On behalf of my co-authors, I thank the correspondents for their interest in our paper. ^{Reference Barker, Taylor, Kader, Stewart and Le Fevre1}

The Edinburgh Crisis Centre is undoubtedly an important resource for the city, but it is not a crisis house in the Camden mould. They have a maximum of four beds, with a maximum duration of stay of 7 days. They do not take referrals from the National Health Service (NHS), and do not share information with the NHS. During a 12-month period, they had only 12 residential clients, who were also working with our intensive home treatment team (IHTT). So to conclude that the Crisis Centre is the reason for a reduction in hospital admissions is simply not accurate, although IHTT values the presence of the Centre.

Dr Bhattacharya will have noted that we demonstrated a 17% reduction in admissions requiring detention during our study (see previous letter by Bhattacharya & McQueen). This is in contrast to Forbes et al and Tyrer et al (see Bhattacharya & McQueen for references). The Forbes study was based on a nurse-led service operating in a semi-rural environment, which already had a good-quality community mental health team and low base rate of detention. We have already mentioned the limitations of the Tyrer study. ^{Reference Barker, Taylor, Kader, Stewart and Le Fevre1} An important point about not conflating crisis teams with home treatment teams is also made, and we believe it is home treatment that can obviate the need for admission.

Finally, Drs Casserly & Palin ^{Reference Casserly and Palin2} quite rightly suggest that our findings or model cannot automatically be generalised to other areas - this may be particularly true in remote or rural areas like Grampian. However, the planned bed closures they allude to would not have occurred without adequate alternative community provision - this was explicit in the strategy. Of course, once beds are closed, raw admission numbers fall, but not necessarily re-admissions or detentions (as we found). Further, mean length of stay has also fallen, consistent with a supportive ‘early discharge’ role. Naturally, we see a lot of dual diagnosis, but record only primary diagnosis. Last, pollsters such as MORI state that any postal survey with a response rate >10% is valid, and 29% of over 700 cases is a reasonable return, with many patients stating that they preferred home to the local psychiatric hospital as their locus of care. It should also be noted that in over 2 years of IHTT working with individuals who are by definition high risk, only one suicide has (tragically) occurred.

So, even in these austere times service innovation can have positive outcomes, but it is important to critically appraise these innovations against existing practice.