Article Text

Download PDFPDF
The need for expanding and re-focusing of statistical approaches in diagnostic research
  1. H Brenner1,
  2. T Stürmer1,
  3. O Gefeller2
  1. 1Department of Epidemiology, German Centre for Research on Ageing, Heidelberg, Germany
  2. 2Department of Medical Informatics, Biometry and Epidemiology, University of Erlangen-Nuremberg, Germany
  1. Correspondence to:
 Dr H Brenner, Department of Epidemiology, German Centre for Research on Ageing, Bergheimer Str 20, D-69115 Heidelberg, Germany;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Fitting statistical methodology to the need of diagnostic research

In his contribution “Misguided efforts and future challenges for research on diagnostic tests”1 Dr Feinstein has identified major gaps and shortcomings in previous and current diagnostic research. While we fully agree with most of his criticisms, we would like to take issue with him over the role of statistics and mathematical formalisation in diagnostic research. In particular, we would like to emphasise the need and potential of expanding and re-focusing rather than abandoning statistical approaches in diagnostic research.

The traditional concepts of sensitivity and specificity as well as of the “posterior probabilities” have certainly been useful as a methodological framework for structuring efforts to quantify accuracy of diagnostic markers in well defined, very special settings in the past, and they may continue to be useful as such in the future. The major limitation of these concepts does not so much lie in their intrinsic properties, but in the uncritical adoption of these concepts to a wide range of different, usually more complex settings. This misapplication along with some misconceptions outlined below have often been severely misguiding indeed.


An important example is the dogma still found in most textbooks …

View Full Text