Blog

There's no ignoring a meta-analysis

Pharmaceutical companies should proceed with great care in deciding whether to conduct a meta-analysis, and the results should be used "ever so cautiously, if at all, in making decisions", a former industry exec believes.

"If you do a meta-analysis, it will count," Dr Ronald Krall told a recent US drug safety conference sponsored by the Drug Information Association, the FDA and PhRMA. One should think about how results will be interpreted before embarking on a meta-analysis, because once done, "you won't be able to ignore the answer".

Dr Krall is retired as chief medical officer of GlaxoSmithKline, a company whose products have been the subject of some recent negative meta-analyses, including the FDA's review of suicidality and anti-epilepsy drugs and Dr Steven Nissen's much-publicised review finding the diabetes drug Avandia to be associated with a significant increased risk of heart attacks.

Dr Krall and FDA officials cited the inherent difficulties and pitfalls in conducting a safety meta-analysis by combining results from trials of disparate populations and durations, using varying doses and analysing different indications and endpoints.

The contribution of a meta-analysis to decision-making will depend, Dr Krall said. If the outcome assessed is the same as in the component trials, and the underlying studies are homogenous in design with informative results and a large effect size, then the meta-analysis is descriptive but may not add much to the individual trials themselves, he said.

In contrast, where the underlying trials are heterogeneous in nature and not individually informative, and the effect size was small, then a meta-analysis may be informative. He also painted a third scenario, exemplified by the anti-epileptic/suicidality review, wherein the outcome assessed is not the same as the intent of the underlying trials, making the meta-analysis results difficult to interpret.

Conference speakers noted that the availability of public clinical trial results databases and health insurers' in-house claims databases make it far easier to conduct meta-analyses than ever before. And yet, there is not enough public discussion and consensus about the methods that should be used, FDA officials said.

"What we need to do ... is to get people together and develop best practices around these analyses," said FDA drugs centre director Dr Janet Woodcock. "We feel we’ll be better off if we all do these together in an open and transparent manner, and then not only publish the data and results, but publish the ... sensitivity analysis, what are the liabilities of these methodologies, rather than ... what’s happening now ¬is we get competing analyses published and people then argue about it in the press.

"We can’t run away from these things because they’re going to be done," Dr Woodcock said. "And the question is are we going to try to do these new analyses in an orderly manner and contribute to the advancement of scientific knowledge, or is this going to become some kind of academic food fight."

The FDA is developing a guidance document on safety meta-analyses that it hopes to issue by the end of 2009.

Posted 27/10/2008 15:36:23 PM
by Sue Sutter,
Washington Editor

Name
Comment

Rate this article - 1 2 3 4 5

Comments

No Comments Yet