Article type
Year
Abstract
Background: Theoretical considerations suggest that randomized clinical trials (RCTs) stopped early for benefit may systematically overestimate treatment effects.
Objective: To assess the prevalence of RCTs stopped early for benefit, and the quality of reporting of methods to inform the decision to truncate the trial.
Methods: We searched MEDLINE, EMBASE, and Current Contents in August 2002 and full text databases in February of 2003 and identified RCTs stopped early for benefit. We documented journal and year of publication, reporting of methods and funding, planned sample size, number and periodicity of interim analyses, stopping rules, and effect size.
Results: We found 98 RCTs stopped early for benefit, most of them (69) published in 5 top medical journals. Typically, these were industry-funded drug trials in cardiology, cancer, and HIV/AIDS. The proportion of all RCTs published in journals contributing at least one eligible RCT that was stopped early for benefit increased from 0.2% between 1988-1992 to 2.1% between 1998-2002 (P for trend <.001). Prestigious journals were overrepresented among journal publishing truncated RCTs - 69 of 98 RCTs were published in the top 5 medical journals. Most RCTs recruited approximately 60% of the planned sample and stopped after a median of 6 months of follow-up, one interim analysis, and a median of 84 patients had experienced the endpoint of primary interest. The median risk ratio for treatment effect among truncated RCTs was 0.54 (interquartile range 0.32-0.69). 54 of 98 truncated RCTs did not report at least one of the following: the planned sample size, the interim analysis after which the trial was stopped, or whether or not a stopping rule informed the decision. Trials with fewer events yielded larger treatment effects (OR 18.9, 95% CI 5.9-60.5).
Conclusions: RCTs stopped early for benefit are becoming more common. The reporting of methods to inform the decision to truncate the trial is often deficient. Clinicians should interpret with caution RCTs stopped early for benefit, particularly those with poor reporting and those that report implausibly large treatment effects, and issues around early stopping should be included in the methodologic quality assessment of trials in Cochrane reviews.
Objective: To assess the prevalence of RCTs stopped early for benefit, and the quality of reporting of methods to inform the decision to truncate the trial.
Methods: We searched MEDLINE, EMBASE, and Current Contents in August 2002 and full text databases in February of 2003 and identified RCTs stopped early for benefit. We documented journal and year of publication, reporting of methods and funding, planned sample size, number and periodicity of interim analyses, stopping rules, and effect size.
Results: We found 98 RCTs stopped early for benefit, most of them (69) published in 5 top medical journals. Typically, these were industry-funded drug trials in cardiology, cancer, and HIV/AIDS. The proportion of all RCTs published in journals contributing at least one eligible RCT that was stopped early for benefit increased from 0.2% between 1988-1992 to 2.1% between 1998-2002 (P for trend <.001). Prestigious journals were overrepresented among journal publishing truncated RCTs - 69 of 98 RCTs were published in the top 5 medical journals. Most RCTs recruited approximately 60% of the planned sample and stopped after a median of 6 months of follow-up, one interim analysis, and a median of 84 patients had experienced the endpoint of primary interest. The median risk ratio for treatment effect among truncated RCTs was 0.54 (interquartile range 0.32-0.69). 54 of 98 truncated RCTs did not report at least one of the following: the planned sample size, the interim analysis after which the trial was stopped, or whether or not a stopping rule informed the decision. Trials with fewer events yielded larger treatment effects (OR 18.9, 95% CI 5.9-60.5).
Conclusions: RCTs stopped early for benefit are becoming more common. The reporting of methods to inform the decision to truncate the trial is often deficient. Clinicians should interpret with caution RCTs stopped early for benefit, particularly those with poor reporting and those that report implausibly large treatment effects, and issues around early stopping should be included in the methodologic quality assessment of trials in Cochrane reviews.
PDF