Telematic Review
Area: Suicidio e suicidologia

Suicidal behaviors and personality

Giulia Grava (1), Guseppe Berti Ceroni (2), Paola Rucci (3), and Paolo Scudellari (4)

1) Psychologist, Department of Psychiatry and Psychology, University of Bologna, Italy.
2) Psychiatrist, Neuropsychiatric Private Hospital “Villa Baruzziana”, Bologna, Italy. Professor of Psychotherapy, School of Psychiatry, University of Bologna, Italy.
3) Statistician, Department of Psychiatry, Neurobiology, Pharmacology and Biotechnology, University of Pisa, Pisa, Italy.
4) Associate Professor, Department of Psychiatry and Psychology, University of Bologna, Bologna, Italy.


This paper suggests the potential clinical usefulness of two projective tests (Rorschach and Object Relation Technique) and of a clinical interview focused on the pathway to suicide, life events and major life difficulties to better define subtypes of patients attempting suicide. Thirty-three hospitalized subjects who had committed a suicide attempt in the previous 6 months were examined using an in-depth assessment of the pathway to suicide. The aims of this study are: (1) to compare, within a clinical sample of suicidal attempters, the psychopathological, personality and psychosocial characteristics of subjects with mood disorders alone, personality disorders and dysfunctional personality alone and mood disorders with personality disorders and/or dysfunctional personality; (2) to analyse the relationship between the method chosen and the clinical characteristics of the subgroups.


Knowledge on suicidal phenomena is constantly increasing. In the last decades studies have focused on suicidal acts and attempts in patients with personality disorders (Linehan, 1986; Soloff, Lis, Kelly, Cornelius, & Ulrich, 1994; Mann, Waternaux, Haas, & Malone, 1999; Stanley, Gameroff, Michalsen, & Mann, 2001; Bertolote, & Fleischmann, 2002), on the continuity between successful suicides and those attempted or aborted (Beck, & Lester, 1976; Maris, 1981; Malone, Haas, Sweeney, & Mann, 1995; Barber, Marzuk, Leon, & Portera, 1998) and on the need to examine in every subject the pathway to suicide using an in-depth clinical approach (Hendin, Maltsberger, Lipschitz, Haas, & Kyle, 2001; Maltsberger, 2002; Maltsberger, 2004) and a series of tests (Rudd, Ellis, Rajab, & Wehrly, 2000).

Suicide is largely associated with depressive states (Goodwin, & Jamison, 1990; Malone, Haas, Sweeney, & Mann, 1995). Recent epidemiological studies provided consistent evidence that depressive states are the most frequent suicide cause but that borderline personality disorder is the second cause (Bertolote, & Fleischmann, 2002). Soloff et al. (2000) compared the characteristics of suicidal attempters with major depression, borderline personality disorder or with both disorders to determine if the characteristics of depression (hopelessness) and of borderline personality disorder (impulsivity, i.e. bad impulse management, and helplessness) were predictive of the number of suicide attempts, wish to die, planning degree, physical damage and the suicidal method chosen. They found that patients with major depression and borderline personality disorder realized a higher number of attempts and had a higher level of planning.

Linehan (1986) was the first to stress that suicide attempts in personality disorders are in a continuum with self-mutilating acts in all characteristics except the wish to die and Stanley et al. (2001) confirmed that patients with a borderline personality disorder were fully intentional about committing suicide also when they adopted the same methods used for self-mutilating acts. This has relevant clinical implications because clinicians usually interpret these suicide attempts as not fully intentional (Bongar, Peterson, Golan, & Hardimann, 1990).

Barber, Marzuk, Leon, & Portera (1998) provided evidence that aborted suicidal attempts, i.e. attempts not carried through completely because the individual changed his/her mind immediately before the act, are characterized by the same lethality and/or intentionality degree as those in actual suicide attempts. Moreover, subjects who had made one aborted suicide attempt nearly twice as likely to have made an actual suicide attempt than as subjects with no aborted attempts.. Indeed, aborted suicide attempts have a great predictive value and must be studied, because they fully belong to the personal suicidal history (Beck & Lester, 1976; Maris, 1981; Malone, Haas, Sweeney, & Mann, 1995).

Investigation of the pathway to suicide implies the analysis of ‘crisis markers’ (Hendin, Maltsberger, Lipschitz, Haas, & Kyle, 2001), including cognitive, emotional and psychosocial variables.

A well-known cognitive marker is the ‘cognitive constriction’ described by Shneidman (1985): the more the subject thinks about suicide, the more it seems to be ‘the only way out’.

The emotional states vary from anguish to rage, guilt, narcissistic wounding, sense of loss. Maltsberger (2002; 2004) described ‘desperation’ as an emotional state characterized by anguish and the prompt need of relief. This key aspect of emotional states leads subjects to search for an urgent solution, inducing impulsivity also in people without particular problems in the area of impulse management. Maltsberger described all the phases of the suicidal path: the subject is able to manage his ‘flooding’ feelings and thus can overcome the acute phase of the crisis, or he/she becomes overwhelmed and or attempts to commit suicide. To manage the ‘emotional flooding’ the subject frequently uses means that are usually considered as markers of a bad impulse management, such as substance or food abuse, sexual promiscuity, self-mutilating acts (Apter, Plutchick, & Van Praag, 1993; Hendin, Maltsberger, Lipschitz, Haas, & Kyle, 2001), which, in this condition, are paradoxically to be considered Ego functions and not impulsivity markers (Maltsberger, 2002).

Clinical diagnosis, severity of depression, and psycho-social conditions are easily determined in the usual psychiatric routine. The need to deeply evaluate the personality structure and the path to suicide of subjects admitted for a recent suicidal act led us to use two well-known projective tests, the Rorschach Ink-blot test (Rorschach, 1921) and the Object Relation Technique (Philippson, 1955) and a clinical interview focused on a detailed examination of the suicidal path.

The use of projective tests and of the focused interview may be useful to examine the subjective viewpoint on the method to commit suicide. The most relevant criteria examined in literature (Jamison, 1999) are the method availability, specific cultural aspects, the spreading of a method used by others and made known through the media ( ‘the Werther effect’). These criteria are relatively external to the personality structure and to the psychopathology. Considering the viewpoint of the potential suicidal person, we found that subjects identify some methods as ‘hard’ and others as ‘soft’.

The aims of this study are: (1) to compare, within a clinical sample of suicidal attempters, the psychopathological, personality and psychosocial characteristics of subjects with mood disorders alone, personality disorders and dysfunctional personality alone and mood disorders with personality disorders and/or dysfunctional personality; (2) to analyse the relationship between the method chosen and the clinical characteristics of the subgroups. Our hypothesis is that ‘soft’ methods are related to steady premeditation and suicide planning and ‘hard’ ones to bad impulse management.



Subjects were recruited from patients consecutively admitted over six months. for a suicide attempt to three inpatient units (Psychiatric Clinic, University of Bologna, Villa Baruzziana, Bologna and La Luna, Ferrara) of two towns of North-Eastern Italy. The study sample constitutes 3.7% of patients admitted to the three units. These units mostly admit patients referred from outpatient clinics to monitor the change of treatment or to alleviate the burden of caregivers when needed for a short period of time.

The definition of the suicide attempt we adopted was that used by Mann et al. (1999): a self-destructive act that was sufficiently serious to require medical evaluation and that was carried out with the intention of ending one’s life. Inclusion criteria were: suicide attempt committed within 6 months from the index assessment, age between 16 and 85, ability to speak Italian fluently. Exclusion criteria were: current severe psychotic symptoms, severe cognitive impairment, drug abuse as the only diagnosis, life-threatening physical illness.

All patients were voluntary and provided a written informed consent to participate in the study. None of the patients refused.

The sample includes 33 subjects (28 F, 5 M) with a mean age of 43.4 years (SD 16.2, range 17-82). All subjects were Caucasian, and 32 out of 33 were Italian citizens.

Three percent of subjects were unable to read and write, 21.2% had a primary school diploma, 45.4% had a middle school diploma, 9% had a high school diploma, 3% had a college diploma.

Fifteen percent were students, 9% unemployed, 12% housewives, 46% employed, 18% retired.

Twelve had a current depressive episode, 6 a bipolar disorder, 7 a situational depression, 6 other diagnoses (3 eating disorder, 2 fictitious/histrionic disorder and 1 substance dependence). Ten subjects had borderline personality disorder and 11 other personality disorders. Fifteen subjects were at their first suicide attempt, 5 had committed one previous attempt, 5 two attempts, 8 more than two.


Patients were assessed as soon as the psychotic and melancholic features had remitted (on average 14.7 ± 20.3 days (median 7 days) after the suicide attempt) by means of the Rorschach test (1921) and the Object Relation Technique (O.R.T, Phillipson, 1955). These tests were administered by one of the authors (GG), who also conducted the focused clinical interview. The procedure usually chosen was to administer the Rorschach test first, because of its high emotional impact, then the O.R.T because this thematic test offers a broader view of the personality structure and also smoothly leads to the clinical interview. This interview was focused on all the crisis markers and their sequence in the pathway to suicide and on the subjective feelings accompanying every step. This method was accepted very well: subjects usually appreciated the direct questions and expressed their feelings and views in depth.

Diagnoses were made using ICD-IX-CM criteria (WHO, 1997), by the second author (GBC), a psychiatrist with a long-standing research and clinical experience, who conducted face-to-face clinical interviews with all study participants. The diagnosis was reviewed in small groups including this author, the psychiatrist responsible for the patient during the admission and the psychologists who administered the study instruments until consensus was reached. The diagnostic assessment also included the Hamilton Depression Rating Scale (Hamilton, 1960) and the Italian translation of the Structured Clinical Interview for DSM-IV Personality Disorders (SCID-II) (Mazzi et al., 2003). A neuroimaging examination and multiple cognitive tests to evaluate cognitive impairment were conducted when requested by the clinicians.

The history of suicide attempts and life events and major difficulties was obtained by complementing the medical records with the data obtained during the clinical interview. Life events and major difficulties were assessed using the checklist developed by Berti Ceroni et al. (1996).


The Rorschach Ink-blot test was assessed by using some of the Klopfer and Davidson (1962) indexes and other indexes developed by Argentine authors with reference to the Roman Rorschach School (Herrera, Orcoyen, & Passalacqua, 1995; Passalacqua, 1995; Passalacqua, 1999). These indexes allowed us to determine:

- the quality of the mood, defined by the number of answers, a high number of animal contents, a high number of answers where the shape was the main feature that ‘created’ the answer, the presence of deterioration, few answers determined by textile shading and achromatic color.

- bad impulse management, defined by answers in which chromatic color, and not the shape of the ink-blot, was the most important feature, a high number of animal movements- considered as indicators of impulses not acceptable by the subject because too primitive- and a small number of human movements- considered as indexes of Ego function.

- the degree of aggression, defined by the number of white spaces themselves ‘creating’ the answer, by criticism towards the test and/or the interviewer, by aggressive movements and contents.

The O.R.T and the focused interview contributed to define the personality structure. While stories created by subjects on the thematic figures of the O.R.T gave only suggestions to the interviewer, despite delicate questioning in order to detail and specify some aspects of the stories, the focused interview was characterized by direct questions on the subject’s emotional and cognitive points of view to obtain maximum detail and precision.

In order to define synthetically the personality structure, we developed two new concepts: negative life balance and personality dysfunction.

Negative life balance is a combination of Maltsberger’s concept of ‘desperation’ and Shneidman’s concept of ‘cognitive constriction’: it is not the cognitive side of an affect, as in the case of hopelessness, but rather the conclusion of a coherent and detailed cognitive evaluation of one’s own internal and external conditions. This evaluation has to do with the severity and the chronic course of the mental disorder, with personality dysfunctions and with the relevance of psycho-social conditions. Negative life balance can be derived from the clinical history and from the projective tests but usually it emerges directly during the focused interview as well as during a psychotherapy.

Dysfunctional personality regards other aspects described by Maltsberger (2002, 2004) as markers in the pathway to suicide: the ‘flooding’ of emotions, the impulsive reactions necessary to survive and the need of prompt relief. Dysfunctional personality can be derived from the clinical history and from the focused clinical interview, but, in our experience, its emotional nature can be better captured by the projective tests and, by some indexes of the Rorshach to evaluate mood, bad impulse management and aggression.

In close relation with this new concept, helplessness is intended as the inability to use others to alleviate the emotional state, either for a compelling choice to consciously reject all help or for a conflict between allowing to be helped to live and acting with total self-neglect This state can be better derived from the O.R.T and the focused interview.

The method used to commit the index suicide attempt was recorded during the interview and classified as soft or hard. ‘Hard’ were the methods imagined to cause death through pain, tearing the body to pieces, bleeding to death, choking, i.e. hanging oneself, throwing oneself under a train or out of a window, shooting oneself, poisoning, inducing an embolism. The ‘soft’ methods instead were swallowing medications, cutting one’s veins, abandoning oneself to the current to the point of drowning, suffocating with a plastic bag, turning on the gas. All these ‘soft’ methods can be atrocious, but they are probably thought of not in terms of self-aggressiveness, but as a liberating act: for example, a subject begins cutting his/her veins imagining he/she will invoke a less traumatic death through slow blood loss, to gradually pass out and so die, while instead, he/she stops halfway through as the pain is unbearable.

Statistical analysis

Chi-square with exact significance levels was used to compare the frequency of categorical variables across groups. Non-parametric (Kruskal-Wallis) analysis of variance was used to compare continuous variables across groups. When these tests were significant, post-hoc pairwise comparisons were conducted at the p-level of 0.05 .


Subjects were assigned to 3 mutually exclusive groups based on the clinical diagnosis and the results of the tests and of the clinical interview, the first characterized by mood disorders (MD), the second by an Axis-II disorder and dysfunctional personality (PD) and the third by the presence both of a mood disorder and an Axis-II disorder and/or dysfunctional personality (MD+PD). The demographic and clinical characteristics of the three subgroups are provided in Table 1.

Patients with MD alone were significantly older and less frequently employed than the other groups. ICD-IX-CM diagnoses were bipolar disorder in 2 subjects and unipolar disorders in 6 subjects of the MD group. In the MD+PD group, 4 subjects had bipolar disorder, 6 unipolar disorder and 2 situational depression. The PD group included subjects with situational depression, eating, histrionic and substance dependence disorders (38.4%, 27.1%, 15.4% and 7.7%, respectively).

The severity of depression, measured using the Hamilton depression rating scale, was significantly higher in subjects with MD alone as compared to subjects with PD.

Six subjects with PD and 4 MD+PD had borderline personality disorder. Other personality disorders in the DP group were in cluster B (N=5) or B and C (N=2). In the MD+PD group, personality disorders were in cluster A and C (N=2) or B (N=1) and C (N=1).

The frequency of intentionality and hopelessness did not differ significantly among the subgroups. Helplessness was significantly more frequent in the PD and MD+PD groups than in MD. Negative life balance was significantly more frequent in the MD+PD group than in the MD alone group and was absent in the PD subjects.

Major difficulties were reported significantly more frequently by MD than PD patients, while the opposite was found for life events. Previous attempts were more common in the two subgroups with personality disorders and/or dysfunctional personality, and the choice of a hard method to commit the suicide attempt was more frequent in MD patients than in the other groups (Table 1), but these differences failed to reach significance.


To our knowledge, this is the first study that combines clinical diagnosis, projective tests and a clinical interview focused on the pathway to suicide in order to better define suicide attempts. The Rorschach test is useful for assessing the quality of mood, the degree of bad impulse management and the degree of aggression, thus defining the complexity of personality structure. The O.R.T was used to specifically evaluate feelings of hopelessness and helplessness, the latter being the result of both the personality structure and the environment as perceived. The different nature of the data obtained by the two projective tests clarifies our choice. It was amazing how the path we built theoretically also worked in practice: the severe discipline that the Rorschach requires from both the subject and the interviewer seemed to encourage the subject to speak about the content of the various tables with reference to him/herself. Then the O.R.T created such an interpersonal alliance that, even if the last table was blank, the subject was usually tempted to go beyond the assigned task.. The subject in fact was not dismayed by the white card but was so eager to go on speaking that it was natural to begin with the clinical interview. We think that such a procedure is worth further investigation to verify its validity in larger samples and in follow-up studies.

In this study, three groups of suicide attempters emerged, one with mood disorders alone (MD), one with personality disorders and dysfunctional personality (PD) and one with both mood disorders and personality disorders and dysfunctions (MD+PD). Our study was stimulated by a paper of Soloff et al. (2000), who showed that subjects with major depression and borderline personality disorder have a higher number of attempts and degree of suicide planning than patients with either major depression or borderline personality disorder. Our classification extends that of Soloff et al. to subjects with any kind of personality disorder and/or dysfunctional personality and is grounded on different concepts and measures. In spite of this, our results indicate a higher, although not significantly, number of attempts in patients with MD+PD and suggest that the large majority of these subjects have a negative life balance, which is consistent with a high degree of suicide planning.

Compared to patients with DP and MD+PD, patients with mood disorders alone were older and were experiencing more major difficulties. Patients with PD alone were young or middle-aged and showed a constellation of clinical characteristics associated with bad impulse management, including eating, histrionic and substance dependence disorders, in addition to helplessness and a relevant sensitivity to life events. The MD+PD subgroup was not, as might appear at first glance, a mingled area of characteristics belonging to the other subgroups, but had a coherent profile given by the interweaving of personality disorder and/or dysfunctional personality, psychosocial difficulties and depression. The key feature of this subgroup is to draw up a negative life balance.

An open question is the overlap between the concepts of negative life balance and hopelessness. In our opinion hopelessness is the cognitive feature of depressive affects, while negative life balance combines emotional and cognitive features in a coherent and detailed evaluation of a subject’s internal and external conditions, such as the severity and the chronic course of mental disorders he/she suffered from, the personality dysfunctions and the relevance of negative psycho-social conditions. Negative life balance can be derived from the clinical history and from the projective tests but usually it emerges directly during the focused interview as well as during a psychotherapy. Pessimism, considered by many authors (Mann, Waternaux, Haas & Malone, 1999; Oquendo, Galfalvy, Russo, Ellis, Grunebaum, Burke, Mann, 2004; Mann, 2004), in addition to aggressive/impulsive traits, as the temperamental diathesis to suicidal behavior in unipolar and bipolar patients, may be closely related to our new concept. In the psychoanalytic field, Gabbard (2003) clearly outlines the concept we propose: “Suicidal patients have determined that life has little to offer, and analysis is a dubious proposition. What insight could possibly transform life into a journey worth traveling?” (pag.250).

Negative life balance usually was either expressed by subjects or recognized by the interviewer through the answers to tests or obtained by the psychiatrist from the life history and the medical records. In one case the information collected was incomplete: a subject who reacted to projective tests did not give enough material to allow us to determine a negative life balance and did not express it directly in the focused interview; despite this, his long and dramatic life history and his miserable conditions evoked such strong worries and culpability in the interviewer and in the psychiatrist that they experienced true emotional and behavioral turmoil for some days, as if the subject had dissociated (Maltsberger, 2004) and projected his desperation and bad planning about his life. Dissociation between subjective and objective feelings may represent a problem in the definition of this relevant marker of tendency to suicide.

The large majority of the subjects (27/33) chose a soft method and in particular drug poisoning. Only in the MD group about one third of the subjects used a hard method. The Hamilton score in MD subjects that chose a hard method was not significantly higher than that of subjects who chose a soft method (Mann-Whitney U=13, p=0.792). Our hypothesis that the maximum of premeditation and planning and the minimum of impulsivity that is typical of severe depression correlated with the choice of a soft method and the maximum of impulsivity triggered by a life event correlated with the choice of a hard method is not confirmed.

Classification of suicide attempters into mutually exclusive groups, that Soloff et al. (2000) did on a nosographic basis, and that we propose on a broader basis, may have a heuristic value and inform different treatment and prevention strategies. Maltsberger’s (2002, 2004) and Gabbard’s (2003) refined and persuasive clinical observations, on which our methodology was partly grounded, may not regard all the subjects attempting suicide, but only a part of them. It might be hypothesized that Maltsberger described his ‘descent into suicide’ bearing mostly in mind subjects with dysfunctions in personality and helplessness, who thus end up being desperate, and that Gabbard’s suicide attempters who turn to psychoanalysis, ‘although without any hope’, are those more accustomed to their illness due to its long duration, a good insight and a relatively high level of education.

Our study has an important limitation in the relatively small number of cases, so our results should be considered preliminary and need confirmation in larger samples. A follow-up study which examines the predictive validity of our assessment is needed. Another limitation is that the patients we selected are not representative of all suicide attempters found in psychiatric settings. Therefore, the assessment we propose should be used in patients with other disorders (such as schizophrenia). and also other settings where suicide attempts are very frequent (for instance, jail).


The present study confirms the usefulness of classifying suicide attempters according to the presence/absence of mood disorders, personality disorders and dysfunctional personality, negative life balance, as they are defined in the text. Negative life balance enabled us to identify a subgroup of suicide attempters, but the possible overlap between this concept and the concepts of pessimism, hopelessness and descent to suicide used by other authors deserves further investigation.

The use of multiple projective tests and of a clinical interview specifically focused on the pathway to suicide provided a deep and comprehensive personality structure assessment and allowed us to better characterize subjects with complex dysfunctional personality and subjects with mood disorders and dysfunctional personality. This method has potential clinical implications in the selection of prevention and treatment strategies.


American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th edition). Washington DC, American Psychiatric Press.

Apter A, Plutchick R, & Van Praag HM (1993). Anxiety, impulsivity and depressed mood in relation to suicidal and violent behavior. Acta Psychiatrica Scandinavica 87, 1-5.

Barber ME, Marzuk PM, Leon AC, & Portera L. (1998). Aborted suicide attempts: a new classification of suicidal behavior. American Journal of Psychiatry 155, 385-389.

Beck AT, & Lester D (1976). Components of suicidal intent in completed and attempted suicide. Journal of Psychology 92, 35-385.

Beck AT, Steer RA, Kovacs M, & Garrison B. (1985) Hopelessness and eventual suicide. American Journal of Psychiatry 142, 559-563

Berti Ceroni F, Gherardi S, & Rucci P. (1996). DSM-III mental disorders in general medical sector: Predictors of outcome. International Journal of Social Psychiatry 42, 45-57.

Bertolote GM, & Fleischmann A (2002). Suicide and psychiatric diagnosis. World Psychiatry 1, 181-185.

Bongar B, Peterson LG, Golann S, & Hardimann JJ (1990). Self-mutilation and the cronically suicide patient. Annals of Clinical Psychiatry 2, 217-222.

Brown G, &Harris T (1978). Social Origins of Depression. London, Tavistock.

Gabbard GO (2003). Miscarriage of psychoanalytic treatment. International Journal of Psycho-Analysis 84, 249-261.

Goldberg D, & Huxley P (1980). Mental Ilnness in the Community. London, Tavistock.

Goodwin FK, & Jamison KR (1990). Manic-Depressive Illness. New York, Oxford University Press.

Hamilton M (1960). A rating scale for depression. Journal of Neurology Neurosurgery and Psychiatry 23, 56-62.

Hendin H, Maltsberger JT, Lipschitz A, Haas A, & Kyle J (2001). Recognizing and responding to a suicide crisis. Suicide and Life-Threatening Behavior 31, 115-127.

Herrera MT, Orcoyen D, & Passalacqua AM (1995). Suicidio: final o principio? Asociatiòn latino-americana de Rorschach, 3, 55-64.

Jamison KR (1999). Rapida scende la notte. Capire il suicidio. Italian translation, Milano, Longanesi 2001.

Klopfer B, & Davidson HH (1962). La tecnica Rorshach. Un manuale introduttivo. Italian translation, Firenze, Organizzazioni Speciali 1971.

Linehan M (1986). Suicidal people: one population or two? Annals of the New York Academy of Science 487, 16-33.

Malone KM, Haas GL, Sweeney JA, & Mann JJ (1995). Major depression and the risk of attempted suicide. Journal of Affective Disorders 34, 173-185.

Maltsberger JT (2002). The descent into suicide, paper presented at the Austen Riggs Center Conference Dancing with death, Stockbridge (MA).

Maltsberger JT (2004). The descent into suicide. International Journal of Psycho-Analysis 85, 653-667.

Maltsberger JT, Hendin H, Haas AP, & Lipschitz A (2003). Determination of precipitating events in the suicide of psychiatric patients. Suicide and Life-threatening Behavior 33, 111-119.

Mann JJ (2004). Searching for triggers of suicidal behavior. American Journal of Psychiatry 161, 395-397.

Mann JJ, Waternaux C, Haas GL, & Malone KM (1999). Toward a clinical model of suicidal behavior in psychiatric patients. American Journal of Psychiatry 156, 181-189.

Maris RW (1981). Pathways to suicide: a survey of self-destructive behavior. Baltimore,John Hopkins Press.

Mazzi F, Morosini P, de Girolamo G, & Guaraldi GP (2003). Intervista clinica strutturata per l’asse II del DSM-IV. Firenze, OS.

Oquendo MA, Galfalvy H, Russo S, Ellis SP, Grunebaum MF, Burke A, & Mann JJ. (2004). Prospective study of clinical predictors of suicidal acts after a major depressive episode in patients with major depressive disorder or bipolar disorder. American Journal of Psychiatry 161.1433-41.

Passalacqua AM (1995). La depresiòn en adolescentes. Evaluation de los signos de riesguo suicida. Rorschachiana, 17, 125-142.

Passalacqua AM (1999). The Rorschach test in the detection of suicidal potential. Rorschachiana, 23, 1-17.

Phillipson H (1955). Il test delle relazioni oggettuali. Italian translation, Firenze, Organizzazioni Speciali 1974.

Rorschach H (1921) Psicodiagnostica. Metodologia e risultati di un esperimento diagnostico basato sulla percezione. Italian translation, Roma, Kappa 1981.

Rudd MD, Ellis TE, Hasan Rajab N, & Wehrly T (2000). Personality types and suicidal behavior: an esplorative study. Suicide and Life-Threatening Behavior 30,199-212

Shneidman ES (1985). Definition of Suicide. New York, Wiley.

Soloff PH, Lis JA, Kelly T, Cornelius J, & Ulrich R (1994). Risk factors for suicidal behavior in borderline personality disorder. American Journal of Psychiatry 151, 1316-1323.

Soloff PH, Lynch KG, Kelly TM, Malone KM, & Mann JJ (2000). Characteristics of suicide attempts of patients with major depressive episode and borderline personality disorder. A comparative study. American Journal of Psychiatry 157, 601-608.

Stanley B., Gameroff MJ, Michalsen V., & Mann JJ (2001). Are suicide attempters who self- mutilate a unique population? American Journal of Psychiatry 158, 427-432.

WHO (1997). International Classification of Disease (9th Edition, Clinical Management). Geneva, WHO.

Table 1 - Demographic and clinical characteristics of the three subgroups