Several readers have responded with comments and concerns regarding my column, "Do Physicians Use Practice Guidelines?" (Psychiatric Times, June 2006, page 13). The concerns center on the relevance of clinical practice guidelines (CPGs) to the practice of psychiatry; specifically, what makes a CPG more valid than other forms of research such as literature, meetings, and lectures, and how can physicians be sure that CPGs are free of bias from managed care and insurance companies? Since the issues these readers raised are important and concern many psychiatrists, they merit some discussion.
Expert consensus
Many CPGs are created through a process of expert consensus in which clinicians who specialize in studying a specific psychiatric illness are asked to rate the appropriateness of various treatments for the disorder. For example, in the development of a consensus guideline for the pharmacologic treatment of psychosis, 50 expert clinicians were asked to rate the appropriateness of 11 antipsychotic medications for the treatment of an initial episode of psychosis with predominantly positive psychopathology.1 Medications were subsequently designated as first-, second-, and third-line choices based on statistical analysis of the combined expert
ratings.
It is reasonable to question how these experts were selected and why they should be considered more qualified to draw conclusions from the research data than practicing psychiatrists who are, by definition, also experts.2 Clinicians and researchers, like those who create CPGs, often disagree among themselves about the meaning of research results and the relative effectiveness of the various treatments for a psychiatric disorder. Sometimes they cannot provide convincing evidence that certain treatments are better than others.
However, these problems don't necessarily invalidate the use of CPGs. In an ideal world, research would provide definitive answers about the effective treatment of various disorders. Every clinician would know this and could appropriately apply it in patient care.
Unfortunately, the world of psychiatric treatment is far from ideal. Research is a messy proposition that often produces tentative and inconclusive results that are open to various interpretations. If disagreement about research results were the criterion for rejection, science would make little progress in discovering successful treatments for medical disorders. In fact, science actually thrives on disagreement because it drives further research and exploration. The challenge for clinicians and researchers is to provide the best possible interpretation of the data while controlling for the cognitive biases that inevitably enter into any such judgment process.3,4
I don't know exactly how clinical experts are chosen, but I presume they are selected by their colleagues based on who has published the most relevant papers on the diagnosis and treatment of specific psychiatric disorders. There is certainly potential for bias in this process, but I cannot think of a better way to identify experts in the field. Selecting a large number of experts ensures that the guidelines will include the broadest possible knowledge about treating the disorder. It also provides a peer review process that decreases the likelihood of bias in the interpretation of research results. This does not mean that bias is extinguished—it is simply reduced.
Physicians often argue that it is sufficient to attend professional meetings, read the literature, listen to lectures, and consult with colleagues in order to draw their own conclusions about how to treat their patients. They maintain that they do not need anyone to tell them how to practice medicine. Yet practitioners, like experts, need to guard against bias in their own interpretation of clinical and research data. Individual clinical experts may be no more skillful than practicing psychiatrists in the management of a complex case, but the collective wisdom of a large number of experts is more likely to be correct than is the judgment of a single clinician.
An expert consensus panel is also a useful method for filling in the gaps in formal research. No study can address all of the questions associated with the clinical treatment of a disorder. As the number of treatment subgroups in a research study increases, a greater number of subjects will be required to provide the statistical power necessary to detect differences among the subgroups. Eventually, the number of subjects and the cost of the study will become too large to manage. An expert consensus guideline provides some direction to practitioners about treatments that are not included in the formal research.
