[How can "significance" and "relevance" be combined?]

Clinical trials are aimed at providing results which enable improvements in patient care. It is widely criticized, however, that the characterization of results as "significant" or "non-significant" does not allow any assessment of their clinical relevance. To counter this criticism 2 biostatistical concepts are available: the use of confidence intervals and the application of statistical tests with shifted null-hypotheses. Possibilities and limitations of these concepts are discussed in this contribution.