Secondary Logo

Journal Logo

Original Studies

Rapid Testing Algorithm Performance in a Low-Prevalence Environment

Martin, Eugene G. PhD; Cornett, Julia MD†,‡; Mohammed, Debbie Y. DrPH§; Salaru, Gratian MD

Author Information
Sexually Transmitted Diseases: May 2020 - Volume 47 - Issue 5S - p S35-S40
doi: 10.1097/OLQ.0000000000001138
  • Free

Among marginalized groups, stigma and poverty often interfere with acceptance of laboratory-based HIV testing in traditional health care settings.1 This has led to innovative approaches using simple rapid testing algorithms (RTAs) to achieve outreach, identification, and linkage to medical care for persons living with HIV (PLWHs). Many initiatives rely on rapid HIV diagnostic tests (RDTs) administered in community-based settings by operators not formally trained in laboratory technology.

Conducted with adequate quality assurance, and clearly defined protocols and procedures,2 RDTs can be performed with great sensitivity and specificity. Nonetheless, they are susceptible to operator errors including pipetting, timing, and reading-related issues. Adding a second “confirmatory” rapid test (RT#2) to an initial positive (RT#1) has advantages compared with the delayed confirmation associated with standard laboratory testing.3 An RTA can presumptively and reliably confirm an initial positive RT result in a high-prevalence setting. Although the World Health Organization's protocols allows for a third, tie-breaker rapid test to determine an HIV diagnosis in high-prevalence, resource-limited settings,4 in high-resource communities, an HIV diagnosis hinges upon subsequent laboratory-based confirmation.

In settings of very low HIV prevalence, the predictive value of a single positive rapid test result can be potentially improved by a strategy that uses a second, different manufacturer's rapid test to achieve independent (orthogonal) verification of the initial preliminary positive result.

In the United States, discordant RTA results are commonly resolved by use of the Centers for Disease Control and Prevention (CDC) laboratory algorithm,5 which recommends an instrumented antigen/antibody HIV combination screening immunoassay and, if reactive, followed by an HIV-1/HIV-2 antibody differentiation immunoassay. Instrumented antigen/antibody tests are preferred over a rapid combination (Determine Combo [DC] Ag/Ab) test, as they are more sensitive to the presence of p24 antigen, an early marker of an acute HIV infection.6,7 When the differentiation assay is interpreted as negative or indeterminate for HIV-1, a reflex to a qualitative HIV-1 nucleic acid test is recommended.

The New Jersey (NJ) RTA provides verification of initial, positive HIV1/2 antibody screening results, using an orthogonal rapid test (RT#2) as well as an expedited pathway into care for those identified as possibly very infectious (free p24 Ag ONLY). By modifying the RTA for use with a rapid HIV1/2 Ag/Ab test, we hoped to increase the identification of early infections. During a 3-year period (2015–2018), we sought to evaluate the performance of the algorithm and the utility of the antigen biomarker in a low-prevalence setting. We were interested in determining how often concordant rapid test results were confirmed as true infection, how often a discordant result portended a false-positive (FP) screen result, and how often a free p24 antigen-positive result was observed and subsequently confirmed.


Clinical faculty from Rutgers University– Robert Wood Johnson Medical School (RWJMS) (New Brunswick, NJ) support a statewide rapid screening network consisting of 136 sites (92 operated under RWJMS Clinical Laboratory Improvement Amendments [CLIA] waiver), whereas an additional 44 sites operate under CLIA waivers held by other entities. All sites use the NJ HIV quality assurance framework ( and procedures provided by RWJMS.

Rapid HIV test shipments are validated centrally using commercially available HIV performance panels and distributed from a single location to testing facilities throughout New Jersey. New Jersey HIV provides a centralized, quality assurance program and product training, as well as hands-on training in the use of all approved RT devices. A group of 5 medical technologists and 2 laboratory professionals provides technical support to sites engaged in rapid HIV screening, including a centralized program providing RTA discordant resolution. Monthly reports document the distribution of inventory and aggregate rapid test performance. Quality control performance is documented on log sheets and reviewed by assigned medical technologists for RWJMS sites. Approximately 30% of all rapid tests purchased for the program are used at sites for weekly quality control testing. In addition, all sites participate in the American Association of Bioanalysts proficiency testing programs for the 2 rapid tests used in the RTA. The 136 screening venues includes 26 community-based organizations, 12 mobile vans, multiple health care clinics, county correction facilities, federal qualified health centers, health departments, hospitals, university clinics, and substance abuse treatment facilities.

A common lexicon is used throughout this article: RT#1 refers to the Abbott/Alere (Scarborough, ME) DC HIV-1/2 Ag/Ab. This rapid screening test uses 2 biomarkers, one identifying free p24Ag and the other HIV1/2 antibodies. If either biomarker was positive, according to CDC surveillance terminology, the client was designated as PRELIMINARY POSITIVE for HIV infection. Those identified by RT#1 as AG+(ONLY) were immediately referred for care and laboratory diagnostic confirmation at a health care facility. All other preliminary positive screens, including any HIV-1/2 antibody-positive screen (with or without free p24Ag), were subject to a second rapid test (RT#2) to achieve orthogonal verification. RT#2 was usually a Trinity Biotech Uni-Gold Recombigen HIV-1/2 (Bray, Co., Wicklow, Ireland). Infrequently, an Orasure OraQuick Advance Rapid HIV-1/2 Antibody rapid test (Orasure Technologies, Bethlehem, PA) was also used. These results were either concordant (RT#1 and RT#2 both positive for HIV1/2 antibodies) and the client designated presumptive positive or they were discordant: (RT#1 positive and RT#2 negative) requiring resolution using the CDC HIV laboratory diagnostic algorithm.

Laboratory confirmation is completed at the linkage facility. The linkage site is asked to return documentation on any negative confirmatory test results that they might obtain during follow-up laboratory confirmation of the infection.

Regardless of the outcome, a Preliminary Positive Tracking Form is sent to NJ HIV for each positive rapid test result. This form provides information on all rapid tests performed, including specimen type, rapid test results, site numbers involved in testing, operator numbers, test lot numbers, and information regarding client referral to care.

When the initial RT#1 includes an antibody-positive biomarker, a second rapid test is performed. When the RT#2 is negative, the results are considered discordant. Additional whole blood is collected, spun, and either refrigerated (serum separator tube) or frozen (white top: K2EDTA and gel for plasma separation tubes), collected by the NJ HIV support team, and forwarded to a contracted reference laboratory (Quest Diagnostics, Teterboro, NJ) for resolution. Specimens do not contain any personal identifiers; an anonymized test number is used to identify discordant specimens. Non-RWJMS locations are required to forward deidentified discordant resolution data when they are collected and processed at another facility.

Mandatory use of common format test logs and the communication of positive testing data through a standardized Preliminary Positive Tracking Form provides documentation of all HIV screening results including initial (RT#1) and second (RT#2) RT results, as well as documentation of confirmatory data obtained by the site, deidentified and returned to NJ HIV in Somerset. The form allows for outcome tracking for initial DC screens reactive to p24Ag+, HIV1/2 Ab+, or both.

Treatment sites follow updated recommendations for diagnosis of HIV infections5 and confirm results before initiating HIV genotyping and other testing associated with the initiation of antiretroviral therapy. Ultimately, the medical provider at the linkage facility is responsible for ordering these tests and reporting these results to the NJ Department of Health HIV Surveillance Program, as well as for faxing confirmatory information to NJ HIV Somerset offices.

All statistical calculations including HIV sensitivity, specificity, accuracy, and positive predictive value were acquired using Medcalc, an online calculator.8 Confidence intervals (CIs) for sensitivity, specificity, and accuracy are “exact” Clopper-Pearson CIs. Confidence intervals for reported predictive values are standard logit CIs.9


Between 2015 and 2018 (Fig. 1), 310,785 rapid DC Ag/Ab screening tests were performed identifying 2400 as preliminary positive screens. Of these, 2191 (91.8%) were verified with a second, different rapid test. Those found to be only DC AG-POS (n = 13) on initial screen were immediately linked to medical care. After laboratory confirmation, only 1 (7.7%) subsequently confirmed as HIV infected after additional testing.

Figure 1
Figure 1:
Illustrates the 3 pathways used in the NJ HIV rapid testing algorithm. Initial rapid screening performed with RT#1 (DC Ag/Ab) identified 2400 preliminary positive clients. Of these, 13 presented as free p24Ag + ONLY pathway and were immediately linked to health care for laboratory confirmation of infection and potential HIV staging. The remaining 2387 individuals were subject to RT#2 at the screening facility using an orthogonal rapid test (RT#2) to verify the presence of HIV antibodies via the RTA concordant pathway. These were linked to care within the same or the next business day. According to CDC surveillance criteria, they were designated presumptively positive and referred to care for diagnostic laboratory confirmation, staging, and initiation of antiretroviral therapy. Of the 2387 RTAs performed, 196 (8.2%) failed to verify and were handled by the RTA discordant pathway. Specimens were drawn, collected and transported to NJ HIV, and forwarded to a commercial laboratory for additional laboratory testing. Of these, 182 (92.9%) were determined to be initial FP RT#1. Only 14 (7.1%) were subsequently determined to be truly HIV infected. These were linked to care with a delay averaging 4 days.

Of the 196 discordant RTA results, 182 (92.9%) were determined to be FP screens, including 13 specimens that were DC AG (POS)/AB (POS) by rapid testing. Fourteen discordant specimens were resolved to be DC HIV true-positive (TP) infections. The average quantitative HIV-1 viral load was 277,385 copies/mL, although 5 (35.7%) of the 14 cases had viral loads <1000 copies/mL.

Table 1 provides a summary of rapid HIV screening in New Jersey between 2015 and 2018. It details the number of preliminary positives identified by RT#1 (2400), biomarkers associated with the overall outcome (p24 antigen, HIV1/2 antibodies, or both in the initial screen). It details the number of RTAs performed (2387) and outcomes associated with either a concordant RTA or a discordant RTA.

Summary: NJ HIV Rapid Testing Algorithm (2015–2018)

Among the 2191 “presumptive positive” by the RTA, 3 FP RTAs were subsequently reported, with both rapid tests being positive, whereas the laboratory-based HIV-1/2 Ag/Ab assay and quantitative HIV-1 viral load were both negative. Among 2203 individuals living with HIV in a low-prevalence setting (0.71%), the sensitivity of the RTA (TP rate) was 99.36% (95% CI, 98.94%–99.65%), whereas the specificity was 98.38% (95% CI, 95.33%–99.66%). The overall probability that a tested client was correctly classified by the RTA (accuracy) was 98.39% (95% CI, 97.80%–98.85%). The positive predictive value of RT#1 was 30.47% (95% CI, 12.48%–57.38%).


Globally, RDT using an RTA has made HIV diagnosis and treatment more accessible while reducing the number of PLWHs lost to care. However, multiple factors, including operator and clerical errors, inconsistent visual interpretation by operators,10,11 problems with the implementation of testing strategies, and neglected postscreen follow-through, all contribute to missed cases of infection and false diagnoses that waste significant clinical resources12 and ultimately delay treatment.

Clients screened as preliminary positive with a single reactive rapid test often fail to return to receive definitive test results, are not linked to care, are delayed in their entry into care, and/or fail to enter care at all.3 By combining a second rapid test along with a collaborative network of linkage and treatment facilities, 91.8% of presumptive positive screens in New Jersey were identified and linked to care within 7 days, supporting the National HIV/AIDS Strategy goal of linking at least 85% of PLWH into medical care within 30 days.13,14

During a 3-year period, the NJ RTA algorithm presumptively identified 2204 PLWHs in a low-prevalence setting (0.71%). Of 196 discordant RTAs, only 14 instances were determined to be truly positive, a reflection of the intrinsic value of orthogonal testing, as well as a potential reflection of the robust training and quality assurance procedures extant statewide. These procedures include hands-on training, monthly site visits from a trained medical technologist, mandatory proficiency testing, and centralized discordant management.

Fingerstick-based DC Ag detection has not been as successful as anticipated. We report a single instance in which laboratory testing confirmed an early HIV infection. Although performance of the DC Ag biomarker using plasma specimens is less sensitive than an instrumented laboratory-based Ag/Ab combo assay, detection of acute HIV infection has been described.15,16 The limited number of positive DC fingerstick antigen biomarkers may represent the outcome of screening in a low-prevalence setting, although a pooled NAAT quality assurance project completed in Newark, NJ, after a negative rapid antibody screening result (StatPak; Chembio) identified 8 cases of acute HIV infection among 6845 screened negative by a rapid antibody test.17

As initial screening, rapid tests are used in a variety of risk settings.18 The positive predictive value of even the most specific tests is highly dependent on disease prevalence.19 In a high-prevalence setting, it is more likely that persons who test positive truly have a disease than if the test is performed in a population with low prevalence.

Although highly resourced countries ultimately rely on laboratory-based confirmatory algorithms, the use of an RTA at CLIA-waived screening sites permits rapid identification and credible linkage for individuals who might otherwise refuse additional testing. Expanding screening opportunities and eliminating barriers that impact linkage ultimately decrease morbidity, mortality, and transmission.

In an earlier, large-scale retrospective evaluation, we reported a rate of 1 FP result in 51,413 tests20 using an RTA. The current study suggests that FP results may be even less frequent. This study reports a rate of 1 FP results in 103,595 tests after initial screening by DC Ag/Ab.

Of discordant RTA specimens, 92.9% were found to be FP by initial rapid DC screens. Only 7.1% of the 196 discordant specimens resolved as true HIV1/2-positive infections (i.e., false-negative second rapid Trinity Unigold) results. Although differential rapid test sensitivity might explain the 14 instances in which detection of HIV infection by DC (RT#1) was discordant with the negative results of RT#26, the eventual false positive outcome (92.9%) of most RTA discordants is not attributable to differential sensitivity of orthogonal rapid tests.

Messaging at the point-of-screening needs to accurately reflect both residual risk and the likely outcome of definitive testing. Given the likelihood (92.9%) that a discordant result will resolve as an FP initial screen provides both realistic caution and some reassurance during the initial screening encounter. On the other hand, verification of a preliminary positive at the point of screening clearly provides near-certain assurance (99.9%) of likely infection.

The CDC has recommended universal screening in the United States since 200621 with additional recommendations for annual rescreening for those at increased risk for infection.22 Such recommendations acknowledge the importance of identifying cases associated with both ongoing high-risk behaviors and cases that could potentially go undetected until late in the course of the disease.23 In the absence of strategies designed to improve the accuracy of screening algorithms and credible referral processes, an increase in screening from low-prevalence communities could challenge the credibility of point-of-care screening.

Although options for efficient population screening need to be expanded and testing barriers confronted, appropriate entry into care and initiation of timely therapeutic processes are also critical. Currently, less than 40% of the US population has been screened for HIV exposure, and screening in the 50 US counties where more than half of all HIV diagnoses occur suggests that less than 35% of people recommended for annual HIV screening actually received testing in the preceding year.24 Among marginalized groups, the ability to provide screening for exposure in less threatening, more accessible venues including community-based venues maybe essential to obtaining adequate population coverage.

An RTA should refine diagnostic probabilities. Definitive, accurate diagnosis of HIV infection is critical to credible linkage, as is effective initiation of care. Key benefits of a sequential, 2-step, RTA are the elimination of confirmatory delays and the improvement of linkage processes by reducing FP referrals.

Linkage and initiation of antiretroviral therapy are the first steps. Whether modeled mathematically25 or tested through field trials,26 screening without subsequent linkage, treatment, adherence to treatment, and retention in care, initial viral suppression is ineffective at reducing HIV incidence. In 2018, the Treatment as Prevention trial increased HIV screening, but viral suppression was low because of inadequate linkage to care, especially among men and young people.27

The use of an RTA allows for immediate, on-site verification of most preliminary HIV antibody-positive results and permits an expedited linkage into care through collaborative linkage programs. By modifying the RTA process to include immediate referral to care for all preliminary positive results associated with free p24 Ag, we sought to maximize potential sensitivity at the point of screening and to engage those with early HIV infections in care. In 310,785 screens, only 13 instances of a solitary, free p24 Ag marker occurred. Of these, only a single case was established during laboratory confirmation. Such findings suggest that immediate referral into care based on a positive fingerstick free p24 antigen marker is unwarranted based on the likelihood of an FP outcome.

There are several advantages to using an RTA in a high-resource, low-prevalence environment. It is important to note that it maximizes the ability of HIV prevention programs to tailor screening activities in communities at risk, allows for expedited entry into care with assurance that the referral will subsequently confirm, identifies problematic referrals at the time of the rapid screening event, reduces the number of screen-positive clients lost to care, and permits definitive diagnosis before entry into care. In addition, 91.8% of initial preliminary positive screens identified by DC Ag/Ab were confirmed by a second RT, referred to surveillance as “presumptively confirmed” and linked into the care continuum within a matter of days. An RTA should refine diagnostic probabilities, and improve the credibility of patient referrals and the likelihood that the patient will ultimately begin treatment expeditiously.

The significance of the quantitative viral load for the 14 discordant specimens resolved to be truly positive is unclear. Although the average quantitative HIV-1 viral load was 277,385 copies/mL across all HIV-positive, discordant specimens, 35.7% (5) had viral loads <1000 copies/mL, which may have a variety of explanations including failure to divulge existing treatment at the time of screening, the presence of elite controllers in the screening population, and others.

There are 3 significant limitations to consider.

Limitation 1: The number of presumptive positives that are ultimately identified as biologic FPs may underestimate the true number that occur because we rely on self-reporting by clinical staff and the actual number is not automatically captured with internal program reporting. Our data provide a measure of performance applicable to sites with a well-established quality assurance system. Although our preliminary positive tracking system provided a means for clinical sites to report concordant rapid tests that fail to subsequently confirm, we rely on clinical sites to initiate reports of inconsistent outcomes at the treatment site. This may not ensure universal compliance because we are dependent on sites to alert us to situations in which both RTs were reported as positive, the client was referred into care and the confirmatory laboratory tests subsequently failed to support a conclusion of HIV1/2 infected.

Limitation 2: Rapid testing does not provide a measure of the number of acute HIV infections or negative rapid screens (missed infections) associated with less than optimal test sensitivity. Because of the absence of a more sensitive criterion standard such as a pooled nucleic acid test, the negative predictive value is not truly measurable in this study. An ongoing quality assurance project after a negative DC Ag/Ab screening result has demonstrated cases of missed HIV infection presumably because of underreading by operators of faintly positive lines on the rapid test, but has not demonstrated substantial numbers of early missed infections.

Limitation 3: The 2188 concordant positives are entered into CDC Evaluation Web (Luther Consulting, LLC) as a presumptive positive diagnosis by point-of-care site personnel. Although rapid testing sites are asked to inform the program if laboratory confirmation at the health care facility is unsuccessful, we have limited access to laboratory confirmatory data.

Even in settings of low prevalence, RTAs provide an important tool in achieving the National HIV/AIDS Strategy goal of linking at least 85% of PLWHs into medical care within 30 days. In addition, by improving the credibility of the referral process, sites maximize available clinic time for patients truly in need of a clinical encounter.


1. Pellowski JA, Kalichman SC, Matthews KA, et al. A pandemic of the poor: Social disadvantage and the U.S. HIV epidemic. Am Psychol 2013; 68:197–209.
2. Pai N, Wilkinson P, Deli-Houssein S, et al. Barriers to implementation of rapid and point-of-care tests for human immunodeficiency virus infection: Findings from a systematic review (1996–2014). Point Care 2015; 14:81–87.
3. Delaney KP, Rurangirwa J, Facente S, et al. Using a multitest algorithm to improve the positive predictive value of rapid HIV testing and linkage to HIV care in nonclinical hiv test sites. J Acquir Immune Defic Syndr 2016; 71:78–86.
4. Kosack CS, Shanks L, Beelaert G, et al. Designing HIV testing algorithms based on 2015 WHO guidelines using data from six sites in Sub-Saharan Africa. J Clin Microbiol 2017; 55:3006–3015.
5. Branson BM, Owen SM, Wesolowski LG, et al. Centers for Disease Control and Prevention and Association of Public Health Laboratories. Laboratory Testing for the Diagnosis of HIV Infection: Updated Recommendations. Available at: Accessed July 22, 2019.
6. Delaney KP, Hanson DL, Masciotra S, et al. Time until emergence of HIV test reactivity following infection with HIV-1: Implications for interpreting test results and retesting after exposure. Clin Infect Dis 2017; 64:53–59.
7. Masciotra S, Luo W, Westheimer E, et al. Performance evaluation of the FDA-approved Determine HIV1/2 Ag/Ab Combo assay using plasma and whole blood specimens. J Clin Virol 2017; 91:95–100.
8. Medcalc Software. [computer program]. Version 19.1.3. Ostend, Belgium: MedCalc Software, 2019: Available at: Accessed August 31, 2019.
9. Mercaldo ND, Lau KF, Zhou XH. Confidence intervals for predictive values with an emphasis to case-control studies. Stat Med 2007; 26:2170–2183.
10. Grebe E, Owen R, Facente S, et al. Independent assessment of the Sedia Asanté HIV-1 Rapid Recency Assay. Presented at: 2019 HIV Diagnostics Conference, Atlanta, GA, March 25–28, 2019. Available at:é-Evaluation.pdf. Accessed August 29, 2019.
11. Martin EG, Cappello J, Ippolito T, et al. Side by side comparison: DPP-HIV syphilis assay system vs. syphilis health check. Presented at: 2019 HIV Diagnostics Conference, Atlanta, GA, March 25–28, 2019. Available at: Accessed August 29, 2019.
12. Khan S, Marfa E, Pasipamire M, et al. Identification of misdiagnosed HIV clients in an early access to ART for all implementation study in Swaziland. J Int AIDS Soc 2017; 20(Suppl 6):21756.
13. Mohammed D, Davidow A, Martin E, et al. Effect of a rapid test algorithm on linkage to medical care and viral suppression among New Jersey residents, 2007–2015. Sex Transm Dis 2018; 45:476–481.
14. White House Office of National AIDS Policy. National HIV/AIDS Strategy: Updated to 2020. 2015. Available at: Accessed April 3, 2019.
15. Delaney K, Violette L, Ure G II, et al. (2018). Estimated time from HIV infection to earliest detection for 4 FDA-approved point-of-care tests. Available at: Accessed September 24, 2019.
16. Stekler JD, Ure G, O'Neal JD, et al. Performance of Determine Combo and other point-of-care HIV tests among Seattle MSM. J Clin Virol 2016; 76:8–13.
17. Martin EG, Salaru G, Mohammed D, et al. Finding those at risk: acute HIV infection in Newark, NJ. J Clin Virol 2013; 58(Suppl 1):e24–e28.
18. False-positive HIV test results. May 2018. Available at: Accessed September 23, 2019.
19. Greig JR, Batchelor D, Wallis M. Positive predictive value is poor in low risk populations seen in universal screening for HIV infection. BMJ 2013; 346:f3575.
20. Martin EG, Salaru G, Paul SM, et al. Use of a rapid HIV testing algorithm to improve linkage to care. J Clin Virol 2011; 52(Suppl 1):S11–S15.
21. Branson BM, Handsfield HH, Lampe MA, et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep 2006; 55(RR-14):1–17; quiz CE1–4.
22. DiNenno EA, Prejean J, Irwin K, et al. Recommendations for HIV screening of gay, bisexual, and other men who have sex with men—United States, 2017. MMWR Morb Mortal Wkly Rep 2017; 66:830–832.
23. Sugarman J. Implications of universal screening for HIV infection. BMJ 2013; 346:f2957.
24. Pitasi MA, Delaney KP, Brooks JT, et al. HIV testing in 50 local jurisdictions accounting for the majority of new HIV diagnoses and seven states with disproportionate occurrence of HIV in rural areas, 2016–2017. MMWR Morb Mortal Wkly Rep 2019; 68:561–567.
25. Perry A, Kasaie P, Dowdy DW, et al. What will it take to reduce HIV incidence in the United States: A mathematical modeling analysis. Open Forum Infect Dis 2018; 5:ofy008.
26. Abdool Karim SS. HIV-1 epidemic control—Insights from test-and-treat trials. N Engl J Med 2019; 381:286–288.
27. Iwuji CC, Orne-Gliemann J, Larmarange J, et al. Universal test and treat and the HIV epidemic in rural South Africa: A phase 4, open label, community cluster randomised trial. Lancet HIV 2018; 5:e116–e125.
Copyright © 2020 American Sexually Transmitted Diseases Association. All rights reserved.