Views 3,572
Citations 0
June 7, 2016

The Potential for Postrandomization Confounding in Randomized Clinical Trials

Author Affiliations
  • 1Division of Preventive Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
  • 2Barbra Streisand Women’s Heart Center, Cedars-Sinai Heart Institute, Los Angeles, California
  • 3Departments of Biostatistics and Epidemiology, Harvard T. H. Chan School of Public Health, Boston, Massachusetts

Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.

JAMA. 2016;315(21):2273-2274. doi:10.1001/jama.2016.3676

Randomized clinical trials (RCTs) are considered the “gold standard” for evaluating an intervention’s efficacy and safety.1 Although large-scale RCTs are nearly always free of baseline confounding because randomization balances the distribution of measured and unmeasured risk factors across treatment groups, biases may emerge after randomization because of differential use of nonstudy medication or treatments, imbalanced rates of disease screening, differential loss to follow-up, and other differences between treatment groups. Such postrandomization “confounding” is more likely to occur in long-term trials or pragmatic trials taking place in typical patient care settings. Because community-based long-duration pragmatic trials are increasingly being used to evaluate sustained interventions, it is important for researchers to use available methodological approaches to assess the presence of such biases1 and for clinicians to be aware of potential sources of confounding when interpreting results of RCTs.

First Page Preview View Large
First page PDF preview
First page PDF preview