Background: Authors' conflicts of interest may affect the content of medical guidelines. In April 2010, the Association of Scientific Medical Societies in Germany (AWMF) issued recommendations on how such conflicts of interest should be dealt with. Most AWMF guidelines are so-called S1 guidelines developed by informal consensus in a group of experts. We now present the first study to date on the management of conflicts of interest in S1 guidelines.
Methods: On 2 December 2013, we selected the guidelines that had appeared from 1 November 2010 to 1 November 2013 among the 449 current S1 guidelines on the AWMF website. We extracted information about conflicts of interest from the guideline texts, reports, and/or conflict of interest statements and evaluated this information descriptively.
Results: There were 234 S1 guidelines in this category, developed by a total of 2190 experts. For 7% (16/234) of the guidelines and 16% (354/2190) of the experts, no individual conflict of interest statement could be found. Where conflict of interest statements were available, conflicts of interest were often declared--in 98% (213/218) of the guidelines and by 85% (1565/1836) of the authors. The most common type of conflict of interest was membership in a specialist society or professional association (1571/1836, 86%). Half of the experts acknowledged a financial conflict of interest (911/1836, 50%). Conflicts of interest were more common among experts contributing to guidelines that mainly concerned treatment with drugs or other medical products than in guidelines that did not have an emphasis of this type (397/663, or 60%, versus 528/1173, or 45%). The conflicts of interest were assessed in 11% (25/234) of the guidelines, with practical consequences in a single case.
Conclusion: Conflicts of interest are often declared in the S1 guidelines of the AWMF, but they are only rarely assessed by external evaluators. Clear rules should be issued for how experts' declared conflicts of interest should be acted upon, whether they are of a financial nature or not.