Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

On the multiple imputation variance estimator for control-based and delta-adjusted pattern mixture models.

Yongqiang Tang1

  • 1Shire, 300 Shire way, Lexington, Massachusetts, U.S.A.

Biometrics
|April 14, 2017
PubMed
Summary

Sensitivity analyses using pattern mixture models (PMMs) in clinical trials with non-ignorable dropout require careful variance estimation. Delta-adjusted PMMs offer approximately valid inference, unlike control-based PMMs, with a proposed variance estimator.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Sample size calculation for the Andersen-Gill model comparing rates of recurrent events.

Statistics in medicine·2022
Same author

Complex survival trial design by the product integration method.

Statistics in medicine·2021
Same author

MOVER confidence intervals for a difference or ratio effect parameter under stratified sampling.

Statistics in medicine·2021
Same author

Sample size calculation for recurrent event data with additive rates models.

Pharmaceutical statistics·2021
Same author

Comments on "Equal-tailed confidence intervals for comparison of rates".

Pharmaceutical statistics·2021
Same author

Some new confidence intervals for Kaplan-Meier based estimators from one and two sample survival data.

Statistics in medicine·2021

Area of Science:

  • Biostatistics
  • Clinical Trials Methodology
  • Longitudinal Data Analysis

Background:

  • Pattern mixture models (PMMs) and delta-adjusted PMMs are sensitivity analyses for clinical trials with non-ignorable dropout.
  • Standard multiple imputation (MI) inference using Rubin's variance estimator can be biased when imputation and analysis models are uncongenial.

Purpose of the Study:

  • To quantify the bias of Rubin's variance estimator in control-based and delta-adjusted PMMs for longitudinal continuous outcomes.
  • To propose a variance estimator for asymptotically valid MI inferences in PMMs.

Main Methods:

  • Derivation of analytic expressions for MI treatment effect estimators and Rubin's variance for PMMs and MMRM with infinite imputations.
  • Comparison of a proposed variance estimator with bootstrap variance.
Keywords:
Control-based pattern mixture modelDelta-adjusted imputationMissing not at randomMixed effects model for repeated measuresRubin's variance estimatorUncongeniality

Related Experiment Videos

  • Illustration using an antidepressant trial dataset and a simulation study.

    • Main Results:

    • Asymptotic bias of Rubin's variance estimator is small in delta-adjusted PMMs but can be sizable in control-based PMMs.
    • Inference based on Rubin's rule is approximately valid in delta-adjusted PMMs.
    • The proposed variance estimator ensures asymptotically valid MI inferences.

    Conclusions:

    • Delta-adjusted PMMs provide more reliable sensitivity analyses in clinical trials with non-ignorable dropout compared to control-based PMMs.
    • The proposed variance estimator is effective for ensuring valid multiple imputation inferences in pattern mixture models.
    • The findings are supported by both theoretical derivations and empirical evaluations.