Killaspy et al's longer-term follow-up ^{Reference Killaspy, Kingett, Bebbington, Blizzard, Johnson and Nolan1} to the REACT study ^{Reference Killaspy, Bebbington, Blizard, Johnson, Nolan and Pilling2} replicated their original finding that ACT teams had no advantage over CMHTs in reducing in-patient care and concluded by questioning further investment in ACT in the UK. We found this interesting because we have evidence for a reduction in in-patient bed use locally, albeit using a different methodology.

The Sandwell Assertive Outreach Team has been operating for over 5 years, serving an ethnically and socioeconomically diverse urban population of approximately 280 000. The team has remained adherent to the Department of Health Policy Implementation Guide ³ and has a mean score of 3.7 on the Dartmouth Assertive Community Treatment Scale. ^{Reference Teague, Bond and Drake4} We retrospectively reviewed our performance in terms of number of admissions and bed-days for all 73 patients who have been with our service for over 3 years. We compared these results with data for the same population in a similar period prior to transfer of care to our team. The results are summarised in Table 1.

Table 1 Bed usage for 73 patients of the Sandwell Assertive Outreach Team before and after transfer to the team

	Year prior to transfer	Year after transfer	3 years prior to transfer	3 years after assertive outreach treatment
Admissions per patient	0.92	0.48	2.39	1.21
Admissions per patient per year	0.92	0.48	0.8	0.4
Bed-days per patient	63.6	30.5	156.7	80.1
Bed-days per patient per year	63.6	30.5	52.2	26.7

We are conscious of a local trend for referrals to our service to be initiated as patients relapse and therefore transfer of care often occurs on discharge from hospital. Improvements seen in 1-year figures may be due to a period of remission in keeping with the natural history of the illness, but the fact that improvements are maintained over 3 years in patients with frequent relapses would suggest that this is less likely to be a significant factor.

A possible explanation for the reduction in bed use might be that our assertive outreach team offers daily home treatment for patients in relapse and at risk of admission instead of involving the crisis and home treatment team. We are not aware of this aspect of assertive outreach being reported elsewhere in the literature about UK services and suggest it produces better outcomes by preventing patients with a history of disengaging from mental health services having to develop a therapeutic relationship with a new team at a time of crisis.

We feel that these before-and-after findings provide evidence to suggest that assertive outreach was locally responsible for reducing bed usage over several years in a population previously characterised by poor engagement and multiple admissions. Burns et al ^{Reference Burns, Catty, Dash, Roberts, Lockwood and Marshall5} found that fidelity to ACT staffing practices did not explain variation in outcome between trials and concluded that we should research the practices of teams. It would be interesting to know whether other services report a reduction of in-patient bed use and whether a programme of active daily visiting with medication in relapse played a part. We suggest that this aspect of assertive outreach could be incorporated in future research into effective components of the model.