Session Item

A benchmarking exercise in clinical trials is a simple tool designed to ensure protocol compliance among trial participants before recruitment start. An anonymous patient dataset is distributed to all participants for delineation and planning; the final plan is then peer reviewed to assess the understanding of the protocol mandates by all trial participants.

This talk will cover the most common methods used to analyze benchmarking datasets and how proper analysis of the results can lead to much more than a simple performance measure:

•    Benchmarking can reduce interobserver variability across investigators

•    It can reduce interobserver variability across peer reviewers

•    It can estimate the effect of variations on patient outcome and sample

•    It highlights ambiguous or unrealistic protocol requirements (e.g. too strict OAR constraints)

•    Finally, it can help disseminating new or complex techniques.

Benchmarking is an important and powerful tool available to trialists, and its potential should be well understood to be used properly.