Choosing outcomes for clinical trials: a pragmatic perspective

Abstract

Purpose of review: The turn to evidence-based medicine in critical care has lead to a dramatic increase in the number of randomized clinical trials. Yet many of these trials are not showing differences on an appropriately prespecified primary outcome. In light of this, there have been some heated arguments as to what sorts of clinical trials should be conducted. We synthesize pragmatic recommendations from two governing bases: rigorous statistical practice and a commitment to insuring trials provide information that should help guide patient-centered decision-making.

Recent findings: We suggest six principles for the selection of an randomized clinical trial primary outcome: (1) your intervention should plausibly change your primary outcome. (2) The primary outcome should be cared about by your audience. (3) The primary outcome should be relevant to patients. (4) The primary outcome should be measured efficiently. (5) The primary outcome should be measured reliably. (6) When possible, use a well understood continuous scale. In addition, principles for selecting secondary outcomes are described.

Summary: It may be of value, when proposing a trial, to present preliminary evidence documenting the extent to which a proposed primary outcome actually accords each of these principles.