Partisan Perspectives in the Medical Literature: A Study of High Frequency Editorialists Favoring Hormone Replacement Therapy
Unfavorable results of major studies have led to a large shrinkage of the market for hormone replacement therapy (HRT) in the last 6 years. Some scientists continue to strongly support the use of HRT.
We analyzed a sample of partisan editorializing articles on HRT to examine their arguments, the reporting of competing interests, the journal venues and their sponsoring societies.
Through Thomson ISI database, we selected articles without primary data written by the five most prolific editorialists that addressed clinical topics pertaining to HRT and that were published in regular journal issues in 2002–2008.
We recorded the number of articles with a partisan stance and their arguments, the number of partisan articles that reported conflicting interests, and the journal venues and their sponsoring societies publishing the partisan editorials.
We analyzed 114 eligible articles (58 editorials, 16 guidelines, 37 reviews, 3 letters), of which 110 (96%) had a partisan stance favoring HRT. Typical arguments were benefits for menopausal and related symptoms (64.9%), criticism of unfavorable studies (78.9%), preclinical data that showed favorable effects of HRT (50%), and benefits for major outcomes such as osteoporosis and fractures (49.1%), cardiovascular disease (31.6%), dementia (24.6%) or colorectal cancer (20.2%), but also even breast cancer (4.4%). All 5 prolific editorialists had financial relationships with hormone manufacturers, but these were reported in only 6 of the 110 partisan articles. Four journals published 15–37 partisan articles each. The medical societies of these journals reported on their websites that several pharmaceutical companies sponsored them or their conferences.
There is a considerable body of editorializing articles favoring HRT use and very few of these articles report conflicts of interest. Full disclosure of conflicts of interest is needed, especially for articles without primary data.