A critical review of papers from clinical cancer research

A review of 75 original articles from clinical cancer research in Norway is presented. Articles published in 1993 and with at least one Norwegian author were included in the review. Sixty papers were observational, whereas 15 were experimental. Of the observational studies 44 were retrospective. Most of the papers concerned prognostic factors. Prior hypotheses were explicitly defined in 16 papers only, and less than half of the articles described inclusion and exclusion criteria. Sample size calculations were performed in four papers only. The choice of statistical method was considered to be suitable in 22 of the 58 articles presenting statistical inferences. Problems related to multiple significance testing were rarely addressed, although the median number of p-values reported was as high as 8. Confidence intervals for main findings were presented in 14 papers. For proper planning of studies, as well as for analysis and interpretation of study results, statistical advice is indeed required.