In November 1999 a study from Bristol, published in the British Journal of Psychiatry, reported a proverbial cup-half-empty or half-full result: when provided with the ICD—10 PHC mental health guidelines, a group of 14 general practitioners (GPs) (volunteers) ‘showed no improvement in overall detection of mental health problems, in accuracy of diagnosis or the prescription of antidepressants’ (Reference Upton, Evans and GoldbergUpton et al, 1999). There was, however, a significant increase in the number of patients diagnosed with depression or unexplained somatic symptoms. Among the invited comments that were published alongside this first Bristol study were remarks such as, ‘At first glance... one is tempted to say that it merely tells us what we think we already know, namely that GPs are poor at detecting and treating depression’.

Anyone reading the new Bristol study reported in this issue could be forgiven for a slight feeling of déjà vu, in that Croudace et al (Reference Croudace, Evans and Harrison2003, this issue) have given the GPs of Bristol another working-over in a bigger and better-designed study, with the same largely negative result. Is the same disappointing general conclusion still justified? To some extent the answer has to be yes, but there are important differences in the two studies.

This time the study was not limited to GPs who were positive volunteers, and it included a comparison group of GPs who were not provided with the extra information. A brief quality-of-life assessment (of the patients) was also included. Technically, the study was a cluster randomised controlled trial involving several sophisticated statistical checks and procedures (which only the statistically minded can understand). The important issue of getting the GPs involved in some kind of active educational activity was dealt with by inviting 1 GP from each of the 15 intervention practices to join an editorial team which revised, amended and extended the ICD—10 PHC 1996 guidelines, so that the GPs would feel that the resulting document was a shared-ownership model. That sounds like a good idea but, curiously enough, no attempt was made to check whether any of the intervention GPs actually used the guidelines in their daily work during the period of the study. In theory at least, some of the GPs could have left the guidelines unused in the desk drawer or on the shelf while they got on with their daily surgeries in the usual way.

This second study concluded that the educational intervention had even less effect than in the first study. This may have been because the second study did not feature a special intensive study day for all the GPs involved. In the first study, structured role-play using the guidelines enabled the GPs to practise (under the eagle eye of Professor Sir David Goldberg) new consulting behaviours, as outlined in the guidelines. In the second study, however, for each participating practice a (presumably less-intensive) meeting was held between the research team and the GPs, at which the content and purpose of the guidelines were explained. This probably important difference was no doubt due to practical problems in setting up such a complete study day for the larger number of GPs involved in the second study (14 individuals in the first study, and all 56 GPs from 15 intervention practices in the second). The importance of detailed and personal involvement of all participating GPs was demonstrated, for instance, in the study by Tiemens et al (Reference Tiemens, Ormel and Jenner1999), which did find an improvement in the short-term outcome and level of daily functioning of general practice patients. The 17 GPs in this Dutch study had all taken part in eight training sessions of 2.5 hours each.

This editorial is not the place to indulge in a systematic review of the many studies of how GPs react to extra stimulation about the psychiatric illnesses of their patients, but one conclusion stands out in the very varied results reported. This is that the useful effects of extra stimulation are always associated with the involvement of the GPs in something quite active, that is directly related to their interviewing behaviour. In other words, behavioural change is best achieved by means of behavioural intervention. ‘You can take a horse to water but you can't make it drink’ is too crude to be a summary conclusion of the many studies on this subject, but the old saw does at least serve as a reminder of some of the main problems and frustrations awaiting anyone planning further studies.

We must also remember that, for a variety of reasons, not all patients with the less-severe types of psychiatric disorder need (or wish) to be treated. In addition, it is possible that the studies reported up to now are missing a variety of subtle but still worthwhile effects that are of benefit to patients, but that are too difficult to measure. It is, of course, easy to go on thinking up reasons for the generally disappointing results of many of the published studies. Nevertheless, whatever the explanation, the patient—GP consultation as currently found in countries with well-developed health services appears to be remarkably resistant to attempts to change in this particular respect. Other background issues closely related to this whole topic were recently reviewed in the Journal (Reference Chilvers, Harrison and SiposChilvers et al, 2002).

There are other ways of looking at this topic which anyone interested in planning further studies should consider. One way is to give more attention to the process of the patient—GP consultation, since up to now emphasis has been mainly upon measuring aspects of the input and the outcome. A similar way is to recommend what some would call a systems approach. To do this means stepping back from the immediate details and considering, in the complex of interactions that constitute an encounter between patient and GP, who does what to whom, and why. For instance, the GP has had a particular type of medical education and so has a personal concept of the functions of a GP; the content and the result of the consultation is also affected by what the patient wants, and what the patient thinks the GP can do and should do. All of these are also constrained by the time available for the consultation and its physical setting. To achieve any worthwile and lasting change in only one or two of the elements in this complex system would almost certainly require that some attention be given to related elements. The key question is which elements and what to do about them. No doubt the authors of some of the published studies have thought along these lines, and it is to be hoped that future work will be based more overtly upon this broader view of what exactly needs to be studied. For instance, one possibility would be to carry out smaller but more-intensive studies of why the psychiatrist and GP agree or disagree about the patient, and to examine whether this has any detectable effect upon the outcome.

All the studies discussed above have been performed in countries where the standards of general medical education are comparatively high, and medical services are well developed. However, we must remember that the ICD—10 PHC classification and guidelines were not developed with these prosperous countries particularly in mind. They are for the whole world, and in the majority of countries there are few psychiatrists and few trained GPs. In some countries most of the doctors and other health workers that are available have had little or no training in clinical psychiatry. As yet, nobody knows what effects the systematic provision and use of the ICD—10 PHC guidelines would have in those countries where they are likely to be most needed. Other and simpler educational packages are also available, such as the WPA/PTD Educational Program on Depressive Disorders (World Psychiatric Association, 1997).

To take an even wider viewpoint, it has been known for many years that the problem of low rates of detection of psychological problems and psychiatric disorders in primary care is an international one. Two notable and large-scale studies in point are the World Health Organization (WHO) Collaborative Study on Strategies for Extending Mental Health Care in four countries (Reference Harding, de Arango and BaltazarHarding et al, 1980), and the more-recent WHO Collaborative Study in 14 countries (Reference Sartorius and UstunSartorius & Ustun, 1994). Of the many possible reasons for the problem being so widespread and persistent, one is particularly interesting and challenging: is what psychiatrists regard as an underreporting of psychiatric disorders in primary and general health care actually best regarded as a troublesome side-effect of so-called modern scientific medical education? It seems quite possible that this is so, at least from the standpoint of the practice of primary care medicine, since this type of medical education is one of the few features common to all the centres that have reported studies of the problem. Any long-term remedies will concern the basic design of both undergraduate and post-graduate education of medical professionals, in preference to post hoc attempts to remedy the deficit. For the near future, all that can be done is to try different ways of remedial education. There is no doubt that efforts to improve the lot of the patients and families affected by depressive disorders are justified. The World Bank is not the sort of organisation that one would normally associate with statements about psychiatric issues but, in their 1993 report, the very significant contribution of depressive disorders to the overall burden on the economy of the world was clearly documented (Reference Broadhead, Blazer and GeorgeBroadhead et al, 1990; World Bank, 1993). Action of some sort is an obligation for the medical profession.