A pragmatic evaluation of a nationwide mental health outreach service in Japan raises more questions than answers

What is already known on this topic

Intensive forms of community treatment for people with severe mental illness include Assertive Community Treatment (ACT) and Assertive Outreach (AO). They are characterised by caseloads of <20 patients per clinician. These interventions can reduce admissions to hospital, increase retention in care and improve social functioning, but the effect on mental state and quality of life is less clear.1 Any benefit is greatest for patients with high rates of hospital admission.1 There is less evidence when comparing less intensive forms of case management with standard care.1 Japan has not had a nationwide implementation of ACT and there are many patients who lose contact with mainstream services.2

Methods of the study

Kayama and colleagues evaluated the effectiveness of the nationwide Japanese Outreach Model Project (JOMP), established in 2011 by the Japanese Ministry of Health to provide multidisciplinary outreach services for eligible patients to prevent them from repeated hospitalisations.2 Patients, who do not or will not use regular Japanese outpatient care funded by public insurance, but are at high-risk for hospitalisations are the target population of JOMP. The multidisciplinary JOMP programme focuses on activities of daily living, social interactions, and mental and physical health. Although not ACT, community-based services are provided on a 24 h basis; caseloads vary from 3 to 20 and patients are re-evaluated every 6 months to see if they continue to need the service. The design of the study by Kayama and colleagues was a longitudinal retrospective survey of 162 participants from 32 JOMP outreach teams across the country. Outcomes were assessed at baseline, 6 and 12 months. These included readmission, Global Assessment of Functioning (GAF) scores and problematic behaviours measured by the Social Behavioural Schedule (SBS). Data were compared to previously published studies from Japan and abroad on ACT or AO (caseload=10–20) or standard care (caseload=80).

What does this paper add?

• Patients spent more time on the programme engaged on psychosocial interventions with staff than before the study. Mean GAF scores improved from 37.6 (SD=13.4) at baseline to 41.9 (SD=16.2) after 12 months (F=20.0, p<0.01). The SBS fell from a mean of 25.1 (SD=11.4) to 18.5 (SD=12.8) over the same period (F=17.5, p<0.01).

• The JOMP achieved similar outcomes in terms of subsequent inpatient service use as previously published evaluations of more formal ACT or AO. This was despite patients in the JOMP having lower levels of social function on entry to the study than those reported in Japanese ACT studies.

• Results were also similar to published data on standard care even though, as noted before, the participants had potentially a worse prognosis because they had disengaged from mental health service for at least 3 months.

Limitations

• There were no controls, so regression to the mean cannot be excluded. As a result, comparisons were with previously published research and so no adjustment could be made for any differences between studies in terms of demographic characteristics, clinical features, psychosocial functioning or health service use.

• The effect of previous admissions on outcomes was not studied although this is an important determinant in response to more intensive forms of case management such as ACT. Where data on admissions in the 18 months prior to the study were presented (16.7%), the programme did not achieve any reduction given the readmission rate was 27.2% over 12 months of follow-up. Furthermore, follow-up was limited to 12 months and the statistically significant improvements in psychosocial functioning and behaviour were of unclear clinical significance. Finally, data on problem behaviour at 12-month follow-up were only available on 57 participants.

• There was considerable overlap in the characteristics of the teams participating in the JOMP and those of the comparison ACT studies in terms of staff mix and caseload. For instance, although caseloads in the JOMP ranged from 3 to 20, the average was 6. On a related issue, the wide range of caseloads raises questions about the fidelity of participating teams to the JOMP model and whether the 162 participants from 32 teams were receiving similar care. In addition, it is unclear whether any additional benefit from the JOMP when compared with standard care was to do with the increased number or quality of staff. Levels of inpatient health service use reported in this study may have been influenced by the fact that Japan has one of the highest numbers of psychiatric beds per 1000 of the general population worldwide.3 This may therefore limit the generalisability of findings to other countries.

What next in research

More head-to-head randomised comparisons are needed for intensive versus non-intensive case management. This study does not resolve the issue.

Do these results change your practices and why

The limitations mean that the study is unlikely to change practice and it does not assist in further determining those aspects of community care that are associated with better outcomes.