0
Correspondence |

Response FREE TO VIEW

Bart J. Harvey, MD, PhD; Thomas A. Lang, MA
Author and Funding Information

From the Department of Family and Community Medicine, and the Department of Surgery (Dr Harvey), Dalla Lana School of Public Health, University of Toronto; and Tom Lang Communications and Training (Mr Lang).

Correspondence to: Bart J. Harvey, MD, PhD, Dalla Lana School of Public Health, University of Toronto, Room 688, 155 College St, Toronto, ON, M5T 3M7, Canada; e-mail: bart.harvey@utoronto.ca


Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Harvey receives royalties from the statistics book, Statistics for Medical Writers and Editors. Mr Lang receives royalties from the statistics book, How to Report Statistics in Medicine.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (http://www.chestpubs.org/site/misc/reprints.xhtml).


© 2011 American College of Chest Physicians


Chest. 2011;139(3):724. doi:10.1378/chest.10-2787
Text Size: A A A
Published online

To the Editor:

We appreciate Drs Maldonado and West’s interest in our article in CHEST (September 2010).1 They are correct to point out that our article did not address Bayesian analysis of research studies, but we did not intend it to. Although Bayesian approaches are often used in analyzing diagnostic test characteristics, as the authors note, Bayesian applications in analyzing research studies are not as straightforward. To begin with, the a priori (ie, pretest) probabilities required for Bayesian statistics are usually more easily available for diagnostic tests than they are for research studies. In the simplest instance, the a priori probability for a diagnostic test can be estimated by the rate of occurrence of the disease of interest in the population. In addition, the results of any previous tests (eg, a chest radiograph) will further refine the a priori probability for subsequent tests (eg, a chest CT scan). Such a priori probabilities are usually not available for research studies,2,3 although 50% can be chosen to reflect the uncertainty of the ultimate outcome. Of note, Rosenthal4 illustrates how both Bayesian and frequentist analyses can produce similar results when an a priori probability of 50% is chosen for the Bayesian analysis.

Drs Maldonado and West’s advice to consider “probable” and “improbable” hypotheses has merit, but if the a priori probability for a given outcome is overly “probable” or “improbable,” is it ethical to conduct the study? Is there truly equipoise?5,6 Thus, their advice should be tempered when applying it to clinical research. We do, however, completely agree with them that P values are only one of many factors that should be considered when interpreting the results of a research study. The hypothesis tested and the reasons for testing it; the estimated effect size; the precision of the estimate; the research design; the probable sources of error, confounding, and bias; and the biologic plausibility of the results should also inform the overall interpretation of the research.

Harvey BJ, Lang TA. Hypothesis testing, study power, and sample size. Chest. 2010;1383:734-737. [CrossRef] [PubMed]
 
Lewis RJ, Wears RL. An introduction to the Bayesian analysis of clinical trials. Ann Emerg Med. 1993;228:1328-1336. [CrossRef] [PubMed]
 
Jonson NE. Everyday diagnostics—a critique of the Bayesian model. Med Hypotheses. 1991;344:289-295. [CrossRef] [PubMed]
 
Rosenthal JS. Struck by Lightning: The Curious World of Probabilities. 2005; Toronto, ON, Canada HarperCollins Publishers Ltd.:217-218
 
Lilford RJ. Ethics of clinical trials from a Bayesian and deci­sion analytic perspective: whose equipoise is it anyway? BMJ. 2003;3267396:980-981. [CrossRef] [PubMed]
 
Freedman B. Equipoise and the ethics of clinical research. N Engl J Med. 1987;3173:141-145. [CrossRef] [PubMed]
 

Figures

Tables

References

Harvey BJ, Lang TA. Hypothesis testing, study power, and sample size. Chest. 2010;1383:734-737. [CrossRef] [PubMed]
 
Lewis RJ, Wears RL. An introduction to the Bayesian analysis of clinical trials. Ann Emerg Med. 1993;228:1328-1336. [CrossRef] [PubMed]
 
Jonson NE. Everyday diagnostics—a critique of the Bayesian model. Med Hypotheses. 1991;344:289-295. [CrossRef] [PubMed]
 
Rosenthal JS. Struck by Lightning: The Curious World of Probabilities. 2005; Toronto, ON, Canada HarperCollins Publishers Ltd.:217-218
 
Lilford RJ. Ethics of clinical trials from a Bayesian and deci­sion analytic perspective: whose equipoise is it anyway? BMJ. 2003;3267396:980-981. [CrossRef] [PubMed]
 
Freedman B. Equipoise and the ethics of clinical research. N Engl J Med. 1987;3173:141-145. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543