Flexible designs for diagnostic studies - from diagnostic accuracy to personalized medicine

• Applicant: Professorin Dr. Antonia Zapf
• Subject Area: Epidemiology and Medical Biometry/Statistics
• Term: from 2017 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 333052698

Studies for the evaluation of diagnostic tests, i.e. diagnostic studies, are an important form of clinical trials. After all, a correct diagnosis is the prerequisite for a successful therapy. In addition, diagnostic studies are an elementary component of research in the field of personalized medicine, as its aim is to apply only therapies which give reasons to expect an effect individually (for example because of a genetic predisposition or because of the subtype of the disease).The development of a diagnostic test follows, just like the development of therapeutics, several phases. The crucial development phases include confirmatory diagnostic accuracy studies and randomized diagnostic studies. A special characteristic of diagnostic accuracy studies is that there are two co-primary endpoints (true positive and true negative rate), while in randomized diagnostic studies a patient-relevant endpoint (for example mortality) is selected.Diagnostic accuracy studies as well as randomized diagnostic studies require a sample size calculation. Sample size calculations are based on assumptions derived from previous studies and/or from the literature. If these assumptions are wrong, this leads to an over- or underestimation of the required sample size. Hence it is reasonable to apply a flexible (adaptive) study design which allows for a subsequent adjustment of the sample size as a result of an interim analysis. Adaptive designs have a high practical relevance as they can reduce time and costs of a study and the chance of a successful study can be increased. The various methods for adaptive designs that have been developed for therapeutic studies are only of limited use for diagnostic studies. Particularly on the topic of adaptive designs for diagnostic studies, only very little methodological research has been done yet.The requested project aims are developing and evaluating methods for adaptive study designs for various questions in diagnostic studies. A distinction is made between diagnostic accuracy studies and randomized diagnostic studies, between interim analyses with and without unblinding regarding the true disease state and between studies with and without a standard test as comparator. Moreover, methods for adaptive seamless designs for diagnostic studies shall be developed and evaluated. In this very efficient study design, two study phases can be linked and thus a helpful diagnostic test can be applied more quickly. For practical application of the methods developed, computer programs (R and/or SAS) shall be written for all designs and be provided for the public.

• DFG Programme: Research Grants