Pooled testing for HIV screening: Capturing the dilution effect

We study pooled (or group) testing as a cost-effective alternative for screening donated blood products (sera) for HIV; rather than test each sample individually, this method combines various samples into a pool, and then tests the pool. A group testing policy specifies an initial pool size, and based on the HIV test results, either releases all samples in the pool for transfusion, discards all samples in the pool, or divides the pool into subpools for further testing. We develop a hierarchical statistical model that relates the HIV test output to the antibody concentration in the pool, thereby capturing the effect of pooling together different samples. The model is validated using data from a variety of held studies. The model is embedded into a dynamic programming algorithm that derives a group testing policy to minimize the expected cost due to false negatives, false positives, and testing. Because the implementation of the dynamic programming algorithm is cumbersome, a simplified version of the model is used to develop near optimal heuristic policies. A simulation study shows that significant cost savings can be achieved without compromising the accuracy of the test. However, the efficacy of group testing depends upon the use of a classification rule (that is, discard the samples in the pool, transfuse them or test them further) that is dependent on pool size, a characteristic that is lacking in currently implemented pooled testing procedures.