Abstract

Missing data due to loss to follow-up or intercurrent events are unintended, but unfortunately inevitable in clinical trials. Since the true values of missing data are never known, it is necessary to assess the impact of untestable and unavoidable assumptions about any unobserved data in sensitivity analysis. This tutorial provides an overview of controlled multiple imputation (MI) techniques and a practical guide to their use for sensitivity analysis of trials with missing continuous outcome data. These include δ- and reference-based MI procedures. In δ-based imputation, an offset term, δ, is typically added to the expected value of the missing data to assess the impact of unobserved participants having a worse or better response than those observed. Reference-based imputation draws imputed values with some reference to observed data in other groups of the trial, typically in other treatment arms. We illustrate the accessibility of these methods using data from a pediatric eczema trial and a chronic headache trial and provide Stata code to facilitate adoption. We discuss issues surrounding the choice of δ in δ-based sensitivity analysis. We also review the debate on variance estimation within reference-based analysis and justify the use of Rubin's variance estimator in this setting, since as we further elaborate on within, it provides information anchored inference.

Highlights

  • Since the true values of missing data are never known, it is necessary to assess the impact of untestable and unavoidable assumptions about any unobserved data in sensitivity analysis. This tutorial provides an overview of controlled multiple imputation (MI) techniques and a practical guide to their use for sensitivity analysis of trials with missing continuous outcome data

  • When missing data occurs complexity arises, since any statistical analysis necessarily makes an untestable assumption about the distribution of the unobserved

  • We have described and illustrated how sensitivity analysis can be conducted to explore departures from an MAR assumption for unobserved continuous outcome data using controlled MI

Read more

Summary

N T RO DU CT ION

In late-phase clinical trials, loss to follow-up and intercurrent events—such as treatment withdrawal or partial compliance—are almost inevitable. This controlled MI method is appealing since it avoids having to specify any numerical sensitivity analysis parameters; only qualitative assumptions are required In any trial, such as adapt or the acupuncture trial, we can only begin to think about missing data once we know the precise treatment effect we are aiming to estimate. The strong MCAR assumption is often unlikely to be valid in the clinical trial context where drop out may be effected by treatment and observed responses This is likely in longitudinal settings when data are missing due to uncontrollable events such as receipt of rescue medication, since these events are often associated with the study variables. For imputation k, using uninformative priors, draw β0k , β1k , β 2 and σ 2.1 from the Bayesian posterior of β0 , β1 , β 2 and σ 2.1 , k

Draw missing data from
Method
DISCUSSION
CONFLICT OF INTEREST
Peer review trial
Findings
Analysis
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call