Get started
Work

Living Will

Year: 2012
Type: article
Abstract: 2558 correspondence which will depend, in part, on what one wishes to estimate.First, define the outcome in terms of information collected while the patient is receiving the assigned treatment; for example, count a discontinuation of treatment as a treatment failure.This option does not require outcome measurements after discontinuation, though it may have limitations in other respects.Second, continue to collect and use information about outcomes after treatment discontinuation.Measured outcomes may reflect effects of treatments administered after study-treatment discontinuation; the impact of this can be limited, or better understood, by specifying in the protocol "rescue" treatments to be used as part of the treatment regimen.Third, apply an analysis method that in effect imputes unobserved outcomes after treatment discontinuation.This option requires careful attention to assumptions underlying the imputations.As Hirman and Flyer point out, an analysis that assumes that data are missing at random and relies on data from patients who continued to receive treatment to impute outcomes in those who discontinued treatment can be problematic.Such an approach would estimate counterfactual outcomes (i.e., what might have happened had the patients who discontinued treatment actually continued), and its usefulness depends on whether the counterfactual outcomes are of interest.If patients who discontinue treatment revert to "standard care," an analysis that imputes observed outcomes with information from a group receiving "standard care" may be more relevant.Such an analysis can be facilitated by collecting outcomes data on some participants who discontinued treatment.A sensitivity analysis that models realistic departures from data that are missing at random may help to assess how robust the findings are with regard to differing reasonable assumptions regarding the missing data.
DOI: 10.1056/nejmc1207998
Author: Rena Xu
Language: English
Open Access status: bronze
OpenAlex ID: W4256636643
FWCI:
Cites:
Cited by:
Related to: 10