Factors that Predict Severe and Repeated Aggression in Forensic Psychiatric Inpatients

Shannon M. Bader, California Department of State Hospitals, Sacramento, California, USA
Sean E. Evans, Psychology Department, La Sierra University, Riverside, California, USA; Psychology Department, Patton State Hospital, Patton, California, USA

Institutional aggression within forensic psychiatric facilities is a common problem that has not been adequately researched. Several studies have shown that a relatively small portion of the at-risk patient population is responsible for the majority of inpatient aggression. The present study compared a sample of forensic inpatients who engaged in repeated and severe aggression towards staff or peers with a sample of non-aggressive inpatients that were matched for age, gender, legal commitment type, and hospital length of stay. The discriminant model correctly predicted 74.2% of those classified as perpetrating severe and repeated aggression and 87.1% of those classified as non-aggressive within the institution. The model accounted for 51% of the variance. The predictors included variables traditionally associated with violence risk, such as the presence of a personality disorder. However, the model also included suicide attempts and psychiatric hospitalization as a juvenile; two predictors that have not been commonly included in community violence measures. These findings identify unique predictors of highly aggressive patients who cause serious injuries to other patients and staff. The early identification and management of individuals with increased risk for severe and repeated aggression is an important consideration for triaging inpatients at risk for this type of aggression.

inpatient aggression, discriminant function, forensic, psychiatric, inpatient, severe, repeated aggressors

The demand for research on violence and aggression in psychiatric inpatients stems from the very real issue that a small portion of mentally ill inpatients pose a danger to each other and to the staff in these institutions. “The reported base rate of inpatient aggression varies considerably in the literature and is likely due to a host of factors, including the definition of violence, detection and documentation methods, and patient characteristics; notwithstanding, many studies have indicated that approximately 25–35% of inpatients exhibit violent behavior while in the hospital. Research has begun to identify some commonly appearing traits in those patients who exhibit repeated violent behavior, such as neurological impairment or organic brain disorder, suicide attempts, drug abuse, male gender, and presence of a personality disorder. Additionally, “Inpatient aggression examined within a maximum-security psychiatric hospital revealed that physical aggression was positively correlated with psychotic symptoms (e.g., hallucinations and delusional thoughts), borderline personality disorder, and required assistance in daily living activities; moreover, the age of the patient, symptoms of depression, and total scores on the Psychopathy Checklist – Revised were negatively correlated with physical aggression” (p. 111). Still, findings regarding associations between diagnosis and inpatient aggression demonstrate mixed support, especially those surrounding a diagnosis of schizophrenia. This has led to a shift towards identifying specific symptoms as correlates of violent and aggressive behavior in inpatients, as opposed to broader diagnoses.

The current study compares a group of inpatients in a maximum-security forensic psychiatric hospital “who were responsible for repeated assaults that resulted in major or severe injury to the victim” with “matched forensic inpatients who had no documented acts of aggression while hospitalized” (p. 112). The authors identified n=31 severe repetitive aggressors (SRA) including 21 males and 10 females ranging in age from 19-60 with a mean of 33.6 years. The individuals in this group were selected based on the perpetration of two or more violent actions that resulted in severe injury and were rated as a 3 or higher on a severity scale of 1-4 (as recorded in special incident paperwork completed by hospital staff). Using a matched-groups design, the researchers compared these patients with nonaggressive individuals from the same facility. “In order to control for demographic variables that may be related to violence, each identified SRA patient was matched to a comparison non- aggressive patient by age (+/-1 year), gender, type of legal commitment, and length of stay at the hospital” (p. 114).

“The clinical charts for both the SRA and comparison groups were read and cataloged by the primary authors and trained clinical psychology graduate student research assistants using a 50-item coding manual. The coding manual included relevant demographic information, factors derived from the prior literature on institutional violence (clinical diagnosis and verbalized threats) as well as static and dynamic risk factors found on commonly used community violence measures (ever married, medication adherence, antisocial attitudes, under age 25 at first violent incident). With the exception of demographic information such as race, gender, diagnosis, or legal commitment, all study variables were coded as present or absent” (p. 114).

Analyses revealed that the SRA group was significantly more likely to have a mood disorder, a personality disorder, and a neurodevelopmental disorder. The SRA group was significantly younger and was admitted to the hospital more recently, and many had a current or past psychosis diagnosis or mood disorder diagnosis. Of the 23 individuals in the SRA group diagnosed with a personality disorder, all had either Antisocial Personality Disorder or Borderline Personality Disorder. Interestingly, the SRA group was significantly more likely to have a history of psychiatric hospitalization as a juvenile and a history of suicide attempts prior to the current hospitalization. “Perhaps as interesting than these differences, were the similarities between the two groups. For example…the SRA group and the non-aggressive control sample both had similar numbers of diagnoses for psychosis, past drug or alcohol abuse, and violence before age 25” (p. 115).

A discriminant function analysis of the two groups revealed that “three predictors, a history of a personality disorder diagnosis, a history of psychiatric hospitalization as an adolescent, and a history of suicide attempts contributed most to distinguishing between those who were severe and repeatedly aggressive and those who were not” (p. 115). “The discriminant model correctly predicted 74.2% of those classified as SRA and 87.1% of those classified as non-aggressive” (p. 115).

The findings of this study are unique in that they not only identify specific factors as predictors of inpatient aggression, they compare the violent group with a non-violent group by matching pairs and controlling for variables that may have led to discrepant findings in previous research (i.e., age, gender, legal commitment type, length of stay). The ability of the three factors (presence of personality disorder, hospitalization as a juvenile, prior suicide attempts) to meaningfully predict aggressive inpatient behavior has important implications for hospital staff and practitioners. Recent support for structured professional judgment (SPJ) models of risk assessment highlight the importance of reliance on empirically validated measures alongside clinical observation. With confusion and disagreement on how to capture likelihood of inpatient aggression in forensic psychiatric settings, the factors identified in the current study might aid the clinical observations of practitioners utilizing the SPJ model. Improving accuracy when triaging patients into maximum-security facilities will allow these hospitals a greater ability to identify and manage individuals on a patient-to-patient basis. Further, the safety of staff and other patients may be better protected.

Equally as important as identifying correlates of inpatient aggression is parsing out which factors are not as highly influential as previously believed. Future research on this topic should expand on the matched-groups methodology and better control for differences between groups. “Ideally, all participants in the study, regardless of whether they were SRA’s or non-aggressive, would have had the same assessments completed; however, this study was limited to the assessments available in the medical record” (p. 117). Replications of this study should aim to better compare and contrast SRA’s and non-SRA’s in order to validate predictive factors, but also to clarify which traits, factors, and symptoms, are not necessarily strong predictors of inpatient aggression. Though settling conflicting research findings will take many more studies, clarity on this topic with provide more efficient management of forensic psychiatric inpatients and better protection of the staff that care for them.

As always, please join the discussion below if you have thoughts or comments to add!