gms | German Medical Science

In clinical research, adaptive seamless designs provide a way to combine two study phases of clinical development in one trial. The advantage over a conventional design is that all data from both stages is used in the final analysis. This leads to a reduced total number of patients and costs and, so, to a shorter development time.

Currently, seamless designs are well known for interventional studies. In contrast, in diagnostic studies, this study design received little attention so far, and there is - to the best of our knowledge - very little methodological research in this field. In addition, a transfer of the methodology from interventional to diagnostic studies is not that easily possible, because of some specific features of diagnostic studies (for example different endpoints in different study phases).

The aim of this research project is to combine an early case-control with a confirmatory cohort diagnostic accuracy study in one trial without compromising the integrity of the trial. The challenge is that the prevalence in the case-control study is in general not representative for the target population. The idea is to recruit a small representative sample in the first stage and to enrich the population with additional cases or controls to achieve a balanced case-control ratio for the interim analysis at the end of stage 1. At this point in time the trial can be stopped for futility if e.g. the co-primary endpoints of sensitivity and specificity have failed. In case the trial is continued, for the final analysis the additional cases/controls from stage 1 will be omitted.

We performed a simulation study to investigate if the results of the final analysis are unbiased and if the type-one error is preserved. Furthermore we compare the seamless designs with the classical design regarding for example sample size and statistical power. In the talk we will explain the approach in detail and present the results of the simulation study.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.