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

Adaptive regression.

Michael A Proschan1, Eric Leifer, Qing Liu

  • 1National Heart, Lung, and Blood Institute, Bethesda, Maryland, USA. ProschaM@mail.nih.gov

Journal of Biopharmaceutical Statistics
|July 19, 2005
PubMed
Summary
This summary is machine-generated.

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

Proteomics risk scores and mortality in heart failure: Generalizability across populations.

PloS one·2026
Same author

Donor-Specific Transplant Outcomes from BMTCTN 1702: A Multi-Center Prospective Biological-Assignment Trial.

Blood advances·2026
Same author

A Single Dose Study of Pembrolizumab in People With HIV Infection Who are Immunologic Nonresponders.

Open forum infectious diseases·2026
Same author

Enhanced accuracy of NIRS-vascular occlusion testing through incorporation of conduit artery diameter.

JPhys photonics·2026
Same author

Response-adaptive randomization with imperfect intermediate endpoints.

Clinical trials (London, England)·2026
Same author

Event-driven planning of two-armed trials with a binary endpoint.

Clinical trials (London, England)·2026
Same journal

Correction.

Journal of biopharmaceutical statistics·2026
Same journal

Leveraging external controls in clinical trials: estimands, estimation, assumptions.

Journal of biopharmaceutical statistics·2026
Same journal

Special issue of nonclinical statistics in regulatory applications guest editors' notes.

Journal of biopharmaceutical statistics·2026
Same journal

Comparison of flexible parametric modeling and nonparametric methods to estimate restricted mean survival time: A simulation study.

Journal of biopharmaceutical statistics·2026
Same journal

Simulated treatment comparisons with jackknife pseudo values for estimating population-adjusted marginal treatment effects.

Journal of biopharmaceutical statistics·2026
Same journal

Sample sizes for randomized controlled trials utilizing Bayesian response adaptive randomization for continuous outcomes.

Journal of biopharmaceutical statistics·2026
See all related articles

Adjusting for prognostic covariates in clinical trials improves precision but can be counterproductive if overused. This study examines valid methods for covariate selection to balance precision and avoid bias.

Area of Science:

  • Biostatistics
  • Clinical Trial Design
  • Statistical Analysis

Background:

  • Adjusting for prognostic covariates in clinical trials enhances statistical precision and corrects for random treatment imbalances.
  • Over-inclusion of covariates, particularly those selected based on post-randomization data, can introduce bias and reduce the validity of trial results.
  • Methods like stepwise regression, relying on outcome correlation or treatment imbalance, are commonly proposed for covariate selection but their validity is questioned.

Purpose of the Study:

  • To critically evaluate the validity of covariate selection methods in clinical trials that utilize post-randomization data.
  • To investigate whether common covariate inclusion criteria (outcome correlation, treatment imbalance) are statistically sound.
  • To identify and propose valid analytical approaches for handling prognostic covariates in randomized controlled trials.

Related Experiment Videos

Main Methods:

  • Review and critique of existing covariate selection strategies, including stepwise regression based on post-randomization data.
  • Theoretical examination of the statistical implications of selecting covariates based on outcome association or treatment group differences.
  • Exploration of alternative, statistically valid methods for covariate adjustment in the presence of potential selection bias.

Main Results:

  • The validity of covariate selection methods based on post-randomization data, such as strength of correlation with outcome or degree of treatment imbalance, is questionable.
  • Stepwise regression and similar data-dependent covariate selection approaches can lead to biased estimates and inflated Type I error rates.
  • The study highlights the need for pre-specified covariate adjustment strategies to maintain the integrity of clinical trial analyses.

Conclusions:

  • Covariate selection based on post-randomization data is generally not a valid approach for clinical trial analysis.
  • Reliance on outcome correlation or treatment imbalance for covariate inclusion can compromise the reliability of trial findings.
  • Pre-specification of prognostic covariates before data analysis is crucial for valid and unbiased results in clinical trials.