There is a problem with medium secure psychiatric care in London. As Lelliott et al (2000) show, patients from Lambeth, Southwark and Lewisham Health Authority who find themselves in medium security are as likely to be looked after by the private sector as by the National health Service (NHS). Does this matter? After all, many patients in the NHS hospitals would happily upgrade to a private bed. Within medium security, the physical environment and range of specialist treatments may in some cases be superior to their NHS equivalents. In this sense, it does not matter that patients are treated privately. In fact, as forensic psychiatrists working in medium security are sometimes prone to complacency, it could be argued that the existence of competing private units has an important influence in raising the overall standard of care.

Problems arise because of the location of the private units. The patients are usually in London; the private secure units are not. Some are several hundred kilometres away, presenting a formidable obstacle to patients who wish to maintain contact with their families. It is difficult enough to maintain these ties anyway, when admissions often exceed a year in length. In addition, the psychological impact on patients and relatives should not be underestimated. It is bad enough to have to explain the transfer to a forensic psychiatry service, without having to introduce the idea that the patient will now be located a great distance from home. This degree of separation from the familiar is every bit as frightening to patients and relatives as the ‘big house on the hill’ image that psychiatry was meant to have left behind. Furthermore, as the study notes, many of these patients have not committed offences, but are coming from general psychiatric settings, where their behaviour has become unmanageable.

There is a further problem of rehabilitation, which is bound to be more difficult over such vast distances. This was highlighted in at least one public enquiry as a contributory factor to a homicide by a former patient. Such dramatic problems are rare, but the everyday reality is a waste of staff time in travelling to distant parts of the country, and unnecessary delays in rehabilitation, which have financial implications in one of the most deprived parts of the country.

There are many hidden costs of this model of care provision. As the health authority is spending money on distant services, it does not derive any of the benefits that come from providing such services locally. When a forensic service is doing its job properly, it provides advice to general psychiatric teams on risk management, and may fulfil a similar role in relation to the probation service and specialised hostels for mentally disordered offenders. There should also be community treatment and supervision for sex offenders, and for those with personality disorders. All of these activities tend to revolve around a medium secure unit, and all are lost when the unit is located elsewhere.

The paper by Lelliott et al raises other issues, including the reasons for the statistical overrepresentation of ethnic minorities among patients admitted to medium security. Unfortunately, like other work in the area, it cannot answer the questions it raises. The sample is too small. It is also a cross-sectional snapshot, providing no impression of movement through the system, and no information about referrals, or the context in which the service operates. This criticism applies to much research in forensic psychiatry. Researchers are often hunting data that any decent service would collect as a matter of routine. The sad fact is that the NHS has never bothered to do so. The added legacy of 18 years of manufactured competition means that researchers collect data in isolation, and similar exercises to this one have been carried out by health authorities throughout the UK. The use of different methods prevents the pooling of data, and limits the questions that can be asked.

Amidst all the gloom, there are at last some encouraging signs. The rebirth of regional commissioning makes it possible to take sensible decisions about in-patient services for mentally disordered offenders. There are plans for a statistical observatory for mental health in London, which would bring together data across the capital. Forensic services are ideally suited to such an approach, as the numbers are relatively small. It is estimated that fewer than 2000 residents of London are held in medium- or high-security services at any one time, and turnover is slow. It should be possible to develop and maintain a system that will provide basic information, to inform service planning.

Any researchers who fear unemployment should be reassured. The existence of such basic data would free them to address more interesting questions. For example, the difficult issues relating to medium secure care in London emerge only when one starts to compare services in different parts of the capital. Why does one part of the city make so much greater use of private beds than another? What characteristics of general psychiatric services enable an area to manage with fewer medium secure beds? These are the thorny questions with which researchers should be grappling and a database would provide them with the basic tools to make a start.

The growth of a forensic psychiatric database for London would also open up a number of other possibilities. Given adequate safeguards around confidentiality, would it be possible to share information about individuals? The ethical objections may be considerable, but there should be no doubting the strength of the arguments in favour. The single most common message from homicide enquiries is that the sharing of information is a major deficiency in our present arrangements for risk management. Are we seeking the Holy Grail of a perfect risk assessment tool, when our real failure lies in not using the information we already have?

Whatever the form of the system that may eventually emerge, there is a need for investment in the infrastructure of information technology. The relatives planning their visit to the distant medium secure unit will be able to make all their travel arrangements electronically. What are they to make of the fact that, once their relative is back in the community, any continuing risk will be managed entirely by means of people writing on pieces of paper, and hoping that other people will read them?