First-onset psychosis has been the focus of intensive and increasing research interest that promises to yield dividends in both the understanding and treatment of psychotic illnesses (Reference LewisLewis, 2002). A prevailing hypothesis is that earlier and more effective initial treatment of psychosis will improve outcomes. Although evidence is not strong that untreated psychosis is neurotoxic, early treatment can clearly be justified on the grounds that it should reduce suffering (Reference Ho, Alicata and WardHo et al, 2003).

The development of a number of pioneering and high-profile dedicated early-onset services has resulted in a dramatic shift in mental health policy in England, albeit one that is the subject of significant debate (Reference Pelosi and BirchwoodPelosi & Birchwood, 2003). There is a requirement in The Mental Health Policy Implementation Guide to establish a network of specialist early intervention services across the country, structured according to a clear service description (Department of Health, 2001). Singh et al (Reference Singh, Wright and Joyce2003, this issue) have provided a valuable road map for the many health and social care economies in England that are struggling to implement this requirement. They present a useful ‘rough and ready’ methodology for identifying first-episode psychosis. They also found an extraordinary 18-fold variation in the incidence of psychosis across catchment areas in south and west London. This is considerably greater than the degree of variation reported by other investigators: a 2.5-fold variation between sites in an international study (Jablensky, 2003) and an approximately 4.5-fold variation between electoral wards in south London (Reference Boydell, Vanos and McKenzieBoydell et al, 2001). The mean incidence of referred psychosis across a broad swathe of south and west London (21/100 000 total population) is, however, close to the annual incidence of treated psychosis that we have identified from case records of people referred to community mental health services in Croydon. In this demographically-heterogenous outer-London borough, in 2001 we found an incidence of 24/100 000 total population first-onset cases referred to child and adolescent mental health services (CAMHS) and the local community mental health teams (CMHTs). We also found marked, though less pronounced, variations in incidence within the Croydon sectors that appeared to relate to socio-demographic factors, particularly poverty, high social mobility and patterns of substance misuse. One important lesson is that the incidence of psychosis is likely to vary between catchment areas, possibly quite markedly. Commissioners and providers will have to carry out local needs assessment and not rely on the centrally-derived figure of 15/100 000 total population per year, which is likely to be significantly too low for urban areas. This needs assessment will also have to be continually updated, since experience suggests that improved access to services will lead to an increase in referrals, at least initially (Reference Yung, Phillips, Drew, McGorry and JacksonYung et al, 1999).

Singh et al (this issue) also present important naturalistic data about the care received by their first-episode cohort. Just over 50% of CMHT patients and less than 4% of CAMHS clients were on an enhanced Care Programme Approach. The latter figure is likely to reflect the traditional way of working within CAMHS and suggests an urgent need to develop capacity to manage psychosis in adolescents within south and west London. Given the low numbers of first-onset cases and the large population served, spanning seven boroughs, this represents a significant challenge to local service planners and providers. The Policy Implementation Guide solution, to add consultant adolescent psychiatry sessions to a generic first-onset service, might not be the best option after all. Since allocation to an enhanced Care Programme Approach, and hence a care coordinator, is likely to be demand-led, the rather low figure of CMHT patients on enhanced Care Programme Approach suggests a further problem with the Policy Implementation Guide service specification. It requires case-loads for early psychosis services comparable to those suggested for assertive community treatment services. In reality, first-onset patients are very heterogeneous in terms of their psychopathology, response to treatment, and both immediate and long-term needs for care. Many simply neither want nor need the intensive inputs recommended in the guide. Singh et al show clearly that neither CMHTs nor CAMHS contain staff members with the skills to offer specialised interventions, such as cognitive-behavioural therapy, family therapy, vocational employment services and dual diagnosis care, should these be indicated. This skills gap is, of course, as the National Institute for Clinical Excellence Schizophrenia Guideline spells out, relevant to the care of all patients with psychosis (National Institute for Clinical Excellence, 2002).

The Croydon Outreach Assessment and Support Team, a dedicated first-onset service, has been in place in Croydon since April 2000. The practical experience of the team (which is exclusively community-based) offers reassurance about some of the concerns raised by Singh et al. First, it has not proven difficult to recruit and retain skilled and motivated staff, some of whom have been able to make regular use of training they have previously received but had been unable to employ within very over-stretched generic CMHTs. Second, once local professionals have grown used to the presence of a specialist service, they have been more than happy to refer on to the team, in the knowledge that this ensures their patients receive a high standard of care. Gone are the times when we could afford to treat patients because they were ‘interesting’. Third, issues of the interface between CMHTs and the early onset service are minimised by the fact that the Croydon Outreach Assessment and Support Team relates to and identifies with the local mental health service within the Borough rather than the currently fashionable million-plus aggregation. Finally, we have developed a joint care coordinator post that spans an effective adolescent service oriented to the assessment and management of psychosis and the Croydon Outreach Assessment and Support Team to ease the transition between CAMHS and adult services. This has proved to be a significant asset to both teams.

There remain many unanswered questions about the management of early psychosis, including the methods that should best be used to decrease the duration of untreated psychosis prior to entering the care system. Evolving early intervention services should be subject to careful evaluation.