Conflicts of interest and the quality of recommendations in clinical guidelines.

BACKGROUND There is increasing concern that conflicts of interest affect the development process of clinical practice guidelines. We evaluated The American Psychiatric Association's Practice Guideline for the Treatment of Patients with Major Depressive Disorder to determine the existence of financial and intellectual conflicts of interest and examine their possible effects. We selected this guideline because of its influence on clinical practice and because this guideline recommends pharmacotherapy for all levels of depression, despite controversies over the evidence base. METHODS AND FINDINGS We determined the number and type of financial conflicts of interest for members of the guideline development group as well as for the independent panel charged with mitigating any effect of these conflicts. We also quantified the potential for intellectual conflicts of interest. We examined the quality of references used to support recommendations, as well as the degree of congruence between the research results and the recommendations. Fewer than half (44.4%) of the studies supporting the recommendations met criteria for high quality. Over one-third (34.2%) of the cited research did not study outpatients with major depressive disorder, and 17.2% did not measure clinically relevant results. One-fifth (19.7%) of the references were not congruent with the recommendations. Financial ties to industry were disclosed by all members (100%) of the guideline development committee with members reporting a mean 20.5 relationships (range 9-33). The majority of the committee participated on pharmaceutical companies' speakers' bureaus. Members of the independent panel that reviewed the guidelines for bias had undeclared financial relationships. As a marker of intellectual conflict of interest, 9.1% of all cited research and 13% of references supporting the recommendations were co-authored by the six guideline developers. CONCLUSIONS The prevalence of conflicts of interest among panel members was high. The quality of the evidence cited raises questions about the validity of the recommendations. Attention to the quality of cited studies and to the risk of bias resulting from conflicts of interest should be a priority for guideline development groups.