gms | German Medical Science

21. Jahrestagung des Deutschen Netzwerks Evidenzbasierte Medizin e. V.

Deutsches Netzwerk Evidenzbasierte Medizin e. V.

13. - 15.02.2020, Basel, Schweiz

Sample Size Estimation, beyond bare numbers

Meeting Abstract

Suche in Medline nach

  • Thomas Fabbro - Universitäts Basel, Departement Klinische Forschung, Basel, Schweiz
  • Gilles Dutilh - Universitäts Basel, Departement Klinische Forschung, Basel, Schweiz

Nützliche patientenrelevante Forschung. 21. Jahrestagung des Deutschen Netzwerks Evidenzbasierte Medizin. Basel, Schweiz, 13.-15.02.2020. Düsseldorf: German Medical Science GMS Publishing House; 2020. Doc20ebmPreWS-5-01

doi: 10.3205/20ebm142, urn:nbn:de:0183-20ebm1425

Veröffentlicht: 12. Februar 2020

© 2020 Fabbro 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



Description: The estimation of sample size is a crucial step in planning a clinical study and has many implications. Therefore, the evaluation of the consequences of using different analysis methods and different assumptions is crucial. Nevertheless, especially at this early stage of planning, constrains in time and budget demand a quick and low cost solution. This was our motivation to develop an R-package that allows for evaluating the sample size for ranges of different parameters and different methods in a few steps. The presented framework does not only allow to use any available function for power calculation, but also to easily perform resampling-based sample size estimation. The latter feature makes it possible to easily estimate the sample size in situations where no closed formulas are available or where assumptions of standard approaches are known to be violated.

To allow a fast and reliable reporting and discussion of the results, the package offers a convenient plotting function to show the estimated sample size with respect to the assumptions to visualize the sensitivity of the estimation to these assumptions. Additional functions help to extract the calculated information directly for integration in reports. Thanks to this utility, adaptations in the calculation method can be automatically transferred to the report, allowing a safer and faster workflow.

Intended methods: In the workshop the basic steps will be presented in different scenarios:

a) Sample size estimation based on a parametric power function

b) Sample size estimation based on resampling

c) Sample size estimation for achieving a certain estimation accuracy.

Participants who have R installed on their notebook will be able to practice the steps. It will also be possible to modify the basic examples and to discuss questions. An elaborate example will be presented to show additional details and possibilities of adjustment.

Competing interests: none


Vignette of the sse-package. Available from: Externer Link