|
Abstract
|
Courts and legislators continue to assume psychiatrists are able to predict dangerousness, but research has shown they have no special ability to do so. In this study, two psychiatrists examined 47 new inpatient admissions to a short-term psychiatric treatment unit and predicted whether they would commit battery or demonstrate threatening or suicidal behavior within seven days. The psychiatrists were not accurate in predicting battery or suicidal behavior but had some efficacy in predicting threatening behaviors. The presence of assaultive or threatening behavior on admission, hallucinations on mental status examination, and a discharge diagnosis of mania were useful for predicting battery. A discharge diagnosis of mania was useful for predicting threatening behavior. The use of likelihood ratios to conceptualize predictive data is described.
VioLit summary:
OBJECTIVE:
The purpose of this study by Janofsky et al. was to examine the ability of psychiatrists to accurately predict violent behavior among patients.
METHODOLOGY:
The authors employed a quasi-experimental longitudinal prospective design with a non-probability sample of 47 new inpatient admissions to a short-term psychiatric facility in Baltimore over the eight weeks from November through December, 1985. Within 24 hours of admission, the patient had been interviewed, with historical, demographic and mental status examination data being collected. Immediately following the interview, one or both of the attending physicians predicted whether the patient would engage in battery or threatening or suicidal behavior over the course of the next seven days. Demographic information included age, sex, race and education. Histories involved collection of information about past violent behavior against others, suicide attempts or self-mutilation, non-assaultive crimes, illicit drug or alcohol abuse and whether the patient lived in a violent subculture. Information was also gathered as to whether assaultive, threatening or suicidal behavior was the cause of this latest admission. Examination of the patients' mental status involved information concerning the presence of hallucinations, of delusions, of thought disorder and an assessment of the immediate risk of violence. Patients were monitored by nurses 24 hours a day for seven days following their initial admissions. Nurses recorded any assaultive, threatening or suicidal behavior using the 'violent-incident instrument', and also provided a brief description of each violent act. Battery was defined as conducting an assault and making impact using the body, a weapon or an object. Threatening behavior was considered as threatening someone with words, with the body or with a weapon. Suicidal behavior included an attempt or threat of suicide, or self-mutilation. Discharge diagnoses were recorded and categorized. Analyses included the Fisher's exact test, sensitivities and specificities, likelihood ratios and positive predictive values.
FINDINGS/DISCUSSION:
Of the 47 subjects, 57% were male, 72% were white and the mean level of education was 10 years. At the time of discharge from the facility, 17% were diagnosed as schizophrenic, 21% as suffering from depression, 23% from bipolar disorder, 17% from adjustment disorder, 9% from other psychotic disorders and 13% were found to be suffering from other nonpsychotic disorders. Throughout the duration of the study, nine incidents of battery were reported, fifty incidents of threatening behavior and nine of suicidal activities. 21 patients accounted for all these behaviors. Results showed that physicians' predictions were not related to behavioral outcomes for battery and suicidal behavior, but they were moderately significantly related for threatening acts. Assaultive behavior at the time of admission, as well as the presence of hallucinations and a discharge diagnosis of bipolar disorder or mania were found to be useful for the prediction of battery and threatening behavior. The authors concluded that the physicians had no ability to predict either assaultive or suicidal behavior, although they did have some success in accurately predicting threatening behavior. However, the authors cautioned that the prediction of any behavior with low prevalence is extremely difficult, and the variation in prevalence can affect the power and utility of a test of prediction in any setting. Both battery and suicidal behavior exhibited low prevalence in the present study. As well as this factor, the authors suggested that the difficulties in prediction might be due to the voluntary nature of the admitted patients in their study. Prediction of violence among patients who had been involuntarily committed might result in better predictions than for those who had been admitted voluntarily.
AUTHORS' RECOMMENDATIONS:
The authors suggested that further research be conducted on different populations, within an uncontrolled clinical environment and in which no precautions had been taken to prevent violent behavior.
EVALUATION:
The authors present an interesting and important contribution to knowledge in the field of prediction of dangerousness. Although the sample is small and of a highly specialized nature, which precludes widespread generalization, the use of sophisticated analyses and attempts to minimize some of the methodological flaws of existing research helps to make this study a valuable basis for future research. The examination of alternative explanations for the findings is a helpful addition, although a discussion of implications of the findings for policy planning would have been appreciated. (CSPV Abstract - Copyright © 1992-2007 by the Center for the Study and Prevention of Violence, Institute of Behavioral Science, Regents of the University of Colorado)
KW - 1980s
KW - Maryland
KW - Mental Health Institution
KW - Mental Health Personnel Perceptions
KW - Mental Illness
KW - Mentally Ill Adult
KW - Mentally Ill Offender
KW - Mentally Ill Patient
KW - Violence Prediction
KW - Patient Violence
KW - Dangerousness Prediction
KW - Adult Violence
KW - Adult Offender
KW - Adult Patient
Language: en |