DÄ internationalArchive23/2008Benefit and Risk of Mammography Screening – Considerations from an Epidemiological Viewpoint: Inaccurate Estimation of Everyday Risks

Correspondence

Benefit and Risk of Mammography Screening – Considerations from an Epidemiological Viewpoint: Inaccurate Estimation of Everyday Risks

Dtsch Arztebl Int 2008; 105(23): 419. DOI: 10.3238/arztebl.2008.0419a

Beise, R

LNSLNS Conditional probabilities versus absolute frequencies: what is this really all about?

We know from psychological research that the manner in which facts are presented has a major influence on the thoughts and actions of the persons affected by them. Facts are assessed differently if they are presented as a potential gain or a potential loss. This is the central point: whoever controls the manner in which study data are presented will essentially determine whether a proposed screening program will be accepted.

The authors object to presenting the benefit of screening with absolute frequencies and advocate using conditional probabilities instead. They postulate that the general population is well versed in the use of conditional probabilities because certain risks that are permanently present in everyday life require constant assessment (e.g., building insurance).

I do not share this opinion. Many studies in social psychology have shown that most people generally estimate everyday risks inaccurately. Gigerenzer (1) has shown in several studies that neither laypersons nor experts can draw correct conclusions from conditional probabilities. This being known, the authors' argument that the general acceptance of building insurance reflects a correct understanding of conditional probabilities seems rather weak. It is, furthermore, good scientific practice to express study data in terms of frequencies. No critically thinking physician would assess a drug trial, for example, on the basis of relative risk reduction alone; rather, the actual benefits and potential risks would need to be assessed in the light of the frequency parameters NNT (number needed to treat) and NNH (number needed to harm). Why should this be any less obvious when a screening test comes under scrutiny?
DOI: 10.3238/arztebl.2008.0419a


Dr. med. Reinhard Beise
Brunnenstr. 21, 85570 Otten
1.
Gigerenzer G: Das Einmaleins der Skepsis. Über den richtigen Umgang mit Zahlen und Risiken. Bvt Berliner Taschenbuch Verlag 2004.
1. Gigerenzer G: Das Einmaleins der Skepsis. Über den richtigen Umgang mit Zahlen und Risiken. Bvt Berliner Taschenbuch Verlag 2004.