To the Editor: We read the article by Dr. Nielssen and colleagues (1) in the July issue with interest. The authors' analysis indicated that 138 persons with psychosis were found not guilty of homicide on the grounds of mental illness (NGMI). In a large database of 21,857 Dutch pretrial psychiatric reports made between 2000 and 2006, we found that 61 homicides were committed by people with a psychotic illness (2).

Unlike other Western countries, the Netherlands applies a 5-point scale of accountability (3). Of the 61 reports that we investigated, 36 psychotic defendants (59 %) were considered unaccountable, 14 (23%) had strongly diminished accountability, four (7 %) had diminished accountability, and one (2%) was considered fully accountable. (The fifth category is slightly diminished accountability, and no defendants were judged to have this level of accountability.) Most of the suspects did not have a violent criminal past and were not under psychiatric treatment at the time of the offense, in line with the findings of Dr. Nielssen and colleagues.

In conclusion, the rate of violent offenses among persons with psychosis is even higher than the number of persons found NGMI. For example, some psychotic defendants commit a homicide because they need money to support a comorbid addiction. Although not all violent offenses by persons with psychosis are caused by the underlying psychosis, we fully agree with the authors that the early recognition and treatment of psychosis may prevent violent offences.