A pair of articles in the Annals of Internal Medicine discuss from opposing viewpoints the ethical aspects of stopping clinical trials early because interim monitoring suggests that one treatment has significant benefits over the other.
This practice has increased considerably, more than doubling since 1990.
One group of authors takes a wide view, suggesting that early cessation reduces scientific validity and violates the ethical requirements of social value and a favourable risk-to-benefit ratio. They note that truncated trials overestimate treatment effects, and where the number of events is small this overestimate may be large.
They suggest researchers ensure that a large number of events is accrued before the trial is stopped, and continue recruitment and follow-up for long enough after the cessation rule is satisfied to ensure that the trend continues. Overall, these authors suggest that stopping trials early for apparent benefit is often unethical and can be justified only under restricted circumstances.
The author holding the opposing viewpoint notes that the task of data monitoring committees is very difficult. He argues that if appropriate cessation boundaries are used, the resulting overestimate should be small, and that the use of Bayesian techniques – where previous knowledge is applied to generate a range of plausible outcomes – can minimise the risk of accepting wildly incorrect results.
He suggests that such methods formalise the “evidence-based skepticism” promoted in the other paper.
Ann Intern Med 2007;146:878-81;882-7