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 confidentiali ty was earlier proposed, based on testing blood samples in groups of s ize 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 ex tended to higher levels of prevalence (up to about 30 percent). This i s done by taking two samples per individual and forming two independen t sets of groupings which are then tested. The results of the tests ar e averaged to yield a new estimate.