course of violence in patients with schizophrenia: relationship to

14
Course of Violence in Patients With Schizophrenia: Relationship to Clinical Symptoms by Menahem Krakowski, Pal Czobor, and James C.-Y. Chou Abstract To understand the heterogeneity of violent behaviors in patients with schizophrenia, one must consider underlying clinical symptoms of the illness and their change over time. The purpose of this study was to examine persistence and resolution of violence in rela- tion to psychotic symptoms, ward behaviors, and neu- rological impairment. Psychiatric symptoms and ward behaviors were assessed in violent inpatients with schizophrenia or schizoaffective disorder and in nonvi- olent controls on entry into the study. Patients were followed for 4 weeks; those who showed resolution of assaults over this time were classified as transiently violent, and those who remained assaultive were cate- gorized as persistently violent. At the end of the 4 weeks, psychiatric symptoms, ward behaviors, and neurological impairment were assessed. Overall, the two violent groups presented with more severe psychi- atric symptoms and were judged to be more irritable than the nonviolent control subjects, but the tran- siently violent patients showed improvement in symp- toms over time. At the end of 4 weeks, the persistently violent patients had evidence of more severe neurologi- cal impairment, hostility, suspiciousness, and irritabil- ity than the other two groups. Canonical discriminant analyses identified two significant dimensions differen- tiated the groups. The first, characterized by positive psychotic symptoms, differentiated the violent patients from the controLsubjects;_the second,xharacterized by neurological impairment and high endpoint score for negative symptoms, differentiated the transiently from the persistently violent patients. Identification of cer- tain symptoms associated with different forms of vio- lence has important implications for the prediction and differential treatment of violent behavior in patients with schizophrenia. Key words: Schizophrenia, violence, psychiatric symptoms, neurological impairments. Schizophrenia Bulletin, 25(3):505-517,1999. Violence has been reported to be more frequent in patients with serious and persistent mental illnesses than in the rest of the population. Using data from the National Institute of Mental Health's Epidemiologic Catchment Area surveys, Swanson and colleagues (1990) showed that the occurrence of major mental illness was associated with increased risk of violent behavior in the preceding year. Active psychiatric symptoms, not merely a history of mental illness, are implicated in this association of mental illness with assaultive behavior (Swanson 1994). There- fore, an understanding of the occurrence of violence in this population requires careful consideration of these psychiatric symptoms. A positive relationship between violence and various psychotic symptoms, such as delusions, hallucinations, and thought disorder, has been repeatedly demonstrated in patients with schizophrenia and in those with other psy- chiatric conditions (Tardiff and Sweillam 1980; Yesavage et al. 1981; Yesavage 1983; McNiel and Binder 1994). However, these symptoms are not invariant; they change over time as a function of the underlying illness. Thus, to better understand their relationship to violence, we must consider when in the course of mental illness the violent behavior occurs, as well as the progression or resolution of the underlying symptoms (Krakowski et al. 1986). The timing of the violent acts within the course of the illness is often not reported; however, many of these previous studies were conducted around the time of the hospital admission, which often represents a phase of acute decompensation. Violence that ^precedes admission is related to violence that occurs during early hospitalization (Beck and Bonnar 1988; McNiel et al. 1988), and both are related to positive psychotic symptoms (McNiel et al. 1988; Beck et al. 1991). Patients who became assaultive early in their hospital stay had more positive psychotic symptoms at the time of admission (Lowenstein et al. Reprint requests should be sent to Dr. Menahem Krakowski, Nathan Kline Institute for Psychiatric Research, Orangeburg, NY 10962; fax (914) 398-6545. 505 Downloaded from https://academic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Upload: others

Post on 25-Mar-2022

3 views

Category:

Documents


0 download

TRANSCRIPT

Course of Violence in Patients With Schizophrenia:Relationship to Clinical Symptoms

by Menahem Krakowski, Pal Czobor, and James C.-Y. Chou

Abstract

To understand the heterogeneity of violent behaviorsin patients with schizophrenia, one must considerunderlying clinical symptoms of the illness and theirchange over time. The purpose of this study was toexamine persistence and resolution of violence in rela-tion to psychotic symptoms, ward behaviors, and neu-rological impairment. Psychiatric symptoms and wardbehaviors were assessed in violent inpatients withschizophrenia or schizoaffective disorder and in nonvi-olent controls on entry into the study. Patients werefollowed for 4 weeks; those who showed resolution ofassaults over this time were classified as transientlyviolent, and those who remained assaultive were cate-gorized as persistently violent. At the end of the 4weeks, psychiatric symptoms, ward behaviors, andneurological impairment were assessed. Overall, thetwo violent groups presented with more severe psychi-atric symptoms and were judged to be more irritablethan the nonviolent control subjects, but the tran-siently violent patients showed improvement in symp-toms over time. At the end of 4 weeks, the persistentlyviolent patients had evidence of more severe neurologi-cal impairment, hostility, suspiciousness, and irritabil-ity than the other two groups. Canonical discriminantanalyses identified two significant dimensions differen-tiated the groups. The first, characterized by positivepsychotic symptoms, differentiated the violent patientsfrom the controLsubjects;_the second,xharacterized byneurological impairment and high endpoint score fornegative symptoms, differentiated the transiently fromthe persistently violent patients. Identification of cer-tain symptoms associated with different forms of vio-lence has important implications for the predictionand differential treatment of violent behavior inpatients with schizophrenia.

Key words: Schizophrenia, violence, psychiatricsymptoms, neurological impairments.

Schizophrenia Bulletin, 25(3):505-517,1999.

Violence has been reported to be more frequent in patientswith serious and persistent mental illnesses than in therest of the population. Using data from the NationalInstitute of Mental Health's Epidemiologic CatchmentArea surveys, Swanson and colleagues (1990) showedthat the occurrence of major mental illness was associatedwith increased risk of violent behavior in the precedingyear. Active psychiatric symptoms, not merely a history ofmental illness, are implicated in this association of mentalillness with assaultive behavior (Swanson 1994). There-fore, an understanding of the occurrence of violence inthis population requires careful consideration of thesepsychiatric symptoms.

A positive relationship between violence and variouspsychotic symptoms, such as delusions, hallucinations,and thought disorder, has been repeatedly demonstrated inpatients with schizophrenia and in those with other psy-chiatric conditions (Tardiff and Sweillam 1980; Yesavageet al. 1981; Yesavage 1983; McNiel and Binder 1994).However, these symptoms are not invariant; they changeover time as a function of the underlying illness. Thus, tobetter understand their relationship to violence, we mustconsider when in the course of mental illness the violentbehavior occurs, as well as the progression or resolutionof the underlying symptoms (Krakowski et al. 1986). Thetiming of the violent acts within the course of the illnessis often not reported; however, many of these previousstudies were conducted around the time of the hospitaladmission, which often represents a phase of acutedecompensation. Violence that ^precedes admission isrelated to violence that occurs during early hospitalization(Beck and Bonnar 1988; McNiel et al. 1988), and both arerelated to positive psychotic symptoms (McNiel et al.1988; Beck et al. 1991). Patients who became assaultiveearly in their hospital stay had more positive psychoticsymptoms at the time of admission (Lowenstein et al.

Reprint requests should be sent to Dr. Menahem Krakowski, NathanKline Institute for Psychiatric Research, Orangeburg, NY 10962; fax(914) 398-6545.

505

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Schizophrenia Bulletin, Vol. 25, No. 3, 1999 M. Krakowski et al.

1990). Florid psychotic symptoms predominate duringearly hospitalization and can easily escalate intoassaultive behavior (Lion et al. 1981). In one study ofmanic patients (Carlson and Goodwin 1973), psychophar-macological treatment was postponed so that investigatorscould observe the natural course of the illness. The initialsymptoms, which included irritability and agitation, even-tually progressed to open hostility and anger, and theaccompanying behavior was frequently assaultive andexplosive. Violence presumably remits as these acutesymptoms improve.

When studies are not limited to the period of acutedecompensation, violence appears to be associated with amore restricted set of symptoms, including hostility, sus-piciousness, and more specific and organized delusions,often focused on specific people and having persecutorythemes (Junginger 1996). One study (Virkkunen 1974)differentiated between two groups of violent schizophre-nia patients. In one group, violent acts occurred duringpsychotic episodes; were accompanied by hallucinations,delusions, and other positive symptoms; and were seldompreceded by a hostile attitude toward the victim. In theother group, the assaults were not restricted to psychoticepisodes; hostility and resentment toward the victim hadexisted for a long time and eventually culminated in actsof violence. In another study (Nestor et al. 1995), severelyviolent psychotic patients were more likely to havechronic and organized delusional beliefs involving per-sonal targets, compared with less violent psychoticpatients. In a study that followed psychiatric patients inthe community (Link and Stueve 1994), three specificsymptoms—namely, feeling that others wished one harm,believing that one's mind was dominated by forcesbeyond one's control, and believing that others' thoughtswere being put into one's head—were strongly associatedwith violence. Violence is thus more likely when psy-chotic symptoms cause the patient to feel personallythreatened or when they involve the intrusion of thoughtsthat can override self-controls. These findings were basedon patients' retrospective self-reports of symptoms over a1-year period; there is therefore no information to showwhen these symptoms occurred in the course of thedisorder.

Other important clinical symptoms associated withviolence include neurological abnormalities. Such abnor-malities have been reported in various violent populations(Wolfgang 1975; Monroe et al. 1977), especially withmore persistent or recidivistic forms of violence (Yeudallet al. 1982; Krieger 1985; Volkow and Tancredi 1987). Inpatients with schizophrenia, violence is also related tovarious measures of neurological dysfunction, includingelectroencephalographic abnormalities (Heath 1982) andneuropsychological impairment (Adams et al. 1990). In a

previous study that was conducted on a special unit forthe management of violent behavior (Krakowski et al.1989), persistently violent patients with schizophreniapresented with more neurological impairment than nonvi-olent and transiently violent patients. No distinct patternof neurological dysfunction was found, but there weremore abnormalities in higher motor and sensory integra-tion. In a study of schizophrenia outpatients (Lapierre etal. 1995), no relationship was found between neurologicalimpairment and lifelong history of physical violence.

In general, patients with schizophrenia have signifi-cantly more hard and soft neurological signs than nonpsy-chiatric control subjects or patients with mixed psychi-atric disorders and affective disorders (Heinrichs andBuchanan 1988; Woods et al. 1991; Rubin et al. 1994;Ismail et al. 1998). These neurological signs appear to beindependent of medication effects because they are foundin schizophrenia patients who have never received neu-roleptics (Gupta et al. 1995). Neurological impairment inpatients with schizophrenia has been associated with cer-tain symptoms or symptom profiles. There is some evi-dence that neurological impairment is more prominent inpatients with negative symptoms (Merriam et al. 1990;Wong et al. 1997), especially enduring negative symp-toms (Buchanan et al. 1990). Neurological abnormalitiesare also correlated with thought disorder (Manschrek etal. 1981) and with poor social functioning (Wong et al.1997). In patients with schizophrenia, these abnormalitieshave also been shown to be associated with other charac-teristics of the illness, such as treatment resistance(Wagman et al. 1987).

In this study, we investigated the course of violenceover time in relation to psychiatric symptoms, wardbehaviors, and neurological symptoms. We hypothesizedthat persistent violence would be associated with morepersistent psychotic symptoms and more severe neurolog-ical symptoms and that transient violence would be moreoften accompanied by florid psychotic symptoms thatresolve over time.

Methods

Subjects. The subjects were newly admitted patients onall admission wards in two State psychiatric hospitals,Manhattan Psychiatric Center (New York, NY) andRocklahd Psychiatric Center (Orangeburg, NY). Patientswere 18 to 55 years old and were diagnosed with schizo-phrenia or schizoaffective disorder as assessed by theStructured Clinical Interview for DSM-III-R (SCID;Spitzer et al. 1987) by research doctors. They had no sig-nificant systemic (e.g., pulmonary or cardiovascular) dis-ease. These patients were observed for physical assault.Patients who were assaultive within 2 months of admis-

506

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Violence in Schizophrenia and Clinical Symptoms Schizophrenia Bulletin, Vol. 25, No. 3, 1999

sion were evaluated and observed for another 4 weeks; ifan additional physical assault occurred during that period,they were included in the final sample. The sample ulti-mately included 96 violent patients. In addition, 81 non-violent patients were included for comparison; they werematched to randomly selected violent patients on the basisof age, ethnic background, and sex.

Study Design and Evaluations. Informed consent wasobtained from the violent patients and nonviolent controlpatients. An initial evaluation of psychiatric symptomsand ward behaviors was then performed (within 3 days ofthe first physical assault for the violent patients). Thepatients were then followed for 4 weeks; all physical andverbal assaults were recorded, and all administered med-ications were monitored. At the end of the 4 weeks, allpatients received a comprehensive examination, whichincluded assessment of psychiatric symptoms and wardbehaviors, as in the initial evaluation, and assessment ofneurological symptoms. Demographic and historical data,including psychiatric history, drug and alcohol abuse, andhead trauma, were obtained through patient interviewsand a review of medical records.

Joint rating sessions that included all raters were con-ducted before the study to establish interrater reliability.Raters' performance was tested through additional ses-sions; those who had an intraclass correlation coefficient(ICC; Bartko and Carpenter 1976)—a measure of inter-rater reliability—of at least 0.8 participated in the study.ICCs for the individual scales are given below.

MeasuresInpatient assaults. For the purpose of this study,

assault was defined as actual physical contact (striking,kicking, pushing, or scratching). Physical assaults wereassessed by two licensed research nurses through dailyreview of nursing reports (ward journals), patients' charts,and interviews with ward staff present at the time of theincidents. The assaults were then recorded using theModified Overt Aggression Scale (MOAS; Kay et al.1988), which is based on the Overt Aggression Scale(OAS; Yudofsky et al. 1986). The patient had to be anactive assailant to be included in the study, but the assaultcould have been initiated by the patient or could haveoccurred in response to some provocation, such as aninsult.

In addition to physical assault, which was the primaryvariable of interest, verbal assaults—curses, insults, andthreats of violence—were also recorded. This additionalrecord was intended to further describe the patients' behav-iors. Raters were trained before the study. ICCs for inter-rater reliability, obtained before the study with three raters,were 0.94 for physical assaults and 0.89 for verbal assaults.

Psychiatric symptoms. The Brief PsychiatricRating Scale (BPRS; Overall and Gorham 1962) was usedto evaluate psychiatric symptoms. It provides an assess-ment of five major symptom dimensions, or factors: anxi-ety-depression, anergia, thought disturbance, activation,and hostility-suspiciousness. The last three factorstogether represent a measure of positive psychotic symp-toms. Anergia, which is highly correlated with overallmeasures of negative symptoms (Czobor et al. 1991), wasused as a measure of negative symptoms.

The BPRS was administered by research doctors whodid not know whether the patient was persistently or tran-siently violent or whether the patient was a nonviolentcontrol subject or a violent patient. The raters wereinstructed not to inquire about actual violence; therefore,the hostility item was limited to an assessment of hostileattitude. Because detailed information about incidents wascollected separately, these assessments of hostility pro-vided independent information. ICCs for interrater relia-bility were obtained before the study; they rangedbetween 0.87 and 0.98 for BPRS total score.

Ward behaviors. The irritability, routine, andsocial subscales from the Nurses' Observation Scale forInpatient Evaluation (NOSIE; Honigfeld and Klett 1965)were administered. These three subscales assess behaviorsrelevant to the occurrence of assaults and do not overlapwith the information obtained through the BPRS. The irri-tability subscale contains such items as "is impatient,""gets angry or annoyed easily," "becomes easily upset ifsomething does not suit him." The routine subscaleassesses compliance with ward routine and regulationsand includes such items as "refuses to do ordinary thingsexpected of him," "has to be told to follow hospital rou-tine," and "has difficulty completing even simple tasks onhis own." The social subscale assesses social interest; itincludes such items as "tries to be friendly with others,""shows interest in activities around him," and "starts upconversation with others." These subscales are scored sothat higher values indicate less pathology. The NOSIEmean score reported in this study is the mean score ofthese three subscales.

The licensed research nurses interviewed the wardstaff. The ICCs for interrater reliability were obtainedbefore the study and were higher than 0.80 for all threesubscales. Because these ratings were based on patients'ward behaviors, the raters were not blind to the violence.

Neurological abnormalities. The QuantifiedNeurological Scale (QNS; Convit et al. 1994) was admin-istered to assess neurological symptoms. The QNS is acomprehensive quantitative assessment of neurologicalfunction that includes assessment of motor sequencing,motor integrative activity, cranial nerves, and cortical andsoft signs, as well as the cancellation test.

507

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Schizophrenia Bulletin, Vol. 25, No. 3, 1999 M. Krakowski et al.

The QNS was administered by the same researchdoctors who administered the BPRS; they were blind tothe patients' violence. ICCs for interrater reliability forindividual items was obtained before the study and rangedfrom 0.69 to 1.00; 87 (90%) of the 96 items had ICCs of0.75 or greater.

Course of violence and group formation. In ourstudy, 96 patients committed two or more physicalassaults during the 4-week period. In patients with two ormore assaults, subsequent assaults often occurred soonafter the first one; specifically, 50 percent of subsequentassaults occurred within 10 days after the first assault. Theresolution of this early violence varied among patients. Insome, it decreased rapidly after the initial period of 10days; in others, it persisted over the four weeks. To furthercharacterize this heterogeneous course, we classified theviolent patients on the basis of change in violent behavior,i.e., whether there was a greater than 50 percent decreasein the number of physical assaults compared with thebaseline measures obtained during the 10-day period afterthe first assault. Two groups of patients were thus identi-fied: the "persistently violent" patients (n = 44) showedno substantial decrease (i.e., showed a < 50% decrease) inviolence during the second period (18 days) comparedwith the initial 10 days; the "transiently violent" patients(n = 52) showed a marked (> 50%) decrease in violenceduring the second period.

Data Analysis. The purpose of this study was to under-stand symptoms associated with different courses of vio-lence in schizophrenia patients, including positive andnegative psychotic symptoms, ward behaviors, and neuro-logical symptoms. Analysis of variance (ANOVA) wasused to compare the groups on the QNS. For the BPRSand NOSEE, comparisons among the groups were done attwo different time points: on entry into the study (baselineevaluation) and after 4 weeks (endpoint evaluation).Repeated-measures ANOVA was used to compare thegroups on the main variables, which included the BPRSfactors and the NOSIE subscales. The main effect ofgroup (persistently violent, transiently violent, or nonvio-lent) and the interaction effect between group and timeconstituted the principal interest of these analyses. Wecompared subscales or factors only when the groups dif-fered in global scores.

Significant group and interaction effects were furtherinvestigated in post hoc tests in two ways. We comparedeach pair at both baseline and endpoint and usedBonferroni corrections to adjust significance for multiplecomparisons. In addition, to examine overall improve-ment, we investigated changes in BPRS or NOSIE scoresover time in each of the three groups.

Canonical discriminant analysis was performed todetermine how psychotic and neurological symptoms,considered together, distinguished among the threegroups. The clinical variables consisted of the BPRS base-line and endpoint measures of positive (thought disorder,activation, and hostility-suspiciousness factors) and nega-tive (anergia factor) psychotic symptoms, as well as neu-rological symptoms (total QNS score). The NOSIE sub-scales were not included in the discriminant analysisbecause the raters were not blind to the patients' violentbehavior. Classification accuracy was assessed by thejackknife ("leave-one-out") procedure.

The number of patients in these analyses variedbecause of missing data. Fewer patients had QNS andNOSIE data than BPRS data. Patients with missing dataon the QNS or NOSEE were compared with patients whohad no missing data on the demographic variables and onBPRS total score; there were no significant differences.

Results

Basic Descriptive Data on the Three Groups. Table 1shows the demographic, diagnostic, and clinical charac-teristics of the nonviolent, persistently violent, and tran-siently violent groups. No significant differences in demo-graphic or background characteristics were recordedamong the three groups. The average daily dosage (inchlorpromazine equivalents) of antipsychotic medicationfor the first and last weeks of the study are provided in thetable. There were no significant differences, but the nonvi-olent patients tended to receive less medication and thetransiently violent patients more during the first week ofthe study.

Data about the number of physical and verbalassaults during the first 10 days and the subsequent 18days of the 4-week period are provided in the table.Significant differences in verbal assaults were notedamong the groups: the transiently violent patients pre-sented with more verbal assaults than the other twogroups in the initial 10-day period, and the persistentlyviolent patients presented with more assaults than theother two groups in the subsequent 18 days.

Comparison of Clinical Symptoms Among the ThreeGroups. The results for the comparisons among the per-sistently violent, transiently violent, and nonviolentpatients on psychiatric symptoms (BPRS), ward behaviors(NOSIE), and neurological symptoms (QNS) are pre-sented in table 2. Repeated-measures ANOVA was per-formed to compare the groups on the BPRS and NOSIEsubscales. The table includes group effects and interactioneffects of group X time for the BPRS total score, the

508

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Violence in Schizophrenia and Clinical Symptoms Schizophrenia Bulletin, Vol. 25, No. 3,1999

Table 1. Sociodemographic and clinical characteristics of the nonviolent, persistently violent,transiently violent groups

and

Characteristic

Male

RaceWhiteBlackHispanic

DiagnosisSchizophreniaSchizoaffective disorder

Substance use history

Head trauma history

Age, yrs.

Duration of illness, yrs.

Length of present hospitalization, days

Physical assaults1

Early2

Late3

Verbal assaultsEarly2

Late3

Antipsychotic medication dose4

First weekFourth week

Nonviolent(17 = 81)

55 (67.9)

27 (33.3)36 (44.4)18(22.2)

56(69.1)25 (30.9)

42(51.9)

12(18.2)

35.2 ± 6.8

14.4 ±6.8

28.7 ± 22.8

0.1 ±0.50.4 ±1.3

789.4 ± 587.3806.0 ± 651.8

Persistentlyviolent (n = 44)

30 (68.2)

9 (20.5)27(61.4)

8(18.2)

28 (63.6)16(36.4)

31 (70.5)

7(18.4)

33.2 ± 7.8

12.5 ±8.0

23.7 ±17.2

2.1 ±0.93.0 ±2.2

2.3 ± 2.53.3 ± 3.3

912.6 ±596.6956.8 ± 663.2

Transientlyviolent (n = 52)

29 (55.8)

16(30.8)26 (50.0)10(19.2)

29 (55.8)23 (44.2)

31 (59.6)

8 (20.0)

33.2 ± 7.7

14.0 ±7.3

28.4 ±19.1

4.7 ± 2.00.5 ± 0.7

5.0 ±4.71.5 ±1.7

1127.7 ±1028.31037.0 ±793.6

P0.30

0.47

0.29

0.13

0.97

0.19

0.46

0.24

<0.001<0.001

0.060.20

Note.—Data are n (%) or mean number ± standard deviation. Difference between groups was tested by X2 for the categorical variablesand by analysis of variance for continuous variables.1 Variable used to dichotomize the groups.2 First 10 days after initial physical assault.3 Subsequent 18 days.4 Average daily dose in chlorpromazine equivalents during the first and fourth weeks of the study.

BPRS factors, the NOSIE mean score, and the threeNOSIE subscales. Group X time interaction effects wereinvestigated further in two ways: First, the groups werecompared pair by pair at both baseline and endpoint onthe various scales and subscales. Second, overall changesin BPRS and NOSIE scores over time in each of the threegroups were examined. We used ANOVA to compare thethree groups on the QNS total score.

Psychiatric symptoms. Table 2 indicates that over-all, the violent patients presented with more severe psy-chiatric symptoms than the nonviolent control patients.On the repeated-measures ANOVA, there were significantoverall group effects (F = 9.79, df = 2,159, p < 0.001) fortotal BPRS score and for all individual BPRS factorsexcept anergia. Significant group X time interactionswere found for the total BPRS score (F - 3.44, df = 2,159,p = 0.03), as well as for the anergia, activation, and hostil-ity-suspiciousness factors.

Post hoc comparisons (t tests) indicated that at base-line, the persistently and transiently violent patients pre-sented with significantly more psychiatric symptoms, asmeasured by total BPRS score, than the nonviolentpatients; however, the two groups of violent patients didnot differ significantly from each other. At endpoint, thepersistently violent patients presented with significantlymore psychiatric-symptoms, as measured by total-BPRSscore, than the nonviolent and transiently violent groups.The latter two groups, however, no longer differed signifi-cantly from each other.

These differences between the groups in total BPRSscore were further elucidated by post hoc Bonferroni cor-rected t tests, which were performed to compare thegroups on the individual BPRS factors. At baseline, thepersistently violent patients were significantly moreimpaired than the nonviolent control patients in anxiety-depression, thought disturbance, and hostility-suspicious-

509

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Schizophrenia Bulletin, Vol. 25, No. 3, 1999 M. Krakowski et al.

Table 2. Comparisons of nonviolentpatients by clinical measures

Scale, subscales NV

(NV), persistently

PV

violent (PV),

TV

and transiently violent

Group

F P

(TV)

GroupTime

F

X

PBPRSTotal score

BaselineEndpoint

Anxiety-depressionBaselineEndpoint

AnergiaBaselineEndpoint

Thought disturbanceBaselineEndpoint

ActivationBaselineEndpoint

Hostility-suspiciousnessBaselineEndpoint

NOSIEMean score

BaselineEndpoint

RoutineBaselineEndpoint

SocialBaselineEndpoint

IrritabilityBaselineEndpoint

QNS

Total score

n = 81/731

39.3 (9.4)38.1 (10.7)

8.2 (2.8)7.9 (3.3)

9.4 (3.2)9.0 (3.2)

9.4 (4.0)9.3 (4.3)

5.4 (2.4)5.4 (2.3)

6.8 (2.9)6.5 (3.2)

n = 66/641

5.0 (1.2)5.4 (1.2)

5.4 (1.5)5.7 (1.6)

2.93(1.7)3.63(1.8)

6.7 (1.9)6.8 (1.7)

n = 671

10.2 (9.6)

n = 44/42

46.32

46.62'3

9.54

9.0

9.39.83

12.42

11.1

6.56.7

8.72

9.82'3

(10.0)(11.9)

(2.7)(2.8)

(3-1)(3.0)

(4.1)(5.0)

(2.8)(3.3)

(3.6)(3.7)

n = 41/41

3.72

4.02'3

4.64

4.74

3.43.73

CO

C

O

ro

ro

15.43'4

(1.8)(1.7)

(1.7)(1.8)

(2.0)(1.7)

(2.7)(2.6)

37

(14.5)

n = i

45.82

41.1

8.78.1

9.38.2

11.110.8

7.62

6.1

9.22

7.9

52/45

(10.3)(11.9)

(3.0)(2.8)

(3.1)(2.6)

(5.1)(5.0)

(2.6)(2.6)

(3.0)(3.7)

n = 47/42

3.82

4.9

4.64

5.4

3.94.7

2.82

4.52

n =

9.4

(1.5)(1.5)

(1-9)(1.7)

(2.2)(1.9)

(2.2)(2.4)

41

(8.6)

9.79 >0.001 3.44 0.03

3.05 0.05 0.28 0.76

0.62 0.54 4.4 0.01

4.82 0.009 1.43 0.24

6.84 0.001 3.80 0.02

12.1 <0.001 4.71 0.01

15.8 <0.001 2.79 0.06

5.02 0.008 1.55 0.22

4.73 0.01 0.85 0.43

50.02 <0.001 4.49 0.01

3.74 0.03 — —

Note.—Data are mean ± standard deviation; repeated-measures analysis of variance (ANOVA) was used for BPRS and NOSIE scalesand subscales, ANOVA was used for QNS score, and Bonferroni corrected t tests were for paired comparisons; BPRS = Brief PsychiatricRating Scale; NOSIE = Nurses' Observation Scale for Inpatient Evaluation (NOSIE mean is the mean of the three subscales; higher val-ues indicate less pathology); QNS => Quantified Neurological Scale.1 Number of patients at baseline/endpoint; number is smaller for NOSIE and QNS because of missing data.2 Greater impairment than in NV group <0.01.3 Greater impairment than in TV group <0.05.4 Greater impairment than in NV group <0.05.

510

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Violence in Schizophrenia and Clinical Symptoms Schizophrenia Bulletin, Vol. 25, No. 3, 1999

ness. The transiently violent patients presented with moreprominent activation and hostility-suspiciousness than thenonviolent control patients. There were no significant dif-ferences between the two violent groups on any BPRSfactor.

At endpoint, the persistently violent patients weresignificantly more impaired than the nonviolent controlpatients in hostility-suspiciousness and more impairedthan the transiently violent patients in both hostility-suspi-ciousness and anergia. There were no significant differ-ences between the transiently violent patients and the non-violent control patients.

The change in psychiatric symptoms from baseline toendpoint was examined in each group separately. Thetransiently violent patients, but not the other two groups,showed significant overall improvement in psychiatricsymptoms. The total BPRS score decreased significantlyover the course of the study in the transiently violentpatients (t = -3.03, df = 44, p = 0.004), but not in the per-sistently violent (t - 0.01, df = 41, p = 0.99) or nonviolent{t = -0.07, df =12,p = 0.49) patients. In line with the sig-nificant group X time interactions reported above on therepeated-measures ANOVA, significant changes occurredover time in the transiently violent group for anergia (t =-2.92, df = 44, p = 0.006), activation {t = -3.06, df = 44,p = 0.004), and hostility-suspiciousness (t - -2.17, df =44, p = 0.04).

Ward behaviors. Table 2 indicates that, overall,the violent patients presented with more disturbed wardbehavior than the nonviolent control patients; there was asignificant overall group effect (F = 15.8, df = 2,146, p <0.001) for the NOSIE mean score (on the three subscales).Significant overall group effects were also found for eachof the three NOSIE subscales; however, while the nonvio-lent control group scored better than the other two groupson the routine and irritability subscales, the transientlyviolent patients scored better on the social subscale. Atrend was found for the interaction effect of group X timefor the NOSIE mean score (F = 2.8, df = 2,146, p = 0.06).The group X time interaction effect reached significanceonly for the irritability subscale.

Post hoc Bonferroni corrected t tests indicated that atbaseline both the persistently and transiently violentpatients were more impaired in their capacity to followward rules and regulations (NOSIE routine subscale) andwere more irritable (NOSIE irritability subscale) than thenonviolent patients. The two violent groups did not differsignificantly from each other.

At endpoint, the persistently violent patients weremore irritable and had more difficulties following wardrules and regulations than the nonviolent control patients.The transiently violent patients were more irritable thanthe nonviolent control patients. The transiently violent

patients showed greater social interest (NOSIE social sub-scale) than the nonviolent patients at both time points andthan the persistently violent patients at endpoint.

The change over time from baseline to endpoint wasexamined in each group separately. In the transiently vio-lent patients, the NOSIE mean score on the three sub-scales decreased significantly over the 4 weeks (r = 3.2,df = 41, p = 0.003), but no significant time effect wasfound in the persistently violent (r = 1.01, df = 40, p =0.32) or nonviolent (t = 0.78, df = 63, p = 0.43) patients.

Neurological symptoms. Table 2 indicates a signif-icant difference among the three groups in QNS totalscore (ANOVA; F = 3.74, df = 2,144, p = 0.03). Post hoct tests indicated that the persistently violent patients pre-sented with more severe neurological symptoms, as mea-sured by the overall QNS score, than the transiently vio-lent patients (t - 2.28, df =77 , p = 0.03) and than thenonviolent control patients (t - 2.23, df = 103, p = 0.03).No significant difference was found between the tran-siently violent patients and the nonviolent control patientsin overall QNS score (t = 0.45, df = 107, p = 0.66).

We found no distinct pattern of abnormal neurologi-cal findings that would implicate a single brain area orfunctional system. The higher QNS score obtained by thepersistently violent patients was due to the cumulativeeffect of multiple items (81% of all items). However, cer-tain individual neurological signs were found significantlymore often in the persistently violent patients than in theother two groups. These signs included motor integrativeimpairment—specifically disturbances in pronation-supination, heel-to-shin movements, walking and gait,tandem gait, and foot tapping. Sequencing of complexmotor acts, especially as assessed by the fist-stretch task,was more frequently impaired. Other abnormalitiesincluded reflex asymmetry, facial asymmetry, and poorerperformance on the cancellation test.

To further elucidate the relationship between neuro-logical symptoms and violence, we investigated correla-tions between QNS total score and frequency of physicalassaults as a continuous variable in the first and secondperiods of the study. We found a significant positive corre-lation between QNS total score and the number of physi-cal assaults occurring later in the^ hospital stay" iif all theviolent patients combined (Pearson's r = 0.32, df = 77,p = 0.005); there was a nonsignificant negative relation-ship between QNS score and the number of physicalassaults occuring in the early period (Pearson's r = -0.14,df =77, p = 0.22). The same pattern was present in thepersistently violent group: QNS score was related posi-tively to late assaults (Pearson's r = 0.39, df = 36, p =0.02), but not early assaults (Pearson's r = -0.11, df = 36,p = 0.53).

511

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Schizophrenia Bulletin, Vol. 25, No. 3, 1999 M. Krakowski et al.

As mentioned above, neurological abnormalities havebeen associated with specific symptoms in schizophrenia;to better understand the role of neurological symptoms inpersistent violence, we investigated the relationshipbetween QNS score and endpoint BPRS factors in thepersistently violent group. QNS total score was relatedsignificantly with endpoint BPRS anergia (Pearson's r =0.43, df =36, p < 0.01) and with endpoint hostility-suspi-ciousness (Pearson's r = 0.53, df = 36, p < 0.001). Wefound no significant correlation with thought disturbance(r = 0.27, df = 36, p = 0.11) or with activation (r = 0.28,d f=36 ,p = 0.10).

Differentiation of the Three Groups by ClinicalVariables. Clinical data from the nonviolent, persis-tently violent, and transiently violent patients were sub-jected to a canonical discriminant-function analysis toexamine the multivariate separation among the threestudy groups. The discriminant analyses correctly classi-fied 64 percent of the nonviolent patients, 59 percent ofthe persistently violent patients, and 59 percent of thetransiently violent patients.

The discriminant analyses identified two significantdimensions (canonical variables) that differentiated thegroups (see table 3). The first canonical discriminant vari-able was characterized by positive psychotic symptoms atboth baseline and endpoint and by negative symptoms(BPRS anergia) at baseline. Neurological symptoms andendpoint negative symptoms had negligible loadings onthis variable. The second canonical variable was charac-terized most prominently by neurological impairment andnegative symptoms at endpoint; there was also a moder-

Table 3. Canonical discriminant variablesbased on BPRS factors and QNS score

Clinical variable

Thought DisturbanceBaselineEndpoint

ActivationBaselineEndpoint

Hostility-suspiciousnessBaselineEndpoint

AnergiaBaselineEndpoint

QNS

Canonicalvariable 1

0.860.84

0.950.82

0.990.79

0.880.08

0.07

Canonicalvariable 2

0.510.54

-0.300.57

-0.080.61

-0.480.99

0.99Note.—BPRS = Brief Psychiatric Rating Scale; QNS ••Neurological Scale.

Quantified

ately high loading for positive symptoms at endpoint. Theopposite sign of the loading for negative symptoms atbaseline as compared to endpoint indicated a markeddivergence between baseline and endpoint values on thiscanonical variable.

We found significant differences among the threegroups on both the first (ANOVA; F = 18.7, df = 2,144,p < 0.001) and the second (ANOVA; F = 14.6, df = 2,144,p < 0.001) canonical variables. The first canonical vari-able differentiated the nonviolent control group from bothviolent groups, but especially from the transiently violentpatients. As the class means indicate, the nonviolentpatients had a high negative score on this canonical vari-able, whereas the violent patients had positive scores; thetransiently violent patients showed the greatest differencefrom the nonviolent group (see table 4). The class meanindicated that the higher loadings on this canonical vari-able for baseline positive symptoms, compared with thoseat endpoints, and the near-zero loading for negative symp-toms and neurological symptoms at endpoint, as indicatedby the canonical coefficients (table 3), were more charac-teristic of the transiently violent patients than of the per-sistently violent group.

The second canonical variable differentiated primar-ily the persistently violent from the transiently violentpatients. The persistently violent patients had a high posi-tive mean on this variable, whereas the transiently violentpatients had a high negative mean; the nonviolent patientshad a near-zero mean.

Discussion

Differences Among the Three Groups at Baseline andEndpoint. Our results suggest that violent behavior inschizophrenic patients is a heterogeneous phenomenonbest understood in the context of other features of the ill-ness and their course over time. Early in hospitalization,positive psychotic symptoms, including thinking distur-bances, excitement, hostility, and suspiciousness, areimportant factors in violent behavior. We found that thesesymptoms were markedly more severe in the transientlyand persistently violent groups compared with the nonvio-lent control group. These findings are consistent withstudies showing an association between violent behaviorand positive symptoms (Tardiff and Sweillam 1980;Yesavage et al. 1981; Yesavage 1983; McNiel et al. 1988;Lowenstein et al. 1990; Beck et al. 1991; McNiel andBinder 1994). The violent patients in our study were alsomore irritable and had more difficulty following wardrules and regulations than the nonviolent control patients.

The transiently violent patients showed significantimprovement in psychiatric symptoms and ward behav-iors. At the end of 4 weeks, they no longer differed signif-

512

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Violence in Schizophrenia and Clinical Symptoms Schizophrenia Bulletin, Vol. 25, No. 3, 1999

Table 4. Means and analysis of variance (ANOVA) for the canonical discriminant variables

Canonical variable

First

Second

Nonviolent(n = 67)

-0.56 ± 0.93

-0.09 ± 0.82

Persistentlyviolent(n = 37)

0.34 ±1.14

0.71 ±1.11

Transientlyviolent(n = 41)

0.60 ± 0.95

-0.53 ±1.14

F

18.7

14.6

ANOVA

P

>0.01

>0.01

Note.—Data are mean ± standard deviation.

icantly from the nonviolent control patients in psychoticsymptoms or in ability to follow ward routines; also, thenumber of threats and insults (i.e., verbal assaults)markedly decreased. The persistently violent patients, onthe other hand, showed little improvement in symptoms.They presented with more severe suspiciousness and hos-tility than the other two groups and with more severe neg-ative symptoms than the transiently violent patients. Theywere more irritable than the nonviolent control patientsand had more difficulty following ward routines. Theydemonstrated less social interest than the transiently vio-lent patients.

The persistently violent patients also presented withmore severe neurological symptoms than the other twogroups. This finding was consistent with the results of anearlier study in a somewhat different population(Krakowski et al. 1989). The subjects in that study werealso patients with schizophrenia and schizoaffective dis-order in a State hospital, but they had been hospitalizedfor a longer period of time and were generally more vio-lent than patients in our study in that they required trans-fer to a special unit for the management of violent behav-ior. Our results, however, differ from those of Lapierreand coworkers (1995), who did not find any relationshipbetween neurological symptoms and violence in schizo-phrenia outpatients. The authors themselves noted thediscrepancy between their findings and previous studies,and they suggested that the subjects in their study werehigh-functioning outpatients who may not have attained alevel of neurological impairment that would contribute toviolence.

Symptom Trajectory and Group Differentiation. Thesymptom trajectories in the three groups were followedfor 4 weeks. Although positive symptoms improved in thetransiently violent group over this time period, they werestill more severe (although not significantly so) than in thenonviolent patients. Thus, in the canonical discriminantanalysis (on the first canonical variable), the violentpatients, including the transiently violent group, were dif-ferentiated from nonviolent control patients on the basisof more severe positive symptoms at both time points. Aprojection of the present symptom trajectory beyond the4-week period indicates that greater resolution of symp-

toms would have been obtained in the transiently violentpatients and that better differentiation among the groupswould have been achieved. It is possible that, over alonger follow-up period, psychotic symptoms would alsohave improved in the persistently violent group, but therewas no indication of amelioration over the 4 weeks.

Persistent negative symptoms and neurological symp-toms differentiated most clearly the persistently from thetransiently violent patients on the canonical discriminantanalyses. This differentiation was indicated by the veryhigh positive coefficient loadings for these clinical mea-sures on the second canonical variable. The first canonicalvariable highlighted the positive symptoms presentthroughout the 4-week period to differentiate the violentgroups from the nonviolent control group. In contrast, thesecond canonical variable accentuated the change insymptoms over time to differentiate persistently fromtransiently violent patients. We found a marked contrastbetween baseline and endpoint coefficient loadings on thiscanonical variable for activation, hostility-suspiciousness,and anergia—symptoms that improved significantly in thetransiently violent patients over the 4 weeks.

Persistent symptoms may be qualitatively differentfrom transient ones. Hostility and suspiciousness, whichbecame more pronounced over time in the persistentlyviolent group, may represent fairly chronic symptoms inthese patients. The beliefs and attitudes they generate maybe held with stronger conviction and may exert a greaterinfluence over behavior. They could constitute an ongoingpredisposition to violence, as with the symptoms reportedin the studies mentioned above (Virkkunen 1974; Nestoret al. 1995; Junginger 1996).

Likewise, the more persistent negative symptoms inthe persistently violent group may differ in nature fromthose in the transiently violent patients, which improvedsignificantly over the 4 weeks. This difference may corre-spond in part to the distinction drawn by certainresearchers between enduring negative symptoms and"state-related" negative symptoms, which often are moreconspicuous in the acute phase of the illness and improvewhen positive psychotic symptoms are diminished(Carpenter and Kirkpatrick 1988; Carpenter et al. 1988).During the initial phase in treatment of patients withacutely exacerbated schizophrenia, amelioration of nega-

513

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Schizophrenia Bulletin, Vol. 25, No. 3, 1999 M. Krakowski et al.

tive symptoms over time was positively related toimprovement in positive symptoms (Czobor and Volavka1996).

These persistent negative psychotic symptoms mayappear to be incompatible with violent behavior, becausepatients with prominent negative symptoms are oftensocially withdrawn and passive and thus are not prone toact out violently. However, patients with a mixed clinicalpicture, in which negative and positive symptoms coexist,are markedly different from patients with predominantlynegative symptoms. They are considerably less coopera-tive with outpatient referral, for example (Miner et al.1997). Patients with negative-syndrome schizophreniaresponded positively to appropriate structure and treat-ment, whereas those with mixed syndromes, that is,marked levels of both positive and negative symptoms,were the most noncompliant of all patient groups (Mineret al. 1997). A high incidence of inpatient assaults andviolent crime was reported for some chronically hospital-ized patients who had prominent negative and positivesymptoms (Krakowski et al. 1993). Future studies shouldassess these symptoms over longer time periods to verifytheir chronicity in violent patients.

Role of Neurological Symptoms in Violent Behavior.Although the present and previous studies (Krakowski etal. 1989) show that neurological symptoms are present inpersistently violent patients, the exact role of these symp-toms in violent behavior cannot be established throughthese findings. The way in which this impairment isexpressed or influences behavior depends on multiple fac-tors, including the available repertoire of responses estab-lished by the patient, the psychiatric symptoms, and thecognitive impairments. Certain inferences can be drawnfrom our study as to possible mechanisms or interactions,which must be elucidated further through future research.

In our study, neurological symptoms were related notsimply to the presence of assaultive behavior, but to itspersistence. Thus, neurological dysfunction may be asso-ciated with reduced ability to modify behavior or withdecreased capacity for self-correction. It may also affectviolent behavior indirectly by affecting response to treat-ment, since neurological dysfunction has been reported toreduce the impact of antipsychotic medications (Small etal. 1987; Wagman et al. 1987; Friedman et al. 1991;Breieretal. 1992).

Possibly, however, the relationship between neuro-logical symptoms and violence is more specific. In thepersistently violent group, neurological abnormalitieswere prominently associated with persistent hostility andsuspiciousness, but not with excitation or thought distur-bance. Neurological impairment may underlie certaincognitive impairments, such as difficulties in processing

cues of interpersonal communication. These difficultiescan then contribute to patients' suspiciousness and hostil-ity and eventually lead to violent behavior.

In agreement with the literature (Buchanan et al.1990; Merriam et al. 1990; Wong et al. 1997), our studyfound neurological symptoms to be associated with morepersistent negative symptoms. Negative symptoms havealso been associated with other measures of neurologicaldysfunction, such as decreased metabolic activity andblood flow in various areas of the brain (Liddle et al.1992; Wolkin et al. 1992). These negative symptoms maysimply represent an epiphenomenon of the neurologicaldysfunction and have no direct relationship to violence.They may, however, facilitate assaultive behavior. Forexample, in certain individuals severe social isolation andemotional detachment were associated with antisocial andviolent behavior that was characterized by a prominentlack of empathy with the victim (Tantam 1988). As men-tioned above, the combination of negative and positivesymptoms can have a greater impact on behavior thaneither set of symptoms alone (Miner et al. 1997). Such acombination of symptoms, then, can also work synergisti-cally to produce violence; emotional detachment anddeficits in empathy can more easily result in violencewhen they occur in conjunction with severe hostility andsuspiciousness.

Conclusion

Initially, both transiently and persistently violent groupspresented with more severe positive psychotic symptoms,were more irritable, and had greater difficulties followingward regulations compared with nonviolent patients. Inthe transiently violent patients, however, assaultive behav-ior was confined to a well-defined period of acute decom-pensation with florid psychotic excitation. In the persis-tently violent group, the violence was not limited to sucha short period; these patients evidenced more persistentnegative symptoms and suspiciousness and presented withmore severe neurological symptoms.

If we extend our findings beyond inpatient settings,such as the one in which our study was conducted, wesurmise that transiently violent patients become assaultivein the community in the context of an acute psychoticdecompensation, which may be precipitated by factorssuch as poor treatment compliance. Patients are then hos-pitalized, respond well to treatment, and can be dis-charged back into the community. The persistently violentpatients, on the other hand, do not respond as well totreatment. They are probably more difficult to dischargefrom the hospital and are therefore more often found inlong-term facilities. Investigating the behaviors and

514

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Violence in Schizophrenia and Clinical Symptoms Schizophrenia Bulletin, Vol. 25, No. 3, 1999

symptoms of these patients in the community over alonger period of time is important in establishing the gen-eralizability of our results across settings. If these differ-ent clinical contexts of assaultive behavior are confirmedin the outpatient setting, they can be helpful for both riskassessment and risk management.

References

Adams, J.J.; Meloy, G.R.; and Moritz, M.S. Neuro-psychological deficits and violent behavior in incarceratedschizophrenics. Journal of Nervous and Mental Disease,178:253-256, 1990.

Bartko, J.J., and Carpenter, W.T. Jr. On the methods andtheory of reliability. Journal of Nervous and MentalDisease, 163:307-317, 1976.

Beck, J.C., and Bonnar, J. Emergency civil commitment:Predicting hospital violence from behavior in the commu-nity. Journal of Psychiatry and Law, 16:379-388, 1988.

Beck, J.C.; White, K.A.; and Gage, B. Emergency psychi-atric assessment of violence. American Journal ofPsychiatry, 148:1562-1565, 1991.

Breier, A.; Buchanan, R.W.; Kirkpatrick, B.; Irish, D.;Brandt, D.; and Carpenter, W.T. Jr. Clozapine in schizo-phrenic outpatients: Efficacy, long-term outcome, andpredictors. Proceedings of the American College ofNeuropsychopharmacology 31st Annual Meeting,December 14-18, 1992; p. 215. Abstract.

Buchanan, R.W.; Kirkpatrick, B.; Heinrichs, D.W.; andCarpenter, W.T. Jr. Clinical correlates of the deficit syn-drome of schizophrenia. American Journal of Psychiatry,147:290-294, 1990.

Carlson, G.A., and Goodwin, F.K. The stages of mania: Alongitudinal analysis of the manic episode. Archives ofGeneral Psychiatry, 28:221-228, 1973.

Carpenter, W.T. Jr., and Kirkpatrick, B. The heterogeneityof the long-term course of schizophrenia. SchizophreniaBulletin, 14:645-652, 1988.

Carpenter, W.T. Jr.; Heinrichs, D.W.; and Wagman, A.M. I.Deficit and nondeficit forms of schizophrenia: The con-cept. American Journal of Psychiatry, 145T578-583, 1988.

Convit, A.; Volavka, J.; Czobor, P.; de Asis, J.; andEvangelista, C. Effect of subtle neurological dysfunctionon response to haloperidol treatment in schizophrenia.American Journal of Psychiatry, 151:49-56, 1994.

Czobor, P.; Bitter, I.; and Volavka, J. Relationshipbetween the Brief Psychiatric Rating Scale and the Scalefor the Assessment of Negative Symptoms: A study oftheir correlation and redundancy. Psychiatry Research,36:129-139, 1991.

Czobor, P., and Volavka, J. Positive and negative symp-toms: Is their change related? Schizophrenia Bulletin,22:577-590, 1996.

Friedman, L.; Knutson, L.; Shurell, M.; and Meltzer, H.Prefrontal sulcal prominence is inversely related toresponse to clozapine in schizophrenia. BiologicalPsychiatry, 29:865-877, 1991.

Gupta, S.; Andreasen, N.C.; Arndt, S.; Flaum, M.;Schultz, S.K.; Hubbard, W.C.; and Smith, M. Neuro-logical soft signs in neuroleptic-naive and neuroleptic-treated schizophrenic patients and in normal comparisonsubjects. American Journal of Psychiatry, 152:191-196,1995.

Heath, R.G. Psychosis and epilepsy: Similarities and dif-ferences in the anatomic-physiological substrate.Advances in Biological Psychiatry, 8:106-116, 1982.

Heinrichs, D.W., and Buchanan, R.W. Significance andmeaning of neurological signs in schizophrenia. AmericanJournal of Psychiatry, 145:11-18, 1988.

Honigfeld, G., and Klett, G. The Nurses' ObservationScale for Inpatient Evaluation: A new scale for measuringimprovement in chronic schizophrenia. Journal ofClinical Psychology, 21:65-71, 1965.

Ismail, B.; Cantor-Graae, E.; and McNeil, T.F. Neurologicalabnormalities in schizophrenic patients and their siblings.American Journal of Psychiatry, 155:84-89,1998.

Junginger, J. Psychosis and violence: The case for a con-tent analysis of psychotic experience. SchizophreniaBulletin, 22:91-103, 1996.

Kay, S.; Wolkenfeld, F.; and Murrill, L. Profiles of aggres-sion among psychiatric patients: I. Nature and prevalence.Journal of Nervous and Mental Disease, 176:539-546,1988.

Krakowski, M.; Convit, A.; Jaeger, J.; Lin, S.; andVolavka, J. Neurological impairment in violent schizo-phrenic inpatients. American Journal of Psychiatry,146:849-853,1989.

Krakowski, M.; Kunz, M.; Czobor, P.; and Volavka, J.Long-term high-dose neuroleptic treatment: Who gets itand why? Hospital and Community Psychiatry,44:640-644,1993.

Krakowski, M.; Volavka, J.; and Brizer, D. Psycho-pathology and violence: A review of literature. Com-prehensive Psychiatry, 27:131-148, 1986.

Krieger, L. Brain damage linked to violent behavior.American Medical News, 25:19-22, 1985.

Lapierre, D.; Braun, C.M.J.; Hodgins, S.; Toupin, J.;Leveillee, S.; and Constantineau, C. Neuropsychologicalcorrelates of violence in schizophrenia. SchizophreniaBulletin, 21:253-262, 1995.

515

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Schizophrenia Bulletin, Vol. 25, No. 3, 1999 M. Krakowski et al.

Liddle, P.F.; Friston, K.J.; Frith, CD.; Hirsch, S.R.; Jones,T.; and Frackowiak, R.S. Patterns of cerebral blood flowin schizophrenia. British Journal of Psychiatry,160:179-186, 1992.

Link, B.G., and Stueve, A. Psychotic symptoms and theviolent/illegal behavior of mental patients compared tocommunity controls. In: Monahan, J., and Steadman, H.J.,eds. Violence and Mental Disorder: Developments in RiskAssessment. Chicago, IL: University of Chicago Press,1994. pp. 137-159.

Lion, R.; Snyder, W.; and Merrill, G. Underreporting ofassaults on staff in a State hospital. Hospital andCommunity Psychiatry, 32:497-498, 1981.

Lowenstein, M.R.; Binder, R.L.; and McNiel, D.E. Therelationship between admission symptoms and hospitalassaults. Hospital and Community Psychiatry,41:311-313, 1990.

Manschrek, T.C.; Maher, B.A.; and Ader, D.A. Formalthought disorder, the type-token ratio, and disturbed vol-untary movement in schizophrenia. British Journal ofPsychiatry, 139:7-15, 1981.

McNiel, D.E., and Binder, R.L. The relationship betweenacute psychiatric symptoms, diagnosis, and short-termrisk of violence. Hospital and Community Psychiatry,45:133-137, 1994.

McNiel, D.E.; Binder, R.L.; and Greenfield, T.K.Predictors of violence in civilly committed acute psychi-atric patients. American Journal of Psychiatry,145:965-970, 1988.

Merriam, A.E.; Kay, S.R.; Opler, L.A.; Kushner, S.F.; andVan Praag, H.M. Neurological signs and the positive-neg-ative dimension in schizophrenia. Biological Psychiatry,28:181-192, 1990.

Miner, C ; Rosenthal, R.N.; Hellerstein, D.J.; and Muenz,L.R. Prediction of compliance with outpatient referral inpatients with schizophrenia and psychoactive substanceuse disorders. Archives of General Psychiatry,54:706-712, 1997.

Monroe, R.R.; Hulfish, B.; Balis, G.; Lion, J.; Rubin, J.;McDonald, M.; and Barick, J.D. Neurologic findings inrecidivist aggressors. In: Shagess, C ; Gershon, S.; andFriedhoff, A.J., eds. Psychopathology and BrainDysfunction. New York, NY: Raven Press, 1977.pp. 241-253.

Nestor, P.G.; Haycock, J.; Doiron, S.; Kelly, J.; and Kelly,D. Lethal violence and psychosis: A clinical profile.Bulletin of the American Academy of Psychiatry and theLaw, 23:331-341, 1995.

Overall, J.E., and Gorham, D.R. The Brief PsychiatricRating Scale. Psychological Reports, 10:799-812, 1962.

Rubin, P.; Vorstrup, S.; Hemmingsen, R.; Andersen, H.S.;Bendsen, B.B.; Stromso, N.; Larsen, J.K.; and Bolwig,T.G. Neurological abnormalities in patients with schizo-phrenia or schizophreniform disorder at first admission tohospital: Correlations with computerized tomography andregional cerebral blood flow findings. Ada PsychiatricaScandinavica, 90:385-390, 1994.

Small, J.G.; Millstein, V.; Small, I.F.; Miller, M.J.;Kellams, J.J.; and Corsaro, C.J. Computerized EEG pro-files of haloperidol, chlorpromazine, and placebo in treat-ment-resistant schizophrenia. Clinical Encephalography,18:124-135, 1987.

Spitzer, R.L.; Williams, J.B.W.; Gibbon, M.; and First,M.B. Structured Clinical Interview for DSM-III-R. NewYork, NY: Biometrics Research Department, New YorkState Psychiatric Institute, 1988.

Swanson, J.W. Mental disorder, substance abuse, andcommunity violence: An epidemiological approach. In:Violence and Mental Disorder: Developments in RiskAssessment. Monahan, J., and Steadman, H.J., eds.Chicago, IL: University of Chicago Press, 1994.pp. 101-136.

Swanson, J.W.; Holzer, C.E.; Ganju, V.K.; and Jono, R.T.Violence and psychiatric disorder in the community:Evidence from the Epidemiologic Catchment Area sur-veys. Hospital and Community Psychiatry, 41:761-770,1990.

Tantam, D. Lifelong eccentricity and social isolation: I.Psychiatric, social, and forensic aspects. British Journalof Psychiatry, 153:777-782, 1988.

Tardiff, K., and Sweillam, A. Assault, suicide, and mentalillness. Archives of General Psychiatry, 37:164-169, 1980.

Virkkunen, M. Observations on violence in schizophrenia.Ada Psychiatrica Scandinavica, 50:145-151, 1974.

Volkow, N.D., and Tancredi, L. Neural substrates of vio-lent behaviour: A preliminary study with positron emis-sion tomography. British Journal of Psychiatry,151:668-673, 1987.

Wagman, A.M.; Heinrichs, D.W.; and Carpenter, W.T. Jr.Deficit and nondeficit forms of schizophrenia: Neuro-psychological evaluation. Psychiatry Research, 22:319-330, 1987.

Wolfgang, M.E. Delinquency and violence from the view-point of criminology. In: Field, W., and Sweet, W., eds.Neural Bases of Violence and Aggression. St. Louis, MO:Green, 1975. pp. 456-490.

Wolkin, A.; Sanfilipo, M.; Wolf, A.P.; Angrist, B.; Brodie,J.D.; and Rotrosen, J. Negative symptoms and hypo-frontality in chronic schizophrenia. Archives of GeneralPsychiatry, 49:959-965. 1992.

516

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Violence in Schizophrenia and Clinical Symptoms Schizophrenia Bulletin, Vol. 25, No. 3, 1999

Wong, A.H.; Voruganti, L.N.; Heslegrave, R.J.; and Awad,A.G. Neurocognitive deficits and neurological signs inschizophrenia. Schizophrenia Research, 23:139-146,1997.

Woods, B.T.; Kinney, D.K.; and Yurgelun-Todd, D.A.Neurological "hard" signs and family history of psychosisin schizophrenia. Biological Psychiatry, 30:806-816,1991.

Yesavage, J.A. Inpatient violence and the schizophrenicpatient. Ada Psychiatrica Scandinavica, 67:353-357,1983.

Yesavage, J.A.; Werner, P.D.; Becker, J.M.; Holman, C ;and Mills, M. Inpatient evaluation of aggression in psy-chiatric patients. Journal of Nervous and Mental Disease,169:299-301, 1981.

Yeudall, L.T.; Fromm-Auch, D.; and Davies, P. Neuro-psychological impairment of persistent delinquency.Journal of Nervous and Mental Disease, 170:257-265,1982.

Yudofsky, S.C.; Silver, J.M.; and Jackson, W. The OvertAggression Scale for the objective rating of verbal andphysical aggression. American Journal of Psychiatry,143:35-39, 1986.

Acknowledgments

This work was supported by grant MH—45454 from theNational Institute of Mental Health. We are grateful toMorris Meisner, Ph.D., Henry Glickman, Ph.D., and

Shang Lin Pin, Ph.D., for critical comments that improvedthis report. We gratefully acknowledge the contributionsof Lorraine Schmader, R.N., Theresa Abad, R.N., JacioneNagel, R.N., Jan Libiger, M.D., Josephine Caparas, M.D.,Michal Kunz, M.D., Hana Papezova, M.D., FrankoStepcic, M.D., Gordana Peric, M.D., and Eugene Zubkov,M.D., for patient evaluations. Jane O'Donnell, R.N., JulieScheurer, M.D., and Milica Stefanovic, M.D., were thestudy coordinators; and Jozsef Vitrai, Ph.D., EvaCzoborne, M.A., and Tovit Krakowski, M.A., managedthe database. We also appreciate the cooperation of theadministrators and staff at the Rockland PsychiatricCenter, Orangeburg, NY, and at the Manhattan PsychiatricCenter, New York, NY.

The Authors

Menahem Krakowski, M.D., Ph.D., is ResearchPsychiatrist, Nathan Kline Institute for PsychiatricResearch, Orangeburg, NY, and Assistant Professor,Department of Psychiatry, New York University School ofMedicine, New York, NY. Pal Czobor, Ph.D., is ResearchScientist, Nathan Kline Institute for Psychiatric Research,and Research Associate Professor, Department ofPsychiatry, New York University School of Medicine.James C.-Y. Chou is Research Psychiatrist, Nathan KlineInstitute for Psychiatric Research, and Assistant Professor,Department of Psychiatry, New York University School ofMedicine.

517

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019

Dow

nloaded from https://academ

ic.oup.com/schizophreniabulletin/article-abstract/25/3/505/1849503 by guest on 05 April 2019