We are delighted that our work is read by such influential figures in the ACT world and that they appreciate our attempt to bring scientific rigour to the area. However, we stand by our conclusions.

It is preceisely because we wished to understand the heterogeneity of results from good services and good studies that we undertook our work. We examined 20 components of home-based care in experimental and control services of published studies. These included, but were not restricted to, ACT studies. Our results (Reference Burns, Knapp and CattyBurns et al, 2001, Reference Catty, Burns and KnappCatty et al, 2002) show that the clear distinction between ACT and other forms of intensive home-based care that Professor Stein and colleagues insist on, and which Marshall and colleagues (Reference Marshall, Gray and LockwoodMarshall et al, 1997; Reference Marshall and LockwoodMarshall & Lockwood, 1998) used, simply will not withstand scrutiny. Many of the ACT teams did not have all the ‘ACT’ characteristics and many of the ‘non-ACT’ teams had more of them. In short, the groups overlap. It is for this reason that we conducted our regression analysis to determine which components are associated with outcome. Many of these ‘effective ingredients’ were present in later study control services. We would argue that this is why big differences are no longer being found in US or European studies.

None of this is to take away from the epoch-making impact of Stein and Test's original study. If we are to progress in our understanding of community care, however, we must be prepared to test more sharply-focused questions with increasingly well-designed studies.