The United States Centers for Disease Control and Prevention (US CDC) previously recommended using an immunoassay with a confirmatory Western blot or indirect immunofluorescence assay for HIV diagnosis.1 In 2014, an updated testing algorithm was recommended that increases sensitivity for detecting acute HIV infections and differentiates between HIV-1 and HIV-2 infection.2 In the current algorithm, samples are first tested using an assay that detects both HIV antigen and antibody (fourth-generation assay). If that assay is reactive, an HIV-1/HIV-2 discriminatory assay is performed. If the discriminatory assay is nonreactive or indeterminate, a nucleic acid test is performed.2
In resource-limited settings, the World Health Organization recommends using 2 HIV rapid tests for diagnosis.3 These tests can be performed at the point of care, facilitating linkage to counseling services and HIV care.3,4 If the results of the 2 rapid tests are discordant (1 reactive and 1 nonreactive), a tie-breaker test is often used3 [eg, a third rapid test or an enzyme immunoassay (EIA)5,6]. Despite the widespread use of a tie-breaker approach for HIV diagnosis and the availability of performance data for individual HIV assays, there are limited data comparing the performance of different testing strategies for determining HIV status in individuals with discordant rapid test results. Previous studies have evaluated individuals with discordant HIV rapid tests in populations at increased risk of HIV infection (eg, sexual partners of HIV-infected individuals,7 women attending prenatal or antenatal clinics,8,9 adults attending clinics for sexually transmitted diseases,10,11 and adults in HIV screening programs with very high HIV prevalence12). In those studies, the frequency of discordant rapid test results ranged from 0.7% to 2.3%, and many participants had acute HIV infection. However, because the window period for acute HIV infection is short,13 individuals with acute HIV infection may comprise a smaller portion of those with discordant rapid tests in cohorts at lower risk of HIV acquisition.
In this study, we analyzed samples from participants in a population-based survey in Tanzania and South Africa. The goals of this study were to compare the performance of different testing strategies for determining HIV status in individuals with discordant HIV rapid test results and to investigate factors associated with discordant rapid tests in HIV-infected individuals.
Study Population and Samples Used for Analysis
Samples were obtained from the National Institute of Mental Health Project Accept trial [HIV Prevention Trials Network (HPTN) 043] (NCT00203749). HPTN 043 was a large, community-randomized clinical trial in Africa and Thailand that assessed the impact of integrated behavioral interventions on HIV incidence.14 HIV incidence was assessed at the end of the trial in a single cross-sectional survey.15 The survey included over 50,000 participants aged 18–32 years, randomly sampled from 48 communities at 5 study sites. All eligible individuals in each household could participate in the survey.
This report describes analysis of samples collected at 3 HPTN 043 sites (Kisarawe, Tanzania; Soweto and Vulindlela, South Africa). In HPTN 043, one 10 mL EDTA-anticoagulated blood sample was collected from each participant. The sample was used for in-country HIV testing and then used to prepare plasma that was frozen at −80°C and shipped to the HPTN Laboratory Center at Johns Hopkins University, Baltimore, MD for further testing; a 4 mL sample was also collected for in-country CD4 cell count testing. Laboratories at the study sites participated in external quality assurance programs; external quality assurance assessments were also performed by the HPTN Laboratory Center. In HPTN 043, 255 of 34,813 samples (0.7%) from Tanzania and South Africa had discordant HIV rapid test results [187 (2.1%) of 9041 samples from Tanzania; 68 (0.3%) of 25,772 samples from South Africa; Table 1].
In-country HIV Rapid Testing
Samples were initially tested in-country at local laboratories using 2 HIV rapid tests performed in parallel on fresh samples. The following rapid tests were used for this initial assessment: Determine HIV-1/2 (Inverness Medical Innovations, Phetchabun, Japan), SD Bioline HIV-1/2 v3 (Youngin-Si, South Korea), or Uni-Gold HIV Test (Trinity Biotech, Bray, Ireland; Soweto only). Samples with discordant rapid test results were analyzed in-country using a tie-breaker test. The site in Tanzania used a third HIV rapid test as the tie-breaker test (Uni-Gold HIV Test). The 2 sites in South Africa used a fourth-generation chemiluminescent immunoassay as the tie-breaker test (the ARCHITECT HIV Ag/Ab Combo assay, Abbott Laboratories, Wiesbaden, Germany; referred to below as the Abbott Combo assay). All other testing described in this report was performed at the HPTN Laboratory Center.
Classification of Samples With Discordant Rapid Test Results as HIV Positive or HIV Negative
Samples were classified as HIV positive or HIV negative using the testing algorithm shown in Figure 1. This algorithm is similar to the current US CDC testing algorithm2 but was modified to include two fourth-generation assays and 2 HIV RNA assays to maximize sensitivity for detecting HIV infection. The two fourth-generation assays were performed in parallel (Abbott Combo assay; and GS HIV Combo Ag/Ab EIA, Bio-Rad Laboratories, Redmond, WA; referred to below as the Bio-Rad Combo assay). If 1 or both assays were reactive, samples were tested using the Multispot HIV-1/HIV-2 rapid test (Bio-Rad Laboratories, Redmond, WA; referred to below as the discriminatory assay). If the discriminatory assay was negative or indeterminate, samples were tested using 2 HIV RNA assays (APTIMA HIV-1 RNA Qualitative Assay, Hologic Gen-Probe Inc., San Diego, CA, referred to below as the Aptima RNA assay; and Roche COBAS AMPLICOR HIV-1 MONITOR test, v1.5, Roche Diagnostics, Branchburg, NJ; referred to below as the Roche viral load assay). Samples were classified as HIV positive if they had any of the following test results: positive discriminatory test, positive Aptima RNA assay, or positive Roche viral load assay (Fig. 1).
Comparison of the Performance of Different Tie-breaker Assays
Samples with discordant rapid tests were tested with 4 tie-breaker assays: (1) a third HIV rapid test (Uni-Gold HIV Test; performed in-country, Tanzania only); (2) a third-generation EIA (VITROS Anti-HIV 1 + 2 assay, Ortho Clinical Diagnostics, Rochester, NY, referred to below as the Vitros EIA); (3) the Bio-Rad Combo assay; and (4) the Abbott Combo assay.
Further Characterization of Samples With Discordant Rapid Test Results
Samples classified as HIV positive were further characterized with 2 additional HIV rapid tests (Uni-Gold Recombigen HIV-1/2; Trinity Biotech, Bray, Ireland; and OraQuick Advance Rapid HIV-1/2 Antibody Test; OraSure Technologies, Inc., Bethlehem, PA), a Western blot test, and a qualitative multidrug screening that detects 15 antiretroviral drugs, including protease inhibitors, nucleoside/nucleotide reverse transcriptase inhibitors, and nonnucleoside reverse transcriptase inhibitors.16,17 Samples classified as HIV negative that had at least 1 reactive (false-positive) serologic assay were also further characterized with a Western blot test and antiretroviral drug screening.
The work described was performed in accordance with the Declaration of Helsinki. The HPTN 043 trial was conducted in close partnership with established community advisory boards and local government departments. Consent was obtained (written consent in South Africa; oral consent in Tanzania) at the community level for trial participation. To approach household members to participate in the postintervention survey, investigators needed permission from the head of the household. Oral consent was obtained from each participant for collection and testing of blood samples.14 The study was approved by ethics committees for each site and by all participating academic institutions.
Classification of HIV Discordant Samples as HIV Positive or HIV Negative
This study evaluated 173 of the 255 HIV discordant samples (67.8%) identified in the HPTN 043 study (Table 1). Eighty-two samples were excluded (80 had insufficient plasma available, 1 was contaminated in-country, and 1 was involved in an in-country sample mix-up). The 173 samples were first classified as HIV positive or HIV negative using the HIV testing algorithm shown in Figure 1. This classification was used as a gold standard for the evaluation of simpler, tie-breaker testing approaches. Twenty-nine of the 173 samples (16.8%) were classified as HIV positive (24/127 from Tanzania; 5 of 46 samples from South Africa, P = 0.3, Fisher exact test); 144 samples were classified as HIV negative.
Comparison of Tie-breaker Approaches for Determining the HIV Status of Samples With Discordant HIV Test Results
We next evaluated 4 tie-breaker approaches for determining the HIV status of individuals with discordant HIV rapid test results (Table 2): (1) a third-generation rapid test (the Uni-Gold HIV Test, performed in-country, Tanzania only), (2) a third-generation EIA (the Vitros EIA), (3) a fourth-generation EIA (the Bio-Rad Combo assay), and (4) a fourth-generation chemiluminescent immunoassay (the Abbott Combo assay). The two third-generation tests had low sensitivity (8.3% and 24.1%) with specificities of 96.1% and 98.6%. The fourth-generation assays were more sensitive (93.1% and 96.6%) but had lower specificity (88.2% and 90.3%). The accuracy of the 4 tests ranged from 79.5% to 91.3% (Table 2).
Further Evaluation of Samples Classified as HIV Positive
Further testing was performed to characterize the 29 HIV-positive samples (Table 3). The mean CD4 cell count of 29 participants was 1050 cells per microliter (range, 222–1826); 2 had a CD4 cell count <350 cells per microliter (case 1: 252; case 4: 222). As noted above, the HIV rapid test used as a tie-breaker in Tanzania had a sensitivity of 8.3% (Uni-Gold HIV test). For comparison, the HIV-positive samples were tested using 2 HIV rapid tests approved by the US Food and Drug Administration (the Uni-Gold Recombigen HIV-1/2 Test and the OraQuick Advance HIV-1/2 Antibody Test). This testing was performed at the HPTN Laboratory Center and included samples from both Tanzania and South Africa (Table 3; note that some samples were depleted in previous testing). The Uni-Gold Recombigen HIV-1/2 test was reactive for 4 of 16 samples tested (sensitivity: 25.0%). The OraQuick Advance HIV-1/2 Antibody Test was reactive for 11 of 26 samples tested (sensitivity: 42.3%). In several cases where these tests were reactive, the test line was faint and could easily have been missed (1 of 4 samples that was reactive with the Uni-Gold Recombigen HIV-1/2 Test had an extremely faint T line; 9 of 11 samples that were reactive with the OraQuick Advance HIV-1/2 Antibody Test had a faint or extremely faint T-line). Sensitivity was even lower when a Western blot test was used to confirm HIV infection (Table 3; range: 4.2%–12.5%).
Twenty-four of the 29 samples (82.8%) were reactive with both fourth-generation assays and the discriminatory test. Three of the remaining 5 samples were reactive with only 1 of the two fourth-generation assays (1 was reactive with the Bio-Rad Combo assay only; 2 were reactive with the Abbott Combo assay only); all 3 of these samples were positive using the discriminatory assay. One or 2 of these 3 samples would have been missed using the testing algorithm recommended by the US CDC,2 depending on which 1 of two fourth-generation assays was used for screening. The remaining 2 samples were reactive using both fourth-generation assays and the third-generation EIA but had nonreactive results using the discriminatory assay.
Samples were next tested using the Roche viral load assay (performed using a validated method for low-volume samples, limit of detection: 400 copies per milliliter; 2 samples did not have sufficient plasma available for testing). Samples with viral loads <400 copies per milliliter were tested using the Aptima HIV RNA assay (limit of detection: 30 copies per milliliter; 1 sample did not have sufficient plasma for testing). Four of 27 samples (14.8%) tested had viral loads >400 copies per milliliter (viral load: 1280–459,246 copies per milliliter); 1 sample with <400 copies per milliliter HIV RNA had a reactive Aptima HIV RNA assay. The remaining 24 samples did not have detectable HIV RNA.
As a final step, samples were tested using a qualitative assay that detects 15 antiretroviral drugs (4 samples did not have sufficient plasma for testing). Only 1 sample had a positive result: a sample from a pregnant women in South Africa with a viral load of 1280 copies per milliliter; lamivudine and nevirapine were detected (case 4). These results indicate that most HIV-infected samples with discordant HIV rapid test results (21/25 = 84.0%) were from individuals who were virally suppressed in the absence of antiretroviral treatment (ART).
Further Evaluation of Samples Classified as HIV Negative
Samples from 144 study participants were classified as HIV negative using the testing algorithm shown in Figure 1. These individuals had a mean CD4 cell count of 986 cells per microliter (range, 481–5833; one value was missing). In 24 of those cases (16.7%), 1 or more serologic assay was reactive (in addition to 1 HIV rapid test initially performed at study sites, Table 4). In 11 of the 24 cases, only 1 of the assays had a reactive tie-breaker test (4 had a reactive rapid test, 2 had a reactive third-generation EIA, 4 had a reactive Bio-Rad Combo assay, and 1 had a reactive Abbott Combo assay). A Western blot test was performed for the 7 samples that had a reactive third-generation EIA or fourth-generation assay; 1 Western blot was indeterminate and 6 were negative. In the remaining 13 cases (cases 12–24), both fourth-generation assays were reactive; a Western blot was indeterminate in 9 cases, negative in 3 cases, and positive in 1 case. In all of the cases, the discriminatory assay and the Aptima HIV RNA assay were negative. As a final step, the samples were tested for antiretroviral drugs (12 samples did not have sufficient plasma for testing); none of the samples had antiretroviral drugs detected. Since longitudinal samples were not collected in the HPTN 043 trial and cellular DNA was not stored for testing (eg, whole blood samples), it was not possible to determine conclusively whether these individuals were HIV infected.
At the 3 HPTN 043 sites included in this study, HIV prevalence ranged from 5.9% to 30.8%, and HIV incidence ranged from 0.78% to 3.90%. Overall, 255 of the 34,813 participants (0.7%) had discordant HIV rapid test results (0.3% in South Africa, 2.1% in Tanzania). This report describes analysis of plasma samples from 173 of the 255 participants; 29 of the samples (16.8%) were classified as HIV positive using a rigorous testing algorithm. Because the number of HIV-positive samples from South Africa was small (N = 5), we did not compare the performance of testing approaches in the 2 countries.
We compared the sensitivity and specificity of 4 tie-breaker assays for determining HIV status in these discordant cases. The sensitivity of third-generation assays was low (HIV rapid test performed in Tanzania: 8.3%; third-generation EIA: 24.1%); specificity was 96.1% and 98.6%, respectively. A higher sensitivity was obtained when samples were tested in the United States using Food and Drug Administration–cleared HIV rapid tests (25.0% and 42.3%); however, many of the reactive test results were very weak and could easily have been missed, especially in an environment with suboptimal lighting. Use of a Western blot to confirm a reactive third-generation assay (HIV rapid test or EIA) further reduced the sensitivity for detecting HIV infection. The sensitivity of fourth-generation assays was considerably higher (Bio-Rad Combo assay: 93.1%; Abbott Combo assay: 96.6%); however, specificity was only 88.2% and 90.3%, respectively. Accuracy was highest for the Abbott Combo assay (91.3%). These results indicate that more rigorous testing approaches (beyond a single tie-breaker assay or a third-generation assay with confirmatory Western blot) should be considered when resolving HIV status in individuals with discordant rapid test results.
Previous studies found that some individuals with discordant rapid tests had acute HIV infection.10–12 Those studies were performed in cohorts at increased risk of HIV acquisition (those attending clinics for sexually transmitted infections or participating in a screening program with very high HIV prevalence).10–12 It is notable that those studies considered individuals with undetectable HIV RNA to be HIV uninfected. In this study, which was performed in the context of a general population survey, almost all of the HIV-infected individuals with discordant HIV rapid tests had undetectable HIV RNA using 2 assays validated for different HIV subtypes.18,19 Since most of these individuals were virally suppressed in the absence of ART, they are likely to have been elite controllers. Criteria used to identify elite controllers usually include documentation of viremic control for at least 1 year.20 Unfortunately, longitudinal samples were not collected in the HPTN 043 trial. DNA polymerase chain reaction testing may also be helpful for confirming a diagnosis of HIV infection in elite controllers; however, some elite controllers have negative DNA polymerase chain reaction tests.21,22 Samples suitable for HIV DNA testing were not collected in the HPTN 043 trial. It was not possible to determine the proportion of elite controllers in the entire HPTN 043 cohort because viral load testing was only performed for a small proportion of the HIV-infected individuals. Other reports suggest that elite controllers represent <1% of HIV-infected individuals20,23; however, higher prevalence of elite controllers has been reported in some studies.24
Viral suppression, whether natural or from ART, has been associated with false negative HIV screening tests.25,26 Downregulation of anti-HIV antibodies in virally suppressed individuals has been documented using serologic assays developed for HIV incidence estimation.27–30 The low levels of circulating virus (and viral antigen) in virally suppressed individuals may also increase the likelihood of obtaining a false negative test result using a fourth-generation assay.
Failure to detect HIV infection in virally suppressed individuals has practical implications. Although most individuals on ART are not likely to be tested for HIV infection (since their HIV status is known), some may chose not to disclose knowledge of their HIV status to care providers.31 Individuals on ART may also be tested for HIV infection in population-based surveys and clinical trials (eg, if self-reported data on HIV status is not collected or if individuals chose not to disclose their HIV status to study staff32). Elite controllers may not be aware of their HIV status and may be incorrectly classified as HIV uninfected because of false negative HIV tests.33 Viral suppression may also complicate identification of newly infected individuals. We recently described a cohort in the United States in which 21% of HIV seroconverters had low or undetectable HIV RNA in the absence of antiretroviral drug use.17 In the same study, HIV infection was missed at multiple visits in 2 virally suppressed participants using an HIV rapid test; in 1 case, HIV infection was also missed using a fourth-generation assay.33 Missed HIV infection may become more common in resource-limited settings that rely on third-generation HIV screening assays and simple HIV testing algorithms, particularly as ART is scaled up for HIV treatment and prevention.34–36
Another factor that may potentially impact the performance of different HIV testing algorithms is HIV-1 subtype. Subtype D HIV infection is associated with lower levels of anti-HIV antibodies and lower avidity of anti-HIV antibodies for target antigens.37,38 In HPTN 043, HIV prevalence and incidence were higher in South Africa than in Tanzania. However, the percentage of individuals with discordant HIV rapid test results and the proportion of the discordant samples that were from HIV-infected individuals were both higher in Tanzania. This may reflect differences in the prevalent HIV subtypes in these 2 countries. In South Africa, the overwhelming majority of HIV infections are subtype C, whereas various viral subtypes are found in Tanzania.39 Previous analysis of samples from a subset of HPTN 043 participants in Tanzania revealed a mixture of HIV subtypes (44% subtype A; 22% subtype C; 24% intersubtype recombinant; and 10% subtype D15). In this study, we were not able to assess the contribution of HIV subtype to HIV test outcomes because most HIV-infected individuals with discordant HIV rapid tests were virally suppressed; in the 3 cases in which HIV subtype was determined, 1 individual had subtype A1 infection and 2 had subtype C infection.
In this study, individuals classified as HIV positive were likely to have been HIV infected because they had positive/reactive tests with numerous HIV assays (1 third-generation rapid test, at least 1 fourth-generation HIV assay, and the discriminatory assay or an RNA assay). We do note, however, that the HIV testing algorithm currently recommended by the US CDC has not been fully evaluated for use in Africa. Further testing is needed to evaluate performance of this algorithm in different settings. In this study, some false positive test results were obtained using fourth-generation assays as tie-breaker tests. A variety of factors have been associated with false positive HIV test results, including the presence of cross-reacting antibodies to other infectious agents.40–43 The high frequency of other infectious diseases in Africa could contribute to a higher rate of false positive test results due to immune stimulation and/or formation of antibodies that cross-react with target antigens in HIV assays. Interestingly, the extended sample testing performed in this study identified 13 study participants who were classified as HIV negative using a rigorous testing algorithm but had reactive results using both the Bio-Rad and Abbott Combo fourth-generation assays. These participants do not meet the criteria for HIV infection using the testing algorithm currently recommended by the US CDC because both the differentiation assay and HIV RNA tests were negative. In HPTN 043, samples were not stored for HIV DNA testing, which may have been helpful in resolving the HIV status in these cases.
Taken together, the findings from this report suggest that although simple testing algorithms are practical for determining HIV status in most individuals in resource-limited settings, extensive, laboratory-based testing may be needed to resolve the HIV status in individuals with discordant HIV rapid tests.
The authors thank the HPTN 043 study team and study participants for providing samples for this study and thank the laboratory staff at the study sites and at the HPTN Laboratory Center with assistance for sample management and testing. The authors thank Laura Robins-Morris for assistance with data management in the HPTN 043 study.
1. Centers for Disease Control and Prevention. Detection of acute HIV
infection in two evaluations of a new HIV
diagnostic testing algorithm—United States, 2011–2013. MMWR Morb Mortal Wkly Rep. 2013;62:489–494.
7. Boeras DI, Luisi N, Karita E, et al.. Indeterminate and discrepant rapid HIV
test results in couples' HIV
testing and counselling centres in Africa
. J Int AIDS Soc. 2011;14:18.
8. Granade TC, Parekh BS, Tih PM, et al.. Evaluation of rapid prenatal human immunodeficiency virus testing in rural cameroon. Clin Diagn Lab Immunol. 2005;12:855–860.
9. Koblavi-Deme S, Maurice C, Yavo D, et al.. Sensitivity and specificity of human immunodeficiency virus rapid serologic assays and testing algorithms in an antenatal clinic in Abidjan, Ivory Coast. J Clin Microbiol. 2001;39:1808–1812.
10. Powers KA, Miller WC, Pilcher CD, et al.. Improved detection of acute HIV
-1 infection in sub-Saharan Africa
: development of a risk score algorithm. AIDS. 2007;21:2237–2242.
11. Fiscus SA, Pilcher CD, Miller WC, et al.. Rapid, real-time detection of acute HIV
infection in patients in Africa
. J Infect Dis. 2007;195:416–424.
12. Bassett IV, Chetty S, Giddy J, et al.. Screening for acute HIV
infection in South Africa
: finding acute and chronic disease. HIV
13. Fiebig EW, Wright DJ, Rawal BD, et al.. Dynamics of HIV
viremia and antibody seroconversion in plasma donors: implications for diagnosis and staging of primary HIV
infection. AIDS. 2003;17:1871–1879.
14. Coates TJ, Kulich M, Zelaya CE, et al.. Outcomes from NIMH Project Accept (HPTN 043): a cluster-randomized trial of community mobilization, mobile HIV
testing, post-test support services, and real-time performance feedback. Lancet Glob Health. 2014;2:e267–e277.
15. Laeyendecker O, Piwowar-Manning E, Fiamma A, et al.. Estimation of HIV
incidence in a large, community-based, randomized clinical trial: NIMH Project Accept (HIV
Prevention Trials Network 043). PLoS One. 2013;8:e68349.
16. Marzinke MA, Breaud A, Parsons TL, et al.. The development and validation of a full scan-high resolution accurate mass spectrometric (HRMS) screening method for the qualitative monitoring of antiretroviral agents in human blood. Clin Chim Acta. 2014;433:157–168.
17. Chen I, Cummings V, Fogel JM, et al.. Low-level viremia early in HIV
infection. J Acquir Immune Defic Syndr. 2014;67:405–408.
19. Michael NL, Herman SA, Kwok S, et al.. Development of calibrated viral load standards for group M subtypes of human immunodeficiency virus type 1 and performance of an improved AMPLICOR HIV
-1 MONITOR test with isolates of diverse subtypes. J Clin Microbiol. 1999;37:2557–2563.
20. Walker BD, Yu XG. Unravelling the mechanisms of durable control of HIV
-1. Nat Rev Immunol. 2013;13:487–498.
21. Lambotte O, Boufassa F, Madec Y, et al.. HIV
controllers: a homogeneous group of HIV
-1-infected patients with spontaneous control of viral replication. Clin Infect Dis. 2005;41:1053–1056.
22. Hatano H, Delwart EL, Norris PJ, et al.. Evidence for persistent low-level viremia in individuals who control human immunodeficiency virus in the absence of antiretroviral therapy. J Virol. 2009;83:329–335.
23. Okulicz JF, Marconi VC, Landrum ML, et al.. Clinical outcomes of elite controllers, viremic controllers, and long-term nonprogressors in the US department of defense HIV
natural history study. J Infect Dis. 2009;200:1714–1723.
24. Goujard C, Chaix ML, Lambotte O, et al.. Spontaneous control of viral replication during primary HIV
infection: when is “HIV
controller” status established? Clin Infect Dis. 2009;49:982–986.
25. Merchant M, Wright M, Kabat W, et al.. Long-term highly suppressed HIV
-infected children and adolescents with negative rapid HIV
tests due to significant antibody loss. J Clin Virol. 2014;59:172–176.
26. Claassen M, van Zyl GU, Korsman SN, et al.. Pitfalls with rapid HIV
antibody testing in HIV
-infected children in the Western Cape, South Africa
. J Clin Virol. 2006;37:68–71.
27. Wendel SK, Mullis CE, Eshleman SH, et al.. Effect of natural and ARV-induced viral suppression and viral breakthrough on anti-HIV
antibody proportion and avidity in patients with HIV
-1 subtype B infection. PLoS One. 2013;8:e55525.
28. Laeyendecker O, Rothman RE, Henson C, et al.. The effect of viral suppression on cross-sectional incidence testing in the Johns Hopkins hospital emergency department. J Acquir Immune Defic Syndr. 2008;48:211–215.
29. Marinda ET, Hargrove J, Preiser W, et al.. Significantly diminished long-term specificity of the BED capture enzyme immunoassay among patients with HIV
-1 with very low CD4 counts and those on antiretroviral therapy. J Acquir Immune Defic Syndr. 2010;53:496–499.
30. Longosz AF, Mehta SH, Kirk GD, et al.. Incorrect identification of recent HIV
infection in adults in the United States using a limiting-antigen avidity assay. AIDS. 2014;28:1227–1232.
31. Sullivan AK, Savage EJ, Lowndes CM, et al.. Non-disclosure of HIV
status in UK sexual health clinics—a pilot study to identify non-disclosure within a national unlinked anonymous seroprevalence survey. Sex Transm Infect. 2013;89:120–121.
32. Marzinke MA, Clarke W, Wang L, et al.. Non-disclosure of HIV
status in a clinical trial setting: antiretroviral drug screening can help distinguish between newly-diagnosed and previously-diagnosed HIV
infection. Clin Infect Dis. 2014;58:117–120.
33. Piwowar-Manning E, Fogel JM, Laeyendecker O, et al.. Failure to identify HIV
-infected individuals in a clinical trial using a single HIV rapid test
for screening. HIV
Clin Trials. 2014;15:62–68.
34. Cohen MS, Chen YQ, McCauley M, et al.. Prevention of HIV
-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505.
35. Vitoria M, Vella S, Ford N. Scaling up antiretroviral therapy in resource-limited settings: adapting guidance to meet the challenges. Curr Opin HIV
37. Mullis CE, Munshaw S, Grabowski MK, et al.. Differential specificity of HIV
incidence assays in HIV
subtypes A and D-infected individuals from Rakai, Uganda. AIDS Res Hum Retroviruses. 2013;29:1146–1150.
38. Longosz AF, Morrison CS, Chen PL, et al.. Immune responses in Ugandan women infected with subtypes A and D HIV
using the BED capture immunoassay and an antibody avidity assay. J Acquir Immune Defic Syndr. 2014;65:390–396.
39. Hemelaar J, Gouws E, Ghys PD, et al.. Global trends in molecular epidemiology of HIV
-1 during 2000-2007. AIDS. 2011;25:679–689.
40. Klarkowski D, O'Brien DP, Shanks L, et al.. Causes of false-positive HIV
rapid diagnostic test results. Expert Rev Anti Infect Ther. 2014;12:49–62.
41. Everett DB, Baisely KJ, McNerney R, et al.. Association of schistosomiasis with false-positive HIV
test results in an African adolescent population. J Clin Microbiol. 2010;48:1570–1577.
42. Swaminathan S, Hanna LE, Sundaramurthi JC, et al.. Prevalence and pattern of cross-reacting antibodies to HIV
in patients with tuberculosis. AIDS Res Hum Retroviruses. 2008;24:941–946.
43. Trama AM, Moody MA, Alam SM, et al.. HIV
-1 envelope gp41 antibodies can originate from terminal ileum B cells that share cross-reactivity with commensal bacteria. Cell Host Microbe. 2014;16:215–226.