countries. We need results from Finnish datasets that cover persons with diabetes treated by diet and include information on the disease (type, treatment, duration) and risk factors. British analyses of (pooled?) population based datasets with a large number of deaths would help in assessing the validity of the findings of Chaturvedi et al. The situation in other countries should also be studied.

Geographical mobility of psychiatric patients in London is high and is particularly marked for those presenting for in-patient treatment. These findings suggest that greater mobility could be one of the most important reasons for the higher than expected demands on psychiatric services and the difficulties in maintaining contact with patients in London in general and inner London in particular. More attention should be paid to geographical mobility as a predictor of psychiatric service use, and it is recommended that it is recorded regularly in information systems.

An important forensic psychiatric measure, contacts with police, was compared in a randomized, controlled trial of 155 patients with severe mental illness with a previous admission within the past two years. The patients, who also had their personality status addressed formally before randomization, were allocated to community multidisciplinary teams or to hospital-based care programs after discharge from in-patient care and were followed up for one year. A total of 138 patients (89%) had at least one post-baseline assessment and of these patients, 16 (12%) had at least one police contact in the year of the study, most of which were emergency assessments. The data showed significantly greater numbers of police contacts in patients with increasing severity of personality disturbance. Patients with such disturbance were six times more likely to have police contacts than those with no personality disorder. There were significantly more contacts in patients with borderline and antisocial (dissocial) personality disorder allocated to community-oriented care compared with hospital-oriented care. These findings have important implications for risk assessment in severe mental illness.

Most of the studies that are frequently cited as examples of effective comprehensive community care, (i.e. they reduce the demand for hospital beds without any loss in treatment efficacy (Stein & Test, 1980; Hoult & Reynolds, 1984; Muijen et al., 1992) were carried out before the introduction of the Care Programme Approach (CPA) in 1991 (Department of Health, 1990) which at present only applies to England. As the CPA derives from these earlier studies the discrepancies between hospital and community based aftercare might be expected to become less, as now all services in England are expected to include a significant community element. However, there can still be important differences between those services focusing on community care as the main priority and those in which the hospital system is paramount.The psychiatric services in the area covered by North West London Mental Health Trust (NWL Trust) represented a natural test of these two approaches as they had parallel hospital and community based teams covering the same catchment areas respectively, North Paddington, in Westminster and Brent, in outer London.At this point it is useful to provide more detailed description of the two geographical areas at the time of the study and the community and hospital based teams that were involved.