JOURNAL OF

PSYCHIATRIC
Journal of Psychiatric Research 39 (2005) 109–115
RESEARCH
www.elsevier.com/locate/jpsychires

Aggression and psychopathology in treatment-resistant

inpatients with schizophrenia and schizoaffective disorder
Karen A. Nolana,b,*, Jan Volavkaa,b, Pal Czobora,b, Brian Sheitmanc,
Jean-Pierre Lindenmayerb, Leslie L. Citromea,b, Joseph McEvoyd, Jeffrey A. Liebermanc
a
Nathan Kline Institute for Psychiatric Research, 140 Old Orangeburg Road, Orangeburg, NY 10962, USA
b
New York University School of Medicine, New York, NY USA
c
University of North Carolina at Chapel Hill, Chapel Hill, NC, USA
d
Duke University, Durham, NC, USA
Received 2 January 2004; received in revised form 17 April 2004; accepted 30 April 2004

Abstract

Positive psychotic symptoms, such as threat/‘‘control-override’’ delusions or command hallucinations, have been related to
aggression in patients with schizophrenia. However, retrospective data collection has hampered evaluation of the direct influence of
psychopathology on aggressive behavior. In this study, we monitored aggressive behavior and psychopathology prospectively and in
close temporal proximity in 157 treatment-resistant inpatients diagnosed with chronic schizophrenia or schizoaffective disorder
participating in a 14-week double-blind clinical trial. Aggressive behavior was rated with the overt aggression scale (OAS). Psy-
chopathology was assessed using the positive and negative syndrome scale (PANSS). At baseline, subjects who would be aggressive
during the study had higher scores on only two PANSS items: hostility and poor impulse control. During the study PANSS positive
subscale scores were significantly higher in aggressive subjects. Total PANSS scores were higher within 3 days of an aggressive
incident, as were positive and general psychopathology subscale scores. However, in a smaller subsample for whom PANSS ratings
were available within 3 days before aggressive incidents, only scores on the PANSS positive subscale were significantly higher. These
findings in chronic, treatment resistant inpatients support the view that positive symptoms may lead to aggression.
Ó 2004 Elsevier Ltd. All rights reserved.

Keywords: Schizophrenia; Aggression; Psychopathology

1. Introduction violence (Bjørkly, 2002a,b) conclude that there is limited

evidence directly linking TCO symptoms and violence.
Schizophrenia is associated with an elevated risk for That specific psychotic symptoms sometimes moti-
aggressive behavior (Volavka, 2002), but the relation- vate aggression in some patients is not disputed, how-
ship between psychopathology and overt aggression is ever consensus has not been reached either on the extent
incompletely understood. Link et al. (1992) reported of their influence nor how they interact with other
that severity of psychotic symptoms was significantly clinical or demographic factors. Several studies have
related to violence in the mentally ill. Threat/control indicated that psychotic symptoms are important de-
override (TCO) symptoms (i.e., delusions of thought terminants of aggressive behavior in mentally ill indi-
insertion, that one is dominated by external forces, or viduals. In Taylor’s (1985) sample of psychotic
that people wish to do one harm) have been associated offenders, the majority were actively psychotic at the
with violence (Link et al., 1998). However, two recent time of the offense and, though only 20% reported that
reviews of the literature on psychotic symptoms and they were directly driven to offend by their psychotic
symptoms, a delusional motive was considered probable
for an additional 23% of the psychotic men. In addition,
*
Corresponding author. Tel.: +845-398-6572; fax: +845-398-6566. delusionally motivated offenses were more likely to
E-mail address: [email protected] (K.A. Nolan). have had serious consequences. Among patients with a

0022-3956/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jpsychires.2004.04.010
110 K.A. Nolan et al. / Journal of Psychiatric Research 39 (2005) 109–115

history of delusions, 40% of those who had been violent vka et al., 2002). We have already analyzed the effects of
retrospectively reported at least one violent incident that these treatments on hostility (Citrome et al., 2001) and
was probably or definitely motivated by a concurrent on acts of overt aggression (Volavka et al., 2004). The
delusion (Junginger et al., 1998). In one small study, secondary analysis presented here was undertaken to
which focused on 25 individuals with persecutory delu- more specifically examine the relationship between
sions, 20% of the subjects reported recent aggression as schizophrenic psychopathology and overt aggression.
‘‘safety behaviours’’ (i.e., behavior carried out with the
goal of reducing persecutory threat) (Freeman et al.,
2001). Although no correlation was found between the 2. Methods
presence of TCO delusions and violence in a prospective
study of patients recently discharged from acute psy- Subjects were 157 inpatient participants in a 14-week
chiatric care (Appelbaum et al., 2000), a significant as- prospective, double-blind trial conducted at four state
sociation was observed when TCO symptoms of less psychiatric facilities (two in New York and two in North
than delusional severity were examined. McNiel et al. Carolina). Each subject provided written informed
(2003) reported that ‘‘anger and suspicious, paranoid consent, after receiving a complete description of the
symptoms’’ were associated with increased violent be- study, according to the guidelines of the local Institu-
havior in psychiatric patients. In that study, community tional Review Board. All met DSM-IV criteria for
violence was retrospectively assessed by self-report but schizophrenia or schizoaffective disorder, were between
the assessments of psychopathology and mood were the ages of 18 and 60, and had a minimum score of 60 on
administered while the subjects were hospitalized. the PANSS at baseline. In addition, all subjects had a
Violent behavior in the community differs from vio- history of suboptimal treatment response, defined by
lent behavior in the hospital in many respects; predictors two criteria, both of which had to be present at baseline:
of community violence such as gender, age and alcohol (1) persistent positive symptoms (either current or doc-
abuse appear to play a lesser role in inpatient aggres- umented in the past) continuing after at least six con-
sion, where psychopathology may be more important secutive weeks of treatment with one or more
(Steinert, 2002). In addition, it may be more feasible to conventional antipsychotics at daily dosages equivalent
study aggression prospectively in an inpatient setting. to at least 600 mg of chlorpromazine and (2) poor level
Cheung et al. (1997) compared 31 inpatients with of functioning over the previous two years as indicated
schizophrenia who had been physically aggressive with by lack of competitive employment or enrollment in an
31 non-aggressive comparison subjects. They reported educational or vocational program and absence of
that aggressive patients had higher positive symptom, age-expected interpersonal relationships involving on-
negative symptom, general psychopathology and total going regular contact outside the biological family.
PANSS scores than non-aggressive patients. Steinert Study participants were randomly assigned to receive
et al. (2000b) reported a significant association between clozapine, olanzapine, risperidone, or haloperidol.
severity of aggression during inpatient treatment and Ninety-one subjects completed the 14-week study; the
PANSS positive subscale at admission. In that study, data reported here were derived from all available data
neither negative subscale scores nor general psychopa- for the 157 subjects who entered the study.
thology subscale scores were associated with aggression. Aggressive behavior was monitored by research staff
Arango et al. (1999) examined PANSS scores obtained (nurses and study coordinators) during the study using
at hospital admission in patients who were or were not the overt aggression scale (OAS) (Yudofsky et al., 1986),
aggressive during the subsequent hospital stay. Patients which categorizes aggressive behavior according to
who were subsequently violent scored higher on six out means of expression and object of aggression: verbal
of seven PANSS positive subscale items, but only one aggression, physical aggression against objects, physical
negative subscale item (poor rapport) and three general aggression against self, and physical aggression against
psychopathology subscale items (uncooperativeness, other people. Within each category there are four levels
lack of judgment and insight, and poor impulse control). of severity, with weighted scores ranging from 1 (lowest
In each of these studies, aggression and psychopathol- level of verbal aggression) to 6 (physical aggression
ogy were assessed at different times. A comprehensive against self or other people, resulting in serious injuries).
prospective assessment of psychopathology in temporal The OAS was completed for each episode of aggressive
proximity to aggressive incidents has not been pub- behavior during the study. All behaviors emitted in an
lished. The primary purpose of this paper is to provide episode were recorded. The overall severity of each ag-
such data. gressive incident was rated by summing the weighted
These data were collected in the context of a study score for the most severe behaviors within each category
comparing the effects of clozapine, olanzapine, risperi- for an event (Silver and Yudofsky, 1991).
done, and haloperidol in treatment-resistant inpatients Incidents of overt aggression were ascertained retro-
with schizophrenia and schizoaffective disorder (Vola- spectively during a baseline period (up to 90 days pre-
 K.A. Nolan et al. / Journal of Psychiatric Research 39 (2005) 109–115 111

ceding randomization) from available documentation, (PANSS total, subscales and factors) were examined by
and then observed prospectively during the study. To one-way ANOVA; a separate analysis was performed
detect incidents of aggression during the study, research for each of the variables.
staff talked to subjects and ward staff and reviewed ward
documentation (such as progress notes, incident reports, 2.1.2. Temporal relationship between psychopathology
shift-to-shift reports) daily. When an aggressive incident and aggression
was detected, research staff interviewed the nursing staff Because PANSS ratings were performed according
who reported or documented the incident, as well as to the schedule specified by the study protocol and
other witnesses to verify the incident and to obtain a aggressive behavior occurred at unpredictable times, in
detailed description. When an incident of aggressive order to examine the temporal relationship between
behavior was confirmed, the research staff then used this psychopathology and aggression it was necessary to
information to complete the OAS rating. adopt a somewhat arbitrary definition of temporal
The PANSS was used to assess severity of psycho- proximity. In the primary analysis, hierarchical linear
pathology. It consists of 30 items, each rated on a scale model (HLM) (Gueorguieva and Krystal, 2004) anal-
of 1 (absent) to 7 (severe). In addition to a total PANSS ysis was applied to all time points data to compare
score, subscale scores can be calculated for positive PANSS ratings performed within 3 days of an ag-
symptoms (7 items), negative symptoms (7 items) and gressive incident to those performed more than 3 days
general psychopathology (16 items). We also analyzed before or after an incident. In the HLM analyses,
PANSS scores in this study according to Lindenmayer repeated assessments of symptom severity, as indexed
et al.’s (1994) 5-factor model of psychopathology which by the total and subscale scores of the PANSS, were
assigns PANSS items to positive, negative, excitement, applied as response (dependent) variables in the
cognitive and depression/anxiety components. The model. The independent variable of interest was vio-
PANSS was administered every week for the first four lence status (aggressive incident ‘present’ or ‘absent’
weeks of the study and every other week thereafter for within 3 days of the rating). In order to account for
the duration of the subject’s participation. The interrater the effect of potential covariates, ‘time’, ‘medication’,
reliability for PANSS total score, estimated by intraclass ‘aggression status’ (aggressive versus non-aggressive
coefficients for paired ratings at each of the four study subjects), and interaction among these factors were
sites, ranged from 0.93 to 0.98. PANSS ratings were included as additional independent variables in the
performed by doctoral-level professional staff. model. A separate analysis was conducted for each
variable of interest including the PANSS total and
2.1. Statistical analyses subscale scores and 5-factor model component
scores. A secondary HLM analysis focused on PANSS
2.1.1. Psychopathology in aggressive versus non-aggres- ratings performed within 3 days before an aggressive
sive subjects incident.
Differences between aggressive versus non-aggressive
subjects in individual symptoms at baseline was inves-
tigated with a mixed model repeated measures analysis 3. Results
of variance. ‘Group’ (aggressive versus non-aggressive)
was applied as a between-subjects factor. ‘Item’, an in- During the study, there were 408 incidents of ag-
dicator variable indexing PANSS item, was used as a gressive behavior; 88 subjects had one or more aggres-
within-subject (repeated measures) factor. Interaction sive incidents. The total number of incidents per subject
between ‘group’ and ‘item’ was included in the model. If ranged from 0 to 27. Among subjects with at least one
the interaction between group and item reached statis- aggressive event, the mean number of incidents was 4.5
tical significance, post-hoc ANOVA analyses were con- (SD ¼ 5.3). Weighted severity scores for individual in-
ducted to investigate differences for each of the cidents ranged from 1 to 10 with an overall mean of 3.4
individual items. The Westfall method (1997) was ap- (SD ¼ 2.1). For individual subjects, mean severity was
plied to control for the inflated probability of the erro- not correlated with the number of incidents (r ¼
0:09,
neous rejection of the null-hypothesis (Type 1 error) due p > 0:05).
to multiple testing (i.e., analysis of individual items of Verbal aggression was most frequent, occurring either
the PANSS scale). This procedure, as compared to the alone ðn ¼ 266Þ or in combination with other forms of
traditional methods of multiplicity correction (e.g., aggression ðn ¼ 98Þ. Physical aggression was the second
Bonferroni adjustment), achieves greater statistical most frequent form of aggression ðn ¼ 96Þ and occurred
power because it incorporates correlations among test- more often in combination with other forms of aggres-
statistics and accounts for potential logical constraints. sion ðn ¼ 68Þ than alone. Physical aggression against
Differences between aggressive and non-aggressive sub- self occurred infrequently during this study ðn ¼ 11Þ and
jects in broad measures of psychopathology at baseline was not correlated with any other type of aggression. All
112 K.A. Nolan et al. / Journal of Psychiatric Research 39 (2005) 109–115

other forms of aggression were highly correlated with All PANSS scores obtained during the study were
each other. compared in aggressive and non-aggressive subjects.
Aggressive behaviors are considered in the rating of the
3.1. Psychopathology in aggressive versus non-aggressive hostility item and potentially contribute to both total
subjects and positive subscale scores. In the analyses that follow,
the hostility item was eliminated from total and positive
For these analyses, subjects who had at least one in- subscale scores, and the maximum possible scores were
cident of aggressive behavior during the study were reduced to 203 (total) and 42 (positive). We subtracted
considered ‘‘aggressive’’. Baseline (study entry) PANSS the hostility item from the 5-factor model excitement
ratings were compared in aggressive and non-aggressive component score.
subjects. There were significant ðp < 0:05Þ between The results of this analysis are summarized in Table
group differences on a number of PANSS items: P2 1. Mean total PANSS scores were not different in ag-
(conceptual disorganization), P4 (excitement), P7 (hos- gressive and non-aggressive subjects, nor were there
tility), N6 (lack of spontaneity), G5 (mannerisms and significant between group differences on scores for either
posturing), G13 (disturbance of volition) G14 (poor the negative or general psychopathology subscales. Po-
impulse control). However, after Bonferroni multiplicity sitive subscale scores were significantly higher in ag-
correction, the group differences remained significant for gressive subjects.
only two items: hostility (adj. p ¼ 0:0026; mean differ- Positive component scores were slightly but signifi-
ence ¼ 0.69 points) and poor impulse control (adj. cantly higher in aggressive subjects. Aggressive and
p < 0:0001, mean difference ¼ 0.92 points). Two addi- non-aggressive subjects did not differ on negative or de-
tional items were marginally significant: conceptual pression/anxiety component scores. Scores for excitement
disorganization (adj. p ¼ 0:0645, mean difference ¼ 0.59 (without the hostility item) and cognitive components
points) and disturbance of volition (adj. p ¼ 0:0699, were significantly higher in aggressive subjects.
mean difference ¼ 0.47 points).
Baseline PANSS scores were factor analyzed sepa-
rately for the two groups. Varimax rotation yielded 5- 3.2. Temporal relationship between psychopathology and
factor solutions for each group, each accounting for aggression
approximately 60% of the variance. However, compar-
ison of the correlation matrices of the 30 PANSS items The results of this analysis are summarized in Table
for the violent and non-violent subjects indicated a sig- 2. Total PANSS scores (hostility item omitted) within 3
nificant lack of homogeneity (v2 ¼ 581.69, df ¼ 60, days of an incident were approximately 5.3 points (0.18
p < 0:0001) between the two groups. For non-aggressive points per item) higher (F ¼ 26.95, df ¼ 1,981, p <
subjects, the factor structure was similar to Lindenma- 0:0001) than those performed more than 3 days before
yer et al.’s (1994) 5-factor model, comprised of (in de- or after an incident. Positive subscale scores (Hostility
creasing order of explained variance) negative, positive, item omitted) performed within 3 days of an aggressive
cognitive, excitement, and anxiety/depression dimen- incident were approximately 1.2 points (0.2 points per
sions. For aggressive subjects, negative and excitement item) higher (F ¼ 12.41, df ¼ 1,981, p ¼ 0:0004). Nega-
dimensions were combined in the first factor, followed tive subscale scores did not differ as a function of
by positive, cognitive, and anxiety/depression. The fifth proximity to an aggressive incident. General psychopa-
factor consisted of a single item (G7 motor retardation). thology scores were approximately 3.5 points (0.22

Table 1
Mean estimated difference in PANSS scores (expressed as points per item) between violent and non-violent subjects
Items Difference Fb p
a
Total score 29 0.14 1.88 0.171
Subscales
Positivea 6 0.25 9.07 0.0027
Negative 7 0.09 1.47 0.2263
General psychopathology 16 0.17 3.01 0.0833
Component scores
Positive 4 0.09 4.04 0.0446
Negative 6 )0.01 2.55 0.1108
Excitementa 3 0.46 9.23 0.0024
Cognitive 5 0.62 7.64 0.0058
Depression/anxiety 5 0.18 0.10 0.7557
a
Hostility item removed.
b
df ¼ 1,981.
 K.A. Nolan et al. / Journal of Psychiatric Research 39 (2005) 109–115 113

Table 2
Mean estimated difference in PANSS scores (expressed as points per item) obtained within 3 days of an aggressive incident
Items Difference Fb p
a
Total score 29 0.18 26.95 <0.0001
Subscales
Positivea 6 0.20 12.41 0.0004
Negative 7 0.11 3.96 0.0469
General psychopathology 16 0.22 34.01 0.0001
a
Hostility item removed.
b
df ¼ 1,981.

Table 3
Mean estimated difference in PANSS scores (expressed as points per item) obtained within 3 days before an aggressive incident
Items Difference Fb p
a
Total score 29 0.15 3.54 0.0602
Subscales
Positivea 6 0.28 4.59 0.0324
Negative 7 0.08 0.47 0.4933
General psychopathology 16 0.13 2.62 0.1059
a
Hostility item removed.
b
df ¼ 1,917.

points per item) higher for PANSS ratings within 3 days nitude of the observed differences is relatively small. The
of an aggressive incident (F ¼ 34.01, df ¼ 1,981, difference in PANSS total scores obtained within 3 days
p < 0:0001). of an aggressive incident compared to those obtained
In order to examine whether increased psychopa- more than 3 days from an aggressive incident is on the
thology precedes aggression, the analysis was repeated order of 5 points. In addition, since PANSS raters were
comparing PANSS scores obtained within 3 days before encouraged to review the chart when making their rat-
an aggressive incident to those obtained more than 3 ings, it is possible that knowledge that aggressive be-
days before or after an incident, the pattern of results havior had recently occurred influenced ratings more
was similar. The sample size was reduced, PANSS rat- generally.
ings within three days before at least one aggressive in- Steinert et al. (2000b) reported a significant associa-
cident available for only 35 of the aggressive subjects. In tion between severity of aggression during inpatient
this analysis, only positive subscale scores were signifi- treatment and PANSS positive subscale at admission. In
cantly higher before aggressive incidents (see Table 3). that study, neither negative subscale scores nor general
psychopathology subscale scores were associated with
aggression. However, their methods differed from ours
4. Discussion in important ways. Steinert et al. administered the
PANSS at a single time point not linked to the occur-
In this study of treatment-resistant inpatients with rence of aggression. In addition, although they rated
schizophrenia, the occurrence of any aggressive behav- aggressive behavior during the subject’s entire inpatient
ior was related specifically to the severity of positive stay, this was done retrospectively, and only the most
symptoms. This relationship held true even when the severe incident was analyzed in relation to psychopa-
hostility item was eliminated from the positive subscale thology. Our analysis of baseline PANSS scores indi-
score. In addition, in the days preceding aggressive ep- cated that subjects who went on to engage in aggressive
isodes, positive symptoms were elevated. These data behavior during the study differed from the nonaggres-
agree with and extend previous reports linking aggres- sive patients on only two items: hostility and poor im-
sion with psychosis. pulse control. Thus, a single, arbitrarily timed
Our data indicate an increase in the overall severity of measurement of psychopathology does not appear to
psychopathology in the time period surrounding ag- capture the complexity of the relationship with aggres-
gressive behavior; PANSS total scores (hostility item sion that emerges in temporally associated ratings.
removed) performed within 3 days of an aggressive It should be noted, however, that our main tem-
event were approximately 5 points higher than those poral variable lacked directional specificity; i.e., we
obtained at times not in proximity to aggressive be- compared PANSS scores on the basis of absolute
havior. Around the time of an aggressive incident, scores temporal proximity to the aggressive behavior. Thus,
on both the positive subscale and the general psycho- scores for ratings performed just prior to aggressive
pathology subscale were elevated. However, the mag- incidents were combined with ratings performed just
114 K.A. Nolan et al. / Journal of Psychiatric Research 39 (2005) 109–115

after the aggression. Proximity of a PANSS assess- very violent patients. Furthermore, the underlying
ment to an aggressive episode was coincidental, not causes of aggressive behavior may be quite different in
planned. In addition, the PANSS raters were not patients with severe cognitive impairments, who
blinded to aggressive behavior and in fact were en- would not have met the capacity requirement for this
couraged to review the progress notes for the week study.
preceding each rating. It is possible that knowledge of However, in another recent study which focused
the occurrence of aggressive behavior could have in- specifically on physically assaultive behavior (Nolan
fluenced ratings generally. et al., 2003) we also found evidence that psychotic
The association between aggression and negative symptoms may play a direct causal role. In that study,
component scores (but not negative subscale scores) was we used a semi-structured interview to elicit reasons for
unexpected. Five of the six items contributing to the assaults from inpatient assailants and their victims
negative component score are drawn from the seven shortly after the assault occurred. Consensus clinical
items comprising the negative subscale. The sixth neg- ratings indicated that approximately 20% of the assaults
ative component item is G16 – active social avoidance – in that sample were directly related to positive psychotic
for which ratings are based on evidence of ‘‘Diminished symptoms. Factor analysis revealed two ‘‘psychosis’’
social involvement associated with unwarranted fear, factors, one related to positive psychotic symptoms and
hostility, or distrust.’’ This suggests that it may not be the other to psychotic confusion and disorganization, as
negativity per se which is increased in temporal prox- well as a third factor that differentiated impulsive versus
imity to aggression, but an emotional state of anxiety, psychopathic assaults. Thus, psychosis also appears to
suspiciousness, or hostility severe enough to have an contribute to more serious physical aggression.
observable effect on social interactions. In addition, It is also important to note that these results were
pointing to the complexity of behavioral correlates of obtained in a sample of chronic, treatment-resistant in-
aggressive episodes, it is possible that individuals who patients. Other studies have demonstrated that aggres-
scored higher on negative symptoms lacked the social sion occurring in the hospital has different correlates
skills to defuse an interpersonal aggressive episode either from aggression in the community and that the majority
verbally or with other adaptive behavior. It is also of inpatient aggressive episodes occur within the first
possible that the association between aggression and week of neuroleptic treatment (Steinert et al., 2000a).
negative component scores observed in this sample is
mediated by secondary rather than primary negative
symptoms. Acknowledgements
Our secondary analysis contrasted PANSS ratings
obtained within 3 days just prior to an incident to rat- NIMH Grant (R10 MH53550) provided the principal
ings that were not obtained near the time of aggressive support for this project. Additional support was pro-
incidents. In this analysis, only positive subscale scores vided by the UNC-Mental Health and Neuroscience
were significantly elevated (approximately 1.7 points). Clinical Research Center (MH MH33127) and the
However, PANSS ratings available within 3 days before Foundation of Hope, Raleigh North Carolina. We
an incident for only 35 of the 88 aggressive subjects. thank Janssen Pharmaceutica Research Foundation, Eli
Since there was a significant reduction of sample size for Lilly and Company, Novartis Pharmaceuticals Corpo-
this analysis, this result should be interpreted cautiously. ration, and Merck and Co., Inc. for their generous gifts
Nevertheless, it does imply that the increase in negative of medications. Eli Lilly and Company contributed
component scores discussed above may be present only supplemental funding that covered approximately 18%
in the post-aggression ratings. of the total cost of the study. However, overall experi-
One of the limitations of our findings is that ag- mental design, data acquisition, statistical analyses, and
gression was not the focus of the study in which these interpretation of the results were implemented without
data were collected. The participants in this study any input from any of the pharmaceutical companies.
were not selected for aggressive behavior and the Linda Kline, RN, MS, CS was the chief coordinator of
relative infrequency of aggression in this sample did the entire project.
not permit a more detailed analysis of the relation-
ships between specific symptoms and aggression or
between symptoms and subtypes of aggression. In References
addition, the severity of aggressive events was gener-
ally low. Although strong correlations between verbal Appelbaum PS, Robbins PC, Monahan J. Violence and delusions: data
from the MacArthur violence risk assessment study. American
and physical aggression have been documented else-
Journal of Psychiatry 2000;157:566–72.
where (Kay et al., 1988; Werner et al., 1983), the re- Arango C, Calcedo BA, Gonzalez S, Calcedo OA. Violence in
lationship between psychopathology and aggression inpatients with schizophrenia: a prospective study. Schizophrenia
may be quite different in persistently aggressive or Bulletin 1999;25:493–503.
 K.A. Nolan et al. / Journal of Psychiatric Research 39 (2005) 109–115 115

Bjørkly S. Psychotic symptoms and violence toward others – a Nolan KA, Czobor P, Roy BB, Platt MM, Shope CB, Citrome LL,
literature review and some preliminary findings Part 2. Hallucina- Volavka J. Characteristics of assaultive behavior in psychiatric
tions. Aggression and Violent Behavior 2002a;7:605–16. inpatients. Psychiatric Services 2003;54:1012–6.
Bjørkly S. Psychotic symptoms and violence toward others – a Silver JM, Yudofsky SC. The overt aggression scale: overview and
literature review of some preliminary findings Part 1. Delusions. guiding principles. Journal of Neuropsychiatry and Clinical Neu-
Aggression and Violent Behavior 2002b;7:617–31. rosciences 1991;3:s22–9.
Cheung P, Schweitzer I, Crowley K, Tuckwell V. Aggressive behaviour Steinert T. Prediction of inpatient violence. Acta Psychiatrica Scan-
in schizophrenia: the role of psychopathology. Australian and New dinavica Supplement 2002:133–41.
Zealand Journal of Psychiatry 1997;31:62–7. Steinert T, Sippach T, Gebhardt RP. How common is violence in
Citrome L, Volavka J, Czobor P, Sheitman B, Lindenmayer JP, schizophrenia despite neuroleptic treatment? Pharmacopsychiatry
McEvoy J, Cooper TB, Chakos M, Lieberman JA. Effects of 2000a;33:98–102.
clozapine, olanzapine, risperidone, and haloperidol on hostility Steinert T, Wolfle M, Gebhardt RP. Measurement of violence during
among patients with schizophrenia. Psychiatric Services in-patient treatment and association with psychopathology. Acta
2001;52:1510–4. Psychiatrica Scandinavica 2000b;102:107–12.
Freeman D, Garety PA, Kuipers E. Persecutory delusions: developing Taylor PJ. Motives for offending among violent and psychotic men.
the understanding of belief maintenance and emotional distress. British Journal of Psychiatry 1985;147:491–8.
Psychological Medicine 2001;31:1293–306. Volavka J. Neurobiology of violence. Washington, DC: American
Gueorguieva R, Krystal JH. Move over ANOVA: progress in Psychiatric Publishing, Inc.; 2002.
analyzing repeated-measures data and its reflection in papers Volavka J, Czobor P, Nolan KA, Sheitman B, Lindenmayer JP,
published in the Archives of General Psychiatry. Archives of Citrome L, McEvoy JP, Cooper TB, Lieberman JA. Overt
General Psychiatry 2004;61:310–7. aggression and psychotic symptoms in patients with schizophrenia
Junginger J, Parks-Levy J, McGuire L. Delusions and symptom- treated with clozapine, olanzapine, risperidone, or haloperidol.
consistent violence. Psychiatric Services 1998;49:218–20. Journal of Clinical Psychopharmacology 2004;24:225–8.
Kay SR, Wolkenfeld F, Murrill LM. Profiles of aggression among Volavka J, Czobor P, Sheitman B, Lindenmayer JP, Citrome L,
psychiatric patients. I. Nature and prevalence. Journal of Nervous McEvoy JP, Cooper TB, Chakos M, Lieberman JA. Clozapine,
and Mental Disease 1988;176:539–46. olanzapine, risperidone, and haloperidol in treatment-resistant
Lindenmayer JP, Bernstein-Hyman R, Grochowski S. A new five patients with schizophrenia and schizoaffective disorder. American
factor model of schizophrenia. Psychiatric Quarterly 1994;65:299– Journal of Psychiatry 2002;159:255–62.
322. Werner PD, Yesavage JA, Becker JMT, Brunsting DW, Isaacs JS.
Link BG, Cullen FT, Andrews H. The violent and illegal behavior of Hostile words and assaultive behavior on an acute inpatient
mental patients reconsidered. American Sociological Review psychiatric unit. Journal of Nervous and Mental Disease
1992;57:275–92. 1983;171:385–7.
Link BG, Stueve A, Phelan J. Psychotic symptoms and violent Westfall PH. Multiple testing of general contrasts using logical
behaviors: probing the components of threat/control-override constraints and correlations. Journal of the American Statistical
symptoms. Social Psychiatry and Psychiatric Epidemiol Association 1997;92:299–306.
1998;33(Suppl. 1):S55–60. Yudofsky SC, Silver JM, Jackson W, Endicott J, Williams D. The
McNiel DE, Eisner JP, Binder RL. The relationship between aggres- Overt Aggression Scale for the objective rating of verbal and
sive attributional style and violence by psychiatric patients. Journal physical aggression. American Journal of Psychiatry 1986;143:35–
of Consulting and Clinical Psychology 2003;71:399–403. 9.