Background and aim: The prevalence of homicide perpetrators with a diagnosis of schizophrenia is 6% in Western countries populations. The relationship between schizophrenia and homicide is complex and cannot be reduced to a simple causal link. The aim of this systematic review was to clarify the role of substance abuse in the commission of murder in people suffering from schizophrenia.

Methods: A systematic English-French Medline and EMBASE literature search of cohort studies, case-control studies and transversal studies published between January 2001 and December 2011 was performed, combining the MeSH terms "schizophrenia", "psychotic disorders", "homicide", "violence", "substance use disorder", and the TIAB term "alcohol". Abstract selection was based on the STROBE and PRISMA checklist for observational studies and systematic and meta-analysis studies, respectively.

Results: Of the 471 selected studies, eight prospective studies and six systematic reviews and meta-analysis studies met the selection criteria and were included in the final analysis. Homicide committed by a schizophrenic person is associated with socio-demographic (young age, male gender, low socioeconomic status), historical (history of violence against others), contextual (a stressful event in the year prior to the homicide), and clinical risk factors (severe psychotic symptoms, long duration of untreated psychosis, poor adherence to medication). In comparison to the general population, the risk of homicide is increased 8-fold in schizophrenics with a substance abuse disorder (mainly alcohol abuse) and 2-fold in schizophrenics without any comorbidities. A co-diagnosis of substance abuse allows us to divide the violent schizophrenics into "early-starters" and "late-starters" according to the age of onset of their antisocial and violent behavior. The violence of the "early-starters" is unplanned, usually affects an acquaintance and is not necessarily associated with the schizophrenic symptoms. Substance abuse is frequent and plays an important role in the homicide commission. In addition, the risk of reoffending is high. In the "late-starters", the violence is linked to the psychotic symptoms and is directed to a member of the family. The reoffence risk is low and it depends on the pursuit of care or not.

Conclusion: Defining subgroups of violent schizophrenic patients would avoid stigmatization and would help to prevent the risk of homicide by offering a multidisciplinary care which would take into account any substance abuse.