|Home | About | Journals | Submit | Contact Us | Français|
Schizophrenia is associated with an increase in the risk of both homicide and suicide. The objectives of this study were to systematically review all published articles that examined the relation between neurocognitive deficits and suicidal or homicidal behaviours in schizophrenia, and to identify vulnerabilities in suicidal and homicidal behaviour that may share a common pathway in schizophrenia.
A systematic review of the literature was performed using MEDLINE to include all studies published up to August 31, 2012.
Among the 1760 studies, 7 neuropsychological and 12 brain imaging studies met the selection criteria and were included in the final analysis. The neuropsychological and functional neuroimaging studies were inconclusive. The structural imaging studies reported various alterations in patients with schizophrenia and a history of homicidal behaviour, including: reduced inferior frontal and temporal cortices, increased mediodorsal white matter, and increased amygdala volumes. Patients with a history of suicidal acts showed volumetric reductions in left orbitofrontal and superior temporal cortices, while right amygdala volume was increased, though, these findings have rarely been replicated. Finally, no study has directly compared neurocognitive markers of suicidal and homicidal risk.
These results suggest that brain alterations, in addition to those associated with schizophrenia, may predispose some patients to a higher risk of homicide or suicide in particular circumstances. Moreover, some of these alterations may be shared between homicidal and suicidal patients. However, owing to several limitations, including the small number of available studies, no firm conclusions can be drawn and further investigations are necessary.
La schizophrénie est associée à un risque accru d’homicide et de suicide. Cette étude avait pour objectifs de mener une revue systématique de tous les articles publiés qui ont examiné la relation entre les déficiences neurocognitives et les comportements suicidaires ou homicides dans la schizophrénie, et d’identifier les vulnérabilités du comportement suicidaire et homicide qui peuvent partager une trajectoire commune dans la schizophrénie.
Une recherche systématique de la littérature a été effectuée dans MEDLINE pour inclure toutes les études publiées jusqu’au 31 août 2012.
Parmi les 1760 études, 7 études neuropsychologiques et 12 études d’imagerie cérébrale satisfaisaient aux critères de sélection et ont été incluses dans l’analyse finale. Les études de neuroimagerie fonctionnelle et neuropsychologiques n’étaient pas concluantes. Les études d’imagerie structurelle rapportaient des altérations variées chez les patients souffrant de schizophrénie et ayant des antécédents de comportement homicide, notamment : des cortex frontal et temporal inférieurs réduits, une matière blanche médiodorsale accrue, et des volumes d’amygdale plus larges. Les patients ayant des antécédents de gestes suicidaires présentaient des réductions volumétriques dans le cortex orbitofrontal gauche et le cortex temporal supérieur, alors que le volume de l’amygdale droite était accru, bien que ces résultats aient rarement été reproduits. Enfin, aucune étude n’a comparé directement les marqueurs neurocognitifs du risque de suicide et d’homicide.
Ces résultats suggèrent que les altérations du cerveau, outre celles associées à la schizophrénie, peuvent prédisposer certains patients à un risque plus élevé d’homicide ou de suicide dans des circonstances particulières. De plus, certaines de ces altérations peuvent être partagées entre les patients homicides et suicidaires. Cependant, en raison de plusieurs limitations, dont le nombre réduit d’études disponibles, aucune conclusion déterminante ne peut être tirée et plus de recherche est nécessaire.
Suicide and homicide are major public health concerns. According to the World Health Organization,1 one million people die by suicide, and 10 to 20 times more people attempt suicide each year around the world. Further, more than half a million people die by homicide each year worldwide.2,3 These acts represent a significant cost to the people involved, their families, and society as a whole.4
To date, several risk factors for suicide and homicide have been identified, the most significant of which is the presence of psychiatric disorders.5–8 Among these, schizophrenia, one of the most debilitating psychiatric disorders, is associated with increased risks of both death by suicide9 and committing homicide.10 Suicide is one of the main causes of premature death among these people (10% to 15% die by suicide).11 A recent meta-analysis also confirmed that the risk of homicide is 20-fold among patients with schizophrenia, compared with healthy control subjects.10 In line with this, it has been estimated that 6% of murderers suffer from schizophrenia.5–7 Although schizophrenia increases the risk of both suicide and homicide, only a few patients with schizophrenia will die by suicide or commit homicide. Therefore, schizophrenia alone is insufficient to explain the full suicidal or homicidal risk.
Additional risk factors for suicidal and homicidal behaviours have been identified in patients with schizophrenia, several of which are identical for both behaviours.6,10,12,13 These factors include being male, younger age,13–18 low educational level, low level or absence of a professional activity,19–25 alcohol abuse,12,26,27 and a history of childhood physical abuse.13,28,29 Impulsivity has also been shown to be a major risk factor of suicide30 and aggression31 in schizophrenia. Moreover, a history of suicide attempts is the strongest predictive factor of future suicide completion,8,32–34 and a prior history of any violent offence is the strongest predictive factor of future homicide.6,10,12,13
These sociodemographic and clinical findings support 2 hypotheses: both suicidal and homicidal behaviours may have specific underlying vulnerabilities distinct from those of schizophrenia; and some of these vulnerabilities may be shared between both suicidal and homicidal behaviours. However, the large number of clinical risk factors of suicidal and homicidal behaviours and the complexity of their interactions limit our ability to predict these behaviours in people suffering from schizophrenia (that is, schizophrenia, schizoaffective, or schizophreniform disorders diagnosed as per the DSM-IV criteria). To overcome this issue, we need to explore other ways, such as neurocognitive impairment, which may be more sensitive in detecting vulnerabilities to these complex acts, keeping in mind that schizophrenia, per se, is associated with a range of cognitive deficits.35 In mood disorders, neuropsychological and neuroimaging studies have yielded interesting findings regarding the neurocognitive basis of the vulnerability to suicidal behaviour.36
Our paper aims to identify the specific neurocognitive factors (from neuropsychological and neuroimaging studies) of suicidal and homicidal behaviours in people with schizophrenia, and to examine possible shared pathways between suicidal and homicidal risk at the neurocognitive level. A systematic literature review was performed to synthesize all published articles assessing the association between neurocognitive markers and suicidal and homicidal behaviours in people suffering from schizophrenia. Both neuroimaging and neuropsychological studies were reviewed. Based on previous findings in suicidal behaviour of mood disorders,36 and on the observation that many risk factors for suicide or homicide are shared across diagnoses,37 we expect patients with schizophrenia and histories of suicidal behaviour or homicide to show impaired decision making and cognitive inhibition in relation to amygdala and orbitofrontal, dorsomedial, and dorsolateral prefrontal cortices impairment.
We performed a systematic MEDLINE literature search of all clinical trials, cohort, case–control, and cross-sectional human studies published in English until August 31, 2012. The MeSH terms “schizophrenia” and “psychotic disorders” were combined with the MeSH terms “violence,” “homicide,” and “suicide.” Then, these were combined with the MeSH terms and with the TIAB terms of neurocognitive pathways. To explore the neurocognitive pathways of suicide and homicide, the following MeSH terms were used: “cognition,” “neuropsychology,” “neuropsychological tests,” “executive function,” “decision making,” “problem solving,” “magnetic resonance imaging,” “diffusion magnetic resonance imaging,” “positron-emission tomography,” “prefrontal cortex,” “tomography, emission-computed, single-photon,” and “diffusion tensor imaging.” TIAB terms also included “neuropsychological functions,” “executive functioning,” “executive performance,” and “neuroimaging.” For completeness, the search also included the Embase database (Ovid interface) from January 1, 1996, to August 31, 2012. An iterative process was used to ensure that all relevant articles were obtained. We also performed a manual search of bibliographical references of extracted papers and existing reviews to identify potential studies not captured in the electronic database searches.
Abstract selection was based on the STROBE checklist,38 which describes items that should be included in reports of cohort studies. Our review followed the STROBE statement guidelines. Abstracts identified through the literature search were independently evaluated by 2 authors for meeting the inclusion criteria. The quality of each study was assessed independently by 2 authors using the Crombie criteria adapted by Petticrew and Roberts.39 Disagreements were resolved by a third author.
Titles, abstracts, and manuscripts were only included in the analyses if they met all of the following 4 inclusion criteria: published in a peer-reviewed, English-language journal; included people with schizophrenia, schizoaffective, or schizophreniform disorders diagnosed as per the DSM-IV criteria; included at least 1 neuropsychological task or an MRI; compared at least 2 groups of patients, one of which comprised patients with a history of suicide attempts (defined as any act carried out with a certain intent to die and different from self-mutilation37) or with a history of homicide (defined as fatal injuries inflicted by a person with intent to injure or kill another person2). The study selection process is shown on a chart flow diagram in Figure 1.
Three studies explored executive functions in relation to homicide in schizophrenia (for a total of 82 patients with a history of homicidal behaviour, 89 patient control subjects, and 61 healthy control subjects) (Table 3). In comparison to patients with schizophrenia without such a history, patients with schizophrenia and a history of homicidal behaviour performed more poorly on several executive functions, including mental shifting, cognitive inhibition, and verbal fluency.40 In addition, executive deficits were more common in people with schizophrenia regardless of their history of violence, compared with patients with antisocial personality disorder only.40 However, 2 other studies reported no such group differences.41,42
In the case of suicidal behaviour, 5 studies were found (including 1 prospective study) that showed contradictory results (328 patients with a history of suicidal behaviour; 453 patient control subjects). In 2 studies, outpatient suicide attempters with schizophrenia tended to outperform nonattempters with schizophrenia in executive functioning, namely, on measures of attention and verbal fluency,43 and cognitive flexibility.43,44 However, the 3 other studies did not find any group difference in executive performances.45–47
Five studies were found in relation to homicidal risk (130 patients with a history of homicidal behaviour, 114 patient control subjects). Patients with schizophrenia and with a prior homicide history showed various brain structural abnormalities, including reduced whole brain volume,48 cortical thinning,49–51 particularly in the medial inferior frontal cortex,49,51 a decrement of temporal48,51 and hippocampus volumes,52 an increment in the volume of the amygdala and the putamen,48 a reduction in the volume of the sensorimotor areas,49 a reduction in the volume of the grey matter in the cerebellum, angular gyrus, and supramarginal gyrus,50 a reduction in the grey matter volume in the right parahippocampal gyrus, and a grey matter augmentation in the mediofrontal cortex,52 compared with patients with no history of homicidal behaviours.
Regarding suicidal acts, 3 studies were found (37 patients with a history of suicidal behaviour, 105 patient control subjects, and 100 healthy control subjects). In patients with schizophrenia and a history of suicide attempt(s), compared with patients with schizophrenia and no suicidal history, there was a significant reduction in the grey matter volume in the left superior temporal lobe and in the left orbitofrontal cortex.53 In addition, there was evidence of significantly larger inferior frontal white matter volumes bilaterally.54 There was also a selective increase in volume in the right amygdala of patients with a history of suicidal acts, compared with patients without such history and with healthy control subjects.55
We found 4 functional neuroimaging studies pertaining patients with homicidal behaviours. These studies reported reduced activation in the right frontal and inferior parietal areas in a working memory paradigm,56 and greater activation in the left medial frontal–cingulate gyrus and bilaterally in the temporal-occipital regions in an experimental manipulation of anticipatory fear induction,57 in patients with schizophrenia with previous homicidal acts, compared with those without. Violent patients with schizophrenia with both comorbid antisocial personality and SUD failed to show higher blood flow in the orbital or basal regions during the inhibition part of the go–no-go task, compared with patients with schizophrenia only and with nonviolent healthy control subjects.58 Further, significantly higher activation in frontal motor, premotor, and anterior cingulate regions were observed in this particular group, compared with patients with schizophrenia only.
No functional neuroimaging studies were found in relation to suicidal acts in patients with schizophrenia.
Our study provides the first systematic literature review, revealing the following findings:
Before a detailed discussion of these findings, we must start by highlighting several main limitations. First, the number of studies available, to date, is small, particularly the ones investigating suicidal acts, thus hindering any definitive conclusions. We selected only studies that compared at least 2 groups of patients; one comprised of patients with either a history of suicide attempts or a history of homicide. This is the best way to evaluate specific vulnerability to suicidal and homicidal behaviour in schizophrenia. Second, the populations studied were very heterogeneous in terms of sociodemographic and clinical characteristics. For instance, some studies included only male subjects,40,42,48,49,51,53,56–58 whereas others grouped male and female participants together.41,43,44,46,50,54,55,59–64 Additional limitations include the small sample sizes; the lack of information on substance abuse or medication; the variability of the primary diagnosis42,46,61,62; and that comorbid antisocial personality disorder was often not assessed.40,48,49,56,58 Moreover, the severity of the psychopathology of a subject, the recurrence of psychotic episodes, as well as the age of the first episode psychosis, are often omitted and may also influence the variability of results. Moreover, substance abuse in schizophrenia is associated with an increased risk of violent homicide in longitudinal studies12,65 and with suicidal behaviour.18
The definitions of suicidal and homicidal behaviours were also heterogeneous. The studies included patients with a history of suicide attempts,43,46,53–55,63 suicide attempters with suicidal ideations,44,62 patients who had committed homicide,58 patients with a history of severe violence, including homicide and attempted homicide.41,42,50,59,64,66 Additionally, some authors differentiated severe from less severe suicidal and homicidal acts.
The methodology of the studies reviewed also varied. For instance, the delay between the suicidal or the homicidal act and the neuropsychological assessment varied. The tools used to assess suicidal and homicidal behaviours and neurocognitive markers also differed. For example, homicidal behaviour was evaluated using the Gunn and Robertson scale, with a cut-off score of 440,48,56 or 5.51,57 In addition, while some authors used only a single neuropsychological test to evaluate cognitive processes, others used a larger battery of tests.
Despite these limitations, there are several important implications that can be derived from our review. First, more homogeneity in assessment is required if we are to understand the neurocognitive basis of the vulnerability to homicidal and suicidal risks in schizophrenia. Second, many more neuropsychological and neuroimaging studies are needed. Data from mood disorders largely support the concept of vulnerability to suicidal behaviour, including but not limited to impairment in decision making67,68 and cognitive inhibition.69–72 In addition, structural and functional neuroimaging have implicated several brain regions, including the orbifrontal, dorsolateral, and dorsomedial prefrontal cortices.36 These cognitive deficits and brain dysfunctions may become part of future assessments and the target of future interventions to reduce the suicidal or homicidal risks.
Among the studies published to date pertaining to patients with schizophrenia and a history of suicidal or homicidal acts, most findings come from structural neuroimaging. Nine studies investigated the homicidal risk in schizophrenia and reported reduced inferior frontal and temporal volume, and increased amygdala volume.48 These results are consistent with previous studies implicating these regions with psychopathy scores, history of violence, antisocial personality disorder diagnosis, or impulsiveness.57,58,73,74 A recent meta-analysis suggested that patients with a history of violent behaviour, including homicide, were associated with significantly reduced prefrontal structure and function (right orbitofrontal cortex, right anterior cingulate cortex, and left dorsolateral prefrontal cortex).75 Moreover, in the review by Naudts and Hodgins,76 male patients with schizophrenia who consistently showed a pattern of aggressive and antisocial behaviour had poorer orbitofrontal functions, an increased size of the putamen, and reduced volumes of the amygdala, the hippocampus, and the orbitofrontal cortex, compared with patients with schizophrenia with no history of violence.
Literature also suggests that, among patients with schizophrenia, those who present with conduct problems from childhood onward perform better on tests assessing specific executive functions. However, they also perform more poorly on assessments of orbitofrontal functions, show fewer neurological soft signs, display larger reductions in the amygdala volume, more structural abnormalities in the orbitofrontal cortex, more abnormalities of white matter in the amygdala–orbitofrontal cortex, and smaller reductions of the hippocampus.76 Only one study58 explored neurocognitive pathways of homicidal behaviour and tried to distinguish subtypes of homicide offenders who differ in comorbidity associated with schizophrenia. Using functional MRI, they showed homicidal people with schizophrenia and comorbid antisocial personality disorder and SUD would present a different pattern of prefrontal functioning than homicidal people with schizophrenia only. Frontal basal cortices were significantly less activated in people with schizophrenia and comorbidity during the execution of a go–no-go task than in people with schizophrenia only and nonviolent people without mental illness. In contrast, significantly higher activations in frontal motor, premotor, and anterior cingulate regions were observed in the schizophrenia plus comorbidity group than in the schizophrenia only group.58 More studies are needed to disentangle possible subtypes of homicidal histories and their neural basis.
Fewer results were found in patients with histories of suicidal acts. Interestingly, decreased orbitofrontal and temporal cortical volumes, and increased amygdala volume were reported, suggesting partial overlap with vulnerability to homicidal behaviour. Also, decreased orbitofrontal cortex volume and increased amygdala volume,77 and altered orbitofrontal cortex activation78 were found in patients with a mood disorder and a history of suicidal behaviour. Biochemical alterations were also found in the orbitofrontal cortex of depressed suicide completers.79 Taken together, these findings suggest that some brain alterations occur across diagnoses.
Schizophrenia is associated with numerous cognitive and brain alterations.80 Our study suggests, without confirming, that particular brain alterations may add to other disease-related deficits (for example, hallucination and temporal cortex) to affect a patient’s ability to perceive and respond adequately to their environment. Specific deficits in the prefrontal–limbic circuitry may predispose some patients with schizophrenia to homicidal behaviour or to suicidal behaviour.
More studies are needed and future research should directly compare neurocognitive markers of suicidal and homicidal risk. Cohort studies are also necessary to determine the predictive power of these markers. Finally, targeting these deficits may be part of future therapeutic interventions in at-risk patients.
None of the authors have any competing financial interests for this manuscript. Dr Richard-Devantoy received a grant from the Canadian Institute for Health Research. Dr Turecki and Dr Jollant received a salary grant from the Fond de Recherche du Québec—Santé (FRQS). Dr Dumais received a fellowship grant from the FRSQ. None of the funding bodies influenced the project, the analyses, and the publication of this article.