Abstract:
|
Recent research has fostered new guidance on preventing and treating missing data. Consensus exists that clear objectives should be defined along with the causal estimands, trial design and conduct should maximize adherence to the protocol specified interventions, and a sensible primary analysis should be paired with plausible sensitivity analyses. An estimand is simply what is to be estimated. Two general categories of estimands are: effects of the drug as actually taken (de-facto) and effects of the drug if taken as directed (de-jure). De-jure and de-facto estimands each have strengths and limitations. An iterative process including objectives, estimands, design, analysis, and sensitivity analyses can be used to guide protocol development. Objectives should reflect the diverse needs of regulators, payers, prescribers, patients, care givers, sponsors, and other researchers. Although design and analysis considerations should not dictate choice of estimand, these considerations should not be ignored. For example, maximizing adherence reduces sensitivity to missing data assumptions for de-jure estimands, but may reduce generalizability of results for de-facto estimands if the methods used to maximize adherence in the trial are not feasible in clinical practice. Both de-jure and de-facto estimands are often needed to understand drug benefit and de-jure estimands will often be the focus of safety evaluations. Newer approaches such as reference based controlled imputation provide useful options for analyses of de-facto estimands. A sequential testing approach starting with a de-jure estimand(s) followed by a de-facto estimand(s) may be useful in assessing drug benefit.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.