[Confidence intervals and their relevance for the interpretation of results. Audit of the journal "Strahlentherapie und Onkology"].

BACKGROUND: The statistical quality of the contributions to "Strahlentherapie und Onkologie" is assessed, aiming for improvement of the journal and consequently its impact factor.

MATERIAL AND METHODS: All 181 articles published during 1998 and 1999 in the categories "review", "original contribution", and "short communication" were analysed concerning the appropriate use of confidence intervals.

RESULT: Forty-four publications were excluded from analysis, because they did not contain quantitative data or because the quotation of a confidence interval would not have been meaningful for other reasons. Of the remaining 137 publications only 27 presented all relevant results with clearly defined and correctly interpreted confidence intervals. This corresponds to a fraction of 20% (95% CI: 13-28%).

CONCLUSION: Authors, peer reviewers, and editors could contribute to improve the quality of the journal by setting value on the documentation of confidence intervals.