Abstract
The first Norwegian tools designed to assess the violence risk of the mentally ill were developed in the late 1980s, though the first national guidelines for both violence and suicide risk assessment were not published until two decades later in 2007. This article reviews the history of the field of forensic risk assessment in Norway from its humble beginnings to the present day. First an overview is provided of the history of forensic psychiatry and the criminal justice system. The main scope, however, is to discuss current research on and practice of risk assessment of violence in Norway, with an emphasis on the development and use of Norwegian risk assessment tools and methods. Particular attention is paid to instruments that follow the structured professional judgment model of risk assessment, as actuarial tools are rarely researched and not routinely implemented in clinical practice in Norway. Finally, a brief analysis is provided of some controversies concerning risk assessment in the expert witness reports on Anders Behring Breivik, who murdered 77 persons in 2011.
There is a great variety of strategies and tools for assessing risk of violence. They may be categorized according to how information is used to reach a decision. Unstructured professional evaluation of risk is substantially a heuristic approach without any predefined structure guide for what information should be analyzed and how it should be combined into a risk decision. Anamnestic risk assessment is very similar to the unstructured approach. Although it provides some guidance, in the sense that the use of anamnestic information is emphasized, it offers no structure for how information should be obtained and combined. Structured professional judgment (SPJ) risk evaluation is guided and structured by evidence-based literature, but the overall risk decision is still based on a discretionary interpretation process. In contrast, actuarial risk assessment instruments are designed to predict a specific outcome (e.g., violence) in a specific population (e.g., psychiatric patients) within a specific time frame, with predefined score rules.
In this article, we address the practice of violence risk assessment in correctional and psychiatric settings, and in the courtroom in Norway. Norway is one of the wealthiest countries in the world per capita, so the economic basis for good practice and research ought to be optimal. We describe (a) the historical development of the forensic psychiatry and correctional systems, (b) current research and practice of risk assessment, and (c) courtroom controversies concerning violence risk assessment in the Anders Behring Breivik (ABB) case.
History of Forensic Psychiatry and Criminal Justice Systems
After a presentation of the history of forensic psychiatry and the criminal justice system, we turn to the development of the legislation that constitutes the legal framework for doing violence risk assessment in psychiatry facilities, prisons, and for the courts.
Forensic Psychiatry
The first forensic mental health hospital in Norway, the Criminal Asylum (Kriminalasylet), was established in 1895 in the city of Trondheim, which is situated in the midwestern part of Norway. It served approximately 35 patients who mostly had been both convicted of a criminal offense and diagnosed with a major mental illness, typically psychosis. The hospital was owned and operated by the Norwegian Ministry of Justice and provided services for the entire nation. The reputation of the institution was bad and life inside was miserable. In 1923, another hospital (Reitgjerdet sykehus) was opened in Trondheim, not for treating criminal offenders per se, but for civil psychiatric patients from all over Norway. The admission criterion was to be “especially difficult and dangerous,” which meant having committed serious violent acts. The hospital merged with the Criminal Asylum in 1961, but in 1980, a board of inquiry appointed by the government (with one of the authors of this article as a member) advised that the facility close due to inhumane living conditions, lack of rehabilitative efforts, and excessive use of coercion, seclusion, and restraint (Kongelig Resolusjon (Royal Decree), 1980). As a result, by the early 1980s, several counties had established medium-security units, and three high-security units were established for different health regions. Reitgjerdet sykehus provided services until 1987, and since then, there has been no national hospital serving forensic populations in Norway.
Today, intramural forensic mental health services are provided by three high-security and 17 medium-security units administrated by the Ministry of Health. Both offenders and nonoffenders are treated in these units, but the regional units have a larger proportion of offenders. Most of the units are staffed by interdisciplinary teams of psychiatrists, psychologists, nurses, occupational therapists, and social workers. Today, there are approximately 180 beds serving a population of 5 million inhabitants, yielding three to four beds per 100,000 persons in Norway. About 50% of these beds are allocated to patients who have been sentenced to mandatory treatment for serious violent crime. The remainder is for patients who have been civilly committed due to severe behavioral problems during civil psychiatric treatment. In Norway, violent acts in psychiatric institutions are rarely reported to the police. Even if they are reported, these cases mostly are not brought in for the courts “due to insufficient evidence.” When discharged to a lower level of security, nonoffender patients are treated in either civil psychiatric facilities or in the community via day/ambulatory treatment centers.
There are three regional centers for research and education in forensic psychiatry. They are located in Bergen (covering the western part of Norway), Oslo (covering the eastern and southern part), and Trondheim (covering the middle and northern part). Their main responsibility is to inspire, generate, and conduct research and to provide the field (frequently also including general psychiatry) with educational programs and clinical advice. Recently, the regional centers were merged into the SIFER network (Nasjonalt nettverk for sikkerhets-, fengsels-, og rettspsykiatri, see www.sifer.no) to enhance coordination of research and education in forensic psychiatry.
There are no mental health courts or independent review boards for discharge from the Norwegian mental health system. Such matters are dealt with in ordinary courts of law in accordance with the Mental Health Act. Persons sentenced to court-ordered psychiatric treatment have the right to apply for discharge once a year with the option to appeal if their application is declined. Regardless of whether an application is made, there is a mandatory court hearing every 3 years for these patients. For civil patients, the discharge request is first reviewed by a psychiatrist or a psychologist with responsibility for discharge decisions on the patient’s ward. If denied, the patient can appeal to the hospital’s supervisory committee, which comprises a lawyer, a doctor with no professional affiliation with the hospital, a former patient, and one or more lay persons. If the discharge request is refused again, the patient has the right to bring it to the courts.
The contemporary forensic mental health care system in Norway emphasizes treatment and recovery. Although there are exceptions, most patients are discharged from medium-security units back to the community. Unlike many other Western European nations (Priebe et al., 2006), Norway has seen an overall decline in the number of forensic beds over the past decade (Sigurjonsdottir, 2009). Even within forensic psychiatry institutions, there is less focus on security measures than in other Western countries, and close surveillance through interaction with mental health professionals is considered better than the security cameras, high outdoor walls, and barbed wire, which are more common in other nations. The nurse-to-patient ratios of 3.5:1 in medium-security and 5.3:1 in high-security units makes this approach more feasible in Norway than in most other countries. No security personnel are employed in these facilities. There is cooperation between the prison and mental health systems in cases where prison services apply for prisoners to be observed for diagnostic evaluation in forensic psychiatry units. Still, generally these services operate separately.
The Norwegian Criminal Justice System
The evolution of the Norwegian criminal justice system is similar to that of many other Western nations. Despite a low crime rate (70 per 100,000 people are incarcerated in Norway compared with 700 per 100,000 in the United States; United Nations Office on Drugs and Crime, 2009), prisons are currently under pressure to increase capacity beyond the current 3,600 beds. This demand is partially driven by the rising incarceration rate of foreigners, which increased from 8.6% in 2000 to 28.7% of prisoners in 2012 (Norwegian Correctional Services, 2013). Another contributor is the incarceration of mentally ill offenders, which, similar to other countries, has been largely driven by the deinstitutionalization movement. Indeed, between 1960 and 2004 there was a population-adjusted 74% decline in the population of mental health facilities, but a 52% increase in the prison population of Norway (Hartvig & Kjelsberg, 2009). The waiting time for inmates to receive health services is considerably shorter than for the general population and is facilitated by all correctional facilities having affiliated medical and psychiatric ambulatory units (Kjelsberg et al., 2006).
Norway also has a problem with providing sufficient treatment of insanity-excused persons who have committed moderate-severity crimes, which do not qualify them for coerced psychiatric treatment. A change in the Penal Code in 2002 made these crimes not serious enough to qualify for sentences to mandatory psychiatric treatment. Under the civil Mental Health Act, felons/patients who are found not to be criminally responsible mostly only get very brief and insufficient treatment in general psychiatry. As they have exemption from punishment they cannot be sentenced to prison or correctional control, and are by the police named “free pass insane criminals.” Initiatives have been taken to change this system, but realistic alternatives are costly.
Legislation
The current Norwegian Mental Health Act builds on the “Insanity Law” (“Sinnsykeloven”) of 1848. This law is regarded as a major change in Norwegian social history because it prescribed very radical humanistic principles for the time, and it persisted because of its emphasis on providing humane and fair treatment for patients during hospitalization. However, it also described the duty and right of society to control unacceptable behavior from persons with serious mental disorders by their involuntary admission to what was then called asylums. This law functioned with the same name and only moderate modifications for more than 100 years until a new law, “Lov om psykisk helsevern,” building on the same principles, was implemented in 1961. The new law was most recently revised in 2001 (The Mental Health Act; “Psykisk helsevernloven”). This act regulates the contemporary use of civil mandatory psychiatric treatment.
The other type of mandatory psychiatric treatment in Norway is court-ordered (the Penal Code of 1902; “Straffeloven”). To be court-ordered to involuntary care two criteria must be met: (a) the person was psychotic or in a state of automatism at the time of the offense and (b) the person committed or tried to commit a serious violent or sexual offense, and there is imminent danger for recidivism of a similar serious crime. The law also extends to less serious violent or sexual offenses if the person previously has committed a similar offense and there is imminent danger for serious reoffending. “Imminent danger” refers to danger of recurrence of a serious violent crime within a certain time frame; there is no requirement that the danger should manifest itself immediately (Jacobsen, 2004). Finally, court-ordered treatment can only be applied if community protection cannot be ensured by civil mandatory psychiatric treatment.
In cases where court-ordered treatment is an option, the court requires expert witnesses to conduct a comprehensive psychiatric examination and an evaluation of future risk of violence. Because of this, there has been a steady increase in using risk assessment tools, such as the Historical Clinical Risk Management (HCR-20: Assessing risk for violence; Webster, Douglas, Eaves, & Hart, 1997). Expert witnesses relied almost completely on unstructured risk assessment up to the turn of the millennium, but a 30% increase in the use of structured risk assessment tools occurred between 2002 and 2006 (Hartvig, 2008). Unfortunately, no current official statistics for the use of risk assessment tools are available. Protection of the community is the sole reason for being sentenced to court-ordered psychiatric treatment instead of being treated under the Mental Health Act (Jacobsen, 2004). Two other types of sentences require expert witness reports to inform the court’s decision: mandatory care for persons with severe intellectual disability (IQ < 55; Penal Code §39a) and preventive detention for criminally responsible persons (Penal Code §39c). The requirement that the person committed or tried to commit a serious violent or sexual offense with imminent danger for recidivism of a similar serious crime is identical for all three sentences (§39, §39a, and §39c).
Preventive detention in prison can only be imposed on criminally responsible persons when ordinary imprisonment is not safe enough to ensure protection of the society. In Norway the time limit in prison for murder is 21 years (30 years if the convicted person is a member of a terrorist group). However, it is an established practice to release people serving time without any violations after two thirds of their prison sentence time. The uncertainty associated with professional risk assessments of persons who are accountable for their behavior is especially emphasized and critically examined in the legislative premises of the Mental Health Act. This underscores the importance of professional competence in expert witnesses and use of adequate risk assessment procedures.
Violence Risk Assessment in Norway
Until the early 1990s, assessment of risk of violence was primarily unstructured and conducted mostly in response to requests from the courts in Norway. Even in forensic psychiatric units, risk of violence was mainly addressed as a security issue related to absconding or serious episodes of violence in the units, rather than as an integrated part of clinical practice. The development away from only relying on unstructured approaches for risk assessment represents an important change in forensic psychiatry in Norway. This development was typically dominated by the use of structured tools and approaches rooted in the SPJ tradition (e.g., Webster et al., 1997). However, the main focus is on risk assessment strategies designed and developed in Norway.
Early Developments (1990-2000)
Psychometric assessment approaches such as the Minnesota Multiphasic Personality Inventory (MMPI) and the Rorschach were still widely used in the late 1980s and early 1990s. The dominant focus in risk assessment was still on intrapsychic features of the “dangerous person.” However, this period marked the beginning of the development of structured approaches to risk assessment and their systematic evaluation in Norway. This research was conducted in forensic psychiatry units and resulted in research publications that came approximately at the same time as the first version of the HCR-20. The risk assessment tools that were developed and implemented in Norway in this period were the Scale for the Prediction of Aggression and Dangerousness in Psychotic Patients (PAD; Bjørkly, 1993) and the Brøset Violence Checklist (BVC; Linaker & Busch-Iversen, 1995).
PAD
The PAD is an SPJ tool that was developed at a medium-security unit at the Molde County Hospital. It was designed to assess the impact of interactional and situational factors on aggressive behavior (Bjørkly, 1993; Bjørkly, Havik, & Løberg, 1996). An interactional understanding means that aggression is analyzed as a vulnerability factor for behaving aggressively given various situational and interactional cues. This aspect of the PAD corresponds to the behavioral analytic approach, which advocates analysis of stimulus control factors. The PAD was developed by collecting detailed descriptions of the situational contexts in which patients showed aggressive behavior. After several revisions, a list of 30 specific situations or types of interpersonal interactions was established and grouped into seven categories. The seven main categories of situations in which there is a higher risk for aggressive behavior are physical contact (i.e., during physical contact with others), limit-setting (i.e., when patients are limit-set, corrected, refused a request), problems of communication (i.e., in response to command hallucinations and paranoid delusions), changes/readjustments (i.e., due to changes such as ward unrest and family complications), persons (i.e., from interacting with unfamiliar ward personnel and being in a one-to-one situation with a woman), high-risk contact (i.e., during opportunities for aggressive contact such as pushing others under cars and train), and drugs/stimulants (i.e., during drug intoxication or abstinence).
In the PAD assessment, the patient’s aggressive potential is indicated by the number of situations considered to be potential precipitants to aggressive behavior, the relative probability of aggressive behavior to occur within each situation, and the estimated severity of the predicted aggressive behavior. Probability and severity are each scored on a 6-point scale that ranges from 0 (no probability; no relevance) to 5 (extremely high probability; lethal violence). Thus, profiles for predicted aggression can be obtained for different situations for each patient and represent the patient’s vulnerability for behaving aggressively given various situational and interactional cues. It is then combined with other risk factors to inform a summary risk assessment within the SPJ tradition. Although it was originally developed for psychotic patients, the PAD has been found to be both reliable and predictively valid for individuals with any mental illness or intellectual disability (Bjørkly, 1993). However, because most persons with a mental disorder have a higher risk of violence during acute phases of illness than during better phases, they are assessed separately for each of these periods. In addition, patients are assessed for “within ward” (current context) and “outside ward” (future context) violence risk. The PAD’s emphasis on fluctuations of risk by phase of illness and different contexts is akin to today’s use of risk formulation and risk scenarios in structured risk assessment.
Preliminary, small-scale studies indicated that the PAD had good predictive validity (Bjørkly, 1994) and high interrater reliability (Bjørkly et al., 1996). Other research presentations focused on using the PAD for assessment of dynamic risk factors (Bjørkly, 2004) and development and implementation of risk management strategies (Bjørkly, 2005, 2007). A prospective, long-term study of the predictive validity of the PAD in forensic psychiatry will be finished in 2014. However, the basic underlying theory of patient vulnerability to aggression appears to have made a substantial impact on how assessment of violence risk is taught, understood, and practiced in Norway. This may help explain why a more rigorous actuarial approach has not yet been established.
BVC
The BVC was created as part of a study in which researchers examined daily nursing reports over a 5-year period for patients admitted to the Regional Secure Unit at Brøset, Trondheim. The scope of the investigation was to explore possible precursors to violent incidents. Within 24 hr prior to violent incidents, the six most frequently recorded warning signs or changes in behavior were confusion, irritability, boisterousness, physical threats, verbal threats, and attacks on objects (Linaker & Busch-Iversen, 1995). The BVC requires nurses to score each of the six items as present (1) or absent (0) to determine short-term risk for violence. A total score of 1 or 2 (moderate risk) indicates that preventive measures should be taken. Immediate preventive measures and activation of plans to handle violent behavior are required for scores of 3 or higher. However, because the BVC belongs to the SPJ tradition, it is a summary, nonactuarial assessment of risk for individual cases. After this seminal publication there were a substantial number of BVC publications based on research in acute psychiatry (e.g., Abderhalden et al., 2008; Almvik, Woods, & Rasmussen, 2000) and forensic psychiatry (e.g., Clarke, Brown, & Griffith, 2010) in Europe. The reported predictive validity, as measured by area under the receiver operating characteristic curve (ROC-AUC; Abderhalden et al., 2006) and survival analysis (Bjørkdahl, Olsson, & Palmstierna, 2006) for BVC were found to be comparable with SPJ and actuarial tools specially developed for more comprehensive risk assessment.
Similarities Between the PAD and BVC
Despite major differences, the BVC and the PAD have three common features. First, both belong to the SPJ tradition. Second, they both use nursing staffs’ observations of individuals’ warning signs and interactional vulnerability to inform risk assessment. Despite a mostly joint understanding that violence in the mentally ill should be framed within person–situation interactions, at that time, mainly psychiatrists and clinical psychologists conducted risk assessments. Even if they asked nurses for information, there were no structured methods to strengthen the validity of nurses’ reports concerning risk of violence. Finally, both tools were designed to identify observable changes in individual patients prior to violent acts for use in risk management interventions. Observations recorded on the BVC may be useful for developing early intervention plans, and ratings on the PAD may inform tailored exposure strategies to enhance coping in interactions that heighten risk of violence. The BVC and the PAD have helped establish the SPJ model as the dominant approach to risk assessment in Norway.
Recent Developments (around 2000-2013)
During the last decade, clinical research and practice in Norway further developed the basic approaches of the PAD and the BVC into structured risk management procedures.
Unlike the risk management section of the HCR-20, neither the BVC nor the PAD had incorporated items to assess specific factors that could mitigate risk. However, after 2000 the forensic field in Norway began to include risk management as an integrated part of risk assessment. The engine in this process was the HCR-20, which had become the leading tool for risk assessment after it was translated into Norwegian (Rasmussen, Jakobsen, & Urheim, 2002). In the beginning only some introductory courses were offered on how to use the tool and some mental health professionals even used the HCR-20 merely as a checklist. In 2004, a part-time interprofessional continuing education program in risk assessment and management was established at the university level, with the HCR-20 as an integrated part of the study. So far about 1,500 clinicians have completed both modules in the two-part, 30-credit program.
Criterion-Triggered Intervention (CtI)
The CtI was developed to use a combination of individual warning signs and situational vulnerability (similar to the PAD) as the basis for interventions to prevent violence (Bjørkly, 2004). The main aim is to use this information to develop and implement de-escalating measures for the actual person and to help him or her establish better coping strategies based on these warning signs and vulnerable interactions.
The Early Recognition Method (ERM)
This method was developed in the Netherlands about 10 years ago (Fluttert et al., 2008) and has been tested in Norway since 2008 (Eidhammer, Knutzen, & Fluttert, 2010). The main aims of the ERM are to (a) identify warning signs of violence in patients’ deteriorating behavior; (b) understand how intrapersonal, interpersonal, and environmental factors affect deterioration; and (c) cooperate with the patient in the development and implementation of intervention strategies to mitigate risk of violence. These early signs can be defined as the subjective perceptions, thoughts, and behaviors of the patient prior to the incidence of aggressive behavior. A patient’s early signs of aggression are assumed to be individual, a phenomenon referred to as “signature risk signs” (Fluttert et al., 2008). The ERM, however, is a procedure to detect and discuss these early signs with the patient. The cornerstone of the method is the development of a written protocol that describes the relevant warning signs and how the patient and nurses shall respond to them. Thus, ERM is aimed at increasing not only patients’ self-awareness but also their self-management skills to decrease the likelihood of behaving aggressively. In a validation study, the number of seclusions and aggressive incidents were compared during a treatment-as-usual period and after ERM was implemented at a forensic unit (Fluttert, Van Meijel, Nijman, Bjørkly, & Grypdonck, 2010). A significant decline in the number of seclusions and a decrease in the mean severity of inpatient aggressive incidents were observed after ERM was introduced. ERM is very similar to the CtI; however, the efficacy of the ERM has been validated more intensively. Whereas the BVC focused on assessment of warning signs for short-term risk for violence, the ERM was developed to be used for both short- and long-term risk assessments. Its clear emphasis on user involvement in planning and implementing preventive strategies is one of the assets of this risk management method.
Short-Term Assessment of Risk and Treatability (START)
The shortage of instruments for assessment and management of risk for inpatient violence between 1 and 3 months prompted the development of the START (Webster, Martin, Brink, Nicholls, & Middleton, 2004). This intermediate period is often characterized by dynamic risk factors that complicate accurate risk assessment. The START has 20 dynamic variables that are assessed as potential risk markers and as possible protective factors. They include social skills and relationships; mental state and substance abuse; and insight, plans, and treatability. The START was translated into Norwegian in 2003 (Nonstad, 2003), and a research group at the regional high-security unit in Trondheim has helped test its clinical utility and conduct other validation research. Preliminary findings from these small-scale investigations indicate good predictive validity as measured by ROC-AUC, and very good clinical utility (Nonstad et al., 2010). In another small-scale study, 43 staff members at a high-security unit reported satisfaction with the contribution of the START to interdisciplinary work, structured risk assessment, and treatment planning (Kroppan et al., 2011). A validation study of implementing the START in the correctional system is currently ongoing.
Violence Risk Screening-10 (V-RISK-10)
From 2002 to 2004 a 33-item screening checklist, the Preliminary Scheme (Hartvig, Alfarnes, Ostberg, Skjonberg, & Moger, 2006), was tested with the purpose of item reduction. The main reason for this research was a need for a brief and not too time-consuming risk assessment tool that could be used in short-term general psychiatry wards to identify patients for a more comprehensive risk evaluation, such as with the HCR-20. The BVC had been used in intramural assessment of risk, but it was not ideal for acute psychiatry wards because it was limited to rating of warning signs for a 24-hr time-of-risk frame. This led to the development of the V-RISK-10 (Hartvig et al., 2007). Investigations conducted in Norway, Sweden, and China suggest that the V-RISK-10 has predictive validity similar to that found in tests of comprehensive risk assessment instruments in prospective investigations (Hartvig, Roaldset, Moger, Østberg, & Bjørkly, 2011; Roaldset, Hartvig, & Bjørkly, 2011; Yao, Li, Arthur, Hu, & Cheng, 2012). It is currently used in many countries outside Norway.
The SAFE Pilot Project
This project (The Prospective Pilot Study of Patients Discharged from Forensic Psychiatry with Structured Risk Assessment and Management Plans; SAFE) is a naturalistic, prospective research study that aims to develop and use a software program to improve the transition of patients from forensic psychiatry to mental health services with lower security (Bjørkly, 2004). Four key areas are measured: psychopathology and general level of functioning, risk assessment, risk management strategies, and recidivism. The main focus is on changes in individual dynamic risk factors such as dissociation, hallucinations, hopelessness, insight, persecutory delusions, and proviolent attitudes. The software program generates criterion-triggered alerts based on changes in an individual patient that indicate increased risk for violence. A software program that alerts providers to changes in dynamic risk factors will aid the clinician in deciding whether to do further risk assessment, make changes in risk management strategies, or both. Patients in the project will have dynamic risk factors measured at least four times, and matched patients who follow treatment and transfer as usual will serve as controls. The project runs until 2015, and a full-scale research project will be planned if the initial results are positive.
Biological Factors
Previous research has identified an association between aggressive behavior and biological factors, such as oxytocin (Ditzen et al., 2009; Sala et al., 2011), serotonin (Siever, 2008), and cholesterol (Golomb, 1998). The most recent development in research concerning risk assessment in Norway is the expansion of the established psychosocial risk models to include biological risk markers or risk factors (Roaldset, Bakken, & Bjørkly, 2011). The main idea behind this is that biological factors may explain other aspects of violent behavior that cannot be attributed to psychosocial factors alone, and that a multifaceted biopsychosocial model may enhance risk assessment decisions both at an individual and group level. For example, impulsivity may be an aspect of aggression, and biological factors such as low total cholesterol may contribute to a better understanding and detection of risk for impulsive aggression (Conklin & Stanford, 2008; Gray, Taylor, & Snowden, 2011; Troisi, 2011; Vevera, Zukov, Morcinek, & Papezova, 2003). Preliminary findings indicate that a multifaceted model including total cholesterol is promising (Roaldset, Hartvig, Linaker, & Bjørkly, 2012). When total cholesterol levels were combined with V-RISK-10 scores, results indicated that total cholesterol had significant incremental validity. Furthermore, in multiple logistic regression analysis, total cholesterol significantly increased the explained variance for violent behavior beyond that which could be explained by V-RISK-10, age, and gender. Naturally, further research is needed to support the feasibility and predictive validity of this and other biopsychosocial models.
Violence Risk Assessment by Expert Witnesses
The minimum age of criminal responsibility is 15 years in Norway. However, because risk assessment of violence for persons between the ages of 15 and 18 years is in its early development in Norway, we demarcate our review of research and practice pertaining to evaluation of violence risk to persons older than 17 years. In Norway, expert witnesses are routinely appointed by the courts after a process in which both the prosecution and defense have been consulted. Although expert witnesses may represent only one of the parties unilaterally, this is very rare in Norway compared with other Western countries. Mental health experts have two roles in court cases: (a) to report on the criminal responsibility of the defendant and (b) to assess the risk of violent recidivism. Experts have a consultative function, but the court has the final decision-making authority.
When evaluating criminal responsibility, experts focus on establishing whether psychosis (typically an International Statistical Classification of Diseases and Related Health Problems 10th Revision [ICD]-10 psychotic disorder), automatism, or severe intellectual disability was present at the time of the index offense. This first criterion has received much attention in the world media following the 2012 trial of ABB, who murdered 77 people and gravely injured many others in 2011 (see below for a discussion of this case). During the trial, two pairs of experienced psychiatrists served as expert witnesses and provided conflicting reports. The first pair concluded that ABB had a diagnosis of paranoid schizophrenia and the second pair claimed that ABB had a narcissistic personality disorder and antisocial traits, but without comorbid psychosis. As the latter was in accordance with statements made by psychiatric teams who observed ABB during his remand prison stay, the judge found the defendant criminally responsible for his actions and sentenced him to preventive detention in prison. This verdict contains the possibility of prolonging an offender’s sentence for the rest of his or her life, but the verdict must be evaluated at regular intervals. Had ABB not been found criminally responsible, but still had been evaluated to be at high risk of future violence, he would have been sentenced to mandatory forensic psychiatric treatment, which can be extended indefinitely if annual reassessments confirm his mental state to qualify for further mandatory treatment.
When it comes to sentencing, violence risk assessments play a pivotal role in determining criminal responsibility. When the risk of recidivism for severe violent crimes is assessed to be very high, the defendant is sentenced to preventive detention. However, there are no exact criteria specified in the Penal Code for this type of sentence. During incarceration, violence risk is also assessed prior to decisions on prolonged imprisonment or release. Unstructured assessments were traditionally used. However, during the last decade regular use of structured risk assessment (typically the HCR-20) has become established in the prison service, but only for prisoners serving preventive detention.
Controversies in the ABB Reports
The fact that two pretrial psychiatric reports reached opposite conclusions about ABB’s mental state at the time of his mass murder of 77 persons in 2011 illustrates some of the controversies regarding expert witness testimony. We discuss the psychiatric evaluations based on an exploration of the reports, but our interpretations must be understood given the limitations of not including other sources of information such as verbal clarifications in the courtroom, not having access to scores on the diagnostic instruments, and above all not having met ABB.
The first evaluation (243 pages) found him to suffer from paranoid schizophrenia; the other evaluation (284 pages) from antisocial and narcissistic personality disorder. The first pair of psychiatrists did not conduct psychopathy evaluation, claiming that it would not make sense because he suffered from a lifelong and current psychosis. The second pair of psychiatrists, however, used the Psychopathy Check List (PCL-R, Hare, 1991) and concluded that he had moderate psychopathic traits. The second pair also assessed ABB with the V-RISK-10 (Hartvig et al., 2007), and his high scores prompted a comprehensive risk assessment with the HCR-20 Assessing Risk for Violence (HCR-20; Webster et al., 1997). Based on these assessments, they concluded that ABB had a very high risk of future violence, but that the risk is mitigated by further stay in the high-security prison. The first pair of experts did not carry out an HCR-20 assessment because they thought it would underestimate ABB’s risk of violence because “his dangerousness appears to be completely associated with his active psychotic symptoms” (author translation). According to this logic, the great majority of violent persons with psychosis would be omitted from risk assessment.
However, the main controversy of the reports was the sanity issue. The court requested that the expert witnesses come to a conclusion concerning ABB’s mental state before and during the mass murder. In spite of this mandate, the reports offered scant information concerning how his mental state actually affected the act at the time of the mass murder. Only eight pages (3%) of the first report and six pages (2%) of the second one covered the bomb explosion at the government building (8 killed) and the shooting at the summer camp at Utøya (69 killed). Moreover, the pages of the reports that actually addressed the violent acts mostly cover practical information concerning food, transport, and so forth. Questions in the text that addressed what his mental state was like and how it had an impact on the act of murdering innocent people are almost absent. The lack of violence-specific information is a major shortcoming because this information might have informed decision making concerning criminal responsibility and risk assessment. We do not know whether an act-centered approach (where ABB was interviewed about his cognitive and emotional responses during each murder) would have provided significant information to decide whether the violent acts were psychosis-driven or personality trait–motivated. Still, we claim that act-centered analysis must not be ignored in assessment of criminal responsibility and risk for violence.
Concluding Remarks
The SPJ tradition is very dominant and we failed to find any established regular practice or research concerning actuarial methods within forensic psychiatry in Norway. The strongest current trend is a marked emphasis on interactional and dynamic factors in risk assessment practice and research. However, in most respects, Norwegian risk assessment research and practice are very similar to that found in other countries in Europe and North America.