GROUP-TESTING FOR SENSITIVE CHARACTERISTICS - EXTENSION TO HIGHER PREVALENCE LEVELS

Motivated by the reluctance of the public to participate in surveys of the prevalence of HIV infection, a procedure preserving confidentiality was earlier proposed, based on testing blood samples in groups of size k > 1. A disadvantage inherent in that procedure was that it could only be applied effectively where the prevalence of the infection was less than ten percent. In this paper the range of applicability is extended to higher levels of prevalence (up to about 30 percent). This is done by taking two samples per individual and forming two independent sets of groupings which are then tested. The results of the tests are averaged to yield a new estimate.