Archived Comments for:
A regret theory approach to decision curve analysis: A novel method for eliciting decision makers' preferences and decision-making
We recently read the article “A regret theory approach to decision curve analysis: A novel method for eliciting decision makers’ preferences and decision making” by Tsalatsanis et al. in BMC Medical Informatics and Decision Making 2010, 10:51. We would like to discuss equation 1, because we are convinced the interpretation that follows in the text is erroneous. The equation states: Pt=1/(1+(U1-U3)/(U4-U2)) where Pt is the treatment threshold, U1 to U4 the utilities to respectively treating a diseased, treating a non diseased, not treating a diseased, and not treating a non diseased. If regret U4-U2 is zero, (U1-U3)/(U4-U2) becomes infinite (not “undefined”), so also 1+ (U1-U3)/(U4-U2). Pt will equal 1/infinite, this is zero. Consequently, our treatment threshold becomes zero, and not one, as the authors state. This is also intuitively so: if administering a treatment is almost at no cost, and without harm, clinicians will treat at the slightest suspicion. This is e.g., what happens for malaria treatment in developing countries. The regret U1-U3 is considerable; malaria is a disease with a high mortality, if not treated. The treatment is cheap, almost without side effects. Hence a very low treatment threshold is applied.
Jef Van den Ende, Md, PhD Olivier Koole, MD, MPH Department of Clinical Sciences Institute of Tropical Medicine Antwerp
Competing interests
None declared
Correction of interpretation of equation 1
Athanasios Tsalatsanis, Center for Evidence-based Medicine and Health Outcomes Research, University of South Florida
12 May 2011
Dear Editor,
I would like to thank Drs. Jef Van den Ende and Olivier Koole for their comment and to acknowledge the correction of the interpretation of equation 1 pointed out.
Sincerely,
Athanasios Tsalatsanis
Competing interests
Athanasios Tsalatsanis is the first author of the paper: A regret theory approach to decision curve analysis: A novel method for eliciting decision makers' preferences and decision-making.
Correction of interpretation of equation 1.
15 March 2011
To the editor,
We recently read the article “A regret theory approach to decision curve analysis: A novel method for eliciting decision makers’ preferences and decision making” by Tsalatsanis et al. in BMC Medical Informatics and Decision Making 2010, 10:51.
We would like to discuss equation 1, because we are convinced the interpretation that follows in the text is erroneous. The equation states:
Pt=1/(1+(U1-U3)/(U4-U2))
where Pt is the treatment threshold, U1 to U4 the utilities to respectively treating a diseased, treating a non diseased, not treating a diseased, and not treating a non diseased.
If regret U4-U2 is zero, (U1-U3)/(U4-U2) becomes infinite (not “undefined”), so also 1+ (U1-U3)/(U4-U2). Pt will equal 1/infinite, this is zero. Consequently, our treatment threshold becomes zero, and not one, as the authors state.
This is also intuitively so: if administering a treatment is almost at no cost, and without harm, clinicians will treat at the slightest suspicion. This is e.g., what happens for malaria treatment in developing countries. The regret U1-U3 is considerable; malaria is a disease with a high mortality, if not treated. The treatment is cheap, almost without side effects. Hence a very low treatment threshold is applied.
Jef Van den Ende, Md, PhD
Olivier Koole, MD, MPH
Department of Clinical Sciences
Institute of Tropical Medicine
Antwerp
Competing interests
None declared
Correction of interpretation of equation 1
12 May 2011
Dear Editor,
I would like to thank Drs. Jef Van den Ende and Olivier Koole for their comment and to acknowledge the correction of the interpretation of equation 1 pointed out.
Sincerely,
Athanasios Tsalatsanis
Competing interests
Athanasios Tsalatsanis is the first author of the paper: A regret theory approach to decision curve analysis: A novel method for eliciting decision makers' preferences and decision-making.