Prognostic factors. Confusion caused by bad quality design, analysis and reporting of many studies

Abstract

In contrast to therapeutic research guidance to design, conduct, analyse and report studies on prognostic factors is less developed and often several deficiencies are stressed. For the assessment of the importance of a factor of interest a systematic review of the corresponding studies would be required, however, this is hardly possible because of many weaknesses in the individual studies. In this article I will discuss several deficiencies of the analysis of prognostic factor studies and shortly discuss problems of reporting and of a summary assessment. By using 3 studies in cancer and a hypotheitcal study as examples I will discuss categorization respectively the determination of a functional form for a continuous factor, sample size, multivariable analysis and data quality. The message of this paper is that serious improvements of prognostic factor studies are required. This can be achieved by a closer collaboration between several disciplines and a closer collaboration at the international level. Specifically, experienced statisticians have to play a central role in the planning, analysis, interpretation and reporting of these studies.