Assessment of causal associations between illness and criminal acts in those who are acquitted by reason of insanity

Christian Aarup Skeie, Kirsten Rasmussen About the authors

The trial of Anders Behring Breivik after the terrorist attack on Utøya island and in the Government Quarter on 22 July 2011 reignited a 150-year old debate on the provision in the Norwegian penal code that regulates impunity by reason of insanity. During the trial, the question of whether the perpetrator was legally responsible was raised, and a forensic psychiatry commission was appointed to assess his mental condition. In the first of two forensic psychiatry reports, Breivik was deemed to have been psychotic at the time of his actions, and the prosecuting authority therefore made a request for transfer to compulsory mental health care.

This gave rise to vehement reactions from a number of political actors, victims of the attack, their next of kin and the media, as well as the general public. How could a person who for years had painstakingly planned and then perpetrated this crime in an apparently stone-cold, calculating manner, while having a clear comprehension of the grossly criminal nature of his acts, be cleared of legal responsibility for his actions?

The criticism voiced in the public sphere in the wake of the first forensic psychiatry report seemed to question the justification of an assessment of soundness of mind based exclusively on a medical diagnosis, and in which no position was taken with regard to the perpetrator’s understanding of his acts and the motives behind them. The controversy over whether legal responsibility should be determined on the basis of the medical or the psychological principle is far from new, however (1).

On the basis of the medical principle, which is applied in Norway, a person is exempt from penalty if he/she was psychotic at the time of the act, irrespective of whether there is a causal relationship between the psychosis and the criminal act. In judicial practice that follows from the psychological principle, on the other hand, such a relationship is required for the defendant to be found not legally responsible (2, 3).

According to some critical voices (4), the 22 July trial has brought to light weaknesses in Norwegian forensic psychiatry. As a result, changes have been called for (5). Is it appropriate to continue applying the medical principle, or is it about time to consider alternative strategies?

As a contribution to this debate, this study investigates a sample of Norwegian criminal proceedings in which the perpetrator has been declared insane when the crime was committed. We have examined in how many of these rulings there appears to be a causal relationship between the perpetrator’s illness and the criminal act with which he is charged (as is required when the psychological principle is applied) and in how many cases no such relationship appears to exist.

Material and method

The data have been retrieved from the Lovdata database (6) and comprise selected rulings by the courts of appeal. The Lovdata database contains anonymised rulings from courts of law, including the courts of appeal, from all over the country. A trial in a court of appeal presupposes that a ruling has been made by a district court. The prosecutor and the defendant may both submit a ruling from the district court to the court of appeal.

As a rule, cases will be appealed if there is disagreement regarding the application of Section 44 of the General Civil Penal Code (7). In this article we are using final rulings by the courts of appeal, since these will constitute a legal precedent for the assessment of such cases.


Two different searches were made in the period from August to September 2012. In the first, searches were made for rulings containing references to Section 39 of the General Civil Penal Code in the summary. This section describes the special sanctions of compulsory mental health care, compulsory care and preventive detention. This search returned 286 hits. These were subsequently reviewed in light of the inclusion criteria (Box 1, Box 2), which left us with 72 rulings.


Inclusion criteria, based on the application of law in the ruling and level of detail in the legal text. The case would be included if:

  • the defendant was deemed psychotic and unfit to plead pursuant to Section 44 of the General Civil Penal Code.

  • clinically relevant information on the disorder that granted impunity (e.g. ICD-10 diagnosis and/or symptomatology) was provided and

  • information on the charges in question, so that

  • the ruling was deemed to contain sufficient information to assess whether a causal relationship existed.


Common reasons for exclusion were:

  • Section 44, second paragraph (mental retardation) or third paragraph (unconsciousness)

  • Section 44 had been considered, but not allowed (resulting in imprisonment)

  • Section 44 in the Execution of Sentences Act – not the General Civil Penal Code

  • Insufficient descriptions of clinically relevant information (e.g. cases that focused on procedural matters, application of the law etc.

  • Cases that had already been coded after a previous search

The second search involved rulings containing references to Section 44 of the General Civil Penal Code in the summary. This section states that persons who were psychotic, unconscious or mentally retarded to a high degree at the time of their crime shall not be liable to punishment. This search returned 98 hits. Of these, 11 had already been coded in the first search, since Sections 39 and 44 were both referred to in the summary.

After a review of the remaining cases in light of the inclusion criteria, three cases were left. The reason why searches for Section 39 returned far more hits than searches for Section 44 was that in the majority of their rulings, the courts had emphasised their consideration of special sanctions, and not merely whether the criteria for impunity on grounds of psychosis had been fulfilled.

We thus included 75 out of a total of 373 rulings that had been made in the period from February 1994 to May 2012. The sample includes initial rulings, appeals against rulings and appeals against extensions of special sanctions.

Scoring of the sample

The rulings follow no standardised pattern in terms of their scope or level of detail. Prior to the review of the rulings we prepared a list of variables that included demographic issues, addiction, illness and symptoms, in addition to the breach(es) of the law.

The texts were carefully reviewed by the first author with regard to a possible causal relationship between the illness and the acts committed. Examples of cases where such a causal relationship was assumed to exist were provided by rulings that included such statements as «the experts are of the opinion that the murder of D. must be understood in the context of the observee’s psychotic state of mind», or «the homicidal act was carried out in an especially brutal manner, showing that when in a psychotic state of mind, the defendant may suffer severe delusions that can lead to serious crimes of violence».

Examples of cases where no causal relationship was judged to be in evidence were provided by rulings that included such statements as «moreover, the Court of Appeal finds it to be proven that this was not an accident or a chance occurrence, but that it was a deliberate and wilful act on the part of the defendant, since he was furious with the victim and wanted to stop him. He was completely aware of stabbing the victim» and «consistent with the views of the experts, the Court of Appeal considers that the acts committed are unrelated to the defendant’s mental disorder…»

A randomised sample of 20 rulings was scored by both authors independently of each other to judge the inter-rater reliability of the assessment of a causal relationship between the disorder and the acts committed.


The sample consisted of 75 rulings from the courts of appeal in which the defendant was declared not legally responsible. Altogether 72 (96 %) of the defendants were men and three (4 %) were women. Their average age at the time of crime was 33.7 years (standard deviation 9.8 years, range 17 – 72 years).

The majority (45; 60 %) had no previous criminal record, whereas 27 (36 %) had previous convictions. In three cases (4 %) the available information was insufficient to determine whether the defendant had any previous criminal record or not.

The rulings refer to use of alcohol in the period immediately preceding the time of the act in nine cases (12 %), use of cannabis in one case (1 %) and use of amphetamine in four cases (5 %). However, this information is based on reported use, and the rulings state that this has not been confirmed by objective tests.

The vast majority of the cases involved the use of violence, occasionally of a very serious nature (Table 1). The number of charges exceeds the total number of cases, meaning that many of the rulings refer to multiple charges.

Table 1  Charges (n = 137) among the defendants in the sample (n = 75) and assessment of the causal relationship between the illness and the criminal act. The majority faced multiple charges


Total number of charges

Charges with a causal relationship between the illness and the criminal act




Violent crime






Attempted homicide



Crimes against property



Possession/sale of drugs






Possession of firearms



Attempted rape






Sexual offence



As regards the defendants, in 43 of the cases (57 %) a previous history of serious mental illness (schizophrenia-type disorders, other psychotic and affective disorders) had been documented, while in 21 cases (28 %) there was no such previous history. In 11 cases (15 %) the information provided was insufficient to establish this. Approximately one-half (n = 38; 51 %) had previously been in contact with mental health services, while 24 (32 %) had no such previous contact. In 13 cases (17 %) the available information was insufficient to determine this with any certainty.

The most frequently occurring diagnoses were under ICD-10 category codes F20 – 29 (8), schizophrenia, schizotypal and delusional disorders (in 48 persons in total, 64 %). Affective disorders were found in four persons (5 %), organic diseases in two (3 %) and adult personality and behavioural disorders in five of the cases (7 %). Disorders caused by intake of psychoactive substances were described as the defendant’s primary diagnosis in three rulings (4 %), while the remaining 13 rulings (17 %) did not specify any diagnosis.

Correlation between the illness and the act

The included court rulings were carefully assessed with regard to whether a causal relationship existed between the defendant’s mental health and the crime in question. Of these, there were 58 cases (77 %) where such a relationship had been considered to exist (hereafter referred to as Group 1), while in 17 of the rulings (23 %) no such relationship was deemed to exist (hereafter referred to as Group 2). A randomised sample of 20 rulings were scored by the authors independently of each other to measure reliability, resulting in a kappa score of 0.783.

Of the 48 defendants who had a diagnosis under ICD-10 category codes F20 – 29, a causal relationship between their illness and actions was deemed to exist in 41 (85 %), while in seven cases (15 %) no such relationship was considered to exist. There is a marked difference between these groups. With regard to the other diagnostic categories, the differences between the groups were less marked.

Larger differences between the groups emerged in the prevalence of charges. There was a higher frequency of charges of threats and violence (violent crimes and homicides, although not attempted homicide) in Group 1. The most obvious difference was found in charges of homicide, whereof only one of a total of 26 charges was assumed to have no causal relationship to the defendant’s illness. Moreover, there was a larger prevalence of crimes against property, sexual offences, rape, arson and drug-related crimes in Group 2. The differences were most prominent with regard to charges of possession and sale of drugs and attempted rape.


The investigation showed that in 17 out of 75 cases no causal relationship between the illness and the criminal act was deemed to exist, which may mean that the medical principle results in impunity in a considerable number of cases in which the perpetrator’s mental health has had no effect on the criminal act. The absence of a causal relationship between psychotic disorders and criminal acts was found also by Taylor in 1985 (9). She found that 43 % of a sample of psychotic perpetrators ascribed their criminal acts to psychotically motivated causes.

Homicides were clearly more likely to be deemed psychotically motivated than to be considered to have no causal relationship to the defendant’s psychosis. This finding is consistent with a number of studies that indicate an increased risk of violence for psychoses in general and a correlation between psychosis and homicide in particular (10 – 12). A major review article by Taylor from 2008 (13) concludes that in the most seriously violent persons with psychosis, the violent behaviour is most often motivated by the psychotic symptoms.

Group 2 also had a larger proportion of charges for possession and use of drugs than Group 1. This finding is also consistent with Taylor’s study (9), in which only 29 % of the less serious crimes were deemed to be related to psychotic symptoms (delusions) in the perpetrator. A similar tendency was found for crimes against property, in that such crimes figured more frequently in Group 2.

As can be seen, the causal relationship between the illness and the acts is elucidated by investigating the charges, rather than the perpetrator. This choice is justified by reference to the focus on the correlation between the illness and the act. It is the defendant’s state of mind at the time of the act that will be considered in determining legal insanity. However, several defendants had been charged with multiple felonies, and some of the rulings failed to state clearly whether separate psychiatric assessments had been undertaken for the different times of the criminal acts. This uncertainty represents a weakness in our study.

At the same time, this illustrates a highly relevant topic in forensic psychiatry, since a psychotic criminal may commit numerous felonies for which he is exempt from penalty, although not all these acts will necessarily be motivated by delusions or in other ways result from his disorder.

Our study presumes that it is possible to assess whether a causal relationship between a psychotic state of mind and a criminal act exists or not. However, the authors of the public reports that form the basis for current legal practice – that follow the medical principle (1, 14, 15) – have questioned whether this is feasible. The public reports state that a psychosis that renders the defendant legally insane affects the entire personality, and there will thus always be a risk that the act is a result of the mental disorder. Excluding a causal relationship between the psychosis and the act will be fraught with a great deal of uncertainty, and this will entail a considerable risk of a wrongful conviction (16).

In addition, our study is based on a material that has not been produced for the purpose of drawing conclusions regarding a possible causal relationship between disorders and acts. A statement by a judge that the act was «deliberate and wilful» is not in itself evidence that the act was not influenced by the psychosis. A retrospective assessment of whether a perpetrator had a clear understanding of his acts or whether these were linked to his mental disorder must of necessity be a subjective judgement. In making such an assessment there will normally be no objective observations to rely on for support.

It deserves mention, though, that the measurement of inter-rater reliability in the assessment of correlations in the present sample revealed a marked consistency. The consistency was not perfect, however, since there was some disagreement between the raters.

The investigated data set consisted of 75 cases and encompassed rulings made over a period of 18 years (1994 – 2012). Cases that did not fulfil the inclusion criteria regarding application of the law or the level of detail in the legal text were excluded – altogether 298 rulings. The sample size is relatively small in relation to the number of defendants who are declared unfit to plead on the grounds of psychosis pursuant to Section 44 of the General Civil Penal Code. During the period of 2007 – 11 alone, there were 366 expert reports that declared a defendant to be psychotic (17).

The reason why only a minority of these cases were available for inclusion in the present study is most likely that a majority of the less serious cases involving Section 44 are dismissed by the prosecuting authority (18). As a result, the sample represents the most serious cases in which a defendant has been found legally insane. This implies that the representativeness of the sample is limited, provided that the purpose of the study had been to generalise the findings to all criminals who are found not guilty by reason of insanity.

Another weakness that deserves mention is that the empirical basis of the study has been retrieved from a dynamic online database which is constantly changing, and this means that the search cannot be identically replicated.

Despite these weaknesses, many of the study’s findings are consistent with those from scientific inquiries that have used a more suitable methodology (9, 13). The Norwegian criteria that are applied for declaring a person legally insane have been widely discussed on a theoretical basis. However, only few empirical studies are available regarding the characteristics of those who in fact have been found not guilty by reason of insanity.

In our opinion, it would be useful to continue undertaking more thorough studies on this topic. It would also be desirable to explore whether practices for declaring defendants legally insane have changed over time, since our material is too small to draw any conclusions in this respect.

Anbefalte artikler