Challenges of Non–Intention-to-Treat Analyses

Over the past 5 decades, the randomized clinical trial has become the gold standard for evaluation of the risks and benefits of new interventions, including drugs, medical devices, and surgical procedures. To justify the use of randomization, it is important to note that in a nonrandomized study com...

Full description

Saved in:
Bibliographic Details
Published inJAMA : the journal of the American Medical Association Vol. 321; no. 2; pp. 145 - 146
Main Authors DeMets, David L, Cook, Thomas
Format Journal Article
LanguageEnglish
Published United States American Medical Association 15.01.2019
Subjects
Causation
Clinical trials
Medical devices
Medical electronics
Medical equipment
Medical treatment
Statistical analysis
Surgery
Online AccessGet full text

Cover

More Information
Summary:Over the past 5 decades, the randomized clinical trial has become the gold standard for evaluation of the risks and benefits of new interventions, including drugs, medical devices, and surgical procedures. To justify the use of randomization, it is important to note that in a nonrandomized study comparing 2 interventions, a small P value (of which P < .05 is generally considered statistically significant) for a statistical comparison between groups can be due to 1 of 3 sources: chance, causation, or confounding. Because randomized assignments cannot be associated with participant characteristics, effective randomization eliminates the third possibility, enabling direct assessment of potential causal relationships provided that the study is designed, conducted, and analyzed properly. Proper trial conduct and analysis include ensuring complete follow-up and unbiased ascertainment of the outcomes of interest among all randomized participants.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0098-7484
1538-3598
DOI:10.1001/jama.2018.19192