Snowden (2001, this issue) comprehensively sets out the evidence for there being a strong positive link between substance misuse, mental illness and violence. He rightly points to the complexity of this interaction, which is mot confined to mentally disordered offenders and will be familiar to general psychiatrists, particularly in inner-city practice. Menezes et al (1996) provides evidence in a south London catchment area of high levels of comorbidity between serious mental illness and substance misuse, which will, as Snowden demonstrates, be linked to risk of violence. I would not disagree that, for the small number of patients with high levels of risk for violent offending (particularly those conditionally discharged under Section 41 of the Mental Health Act 1983), a parallel model of forensic care (Snowden, 1999) may be required, but even among such patients I would argue that integration with local services will be necessary to meet their complex need. For most mentally disordered offenders, there are strong arguments for an integrated model of forensic psychiatry emphasising co-working, shared care arrangements and consultation/liaison, with the emphasis placed on local service provision. What most dual-diagnosis patients, particularly dual-diagnosis mentally disordered offenders, face is formidable barriers to integrated care, and there is a risk that a ‘parallel’ forensic service could be one more barrier preventing patients with complex needs from gaining access to the service they require. The forensic psychiatrist's role should include close working with general psychiatry colleagues to design, commission and deliver services and service models that will meet the needs of mentally disordered offenders, particularly those with dual diagnosis. The next phase in the development of forensic psychiatry may be the creation of local services allowing proper integration between forensic, general psychiatry and substance misuse services, and prisons and other agencies involved with mentally disordered offenders (Reference GroundsGrounds, 1996).

Snowden rightly points to the lack of strong evidence in favour of intensive integrated treatment approaches to dual diagnosis (Drake et al, 1998). However, there is even less evidence for the efficacy of sequential service strategies in which the dual-diagnosis patient has treatment from two separate teams in sequence, or parallel services in which the patient attends simultaneously two different teams for treatment. The evidence suggests that the major difference in philosophy and delivery of care between mental illness and substance misuse services results in dual-diagnosis patients not integrating properly with either. Drake et al (1998) conclude that a service model involving integration of mental illness and substance misuse services around a longitudinal stagewise motivational approach to delivery of services will, over a long period of time, bring results. A typical English multi-disciplinary community team (or local forensic service) should be the ideal setting for integrating services by incorporating substance misuse expertise into the team. Within medium secure units or other longer-term secure settings, specific substance misuse programmes should, according to literature, achieve better results when simultaneously addressing the treatment of mental illness and substance misuse. Given the very clear epidemiological links between substance misuse and mental illness and violence, the separation in training between general, forensic and substance misuse services is surprising. The future training of forensic psychiatrists should include significant exposure to training in substance misuse, but current training structures make this difficult to achieve.

The outcome studies for integrated services have implications for risk management. Mental illness may come under good control rapidly but the substance misuse component may require some years of sustained motivational interventions before substance misuse is effectively tackled. As the substance misuse may be the main risk factor for violence, the risk period for violence may be determined more by the outcome of the substance misuse component of patients' problems. This may go some way to explain the reluctance of psychiatrists generally to engage with patients deemed to suffer from a drug-induced psychosis. Psychiatrists may be wary of positive engagement with dual-diagnosis patients because of the fear of criticism if their patients commit antisocial acts linked to substance misuse over which for a long period of time psychiatrists may have little influence.

Although, arguably, services in prisons should be delivered through general psychiatry, forensic psychiatry will, in the immediate future, play a crucial role in the development of service provision in prisons. The Office of National Statistics survey of psychiatric morbidity among prisoners (Reference Singleton, Meltzer and GatwardSingleton et al, 1998) demonstrates that mentally disordered prisoners have complex needs, with comorbidity being the norm. Service delivery in prisons does not reflect this complexity. Medical, drug dependence and psychological services have all developed separately with different goals and different administrative and philosophical foundations. There is a real risk that this separation of services will result in poorly coordinated and patchy care for prisoners with complex needs. The prison drug assessment/treatment service (CARATS) is targeted only at drug misuse, excluding alcohol. Snowden argues persuasively that alcohol plays a central role in violent crime. Relative neglect of alcohol dependence within policies for substance dependence in prisons is therefore surprising. The forensic psychiatrist's role in prison will include both ensuring coordination of care for people with complex problems through mechanisms such as the Care Programme Approach, and influencing the development of services to ensure proper coordination of different service elements. Imprisonment may offer an opportunity to engage chaotic, difficult-to-reach patients through a proper care management system involving the prison and local area services. Preventive work, particularly with those in the prodromal phase of the development of psychosis associated with substance misuse, could be an important role for adolescent forensic psychiatrists in young offender institutions.

There is little published work on substance misuse programmes in secure settings. Given the length of stay in such places, they should provide unique opportunity to achieve detoxification, stability and work through the cycle of engagement for substance misuse. Such settings would also provide opportunities for testing hypotheses that atypical antipsychotic drugs may have particular benefits in those with comorbid conditions. It may be that the lack of emphasis on training in substance misuse limits forensic psychiatrists' understanding of what can be achieved and therefore their appreciation of the research opportunities in medium and high secure services.