Reporting and interpreting adjuvant therapy clinical trials. International Breast Cancer Study Group (formerly Ludwig...

Identifying effective adjuvant treatments for patients with node-negative breast cancer is made difficult by the heterogeneity of the disease, the relatively low event rate and long follow-up time required, and the small magnitude of effects of current therapies. Several aspects of clinical trials that influence the appropriate reporting and interpretation of statistical results are discussed. We point out that the P value is a measure of the statistical uncertainty of an observed outcome and depends on the number of events available for analysis; it is not a measure of the magnitude of a treatment effect. We recommend that the relative reduction in the risk of an event and its 95% confidence interval be presented as an estimate of the treatment effect size, and that absolute improvements be used to judge whether treatment benefits outweigh the costs for the patient population. We suggest that subgroup analyses are important to define treatment effects within groups with different prognoses, and should be used with the understanding that multiple comparisons increase the chance of a false-positive result. Subgroup analysis should rely on the estimates of relative treatment effect and should avoid the use of the P value to declare incorrectly that "treatment is effective for one subgroup but not for another." We present the meta-analysis (overview) as a powerful method to demonstrate the statistical significance of a modest treatment effect by increasing the number of events available for analysis. The interpretation of overviews should consider the potential for treatment interactions and the validity of indirect comparisons that are not protected by a randomized design.(ABSTRACT TRUNCATED AT 250 WORDS)