In a recent paper Jones et al (Reference Kendler and Karkowski-Shuman2001) reported an association of aggressive behaviour in schizophrenia with catechol-O-methyltransferase genotype. The authors studied a sample of 136 males and 44 females with schizophrenia. Aggression in patients was clinically assessed by means of the Overt Aggression Scale (OAS; Yudofski et al, 1986). The patients in this sample showed a surprisingly high level of aggression: 52% verbal aggression in male patients (46% in females), 39% aggression against objects (25% in females), 23% against self (9% females) and 39% against other people (34% females) as measured by the OAS. Data on prevalence of aggression and violence in people with schizophrenia differ widely depending on definition and assessment period but most clinical studies in this field have shown lower rates, at least of physical aggression (for review see Schanda & Taylor, 2001; see also Reference Pavlovic and MarušicMonahan et al, 2000).

In a recent retrospective study we evaluated the patient files of all patients with ICD-9 schizophrenia admitted to the psychiatric hospital of the University of Munich between 1990 and 1995 (n=2093). Relevant socio-demographic, clinical and psychopathological data were evaluated. Fourteen per cent of patients (n=292) met the criteria for aggression (verbal and physical) on admission (Soyka & Ufer, 2002).

Jones et al feel that aggression may even be underestimated in their sample. Recent data suggest that the risk of violence is indeed overlooked in psychiatric patients. Sanders et al (2000) pointed out that while psychiatric patients are asked about suicidal ideas, aggression and the risk of violence are frequently neglected even in patients with clearly violent thoughts.

I have some concerns over whether a single rating scale can be valid and reliable enough to assess aggression adequately in schizophrenia, especially for genetic studies. Aggression and violence in schizophrenia are difficult to predict (Steadman et al 1998; Wallace et al 1998; Reference Pavlovic and MarušicMonahan et al, 2000) and can be both trait or state phenomena in schizophrenia. The OAS is a sensible instrument in this field but aggression is a multi-dimensional phenomenon. In most studies on that issue data from different sources are utilised (Swansonet al, 2000). Steadman et al (2000) have proposed an actuarial tool for assessing the risk of violence which has been evaluated in civil psychiatric patients (Reference Pavlovic and MarušicMonahan et al, 2000). Beside clinical interviews and specific psychopathological scales a broad number of other diagnostic instruments can be used to assess aggression and the risk for violence in psychiatric patients. A few of them may be mentioned: the Buss—Durkee Hostility Inventory with its sub-scales including ‘physical aggression’ (Reference Evans, Reeves and PlattBuss & Durkee, 1957) or the Brown—Goodwin assessment for lifetime history of aggression (Reference DiekstraBrown et al, 1979). Assessment and classification of such complex phenomena as aggression should not be based on only one scale, especially in gentic studies with their substantial clinical and possibly forensic/legal implications. Nevertheless, we hope that this research may provide more insight into the biological mechanisms underlying aggression.