The importance of early treatment for first-episode psychosis has become increasingly clear during the last two decades. Two comprehensive meta-analytic studies ^{Reference Marshall, Lewis, Lockwood, Drake, Jones and Croudace1,Reference Perkins, Gu, Boteva and Lieberman2} have documented that long duration of untreated psychosis is related to poorer outcome, and the quasi-experimental Scandinavian Treatment and Intervention in Psychosis (TIPS) study indicates that reducing it is helpful. ^{Reference Melle, Larsen, Haahr, Friis, Johannessen and Opjordsmoen3} A crucial question is whether a specialised treatment programme for this group of patients can further improve outcome. This issue of the British Journal of Psychiatry presents two thought-provoking papers about the effect and cost–benefit of such a specialised programme. The papers describe two aspects of the Lambeth Early Onset (LEO) study: the cost-effectiveness of a specialised 18-month treatment programme ^{Reference McCrone, Craig, Power and Garety4} and the 5-year effect. ^{Reference Gafoor, Nitsch, McCrone, Craig, Garety and Power5} A previous paper from the LEO study clearly indicated that an assertive outreach with evidence-based biopsychosocial interventions gave a better 18-month outcome than standard care delivered by community mental health teams. ^{Reference Craig, Garety, Power, Rahaman, Susannah and Fornells-Ambrojo6} At follow-up the participants receiving specialised care were more often in regular contact with the clinical team, and failed to attend a smaller proportion of their appointments. They were more likely to have been offered psychosocial interventions and to be in recovery. They also had fewer admissions and better social and vocational functioning. McCrone and colleagues ^{Reference McCrone, Craig, Power and Garety4} have analysed the cost-effectiveness of the specialised programme. They found that the participants in the specialised programme group had more contacts with psychiatrists, psychologists, healthcare assistants, community mental health nurses and day-care services. On the other hand, they had less need for in-patient services, and their in-patient costs were only two-thirds of the costs of the standard care group. The overall costs were in favour of the specialised care group, but the difference was not statistically significant. Even so, the authors conclude that because of the better outcome for the specialised care group, it is very likely that the specialised care programme is cost-effective.

In the 5-year follow-up presented in this issue the authors collected data for the 18 months prior to 5-year follow-up. ^{Reference Gafoor, Nitsch, McCrone, Craig, Garety and Power5} They found that both groups had an increased number of admissions during the last 18 months of the follow-up compared with the first 18 months at the start of the study. But they found no significant difference between the groups in the chances for any admission, the number of admissions or the number of bed days used during the follow-up period. The data show that the specialised care group has a greater increase than the standard care group both in number and total duration of admissions.

The results of the LEO study are paralleled by those of the Danish OPUS study.

In OPUS, 547 participants with a first episode of schizophrenia-spectrum disorder were randomised to integrated treatment or standard treatment. The intervention was a 2-year assertive community treatment enhanced by better specific content, family involvement and social skills training. A primary team member, designated for each participant, was responsible for maintaining contact and coordinating treatment. The study convincingly demonstrated a beneficial effect after 2 years. The intervention group had a significantly lower level of psychotic and negative symptoms, fewer in-patient days, better treatment adherence and higher level of user satisfaction. ^{Reference Petersen, Jeppesen, Thorup, Abel, Øhlenschlæger and Christensen7}

However, after 5 years, the differences were no longer significant. ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, Øhlenschlæger and le Quach8} For both positive and negative symptoms the significant difference disappeared as a result of a deterioration in the intervention group. The standard care group showed no change. For Global Assessment of Functioning the intervention group remained stable whereas the standard group improved. In contrast to the LEO study, both groups showed a lower use of hospitalisation during follow-up. The drop was greatest in the standard group. However, at follow-up more participants in the standard group were not living independently and they spent on an average more days in protected homes.

What lessons can we learn from recent studies?

One lesson seems quite clear: specialised treatment for people with first-episode psychosis is effective as long as the treatment continues. The study by McCrone et al also clearly indicates that such a programme is cost-effective. ^{Reference McCrone, Craig, Power and Garety4} However, at the same time the lack of long-term effect indicates that the individuals are still vulnerable even after a successful 2-year specialised treatment programme. It is worth noting that both the LEO and the OPUS studies show that the specialised care groups deteriorate between 2 and 5 years. It seems like the 2-year effect is more the result of an ongoing active treatment than a cure. This is particularly important because, as Bertelsen et al state, the termination of the intervention ‘could cause feelings of loss for the patient’. ^{Reference Bertelsen, Jeppesen, Petersen, Thorup, Øhlenschlæger and le Quach8}

This may be an understatement as the patients no longer had a person specially designated to maintain contact and coordinate the therapy. Such changes may explain an interesting finding from the studies. Although both studies report that the intervention group had better treatment adherence than the standard group during the intervention period, neither study reports any significant differences at 5-year follow-up. The LEO study explicitly states that there were no significant differences between groups either concerning time spent with general practitioners or in psychiatric services. Even if none of the studies specifically address the question of treatment alliance at follow-up it is tempting to speculate that the termination of the intervention weakened attachment to treatment in several participants. Such a reaction to perceived loss would be in line with the conclusion of Kreyenbuhl et al in their recent review of disengagement from mental health treatment among individuals with schizophrenia: ‘Data point to the importance of patients' perceptions of treatment in their decision making around whether to remain engaged in care’. ^{Reference Kreyenbuhl, Nossel and Dixon9}

One of the most important reasons for the short-term success of the intervention programmes may have been the very fact that they put so much energy into keeping the individuals in treatment. When they were transferred to an ordinary programme, the intervention group lost this advantage. It is a challenge how to secure long-term involvement in treatment for these individuals. As underlined by Kreyenbuhl and colleagues there seems to be a need for greater emphasis on patient-centred communication and building of treatment alliance. We must develop long-term standard care programmes built on these principles. For the time being it seems unrealistic that most people with schizophrenia can manage without such focused effort, even after successful involvement in an intensive 2-year programme.

This brings another challenge into focus: prevention. It is worth mentioning that even if both the LEO and the OPUS studies are considered to be intervention studies, the duration of untreated psychosis was still fairly long: the LEO groups had a mean of 10.5 and 7.6 months, whereas the OPUS groups had a median of 46 and 53 weeks.

One might hope that earlier intervention could reduce vulnerability and give better 5-year outcomes even after termination of a more intensive 2-year treatment programme. Results from the TIPS study indicate that this may be possible (details available from the author on request).