Introduction
Rapid HIV tests at the point-of-care may increase HIV testing and linkage to care [1]. They are used in countries lacking well resourced pathology services and also in high-income countries in an effort to increase HIV testing in high-risk populations, particularly men who have sex with men (MSM) [1]. Compared with laboratory-based tests, HIV rapid tests spare clients the anxiety of waiting for the HIV result and reduce the number of individuals diagnosed with HIV who do not receive their result [1,2].
Regular and frequent testing of populations at risk of HIV is necessary for timely initiation of antiretroviral treatment to have a preventive effect in populations [3]. Recognizing this, the United Nations Programme on HIV/AIDS has set a target for 2020 of diagnosing 90% of all people living with HIV [4]. However, in populations experiencing ongoing transmission of HIV, increasing the frequency of testing will increase the proportion of people who, when tested, have recently acquired infection lacking detectable antibody (acute infection). In this situation, HIV rapid tests will have reduced sensitivity because of their longer window periods [5]. This has been suggested by recent studies in clinics serving MSM in which the sensitivity of HIV rapid tests compared with fourth-generation enzyme immunoassays (EIA) was 85–87% [1,6,7]. In high-income countries, MSM are likely to be tested more frequently than populations in low-income countries [8–11].
The reported sensitivity of a test also depends on the sensitivity of the comparator or reference standard. Manufacturers of four commonly studied HIV rapid tests quote sensitivities of 99.3–100%, but all were compared with earlier EIAs that are less sensitive than the now widely available fourth-generation EIAs [12–16]. Fourth-generation EIAs can detect HIV p24 antigen as early as 2 weeks postinfection, as well as being able to detect the antibodies that appear later [17]. They are recommended as first-line screening tests in the 2014 Centers for Disease Control and Prevention HIV laboratory algorithm [18]. HIV nucleic acid amplification tests (NAATs), which are capable of detecting HIV RNA in the serum from 10 days postinfection, are used in some specialist centres to diagnose acute HIV infection, before antibody has developed, although they are not recommended for routine screening [18]. Most current HIV rapid tests detect HIV antibodies only. There is one Food and Drug Administration approved HIV rapid test designed to detect both HIV antibody and p24 antigen, the Determine HIV-1/2 Antigen/Antibody Combo (Alere, Japan). This appeared more sensitive than other HIV rapid tests in laboratory-based studies of stored sera from seroconverters; however, in a recent review of clinical studies, the p24 antigen component failed to detect a single case of acute infection [19–21]. Although antibody-only HIV rapid tests will be unable to detect infections before any antibody has appeared, studies of frozen sera from seroconverting individuals demonstrate some variation in window periods between HIV rapid tests [22]. Although the proportion of acute infections can be expected to vary across different populations, there are no studies examining the effect of this on the performance of HIV rapid tests.
Health services that have the option of providing either HIV rapid tests or the more sensitive laboratory assays therefore need to know whether there will be a significant loss of sensitivity if they use HIV rapid tests in their clinical population. Our hypothesis was that the reduction in sensitivity of HIV rapid tests, compared with modern laboratory-based tests, varies according to the populations in which they are used, independently of assay type. We therefore systematically reviewed studies of adults in clinical settings, comparing the sensitivity of HIV rapid test with fourth-generation EIA or NAAT, separately, for their ability to detect HIV infection of any duration, acute or established.
Methods
This meta-analysis was conducted according to the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [23]. (See the checklist in Supplementary file 1, https://links.lww.com/QAD/A923.)
Protocol and registration
The meta-analysis is registered with Prospective Registration of Systematic Reviews, number: CRD42015020154.
Search strategy
Medline, PubMed, Embase, Cochrane Controlled Trials Register, Cochrane Reviews and Cumulative Index to Nursing and Allied Health Literature were searched for papers in English from the earliest date up to 14 July 2015.
The search included Medical Subject Headings terms for ‘HIV’, ‘HIV-1’, ‘HIV infections’, ‘Acquired Immunodeficiency Syndrome’, ‘HIV seropositivity’, ‘Point-of-Care Systems’, ‘Enzyme-linked immunosorbent assay’, ‘HIV Core protein p24/blood’, ‘Nucleic acid amplification techniques’, ‘RNA, Viral/blood’ and free-text terms associated with HIV rapid tests and the desired comparators: ‘rapid test’, ‘ELISA’, ‘EIA’, ‘Enzyme linked Immunoassay’, ‘NAAT’, ‘NAT’, ‘RNA amplification’ and ‘pooled NAAT’.
Inclusion and exclusion criteria
We searched for English language articles in which adults who had point-of-care HIV testing whether via blood or oral fluid HIV rapid test and had parallel comparator tests of a fourth-generation HIV EIA antibody/p24 antigen or HIV NAAT (pooled or undiluted). Articles included must have screened HIV rapid test-negative samples with a comparator and provided sufficient data to calculate HIV rapid test sensitivity. Data enabling calculation of specificity were sought but were not essential for inclusion. To rule out false positive fourth-generation EIA results, studies using this comparator were only included if there was a confirmatory test with a different EIA/Line Immunoassay/NAAT/p24-antigen assay capable of detecting early infection. Studies using fourth-generation EIA comparators that were confirmed only with western blots were excluded because a negative western blot could represent early HIV infection or a false positive EIA. Studies were excluded if they tested stored sera or children aged under 16.
Data extraction process
The following data were extracted: study design, participant numbers, country, study setting, comparator, confirmatory tests, and raw numbers of true and false positive, and true and false negative HIV rapid test results. These data were extracted by WST and checked by TRHR independently. Disagreements were resolved by discussion between the two authors and a third (C.K.F.). For studies requiring further data or clarification on type of EIA, the data were requested from the corresponding author. If there was no response after two emails, the study was excluded. Several studies compared multiple HIV rapid tests, providing multiple estimates of sensitivity and specificity. Each estimate became a single data point in the meta-analysis.
Outcomes
The primary outcome was the sensitivity of the HIV rapid test for all HIV infection (whether acute or established HIV). Sensitivity was calculated by dividing the number of confirmed positive HIV rapid tests by the number of confirmed positive comparator tests. For studies with sufficient data, specificity was calculated by dividing the number of confirmed negative HIV rapid tests by the number of negative comparator tests.
Analysis
Meta-analysis was used to calculate pooled estimates of sensitivity and specificity and 95% confidence intervals (CIs). The I2 statistic was used to determine the proportion of total variability in point estimates that could be attributed to heterogeneity, and the chi-squared-based Cochran Q test was used to assess heterogeneity at the significance level of 0.1. Pooled estimates of sensitivity and specificity were calculated from both fixed effects and random effects models. The two-sample z test was used to compare subgroups.
In prespecified analyses, sensitivity was first stratified by the comparator test used (because NAAT have a shorter window period than EIA) and by whether the study population was known to contain greater than 50% MSM. In a subsequent analysis, sensitivity was stratified by the setting (high-income or low-income country) according to World Bank gross national income per capita data [24]. A subgroup analysis was performed, if data were available, on the sensitivity of the antigen component of the HIV antibody/p24 antigen HIV rapid test in acute HIV (defined as HIV NAAT positive, HIV antibody negative and p24 antigen positive/negative cases). Because a significant proportion of studies in high-income countries using NAAT as comparator used an oral fluid rapid test, this analysis was further stratified into studies of finger stick blood HIV rapid tests and studies of oral fluid HIV rapid tests. Specificity was stratified by comparator test only.
Assessment of bias and quality
We assessed publication bias with a funnel plot using the Begg and Mazumdar rank correlation method at the significance level of 0.05 [25–27]. We used the QUADAS-2 tool by the University of Bristol to assess bias within studies [28]. Four domains were used to assess the risk of bias: patient selection, conduct of index test, conduct of reference test and flow of patients through the study. The quality of each item was categorized as either ‘Low risk (+)’, ‘High risk (+++)’ or ‘Unclear (?)’. Data were analysed using Comprehensive Meta-Analysis Software (version 2.2, Biostat; Englewood, New Jersey, USA).
Results
The review process is shown in Fig. 1 and the selected papers summarized in Table 1[5–7,29–43]. Of the 953 titles identified from the six electronic databases, 44 were duplicates yielding 909 unique references. The 850 references irrelevant to the meta-analysis were excluded on abstract alone. We reviewed 59 full text articles and contacted eight authors for further information and seven responded. Two of these articles were subsequently included in the meta-analysis [41,43], and the remainder were excluded because of ineligibility or lack of author response. Eighteen studies meeting the inclusion criteria were included in the final meta-analysis.
;)
Fig. 1: PRISMA diagram of study selection for meta-analysis.
Table 1: Setting, population characteristics, test types, sensitivity and specificity of individual studies.
Table 1: (Continued) Setting, population characteristics, test types, sensitivity and specificity of individual studies.
Characteristics of included studies
The 18 studies provided 29 estimates of sensitivity and included a total of 110 122 HIV rapid tests performed, of which 12 723 (11.5%) true positives were confirmed by either a fourth-generation EIA or HIV NAAT. There were five EIA comparator studies (370 true positive HIV rapid test) and 13 NAAT comparator studies (12 333 true positive HIV rapid tests).
Nine studies were conducted in high-income countries (893 true positive HIV rapid tests out of 55 298 HIV rapid tests performed) including the United States, the United Kingdom and Australia. Nine studies were conducted in low-income countries (11 830 true positive HIV rapid tests out of 54 824 HIV rapid tests performed) including Africa, Malawi, Swaziland, Tanzania, Mozambique, Burkina Faso and Nigeria. Thirteen of the 18 studies used HIV antibody-only HIV rapid test, whereas five studies used the Determine Combo HIV antibody/p24 antigen HIV rapid test (Table 1).
Five studies were conducted in populations that were greater than 50% MSM, and 12 studies were conducted in sexual health clinics or screening clinics for people at high risk for HIV (Table 1). Two studies were of women only: one in pregnant women attending an antenatal clinic [30], and the other in young women working in the food and beverage industry in Tanzania [32]. One study screened febrile patients for acute HIV infection [41].
Risk of bias
Of the 18 included studies, three were at low risk on all four quality domain indicators (see tables in Supplementary files 2 and 3, https://links.lww.com/QAD/A923). Five had one domain assessed as high risk for bias. None of the included studies were assessed as high risk on all four domains. Ten studies performed multiple HIV rapid tests in parallel without commenting on blinding of the reader of one test to the result another, so we assessed these as having an unclear risk of bias. No significant publication bias was observed among the 18 studies as the P value for Begg's test (continuity corrected) was 0.55.
Sensitivity
The pooled sensitivity of HIV rapid tests in the 13 studies using NAAT as a comparator was 93.7% (95% CI: 88.7–96.5). The pooled sensitivity compared with NAAT in low-income countries was 97.4% (95% CI: 94.1–98.9), which was significantly higher than in high-income countries (85.2%; 95% CI: 80.4–89.0, P < 0.001) (Fig. 2). Because of significant heterogeneity between studies, the random effects model was used. Pooled estimates of sensitivity, stratified by subgroups of comparator, test type and population and corresponding measures of heterogeneity are in Table 2.
Fig. 2: Forest plot of sensitivity of HIV rapid tests compared with nucleic acid amplification tests, in high-income and low-income countries.The centres of the squares represent study estimates, the centres of the diamonds represent pooled estimates and the horizontal lines represent 95% confidence intervals.
Table 2: Meta-analysis of 18 publications comparing the sensitivity of HIV rapid tests to laboratory tests in clinical settings, by national income level and type of laboratory reference standard.
The pooled sensitivity of HIV rapid tests in the five studies using fourth-generation EIA as a comparator was 94.5% (95% CI: 87.4–97.7). The pooled sensitivity compared with EIA in low-income countries (98.8%; 95% CI: 96.5–99.6) was significantly higher (P = 0.001) than in high-income countries (88.2%; 95% CI: 80.0–93.4) (Fig. 3).
Fig. 3: Forest plot of sensitivity of HIV rapid tests compared with fourth-generation enzyme immunoassays, in high-income and low-income countries.The centres of the squares represent study estimates, the centres of the diamonds represent pooled estimates and the horizontal lines represent 95% confidence intervals. The sensitivity in the study by Kania
et al. [35] has been artificially reduced from 100% to enable inclusion in the meta-analysis.
The pooled sensitivity of HIV rapid tests in high-income countries overall (using EIA or NAAT as comparator) was 85.7% (95% CI: 81.9–88.9) and in low-income countries overall sensitivity was 97.7% (95% CI: 95.2–98.9). In studies conducted in high-income countries, five estimates of oral fluid HIV rapid tests had a sensitivity of 82.1% (95% CI: 78.0–85.6), and 10 estimates of finger stick blood HIV rapid tests had a sensitivity of 87.0% (95% CI: 82.1–90.7). The sensitivity of HIV rapid tests from eight estimates in five studies of MSM populations, all in high-income countries, compared with EIA or NAAT, was 80.9% (95% CI: 76.9–84.3) (fixed effects, I2 = 0, P = 0.80) (Table 2).
Ten studies provided information on the proportion of acute HIV among those diagnosed with HIV infection. Meta-analysis of these studies gave the proportions which were antibody negative on EIA but NAAT positive: 13.6% (95% CI: 10.1–18.0) in high-income countries [5–7,34,37,38,42] and 4.7% (95% CI: 2.8–7.7) in studies from low-income countries [40,41,43]. The fixed effects model for proportions was used as there was no significant heterogeneity across the studies. (High-income countries, I2 = 10.6; P = 0.3; low-income countries: I2 = 0; P = 0.8.)
Four studies provided sensitivity data for the p24 antigen component of the fourth-generation HIV rapid test in the small number of seronegative acute HIV, and the overall sensitivity was 0% (95% CI: 0–0.1) [7,31,34,40].
Specificity
The overall specificity from 23 estimates with available data was 99.8% (95% CI: 99.5–99.9) (random effects, I2 = 95.4, P < 0.001). The specificity from 16 estimates using NAAT as a comparator was 98.1% (95% CI: 97.9–98.2) (random effects, I2 = 96.4, P < 0.001). The specificity from seven estimates using EIA as a comparator was 99.6% (95% CI: 99.4–99.7) (fixed effects, I2 = 22.6, P = 0.26).
Discussion
This meta-analysis found HIV rapid tests to be less sensitive when used in clinical settings in high-income countries, compared with low-income countries, regardless of whether they were compared with a NAAT or fourth-generation EIA. The discrepancy between high-income and low-income countries could be attributed to the higher proportion of acute HIV infections (antibody-negative NAAT-positive) in populations tested in high-income countries. This may reflect higher background testing rates and/or a higher incidence of HIV in MSM who were the primary focus of studies in high-income countries. We estimate that in high-income countries about one in seven infections are missed by HIV rapid tests when they are used alone. These infections are likely to be particularly transmissible because of the high HIV viral load in early infection [44,45]. Our data indicate that in high-income countries HIV rapid tests should be used in combination with fourth-generation EIA or NAAT, except in special circumstances.
The lower sensitivity of HIV rapid tests in high-income countries is consistent with the nearly three-fold greater proportion of acute infections in study populations from those countries. All studies of MSM were in high-income countries, with an overall sensitivity of 81%, suggesting that characteristics of this population are a factor. A plausible explanation is that the MSM population has a high incidence and, compared with the less-selected populations studied in low-income countries, a higher rate of HIV testing, contributing to the higher proportion of acute infections and the lower sensitivity of HIV rapid tests [8–11]. A tradeoff between sensitivity and specificity seems an unlikely explanation as the HIV rapid tests are only able to detect antibody, which is absent in acute infection. We cannot exclude other explanations such as differences in training and competence of testers between settings, but this seems unlikely given the consistency of findings across countries and different types of test.
This review has some limitations. First, some studies were designed to increase the detection of acute HIV, rather than to measure the sensitivity of HIV rapid tests. Therefore, they may not have been designed to limit bias for sensitivity calculations. Second, 11 studies compared multiple HIV rapid tests, and it may not have been practical to blind readers to the result of another HIV rapid test conducted in parallel, potentially introducing bias. Third, all studies of oral fluid rapid tests were conducted in high-income countries, and a previous meta-analysis has shown oral fluid HIV rapid tests to be less sensitive than finger stick blood HIV rapid tests [46]. However, when we analysed blood and oral HIV rapid tests separately, the pooled sensitivity of finger stick blood HIV rapid tests was still only 87% in high-income countries.
This is the first systematic review of studies comparing HIV rapid tests to fourth-generation EIA and NAAT. Earlier reviews by Yoo [47] and Pai [46] reported a pooled HIV rapid test sensitivity of 99.7 and 98.0% (oral)/99.7% (blood), respectively, but all studies used less-sensitive earlier generation EIA as comparators that would increase the apparent sensitivity of the HIV rapid test. Lewis et al. recently reviewed four studies of the antigen component of the Determine Combo antigen–antibody test in acute infection [21]. The antigen test had zero sensitivity in acute infection, consistent with our findings. The same studies contribute to this meta-analysis, but we also extracted data for the overall sensitivity for HIV infection, whether acute or established, reflecting how the test would be used in practice. The sensitivities of greater than 99% cited by manufacturers of HIV rapid tests, considerably higher than the 85.8% we observed in high-income countries, come from studies using earlier generation EIA comparators that would not have detected acute infections [14–16].
The finding that HIV rapid tests are less sensitive than current laboratory-based tests, particularly in high-income countries, suggests that services screening high incidence and regularly tested populations such as MSM should use more sensitive tests such as fourth-generation EIA or NAAT where possible. This is not possible or advisable in anonymous testing clinics or other settings in which the same-day result is necessary to link newly diagnosed patients to treatment services. It has been suggested that MSM would test more frequently if offered HIV rapid tests [48]; however, a randomized controlled trial comparing HIV rapid tests to laboratory-based testing found no difference in their frequency of testing over time [49]. Our data suggest that HIV rapid tests are acceptably sensitive in low-income countries where they are often the only form of HIV testing available, and acute infections could be addressed by ensuring those at increased risk are re-tested after a suitable period.
As health programs around the world strive to maximize the preventive impact of treatment, the frequency of testing will increase. The challenge for low-income countries will be to monitor for any resulting decline in HIV rapid test sensitivity. The challenge for high-income countries is to identify other ways of increasing the frequency of testing among high-risk groups, without relying solely on HIV rapid tests, and to combine HIV rapid tests with more sensitive laboratory assays, perhaps using dried blood spot samples or other novel methods [35]. Meanwhile, health workers in all settings need to be aware of the longer window period for these tests and counsel clients appropriately.
Acknowledgements
Financial support: The Australian National Health and Medical Research Council supports EPFC and TRHR with Early Career Fellowships (EPCF No. 1091226, TRHR No. 1091536).
Contributors: C.K.F. conceived the study. W.S.T. and T.R.H.R. did the literature search, reviewed the studies and co-drafted the first version of the report. W.S.T. and E.P.F.C. performed the meta-analysis. All authors interpreted the results, critically revised the report for scientific content, and approved the final version.
Conflicts of interest
There are no conflicts of interests.
References
1. Pottie K, Medu O, Welch V, Dahal GP, Tyndall M, Rader T, et al.
Effect of rapid HIV testing on HIV incidence and services in populations at high risk for HIV exposure: an equity-focused systematic review.
BMJ Open 2014; 4:e006859.
2. Choko AT, MacPherson P, Webb EL, Willey BA, Feasy H, Sambakunsi R, et al.
Uptake, accuracy, safety, and linkage into care over two years of promoting annual self-testing for HIV in Blantyre, Malawi: a community-based prospective study.
PLoS Med 2015; 12:e1001873.
3. Birger RB, Hallett TB, Sinha A, Grenfell BT, Hodder SL.
Modeling the impact of interventions along the HIV continuum of care in Newark, New Jersey.
Clin Infect Dis 2014; 58:274–284.
4. UNAIDS.
90-90-90: an ambitious treatment target to help end the AIDS epidemic. 2014; Geneva: United Nations Programme on HIV/AIDS,
www.unaids.org/sites/default/files/media_asset/90-90-90_en_0.pdf
5. Stekler JD, Swenson PD, Coombs RW, Dragavon J, Thomas KK, Brennan CA, et al.
HIV testing in a high-incidence population: is antibody testing alone good enough?.
Clin Infect Dis 2009; 49:444–453.
6. Chan D, Stewart M, Smith M, Price T, Lusk J, Ooi C, et al.
The rise of targeted HIV oral rapid testing in Australia.
Med J Aust 2015; 202:251–255.
7. Conway DP, Holt M, McNulty A, Couldwell DL, Smith DE, Davies SC, et al.
Multicentre evaluation of the Determine HIV Combo assay when used for point of care testing in a high risk clinic-based population.
PLoS ONE 2014; 9:e94062.
8. Cooley LA, Oster AM, Rose CE, Wejnert C, Le BC, Paz-Bailey G.
Increases in HIV testing among men who have sex with men – National HIV Behavioral Surveillance System, 20 U.S. Metropolitan Statistical Areas, 2008 and 2011.
PLoS ONE 2014; 9:e104162.
9. Katz DA, Dombrowski JC, Swanson F, Buskin SE, Golden MR, Stekler JD.
HIV intertest interval among MSM in King County, Washington.
Sex Transm Infect 2013; 89:32–37.
10. Prendergast AJ, Essajee S, Penazzato M.
HIV and the Millennium Development Goals.
Arch Dis Child 2015; 100 (suppl 1):S48–S52.
11. TACAIDS, ZAC, NBS, OCGS, and ICF International, Tanzania Commission for AIDS.
Tanzania HIV/AIDS and Malaria Indicator Survey 2011–2012. 2013;
https://dhsprogram.com/pubs/pdf/AIS11/AIS11.pdf
12. Centers for Disease Control and Prevention Rapid HIV tests suitable for use in clinical settings (CLIA-moderate complexity). Centers for Disease Control and Prevention.
http://www.cdc.gov/hiv/pdf/testing_rapidtests.pdf [Accessed 27 April 2016].
13. Centers for Disesase Control and Prevention. Rapid HIV tests suitable for use in nonclinical settings (requires CLIA waiver certificate). Centers for Disesase Control and Prevention.
www.cdc.gov/hiv/pdf/testing_nonclinical_clia-waived-tests.pdf [Accessed 27 April 2016].
14. Orasure. Oraquick Advance Rapid HIV-1/2 Antibody Test package insert.
http://www.fda.gov/downloads/BiologicsBloodVaccines/ucm091917.pdf [Accessed 27 April 2016].
17. Centers for Disease Control and Prevention. Advantages and disadvantages of different types of FDA-approved HIV immunoassays used for screening by generation and platform. Centers for Disease Control and Prevention.
http://www.cdc.gov/hiv/pdf/testing_Advantages&Disadvantages.pdf [Accessed 27 April 2016].
18. Centers for Disease Control and Prevention. Laboratory testing for the diagnosis of HIV infection – updated recommendations. Centers for Disease Control and Prevention.
http://www.cdc.gov/hiv/pdf/hivtestingalgorithmrecommendation-final.pdf [Accessed 27 April 2016].
19. Masciotra S, Luo W, Youngpairoj AS, Kennedy MS, Wells S, Ambrose K, et al.
Performance of the Alere Determine HIV-1/2Ag/Ab Combo rapid test with specimens from HIV-1 seroconverters from the US and HIV-2 infected individuals from Ivory Coast.
J Clin Virol 2013; 58 (suppl 1):e54–58.
20. Faraoni S, Rocchetti A, Gotta F, Ruggiero T, Orofino G, Bonora S, et al.
Evaluation of a rapid antigen and antibody combination test in acute HIV infection.
J Clin Virol 2013; 57:84–87.
21. Lewis JM, Macpherson P, Adams ER, Ochodo E, Sands A, Taegtmeyer M.
Field accuracy of fourth-generation rapid diagnostic tests for acute HIV-1: a systematic review.
AIDS 2015; 29:2465–2471.
22. Masciotra S, McDougal JS, Feldman J, Sprinkle P, Wesolowski L, Owen SM.
Evaluation of an alternative HIV diagnostic algorithm using specimens from seroconversion panels and persons with established HIV infections.
J Clin Virol 2011; 52 (suppl 1):S17–22.
23. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al.
The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration.
PLoS Med 2009; 6:e1000100.
24. The World Bank, The World Bank.
Country and lending groups. 2015;
http://data.worldbank.org/about/country-and-lending-groups
25. Song F, Khan KS, Dinnes J, Sutton AJ.
Asymmetric funnel plots and publication bias in meta-analyses of diagnostic accuracy.
Int J Epidemiol 2002; 31:88–95.
26. Sterne JA, Egger M.
Funnel plots for detecting bias in meta-analysis: guidelines on choice of axis.
J Clin Epidemiol 2001; 54:1046–1055.
27. Thornton A, Lee P.
Publication bias in meta-analysis: its causes and consequences.
J Clin Epidemiol 2000; 53:207–216.
28. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al.
QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies.
Ann Intern Med 2011; 155:529–536.
29. Charurat M, Nasidi A, Delaney K, Saidu A, Croxton T, Mondal P, et al.
Characterization of acute HIV-1 infection in high-risk Nigerian populations.
J Infect Dis 2012; 205:1239–1247.
30. Black V, von Mollendorf CE, Moyes JA, Scott LE, Puren A, Stevens WS.
Poor sensitivity of field rapid HIV testing: implications for mother-to-child transmission programme.
BJOG 2009; 116:1805–1808.
31. Duong YT, Mavengere Y, Patel H, Moore C, Manjengwa J, Sibandze D, et al.
Poor performance of the determine HIV-1/2Ag/Ab combo fourth-generation rapid test for detection of acute infections in a National Household Survey in Swaziland.
J Clin Microbiol 2014; 52:3743–3748.
32. Everett DB, Baisley K, Changalucha J, Vallely A, Watson-Jones D, Cook C, et al.
Suitability of simple human immunodeficiency virus rapid tests in clinical trials in community-based clinic settings.
J Clin Microbiol 2009; 47:1058–1062.
33. Fiscus SA, Pilcher CD, Miller WC, Powers KA, Hoffman IF, Price M, et al.
Rapid, real-time detection of acute HIV infection in patients in Africa.
J Infect Dis 2007; 195:416–424.
34. Jones CB, Kuldanek K, Muir D, Phekoo K, Black A, Sacks R, et al.
Clinical evaluation of the Determine HIV-1/2Ag/Ab Combo test.
J Infect Dis 2012; 206:1947–1949.author reply 1949–1950.
35. Kania D, Bekale AM, Nagot N, Mondain AM, Ottomani L, Meda N, et al.
Combining rapid diagnostic tests and dried blood spot assays for point-of-care testing of human immunodeficiency virus, hepatitis B and hepatitis C infections in Burkina Faso, West Africa.
Clin Microbiol Infect 2013; 19:E533–541.
36. Martin EG, Salaru G, Mohammed D, Coombs RW, Paul SM, Cadoff EM.
Finding those at risk: acute HIV infection in Newark, NJ.
J Clin Virol 2013; 58 (suppl 1):e24–28.
37. McGrath M, Cohen AGM, Klausner J.
Evaluating 4th generation testing for HIV-1 within a nonprofit setting.
Sex Transm Dis 2014; 41 (suppl 1):S31.
38. Patel P, Mackellar D, Simmons P, Uniyal A, Gallagher K, Bennett B, et al.
Detecting acute human immunodeficiency virus infection using 3 different screening immunoassays and nucleic acid amplification testing for human immunodeficiency virus RNA, 2006–2008.
Arch Intern Med 2010; 170:66–74.
39. Pilcher CD, Louie B, Facente S, Keating S, Hackett J Jr, Vallari A, et al.
Performance of rapid point-of-care and laboratory tests for acute and established HIV infection in San Francisco.
PLoS ONE 2013; 8:e80629.
40. Rosenberg NE, Kamanga G, Phiri S, Nsona D, Pettifor A, Rutstein SE, et al.
Detection of acute HIV infection: a field evaluation of the determine(R) HIV-1/2Ag/Ab combo test.
J Infect Dis 2012; 205:528–534.
41. Serna-Bolea C, Munoz J, Almeida JM, Nhacolo A, Letang E, Nhampossa T, et al.
High prevalence of symptomatic acute HIV infection in an outpatient ward in southern Mozambique: identification and follow-up.
AIDS 2010; 24:603–608.
42. Stekler JD, O’Neal JD, Lane A, Swanson F, Maenza J, Stevens CE, et al.
Relative accuracy of serum, whole blood, and oral fluid HIV tests among Seattle men who have sex with men.
J Clin Virol 2013; 58 (suppl 1):e119–e122.
43. Wolpaw BJ, Mathews C, Chopra M, Hardie D, de Azevedo V, Jennings K, et al.
The failure of routine rapid HIV testing: a case study of improving low sensitivity in the field.
BMC Health Serv Res 2010; 10:73.
44. Hollingsworth TD, Anderson RM, Fraser C.
HIV-1 transmission, by stage of infection.
J Infect Dis 2008; 198:687–693.
45. Smith MK, Rutstein SE, Powers KA, Fidler S, Miller WC, Eron JJ Jr, et al.
The detection and management of early HIV infection: a clinical and public health emergency.
J Acquir Immune Defic Syndr 2013; 63 (suppl 2):S187–S199.
46. Pant Pai N, Balram B, Shivkumar S, Martinez-Cajas JL, Claessens C, Lambert G, et al.
Head-to-head comparison of accuracy of a rapid point-of-care HIV test with oral versus whole-blood specimens: a systematic review and meta-analysis.
Lancet Infect Dis 2012; 12:373–380.
47. Yoo SJ, Sohn YH, Choi SE, Oh HB.
Meta-analysis for the pooled sensitivity and specificity of antihuman immunodeficiency virus Ab rapid tests.
Korean J Lab Med 2009; 29:345–352.
48. Chen MY, Bilardi JE, Lee D, Cummings R, Bush M, Fairley CK.
Australian men who have sex with men prefer rapid oral HIV testing over conventional blood testing for HIV.
Int J STD AIDS 2010; 21:428–430.
49. Read TR, Hocking JS, Bradshaw CS, Morrow A, Grulich AE, Fairley CK, et al.
Provision of rapid HIV tests within a health service and frequency of HIV testing among men who have sex with men: randomised controlled trial.
BMJ 2013; 347:f5086.
