Performance characteristics of MAAs and comparison of cross-sectional incidence estimates to longitudinal incidence estimates obtained for three clinical cohorts.

*Includes only LAg-Avidity and BioRad avidity assays; addition of viral load did not impact MAA performance.
Abbreviations: HRM: high resolution melting; MAA: multi-assay algorithm; HPTN: HIV Prevention Trials Network; HIVNET: HIV Network for Prevention Trials.
The table compares performance characteristics of the HRM-based MAA (Figure 1), the sequence-ambiguity-based MAA (Figure 1), and a 2-assay MAA described in a previous report [10]. The 2-assay MAA includes the LAg-Avidity assay (cutoff<2.8 OD-n) and the BioRad-Avidity assay (cutoff<40%); addition of HIV viral load to this MAA did not improve assay performance [10]. For each MAA, the table shows the mean window period, the shadow, and the cross-sectional incidence estimates obtained for each cohort. Methods used to calculate cross-sectional incidence estimates and confidence intervals have been described previously [13]. For each incidence estimate, data presented include the point estimate of incidence (bolded) and the 95% confidence intervals for the incidence estimate (parentheses).
aLongitudinal incidence estimates were obtained previously for the three cohorts, where longitudinal HIV incidence = (number of seroconversion events)/(number of person-years of follow-up) [28], [32], [33]. For HPTN 064 (low incidence cohort), longitudinal incidence was assessed over 6–12 months of follow-up (1,639 person/years); four seroconverters were identified. For HIVNET 001 (medium incidence cohort), longitudinal incidence was assessed between the 12- and 18-month follow-up visits (2,304 person years); 24 seroconverters were identified. For HPTN 061 (high incidence cohort), longitudinal incidence was assessed over 12 months of follow-up (926 person years); 28 seroconverters were identified.
bThe cross-sectional incidence estimates obtained for each MAA were compared to the longitudinal incidence estimates. The percent difference was calculated by the following equation: [(absolute value of the cross-sectional incidence estimate minus the longitudinal incidence estimate)×(100)]/(the longitudinal incidence estimate).
cThe relative survey size shows the size of a cross-sectional survey that would be needed for each of the two new MAAs to obtain the same precision that would be achieved using the previously optimized 2-assay MAA. Because both numbers are <1, a smaller survey would be needed using either of the two new MAAs.