[Investigating the association between psychosis and criminality/violence].

This review systematically describes the methodological difficulties arising at the interpretation of studies on the association between psychoses (major mental disorders) and criminality/violence: diagnostic issues, different definitions of illegal behaviour and of the connection between mental illness and crime, different sources of information, differences in the recruitment of cases and control groups and in the time under study, impact of interfering variables and statistical issues, problems of the transferability of results. On this basis the data of 23 papers published between 1981 and 2004 are discussed in detail. Even under consideration of sociodemographic variables and comorbid substance abuse, most studies show a moderate albeit statistically significant association between major mental disorders and crime. This association is especially proven in subjects suffering from schizophrenia. Their risk of committing a crime increases with the severity of the offence and is -- compared with the general population -- substantially higher for homicide. However, in general much more danger for the public arises from subjects with substance abuse and personality disorders. The severest expressions of violence can be interpreted -- at least partly -- as a direct consequence of the illness. The association between psychosis and criminality in general seems to be rather an indirect one, mediated through an illness-related increased vulnerability for general criminogenic factors as poverty, social deprivation and substance abuse, intensified by deficits of modern mental health care.