Mental health research is now, for the first time in its existence in the UK, receiving proper attention from funding bodies and policy makers. The amount of attention it is receiving is still far from satisfactory but if the first part of finding a solution is to recognise a problem, this interest is to be commended.

What lies behind each of the points raised by Thornicroft and his colleagues (Reference Thornicroft, Bindman and Goldberg2002, this issue) is the infrastructure of the environment whereby mental health research is carried out. Many of the difficulties associated with successful research into mental health services come from sceptical health professionals who believe dogmatically that their interventions are the best that can be achieved and who consider alternatives, particularly those that might be selected by chance (i.e. randomisation), are unethical (Reference Oliver, Piachaud and DoneOliver et al, 2002). We still need to improve the environment whereby professionals welcome research intervention instead of regarding it as an unwelcome intrusion, designed primarily to promote the careers of researchers rather than improve the health service. Of the gaps identified by Thornicroft and his colleagues, I should like to concentrate on three as being of greater importance than the others.

User involvement

In the past 20 years, there has been a tremendous shift towards giving patients a much greater say in their treatment, and this is one of the essential elements of ‘post-modern psychiatry’ (Reference Bracken and ThomasBracken & Thomas, 2001). This is an interesting development, not least because it is not evidence-based but the consequence of changes in society, with the massive growth of consumerism. There is a real danger that the engine of user initiatives in mental health services, although positive in principle, will accelerate out of control and drive mental health research into the sand. A recent review (Reference Crawford, Rutter and ManleyCrawford et al, 2002) has highlighted that, although user involvement has been highly promoted, the impact of such involvement is far from clear. Such a change in policy has always come before research, often as part of political correctness, and no adequate evaluation has been made of the consequences. In the second recommendation of this paper, a review on the ways in which users can directly participate in research is suggested; this ought to be much wider and include all aspects of user involvement in research.

Large-scale randomised controlled trials

Despite great operational difficulties there have been important large-scale randomised controlled trials (RCTs) carried out in mental health services, of which the UK 700 Study is the most recent example (Reference Burns, Creed and FahyBurns et al, 1999). A reason for these not being more widespread is partly a lack of funding (there are relatively few streams for largescale funding of research as funds have also been attracted to stimulating short, pioneering research studies), but, in my view, the main handicap has been the excessive speed of reform in mental health services that has moved far ahead of available evidence. Thus, for example, the introduction of care management, the Care Programme Approach (CPA), assertive outreach teams and crisis resolution teams into the framework of mental health policy have all happened in the absence of adequate research. The research that has been carried out has often suggested that the introduction of these new approaches is counterproductive (Reference Marshall, Lockwood and GathMarshall et al, 1995) and I personally believe that the main reason for the scandal of insufficient beds being available for acute psychiatry in many parts of the UK is a direct consequence of delays in discharge created by the CPA which increases the average duration of admission by more than two-thirds (Reference Tyrer, Morgan and Van HornTyrer et al, 1995). If we are to make significant progress in this area, and create trials that will properly inform practice, we need to hold back the introduction of new policy initiatives until proper research has been carried out.

Some would argue that this is an unrealistic aim as governments have short time-spans and have to influence policy, but there is some room for compromise here. For example, instead of making the introduction of new policies statutory and thereby precluding alternative forms of assessment, some areas could be regarded as research ‘testing beds’ in which different models could be evaluated without the stifling blanket of statutory requirement being imposed.

Improving training

There is no doubt that training must be a priority for mental health research and we have somewhat belatedly come round to this conclusion, after realising that so many interventions that are effective are not having a major impact on the mental health of the nation because there are insufficient people available to administer them effectively. The best example of this is cognitive—behavioural therapy, which in innumerable trials, ranging from common conditions such as anxiety and depression through to health anxiety and somatisation, schizophrenia and bipolar disorder, has been shown to be effective (Reference Butler, Fennell and RobsonButler et al, 1991; Reference Clark, Hawton and CowenClark, 1990; Reference Clark, Salkovskis and HackmannClark et al, 1998; Reference Kingdon, Turkington and JohnKingdon et al, 1994) but has changed ordinary practice very little because there are so few people trained adequately in the therapeutic procedures. Research to improve training and effectiveness is sorely needed and the approach of Kevin Gournay and his colleagues (Reference Gournay, Sandford and Johnson1997) seems to be an important way forward that deserves much investment.

This paper is very useful in opening the eyes of the profession to important issues in mental health research and its message should be taken on board by all mental health professionals, particularly psychiatrists, who have more training in research than any other discipline. It is to be hoped that other disciplines will also receive comparable instruction in research methodology in time. One of the consequences of clinical and research governance, if implemented correctly, is that we can no longer make the distinction between those who are researchers and those who are practitioners. If we practise in mental health, part of our mind should always be attuned to research; to abandon it will lead to stagnation.