There is little with which we would disagree in John Mahoney's analysis of community care services for severely mentally ill people (Letters, 18 December), the principal focus of the Zito Trust's campaign being precisely those areas and issues he describes.

But one aspect of his letter causes us a problem. He refers to research by the late Douglas Bennett which purports to show that, despite an increase in the overall number of homicides, the low number of homicides by mentally ill people has remained 'more or less constant'. Presumably he is referring to the paper in Psychiatric Bulletin (May 1996), which looked at the number of patients convicted of murder or manslaughter and sent to one of the three special hospitals.

Unfortunately, this research does not make convincing reading as it excludes a number of important categories, namely those patients found not guilty by reason of insanity, those found unfit to plead, those convicted of infanticide and a significant number who commit suicide after the homicide. Furthermore, Bennett's work also excludes all those cases not disposed of via the special hospitals, including those sent to medium-secure units or prison, for example.

The Zito Trust will shortly publish a paper which demonstrates how difficult it is to state with any degree of accuracy just how many homicides are committed by mentally ill people. This is based on a thorough analysis of Home Office figures. Perhaps the most reliable data currently available comes from the National Confidential Inquiry into Suicide and Homicide by People with Mental Illness, which recently published a progress report.

Taking a sample of 238 homicides since April 1996, the inquiry found that 43 per cent were committed by people with a mental disorder of some kind, including alcohol and drug abuse. Twenty-five per cent had a history of contact with the mental health services, 12 per cent in the year before the homicide. Based on a conservative estimate of 450-500 homicides a year in total, these figures, if an accurate reflection of the overall picture, are much higher than Bennett's - or anybody else's - and have serious implications for the future implementation of mental health policy, as they suggest a rate of four or five 'community care' homicides every month.

Michael Howlett,

Director,

The Zito Trust,

London WC2.

In response to Michael Howlett's concerns about how community care fails mentally ill people (Letters, 27 November), John Mahoney is right to emphasise the negative effect of poor housing environments on mental illness.

A network of community care with its centre firmly rooted in supported housing is essential for effective acute bed/resource management. Supported housing should be arranged in a care continuum which interfaces with acute services via a 24-hour staffed registered unit at the higher end, down to self-contained flats which receive a couple of hours' floating support each week.

This network, planned locally with housing associations and social services, is as important to keeping people in the community in good mental health as any amount of resource in the health sector.

The challenge for managers is to create ways in which health, housing and social services work together with a shared vision based on a comprehensive needs analysis. Where this is happening there will be fewer 'victims of community care'.

Lee Whitehead,

Service development unit manager,

Housing Care and Support,

London & Quadrant Housing Trust,

London SE3.