25 March 2010
The authors of the study say researchers need to resist pressures to end clinical trials early
From time to time, researchers stop clinical trials because the results seem almost too good to be true. A new international study shows that is probably because those results are not true.
When researchers end a clinical trial early because the results are better than expected, it allows a drug to get on the market and patients to receive it sooner than if the trial were carried out as planned.
It sounds like a good thing, but it may not be.
A new international analysis of 100 clinical trials that ended early found that the results were often wrong and sometimes life-threatening.
Dr. Victor Montori of the Mayo Clinic is one of the authors of the report. "What is happening is they are catching the data at a random high. The data is accumulating and it looks like a big effect, but if you let it go a little bit, that effect may become smaller over time," he explained.
Or, the benefit of the treatment may not even exist.
"Not only were the studies misleading, in terms of how big the treatment effect was really, but also whether a treatment effect existed at all," Dr. Montori said.
The clinical trials that Dr. Montori and his colleagues reviewed ended early because the results of an experimental treatment were much better than those of an existing therapy.
An example is a study on beta-blockers, drugs often prescribed for high blood pressure. Beta blockers slow the heart rate and can prevent heart attack during surgery. An initial trial was stopped early because these drugs looked to be highly effective if given prior to surgery. The drugs seemed to improve survival rates. But a larger trial showed that some patients given the drugs had significantly higher death rates.
Dr. Montori and other researchers compared the 100 trials that were called off early with more than 400 comparable trials that went through to completion. They found that the results were especially misleading in the smaller trials that ended early.
"What the investigators are caught on is when they look at the data at a given time before the study was supposed to finish, and they find a large treatment effect, then they get mislead by that treatment effect, and then in fact, if they decide to stop at that point, they end up misleading everyone else when they publish the results," he said.
Dr. Montori says the false findings discourage other researchers from repeating the study to see if they get the same positive results.
The authors of the study say researchers need to resist pressures to end clinical trials early. They say this will prevent patients and physicians from making treatment choices based on inaccurate information, or even worse, choosing one treatment when another would be far better.
