Objective: This paper compares nine strategies for determining hepatitis C antibody and viral status. They combine two tests for antibodies (enzyme immunoassays (EIA), recombinant immunoblot assays (RIBA)) and one for viremia (reverse transcription polymerase chain reaction (PCR)). Using optical density to divide EIA results into three categories (high positive, low positive, negative) was also considered.
Methods: Decision analysis compared strategies on cost as well as sensitivity and specificity with regard to antibody and viral status. Parameters in the decision tree included antibody prevalence, proportion viremic, sensitivity, specificity, and cost of individual tests.
Results: The two best strategies are EIA followed by PCR (EIA-->PCR); and EIA with three levels of optical density (EIA-OD), followed by RIBA for EIA-OD low positives, and then PCR for all positives (EIA-OD-->RIBA-->PCR). EIA-->PCR has equal viral sensitivity, slightly lower cost, slightly higher antibody sensitivity, but lower antibody specificity compared to EIA-OD-->RIBA-->PCR. The cost per false antibody positive avoided using EIA-OD-->RIBA-->PCR rather than EIA-->PCR is $36 when prevalence is 5%, and $193 when prevalence is 50%. Using EIA-OD-->RIBA-->PCR rather than EIA-->PCR results in 112 false antibody positives avoided for every true antibody positive missed when prevalence is 5%; this ratio is 18:1 when prevalence is 25%; and 6:1 when prevalence is 50%.
Conclusions: EIA-OD-->RIBA-->PCR is the best choice when prevalence in the tested group is below 20%. As prevalence increases, the choice of EIA-OD-->RIBA-->PCR versus EIA-->PCR will depend on the relative importance of avoiding false antibody positives versus missing true antibody positives. Our analysis makes explicit the magnitude of this trade-off.