Patient heterogeneity and the need for randomized clinical trials

Abstract

The hierarchy of strength of evidence concerning treatment efficacy ranges from individual anecdotal case reports, through studies with nonrandomized controls, to randomized clinical trials. The importance of randomized trials is their ability to convince the medical community. With randomization, bias (whether conscious or unconscious) is avoided, prognostic factors (both known and unknown) tend to be balanced across treatment groups, the validity of statistical tests of significance is guaranteed, and problems with time trends are avoided. A study of thyroid carcinoma demonstrates the importance of prognostic factors, and indicates how they can be identified using a Weibull model and used to form a prognostic index. A study of prostatic cancer illustrates how covariate information can be used to select optimal treatment for different patient subgroups in situations where there is no single best overall treatment. The key to successful clinical trials is a renewed effort in developing and selecting the treatments to be studied.