Anger and functioning amongst inpatients with schizophrenia or schizoaffective disorder living in a therapeutic community


Professor Secondo Fassino, MD, Neurosciences Department, Psychiatry Section, University of Torino, Via Cherasco 11, 10126, Torino, Italy. Email:


Aims:  This study explored the functional correlates of anger amongst therapeutic community inpatients.

Methods:  The sample consisted of 44 subjects diagnosed with schizophrenic/schizoaffective disorder who were involved in a community treatment program. Assessment involved administration of the Health of Nation Outcome Scales and the Global Assessment of Functioning as well as self-evaluations using the Social Adaptation Self-evaluation Scale. Psychopathology was assessed with the Positive and Negative Symptoms Scale. Angry feelings and coping skills were self-assessed with the State–Trait Anger Expression Inventory and the Symptom Checklist-90 Hostility Scale. Multiple regression analyses correlated anger with functioning, controlling for psychopathology.

Results:  Angry feelings related to self-harm, hyperactivity, physical problems, and to global weight independently from Positive and Negative Symptoms Scale scores. They also predicted interest and pleasure in housekeeping, quality of social relationships and relational exchanges.

Conclusions:  Results showed that angry feelings were not merely derivations of schizophrenic psychopathology; rather, they were independently related to self-damaging behaviors, to attentional demands towards the staff, to agreement to community tasks and to low quality of social relationships. Indeed, anger was related to adaptation's level in a therapeutic community setting demonstrated by subjects with psychoses and it may represent an indirect measure of their experienced quality of life. Therapeutic and management approaches to anger amongst subjects with schizophrenia are discussed.

POPULAR OPINION ASSOCIATING anger and violence with schizophrenia contributes to the stigma characterizing this disorder.1–3 Although schizophrenia represents one of the psychiatric diagnoses most frequently correlated with rage and hostility,4 some studies have demonstrated that those with schizophrenia are more likely to be victims rather than perpetrators of violent crimes5 and that anger represents a critical management issue in many settings.6,7

Anger has actually been identified as a trigger for self-damaging and hetero-aggressive behaviours8 and it may severely impede treatment and rehabilitation efforts. Anger has been shown to predict poor social integration in terms of social connections, emotional relationships and self-care habits.9–11

Despite the relevance of anger in management of patients with schizophrenia or schizoaffective disorder, some important issues, such as the relationship between anger and social functioning in schizophrenic subjects, have received scant attention. Recent studies have reported that social functioning and quality of life represent the most relevant outcome measures for rehabilitation programs. Many studies have explored relationships between these factors and psychopathology,12 personality traits,13 social adaptation9 and life conditions,14 whilst only a few have explored their relationship with anger. Moreover, current research has tended to demonstrate a bias against exploring the subjective experiences of anger amongst those with schizophrenia or schizoaffective disorder.

In fact many studies have attempted to identify predictors and correlates of anger amongst subjects with schizophrenia,11,15–18 and anger emerged as target symptom for current antipsychotic treatments.19–22 Nevertheless, anger in schizophrenic subjects has often served as a reaction to external dangers23 or violent victimizations,24 including involuntary hospital admissions,25 limit-setting by authority figures26 or poor management.27,28 Social maladjustment, however, was correlated with higher levels of hostility.29

Descending from this evidence, anger might represent a risk factor for many of the common problems affecting individuals with schizophrenia or schizoaffective disorder living in therapeutic communities, impinging on their social relationships and their functioning within their communities. When patients with schizophrenia or schizoaffective disorder are institutionalized or are residing in therapeutic communities, it becomes particularly important that staff bring in-depth knowledge about dynamics underlying the emergence and expression of anger in this population, as well as correlations between physical health and functional consequences of rage and hostility. Such knowledge may help staff in their management and it can guide treatment planning. The State–Trait Anger Expression Inventory (STAXI)30 is a self-rated instrument that measures the subject's anger as a temperament trait. STAXI analyses reactive or nonreactive anger, the frequency with which angry feelings are expressed inwards, outwards or are controlled. It also analyzes the general disposition towards angry feelings defining an accurate and articulated dimensional profile of the disposition towards anger.

This present study explored functional correlates of self-reported feelings of anger amongst subjects with schizophrenia or schizoaffective disorder living in a therapeutic community to identify the areas of functioning related to and impaired by anger, controlling for positive and negative psychotic symptoms.



The sample consisted of 61 inpatients consecutively admitted and living in three therapeutic communities of the Mental Health Department of the Local Health District 3 (ASL-TO3), in Turin, Italy, from 1 May to 30 June 2004.

Selection criteria for inclusion in the study included a diagnosis of schizophrenia (all subtypes), schizophreniform disorder or schizoaffective disorder, according to DSM-IV criteria; sub-acute or stabilized condition (subjects with current severe positive symptoms, high attention deficits or residual dementia were not able to comprehend and complete questionnaires); ability to comprehend and complete self-administered questionnaires; informed consent to participate in the research.

Amongst 61 subjects, five diagnosed with borderline personality disorder and four with bipolar disorders were excluded from the study. Two were in an acute phase of their disorder and three with residual schizophrenia were not able to comprehend and complete questionnaires. Two subjects refused to co-operate in the study for personal reasons.

The final sample consisted of 45 subjects affected by various DSM-IV-TR psychotic disorders (disorganized schizophrenia, paranoid schizophrenia, residual schizophrenia, undifferentiated schizophrenia and schizoaffective disorder). They were classified as currently in remission, even though residual symptoms were present in some cases. Patients were being treated with different antipsychotic, anxiolytic and mood stabilizing medications. The overall participation rate in the study was 75%. All subjects were informed about the aims of the study and they received adequate explanations about how to fill in close-ended question inventories. A staff member in the therapeutic community (an educator or a psychologist) was available for further support and explanations and he assisted patients during test administration. One subject was not able to answer the questions properly and he was thus excluded after completing the tests. The final sample consisted of 44 subjects who understood the test materials. Table 1 shows social and clinical characteristics and diagnostic distributions of the sample.

Table 1.  Social and clinical background characteristics of the sample
  1. GAF, Global Assessment of Functioning; PANSS, Positive and Negative Symptoms Scale; SD, standard deviation.

Age, mean (±SD)46.41 (±10.89) 
Age, range25–75 
Marital status  
 Disorganized schizophrenia920
 Paranoid schizophrenia1636
 Residual schizophrenia37
 Undifferentiated schizophrenia49
 Schizoaffective disorder1227
GAF score, mean (±SD)37 (±12.98) 
PANSS Positive Scale Score, mean (±SD)19.79 (±6.8) 
PANSS Negative Scale Score, mean (±SD)24.76 (±8.7) 
PANSS General Scale Score, mean (±SD)47.50 (±10.83) 
Total Score PANSS, mean (±SD)92.84 (±22.21) 
Duration of illness years, mean (±SD)20.47 (±11.96) 


All subjects were assessed using both self-administered and other-administered inventories.

Patients filled in the STAXI30 questionnaire for self-assessment of anger traits, and the Social Adaptation Self-evaluation Scale (SASS)31 for self-evaluation of functioning. Psychologists, blind to the self-assessed scores of anger and functioning, filled in the Italian version of the Health of Nation Outcome Scales (HoNOS Roma).32 The psychiatrist completed the Positive and Negative Syndrome Scale (PANSS)33 and the Global Assessment of Functioning (GAF).

During three successive staff meetings in each community, scores of self-administered inventories were discussed to evaluate overall consistency of self-evaluations with clinical impressions to exclude data obtained from subjects providing inconsistent self-evaluations. However, no subjects were excluded on this basis. All participants provided informed consent to participate in this study. Patient anonymity was preserved. The research was approved by the institutional review board of San Giovanni Battista Hospital of Torino, in conformity with the 1995 Declaration of Helsinki, as revised in Edinburgh in 2000.


The state-trait anger expression inventory30

STAXI consists of 44 items that are distributed across five main scales. State Anger: a 10-item stand-alone scale that measures the respondent's current feelings of anger. Trait Anger: also contains 10 items that ask questions about his or her disposition towards anger. Trait Anger related to Temperament: a subscale of Trait Anger consisting of four items that generally address disposition to express anger without provocation. Trait Anger Reactive: also a subscale of Trait Anger consisting of four items that ask about disposition to express anger when provoked. Anger Expressed Inwards: this eight-item scale measures how often respondent holds in or suppresses his or her anger. Anger Expressed Outwards: this eight-item scale measures how often respondent expresses anger to other people or objects. Anger Control: another eight-item scale that attempts to measure control of his/her anger's expression. Anger Expression Index: an experimental composite score that represents a combination of Anger Expressed Inwards, Anger Expressed Outwards, and Anger Control and essentially examines overall frequency of anger expression. All items are rated on a four-point scale between 1 and 4. Raw score totals are converted to percentile ranks and t-scores using normative tables. There are separate normative tables for male and female adolescents, adults, and college students. There are also special population normative tables. The normative sample for STAXI is an impressive 9000 subjects.30

The social adaptation self-evaluation scale31

SASS is a 20-item inventory for self-assessment of social, familial, occupational and environmental functioning. Items explore types of occupations, occupational satisfaction, interest in and quality of hobbies, quality of extra- and intra-familial relationships, respect for social rules, degree of active participation in social activities, interest in cultural information, ability to express personal opinions, relevance of physical world, ability to manage objects involved in daily life and sense of mastery.

The Health of the Nation Outcome Scales32

HoNOS Roma is the Italian version of the Health of the Nation Outcome Scale of the British Royal College of Psychiatrists Research Unit developed by Lora et al. This measure requires subjects to place 18 items on a scale in response to questions posed by an examiner to assess functioning in regard to correlates of psychotic psychopathology, physical problems, Axis I disorders, relationships with relatives and friends, autonomy, work/study activity, financial and living conditions, responsibilities pertaining to family of origin, environmental opportunities, parental helpfulness and availability, staff perception of patient's ‘global weight’ and patient's cooperativeness.

The positive and negative syndrome scale33

PANSS is a well known scale administered by an examiner to measure positive, negative and general symptoms of psychotic subjects. PANSS is composed of three scales and a total score. All psychiatrists had been adequately trained in administration of this instrument.

Data analysis

The analysis of the differences between groups with regard to anger was performed with the Kruskal–Wallis and Mann–Whitney tests.

Bivariate correlations between angry feelings [STAXI and Symptom Checklist-90 (SCL-90) anger–hostility scale] and variables related to functioning (HoNOS and SASS scales) were estimated with Spearman's rank correlation test. Bonferroni corrections were made to adjust for multiple comparisons. Although this study was explorative, we nevertheless attempted to reduce Type I errors by excluding from further analyses all HoNOS and SASS scales that did not reach a P < 0.01 level of significance in correlations with anger-related traits.

Multiple stepwise forward regression analyses were used to compare the relationship between self-rated anger measures (STAXI scales and the SCL-90 anger–hostility scale), entered as independent variables, and measures of functioning (HoNOS and SASS scales), entered as dependent variables. Sex, age and psychopathology (PANSS Positive, Negative and General scale scores) were entered into analysis on the first block as confounding variables. An alpha level of P < 0.05 was accepted for these analyses. The software used was SPSS for Windows, version 11.5 (SPSS Inc., Chicago, IL, USA).


The analysis of differences among five diagnostic subgroups (disorganized schizophrenia, paranoid schizophrenia, residual schizophrenia, undifferentiated schizophrenia and schizoaffective disorder) with self-rated measures of anger (STAXI and SCL-90) revealed non-significant differences. Therefore, the sample was not divided into diagnostic subgroups in further analyses.

A. Bivariate correlation of anger with measures of functioning

Significant correlations between feelings of anger (STAXI) and HoNOS scales (Table 2) were found for Trait Anger and Trait Reactive Anger [the latter was correlated with both Hyperactivity (r = 0.409, P < 0.007; r = 0.431, P < 0.004) and Global Weight (r = 0.504, P < 0.001; r = 0.498, P < 0.001)]; Expressed-In Anger, which was correlated with Physical problems (r = −0.467; P < 0.002); and Expressed-Out Anger, which was correlated with Self-harm (r = 0.424, P < 0.005).

Table 2.  Spearman's significant correlations between STAXI and functioning
  • Correlation is significant at the 0.05 level (two-tailed).

  • Correlation is significant at the 0.01 level (two-tailed).

  • AXCON, Anger Control; AXEX, Anger Expression Index; AXIN, Expressed-In Anger; AXOUT, Expressed-Out Anger; HoNOS, Health of Nation Outcome Scales; SASS, Social Adaptation Self-evaluation Scale; SRAB, State Anger; TRAB, Trait Anger; TRABR, Trait Reactive Anger; TRABT, Trait Temperament Anger.

 Physical problems−0.037−0.117−0.074−0.108−0.467−0.1920.004−0.295
 Depressed mood0.3320.2470.1770.2580.1290.094−0.1360.133
 Other symptoms0.1760.3240.3140.328−0.1960.281−0.3240.163
 Family relations0.1580.2580.2710.254−0.1720.260−0.3470.169
 Family help−0.134−0.095−0.144−0.020−0.3850.0360.194−0.268
 Global ‘Weight’0.2750.5040.3590.498−0.0720.349−0.0500.107
 Interest in housekeeping0.0920.012−0.1170.063−0.473−0.0740.336−0.428
 Pleasure in housekeeping0.006−0.029−0.1230.063−0.4350.0110.272−0.322
 Quality of hobbies−0.115−0.110−0.108−0.073−0.3810.021−0.029−0.142
 Quality of family relationships−0.132−0.134−0.138−0.141−0.065−0.2080.364−0.253
 Quality of social relations−0.262−0.172−0.205−0.144−0.170−0.1370.409−0.331
 Relational exchanges−0.0290.026−0.0890.144−0.429−0.0370.057−0.278
 Social activities0.024−0.105−0.109−0.009−0.275−0.1450.209−0.335
Total score−0.1540.058−0.0510.138−0.3810.0070.192−0.245

Significant correlations between feelings of anger (STAXI) and SASS scales (Table 2) were found for Expressed-In Anger, which correlated with Interest (r = −0.473, P < 0.002) and Pleasure in housekeeping (r = −0.435, P < 0.004) and with Relational exchanges (r = −0.429, P < 0.004). Anger Control was correlated with Quality of social relationships (r = 0.409, P < 0.006) and Expressed-Out Anger was correlated with Interest in housekeeping (r = −0.428, P < 0.006).

No correlations at the P < 0.01 level were found between SCL-90 Hostility Scale and measures of functioning, or between GAF and feelings of anger (STAXI); hence, these measures were not considered in further analyses.

B. Multiple linear regression analyses between anger and measures of functioning

Four scales of HoNOS (Hyperactivity, Global Weight, Physical problems and Self-harm) and four scales of SASS (Interest in housekeeping, Pleasure in housekeeping, Relational exchanges and Quality of social relationships), measuring aspects of functioning, were considered as independent variables. All STAXI scales were considered as dependent variables.

Expressed-In Anger was inversely correlated with Physical problems (P < 0.021), Interest (P < 0.003) and Pleasure in housekeeping (P < 0.017) and Relational exchanges (P < 0.003). Anger Control predicted Quality of social relationships (P < 0.018); State Anger predicted Interest in housekeeping (P < 0.011); Trait Anger predicted Hyperactivity (P < 0.004); and Trait Reactive Anger predicted thoughts or behaviors related to Self-harm (P < 0.005) and patient's Global Weight (P < 0.000; Table 3).

Table 3.  Multiple stepwise regression analysis of anger dimensions and quality of life, controlling for sex, age and psychopathology
 βR2Significance of F
  • STAXI scales.

  • PANSS scales.

  • HoNOS, Health of Nation Outcome Scales; PANSS, Positive and Negative Symptoms Scale; SASS, Social Adaptation Self-evaluation Scale; STAXI, State-Trait Anger Expression Inventory.

HoNOS predicted by STAXIand PANSS   
 Trait reactive anger0.4620.2130.005
 Positive symptoms scale0.4850.3020.000
 Trait Anger0.3980.4560.000
Physical problems   
 Expressed-In Anger−0.3780.1430.021
Global ‘weight’   
 Positive symptoms scale0.4040.2780.001
 Trait reactive anger0.5000.4840.000
SASS predicted by STAXI   
Interest in housekeeping   
 Expressed-in anger−0.4650.1520.021
 State anger0.3860.2670.003
 Anger control0.2930.3370.001
Pleasure in housekeeping   
 Expressed-in anger−0.3870.1490.017
Quality of social relations   
 Anger control0.3830.1470.018
Relational exchanges   
 Expressed-in anger−0.4640.2150.003

PANSS Positive Scale (Table 1), introduced as a confounding variable, predicted Hyperactivity (P < 0.000) and Global Weight (P < 0.002; Table 3).

Neither age nor sex predicted any measure of functioning.


Self-rated feelings of anger reported by subjects with schizophrenia or schizoaffective disorder were related to many dimensions of functioning in a therapeutic community. Angry feelings not only predicted violent and damaging behaviors directed against self and others,8 but were also related to other parameters pertaining to functioning and behavior.

In this study we expected to find that inclination to engage in thoughts or behaviors about self-harm would be predicted both by predisposition to experience reactive angry feelings and by a tendency to express these externally. This finding related a lower anger threshold to a biological disposition toward self-harming.17,18 It was also evident that anger and related self-harming behaviors in these patients often occur in reaction to external triggers.23 Adequate management of patients living in therapeutic community settings might contribute to prevention of anger-related self-harming behaviors and their consequences.26,27

Although hyperactive behaviors displayed psychopathological roots, they were also predicted by trait anger. Thus, biological dispositions amongst subjects with schizophrenia or schizoaffective disorder toward the experience of anger might share common roots with some of their psychotic symptoms. Hyperactivity seemed to bridge these two domains. Antipsychotic medications, particularly new antipsychotics, may effectively treat hyperactive behaviors in some cases. This may derive from the effects of these drugs on both positive symptoms and trait dispositions toward anger.19,21,22

We also found a positive correlation between the disposition to experience (both temperamentally and reactively) and that to express anger, and the patient's ‘weight’, as perceived by the staff. The stress experienced by staff was also related to patients' positive symptoms. Nevertheless, the disposition of patients to experience angry feelings represented an independent predictor of stress and related behaviors experienced by staff members. The latter may elicit angry reactions from patients26,29 and thus create a vicious circle. Adequate staff development should include specific skills and techniques to stop this cycle by using a non-instinctive way of reacting to expressions of anger by patients in therapeutic communities,26,27 possibly reducing stress and burnout experienced by the staff.

Negative correlation between physical symptoms and introverted feelings of anger does not support the hypotheses treating ‘somaticisation’ as the expression of anger in physical symptoms. The negative correlation between introverted anger and Interest and Pleasure in housekeeping may contribute to an original interpretation of these phenomena. Indeed, patients often react with anger following involuntary admission to inpatient treatment;24–26 low interest and pleasure in housekeeping may represent a reaction to the involuntary commitment of a patient without physical symptoms to a therapeutic community setting.25 As more physically ill subjects were more interested in housekeeping, even if this work represented a heavier burden for them, therapeutic communities might be more acceptable for subjects with relevant physical problems. This finding is also consistent with the published reports that show that the more that inpatients with schizophrenia experience anger, the less they experience interest and pleasure in self-care,10 including community housekeeping.

Patients' greater interest in the community was associated with greater anger in response to completing questionnaires. Ornstein34 has suggested exploring with patients the meanings they attach to questionnaires and identifying the origins of their anger in order to reduce it.

Anger control skills were positively related to adaptation of subjects with schizophrenia to the needs of the community and the rehabilitation program. According to the published reports,9 anger management has been found to predict better social involvement in community, and thus control over anger may contribute to a positive prognosis for patients entering a therapeutic community. Specific social skills training6 can be added to psychopharmacological treatment to develop anger management strategies prior to admission to a therapeutic community,19,21,22 particularly for those subjects who display less ability to control anger.

Poor social interactions in a milieu setting may represent both a consequence and a cause of expressed-in anger amongst subjects with schizophrenia living in a therapeutic community. According to previous research,9 internalized feelings of anger may impair relational exchanges. Exploring with the patient the meaning and the origin of such internalized anger may reduce the latter and pre-empt its interference with social adaptation.34 Conversely, difficult relational exchanges in a community setting may also elicit expressed-in anger amongst these patients. Hence, the introduction of a patient in a community setting requires particular attention to fostering social relationships.

Clinical implications

Feelings of anger represent possible prognostic factors, influencing community functioning by patients with schizophrenia or schizoaffective disorder. The assessment of anger prior to admission to a therapeutic community may support the success of the therapeutic project.

Feelings of anger may be addressed with specific drugs;19,21,22 nevertheless our findings underscore that feelings of anger experienced by schizophrenic patients living in a therapeutic community may also be related to dynamics that may be addressed in psychotherapy34–36 (e.g. specific skills-training programs).5

Physically ill subjects with schizophrenia seemed to accept community treatment more easily. The selection criteria for patients admitted to community treatment programs may account for this, while issues regarding treatment compliance of physically healthy subjects should be adequately addressed prior to admission to a community. Indeed, a superficial agreement may produce internalized anger that may impair functioning of both community and patient.25–27

Anger may represent both a relevant stressor and a possible consequence of inadequate rage management by staff. Staff should be trained to perform individualized management of schizophrenic patients working with patients to recognize what might underlie their feelings of anger;26,27,34–36 such an approach might also reduce the stress experienced by staff.37

Unexpectedly, hostility (measured by SCL-90) was not correlated with patients' functioning in a community setting. Introverted and expressed-out angry feelings were not exclusively predictors of aggressive behaviors. In fact they played important roles in determining adaptation to a therapeutic community, in terms of both community tasks and social relationships, and they represented indirect measures of quality of life during the therapeutic program.

Limitations and perspectives

The small sample used in this exploratory study requires that this research be replicated. In addition, the exclusive use of a therapeutic community may reduce the generalizability of these results to other treatment settings. Moreover, a distinction between psychotic subtypes may produce diagnosis-specific results, and a control group in a similar setting might be helpful to validate the interpretation of these results. Different study designs are needed to address causal directions characterizing the association between anger and community functioning. Moreover, angry feelings may become targets for experimental drugs and new milieu management styles, specifically addressing anger-related issues arising amongst schizophrenic inpatients living in therapeutic communities.


The authors wish to thank Dr. Carla Barile, MD, and Dr. Maddalena Dotta, PhD, for helping with the assessment of patients; and Dr. Cinzia Sobrero for her help with editing the manuscript.