Intensive care medicine in 2050: clinical trials designs

Excerpt

For a long time, clinical trials have been designed in a fairly standard way. In phase III randomized clinical trials (RCTs), widely considered the top of the evidence pyramid, each patient typically has a 1:1 chance of being allocated to the experimental or the control treatment. The scheme involves a large number of patients, often due to the modest expected benefits, the so-called “effect size”. Such a factor has the greatest impact on both sample size and power computations, and it has been reported to explain the failure of most RCTs in critical care medicine where the desired effect size (10.1% on average) is often largely above the observed one (1.4% on average [1]). Even large sample sizes are no guarantee of positive findings, with small observed benefits possibly explained by several problems, such as protocol deviations or unscheduled crossovers diluting treatment effects [2] as well as true negative findings. In addition, the large sample sizes required in RCTs may limit their feasibility due to excessive cost and/or time. Effectiveness of clinical trials might be improved by adopting a more integrated approach which increases flexibility and maximizes the use of accumulated knowledge to reduce sample sizes and to increase effect sizes for selected patient subpopulations. …