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

Statistical considerations in the intent-to-treat principle.

J M Lachin1

  • 1The Biostatistics Center, Department of Statistics, The George Washington University, Rockville, MD 20852, USA.

Controlled Clinical Trials
|May 24, 2000
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

Effect of GLP-1 and GIP on C-peptide secretion after glucagon or mixed meal tests: Significance in assessing B-cell function in diabetes.

Diabetes/metabolism research and reviews·2017
Same author

Application of robust estimating equations to the analysis of quantitative longitudinal data.

Statistics in medicine·2001
Same author

A flexible stochastic curtailing procedure for the log-rank test.

Controlled clinical trials·2000
Same author

Worst-rank score analysis with informatively missing observations in clinical trials.

Controlled clinical trials·1999
Same author

Group sequential monitoring of distribution-free analyses of repeated measures.

Statistics in medicine·1997
Same author

The efficacy of tolrestat in the treatment of diabetic peripheral neuropathy. A meta-analysis of individual patient data.

Diabetes care·1996

The intent-to-treat (ITT) design in clinical trials ensures unbiased treatment comparisons by analyzing all randomized patients. Post-randomization exclusions, unlike ITT, can severely inflate Type I error rates, leading to false positive results.

Area of Science:

  • Biostatistics
  • Clinical Trial Design
  • Epidemiology

Background:

  • Randomization is crucial for unbiased clinical trial comparisons, but it's insufficient alone.
  • Ensuring all randomized patients contribute to analysis, or that missing data are ignorable, is vital for unbiased treatment effect assessment.
  • The intent-to-treat (ITT) design, following all randomized subjects regardless of treatment adherence, provides a sufficient condition for unbiased comparison.

Purpose of the Study:

  • To describe statistical considerations for intent-to-treat (ITT) design and analysis in clinical trials.
  • To contrast ITT with efficacy subset analysis, highlighting potential biases from post-randomization exclusions.
  • To discuss the impact of subset selection bias on Type I error probabilities and power.

Main Methods:

Related Experiment Videos

  • Comparison of intent-to-treat (ITT) analysis with efficacy subset analysis.
  • Description of potential biases and inflation of Type I error probabilities due to post-randomization exclusions.
  • Discussion of statistical assumptions for censored or incomplete data and their limitations in addressing subset selection bias.

Main Results:

  • Post-randomization exclusions in efficacy subset analyses can severely inflate Type I error rates, potentially exceeding 0.50 even under the null hypothesis.
  • Standard methods for incomplete data do not adjust for bias introduced by post hoc subset selection, as these models are untestable.
  • The ITT analysis can paradoxically have greater power than efficacy subset analysis, especially when treatments have lasting effects.

Conclusions:

  • The intent-to-treat (ITT) design is essential for unbiased clinical trial comparisons, preventing bias from post-randomization exclusions.
  • Efficacy subset analyses introduce significant bias and can severely inflate Type I error rates, compromising trial validity.
  • ITT analysis offers advantages in power, particularly for treatments with sustained benefits, and may be more powerful than subset analyses in specific scenarios.