Main findings
In our sample of 500 reports of forensic evaluations from cases of severe violent crime, schizophrenia was the diagnosis most often given when experts gave an opinion of legal insanity. This is in line with other studies, in which the proportions of insanity acquittals associated with a schizophrenia diagnosis were between 49% and 96% [
19,
24,
50‐
57]. Thus, our study supports the link between schizophrenia and legal insanity. Significantly fewer defendants with other psychotic disorders and none with substance induced psychotic disorders were considered legally insane, which underlines the importance of performing a good differential diagnostic evaluation [
3]. In Norway, self-induced psychosis by the use of drugs will rarely give legal insanity.
Forensic experts did not systematically describe symptoms of psychosis, be it their presence or absence, in their assessments of defendants’ mental state. They described more such symptoms in their assessments of MSE and considerably fewer in their assessments of MSO, which we find alarming. The incomplete symptom descriptions in experts’ assessments of MSO fuel concerns about the validity of the diagnostic process, and thus the formation of the forensic opinion. Indeed, in the evaluation of legal insanity, symptom presentation and severity are more important than the diagnosis [
32]. A diagnostic category is diverse and comprise conditions that are very different in their severity. In addition, diagnostic accuracy and interrater reliability of diagnostic considerations can be quite low [
58]. The actual symptoms that affected the defendant at the time of the crime, and how these symptoms affected his or her behavior are more important for the legal consideration than the diagnosis. Even in Norway, where the law does not demand to establish a connection between the symptoms of the illness and the actual criminal act, the experts are asked to show that symptoms of a mental illness affected the defendant to such a degree that his or her ability to have a realistic understanding of their relationship with their surroundings was significantly impaired. That opinions are formed without referring to sufficient data as explanations for the opinions, are one of the frequent errors in forensic report writing [
30].
Our results support the concern that experts do not separate the time points of examination and offense in their evaluations, but instead use information from MSE to justify their conclusions about legal insanity at the time of the offense [
18,
35]. This is also reported as a common inadequacy of forensic psychiatric investigations [
59].
Delusions are central symptoms of the psychotic conditions and diagnoses [
42,
60]. The term “delusion” entered the field of criminal responsibility when the lawyer Thomas Erskine used it to justify an insanity verdict in the trial of James Hadfield in England in 1800
4 [
61,
62]. Erskine described delusions as “the true character” and “inseparable companion” of real insanity [
4,
61]. Several studies have shown that this symptom is most closely associated with legal insanity and violence, together with hallucinations [
16,
18,
21,
23‐
25,
57,
63,
64]. In the reports we investigated, we found that experts only mentioned the presence or absence of delusions and hallucinations in their assessment of MSO in about one-quarter of the reports. This is in line with the findings from our pilot study [
45] and is cause for concern. In the study by Spencer and coworkers, almost 90% of the reports reported delusions as present at the time of the crime [
23]. This is considerably more than in our study, where we found that 50% (56 out of 112)
5 of the reports concluding with a psychosis diagnosis and with legal insanity, mentioned delusions. However, Donohue and coworkers [
21] found that 64% of reports where insanity was linked to reduced ability to conform to the law referred to delusions, while 40% of reports that linked insanity to lack of volitional control referred to delusions. This last study, with lower proportions of reports referring to delusions or other symptoms related to psychosis, suggests that our findings may be comparable to findings in reports from other countries. The studies by Spencer et al., and Donohue et al. referred to earlier, found significantly more symptoms mentioned in assessments of MSO than we did. However, the low number of studies on this topic limits comparability between samples.
Experts used the collective term “psychotic symptoms” most often to describe defendants’ MSO. This term was associated with a higher probability of legal insanity both in defendants with schizophrenia and with other psychotic disorders. The experts did not define this collective term in the reports. The construct probably included several of the PANSS items we investigated (delusions, conceptual disorganization, and hallucinations), but it is impossible for the reader of the report to know which. We find it to be of great concern that experts in Norway used this term more than the specific symptoms to describe the defendants symptomatology, as the term leaves the court with little specific information to guide their judgment. Not all symptoms of psychosis are equally important when evaluating the effects of symptoms on MSO, and we advise experts to specify the symptoms they either find to be present or not to be present, and not use a general, unspecified term.
Our most surprising and concerning finding was the difference in the mention of symptoms by diagnosis. Defendants with schizophrenia had the same probability to be considered legally insane if delusions were recorded as present as if delusions were recorded as absent or not recorded at all. Defendants diagnosed with any other psychotic disorders, on the other hand, were declared legally insane 2–6 times more often when delusions and hallucinations were recorded as present at the time of the offense. This indicates that the experts gave a more thorough symptom description when they arrived at the opinion that a defendant diagnosed with other psychotic disorders was legally insane than when they considered a defendant diagnosed with schizophrenia to be legally insane. We find this alarming in terms of the legal security of defendants. Ethical principles of fidelity requires that forensic experts clearly describe the data on which their conclusions were formed. Omitting relevant data is contrary to the integrity principle [
65]. Thus, the lack of descriptions of important data for the insanity opinion found in our sample of reports might be a violation of these ethical principles.
In most societies it is seen as a fundamental human and legal right for persons with mental deviances to be granted exemption from criminal responsibility [
4]. Nevertheless, to be assessed as a person who cannot take responsibility for one's own actions may be experienced as degrading and stigmatizing for the person in question. Much of the stigma associated with serious mental illness revolves around perceptions of dangerousness [
66]. Some argue that to prevent violent acts among persons with schizophrenia will reduce stigma, and lack of responsibility might contribute to this [
67]. However, although offenders with schizophrenia are more often considered legally insane, experts must still take into account the narrow definition of legal insanity and use the clinical condition to determine the question of sanity at the time of the offense, not rely solely on a clinical diagnosis [
4,
66,
67]. Stigma can also be inherent in societal structures [
68]. Each person deserves individual assessment and treatment by the forensic experts, both to preserve their legal and human rights and to adhere to the ethical principles in forensic evaluations [
65].
In Norway, the law only requires an assessment of the presence of a severe mental illness at the time of an offense; no connection between that illness and the act must be proven. This is often referred to as the medical or biological principle [
10,
41,
69,
70]. Some have expressed concerns that the medical principle in Norwegian legislation, with its emphasis on psychosis, might lead to an even stronger connection between diagnosis and insanity than in legislations with other principles of insanity. Our results give some support to this concern. This might possibly lead to increased stigma of persons with psychotic disorders [
24,
71]. Other sources suggest that this is a common challenge faced by forensic experts, regardless of legal principle [
1,
3]. A Swedish study of forensic experts’ work methods showed that experts tended to assess many of the aspects of responsibility based on the diagnostic category alone [
72]. Several papers and textbooks emphasize that to equate a diagnostic category with insanity is a common error in insanity evaluations [
1,
3,
16,
32]. Other studies have also found incomplete clinical descriptions in different kinds of forensic psychiatric evaluations [
14,
40].
One possible explanation for our findings may be that forensic experts are influenced by the stigmatized perception that persons with schizophrenia are generally not responsible for their actions [
71]. When this diagnosis is presented early in the evaluation process, experts may reach a forensic conclusion of legal insanity without gathering thorough evidence to demonstrate that the severity of the condition is consistent with insanity. This may be a result of judgment biases, e.g., availability bias, anchoring bias, or confirmation bias [
31,
35]. Another explanation could be that experts see people with schizophrenia as unfit to be in prison, and that the experts use the possible penal reaction as the most important justification for the insanity suggestion.
Limitations
The PANSS was developed as a research tool to monitor changes in symptoms of psychosis during treatment. We modified the use of this instrument to condense and extract written information on targeted symptoms. Although interrater reliability was good for this use of the PANSS, further use would require subsequent reliability testing. Another limitation is that the category “other psychotic disorders” is very diverse, including schizotypal disorders, paranoid psychosis, acute psychosis, affective psychoses, and SIPD.
The chosen selection criterion severity of the offense was another limitation. As our sample was selected based on indictment, some experts wrote many reports while others wrote few. Standard regression models assume independent observations. Since the experts wrote more than one report, the reports were clustered within expert teams, and the assumption of independence was violated. To compare for this limitation, an inter-cluster correlation was introduced for the pair of experts and the risk ratios were analyzed by general linear mixed models. The estimations of risk ratios were based on low cell counts in some cases, which might give non-robust estimates. Even moderate changes in cell counts might lead to considerable differences in the risk ratios.Our study was not designed to validate symptom descriptions, as we did not study the actual evaluation process. Although a validation study of the evaluation process might be of interest, we focused on the reports because they are used as evidence in courts. The interrater agreement for the symptoms in the reports was very high. Therefore, the observed differences in symptom assessment between persons with schizophrenia and those with other psychosis may be considered both reliable and valid.
Symptom descriptions might also be associated with other features, such as the length of the reports, collateral information collected, whether the defendant cooperated with the experts, and the time interval between the crime and the first interviews. We did not include these features in our analyses, which may be considered a limitation of our study.
Future research
We think it would be interesting to see similar studies performed in countries where a connection between the symptoms present and the offense committed must be proven for a defendant to be considered legally insane. This could improve understanding of how a legal principle affects clinical examinations and descriptions. The legal principle for legal insanity in Norway changed after we selected the reports for this study [
10,
12]. To replicate this study using reports written after this change may provide important insight into whether changes in the rule of law affect the clinical practice of forensic experts.
As many forensic experts in Norway also work in clinical settings, our results raise some concerns regarding how mental health personnel in general report on symptom presentations when they consider different diagnoses for their patients. We believe our findings may be relevant for the general mental health practice as well, and similar studies as ours might be conducted with written medical records from clinical mental health care settings.
Future research should analyze how the descriptive features of the reports, such as the report length, are associated with symptom descriptions. When performing research with correlational analyses, the experts that author the reports could be a selection criterion, for a better understanding of actual differences between the experts’ methods. Moreover, it would be interesting to know more about which information the judges refer to in their verdicts when presented with forensic witnesses’ reports and oral statements in court. Investigations have shown that courts usually follow the experts’ conclusions in their verdicts [
10]. The experts’ reporting of symptom descriptions from collateral sources could be studied if the researchers recorded which symptoms they found in these sources and compared them with the experts’ descriptions in the reports.