An approach to evaluating the reliability of diagnostic tests on pooled groups of infected individuals.