Categorization of Aggressive Acts Committed by Chronically Assaultive State Hospital Patients

This study was conducted at Napa State Hospital (NSH), a 1,190-bed long-term-care psychiatric facility located in Napa, California. Approximately 80% of the patients there are treated under a forensic commitment. These include persons acquitted by reason of insanity, those incompetent to stand trial, and offenders with a mental disorder. Offenders with a mental disorder are prisoners transferred to the hospital at the end of their prison term because a court has determined that their mental illness played a role in their crime and that they continue to present a substantial danger to the community if released on parole. The predominant instant (or committing) offense for these patients is a violent felony. The remaining 20% of patients are treated under civil commitments (mental health conservatorship). A majority of these patients were transferred to the hospital because of unmanageable aggression in a community facility. Eighty-five percent of patients at NSH are male, and the patients have a wide variety of primary psychiatric conditions, including psychotic disorders (80%), affective disorders (9%), cognitive disorders (5%), and personality disorders (3%). The average length of stay is more than three years, which permits longitudinal observation of patients. Treatment is provided on 26 different units, with an approximate census of 45 patients per unit. This study was approved by the University of California-Davis Human Research Protection Committee and the state's Committee for the Protection of Human Subjects, and a waiver of informed consent was granted.

Special incident reports. In order to identify the study sample, we utilized data obtained from special incident reports (SIRs). Special incidents are defined as any occurrence that is potentially or actually physically or psychologically harmful to a patient or has an adverse effect on the operation of the state hospital. Hospital staff are provided training in the documenting and reporting of special incidents as part of their orientation program to the hospital and receive additional training on a yearly basis. Staff witnessing or discovering an incident are responsible for making a detailed report describing the incident soon after it occurs. The report is then filed in the interdisciplinary notes section of the patient's chart. A copy of each SIR is sent to administration for the purpose of collecting and analyzing aggregate data, which is used for quality-improvement activities. The administration has maintained a SIR electronic database since November 1998. Information contained in the database regarding incidents involving patient aggression include the date and time of the assault, a general description of the nature of the act (for example, verbal or physical aggression), the victim of the assault (staff or patient), and whether the patient was the aggressor in the incident.

Chronically assaultive patients. Past research has shown that a small percentage of patients are responsible for a majority of assaults in institutional settings and are ten times more likely than other patients to inflict serious injuries ( 30, 31, 32 ). This study focused on these recidivistic assaulters, who were defined as patients committing at least three significant acts of aggression over a one-year period in the hospital. In order to identify patients meeting this criterion, we screened the SIR electronic database over a six-year period (from June 1998 to December 2004) and identified 298 patients who committed three or more physical assaults over the course of one year. In order to derive a random sample, data for each patient was then entered into Microsoft Excel and assigned a randomly generated number with the RAND function. This function returns an evenly distributed random number greater than or equal to 0 and less than 1. By using the sort function, the random numbers that were linked to the patients were sorted in ascending order to create a list of patients with corresponding research numbers. Patients were entered into the study sequentially until data on approximately 1,000 assaults were obtained.

Narrative description of assault. Using the SIR electronic database, we collected for further investigation the specific date and time that each of the study participants committed assaults. The corresponding narrative description was then located in the patient's medical record. The first step involved reviewing the description of the aggressive act as documented in the incident report. These reports are filed in the interdisciplinary notes, which are daily records of observations and treatment provided to patients by nursing staff. After reviewing the report, we checked interdisciplinary notes written in the days preceding and following the assault in order to gain additional information regarding the assault. We often found that a review of the events leading up to the assault, the clinical management of the incident, and the statements made by an assailant after the assault was very useful in providing the level of detail required to categorize an assault when the incident report was lacking in detail. It was not uncommon for assailants to disclose their reasons for an assault hours or even days after the incident.

Assault categorization. The narrative was used to categorize assaults as impulsive, organized, or psychotic by using a modified version of a procedure for classifying aggressive acts through the use of formal records. The procedure was originally developed by Stanford and Barratt ( 29 ) by using descriptions of assaults by prisoners contained in disciplinary reports. Impulsive assaults had the following characteristics: an identifiable, immediate provocation, agitated or out-of-control behavior (for example, pacing, anger, hostility, or yelling), verbal threats, failed attempts to calm the patient, lack of concern about consequences of the act or personal safety, no obvious secondary gain or long-term motive, or expressions of remorse after the incident. Organized assaults had the following characteristics: proactive or involved some degree of planning, unprovoked or minimal provocation, a social motive (for example, asserting dominance) or an external goal, use of a weapon, preceded by little warning or came as a surprise, an absence of agitation from the patient or an appearance of being in control of his or her behavior, or denial of the assault after the event. Psychotic assaults also were unprovoked; however, the attack lacked a rational alternative motive, and the assailant cited a symptom (delusion or hallucination) as the reason for committing the assault. In addition, evidence of psychosis was present before and after the assault.

If the information contained in the narrative description of the aggressive incident was not sufficient to classify a motivation for the assault, we attempted to gain further information by reviewing additional records (for example, treatment team conferences) and through behavioral consultations (for example, a psychological investigation into reasons for assaultive behavior and treatment recommendations). Assaults without a sufficient description to determine a motivation were coded "unable to determine."

Assault subcategorization. For all assaults categorized, we attempted to further subcategorize impulsive assaults by the provocation preceding the assault. We also organized assaults by the motive and organized psychotic assaults by the symptom driving the assault.

Provocations to impulsive assaults. Impulsive assaults occurred after an adverse interpersonal interaction with either a staff member or another patient. Seven subcategories were defined: directed, denied, restrained, rage, argument, disrespected, and bothered. Interactions with a staff member leading to assault of the staff member were broken down into the two subcategories: directed, a patient assaults the staff member after the staff member directs a patient to stop an unwanted behavior or start an activity, and denied, the patient assaults the staff member after the staff member refuses a patient's request. In addition, we created two other subcategories that involved interactions with a staff member leading to assault, but the victim was not the staff member who initially provoked the patient: restrained, a staff member is assaulted in the process of placing an agitated patient into seclusion or restraints; and rage, an agitated patient assaults a bystander patient who is in the vicinity but played no role in provoking the patient.

Impulsive assaults occurring after an adverse interpersonal interaction with another patient were broken down into three subcategories: argument, a patient assaults in the context of an argument that began verbally but escalated into physical violence with the assailant initiating the violence; disrespected, a patient assaults after the victim insults the assailant (for example, name calling) or makes statements that the assailant feels are disrespectful; and bothered, a patient assaults a victim whose behavior is perceived as bothering or irritating to the assailant (for example, following too closely, invading personal space, or talking or yelling incessantly).

Motives for organized assaults. Assaults with an organized motivation were subcategorized by the motive or goal for the aggressive behavior. Six subcategories were identified: retaliation, intimidation, extortion, cold threat, racial, and sexual. In the first subcategory, retaliation, a patient feels that he or she has been wronged and thus assaults in order to get even or exact revenge—for example, to settle a score. Second, with intimidation a patient assaults in order to assert his or her dominance over another individual, for example, bullying behavior. Third, with extortion a patient uses actual or threatened force in an attempt to obtain from another patient a desired item—for example, cigarettes, food, or money—to which the patient is not entitled. Fourth, with cold threat a patient implies that he or she will harm the victim at a future point if the victim does not comply with a request—for example, a patient calmly tells a nurse to "watch your back tomorrow" after a request for benzodiazepine is refused. Fifth, with racial motivation the patient assaults because the victim is a member of a particular ethnic group. And finally, with sexual motivation a patient behaves in a sexually aggressive manner toward another—for example, grabbing or making sexually inappropriate comments.

Symptoms motivating psychotic assaults. We identified five subcategories of assault motivated by psychotic symptoms: beliefs of harm, stealing, poisoning, talking or laughing, and commanding voices. These predominantly involved assaults in which the assailant acted under the influence of a paranoid idea or misinterpretation. With belief of harm as the psychotic symptom, a patient assaults a victim acting under the belief that the victim has harmed or is planning to physically harm the patient. With stealing as the psychotic symptom, a patient assaults a victim because he or she believes that the victim stole from him or her. With belief of poisoning as the psychotic symptom, a patient assaults a victim who is believed to be poisoning him or her—for example, via food or medications. With belief of talking or laughing as the psychotic symptom, a patient assaults a victim believed to be laughing at him or her or talking about him or her in a negative way. Another subcategory, belief of commanding voices, was used when a patient displaying signs of psychotic disorganization (disordered thought and speech or responding to internal stimuli) assaulted a victim without provocation and may have reported commanding auditory hallucinations to assault.

Some assault descriptions contained sufficient information to categorize a motivation, but there was not enough to subcategorize the assault. Alternatively, some assaults did not clearly fall into a particular subcategory. These assaults were coded as unspecified.

The primary author developed the procedure described above for categorizing and subcategorizing aggressive acts in a preliminary review of 400 assaults. On the basis of the review, a training manual was created that described the criteria for impulsive, organized, and psychotic assaults; specific subcategories for each type of assault along with examples of each; and situations in which an assault is unable to be categorized or subcategorized. This manual is available from the primary author. Before raters coded assaults independently, they were trained in the procedure. An interrater reliability was calculated for 93 assaults committed by eight patients, with the primary author as the standard. The kappa statistic was used secondary to the categorical nature of the data. Good consistency was obtained for each rater, both for coding the type of assault and the subcategory ( κ =.805 and .776, respectively, for rater 1 and .789 and .729, respectively, for rater 2). Chi square analyses were performed to determine the relationship between type of assault, whether the victim was a patient or staff, and legal status of the attacker (forensic or civil). All analyses were conducted with SPSS for Windows, version 10.0.5.

Eighty-eight patients were identified as recidivistic assaulters. The sample was predominantly male (75 men, or 85%, compared with 13 women, or 15%). The mean±SD age of the sample was 41.4±10.6 years. The ethnic distribution of the sample was diverse, 38 patients were Caucasian (43%), 26 were African American (30%), 18 were Hispanic (20%), and six were Asian or of another ethnic origin (7%). As shown in Table 1, most patients (57%) were in the hospital under a forensic commitment and 43% were being treated under a civil commitment statute. For the 50 forensic patients, the instant (or committing) offense was predominantly for a violent crime; 70% were charged with assault or battery. Most patients (75%) had a primary axis I chart diagnosis of schizophrenia or schizoaffective disorder; 53% had axis II comorbidity ( Table 1 ).

The 88 chronically aggressive patients committed a total of 983 assaults, representing a mean of 11.2±6.2 assaults per patient. The narrative description was insufficient to determine a category for 144 assaults (15%), so data for 839 assaults were analyzed ( Table 2 ). Impulsive assaults were most common (451 assaults, or 54%), followed by organized assaults (247 assaults, or 29%) and psychotic assaults (141 assaults, or 17%). There was a trend for organized assaults to be more likely to be committed by forensic patients (138 assaults by forensic patients, or 56%, compared with 109 assaults by nonforensic patients, or 44%), although this finding was not significant (p<.07). Additionally, although the overall number of assaults was fairly small for this group of forensic patients, when a patient who was incompetent to stand trial committed an assault, it was more likely to be classified as organized ( χ ² =25.6, df=8, p<.001).

Of the 839 total assaults for which a categorization was determined, patients were more likely to be the victim (506 assaults, or 60%) than staff (333 assaults, or 40%). However, as shown in Figure 1, the type of assault victim was highly dependent on the type of assault ( χ ² =69.59, df=2, p<.001). Patients were more likely than staff to be targets of organized assaults (194 assaults, or 79%, compared with 53 assaults, or 21%) and psychotic assaults (100 assaults, or 71%, compared with 41 assaults, or 29%). Impulsive assaults were nearly equally likely to target staff and patients (239 assaults on staff, or 53%, compared with 212 assaults on patients, or 47%).

As shown in Table 2, the most common precipitants to an impulsive assault on staff occurred when an assailant was directed or asked to perform an activity by a staff member (26%) and when denied something he or she wanted by staff (17%); impulsive assaults on other patients most often occurred in the context of an argument (20%). The most common motive for organized assaults was retaliation or revenge for a perceived past wrong (32%). A vast majority of psychotic assaults were committed because the assailant falsely believed that the victim intended to harm him or her (43%). Other psychotic assaults were driven by similar paranoid delusional beliefs that the victim was talking about or laughing at the assailant (11%), stealing from the assailant (9%), or trying to poison the assailant (4%).