Authors: Alex Molassiotis Evan Kontopantelis
Publish Date: 2013/12/17
Volume: 22, Issue: 4, Pages: 865-866
Abstract
We fully agree with Dranitsaris and Clemons comment on their Letter to the Editor about the need to test predictive models in improving patient outcomes before we can say that a model is clinically useful Otherwise such models remain statistical modelling exercises of little clinical usefulness We are happy to see that the authors are doing so with their own chemotherapyinduced nausea and vomiting CINV prediction tool and we are also planning to do so with our own model in the future The authors criticised the fact that we are only using a combined score for CINV we have analysed the data both for acute and delayed CINV separately alongside having a combined category of acute and delayed CINV and we decided to retain and report only the combined category because it had the highest predictive value and it is a more useful clinical outcome For example clinicians will not want to treat only acute or only delayed CINV but rather both of them together The methods we have used
Keywords: