gms | German Medical Science

GMDS 2015: 60. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

06.09. - 09.09.2015, Krefeld

How to use prior knowledge and still give new data a chance?

Meeting Abstract

Suche in Medline nach

  • Kristina Weber - Medizinische Hochschule Hannover, Hannover, Deutschland
  • Armin Koch - Medizinische Hochschule Hannover, Hannover, Deutschland

GMDS 2015. 60. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Krefeld, 06.-09.09.2015. Düsseldorf: German Medical Science GMS Publishing House; 2015. DocAbstr. 175

doi: 10.3205/15gmds137, urn:nbn:de:0183-15gmds1376

Veröffentlicht: 27. August 2015

© 2015 Weber et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe http://creativecommons.org/licenses/by/4.0/.


Gliederung

Text

In the context of rare disease and therefore limited resources in terms of patients for clinical trials, it

is often assumed that Bayesian analysis is a possibility to achieve evidence for efficacy of new drugs

or therapy principles more efficiently. The main reason for this assumption is the opportunity to use

additional information through a prior believe or knowledge about this efficacy. However, little is said

about the true impact of this prior assumption about efficacy on the evaluation of outcome.

Using example studies we give an overview of different problems connected to Bayesian analysis with

informative and non-informative prior probabilities. We know that using non-informative priors

usually does not lead to substantially different conclusions than in the frequentist approach, although

even the assumption of non-informative priors can promote the aim of a study to demonstrate either

non-inferiority or superiority. When using informative priors based on data from other studies these

may completely determine conclusions of the trial at hand. It is thus important to balance the impact

of the prior assumption against the information to be derived in the current trial, so that collected

data still has a chance to change prior believe if substantially different.