Abstract:
|
The topic of estimands has grown in importance dramatically since the issuance of the National Research Council report on "The Prevention and Treatment of Missing Data in Clinical Trials" (2010). It is so fundamental to clinical trials that it has sparked many conversations and papers on the nature and purpose of clinical trials themselves and philosophical debates about what is/are the most appropriate estimates of treatment effect derived therefrom. While this issue is largely motivated from the bias or confounding of patient outcomes when they discontinue study treatment, estimands do cause us to reflect deeply on the interpretation of statistical analyses of incomplete data and how the resulting conclusions should be communicated to regulators, payers, prescribers and patients. This talk will challenge some traditional orthodoxy about clinical trials - double-blind, randomized, placebo, controlled trial followed by ITT analysis - and focus on practically relevant estimands that may be of greatest interest to the stakeholders involved.
|