|
Abstract:
|
There is currently wide interest in evaluating in vitro diagnostic (IVD) tests of biomarkers. Applications of these IVDs include personalized medical management. In a screening setting, prevalence of a condition of interest (COI) is often 2% or less. This leads to important challenges in performing prospective studies of IVDs, for example, the effort and expense of enrolling a very large number of subjects without the COI in order to obtain enough subjects with the COI to adequately power statistical inference. One way to meet this challenge is to enroll all eligible subjects likely to have the COI, and limit the rate of enrollment of eligible subjects less likely to have the COI. A benefit of the wide interest in biomarkers is the existence of well-curated specimen banks. Using specimens from these banks can save time and money. However, care must be taken to ensure that the specimens are selected in a manner that mitigates biases, such as spectrum bias. One must also be able to justify using specimens provided to the bank from different prospective studies. This talk will discuss considerations related to subject selection and pooling specimens from multiple studies.
|
