Media Analysis

Hospital and Community Psychiatry July 1994 Vol. 45 No. 7 669

The Influence of Social Networks and Social Support on Violence by Persons With Serious Mental Illness

Sue E. Estroff, Ph.D. Catherine Zimmer, Ph.D. William S. Lachicotte, Ph.D. Julia Benoit, M.L.S.

Objective: The authors examined the relationship between violent acts and threats by persons with serious mental illness, the size and

composition of their social net- works, and characteristics oftbe so- cial support they received. Methods: A group of 1 69 respondents with serious mental illness and 59 of

their significant others were inter- viewed using structured and semi- structuredprotocols to elicit data on demographic characteristics, clini-

cal characteristics, characteristics ofrespondents’ social networks and of the social support they received, and perceptions of threat within the social network. Data on acts and threats of vioknce by respon- dents over an 18-month period

were collected from self-reports by

respondents and significant others

Dr. Estroff is associate professor, Dr. Lachicotte is research assis- tant professor, and Ms. Benoit is social research assistant in the de- partment ofsocial medicine at the University of North Carolina at Chapel Hill, CB 7240, Chapel Hill, North Carolina, 27599. Dr. Zim- mer is assistant professor in the department of sociology and an- thropology at North Carolina State University. This paper is based on a presentation at a work- shop on interventions with vio- lent mentally ill persons in the community sponsored by the Na- tional Institute of Mental Health January 14-15, 1993, in Washing- ton, D.C.

andfvom hospital and court records.

Results: Fifty-six respondents either threatened violence or corn-

mitted a violent act during the

study period. Respondents with a

diagnosis of schizophrenia were more likely to commit vioknt acts

but were not more likely to threat-

en violence than were respondents

with other diagnoses. Respondents in larger networks, those with net-

works composedprimarily of rela- tives, and those who lived with un- relatedpersons were more likely to

threaten violence. Financial de-

pendence on family was associated with more violent threats and acts.

Respondents who perceived hos- tility from others were more likely

to engage in violent threats and

acts, and those with confused thinking were less likely to act or

threaten violence. More than half

of the targets of vioknce were re-

spondents’ relatives, particularly mothers living with a respondent.

Respondents who were violent per-

ceived their significant others as

threatening but did not perceive themselves as being threatening in return. Conclusions: The interper-

sonal and social contexts of respon-

dents and theirperceptions of these contexts are important considera- tions in assessing risk for violence

by persons with mental illness.

Mothers who live with an adult off- spring with schizophrenia may be

at increased risk for being a target

of violence. Violence by persons

with psychiatric disorders may be linked to their perceptions and ex-

perience of being threatened by

others.

It [living at home] was awful. 1 mean my mother was always drunk. She was

screaming andyelling. She never verbally

orphysically abusedme. I mean, I beat my

mother up, and that’s why people get

scared ofme. But 1 didn’t know what else

to do, and I was so angry at her. I mean,

when she got sober 1 told her one time, I

said, ‘I wish to God you’d get drunk.’

And nowgoing to Al-Anon I understand

that you’re so addicted to the chaos that

it’s easier to live with. [I told my thera-

pist] 1 don’t like it. It’s calm at home. Igo

home. There’s nobody there bothering me.

. . . 1 don’t miss the beating [from current

boyfriend]. I miss it being all mixed up

andstirredup.

This excerpt from an interview with

a psychiatric patient provides an ex- ample of the complex interpersonal and intrafamilial dynamics that

must be considered along with an in- dividual’s symptoms to understand

the precipitants and meaning of vio- lence in the lives of persons with serious mental illness. This paper ex- plores the usefulness of thinking

about violence by people with men- tal illness as interpersonal and con-

textual issues, rather than as primar- ily neurological or clinical concerns,

as Tardiff(1) has suggested. We take the view that violence

and the risk for violence are best as-

sessed by investigating what kinds of people in what kinds of situations

and social networks, with what qual- ities of social relations, and at what

phase of their lives and illnesses are likely to engage in dangerous behav-

iors toward whom (2- 4). Charac-

teristics ofthe person’s social support

system and the members of the per-

son’s social network constitute risk

factors for violence because they may create or decrease the opportunity or

need for defensive, threatening, or assaultive behavior.

Many individuals with enduring and serious psychiatric disorders live

670 July 1994 Vol. 45 No. 7 Hospital and Community Psychiatry

with or depend substantially on fam-

ily members, primarily parents and siblings (5). Child sexual abuse and

physical violence are unfortunately widespread in families in general (6). Yet researchers who have focused on family violence have seldom investi- gated the extent or nature of violence in families in which one ofthe mem- bers is seriously mentally ill. Re- searchers who study violence among mentally ill persons have rarely con- sidered it as a type of family or

domestic violence. As a result, we

lack empirical studies of violence among individuals with major psy- chiatric disorders that include rela- tives and friends as the interpersonal context within which violence oc- curs, is identified, and is responded to (7,8).

The findings reported here are from an exploratory study of social network and social support factors related to violence among a cohort of 169 seriously mentally ill persons described in detail elsewhere (9,10). We addressed three main questions. First, what were the clinical and de- mographic characteristics and the

characteristics of the social network and social support of respondents who committed violent acts or

threatened violence? Second, toward whom was the violence directed? And, third, what was the nature and quality of the relationships between

the respondents and their significant others, particularly those who were targets ofviolent acts or threats? The investigation is one of very few that

have focused on contextual factors re- lated to violence outside hospitals, used extensive data from both mdi- viduals with mental illness and their relatives, and used multivariate anal- ysis to examine both the risk for vio-

lent acts and the risk for threats of violence.

Background Social networks, social support, and vioknce. Social networks and social

support represent the number and types ofrelationships in which an in- dividual participates and delineate the potential risks or opportunities for violence. A particularly dense network composed primarily of rela- tives in which everyone knows every-

one else may provide the mentally ill

individual with few unaffiliated al-

lies in a conflict. Conversely, a less dense, loosely connected network

may “burn out” or be brittle, leaving the mentally ill individual aban-

doned, isolated, and unnoticed by others when he or she decompensates and becomes dangerous to self or

others. Beels (1 1) observed that persons

with schizophrenia have a “deficit of

initiative” and fail to pursue or main- tam relations with individuals other

than those to whom they have easy

access (“dues-free” relationships),

such as family members. Other social network researchers have reported

that most relationships of psychiatric patients are with casual acquaint-

ances or unrelated adults (12-14).

Most individuals with severe and pers istent mental illnesses never

marry. Even more rarely are they par-

ents. For example, 86 percent of the

respondents in our sample (N =130) had no partner at the time of the

study, and 61 percent (N=92) had never married. Seventy-one percent

(N= 1 1 1) had no children. In the

general population, spouses far ex- ceed any other type of relative as the

target ofviolence (6,15). Lacking

mates, individuals with severe and persistent mental illness may be

more likely to direct violence toward parents or siblings, with whom they

may have their most intimate rela- tionships.

Social support includes the affec- tive or emotional and instrumental

or helping qualities of relationships. People with major mental illnesses

are characterized as having rela- tionships that are less intimate, reci- procal, symmetrical, and durable

than those of the general population (16), due either to the pragmatic

contingencies oflife as a mentally ill

person (17) or to the interpersonal styles associated with their clinical

conditions. Nelson and associates (13) have challenged this view,

reporting considerable reciprocity in

interactions with others in the social

network ofpatients in their study.

Patients in their study also reported that transactions with friends were more supportive than those with

family members or mental health

professionals.

Persons with persistent psychiat-

nc disorders may be at increased risk

for committing violence because of

socioeconomic factors (18) and be- cause ofhow, where, and with whom they live (1 7), rather than because of

their psychiatric disorders. The corn- bination of having a major mental

illness and living in meager, stressful

circumstances may be much more

predictive ofthe characteristics of so-

cia! networks, relationships, and risk for violence than any clinical factor

alone.

Violence infamilies. The follow- ing factors are consistently related to increased risk for family violence: in-

tergenerational transmission of vio-

lence, low socioeconomic status, so-

cial and structural stress, social isola-

tion and low levels of community ernbeddedness, poor self-esteem, and

personality problems and other psy-

chopathology (1 5). These charac- teristics are common among individ-

uals with major mental illnesses and their families, making these families

at high risk for violence. In the literature on family vio-

lence, the areas that are most relevant

to our approach to studying violence

by seriously mentally ill persons are adolescent-to-parent violence (19) and elder abuse (20-22). In both

situations, mothers are the most common targets of violence, and most of the perpetrators are their

sons or spouses. However, there is lit- tie difference between the rate with

which daughters and sons are per- petrators of extreme violence toward

their mothers. Gelles and Cornell (1 9) emphasized that adolescent-to- parent violence is often accompanied

by violence among siblings. They

concluded that “the only important

consistency in the analysis of vio- lence toward parents and other forms

offamily violence is that the presence in a household of one form of vio-

lence is related to the occurrence of other types of violence.”

As for elder abuse, or violence di- rected at family members over the

age of 65 , Finkelhor and Pillemer (22) noted that the abuser is often in

a dependent position vis-#{225}-vis the abused parent. Although such de-

pendent relationships are common

among seriously mentally ill persons

Hospital and Community Psychiatry July 1994 Vol.45 No.7 671

(5), few published reports state the

age of the parent toward whom vio- lence has been directed. Straznickas

and associates (23) found that younger patients who lived at home

were more likely to attack a parent than other people, but they did not report the age of these parents.

In a recent study of family mem- bers who belong to the National Al- liance for the Mentally Ill (NAMI),

Steinwachs and associates (24) iden-

tified single parents, especially sin-

gle female parents, as “vulnerable

caregivers. “ They also reported that

more than one-third of the single

parents in the sample altered their

behavior to avoid upsetting their di- agnosed relative, that is, they feared violence from their relative. The vast

majority of the single parents in the study (86 percent) were female.

A few previously published inves- tigations have touched on the direct

and indirect, causal and preventive influences of social networks and so-

cial support on violence by persons with mental illnesses in community

settings (7,8,23,25-28). Variables examined in these studies have in- cluded family and household corn-

position and climate, employment status, relations with friends, and history of substance abuse, violence,

and psychiatric disorder both of the person with mental illness and of

those in the person’s network. Cha- otic, violent family environments in

which alcohol or substance use is

common, a history of and ongoing

conflict among family members, and a controlling atmosphere were asso-

ciated with violence by persons with mental illness in at least one of these

studies. Parents, especially mothers, were the relatives most likely to live with the mentally ill person, the pri- mary caretakers or instrumental sup-

porters, and, not surprisingly, the most frequent initiators of commit-

ment petitions for dangerous behav- ior (29).

It is clear from these studies that a family tradition ofphysical violence is associated with violence from the

mentally ill persons who are the sub-

jects ofresearch. This tradition is also predictive of violence in the general

population. Thus these studies reveal

little about the specific role of psy-

chiatric disorders in violence in such

families.

Intrafamilial hostility, criticism,

and overinvolvement are the three elements that constitute expressed emotion. Higher levels of expressed emotion in families are associated with a higher frequency of relapse, that is, rehospitalization, among per-

sons with schizophrenia (30-32). Because rehospitalization has been

used as the measure of relapse, it

would be helpful to know how many

ofthe admissions considered in those studies were precipitated by violent incidents within the family. The con-

cept of expressed emotion is prom- ising in developing a contextual ap- proach to the study of violence be- cause measures of intrafamilial hos-

tility are based on extensive observa- tion of interactions among relatives.

Methods

In this study, we compared the de- mographic and clinical character- istics, social functioning, and social networks and social supports of mdi- viduals with major psychiatric disor-

ders who did and did not commit violent acts or threatened violence.

We also examined the relationships between respondents who were vio-

lent and the persons who were the

targets of violence and between vio- lence and perception ofthreat or hos- tility in the social network. On the basis of these analyses, we sought to determine how these variables would

relate to the respondents’ engaging in violent acts and threats.

Sample and procedure. Most study participants were recruited when they were patients in four in-

patient psychiatric settings in or near Chapel Hill, North Carolina. The re- cruitment sites were two state psy- chiatric hospitals and the psychiatric units ofa university hospital and a community hospital. The respon- dents were residents offive largely suburban counties near the hospitals. Nine respondents were recruited from community mental health pro- grams in these counties.

The sample criteria reflect the aims ofour original investigation of pathways to application for and re-

ceipt of disability income. We re- cruited individuals who were early in their careers with a major psychiatric

disorder, had poor work histories,

had a total of less than one year of

psychiatric hospitalization in their

lifetime, and had never received dis-

ability income. The cohort of 169 persons with

major psychiatric disorders in the

study participated in a series ofstruc- tured and semistructured interviews

at six-month intervals over an 18- month period. In addition, we con-

ducted a one-time interview with 59

persons identified by respondents as their most significant other.

The semistructured interviews were conducted by a team offive doc-

toral- and master’s-level researchers, including the first author, all of whom had extensive clinical experi- ence. Each respondent was inter-

viewed by the same researcher at each wave of the study. The interviews were intended to elicit a broad array

ofinformation from the respondents,

including clinical characteristics, use

of mental health services, social and vocational experiences, ideas about mental illness, and the character- istics of their social networks and so- cial support. The interviews were not

explicitly intended to elicit informa-

tion about violence. Threats ofviolence and violent

acts. We collected data on reported acts or threats of violence by respon-

dents during the 18-month period from baseline to the third interview.

With the exception of financial data and ratings of socioeconomic status

(using Hollingshead and Redlich’s

two-factor index [33]), which came from the second interview, baseline data were used in the analyses. Al-

though the semistructured inter- views were not designed to elicit spe- cific information about domestic violence or other violence toward others, the topic often came up spon-

taneously. Reports of threats of vio- lence and violent acts were also oh-

tamed from hospital charts, commit- ment petitions, and court records in the counties where the respondents

resided. In addition, questions about hitting others and fighting were in-

cluded in the antisocial history scale of the self-report Psychiatric Epide- miology Research Interview (34), one ofthe instruments used in col- lecting data.

672 July 1994 Vol. 45 No. 7 Hospital and Community Psychiatry

Respondents were considered to have committed a violent act if they

had been arrested or criminally

charged and adjudicated for assault and battery, manslaughter, or mur- der, or ifthey had been committed to

psychiatric treatment due to danger to others and the commitment order

specified that the respondent hit, hit

with an object or weapon, or sexually assaulted another person or threat-

ened another person with an object or weapon.

Respondents were considered to

have made a threat ofviolence if they

had been arrested or criminally

charged and adjudicated for commu- nicating threats or for threatening

conduct. They were also considered to have made a threat of violence if they had been committed for being a

danger to others and the commit- ment order specified vague ideation

that was threatening or hostile

toward others or verbal threats to-

ward another person.

Reports of violence made by the

respondent or significant others or

listed in hospital charts, if confirmed by another source, were considered

evidence of violent acts or threats of violence. Reports ofviolence in corn-

mitment petitions and court records

did not need to be confirmed by an- other source.

Demographic variabks. We col-

lected data on respondents’ age, race,

gender, marital status, socioeconom- ic status, and work activity from hos-

pital charts and self-report.

Clinical variables. Data on re- spondents’ symptoms were collected

during the interviews using eight scales from the self-report Psychiat- nc Epidemiology Research Inter- view (PERI) (34-36), including

scales measuring perception of hos-

tility, antisocial history, confused thinking, and schizoid personality.

Scores on the PERI scale items can range from 0 to 4, with higher scores

indicating more severity. The PERI

scale items were either read to the re-

spondent, who used a response card to give the rating to the interviewer, or were completed by the respondent

while the interviewer observed. The eight-item checklist from the Brief

Psychiatric Rating Scale (37) (BPRS)

was completed by the interviewer at

the end ofeach interview. Additional

clinical information, such as respon-

dents’ diagnoses, was obtained from

a review of hospital charts. Previous violence variables. To

investigate how a history of violence

by or toward the respondent affected

subsequent violent behavior, we in- cluded three variables. First, prior

violent behavior by the respondent was measured using court records of

arrests or charges for violent crimes,

previous commitments for danger- ousness to others (from the hospital

chart), positive response to a ques- tion about prior violence on the PERI antisocial history subscale, and

self-reported violence toward anoth-

er person. Second, childhood abuse was measured by information about

childhood physical or sexual abuse from hospital charts or from self-re-

ports. Third, current physical or sex-

ual abuse or substance abuse by or mental illness in a partner, relative,

or household member were classified

as adult network problems. Social network variables. Char-

acteristics of respondents’ social net- works and social support were eli-

cited using a structured protocol de- veloped for the study. The social net-

work elicitation permitted respon- dents to identify a network of un-

limited size and drew from specific questions about important people, household members, people who helped the respondent with daily ac-

tivities, and providers of social and

mental health services. Information

about social supports came from queries regarding especially helpful

or problematic relationships and from respondents’ reports about the

frequency and nature ofcontact with their most important others.

We used the Structural Analysis

ofSocial Behavior (SASB) (38-41) to investigate the quality of relation-

ships, including perceptions of threat, between respondents and

their most significant others. The

SASB requires respondents to iden- tify a significant other and to rate both their own behavior in relation

to that other and the other’s behavior

in relation to themselves. The SASB is based on a two-axis model of rela-

tionships that includes a control- autonomy dimension and an affilia-

tion-attack dimension. We used a shortened version of the scale with

36 items that were rated by respon- dents on a scale from 0 (never, not at all) to 100 (always, perfectly), with 50 marking the border between false and true.

The scale yields four attack coeffi- cients that have been rigorously es-

tablished psychometrically. Attack 1 coefficients are ratings of the sig- nificant other in transitive action

toward the respondent (he or she acts). Attack 2 coefficients are the re- spondent’s ratings of himselfor her-

self in intransitive relation to the other (I am). Attack 3 describes the other’s intransitive response (he or she is), and attack 4 represents the re-

spondent’s transitive actions toward

the other (I act). The language about violence in

the scale is primarily metaphorical, referring to emotional aggression,

fear, and rage, rather than to physical attack. One item states that the other “Murders, kills, destroys, and leaves

me as a useless heap,” and another

describes the respondent as “boiling over with rage or fear, I try to escape, flee, or hide from them.” Coefficients

can range from -1 to 1 , with coeffi- cients closer to zero indicating more hostility and disaffiliation and those

closer to -1 indicating friendly, af- fihiative sentiments.

Analyses. Text analysis was used

to examine the interview transcripts. Bivariate and multivariate analysis

was used to model the relationships

between service use; financial, voca- tional, demographic, and clinical data; data on characteristics of social networks and social support; and re- spondents’ engaging in violent acts and threats of violence.

Results

Violent acts and threats. During the 18-month study period, 39 acts ofviolence were committed by 23 re- spondents. An object or weapon was

used in 16 violent acts (41 percent). The remainder involved hitting (19 acts) and four sexual assaults but no rapes.

Seventy-five threats of violence toward persons were made by 52 in- dividuals. More than halfofthese in-

cidents (53.3 percent) involved threats made to or about specific pen-

3 Percentages designate means of the propor-

tions ofrelatives, friends, and mental health

professionals in individual respondents’

networks.

Hospital and Community Psychiatry July 1994 Vol. 45 No. 7 673

Table 1 Characteristics of study r espondents1

Characteristic %

Violence

No threat or act

Violent threat only Violent act

Demographic variables

Race and gender

African-American men

White men

African-American women

White women Married Socioeconomic status (SES)2

SES 1

SES 2

SES 3

SES 4 SES 5

Employed Clinical variables

Diagnosis

Schizophrenia

Affective disorder Personality disorder Other psychotic disorders

Visits to mental health center

None ever Less than one a month One or two a month

Between two and five a

month More than six a month

Substance use comorbidity

Previous violence variables Violent behavior Childhood abuse Adult network problems

Social network variables Composition3

Relatives Friends Mental health professionals

Instrumental support3 Relatives

Friends Mental health professionals

Residence type Alone With family With unrelated person

Lives with spouse or partner

1 Sample size ranges from 148 to 157 for var-

ions items due to missing data. 2 Determined by the Hollingshead and Red-

lich two-factor index(33), with SES 1 being

the highest and SES 5 being the lowest class

pie, and the remainder were instan-

ces ofvague ideation regarding harm to others.

Respondents were assigned to one

of three groups on the basis of their

64 behavior during the 18-month study 2 1 � period: those who had committed no 146 reported violent acts toward a person

. (N=101, 64.3 percent), those who had made threats ofviolence but did

19.4 flot commit a violent act (N=33, 21 30.3 percent), and those who had engaged

9.7 in violent acts (N=23, 14.6 percent). 40.6 The third group was composed of

9.9 two mutually exclusive subgroups:

those who committed a violent act 12.3 only (N=4) and those who both 17.4 threatened violence and committed a � violent act (N= 19). Overall, 56 per-

213 � in the sample, or about one- 803 third (35.6 percent), engaged in. threats ofviolence, violent acts, or

both directed toward other persons

39.5 during the 18-month period. There

33. 1 were nearly twice as many threats as 17.2 acts. Five respondents (3.2 percent of 10.2 the entire sample) were criminally

charged as a result of violent mci- 41.4 dents, but only one of the charges 12.6 was for a violent act (assault and bat- 15.3 tery). Other charges involved com-

24 2 munication of threats or threatening 64 conduct.

185 Demographic and clinical char- . acteristics. Demographic and clini-

41 .4 cal characteristics of the cohort are 31.8 shown in Table 1. The mean age of 67.5 the cohort was 28.6 years. A total of

50.3 percent were female, and slight- ly under 30 percent were African

62.7 American. The mean number of pre- 28.2 vious psychiatric hospitalizations

9. 1 was 2.8, but more than 40 percent of

the sample reported no previous

� mental health center visits.

190 The sample had a mean total score . for the eight BPRS checklist items of

17.2 12.9. The sample’s mean scores on 69.5 the PERI items were 2 for confused 13.2 thinking, 1.2 for false beliefs and 13.9 perceptions, 1.8 for perception of

hostility, and 2.2 for schizoid person-

ality. These mean scores are com- parable to those reported for other

clinical samples (34). Previous violence. Although

more than two-thirds of the sample

did not commit a violent threat or act during the study period, 41.4

percent reported that they had

engaged in some type of violence in

the past, as Table 1 shows. A total of 31.8 percent reported childhood

abuse by someone in their social net- work, and 67.5 percent reported

problems with their social network

as adults, such as spouse abuse or al-

coholism of a family member. Network size, composition, and

targets. The mean network size of 1 1 .9 persons reported by respon-

dents in our sample is comparable to

that reported by respondents in other recent studies (13,14). Table 1 shows

characteristics ofthe respondents’ so-

cial networks. A large proportion of network members were relatives.

Half of the sample (N = 78) reported that more than two-thirds ofthe per-

sons in their network were relatives.

Fifteen percent ofthe sample (N=23) listed no friends in their network. A

mean of 28.2 percent of the persons in the respondents’ networks were

friends, and a mean of62.7 percent were relatives.

Many persons with enduring mental illnesses are unemployed and

financially dependent on their fami-

lies oforigin (5,24). More than 57

percent of all instrumental helpers

identified by respondents were rela-

tives. A small group of respondents

(N=21, 14 percent) said that all of their instrumental support was pro- vided by relatives. Individuals in the

study who were financially depen- dent on their families were signifi-

cantly more likely to have threatened others or behaved violently than

those who were not dependent (x2 4.14, df= 1, p=.O42).

Network size and composition and other social network variables

varied by diagnosis. Respondents

with schizophrenia reported having a mean of 10.3 people in their net- work, compared with 1 2.4 people for

respondents with affective disorder, 1 2 people for respondents with a per-

sonality disorder, and 1 5 .8 people for respondents with other psychotic

disorders. As Table 2 shows, people with schizophrenia had smaller,

more densely kin-based networks

than respondents in other diagnostic

groups and were more likely to live with relatives other than spouses.

Results from the SASB for re-

spondents grouped by diagnosis

Table 2 Characteristics of the social networks of study respondents, by diagnostic group, in percentages

Diagnostic group

Schizo- Affective Personality Other psychot-

Characteristic phrenia disorder disorder ic disorder

Network composition (N= 157)’

Relatives2 73.4 53.6 55.7 63.0

Friends3 19.2 35.1 34.0 30.6

Mental health professionals Instrumental support (N= 148)’

Relatives4

7.4

70.4

1 1.4

49.6

10.3

45.0

6.4

55.7

Friends5 13.0 28.7 35.1 27.1

Mental health professionals 16.7 2 1 .7 19.9 17.2

Liveswithpartner(N=151)

Residence type (N= 15 1)6

8.2 20.0 16.0 13.3

Alone 19.7 14.0 12.0 26.7 With family 77.1 66.0 56.0 73.3

With unrelated persons 3.3 20.0 32.0 0.0

1 Percentages designate means ofthe proportions of relatives,

sionals in individual respondents’ networks.

2F=10.91,df=3, l53,p<.OO1

3F=7.61,df=3, 153,p<.OOl

4F=7.29,df=3, l44,p<.OOl 5 F=5.80, df=3, 144, p <.001 6 %21789 df=6, p<.Ol

friends, and mental health profes-

674 July 1994 Vol. 45 No. 7 Hospital and Community Psychiatry

were as follows. Mean attack 1 coeffi-

cients were -.60 for persons with

schizophrenia, -.57 for persons with

affective disorders, -.48 for persons

with personality disorders, and -.59 for persons with other psychotic dis-

orders. Mean attack 2 coefficients were -. 58 for persons with schizo-

phrenia, -.62 for persons with affec-

tive disorders , -.4 1 for persons with

personality disorders, and -.62 for

persons with other psychotic disor-

ders. There were no significant dif- ferences between diagnostic groups

on attack 1 and attack 2 coefficients. Eighty-seven people were the tar-

gets of violence by respondents dur- ing the study. The number and pro-

portion of targets who were relatives

and the types of relatives who were targets varied from the general popu- lation (1 5). More than halfofthe tar-

gets were relatives, 78 percent were

known to the respondent, and more than one-third were listed by respon- dents as members of their social net- works.

Among the 46 relatives who were

targets, the most common were

mothers (N= 1 3, 28 percent of rela- tives who were targets), followed by

spouses (N=8, 17 percent), siblings

(N=6, 13 percent), children (N=4, 9

percent), and fathers (N=3, 7 per-

cent). These patterns reflect the pre-

dominance of relatives in the social

networks of individuals in the sam- pie, the large proportion of respon-

dents who lived with their mothers, and the comparative absence of

fathers and children from respon- dents’ households. During the study

the households of7O respondents (42

percent) included parents. The households of 67 of those respon-

dents included a mother, but the

households of only 36 respondents

included a father.

Violence among this cohort dif- fered from that in the overall popula- tion in the high number of parents,

particularly mothers, who were among the targets. Child-to-parent violence, albeit perpetrated by adult

children, seemed to be characteristic of this group. Not coincidentally, mothers also played a dominant role in the social networks ofa substantial

portion of the sample, and were the

network members who most fre- quently initiated proceedings for commitment due to respondents’

dangerousness to others. Although few respondents lived

with partners and spouses, these

partners were proportionately at nearly twice the risk ofbeing a target

of violence as were mothers. Thirty- eight percent (N=8) of coresident

partners or spouses were targets,

compared with 19.7 percent (N= 13)

of coresident mothers. This pattern coincides with that reported by Straus and Gelles (6), who note that violence committed by women is

concentrated within their nuclear

family. However, violence toward

spouses constitutes a high propor-

tion of violence committed by wo-

men in the general population, and

only two male spouses or partners were targets in this study. We

suspect that among the women we

studied, mothers and other female relatives are substituting for hus-

bands as targets of violence. As in the population at large,

most ofthe violence in this study was

committed by men and was directed toward women. Sixty percent (N=

1 2) of the violent acts were by men toward women, and only 10 percent

were by men toward men. Male-to- male threats were similarly infre-

quent (8 percent ofthreats), especial- ly within families. Men and women threatened relatives and nonrelatives

in equal proportions, but when they

engaged in violent acts, women were more likely than men to direct vio-

lence toward relatives. Seventy-five percent of violent acts by women

(N = 6 acts) were directed toward rel- atives, compared with 56 percent of the violent acts by men (N= 14 acts).

Perceptions ofthreat within the

social network. More than half of the respondents (N=85, 54 percent)

identified their mothers as the pri- mary significant other when com-

pleting the SASB, and an additional

1 1 percent (N= 17) rated parents as a

unit. Spouses were rated by 29 re- spondents, or 17 percent ofthe sam-

pie. Interviews with significant others produced 42 matched sets of ratings, including ratings by 16 in-

dividuals who became targets of vio-

lence. Results from the SASB were asso-

ciated with violence in various ways.

The mean attack 1 (-.57) and attack 2 (-.57) coefficients for the entire

sample exceeded the norms reported

Table 3 Mean ratings on the Structural Analysis ofSocial Behavior(SASB)by respondents who were and were not violent

Rated group

SASB rating’

Attack 1 Attack 2

Respondents who were violent (N= 56) Enacted violence (N=23) Threatened violence (N= 33)

Respondents who were not violent (N= 101) Significant others (N= 1 57)

Targetofviolence(N=16) Not a target ofviolence (N= 141)

-.50 -.46 -.53 -.61

-.60

-.63 -.67 -.61 -.53

�.55 -.57

1 The attack 1 coefficient reflects the other in transitive action toward the respondent; attack 2

rates the respondent in intransitive relation to the other. Ratings ofSASB subscales range from

-1 to 1 , with coefficients closer to 0 indicating more hostility and disaffiliation and those closer

to -1 indicating more friendliness and affiliation.

* Significant difference between ratings of significant others who were targets ofviolence and

significant others who were not targets (t= 1.66, df= 1 55, p< .10)

Table 4 Mean ratings on subscales of the Structural Analysis ofSocial Behavior spondents who were or were not violent and by their mothers

(SASB) by re-

Mothers

Respondents Of respon- Of respon-

Not dents who dents who Violent violent were violent

Subscale1 (N=21) (N=34) (N=20) were not violent(N=15)

Attack 1 (other acts) -.47 -.48 -.41 Attack 2 (respondent is)2 -.46 -.57 -.42

Attack 3 (other is)3 -.47 -.65 -.35

-.64 -.75

-.66 Attack 4 (respondent acts) -.68 -.73 -.73 -.76

I SASB coefficients range from -1 to 1 , with coefficients closer to 0 indicating more hostility and

disaffiliation and those closer to -1 indicating more friendliness and affiliation. The attack 1 coefficient represents the other in transitive action toward the respondent, attack 2 the respon-

dent in intransitive relation to the other, attack 3 the other in intransitive relation to the re-

spondent, and attack 4 the respondent’s transitive action toward the other. 2 Significant difference between mothers ofrespondents who were violent and mothers of respon-

dents who were not violent (t=2.16, df=33, p <.05) 3 Significant difference between mothers ofrespondents who were violent and mothers of respon-

dents who were not violent (t=2. 19, df=33, p <.05)

Hospital and Community Psychiatry July 1994 Vol. 45 No.7 675

for the scale (4 1), derived from a non-

clinical group of respondents who rated themselves and their signifi-

cant others at their worst. (Coeffi-

cients closer to zero indicate more

hostility and disaffiliation, and those

closer to -1 indicate friendly, affihia- tive sentiments.) Table 3 summarizes

baseline SASB attack coefficients for

respondents who were and were not

violent during this study and for the

targets ofviolence who were rated by the respondents. Individuals who

were violent, particularly those who

engaged in violent acts and those

who rated individuals who were tar-

gets of violence, scored well above

the norms for attack 1 and attack 2

coefficients. A striking and consistent finding

is that respondents who were violent rated their significant others as more

attacking and menacing than did re- spondents who were not violent, but

did not describe themselves as more hostile, defensive, or fearful than the

respondents who were not violent. In

other words, the respondents who were violent felt threatened and at- tacked, but they did not perceive

themselves to be more threatening or

hostile than individuals who did not

behave violently. The 56 persons in

the cohort who were violent de- scribed their significant others on

the attack 1 coefficients as more hos-

tile (-.50 versus -.61), but rated themselves as more affihiative and

less fearful or defensive than the re-

spondents who were not violent on the attack 2 coefficient (-.63 versus -. 53). They described themselves as more friendly and less hostile in demeanor than the rest of the sample

but saw their significant others as

substantially more attacking. Similarly, respondents who en-

gaged in violent acts rated their sig- nificant others as more menacing (mean attack 1 = -.46) than either re-

spondents who threatened others (mean attack 1 =-.53) or respondents

who were not violent (mean attack 1 =-.61). These differences within the group of respondents who were

violent, and between them and the

entire sample, did not reach statisti- cal significance.

However, when respondents rated significant others who eventually be- came targets of violence, the differ- ences in mean SASB coefficients be- tween the two groups were signifi- cant. The 1 6 significant others who were targets ofviolence were rated by respondents as markedly more at- tacking (mean attack 1 = -.34) than

the 141 significant others who were not targets of violence (mean attack 1= -.60)(t=1.66,df=155,p<.10). Respondents rated themselves as

more angry and wary in response to

significant others who became tar- gets ofviolence (mean attack 2= -. 55) than did respondents whose significant others did not become targets (mean attack 2=-.57), but this difference was not significant.

Further analysis ofSASB findings focused on mothers and respondents who rated each other. As Table 4

shows, the mothers of respondents who were violent rated their children

(attack 3 , he or she is) and themselves (attack 2, I am) as more intransitively hostile than the mothers of persons who were not violent. Both of these

differences were significant. Similar- ly, respondents who were violent rated themselves and their mothers as more attacking in a defensive

mode than respondents who were

Table 5 Variables contributing significantly to the odds of by polytomous logistic re’2

violent acts and threats, identified

Variable Violent acts Threats

Demographic variables

Married - 0.0001**

Employed - 0.046*

Clinical variables

Schizophrenia 28.250*** Previous hospitalization - 1.882*

Scores on Psychiatric Epidemiology Research Interview subscales

Confi.tsedthinking 0.318* 0.189*

False beliefs and perceptions 0.096* Perceivedhostiiity 7.283* 29.447***

Schizoidpersonality - #{216}.#{216}35*** Score on BniefPsychiatnic Rating Scale checklist - 1.279*

Previous violence variables Childhood abuse - 0.05 1 *

Social network variables Networksize - 1.404***

Percentage ofrelatives in network - 1.097*

Structural Analysis ofSocial Behavior ratings Attack 1 (other acts) - 55 . 1 19**

Attack 2 (respondent is) - 0.012** Living with unrelated persons - 50.249*

1 Adapted with permission from Estroffand Zimmer (10)

2 Values less than 1 indicate variables that lower odds ofviolence; values greater than 1 indicate

variables that increase the odds ofviolence. Model �2= 126.78, df=68, p <.01

676 July 1994 Vol. 45 No. 7 Hospital and Community Psychiatry

* .05<p<.10 ** .01<p<.05

*** p<.01

not violent, although these differen-

ces were not significant. These re- suits suggest that both respondents and their mothers felt threatened and hostile, but neither accused each other directly ofbeing threatening or attacking.

A second measure of perceived threat from others derives from the PERI perception of hostility scale (34). Of the five items in the scale,

one that asked whether respondents felt that people were picking fights with them had by far the strongest

correlation with engaging in a vio- lent threat or act. Other items con-

cerning perceptions of people talk- ing behind one’s back, staring at,

avoiding, or cheating the respondent had weaker correlations.

The SASB and PERI findings strongly suggest that perceived threat and hostility from significant others are linked to violence by the respondents. The respondents who were violent described focal rela-

tionships in which they experienced

significant emotional danger and

damage while perceiving themselves to be less angry, defensive, or offen- sive in response than one would ex-

pect. Their commitment to the hos- pital as dangerous to others chal- ienged this perception of themselves.

Multivariate analyses. On the basis of these descriptive findings,

we conducted multivariate logistic regression analyses to model who would engage in violent acts or

threats. The results of these analyses

are reported in Table 5 . Clinical van- ables proved to be the only sig- nificant predictors ofwho would en-

gage in violent acts. Who would en-

gage in violent threats was influ- enced by social functioning, clinical variables, previous violence, and

characteristics ofthe respondents’ so- ciai network and social support.

Individuals with a diagnosis of schizophrenia were substantially more likely than all others to commit acts of violence toward others but

were no more likely to threaten

others than people with other diag-

noses. Perceived hostility from iden- tified and unidentified others, as

measured by the PERI perception of hostility scale, substantially in-

creased the risk for violent acts, but

confused thinking and isolation, as measured by the PERI confused

thinking and schizoid personality

scales, decreased the likelihood of violent acts.

Being married and working low-

ered the odds for threatening vio- lence. Diagnosis was not related to

threats, but respondents with more

previous hospitalizations and more symptoms, as indicated by a higher

BPRS score, were more likely to make threats. Respondents’ percep-

tion ofhostiiity from others, as mea-

sured by the PER! perception of hos- tility scale and SASB attack 1 coeffi-

cient, was related to increased likeli-

hood ofthreatening others. Confused thinking and withdrawal and isola-

tion, as measured by the PERI con-

fused thinking and schizoid person- aiity scales, lowered the odds for making threats. Higher concentra-

tions of relatives in the social net-

work increased the odds for threaten-

ing others, as did increased network

size. Living with unrelated others

also increased the likelihood of

threatening behavior. Scales such as the PERI and SASB

are typically considered to measure

clinical characteristics. Here we con-

sider them also to be important mea-

sures of the quality of social relations or social support. Results of the PERT scale measuring perception of

hostility are reinforced by the SASB results. People who threatened

others described their primary sig- nificant other as attacking and quite

hostile while perceiving themselves

to be comparatively friendly toward

their significant other.

We included in the regression model factors such as demographic

and clinical characteristics and pre- vious violence variables that have

been thought to influence violence

among persons with mental illness.

We then added characteristics of re-

spondents’ social networks and social support. The social network and so-

cial support variables added sig-

nificantiy (.lO<p<.O5) to the

Hospital and Community Psychiatry July 1994 Vol. 45 No. 7 677

model chi square, raising it from 95.5 to 126.8, a difference of 31.3

(df=22). We think this statistical

difference, along with our descnip- tive findings on social network and

social support factors, provide per- suasive evidence that these factors in- crease our understanding ofhow con- text affects violence among mentally ill persons.

Discussion The findings support our contention that the interpersonal and social con-

text ofseniously mentally ill individ-

uais, the quality of their relation- ships with others, and their sub jec- tive experience or clinical condition, as both interrelated and independent

factors, are important considerations in assessing risks for violence. A con- textual approach to understanding

risk for violence considers the social network as the opportunity to en- gage in violence and considers social

support or the quality of relation-

ships as the provocation or perceived need for such behaviors.

Perhaps the most persuasive evi- dence in support of this view is the numerous indications that the re- spondents who were violent felt malice and danger from significant

others and perceived and experienced hostility in their interpersonal net- works. The results of the text analy- sis and the measures obtained using two psychological instruments con- firmed that respondents who threat- ened violence felt threatened. Indi- viduals in the study who committed

violent acts also perceived hostility from others in their networks, albeit to a lesser extent than respondents

who threatened violence. In contrast to findings from re-

cent influential studies (3,42-44), substance abuse comorbidity was not

significantly related to violence in our study. The measures of threat we used may be more direct indicators of what incites violence. We found that an individual who has a thought dis- order and is withdrawn is less likely

to be violent than someone who feels threatened.

This negative impact of primary

symptoms (measured by the PERI scales for confused thinking and schizoid personality) on the risk for violence, particularly on the risk for

violent acts, diverges from previous

findings (44) and from clinical wis- dom associating higher levels of

clinical symptoms with higher risk for violence. Our findings suggest

that violent behavior requires a cer- tam amount of organization, oppor- tunity, and perceived need to defend

oneself.

The issue of threat as it is per- ceived by persons with serious men-

tal illness deserves considerably more

attention. If individuals with mental illnesses are living in situations-in

many cases, family situations-that

they experience as threatening, re- searchers and clinicians may need to

shift their analytic and treatment ap- proaches to address these individual

circumstances. To accurately identify

high- risk situations requi res more careful assessment of how fearful and victimized individuals feel and are in

their household or social network. Several recent investigations have shown that current domestic vio-

ience (45) and prior sexual abuse are

not uncommon among psychiatric

patients, but clinicians and research- ers rarely give adequate attention to these experiences (46-49). We think

it is a mistake to categorize people as violent or to conceptualize violence

as a characteristic ofa person without

cons idening the interpersonal and

clinical processes and social contexts

that individuals experience over time (50,51).

The inverse relationship between

childhood trauma and violence in

our results implies that the timing of the threat or attack from others is

crucial. Those who were threatened and attacked in childhood were less

likely to threaten or act with violence toward others as adults. Some people

who were harmed physically and

psychologically by others may have

subsequently harmed themselves or ironically may have responded to

abuse with the symptoms and behav-

ions that resulted in their being diag-

nosed as mentally ill. It is clear from our investigation

and from those of others that moth-

ens bear considerable responsibility

for caring for relatives with mental

illness (5,23,24,29) and are at high risk of being the targets of repeated

violence by those relatives (19). All

of the mothers who were targets in

our study lived with their adult men- tally iii children. A high proportion

of the respondents lived in a house- hold with a parent, and thus the op-

portunities for violence toward a

parent were comparatively high. Households that include a mother

and adult child are at a higher risk for violence, especially if there is no

other parent present (7,8). It is not premature to propose pre-

ventive interventions for these fami-

lies. First, very few residential ser-

vices such as supported housing for

persons with disabling mental illness were available in the counties where

the study took place. The oppor- tunity and need for intrafamilial vio-

lence might be reduced ifplaces of residence where persons with mental illness felt safe were available.

Individuals who listed mental health professionals in their social

networks were less likely to be vio-

lent. This finding suggests that a

trained person who is involved enough with a seriously mentally ill

person to be considered a member of the person’s social network may be

able to deter violence and to inter-

vene if signals or precursors of vio-

lence appear. It is possible that treat- ment personnel may have been more

willing to be closely involved with

individuals who were not threaten-

ing or attacking or that respondents who were less likely to be violent

were for other reasons more likely to

be involved with treatment person- nel. Nevertheless, our findings sup-

port assigning intensive case mana- gets trained to detect and prevent violence to individuals with a diag-

nosis of schizophrenia who are iso-

lated and fearful and who live with relatives, particularly with their

mothers.

We found significant differences in the characteristics of the social

networks of persons with various di-

agnoses. In view of the strong associ- ation between schizophrenia and vio- lent acts, these distinctions should be investigated further. We need to

know much more about the compli-

cated i nteractions among clinical, social, and social network charac- tenistics in relation to violence (13,

43,44,52).

678 July 1994 Vol. 45 No. 7 Hospital and Community Psychiatry

Conclusions The most significant finding of the

study for purposes ofplanning inter-

ventions is the association between

mother-adult child coresidence and

violence. This association suggests that the parental activity of the

mother, along with her degree of

proximity to her seriously mentally

iii son or daughter, involvement in

that person’s daily living, and vul-

nerabiiity, creates the opportunity

for violence. (Fathers, even if coresi-

dent with their seriously mentally iii

adult children, are rarely reported to

be attacked or threatened.) Our find- ings suggest that risk for violence by

persons with psychiatric disorders

encompasses elements of vuiner-

ability, opportunity, mutual threat,

and dependency within their social

networks.

Acknowledgments

This work was supported by grant MH- 40314 from the National Institute of

Mental Health and by a grant from the

the MacArthur Foundation Research Network on Mental Health and the Law.

References

1. TardiffK: The current state of psychiatry

in the treatment of violent patients. Ar-

chives ofGeneral Psychiatry 49:493-503,

1992

2. Monahan j: Risk assessment of violence among the mentally disordered: generat-

ing useful knowledge. International jour-

nal of Law and Psychiatry 11:249-257,

1988

3. Monahanj: Violence and mental disorder:

perceptions and evidence. American Psy- chologist 47:511-521, 1992

4. Klassen D, O’Connor WA: Crime, inpa- tient admissions, and violence among

mental patients. International journal of Law and Psychiatry 1 1:305-3 12, 1988

5 . Lefley H: Aging parents as caregivers of

adult mentally ill children. Hospital and

Community Psychiatry 38:1063-1070,

1987

6. Straus MA, Gelles Rj: How violent are American families? Estimates from the

National Family Violence Resurvey and other studies, in Family Abuse and Its

Consequences. Edited by Hotaling G, Finkelhor D, Kirkpatrick jT, et al. New-

bury Park, Calif, Sage, 1988

7. BinderR, McNielD: Victims and families

ofviolent psychiatric patients. Bulletin of the American Academy ofPsychiatry and

theLaw 14:131-139, 1986

8. Gondolf E, Mulvey E, Lidz C: Charac-

teristics of perpetrators of family and

nonfamily assaults. Hospital and Com-

munity Psychiatry 41:191-193, 1990

9. Estroff SE, Lachicotte WS, Illingworth,

LC, et al: Everybody’s got a little mental

illness: accounts ofillness and selfamong

persons with severe, persistent mental

illness. Medical Anthropology Quarterly

5:331-369, 1991 10. Estroff SE, Zimmer C: Social networks,

social support, and violence among per-

sons with severe, persistent mental ill-

ness, in Violence and Mental Disorder:

Developments in Risk Assessment.

Edited by Monahanj, Steadman H. Chi-

cago, University ofChicago Press, 1993

1 1 . Beds C: Social support and schizo- phrenia. Schizophrenia Bulletin 7:58-

72, 1981 12. Sokolovsky j, Cohen C, Berger D, et a!:

Personal networks of ex-mental patients

in a Manhattan SRO hotel. Human Or-

ganization 37(1):5-15, 1978

13. NelsonG,HalIGB, SquireD,etal: Social

network transactions of psychiatric pa-

tients. Social Science and Medicine 34:

433-445, 1992

14. Hamilton NG, Ponzoha C, Cutler D, et

a!: Social networks and negative versus

positive symptoms of schizophrenia.

Schizophrenia Bulletin 15:625-632,

1989 15. Gelles R: Violence in the family: a review

ofthe research, in Family Violence, 2nd

ed. Edited by Gelles R. London, Sage, 1987

16. Hammer M: The role ofsocial networks

in schizophrenia, in Handbook ofStudies

on Schizophrenia, Part 2. Edited by Bur-

rows G, Norman TR, Rubinstein G.

London, Elsevier, 1986

17. EstroffSE: Making It Crazy: An Ethnog- raphy ofPsychiatric Clients in an Amer-

ican Community. Berkeley, University of

California Press, 1985

18. Cohen C: Poverty and the course of

schizophrenia. Hospital and Communi-

ty Psychiatry 44:951-957, 1993

19. Gelles R, Cornell CP: Adolescent-to-

parent violence, in Family Violence, 2nd

ed. Edited by Gelles R. London, Sage,

1987

20. Gelles R, Cornell CP: Elder abuse: the

status of current knowledge, ibid

21 . PillemerK, Frankel S: Domestic violence

against the elderly, in Violence in

America. Edited by Rosenberg M, Fen-

ley MA. New York, Oxford University

Press, 1991 22. FinkelhorD, PillemerK: Elderabuse: its

relationship to other forms of domestic

violence, in Family Abuse and Its Con-

sequences: New Directions in Research.

Edited by Hotaling GT, Finkelhor D,

Kirkpatrick jT, et al. London, Sage,

1988 23. Straznickas KA, McNielDE, Binder RL:

Violence toward family caregivers by

mentally ilirelatives. Hospital and Com- munity Psychiatry 44:385-387, 1993

24. Steinwachs D, Kasperj, Skinner E: Fam-

ily Perspectives on Meeting the Needs

for Care of Severely Mentally Ill Rela-

tives: A National Study Final Report.

Baltimore, johns Hopkins University,

Center on the Organization and Financ-

ing ofCare for the Severely Mentally Ill,

1992

25 . Hill RW: Is arson an aggressive act or a

property offence? A controlled study of

psychiatric referrals. Canadianjournal of

Psychiatry 27:648-654, 1982

26. Splane 5, Monahan j, Prestholt D, et al:

Patients’ perceptions of the family’s role

in involuntary commitment. Hospital

and Community Psychiatry 33:569-

572, 1982 27 . Klassen D, O’Connor WA: A prospective

study of predictors of violence in adult

male mental health admissions. Law and

Human Behavior 12:143-158, 1988 28. Hiday VA, Scheid-CookTL: A follow-up

of chronic patients committed to out-

patient treatment. Hospital and Com-

munity Psychiatry 40:52-59, 1989

29. CookjA: Who “mothers” the chronically

mentally ill? Family Relations 37:42-

49, 1988

30. Barrowclough C, Tarrier N: Psychosocial

interventions with families and their ef-

fects on the course of schizophrenia: a

review. Psychological Medicine 14:629-

642, 1984 3 1 . Falloon IRH: Expressed emotion: cur-

rent status. Psychological Medicine 18:

269-274, 1988 32. Vaughn CE, Snyder KS, jones 5, et a!:

Family factors in schizophrenic relapse.

Archives of General Psychiatry 41:

1169-1177, 1984 33. Hollingshead A, Redlich F: Social Class

and Mental Illness. New York, Wiley,

1958 34. Dohrenwend BP, Levav IL, Shrout PD:

Screening Scales From the Psychiatric

Epidemiology Research Interview

(PERI). Rutgers, NJ, Rutgers Univer-

sityPress, 1983 35. Dohrenwend BS, Dohrenwend BP, Link

B, et al: Social functioning of psychiatric

patients in contrast with community

cases in the general population. Archives

of General Psychiatry 40:1174-1182,

1983 36. Shrout PE, Dohrenwend B, Levav IL: A

discriminant rule for screening cases of

diverse diagnostic types: preliminary re-

suits. Journal ofConsulting and Clinical

Psychology 54:314-319, 1986

37. OverallJ, Gorham D: The Brief Psychi-

atric Rating Scale. Psychological Re-

ports 10:799-812, 1962 38. Benjamin IS: Structural analysis of social

behavior. Psychological Review 81:392-

425, 1974 39. Benjamin LS: Structural analysis of dif-

ferentiation failure. Psychiatry 42:1-23,

1979

40. Benjamin LS: Interpersonal Diagnosis

and Treatment of Personality Disorders.

New York, Guilford, 1993 41 . Benjamin LS: SASB Short Form User’s

Manual. Madison, Wisc, Intrex Interper-

sonal Institute, 1987

42. Swanson j, Holzer CE, Ganju VK, et a!:

Violence and psychiatric disorder in the

community: evidence from the ECA

Hospital and Community Psychiatry July 1994 Vol. 45 No. 7 679

studies. Hospital and Community Psy-

chiatry 41:761-770, 1990 43 . Swanson j: Alcohol abuse, mental disor-

der, and violent behavior. Alcohol,

Health, and Research World 17:123-

132, 1993 44. Link BG, Andrews H, Cullen FT: The

violent and illegal behavior of mental

patients reconsidered. American Socio- logical Review 57:275-292, 1992

45. Carmen E, Ricker PP, Mills T: Victims

of violence and psychiatric illness. Amer-

icanJournal ofPsychiatry 141:378-382, 1984

46. Jacobson A: Physical and sexual assault

histories among psychiatric outpatients.

American journal of Psychiatry 146:

755-758, 1989 47. Jacobson A, Herald C: The relevance of

childhood sexual abuse to adult psychi-

atric inpatient care. Hospital and Com-

munity Psychiatry 4 1 : 1 54-1 58, 1990

48. jacobson A, Richardson B: Assault expe-

rience of 1 00 psychiatric inpatients: cv-

idence of the need for routine inquiry.

American Journal of Psychiatry 144:

908-912, 1987 49. Rose S, Peabody C, Stratigeas B: Un-

detected abuse among intensive case management clients. Hospital and Com-

munity Psychiatry 42:499-503, 1991

50. Lindsey KP, Paul G, Mariotto MJ: Urban

psychiatric commitments: disability and

dangerous behavior of black and white

recent admissions. Hospital and Com-

munity Psychiatry 40:286-294, 1989 5 1 . Cazenave NA, Straus MA: Race, class,

network embeddedness and family vio-

lence: a search for potent support sys-

tems. Journal of Comparative Family

Studies 10:282-298, 1979 52. Holmes-Eber P. Riger 5: Hospitalization

and the composition of mental patients’

social networks. Schizophrenia Bulletin

16:157-164, 1990

Using Intensive Case Management to Reduce Violence by Mentally Ill Persons in the Community

Joel A. Dvoskin, Ph.D. Henry J. Steadman, Ph.D.

Aggressive and intensive case man-

agement and a comprehensive ar-

ray of community support services are the keys to reducing the risk of

violence by people with serious

mental illness in the community. The authors describe the elements of intensive case management for

potentially violent clients, includ-

ing use of individual case mana-

gers responsiblefor small caseloads, 24-hour availability ofcase mana-

gers, and strong linkages to agen-

cies providing mental health ser-

Dr. Dvoskin is associate commis-

sioner for forensic services with the New York State Office of

Mental Health, 44 Holland Aye- nue, Albany, New York 12229. Dr. Steadman is president of Pol-

icy Research Associates in Del- mar, New York. An earlier ver-

sion of this paper was presented

at a workshop on interventions with violent mentally ill persons

in the community sponsored by the National Institute of Mental

Health January 14-15, 1993, in Washington, D.C.

vices, substance abuse treatment,

and social services as well as to the

criminaljustice system. They sum-

marize the results of three recent

studies of intensive case manage-

mentprograms suggesting that this

intervention is effective in reduc-

ing clients’ dangerousness in the

community. They discuss cultural

and human resource issues that af-

f ect planning of intensive case

management services. intensive

case managers need to be “bound-

ary spanners” with the training,

experience, and personality to

bridge the often-broadgap between

human service and criminal jus-

tice systems.

On December 1 3, 1992, nearly one- third of the television program 60

Minutes was devoted to the case of

Larry Hogue, a 48-year-old African-

American man living in New York

City. According to the press (1-3), he annually received $36,000 in dis-

ability payments from the Depart-

ment of Veterans Affairs, but he did

not use the benefits to gain housing

or other basic necessities. Instead, he

spent his income on alcohol, man-

juana, and crack cocaine, and he was

chronically homeless. It was reported that when he was

under the influence of these substan-

ces, his behavior terrorized the entire

Upper West Side of Manhattan. He

was reported to throw garbage and

feces at passers-by, destroy property,

and light fines under automobiles or

stuff rags in their gas tanks. He was

once convicted in a jury trial of neck-

less endangerment for pushing a

young girl in front ofan oncoming

truck, which barely managed to stop without hitting her. Yet, when he

was civilly committed to inpatient

psychiatric treatment and was away

from street drugs, it was reported

that his behavior became peaceful

and even docile, and hospital admin-

istrators concluded that he should be

released.

If there are treatments available

that will reduce violence associated

with mental disorder, how can they

be delivered most effectively? How

can the Larry Hogues across the U.S.

be managed while both their rights

to liberty, due process, and least restrictive setting and the public’s

right to be safe are properly bal-

anced? This paper examines these

questions and proposes that inten-

sive case management is an effective

intervention to reduce the risk of vio-

lent behavior by mentally ill persons

in the community. Case manage-