NEW YORK (Reuters Health) - Claims derived from observational epidemiologic studies often continue to be supported in the medical literature despite strong contradictory evidence from randomized controlled trials, investigators report in the Journal of the American Medical Association for December 5.

"Very strong resistance exists in a considerable section of the specialist literature to accept that some promising findings have been refuted, even with very large-scale, well-conducted randomized trials," Dr. John P. A. Ioannidis told Reuters Health.

"A very large number of counter-arguments are raised to defend the original claims at all cost," he added. "When a long time passes from the refutation, this specialist literature even avoids mentioning the contradicting randomized trials, continuing the same line of research in a selective vacuum of evidence."

Dr. Ioannidis and associates at the University of Ioannina School of Medicine in Greece conducted a "citation content analysis" for two observational studies published in 1993 that proposed major cardiovascular benefits associated with vitamin E. The two articles have been cited more than 2600 times. In 2000, the randomized, controlled HOPE trial contradicted the benefits of vitamin E.

The investigators evaluated 165 articles, published in 1997, 2001, and 2005, that cited the two epidemiology papers, categorizing their stance on vitamin E for cardiovascular disease prevention.

"The proportion of unfavorable articles increased from 1.9% in 1997 to 14.3% in 2001 and to 33.9% in 2005," they report. "Even among articles that cited the contradicting HOPE trial ... the majority in 2005 still could not conclude that vitamin E was ineffective."

Common arguments authors used to

justify their position included claims of selection bias, characteristics of the study participants (genetic background, lifestyle, dietary habits, stage of disease), form and dose of vitamin E, and duration of follow-up.

Specialty journals were far more likely than general/internal medicine journals to maintain a favorable response to the putative benefits of vitamin E.

"It is very difficult to promote in a general journal a theory that has been refuted repeatedly by large randomized trials, because of the danger of influencing clinical practice and harming patients," Dr. Ioannidis said. On the other hand, "specialist journals may not be interested in the clinical application or don't see the practical consequences. Instead, they see (the original theory) as a continuation of an interesting biological speculation."

To examine the generalizability of their findings, the investigators conducted similar citation analyses for observational studies supporting the benefits of beta-carotene on cancer (published in 1981, refuted in the mid-1990s) and of estrogen on Alzheimer disease (published in 1996, refuted in 2004), and observed similar results. In 2006, more than 60% of articles referencing the original epidemiologic studies were still favorable.

Observational studies provide useful data and are valuable for generating hypotheses, Dr. Ioannidis pointed out. "The problem is more an issue of interpretation of the results and how strong the interpretation is, especially when we have very strong evidence from randomized trials."

"Better communication between evidence-based medicine principals and specialists with highly circumscribed research interests may help achieve a more judicious interpretation of research findings."

JAMA 2007;298:2517-2526.