Abstract Body

Due to high sensitivity of enzyme immunoassays (EIA) some level of false positivity is expected when used alone or in an EIA-based testing algorithm for HIV diagnostic testing. This is problematic for antenatal clinic (ANC) surveys whose data are used for surveillance purposes as it can lead to inflated prevalence estimates. DBS are preferred specimen for HIV surveillance surveys as they are simple, inexpensive, and can be prepared in non-laboratory settings. However, use of DBS specimens requires modification and optimization of EIA procedures. We present results from an EIA testing algorithm compared to the US Centers for Disease Control (CDC) external quality assurance (EQA) testing algorithm applied to specimens from the South Sudan 2012 ANC survey.

DBS specimens collected for the South Sudan 2012 HIV ANC Survey were tested at the Juba Teaching Hospital Laboratory. Of the 11,155 specimens collected in the ANC survey, 1059 HIV-negative and 285 HIV-positive specimens were sent to CDC Atlanta for EQA testing. As part of the South Sudan testing algorithm, the specimens were screened on Vironostika Uni-form II plus EIA (Biomerieux, France) and reactives were confirmed by Murex HIV-1.2.O EIA (Abbott, Germany). Retesting at CDC was performed using an optimized DBS elution protocol on Genetic Systems HIV-1/2 plus O (GS 1-2-O, Bio-Rad, California) followed by confirmation of positives on Cambridge Biotech HIV-1 Western Blot (WB) (Maxim Biomedical, Maryland). Specimens with discrepant final classification between the two algorithms were further confirmed by WB.

Of the 1344 of specimens tested, the overall agreement of the South Sudan and CDC algorithms was 97.5%; however, 97.2% and 89.1% agreements were observed amongst the negatives and positives, respectively. A total of 33 specimens (2.5%) were discrepant between the algorithms and of those, 29 (88%) specimens were negative and 4 (12%) specimens were indeterminate according to WB results. The optical density values of most false positive (FP) specimens were low on EIAs indicating inherent false positivity of EIAs that is magnified when DBS specimens are used.

Our result suggests a high FP rate with EIA-testing algorithm alone.  These results highlight the need to incorporate a more specific assay such as WB or similar test in the testing algorithm to confirm the HIV status and ensure the accuracy of HIV prevalence data. Use of DBS specimens requires additional optimization steps to ensure accurate results.