The association between aggression turned outward and aggression turned inward has been suggested for some time and was emphasized by Freud. 1 Several recent studies report a relationship between violence and suicide. In one study, suicide attempters had significantly higher scores for lifetime aggression than non-attempters. 2 Longitudinal investigations have demonstrated associations between suicide and self-reported aggression, 3 as well as between suicide and hostility. 4 These results have been most evident in patients with personality disorders, especially borderline personality disorder. 5
Other authors have seen suicide and violence as being parts of different types of pathology. Suicide is considered an internalizing problem, associated with symptoms such as anxiety and depression. It is contrasted with externalizing behaviors, which consist predominantly of violence and antisocial behaviors. 6,7 However, in a study of depressed outpatients, subjects who had made suicide attempts or had had suicidal ideation did not differ from those without suicidal attempts or ideation in hostility or aggression. 8
In order to better understand the occurrence of suicide and its relationship to violence, it is important to consider the clinical symptoms underlying suicide and violence. A relationship between violence and various positive psychotic symptoms has been repeatedly demonstrated in patients with major psychiatric disorders. 9,10 The relationship between psychotic symptoms and suicide, on the other hand, is not as clear and there have been conflicting results. For example, some studies have reported an association between psychotic symptoms and suicide in patients with schizophrenia, 11 while others have not. For example, Grunebaum et al. found no relationship between delusions and history of suicidal ideation or suicide attempts in patients with major mood disorders or schizophrenia. 12
In contrast to positive psychotic symptoms, depressive symptoms have been strongly associated with suicide attempts but not with violence directed at others. The relationship with suicide is found in various diagnostic groups, 13 including schizophrenia. 14,15 Depressive symptoms are the most common clinical correlates of suicidality in schizophrenia. 16
In this study, we investigated the relationship between suicide attempts and physical assaults in patients with schizophrenia, schizoaffective disorder, or bipolar disorder. We hypothesized that suicidal patients would be more violent than non-suicidal patients. We also wanted to compare the psychiatric symptoms associated with suicide with those associated with violence.
The subjects were 18- to 55-year-old inpatients at two state psychiatric hospitals who had a DSM-III-R 17 diagnosis of schizophrenia, schizoaffective disorder, or bipolar disorder. DSM-III-R psychiatric disorders were diagnosed by using the Structured Clinical Interview for DSM-III-R (SCID). 18 Patients who had had at least one physical assault directed at others (other patients or staff) in the first 2 months of hospitalization were included as violent patients. A non-violent comparison group was also included. The violent and non-violent patients were dichotomized on the basis of history of suicide attempts, which was determined through a review of all hospital charts and records, as well as through interviews, in which patients were asked if they had ever attempted suicide or been hospitalized for a suicide attempt. A suicide attempt was defined as a self-destructive act that was carried out with the intent to end one’s life and was sufficiently serious to require medical evaluation. Subjects were given a complete description of the study, after which written informed consent was obtained. The protocol was approved by the IRBs in both hospitals.
Study Design and Evaluations
Upon entry into the study, an initial evaluation of each subject’s psychiatric symptoms was done. Patients were then followed for 4 weeks. All physical and verbal assaults were recorded, and all administered medications were noted down. At the end of 4 weeks, psychiatric symptoms were again assessed. Demographic and historical information was obtained through chart reviews and patient interviews. This included information about psychiatric history, criminal history and community violence, substance use, and substance abuse history.
To establish interrater reliability, joint rating sessions that included all raters were conducted prior to the study; these sessions included all scales that were used as well as the SCID. The raters’ performance was tested in additional sessions; those who had an intraclass coefficient (ICC), 19 which is a measure of interrater reliability, of at least 0.8 participated in the study. ICCs are reported below for the individual scales.
Assessment of violence.
Physical assaults, defined as actual physical contacts (striking, kicking, pushing), were assessed on a prospective basis by two licensed research nurses through daily review of nursing reports, patients’ charts, and interviews with ward staff present at the time of the incidents. These incidents were recorded using the Modified Overt Aggression Scale (MOAS). 20 which is based on the Overt Aggression Scale. 21 Verbal assaults were also recorded. Raters were trained prior to the study on the use of the MOAS. ICCs for interrater reliability, obtained prior to the study with three raters, were 0.94 for physical assaults and 0.89 for verbal assaults.
The Brief Psychiatric Rating Scale (BPRS) 22 was used to evaluate psychiatric symptoms. The BPRS consists of five basic factors: Anxiety-Depression, Anergia, Thought Disturbance, Activation, and Hostile-Suspiciousness. 23 The last three factors measure positive psychotic symptoms. The BPRS was administered by research doctors who were blind to both violent and suicidal behaviors. Because they were blind to violence, they were instructed not to inquire about actual violence; the BPRS hostility item (which is part of the BPRS Hostile-Suspiciousness Factor), therefore, was limited to an assessment of hostile attitude. ICCs for interrater reliability were obtained prior to the study and ranged from 0.87 to 0.98 for BPRS total score.
Patients who were present throughout the 4-week observation period were included. The endpoint BPRS was used in the analysis because patients were more psychiatrically stable at that point in time, and we were interested in more enduring symptoms. Differences between the groups on the BPRS factors were tested through analysis of covariance (ANCOVA). The five BPRS factors served as the dependent variables. In addition to the factors, we also included the BPRS Hostility item (which is part of the Hostile-Suspiciousness Factor) to further examine the relationship between suicide attempts and hostility.
The independent variables included the classification as suicide attempters versus non-attempters, the classification as violent toward others versus non-violent toward others (each classification serving as a covariate), and the interaction between these two classifications (i.e., further subdivision into violent patients who are suicide attempters, violent patients who are non-attempters, non-violent patients who are suicide attempters, and non-violent patients who are non-attempters). Gender, diagnosis, ethnicity, and age were entered as covariates in the analyses.
Overall Characteristics of the Suicidal and Non-Suicidal Groups
There were 253 violent patients with a DSM-III-R diagnosis of schizophrenia, schizoaffective disorder, or bipolar disorder. Of these, 15 refused to participate in the study and 238 agreed. We tried to match every other patient to a non-violent patient on diagnosis, ethnic background, gender, and age (within 5 years), but in some cases we did not find matching controls. The actual ratio of violent to non-violent patients was approximately 2.5. The study group consisted of 238 violent patients and 93 non-violent controls. Twenty-two (9.2%) of the violent patients and 12 (12.9%) of the non-violent controls did not complete the 4 weeks of inpatient treatment after study entry. The final sample which provided the information for this study therefore included 297 patients, 216 violent patients and 81 non-violent controls. Of the 297 patients, 103 had a history of suicide attempts (“attempters”) and 194 did not (“non-attempters”). Of the attempters, 47 (45.6%) had made two or more suicide attempts. Table 1 shows the demographic and diagnostic characteristics of the patients organized on the basis of both suicide attempts and violent behavior.
The attempters differed significantly from the non-attempters in gender, ethnicity, diagnosis, and age. They did not differ from the non-attempters in duration of illness. Of the 94 women in the sample, 52.1% had attempted suicide compared with 26.6% of the 203 men; of the 80 Caucasians, 46.2% had attempted suicide compared with 26.7% of the 157 African Americans; 45.8% of the 48 patients with bipolar disorder and 43.5% of the 92 patients with schizoaffective disorder had attempted suicide compared with 26.1% of the 157 patients with schizophrenia. The attempters were older than the non-attempters. The violent patients differed from the non-violent comparison group in age but not on the other characteristics on which they were matched. Logistic regression was used with suicide classification as the dependent variable and gender, ethnicity, diagnosis, and age as the independent variables. When these variables were considered together, the only difference between attempters and non-attempters that remained significant was in gender (Wald’s 2 = 11.2, df = 1, p < 0.001): a larger percentage of women than men had attempted suicide.
There were no significant differences between attempters and non-attempters and between violent and non-violent patients in the percentage of patients with a history of substance use (76.7% of the attempters versus 73.0% of the non-attempters; 77.4% of the violent patients versus 66.7% of non-violent patients).
Violent Behaviors in the Suicidal and Non-Suicidal Groups.
There was no difference between the attempter and non-attempter groups in the percentage of patients who were physically assaultive in the hospital; 73.8% (76 out of 103) of the attempters versus 72.2% (140 out of 194) of the non-attempters were physically assaultive (Wald’s 2 = 0.41, df = 1, p = 0.52, with age, gender, diagnosis and ethnicity as covariates) (odds ratio [OR] = 1.21; confidence intervals [CI] = 0.68–2.16). There were no differences in the total number of physical assaults (Wald’s 2 = 0.74, df = 1, p = 0.39) (OR = 1.02; CI = 0.98–1.05) or in the severity of these assaults during the 4-week period. There were no differences in the number of verbal assaults.
History of physical assaults in the community was evaluated as a discrete variable; 33.3% (34 out of 102) of the attempters versus 27.8% of the non-attempters (53 out of 191) had a history of community violence (data missing on 4 patients). Logistic regression (with diagnosis, age, gender, and ethnicity as covariates) indicated that there was no relationship between community violence and suicide (Wald’s 2 = 1.38, df = 1, p = 0.24; OR = 1.40; CI = 0.80–2.45).
The attempters were slightly more hostile, as measured by the BPRS Hostility item (mean = 2.81, SD = 1.7 versus mean = 2.55, SD = 1.7), but this difference was not significant (ANCOVA; F = 1.91; df = 1,291, p = 0.17).
Psychiatric Symptoms Associated with Suicide Attempts and with Violence
The independent variables in the ANCOVA included classification as suicide attempters and non-attempters, classification as violent and non-violent, and the interaction between these two classifications (i.e., violent patients who are suicide attempters, violent patients who are non-attempters, non-violent patients who are suicide attempters, and non-violent patients who are non-attempters). None of the interaction terms was significant, indicating that the presence of these two outcomes together (suicide attempts and violent behavior) was not associated with increased severity or a different pattern of symptoms. Therefore, in the following discussion, we report comparisons between attempters and non-attempters and between violent and non-violent patients.
Comparisons between suicide attempters and non-attempters on the BPRS factors indicated that the attempters had significantly higher scores on the Anxiety/Depression Factor (mean = 9.13, SD = 3.1, versus mean = 7.59, SD = 2.6) (ANCOVA; F = 11.68, df = 1,291, p < 0.001). These results remained significant when we limited the analyses to patients who had attempted suicide 1 year or more prior to the present evaluation. Furthermore, analyses of the individual items revealed significant differences on each of the four items that make up the BPRS Anxiety/Depression Factor, including depressive mood (F = 7.72, df = 1,291, p < 0.01), anxiety (F = 5.14, df = 1,291; p = 0.02), somatic concern (F = 4.64, df = 1,291; p = 0.03) and guilt (F = 17.2, df = 1,291, p < 0.001).
The attempters had significantly higher scores on the Activation Factor than the non-attempters (mean = 6.64, SD = 2.8, versus mean = 6.01, SD = 2.7) (F = 3.89, df = 1,291, p < 0.05), but the only item on that factor that reached statistical significance was tension (F = 5.33, df = 1,291, p = 0.02). There were no significant differences between the attempters and non-attempters on the other three BPRS Factors.
Comparisons between the violent and non-violent patients revealed significant differences on all three BPRS factors that measure positive psychotic symptoms. The violent patients had significantly higher scores on Thought Disturbance (mean = 10.41, SD = 4.8, versus mean = 8.75, SD = 4.0) (F = 14.16, df = 1,291, p < 0.001), on Activation (mean = 6.52, SD = 2.9, versus mean = 5.44, SD = 2.3) (F = 7.39, df = 1,291, p < 0.01) and on Hostile-Suspiciousness (mean = 8.33, SD = 3.7, versus mean = 6.10, SD = 2.8) (F = 24.42, df = 1,291, p < 0.001). There were no differences between violent and non-violent patients on the BPRS Anxiety-Depression or Anergia Factors.
Association Between Suicide Attempts and Violence
The first goal of this study was to determine whether attempters and non-attempters differed on various indices of violence against others. The attempters did not differ in physical inpatient assaults (including presence or absence of assaults and frequency or severity of assaults) or verbal inpatient assaults. They did not differ in physical violence in the community. Attempters had a somewhat higher score on the BPRS hostility item, but this did not reach statistical significance. These findings are at variance with some findings in the literature that have indicated an association between the two problems. 2–4 However, they agree with the results of the study of depressed outpatients discussed above 8 as well as with the results of studies that have classified suicide as an internalizing problem and contrasted it with violence, which was categorized as an externalizing behavior. 6,7 The discrepancy with studies reporting an association between violence and suicide attempts may be due in part to differences in diagnoses. The patients in our study were diagnosed with major psychiatric disorders, with the majority having a diagnosis of schizophrenia or schizoaffective disorder, whereas many of the studies that have reported an association between suicide and violence were conducted in subjects with personality disorders, especially borderline personality disorder. 5 For example, Mann et al. 2 noted that a diagnosis of borderline personality disorder and aggression were both robust predictors of attempter status but that the two were strongly interdependent and could not be separated from each other. Similarly, in another study, suicide and violence were specifically related to each other and to serotonergic dysfunction in patients with personality disorder, but not in patients with major depressive disorder. 24
Another factor that may explain the discrepancy in findings is the way in which aggression is reported. The association between suicide and violence may be higher when based on self-reported aggression, since some patients are more likely to present themselves as having multiple problems. Several of the studies relied on patients’ self-reports of violence, whereas in our study violent behavior was assessed on a prospective basis through observation.
Symptoms Associated with Suicide Attempts and Violence
The second goal of this study was to identify symptoms associated with suicide attempts and with violence against others. The fact that suicidal behavior and violence are a function of different psychiatric symptoms explains, in part, the lack of relationship between these two phenomena. Attempters had significantly higher scores than non-attempters on the BPRS Anxiety-Depression Factor, but the assaultive patients did not differ from the non-assaultive ones on these symptoms.
Positive symptoms, on the other hand, were strongly associated with violence, but except for a single symptom, were not related to suicide attempts. In line with the literature discussed above, 9,10 the physically assaultive patients presented with more severe impairments than the non-assaultive patients on all three measures of positive symptoms, including Thought Disturbance, Activation, and Hostile-Suspiciousness. The suicide attempters had higher scores on the Activation Factor compared with the non-attempters, but this was due mostly to increased tension. While this item is considered part of the Activation Factor, it may represent the motor component of the increased anxiety present in these patients.
Our findings concerning symptoms associated with suicide attempts agree closely with findings from a study in which follow-up data were obtained on 1,357 patients with schizophrenia. 25 When patients who committed suicide were compared with those who did not, suicide was associated with depressive symptoms and psychomotor agitation, but was not related to delusions or hallucinations.
The strong association between suicide attempts and the BPRS Anxiety-Depression Factor seen in our study is all the more striking given the extensive nature and temporal stability of these findings. Each of the four symptoms that make up the BPRS Anxiety-Depression Factor (i.e., depressive mood, anxiety, guilt, and somatic concern) was significantly more severe in the suicide attempters than in the non-attempters.
Strengths and Limitations of the Study
The strengths of this study include the fact that the raters were blind to the violence and suicide status of the patients and that the data on assaults were gathered on a prospective basis and did not rely on self-reports. The limitations of this study include the retrospective investigation of the suicidal behavior.
Depressive mood, anxiety, guilt, and somatic concern were significantly more severe in patients with a history of suicide attempts, although for the vast majority of these patients, the attempts had occurred many years prior to the current evaluation. These symptoms may be chronic or recur under stress or with psychotic decompensation. They may represent a depressive diathesis or a chronic dysregulation of affect. It is very important for the clinician to inquire carefully about suicide attempts in the past, even when there is no active suicidal ideation at present, because the depression or mood dysregulation may persist or recur and must be an important focus in the overall treatment of these patients.
1. Freud, S. Mourning and melancholia. In: Strachey J, Freud A, Rothgeb CL, et al., eds. The standard edition of the complete psychological works of Sigmund Freud. Vol 14. London: Hogarth Press; 1953:152–72.
2. Mann JJ, Waternaux C, Haas GL, et al. Toward a clinical model of suicidal behavior in psychiatric patients. Am J Psychiatry 1999;156:181–9.
3. Angst J, Clayton P. Premorbid personality of depressive, bipolar, and schizophrenic patients with special reference to suicidal issues. Compr Psychiatry 1986;27:511–31.
4. Romanov K, Hatakka M, Keskinen E. Self-reported hostility and suicidal acts, accidents, and accidental deaths: A prospective study of 21,443 adults aged 25–59. Psychosom Medicine 1994;56:328–36.
5. Brodsky BS, Malone KM, Ellis SP. Characteristics of borderline personality disorder associated with suicidal behavior. Am J Psychiatry 1997;154:1715–9.
6. Campbell SB. Behavior problems in preschool children: A review of recent research. J Child Psychol Psychiatry 1995;36:113–49.
7. Cicchetti D, Toth SL. A developmental perspective on internalizing and externalizing disorders. In: Cicchetti D, Toth SL, eds. Rochester Symposium on Developmental Psychopathology, Vol 16: Internalizing and externalizing expressions of dysfunction. Hillsdale, NJ: Erlbaum; 1991:1–19.
8. Castrogiovanni P, Pieraccini F, Di Muro A. Suicidality and aggressive behaviour. Acta Psychiatr Scand 1998;97:144–8.
9. McNiel DE, Binder RL. The relationship between acute psychiatric symptoms, diagnosis, and short-term risk of violence. Hosp Com Psychiatry 1994;45:133–7.
10. Tardiff K, Sweillam A. Assault, suicide and mental illness. Arch Gen Psychiatry 1980;37:164–9.
11. Kaplan KJ, Harrow M. Positive and negative symptoms as risk factors for later suicidal activity in schizophrenics versus depressives. Suicide Life Threat Behav 1996;26:105–21.
12. Grunebaum MF, Oquendo MA, Harkavy-Friedman JM. Delusions and suicidality. Am J Psychiatry 2001;158:742–7.
13. Lewinsohn PM, Rohde P, Seeley JR. Psychological risk factors for future adolescent suicide attempts. J Consult Clin Psychol 1994;62:297–305.
14. Harkavy-Friedman JM, Restifo K, Malaspina D. Suicidal behavior in schizophrenia: Characteristics of individuals who had and had not attempted suicide. Am J Psychiatry 1999;156:1276–8.
15. Schwartz RC, Cohen BN. Psychosocial correlates of suicidal intent among patients with schizophrenia. Compr Psychiatry 2001;42:118–23.
16. Siris SG. Suicide and schizophrenia. J Psychopharm 2001;15:127–35.
17. American Psychiatric Association. Diagnostic and statistical manual of mental disorder, 3rd edition, revised. Washington, DC: American Psychiatric Association; 1987.
18. Spitzer RL, Williams JBW, Gibbon M. Structured Clinical Interview for DSM-III-R Patient Version (SCID-P). New York: New York State Psychiatric Institute, Biometrics Research; 1990.
19. Bartko, JJ, Carpenter WT. On the methods and theory of reliability. J Nerv Ment Dis 1976;163:307–17.
20. Kay S, Wolkenfeld F, Murrill L. Profiles of aggression among psychiatric patients, I. Nature and prevalence. J Nerv Ment Dis 1988;176:539–46.
21. Yudofsky SC, Silver JM, Jackson W. The overt aggression scale for the objective rating of verbal and physical aggression. Am J Psychiatry 1986;143:35–9.
22. Overall JE, Gorham DR. The Brief Psychiatric Rating Scale. Psychol Reports 1962;10:799–812.
23. Guy W. ECDEU assessment manual for psychopharmacology, revised (DHEW Publication No. ADM 76-338). Rockville, MD: U.S. Department of Health and Human Services; 1976.
24. Coccaro EF, Siever LJ, Klar HM. Serotonergic studies in patients with affective and personality disorders: Correlates with suicidal and impulsive aggressive behavior. Arch Gen Psychiatry 1989;46:587–99.
25. Stephens JH, Richard P, McHugh PR. Suicide in patients hospitalized for schizophrenia: 1913–1940. J Nerv Ment Dis 1999;187:10–4.