|
Abstract:
|
Patients on HIV treatment require regular monitoring of viral load (VL), which quantifies the number of copies of the virus per ml of plasma. The objective of VL monitoring is to detect whether VL has exceeded a certain threshold, at which point an intervention is needed (such as adherence counseling or change in treatment).
In low income countries, cost of VL assays may lead to less than optimal measurement frequency. One way to address this issue is to use assays of pooled blood samples. Pooled testing can lead to substantial reductions in the number of assays that need to be carried out. The magnitude of the reduction depends on the distribution of VLs that are included in a pooled assay.
In this talk, we demonstrate statistical methods that can be used to evaluate the use of pooling at the program level, in order to assess the impact as it relates to number of assays needed, sensitivity and specificity of individual tests, and cost to the program. Moreover, we investigate how to construct an algorithm to provide guidance, at the patient level, about which patients should have individual testing, should be tested first, and which should be used in pooled assays.
|
