Evaluation of the Nigeria national HIV rapid testing algorithm

**Background:** HIV diagnosis is the critical entry point to HIV care and treatment. Nigeria adopted a two-test national HIV rapid testing algorithm in 2007 using Determine (T1), Unigold (T2), and Stat-Pak (T3 tie-breaker). The 2018 Nigeria HIV/AIDS Indicators and Impact Survey (NAIIS) showed a much lower national HIV prevalence of 1.4% compared to 4.1% from previous surveys, and revealed a low concordance rate between Test 1 and Test 2 (56.5%) with a PPV of 94.5%. WHO recommends three consecutive reactive tests when national HIV prevalence is <5% due to anticipated low PPV, and recommends that national HIV testing algorithms achieve PPV ≥99%, sensitivity ≥99%, and specificity ≥98%. This necessitated a reevaluation of the existing algorithm across different prevalence levels in Nigeria.

**Methods:** The study employed a dual approach. The retrospective component involved a desktop review of program data (n=492,880) from patients attending routine HIV testing services across six geopolitical zones from January 2017 to December 2019. Twelve states (two per zone) were purposively selected based on NAIIS 2018 HIV prevalence, classified as low (0.1%-1.0%), medium (1.1%-3.0%), and high (>3.0%). The prospective component collected samples (n=2,895) from 18 purposively selected HIV testing sites across six states (one per geopolitical zone). Samples were tested using the national serial HIV rapid testing algorithm in the field, then transported to the National Reference Laboratory (NRL) in Abuja for re-testing using the same algorithm. All HIV-positive and discordant samples were tested with Geenius HIV-1/2 supplementary assay. All 2,895 specimens were tested using an HIV multiplex bead assay. Statistical analysis included PPV, NPV, Kappa statistics, agreement rates with 95% confidence intervals, and McNemar's test.

**Key Results:** The retrospective component showed an overall HIV positivity rate of 5.7% (28,319/492,880) and a discordant rate between Test 1 and Test 2 of 1.1%. The highest discordant rate was in Ebonyi (4.7%), followed by Delta (2.9%), with none observed in Katsina (0%) and Enugu (0%). By testing point, pediatric services showed the highest discordant rate (6.4%). In the prospective component, the overall agreement rate between field and NRL testing was 98.2% (95% CI: 96.2-100.0), with a Kappa statistic of 0.96 (95% CI: 0.93-0.99) and no statistically significant difference (McNemar's P=0.327). When compared with Geenius HIV-1/2 supplementary assay, field results showed an agreement rate of 95.2% (95% CI: 93.2-97.2), while NRL results showed 99.3% (95% CI: 97.3-100.0). Compared with HIV multiplex assay, field results showed sensitivity of 96.6%, specificity of 98.2%, PPV of 97.0%, NPV of 98.0%, and Kappa of 0.95 (95% CI: 0.93-0.97). NRL results showed higher performance: sensitivity 99.2%, specificity 99.4%, PPV 99.0%, NPV 99.5%, and Kappa 0.99 (95% CI: 0.97-1.00). The false-positive rate was higher in the field (1.8%) than NRL (0.6%), and the false-negative rate was also higher in the field (3.4%) than NRL (0.8%).

**Clinical Implications:** The Nigeria national serial HIV rapid testing algorithm performed well overall, but field performance was notably lower than under controlled laboratory conditions. The algorithm met WHO performance thresholds (sensitivity ≥99%, specificity ≥98%, PPV ≥99%) only when performed in the laboratory by trained professionals. The higher false-positive rate (1.8%) and false-negative rate (3.4%) in field testing could lead to unnecessary ART initiation or missed diagnoses. Nigeria's policy of re-testing all HIV-positive individuals before ART initiation is critical to mitigate these risks. The study underscores the urgent need for comprehensive HIV serology refresher training for field testers, routine quality assurance implementation, and consideration of WHO's recommendation for a three-test reactive algorithm in low-prevalence settings (<5%). Staff attrition of trained field personnel was identified as a likely contributor to suboptimal field performance.