COMMENTARY

This is an important paper which should be highly visible. Many studies generate data that physicians or mental health providers do not put into practice. Some have suggested that physicians lack the knowledge required to do their job properly. Consequently, a multitude of guidelines have been developed to improve the behaviour of physicians and mental health providers. While conventional medical textbooks aim to help physicians understand the complexities of medicine and personality in the hope that this will lead to appropriate action, guidelines aim to directly instruct physicians what to do and how to do it. Guidelines are typically “evidence-based” and authored by “experts.” Guidelines now serve as one basis upon which to judge physicians.

There are growing concerns about the potential harms of guidelines. Some suggest that guidelines can interfere with the clinical competency of the treating physician, increase costs, lead to non-individualised treatments, be inappropriate quality monitoring criteria, and even harm patients.

Guidelines are intervention instruments, therefore their impact should be studied empirically. One could argue that, rather than being developed from the findings of high quality studies, a guideline is only “evidence-based” if it has been tested in a controlled trial comparing “guideline exposed” clinicians with controls. This is exactly what Croudace et al have done.

Most people with mental health problems present initially to primary care. Croudace et al exposed physicians to a guideline to assist in the identification of mental disorders. The guidelines did not increase the identification and treatment of mental disorders in primary care. There was a trend towards increased patient satisfaction, but worse patient outcome and quality of life.

There are a number of implications: first, this study should be used as a model when developing and testing future guidelines. All guidelines should be tested in this way; second, we need to think critically before chastising physicians for not following guidelines in depression and other areas. In this study, physicians who were not exposed to guidelines were as knowledgeable as those who were specially trained. Instead of criticising physicians for not following guidelines, researchers should clarify why physicians do what they do and why they may behave differently from guideline specifications. There is a large body of empirical evidence on medical decision making which could help to answer this question. Finally, regarding the problem of recognising depression, the question is whether a simple GHQ score is the correct basis on which to decide who should get medical attention. The poorer outcome in the intervention group could be explained by a greater focus on psychological problems in this group, leading patients to complain and therefore feel worse.

In summary, this paper may stimulate new thinking about guidelines, a new definition of “evidence-based” and high quality studies on this topic. Hopefully this study may act as a model for many other investigations of this nature.