follow-up. The problem becomes more relevant when using data. Suppose a method was available to correctly identify 80% of suicides. Burtch and Ericson (1979) calculated a suicide rate in Canadian prisons as 95.9 per 100,000 prisoners (the convention for reporting suicides is a rate per 100,000). CSC has an institutional population of approximately 15,000. ^{Footnote 12} Using the above method, 11 of the 14 suicides in the offender population would be correctly identified, but 2,989 non-suicidal offenders would be incorrectly identified as suicidal.^{Footnote 13} Even with these cases spread across the country, this large number of false positives would present significant problems. The situation would be further exacerbated by the constant intake of new offenders to the system.

A second issue in suicide prediction is clinical versus actuarial predictors. Demographic variables must by necessity ignore individual traits. However, even the application of a large number of demographic variables will not specify the individual very adequately (Lester, 1974). For example, being white, male, alcoholic and around 40 years of age are all demographic predictors of suicide (Murphy & Wetzel, 1990; Roy & Linnoila, 1986), but clearly these variables would identify a very large sample of clients. If we add the additional variables of previous suicide attempts and divorce, we are able to decrease the group somewhat but are still left with an extremely large sample. At best, actuarial variables represent the identification of specific populations at risk.

Clinical prediction, on the other hand, represents professional judgment about the behaviour of a specific individual at a specific point in time. Psychological assessment tools, an understanding of actuarial data, educational and experiential background, and theoretical orientation are all elements that influence what is essentially hypothesis testing to evaluate the behaviours presented by a unique individual. The two methods of prediction go hand-in-hand and are essential in the prevention of suicide.

Murphy (1984) suggests that the actuarial prediction of suicide is not a problem of the inadequacy of the data base or the analysis of these data, but rather the characteristics of low base rate behaviours. He further suggests that information on a person's potential for killing him/herself at some uncertain time in the future is of little use to the clinician needing to know what to do now. In clinical practice, "the decision is not what to do for all time, but rather what to do next, for the near future" (Pokorny, 1983, p.251). The process is a sequence of small decisions that involves alerting signs, further investigation, additional alerting signs or confirmatory indicators, determination of the level of intervention, type of interventions (e.g. medication, inpatient, outpatient, 24 hour observation), continual feedback to review risk assessment and intervention techniques. As with the researcher, the clinician must be concerned about false positives. For example, if the treatment procedure is very intrusive, such as involuntary hospitalization, justification of the treatment becomes difficult. In a prison setting, false positives are of additional concern given the already limited freedoms of offenders.

iv) Practical Issues
In most disorders of human behaviour, the dysfunction is an ongoing process that allows examination of its structure at various stages. In suicide, we are presented with the end result, a death, which was arrived at via the interaction of a number of multifaceted factors (Baton, 1987). The current state of the suicide prediction literature suggests that at this time, suicide cannot be reliably predicted (Rosine, 1991). This is not meant to be alarmist but rather to present a realistic picture of the present state of the art. However, such statements do little to help the mental health professional sitting face to face with a potentially suicidal individual, particularly in prison settings where the expectations of prevention are high.

Suicide in prison has been recognized as a major cause of death for offenders, with the suicide rate being several times higher than that of the general population (Bonner, 1992). A review of the literature on offender suicides (Lloyd, 1990) offered the following summary:

• gender - males were more likely than females to commit suicide.
• age - results were unclear;
• marital status - suiciders were more likely to be single;
• penal status - high rate of suicide among remand prisoners;
• length of sentence - lifers and those serving long sentences were more at risk;
• offense type — murderers were over represented in completed suicides;
• time served — most suicides occurred soon after incarceration, with the first two weeks in custody being the highest risk;
• mental disorder — relationship is unclear, but British findings indicate that approximately one-third of suicide completers had been in-patients prior to incarceration; and
• previous suicide attempts — those who had threatened or attempted suicide before were at greater risk to complete.

A quantitative review of the general suicide literature (Rosine, 1991) identified a number of risk factors in completed suicides. These more general findings have implications for offenders residing in the community under supervision, as well as for the institutional population:

• age — young males and the elderly were at greater risk;
• alcohol abuse — increased the risk of suicide, with alcoholics being at 5 times the risk of non-alcoholics;
• prior suicidal behaviour;
• depression;
• hopelessness;
• marital status — single individuals were at greater risk;
• living arrangement — those living alone or with a non-family member were at greater risk, likely reflecting a diminished level of emotional support;
• mental illness — increased the risk of completion of suicide;
• level of education — higher academic achievement had a positive relationship with suicide risk; and
• unemployment — appears to elevate suicide risk.

Demographic data delineate the population at-risk over the long-term, while dispositional variables help to bring the focus of the problem to the more immediate, short-term risk. However, even dispositional variables have limited use in the prediction of the individual case. Additional research and clinical factors need to be brought to bear on the task of suicide risk prediction.

Historically, men have exhibited higher rates of completed suicides than women, while women have had higher rates of attempted suicide. An important factor in understanding these gender differences in suicide rates is the tendency of men to use more immediate and more lethal methods of suicide (Marks & Abernathy, 1974; Ramsay, Tanney, Tierney & Lang, 1987; Stengel, 1971). However, when lethality of method was controlled, the rates of suicide for males and females were the same (Wilson, 1981). This points to the need to assess lethality during suicide assessments, particularly when assessing women. Lethality needs to be evaluated both in terms of previous attempts and any current plan (Ramsay et al., 1987).

An additional problem in the assessment of suicidal behaviour with offenders is that some offenders are reluctant to reveal their suicidal thoughts and behaviours because they fear the consequences of such revelations. Rosine (1994) found that 14% of a sample of 113 offenders indicated they would be unwilling to disclose potential suicidal behaviour to psychological staff. ^{Footnote 14} The following represent offender perceptions that may influence willingness to reveal suicidal ideation:

1. The decision to place an at-risk individual under observation may be viewed as a punitive rather than a precautionary measure (Task Force on Federally Sentenced Women, 1990).
2. The suicidal person may also have serious questions regarding the competency of the people who are to provide care, making it less likely for them to accept the support that is offered (Task Force on Federally Sentenced Women, 1990).
3. There is a risk mythology within the offender culture of individuals who have been denied temporary passes or release because of current or past file information identifying them as suicidal. This information is perceived as being used to impede progress in returning to the street: ^{Footnote 15}

Steer, Beck, Garrison & Lester (1988) and Beck & Steer (1989) reported that individuals who eventually killed themselves had been more careful not to be discovered in previous suicide attempts than did other suicide attempters. The taking of precaution against discovery has serious ramifications for the clinician attempting to do a risk assessment that is further exacerbated by the issue of social desirability.

There is currently a debate in the literature regarding the role of social desirability (SD) as a suppresser variable in the assessment of suicide. Strosahl, Linehan and Chiles (1984) suggest that a SD response set can result in information about current or past suicidal behaviour not being fully divulged, or being reorganized in such a manner that the clinician is seriously misled. The role of SD is particularly relevant when dealing with clinical populations (Cole, 1988; Holden, Mendonca & Serin; 1989). To further complicate the assessment process, Evenson (1983) and Pallis and Birtchnell (1977) found non-serious attempters and threateners to be seen as more pathological than serious suicide attempters. In other words, those most at-risk take more precaution against discovery, are more likely to mislead the clinician, and probably present as less disturbed.

Depression has long been considered an important variable in the prediction of suicide. However Beck, Kovacs and Weissman (1975) and Wetzel (1976) contend that the association between suicide and depression results from their common relationship with hopelessness. Wetzel, Margulies, Davis and Karam (1980) studied the relationship between hopelessness and depression with suicidal intent. Suicidal intent and hopelessness were highly correlated (r =.76), as was suicidal intent and depression (r =.36). When the effect of depression was statistically partialed out, the correlation between hopelessness and suicidal ideation remained high (r =.72). When the effect of hopelessness was partialed out, the relationship between suicidal intent and depression was eliminated (r =-.10). In a recent study, Beck, Steer, Beck and Newman (1993) found that hopelessness was 1.3 times more important than depression when explaining suicidal ideation.

It is likely that measurement of hopelessness is not independent of SD (Holden & Mendonca, 1984). Again, a SD response set would mean that the at-risk person may not fully divulge information about current or past hopelessness, or may seriously mislead the clinician. Further confounding this issue is that in the "normal" population, the ability to respond in a socially desirable manner appears to reflect social and psychological adjustment (Nevid, 1983). It is suggested that the more similar the at-risk-offender is to the general offender population, the more weight clinicians give to the potential of being misled in the direction of under-estimating risk.

A long and complex philosophical tradition regarding the permissibility of suicide has influenced modem attitudes and theories of suicidal behaviour. By the beginning of the 20th century, the prevailing belief was that suicide was the behaviour of the mentally ill. This premise has caused considerable debate in more recent times. ^{Footnote 16}

In clinical practice, the perception of the psychiatric state of the suicidal person is important in both the assessment of risk and in intervention strategies. Clinical experience teaches that basing suicide interventions on the premise that an individual who threatens to commit suicide or who commits suicide is irrational or mentally ill is patently false. It is clear that there are some suicidal individuals who are both irrational and/or mentally ill, however many others, while very troubled emotionally, are neither irrational nor mentally ill. In the case of irrational or mentally ill individuals manifesting overt difficulties, the decision to "protect them" from themselves is usually relatively clear. It is the second group of individuals who are neither irrational nor mentally ill who present the clinician with difficult decisions, both in terms of assessment and intervention. Given limited ability to predict "dangerousness" accurately, at what point is the danger clear and imminent? ^{Footnote 17} At what point does the psychologist intervene?

Suicide intervention training programs are based on the assumption that not all people who threaten suicide are mentally ill (Ramsey et al., 1987). Suicide risk assessment and subsequent intervention focus on the ambivalence that many suicidal people experience. Ambivalence being defined as the emotional shifting of the suicidal person between wanting to die and wanting to live. Suicide is viewed as a process with level of risk being assessed along a continuum from low to high, with the type of intervention determined by how far along this continuum the individual has progressed.

A multimodal assessment process will allow for a more accurate and informed assessment of suicide risk. Psychometric tests, information and behaviourial observations provided by front line staff, ^{Footnote 18} file review and the clinical interview augment the assessment process and provide necessary and important information.

i) Psychometric tests

The use of projective tests have demonstrated little success in the prediction of suicide (Exner & Wylie, 1977; Kendra, 1979; Neuringer, 1974).

Similar negative results have been found with the MMPI (Clopton & Baucom, 1979; Clopton & Jones, 1975; Watson, Klett, Walters & Vassar, 1984).

However, the use of specific scales designed to measure various aspects of suicidal behaviour offer promise (Pierce, 1981; Pallis, Gibbons & Pierce, 1984; Shaffer, Perlin, Schmidt & Stephens, 1974).

The following are a sample of psychometric tools used in the assessment of suicide. These scales can be used in forensic settings:

• Beck Depression Scale;
• Beck Hopelessness Scale; Risk-Rescue Rating (see Bongar, 1991b, p. 256);
• The Los Angeles Suicide Prevention Center Scale (see Bongar,1991b,p.249); and
• The Suicide Intent Scale (Beck, Schuyler & Herman, 1974).

ii) The clinical interview
The clinical interview is the corner stone of the suicide assessment process. "Even the most sensitive clinician cannot safely assume that he or she fully understands the meaning of suicidal ideation without directly asking the patient about the ideas and discussing their meaning" (Freeman & Reinecke, 1993, p. 44).

The following identify a number of areas that should be explored during the clinical interview:

1. Stress: What situational or personal stressors are occurring in the person's life? How are these stressors perceived by the person? (crisis is not caused by events but by the feelings that are generated by the at-risk person's interpretation of the events and what those feelings make the person want to do.) How long standing or unmanageable are the difficulties that precipitated the current crisis?
2. Symptoms: How is the person responding to the stressors?
3. Emotional: Is there mood instability, despondency, depression, emotional numbing, crying, or hopelessness? Is the person feeling more or less troubled than in the past?
4. Cognitive: Is thinking disturbed, bizarre, disorganized, constricted, or irrational? Does the person see few options? Are they overwhelmed? Is judgment impaired? Is there suicidal ideation?
5. Behaviourial: Is there withdrawal, agitation, extreme fatigue, impulsiveness, recklessness, evidence of self-injury, or substance abuse?
6. Physical: Has there been a change in appetite, sleeping patterns, weight, or general appearance? Are there biophysical complaints?
7. Coping Style: Is there a history of poor coping skills (i.e., substance abuse), instability or reckless lifestyle? How has the person dealt with problems in the past and how successful did the person perceive his/her efforts to be. Is the person able to generate options? Does the person have poor decision making skills? Does the person demonstrate cognitive rigidity? Is there a specific psychiatric disorder? Is the individual competent? Individuals who manifest some type of psychiatric disorder are at greater risk to complete suicide (Freeman and Reinecke, 1993).
8. Current Suicide Plan: Is the person considering suicide? Is there a plan? How lethal is the plan? Does the person have the means available? How prepared is the person to carry out the plan? How soon? Is rescue possible? Has the person put things in order for death? In addition to asking the whether the person wants "to commit suicide," it is also useful to ask whether the person wants "to die." These appear to be conceptually different constructs to some suicidal people and may offer a mechanism to tap into the person's "desire to live." Are there unstated beliefs or assumptions about the inevitability of the person's suicide (e.g., I deserve to die. I was never meant to live past the age of 30. My father committed suicide so I have no choice, etc.)?
9. Prior Suicidal Behaviour: Is there a history of prior suicidal behaviours - ideation, gestures, attempts (the ratio of completers to suicide attempters is approximately 1 to 100 [Ramsay et al., 19871). If yes, were precautions taken against discovery? Was the prior attempt(s) carefully planned or impulsive? Was the attempt such that rescue was anticipated or likely?
10. Resources: What internal and external resources does this person have — friends, family, children, interests, strong beliefs against suicide, openness to intervention? Has there been prior counselling or therapy? If yes, how effective did the person perceive it to be? What coping skills does the person bring to this situation (see coping above)? The fewer perceived internal and external resources the individual has, the more at risk the individual becomes. The issue here is to identify the factors that will tap into the individual's "desire to live." 'What are the factors in this person's life that keeps him/ her from killing him/herself? It is hope for the future that keeps human beings alive. What is it that motivates this particular individual to stay alive?
11. Risk to Others/Homicide: -What risk does the suicidal individual pose to others? Homicidal behaviour and suicidal behaviour can co-exist in the same individual, but there is no evidence in the literature at this time to suggest that suicidal individuals are at any greater risk to harm others than are non-suicidal persons.

Each of the above global areas should be explored during the risk assessment phases. The corresponding questions are meant as prompts to cue the clinician to potential areas of exploration. The questions are meant to provide guidelines. They are NOT meant to be proscriptive. It is unlikely that all questions would be appropriate or pertain to all cases.

iii) Clinical judgment
The assessment of suicide risk is a clinical judgment about the future behaviour of another human being. Risk is assessed along a continuum, from low to high risk. Decisions regarding what will be done next are determined by the level of risk assigned to the case. The specific questions are: .

• To what degree is the offender safe from immediate self-generated harm?.
• Has the at-risk person been open and honest in discussing concerns and feelings?.
• Is there any serious psychopathology present?.
• Is the crisis situation likely to escalate in severity to a degree the person would not be able to tolerate?.

"Although no empirical evidence exists that good clinical judgment saves lives, the common consensus among experts remains that this is still our best form of suicide prevention" (Bongar, Maris, Berman, Litman, & Silverman, 1993, p. 255). Psychological assessment tools, an understanding of actuarial data, educational and experiential background, and theoretical orientation are all elements that influence what is essentially hypothesis testing to evaluate the behaviours presented by a unique individual.

The decision that an individual is at imminent risk is extremely difficult to make under any circumstances; when dealing with an incarcerated population, many of the factors discussed above serve to further confound the obscure. In a crisis situation, decisions must be made quickly. Decisions regarding a specific suicidal person cannot be made before hand, so it is necessary to develop a framework for making decisions under adverse conditions.

The following is a suggested framework for formulating decisions for dealing with suicidal individuals in a correctional setting:

1. Establish the parameters of the situation in your work setting. Review the literature, CPA code of ethics and CSC directives,^{Footnote 19} and speak with colleagues so that you develop and are clear about the basis from which you are making your decisions.
2. Know and understand any policies and procedures for handling suicidal clients in your place of employment.
3. Conduct an assessment and determine the offender's level of suicidal risk.
4. Generate alternative decisions. Placing the individual in an observation cell is only one alternative. Can the staff, family and/or other offenders^{Footnote 20} monitor the individual? For example, the individual may be double bunked with another person. Can the individual be seen everyday and a contract with the individual for non-suicidal behaviour be ensured between contacts?
5. Involve the suicidal person as much as possible in the generation of alternatives.
6. Evaluate each alternative in light of the degree of risk and the degree of compliance from the suicidal person.
7. Make a decision.
8. Evaluate your decision in terms of the possible outcomes. If your decision is to place the individual in the observation cell, what are the implications in view of the therapeutic and environmental outcomes? If your decision is to not place the individual in an observation cell, imagine yourself on the stand at an inquest^{Footnote 21} — can you defend your decision? If you believe you cannot, return to step three.
9. Implement your decision and carefully document what you did and the rationale for your decision. What isn't documented is difficult to defend.
10. Inform the appropriate people of the assessed level of risk and any recommended intervention strategies.
11. Monitor the individual until the risk of suicide has abated, continually evaluating your decision in light of any new information.

It must be recognized "that any reasonable standard of care must acknowledge the vital importance of clinical judgment in working with a unique individual; that every decision in clinical practice has both risk and benefit to the patient, and that the task of the clinician, time and again, is to weigh these issues and make a judgment that is competent, prudent, and reasonable" (Bongar et al., 1993, p.246). Clinicians must examine their own practice and ensure their decisions and risk-benefit decisions focus on optimal care and safety for their clients, not defensive reactions based on fear (Klerrnan, 1989). There is "no such thing as a suicide-proof unit" (Simon, 1988, p.95).

In clinical practice, the decision regarding the suicidal individual "is not what to do for all time, but rather what to do next, for the near future" (Porkorny, 1983, p.251). The suicide intervention process is a sequence of small decisions that involves alerting signs, further investigation, additional alerting signs or confirmatory indicators, determination of the level of intervention, type of interventions (e.g., medication, inpatient, outpatient, 24 hour observation) and continual feedback to review the risk assessment and the intervention techniques.

The type of intervention used will be based on the level of assessed risk. In the case of a high risk person, the first task is to reduce the immediate risk of selfharm. Institutionally, this generally means admission to an observation cell. Once the immediate risk has been addressed, a long-term treatment plan needs to be developed to facilitate more adaptive coping. For a low- or moderate- risk individual, the treatment plan goals are to prevent elevation of the suicide risk level.

An antisuicide contract is a necessary component of any prevention or post crisis intervention treatment plan. The antisuicide contract should include:

• agreement that the clinician be informed of any change in suicidal thinking;
• agreement that there be no suicidal behaviour between the current meeting and the next appointment.^{Footnote 22} Time between contacts will depend on the offender's degree of risk;
• the names and roles others can play in supporting the suicidal person; and
• alternatives and/or back-up plans should any part of the plan fail.^{Footnote 23}

As a preventative measure,^{Footnote 24} it is suggested that an antisuicide contract be established whenever one is providing therapy for childhood abuse (physical, sexual, psychological), Post Traumatic Stress Disorder (PTSD) or other situations that previously gave rise to suicidal feelings, thoughts or behaviours. It is particularly important that the limits of confidentiality be clearly stated (Simon 1988, p.60-61) during both the assessment and treatment phases.

When developing a treatment plan, Baumeister (1990) provides an excellent theoretical model for conceptualizing suicide. Suicide is considered in terms of its intrapersonal dimensions. Environmental and interpersonal variables have an impact on the individual, but it is the person's interpretation of those events that influence his/her behaviour (Bandura, 1977; Baumeister, 1990; Ellis & Grieger, 1977; Meichenbaum, 1985; Ramsay et al. 1987). Borrowing from the risk, need, responsivity literature (Andrews, 1980; Andrews, 1982; Andrews, Bonta, Hoge, 1990), clinical interventions with suicide can be similarly conceptualized.

Risk factors related to the prediction of suicide can be dichotomized in terms of static versus dynamic variables. Static risk factors such as age, gender, prior suicidal behaviour, family history of suicidal behaviour and marital status cannot be manipulated (recognizing that age and marital status can change over time). However, the individual's perception of these factors (i.e., suicide is an inherited trait) and the meaning ascribed to the factors in the person's life (i.e., because my father and grandfather committed suicide, I will also die from suicide) can be altered through therapeutic interventions such as cognitive restructuring. Dynamic variables can be altered by both therapeutic intervention and environmental manipulation. Such measures as job training to address unemployment, medication for depression and grief counselling to deal with the loss of a loved one can be provided. Determining the appropriate intervention strategy involves a systematic needs assessment.

The third component of the analysis is responsivity. Baumeister's (1990) model suggests that as the at-risk person moves along the behaviour chain to the final escape behaviour of suicide, responsivity to the external environment becomes less open. The implications of this model are that responsivity in the suicidal individual can be evaluated and interventions used to increase responsivity.

The two components of need and responsivity are interactive. In escape theory, suicidal behaviours develop in response to the negative affects of an aversive state of high self-awareness. In the model's first stage, the individual's current circumstances fall below a personal standard "produced either by unrealistically high expectations or by recent problems or setbacks, or by both. Internal attributions are made, so that these disappointing outcomes are blamed on the self and create negative implications about the self' (Baumeister, 1990, p.91). If the intervention measures taken do not alter the evaluation model of the individual (i.e., the unrealistically high expectations) and the discrepancy between the internalized standard and circumstances is not reduced, the negative affect will continue. If the recent problems or setbacks were external to the individual, such as unemployment or limited social interaction, and the intervention did not bring about some change in the social and/or economic environment of the individual, the negative affect brought about by the discrepancy between the internalized standard and circumstance will reoccur post intervention. Theoretically, by matching the needs and responsivity of the at-risk individual with the intervention strategy, the probability of a successful intervention will dramatically increase.

The following are recommended for further information on the treatment of suicidal behaviour:

• Baumeister, R.F. (1990). "Suicide as an escape from self," Psychological Review, 97(1), 90- 111.
• Freeman, A. & Reinecke, M.A. (1993). Cognitive Therapy of Suicidal Behavior: A Manual for Treatment. New York: Springer Publishing Co.

The assessment and treatment of suicidal people is a very difficult task and requires that the psychologist have specialized training in suicide intervention. Few graduate schools in psychology offer specific training in suicide intervention^{Footnote 25} (Report on the National Task Force on Suicide in Canada, 1987), yet failure to prevent suicide is one of the leading reasons for successful malpractice suits against professionals working in the mental health field (Daly, 1993; Szasz, 1986). Increasingly, licensing boards are scrutinizing the training of their members in the area of suicide intervention and, unfortunately, some are being found to be deficient in the skill. ^{Footnote 26}

They suggest that the minimum informational requirement for working with suicidal individuals is a course or workshop on suicide intervention which includes a framework for evaluating level of risk and an intervention model such as the Suicide Intervention Workshop provided by Living Works Education, Inc., Calgary. This two-day workshop has been adopted by CSC and will provide the intervener with the basic skills in suicide intervention. Additionally, reading the following will augment your knowledge base:

• Bongar, B. (1991). The Suicidal Patient: Clinical and Legal Standards of Care. Washington, D.C.: American Psychological Association.
• Maris, R.W., Berman, A.L., Maltsberger, J.T. & Yufit, R.I.(Eds.). (1992). Assessment and Prediction of Suicide. New York: Guilford.

However, clinical competency in managing suicidal clients is not restricted to informational competence alone (Berman, 1986). Bongar (1992) discusses the ethical need for clinical competence in the assessment and management of suicidal persons prior to beginning independent practice in clinical settings. A list of specific competencies are identified to facilitate the evaluation of one's own technical and personal abilities.

The successful suicide of an at-risk individual is not widely discussed in the literature.^{Footnote 27} Death of a client while under the care of a professional is deemed by clinicians to be one of the most stressful occupational events they are likely to experience (Deutsch, 1984). Chemtob, Hamada, Bauer, Torigoe and Kinney (1988) found that psychologists who had lost a client to suicide responded to the loss in a similar manner to individuals who had experienced the death of a family member.

A sense of helplessness occurs when we realize that when a person is determined to die, there is little we can do to prevent the suicide. Suicide intervention and prevention programs are based on the assumption that the suicidal individual is ambivalent and that through therapeutic intervention she or he will move from a past-oriented, death-focused goal, to a future-oriented, life-focused goal (Ramsay et al., 1987). But there are those who make the decision to die in spite of all our care, support, therapy and hospitalization.^{Footnote 28} They simply wait us out until they find the right time, place or circumstance where we are unable to prevent their deaths. Their suicide leaves us with the question, did we do everything we could?

Andrews, D.A. (1980). "Some experimental investigations of the principles of differential association through deliberate manipulations of the structure of service systems," American Sociological Review, 45, 448-462.

Andrews, D.A. (1982). A Personal, Interpersonal and Community- Reinforcement Perspective on Deviant Behaviour (PIC-R). Toronto: Ministry of Correctional Services, Planning and Research Branch. ISBN 0-7743-8181-7.

Andrews, D.A., Bonta, J. & RD. Hoge. (1990). "Classification for effective rehabilitation: Rediscovering psychology," Criminal Justice and Behavior, 17 (1), 19-52.

Atkinson, M., Kessel, N. & J.B. Dalgaard. (1975). "The comparability of suicide rates," British Journal of Psychiatry, 127, 247-256.

Balon, R. (1987). "Suicide: Can we predict it?," Comprehensive Psychiatry, 28 (3), 236-241.

Bandura, A. (1977). "Self-efficacy: Toward a unifying theory of behavioral change," Psychological Review, 84 (2), 191-215. Baumeister, R.F. (1990). "Suicide as escape from self," Psychological Review, 97 (1), 90-113.

Beck, A.T., Kovacs, M. & A. Weissman, A. (1975). "Hopelessness and suicidal behavior: An overview," Journal of the American Medical Association, 234, 1146-1149.

Beck, A.T., Schuyler, D. & I. Herman. (1974). "Development of suicidal intent scales," in A.T. Beck, H.LP. Resnik and D.J. Lettiere (Eds.), The Prediction of Suicide. Bowie, MD: Charles Press Publishers.

Beck, A.T. & Steer R.A. (1989). "Clinical predictors of eventual suicide: A five-10 year prospective study of suicide attempters," Journal of Affective Disorders, 17, 203-209.

Beck, A.T., Steer, R.A., Beck, J.S. & C.F. Newman. (1993). "Hopelessness, depression, suicidal ideation and clinical diagnosis of depression," Suicide & Life-Threatening Behavior, 23 (2), 139-145.

Berman, A. (1986). "A critical look at our adolescence: Notes on turning 18 (and 75)," Suicide and Life-Threatening Behavior, 16 (1), 1- 12.

Bongar, B. (St. M. Harrnatz. (1989). "Graduate training in clinical psychology and the study of suicide," Professional Psychology: Research and Practice, 20 (4), 209-213.

Bongar, B. (1991a). "Clinical psychology graduate education in the study of suicide: Availability, resources, and importance," Suicide and Life-Threatening Behavior, 21(3), 231-244.

Bongar, B. (1991b). The Suicidal Patient: Clinical and Legal Standards of Care. Washington, D.C.: American Psychological Association.

Bongar, B. (1992). "The ethical issue of competence in working with the suicidal patient," Ethics and Behavior, 2 (2), 75-89.

Bongar, B., Maris, R.W., Berman, A.L., Litman, R.E. & M.M. Silverman. (1993). "Inpatient standards of care and the suicidal patient. Part I: General clinical formulations and legal considerations," Suicide and Life-Threatening Behavior, 23 (3), 245-256.

Bonner, R.L. (1992). "Isolation, seclusion and psychosocial vulnerability as risk factors for suicide behind bars," in R.W. Maris, A.L. Berman, J.T. Maltsberger & R.I. Yufit (Eds.), Assessment and Prediction of Suicide (pp. 398-419). New York: Guilford Press.

Brugha, T. & D. Walsh. (1978). "Suicide past and present - the temporal constancy of under-reporting," British Journal of Psychiatry, 132, 177-179.

Burtch, B.E. (Si. R.B. Ericson. (1979). The Silent System: An Enquiry into Prisoners Who Suicide. Toronto: University of Toronto, Toronto Centre of Criminology.

Canadian Psychological Association. (1986). A Canadian Code of Ethics for Psychologists.

Chemtob, C.M., Hamada, R.S., Baiser, G.B., Torigoe, R.Y. & B. Kinney. (1988). "Patient suicide: Frequency and impact on psychologists," Professional Psychology: Research and Practice, 19 (4), 421-425.

Clopton, J.R. & D.H. Baucom. (1979). "MMPI ratings of suicide risk," Journal of Personality Assessment, 43 (3), 293-296.

Clopton, J.R. & W.C. Jones. (1975). "Use of the MMPI in the Prediction of Suicide," Journal of Clinical Psychology, 31, 52-53.

Clum, G.A. (1987). "Abandon the suicidal? A reply to szasz," American Psychologist. September, 883-885.

Cole, D.A. (1988). "Hopelessness, social desirability, depression, and parasuicide in two college student samples," Journal of Consulting and Clinical Psychology, 56 (1), 131-136.

Cook, B.L. (1984). "Suicide: A social enigma or an opportunity for further development in prevention?" New Directions for Mental Health Services, 23, 59-69.

Daly, R.J. (1993). "Suicide in depressed patients: Medico-legal issues," British Journal of Psychiatry, 163 (20), 29-32.

Davis, F.B. (1967). "The relationship between suicide and attempted suicide," Psychiatric Quarterly, 41, 752-765.

Deutsch, C.J. (1984). "Self-report sources of stress among psychotherapists," Professional Psychology: Research and Practice, 15, 833-845.

Diggory, J.C. (1974). "Predicting suicide: Will-o-the-wisp or reasonable challenge?," in A.T. Beck, H.L.P. Resnik & D.J. Lettiere (Eds.), The Prediction of Suicide. Bowie, MD: Charles Press Publishers.

Dorpat, T.L. & H. Ripley. (1967). "The relationship between attempted suicide and completed suicide," Comprehensive Psychiatry, 8, 74-89.

Douglas, J.D. (1967). The Social Meaning of Suicide. Princeton: Princeton University Press.

Ellis, A. & R. Grieger. (1977). Handbook of Rational-Emotive Therapy. New York: Springer.

Evenson, R.C. (1983). "Community adjustment of patients who threaten and attempt suicide," Psychological Reports, 52, 127-132.

Exner, J.E. & J. Wylie. (1977). "Some Rorschach data concerning suicide," Journal of Personality Assessment, 41(1), 339-347.

Farber, M.L. (1968). Theory of Suicide. New York: Funk and Wagnalls.

Farberow, N.L. (1974). "Use of the Rorschach in prediction and understanding suicide," Journal of Personality Assessment, 38 (5), 411- 419.

Kleespies, P.M. (1993). "The stress of patient suicidal behavior: Implications for interns and training programs in psychology," Professional Psychology: Research and Practice, 24 (4), 477-482.

Klerman, G.L. (1989). "Psychotherapy on Trial? Osheroff versus Chestnut Lodge," Harvard Medical School Mental Health Letter, 6 (1), 4.

Laprade, M. (1992). "The question of liability in inmate suicides," Forum on Corrections Research, 4 (3), 37-39.

Lester, D. (1970). "Relation between attempted suicide and completed suicide," Psychological Reports, 27, 719-722.

Lester, D. (1974). "Demographic versus clinical prediction of suicide behaviors: A look at some issues," in A.T. Beck, H.L.P. Resnik, and D.J. Lettiere (Eds.), The Prediction of Suicide (pp. 71-85). Bowie, MD: Charles Press Publishers.

Liberakis, E.A. & J. Hoenig. (1978). "Recording of suicide in Newfoundland," The Psychiatric Journal of the University of Ottawa, 3 (4), 254-259.

Lloyd, C. (1990). "Suicide and self-injury in prison: A literature review," Home Office Research and Planning Unit Report. London: Her Majesty's Stationery Office.

Marks, A. & T. Abernathy. (1974). "Toward a sociocultural perspective on means of self-destruction," Life Threatening Behavior, 4, 3-17.

Mather, D.B. (1987). "The case against preventing suicide prevention: Comments on Szasz," American Psychologist, September, 882-883.

McCarthy, P.D. & D. Walsh. (1975). "Suicide in Dublin, part I: The under-reporting of suicide and the consequences for national statistics," British Journal of Psychiatry, 126, 301-308.

Meichenbaum, D. (1985). Stress Inoculation Training. New York: Pergamon.

Murphy, G.E. (1974). "The clinical identification of suicidal risk," in A.T. Beck, H.LP. Resnik and D.J. Lettiere (Eds.), The Prediction of Suicide (pp. 109-118). Bowie, MD: Charles Press Publishers.

Farberow, N.L. & D. MacKinnon. (1974). "A suicide prediction schedule for neuropsychiatrie hospital patients," The Journal of Nervous and Mental Disease, 158 (6), 408-419.

Farberow, N.L., MacKinnon, D. & F.L. Nelson. (1977). "Suicide: Who's counting?" Public Health Reports, 92 (3), 223-232.

Freeman, A. & M.A. Reinecke. (1993). Cognitive Therapy of Suicidal Behavior. New York: Springer Publishing.

Heney, J. (1990). "Self-injurious behaviour and the role of sexual abuse in this phenomena," in L. Rosine (Chair), The Victim Cycle. Symposium conducted at the Ontario Region Case Management Conference, Honey Harbour, Ontario.

Herman, J.L. (1992). Trauma and Recovery. New York: BasicBooks.

Holden, R. & J.D. Mendonca. (1984). "Hopelessness, social desirability and suicidal behavior: A need for conceptual and empirical disentanglement," Journal of Clinical Psychology, 40 (6), 1342-1345.

Holden, R.R., Mendonca, J.D. & RC. Serin. (1989). "Suicide, hopelessness and social desirability: A test of an interactive model," Journal of Consulting and Clinical Psychology, 57 (4), 500-504.

Holmes, C.B. (1987). "Comment on Szasz's view of suicide prevention," American Psychologist, September, 881-882.

Kendra, J.M. (1979). "Predicting suicide using the Rorschach Inkblot Test," Journal of Personality Assessment, 43 (5), 452-456.

Murphy, G.E. (1984). "The prediction of suicide: Why is it so difficult?," American Journal of Psychotherapy, 38 (3), 341-349.

Murphy, G.E, & RD. Wetzel. (1990). "The lifetime risk of suicide in alcoholism," Archives of General Psychiatry, 47, 383-391.

Nevid, J.S. (1983). "Hopelessness, social desirability and construct validity," Journal of Consulting and Clinical Psychology, 51, 139-140.

Neuringer, C. (1962). "Methodological problems in suicide research," Journal of Consulting Psychology, 26, 273-278.

Neuringer, C. (1974). "Suicide and the Rorschach: A rueful postscript," Journal of Personality Assessment, 38 (6), 535-539.

Nuffeld, J. (1982). Parole Decision Making in Canada: Research Towards Decision Guidelines. Ottawa: Solicitor General of Canada.

Ovenstone, I.M.K. (1973). "A psychiatric approach to the diagnosis of suicide and its effect upon the Edinburgh statistics," British Journal of Psychiatry, 123, 15-21.

Pallis, D.J. & J. Birtchnell. (1977). "Seriousness of suicide attempt in relation to personality," British Journal of Psychiatry, 130, 253-259.

Pallis, D.J., Gibbons, J.S. & D.W. Pierce. (1984). "Estimating suicide risk among attempted suicides II. Efficiency of predictive scales after the attempt," British Journal of Psychiatry, 144, 139-148.

Pierce, D.W. (1981). "The predictive validation of a suicide intent scale: A five year follow-up," British Journal of Psychiatry, 139, 391- 396.

Pokorny, A.D. (1974). "A scheme for classifying suicidal behaviors," in A.T. Beck, H.L.P. Resnik and D.J. Lettiere (Eds.), The Prediction of Suicide (pp. 29-44). Bowie, MD: Charles Press Publishers.

Pokorny, A.D. (1983). "Prediction of suicide in psychiatric patients," Archives of General Psychiatry, 40, 249-257.

Ramsay, R.F., Tanney, B.L., Tierney, R.J. & W.A. Lang. (1987). A Suicide Prevention Training Program: Trainer's Handbook.

Report on the National Task Force on Suicide in Canada. (1987). Ottawa: Health and Welfare Canada, H39-107/1987E.

Rosen, A. (1954). "Detection of suicidal patients: An example of some limitations in the prediction of infrequent events," Journal of Consulting Psychology, 18, 397-403.

Rosine, L. (1991). Predictor Variables in Completed Suicides. Unpublished manuscript, Carleton University, Ottawa, Ontario.

Rosine, L. (1994). Childhood Abuse, Suicide and Self-injurious Behaviours Data. Unpublished raw data.

Roy, A. & M. Linnoila. (1986). "Alcoholism and suicide," Suicide and Life-Threatening Behavior, 16 (2), 244-273.

Shaffer, J.W., Perlin, S., Schmidt, C. & J.H. Stephens. (1974). "The prediction of suicide in schizophrenia," Journal of Nervous and Mental Disease, 159 (5), 349-355.

Shneidman, E. (1985). Definition of Suicide. New York: John Wiley & Sons.

Simon, RI. (1988). Concise Guide to Clinical Psychiatry and the Law. Washington, DC: American Psychiatric Press.

Simon, R.L. & P.I. Murphy. (1985). "Sex differences in the causes of adolescent suicide ideation," Journal of Youth and Adolescence, 14 (5), 423-434.

Smith, K. (1982-1983). "Using a battery of tests to predict suicide in a long-term hospital: A clinical analysis," OMEGA, 13 (3), 261-275.

Steer, R.A., Beck, A.T., Garrison, B. & D. Lester. (1988). "Eventual suicide in Interrupted and uninterrupted attempters: A challenge to the cry-for-help hypothesis," Suicide and Life-Threatening Behavior, 18 (2), 119-128.

Stengel, E. (1971). Suicide and Attempted Suicide. Middlesex, England: C. Nicholls & Co.

Strosahl, K.D., Linehan, M.M. & J.A. Chiles. ( 1984). "Will the real social desirability please stand up? Hopelessness, depression, social desirability and the prediction of suicidal behavior," Journal of Consulting and Clinical Psychology, 52 (3), 449-457.

Szasz, T. (1986). "The case against suicide prevention," American Psychologist, 41 (7), 806-812.

Szasz, T. (1987). "Response to Clum," American Psychologist, September, 885-886.

Creating Choices: The Report of the Task Force on Federally Sentenced Women. (1990). Ottawa: Correctional Service of Canada.

Tomlinson-Keasey, C., Warren, J.W. & J.E. Elliott. (1986). "Suicide among gifted women: A prospective study," Journal of Abnormal Psychology, 95 (2), 123-130.

Watson, C.B., Klett, W.G., Walters, C. & P. Vassar. (1984). "Suicide and the MMPI: A cross-validation of predictors," Journal of Clinical Psychology, 40 (1), 115-118.

Weisman, A.D. (1976). "Is suicide a disease?," Life Threatening Behavior, 1 (4), 219-231.

Wetzel, R.D. (1976). "Hopelessness, depression and suicide intent," Archives of General Psychiatry, 33, 1069-1073.

Wetzel, R.D., Margulies, T., Davis, R & E. Karam. (1980). "Hopelessness, depression, and suicide intent," Journal of Clinical Psychiatry, 41 (5), 159-160.

Wilkins, J. (1967). "Suicidal behavior," American Sociological Review, 32, 286-298.

Wilson, M. (1981). "Suicidal behavior: Toward an explanation of differences in female and male rates," Suicide and Life-Threatening Behavior, 11 (3), 131-140.

Zubin, J. (1974). "Observations on nosological issues in the classification of suicidal behavior," in A.T. Beck, H.L.P. Resnik and D.J. Lettiere (Eds.), The Prediction of Suicide (pp. 3-28). Bowie, MD: Charles Press Publishers.