Evidence should not be viewed in isolation.

After decades of negative multicenter, randomized, placebo-controlled, double-blind clinical trials in patients with critical illness, the last decade witnessed the publication of a series of studies demonstrating an improved outcome for a variety of interventions. These positive studies and interventions have subsequently been brought together into "bundles" of care. This rush to change clinicians' practice has often been with little regard for the context in which the evidence was generated and the problems created by the difference between efficacy and effectiveness. This methodology created a bypass mechanism through which implementation of a package of interventions could be fast-tracked to a much larger group of critically ill patients that differ significantly from those in whom the evidence was generated, without being validated as a bundle. The risk of this process is that the bundle tools will be used for issues such as quality control, performance evaluation, and legal and regulatory purposes without the necessary background of robust validation studies. This has resulted in a "two-weight, two-measures" situation that needs to be resolved to avoid the risk of taking interventions completely out of the context in which the positive results were generated and putting patients and professionals at risk of being harmed by the package of interventions.