Magazine article Clinical Psychiatry News

The Tribulations of Trials: How to Read the Studies. (Watch for Inconsistencies)

Magazine article Clinical Psychiatry News

The Tribulations of Trials: How to Read the Studies. (Watch for Inconsistencies)

Article excerpt

NEW YORK--In evaluating the relative efficacy of antidepressant medications, research studies should be a useful guide. But the unwary can be easily misled by inconsistencies in analysis and reporting, Dr. Norman Sussman said at a conference on psychopharmacology sponsored by New York University.

Differences in the disease severity at entry, the definition of recovery or remission, the relative dosages, and the method used to compute results can distort comparisons among studies of different agents, said Dr. Sussman, professor of psychiatry at the university.

Most medications approved for depression are considered roughly equivalent in efficacy but surveys suggest that psychiatrists generally perceive differences among them, with often idiosyncratic views derived from personal experience, marketing strategies, or word of mouth.

The literature can provide a broader basis for comparing the relative merits of different agents-if a critical eye is used.

Open-label studies are generally without value. "Everybody wants [the drug] to work," enhancing the likelihood of a substantial placebo effect, he said. In addition, most are too small to yield useful data.

Single randomized controlled trials are, for the most part, similarly underpowered for comparison purposes: For significant differences to be found among two active agents and placebo, 900-1,000 subjects have to be enrolled, Dr. Sussman said.

In addition, publication is selective: More than half of controlled trials of antidepressants are failed or negative, but 90% of published studies are positive. "You only see the good stuff," he said.

Thus comes the value of metaanalyses, which combine data from several studies. Ideally these should include pooled analysis of all original data, rather than just selection of studies, which can be biased. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.