Nearly 70 years into the modern era of randomized clinical trials, we have learned a great deal about how to ensure valid results, such as establishing intention-to-treat analysis as the primary analytical approach, developing sequential designs to allow for interim decision-making while controlling error probabilities, and specifying primary outcomes prior to reviewing comparative data. We have also developed designs to meet new needs, such as cluster-randomized trials, adaptive and multi-stage designs to prioritize agents in drug development, and SMART designs to evaluate strategies for treatment of chronic conditions requiring regular modification of treatment regimens. At the same time, we continue trying to improve our approaches to long-standing challenges such as validation of surrogate endpoints and handling missing data.
But the clinical trials community faces many new challenges. Among the most widely discussed new issues are those related to precision medicine. The concept of optimizing treatment based on individual patient characteristics has long been a goal, motivating subset analyses and use of analytical tools such as testing for treatment by covariate interaction, but the new area of genomics, bringing the hope of vastly increased capability of targeting treatment to very specific subpopulations, has and will continue to lead to new approaches to study designs in both early and late phases of development.
Another emerging issue is the serious effort to reduce the costs and burdens of conducting trials. Pragmatic trials have been advocated to produce “real-world data,” that can perhaps better inform public health policies as well as reducing the costs and burdens of clinical trials by limiting data collection to key outcomes and leaving many aspects of the trial, such as patient entry criteria and use of concomitant medications, to the discretion of treating physicians. In addition, recent changes in some U.S. regulations, such as requirements for use of a single IRB of record for multicenter trials, are intended to simplify conduct of trials funded by federal health agencies. Clinical trialists will need to navigate these new policies and approaches, learning what types of trials best meet specific needs and how to handle new issues such as finding appropriate approaches to informed consent in cluster-randomized trials.
Finally, we face attacks on fundamental assumptions underlying our approaches to determining optimal treatment strategies. A medical journal editor has declared war on statistical inference, insisting that all inferential approaches are suspect; others argue that the only villain is classical hypothesis testing and reliance on p-values. Perhaps of more serious concern are calls for greatly increased access to experimental treatments at early stages of drug development, and even removing the efficacy requirement for drug approval. Clinical trialists will have to take on the challenge of effectively communicating the importance of rigorous research to individual and public health.