In Clinical Trials

The use of prognostic factors in clinical trials usually corresponds to three purposes: study definition, stratification before randomization, and analysis of the trial results. In the definition of the study, prognostic factors are used as entry or exclusion criteria. This attitude is often based on ethical considerations since eligible patients are those for whom the benefits and risks of treatments are uncertain enough to justify randomization. It is the case in deescalation trials or in those in which better results are expected from more aggressive therapy. However, the use of prognostic factors in selecting patients may bias the trial recruitment since patient inclusion lies with the responsible local investigator who could decide to exclude borderline cases from the trial. Even central randomization does not prevent such a bias. Therefore, prognostic factors must be simple, clearly defined, easy to assess, and well accepted by all investigators participating in the trial.

Randomization is the only method to ensure comparability of the treatment groups concerning known and unknown factors. Currently randomization techniques provide adequate distribution between treatment groups especially when large numbers of patients are included. In small randomized trials, stratified randomization is designed to balance treatment allocation within predefined subgroups. However, the number of strata should be limited to avoid overstratification.

Prognostic factors not only serve in the description of the study population, they also play a role in analysis of the trial results. The omission of important known or unknown prognostic factors may bias the estimate of the treatment effect or reduce the statistical power.45 The analysis of the results of a trial should include both univariate and multivariate tests for treatment effect. Consistent results will make the conclusion more convincing.

0 0

Post a comment