Wee-Shian Chan, MD, MSc; Tim Ramsay, MSc, PhD; Jeffrey S. Ginsberg, MD
Potential Conflicts of Interest: None disclosed.
Chan W, Ramsay T, Ginsberg JS. Prediction Rules Must Be Developed According to Methodological Guidelines. Ann Intern Med. 2010;152:263-264. doi: 10.7326/0003-4819-152-4-201002160-00015
Download citation file:
Published: Ann Intern Med. 2010;152(4):263-264.
We appreciate Dr. Janssen and colleagues' comments. They raise some important and valid concerns, which we address below.
As Dr. Janssen and colleagues point out, prediction rules are important to guide clinical practice in many instances, because they provide absolute risks for individual patients that can be used to determine treatment choices or further diagnostic work-up. We agree that before any rule can be adopted, it must be properly validated (which we did not do in our initial study) and must be derived from solid methodology.
It may be true that the c-statistic may not be the most sensitive measure of how much an additional variable contributes to estimates of individual risk. However, because the overall c-statistic for the final model with 3 variables was 0.943 (corresponding to the area under the receiver-operating characteristic curve), it is unlikely that any other significant predictors were missed by using this method.
to gain full access to the content and tools.
Learn more about subscription options.
Register Now for a free account.
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only