Click on the links below to access all the ArticlePlus for this article.
Please note that ArticlePlus files may launch a viewer application outside of your web browser.
HIV prevention requires individual risk perception and knowledge of status, and detecting HIV infection is key to reducing HIV transmission.1,2 Individuals who tested positive reduced high-risk behavior compared with untested or HIV-negative individuals.3 Various studies have raised concern that many cases are detected late in the disease, however.4-7 Studies have also shown that up to 60% of those who tested for HIV did not receive results.8-11 The prolonged asymptomatic disease stage for HIV increases the transmission risk while individuals are infected but undiagnosed.
HIV rapid tests may overcome underdiagnosis by increasing diagnosis rates through providing immediate results with high sensitivity and specificity.12-17 Rapid tests significantly increased testing rates and results received, leading to early HIV detection.8,12 Their initial use was restricted to structured care delivery sites, but there is now interest in self-testing using rapid tests.18 Spielberg et al19 showed that 20% of at-risk individuals preferred self-testing over a variety of in-clinic testing options, and those who had never tested were significantly more likely to prefer self-testing.
Countries are considering implementing self-testing, but there are no existing studies on the effectiveness and feasibility of HIV self-testing. This study aims to bridge this gap by comparing user acceptability and feasibility of rapid tests in self-testing versus point-of-care rapid testing by trained personnel. It also explores key issues and concerns during implementation.
A cross-sectional study was conducted in 2 major HIV test centers in Singapore: the Communicable Disease Center (CDC) and the Department of Sexually Transmitted Infection Control (DSC) center, from July 2005 to January 2006. Study participants comprised 2 distinct groups: known HIV-positive cases at the CDC (Singapore's main HIV treatment center) and at-risk individuals with unknown HIV status seeking treatment for sexually transmitted infections at the DSC. Participants were selected by systematic sampling based on their sequential appointment time (1 per hour) and were recruited based on their willingness to participate. At-risk individuals had to agree to confirmatory laboratory tests before selection. Minors (<21 years old) and elderly persons (>65 years old) were excluded; the latter group was excluded to reduce confounding, because elderly individuals had different local educational and socioeconomic profiles. HIV-positive individuals were included because their demographic patterns differed from those of the general population and their opinions and self-test performance characteristics may be closer to those of unknown positives in the community.20,21
After providing informed consent, persons were asked to complete a survey on demographics, knowledge, attitudes, and practices regarding HIV and rapid testing. To assess the feasibility of self-testing, interrater agreement between the results of self-testing and testing by trained personnel was measured. Participants were asked to perform self-testing, and the test was repeated on participants by trained personnel. Results of both tests were read by trained personnel for accuracy and consistency. To determine participants' ability to interpret self-test results, participants were then provided with 4 sample rapid test results and asked to identify them based on the instructions provided. Participants were then surveyed on user acceptability and attitudes toward rapid testing.
The Abbott Determine HIV 1/2 rapid test (Abbott Laboratories, Abbott Park, IL), a blood-based rapid test approved for point-of-care testing in Singapore, was used. A study kit was created, which included a test strip, a lancet, a capillary tube, buffer solution, and wipes as well as a tailor-made, pretested, 14-step pictorial instruction sheet based on the manufacturer's recommended steps, with basic annotations in English and Mandarin. The instruction steps included test kit preparation, blood sampling via fingerprick and drawing of blood via a capillary tube, application of blood and buffer solution to the test strip, and waiting for results. Separate instructions were provided for test result interpretation, with 4 pictorial representations of the sample test results: a positive result (control and test lines present), a negative result (only control line present), an invalid result (test line present without control line), and another invalid result (no lines present). Participants were required to articulate the sample results based on the pictorial representations.
The questionnaires followed the language and framework of validated questionnaires in the United States and Singapore, and the instruction sheet and survey were pilot tested. This study received ethics review board approval and was conducted with each participant's informed consent, with pre- and posttest counseling provided by trained personnel. During counseling for known HIV-positive individuals, we emphasized that their HIV-positive status remained irrespective of study test result, and that retesting was unwarranted.
Sample size calculation was based on the level of agreement between results from self-testing and trained personnel testing, using the κ statistic. Based on a postulated κ value from 0.2 to 0.8 and a standard error from 0.01 to 0.02, a sample size of 90 would provide a significance level of 0.05. To increase the level of accuracy for the at-risk group, in which HIV-positive cases were less frequent, the sample size was increased 3-fold.
Participants' knowledge, attitudes, and practices and their ability to identify results were analyzed. Demographics were compared with those of the population.19,20 The χ2 and Fisher exact tests were used to compare categoric variables, and the nonparametric Mann-Whitney test was used to compare medians for continuous outcomes. Test kit sensitivity and specificity were computed from tests performed by trained personnel compared with confirmatory test results. In addition, logistic regression models explored factors associated with reading all sample results correctly and the equivalence of self-test results and trained personnel test results. All demographic variables were considered as covariates, and continuous variables were tested for linearity. Starting from the most significant variable in univariate analysis, the likelihood ratio test was used to determine if inclusion of a covariate significantly improved the model's fit.
Tests were performed with STATA Release 9.0 (Stata Corporation, College Station, TX) at the 5% level of significance.
A total of 420 individuals, 141 from the CDC and 279 from the DSC, were selected for this study. Of these, 88 from the CDC (known HIV-positive individuals) and 262 from the DSC (at-risk participants) agreed to participate (response rates of 62% and 94%, respectively). Nonparticipation at the CDC was attributable to individuals' lack of time, whereas CDC participants constituted 7% of known HIV cases in Singapore.
There was no significant difference between the demographics of known HIV-positive participants compared with the general HIV population.19 Education and income levels among at-risk participants were higher than those of known HIV-positive participants (P < 0.01; Table 1). Adjusted median income levels of at-risk participants were similar to those of Singapore's population. More known HIV-positive participants were men who have sex with men (MSM; 18%) compared with at-risk participants (7%; P < 0.01). Approximately half of the at-risk participants never had an HIV test; of these, 51% did not perceive themselves at risk for infection and 18% stated inconvenience and waiting times as deterrents.
From Table 2, 54% of at-risk participants and 15% of known HIV-positive participants had heard of rapid tests. Almost all participants preferred to obtain results in <1 hour. Although 89% preferred to take the test in private (by themselves), 87% indicated that confidential pretest counseling by trained counselors was necessary. Ninety-four percent of known-HIV positive participants and 89% of at-risk participants thought that confirmatory testing was necessary. After self-testing, 88% of participants found the kit easy to use and 91% found the instructions easy to understand. Nevertheless, 67% indicated blood sampling and transfer as most difficult. From trained personnel observations, 61% of known HIV-positive participants and 92% of at-risk participants failed to perform all steps correctly (9% and 35%, respectively, failed to prepare the test kit correctly, and 53% and 79%, respectively, failed to perform blood sampling and transfer correctly). Twenty percent of known HIV-positive participants and 68% of at-risk participants had invalid self-test results (no lines were present in all but 1 test) compared with only 1% when tests were performed by trained personnel. Although 79% thought that posttest counseling was necessary, 88% thought that the kit should be sold in public outlets, with most willing to pay between US $7 and US $13.
The κ value for interrater agreement between results of tests performed by participants and trained personnel was 0.28 (P < 0.01; Table 3). The test kit sensitivity was 98.9% when the test was performed by trained personnel, and the specificity was 99.6%. The 1 false-negative rapid test result occurred in a known HIV-positive individual on antiretroviral therapy.
For the other main outcome, 12% of all participants (10% of at-risk participants and 19% of known HIV-positive participants) could not correctly identify all sample results (Table 4). Two percent and 7% of all participants could not correctly identify the “positive” and “negative” samples, respectively.
From the univariate analysis, age, educational level, and study site were significantly associated with participants' ability to interpret all sample results correctly. Age, marital status, educational level, study site, and previous HIV testing were significantly associated with test result agreement between participants and trained personnel. From the multivariate model, only age (odds ratio [OR] = 0.92, 95% confidence interval [CI]: 0.89 to 0.95 for each 1-year increase in age) was independently and significantly associated with correctly interpreting all sample results. Educational level (OR = 2.62, 95% CI: 0.97 to 7.12 for tertiary-level education and OR = 3.21, 95% CI: 1.13 to 9.08 for secondary-level education compared with primary education) and study site (OR = 15.74, 95% CI: 6.87 to 36.02 comparing CDC with DSC participants) were independently and significantly associated with test result agreement between self-testing and trained personnel testing.
HIV rapid tests have increased HIV testing rates and communication of results,8,18,22 and research is needed to determine if self-testing increases testing rates and assists in prevention.12 Current use of rapid tests in Singapore is limited to anonymous trained personnel testing. Eighteen percent of participants who never tested cited inconvenience and long waiting times as deterrents. Self-testing offers a possible solution, because many participants preferred testing in private and receiving results promptly.
During the test, most participants found the kit easy to use, but 85% failed to perform all steps correctly, resulting in 56% having invalid results. This suggests that the Abbott Determine HIV 1/2 test, involving blood sampling via fingerprick and collection via a capillary tube, was difficult for participants. Although at-risk participants were younger and had higher education levels, their ability to perform the test correctly was lower than that of known HIV-positive participants, as shown by test result agreement on multivariate analysis. This may reflect the latter's exposure to, and experience with, blood tests. Multivariate analysis also showed that younger people had better ability in interpreting results and that those with higher education levels had better test result agreement with trained personnel. In addition, 1% of trained personnel tests were invalid, which may be attributable to inaccurate blood or buffer amounts used, performance error, or defective kits. Future test designs should enable easy use across all ages and educational levels.
For test kit effectiveness, we showed that the sensitivity and specificity were similar to the manufacturer's claims. Interrater agreement between participant self-test results and trained personnel test results was poor, however, as evidenced by the many invalid results attributable to difficulty in using this test kit, which was not specifically designed for self-testing.
Assessing participants' ability to read test results, 12% could not correctly identify sample results, including 2% and 7% who could not interpret positive and negative samples, respectively, correctly. The test kit results format is commonly seen in other kits, however. This implies that some individuals may inaccurately interpret their own results, even if the test is performed correctly. False-positive interpretation can cause psychologic stress, and false-negative interpretation may result in a false sense of security with continued high-risk behavior.23
In addition, most participants thought that pre- and posttest confidential counseling was necessary, which poses challenges for self-testing. Previously proposed pretest counseling solutions include requisite telephone-in, computer-based, or in-person counseling before purchase,12 but initiating posttest counseling remains the end user's responsibility. Other issues with self-testing include providing confirmatory testing (10% did not think confirmatory testing was necessary) and referrals for individuals with positive test results. Nevertheless, most participants still thought that self-test kits should be publicly sold and would recommend the kit to others.
The key study limitation was the use of a blood-based kit not specifically designed for self-testing. As such, test performance results may not be generalizable to those obtained with other kits. Further studies should compare the effectiveness of newer oral self-test kits.18,24,25 Health care workers were not blinded to patient participation sites, which may introduce reporting bias, but this was minimized by providing clear instructions and training. Other limitations include selection biases, because only 2 centers were chosen, including known HIV-positive participants who may be influenced by HIV care services. Participants' demographics were not significantly different from those of the local population, however, and future screening programs would target similar at-risk individuals. Results may not be extrapolated to risk groups underrepresented in this study, such as intravenous drugs users, who constitute only 2% of HIV cases locally.4 Future studies should investigate the performance and interpretation ability of underrepresented groups and the general population.
Implementation of self-testing should be reconsidered until kit design and downstream issues have been adequately addressed. Key considerations before implementation should include improving and validating the test kit effectiveness, including test performance and interpretation of results, and confirmatory tests, referrals, and counseling. Until these issues are resolved, point-of-care rapid testing by trained personnel may provide more accurate test performance, interpretation of results, counseling, and education.
The authors thank Mark Chen for his invaluable advice, Lalitha Nair and Phyllis Chew for their contributions to testing and collecting data from participants, and staff at the participating sites for their support.
1. Centers for Disease Control and Prevention (CDC). Adoption of protective behaviors among persons with recent HIV infection and diagnosis-Alabama, New Jersey, and Tennessee, 1997-1998. MMWR Morb Mortal Wkly Rep
2. Campbell CH Jr, Marum ME, Alwano-Edyegu M, et al. The role of HIV counseling and testing in the developing world. AIDS Educ Prev
3. Weinhardt LS, Carey MP, Johnson BT, et al. Effects of HIV counseling and testing on sexual risk behavior: a meta-analytic review of published research, 1985-1997. Am J Public Health
4. Cutter JL, Lin WY, Ang LW, et al. HIV in Singapore-past, present, and future. AIDS Educ Prev
5. Brannstrom J, Akerlund B, Arneborn M, et al. Patients unaware of their HIV infection until AIDS diagnosis in Sweden 1996-2002-a remaining problem in the highly active antiretroviral therapy era. Int J STD AIDS
6. Wortley PM, Chu SY, Diaz T, et al. HIV testing patterns: where, why, and when were persons with AIDS tested for HIV? AIDS
7. Bellamy R, Sangeetha S, Paton NI. AIDS-defining illnesses among patients with HIV in Singapore, 1985 to 2001: results from the Singapore HIV Observational Cohort Study (SHOCS). BMC Infect Dis
8. Hutchinson AB, Branson BM, Kim A, et al. A meta-analysis of the effectiveness of alternative HIV counseling and testing methods to increase knowledge of HIV status. AIDS
9. Centers for Disease Control and Prevention. Update: HIV counseling and testing using rapid tests-United States, 1995. MMWR Morb Mortal Wkly Rep
10. Tao G, Branson BM, Kassler WJ, et al. Rates of receiving HIV test results: data from the US National Health Interview Survey for 1994 and 1995. J Acquir Immune Defic Syndr
11. Sullivan PS, Lansky A, Drake A, for the HITS-2000 Investigators. Failure to return for HIV test results among persons at high risk for HIV infection. Results from a multistate interview project. J Acquir Immune Defic Syndr
12. Spielberg F, Leving RO, Weave M. Self-testing for HIV: a new option for HIV prevention? Lancet Infect Dis
13. Keenan PA, Keenan JM, Branson BM. Rapid HIV testing. Wait time reduced from days to minutes. Postgrad Med
14. Were W, Mermin J, Bunnell R, et al. Home-based model for HIV voluntary counselling and testing. Lancet
15. Kassler WJ, Alwano-Edyegu MG, Marum E, et al. Rapid HIV testing with same-day results: a field trial in Uganda. Int J STD AIDS
16. Spielberg F, Kassler WJ. Rapid testing for HIV antibody: a technology whose time has come. Ann Intern Med
17. Menard D, Mairo A, Mandeng MJ, et al. Evaluation of rapid HIV testing strategies in under equipped laboratories in the Central African Republic. J Virol Methods
18. Wright AA, Katz IT. Home testing for HIV. N Engl J Med
19. Spielberg F, Branson BM, Goldbaum GM, et al. Overcoming barriers to HIV testing: preferences for new strategies among clients of a needle exchange, a sexually transmitted disease clinic, and sex venues for men who have sex with men. J Acquir Immune Defic Syndr
20. Ministry of Health, Singapore. Statistics HIV. 2005. Available at: http://www.moh.gov.sg/corp/publications/statistics/hiv.do
. Accessed April 30, 2006.
21. Singapore Department of Statistics. Key statistics. Available at: http://www.singstat.gov.sg/
. Accessed April 15, 2006.
22. Kassler WJ, Dillon BA, Haley C, et al. On-site, rapid HIV testing with same-day results and counseling. AIDS
23. Ryder K, Haubrich DJ, Calla D, et al. Psychosocial impact of repeat HIV-negative testing: a follow-up study. AIDS Behav
24. Reynolds SJ, Muwonga J. OraQuick ADVANCE Rapid HIV-1/2 antibody test. Expert Rev Mol Diagn
25. Liang TS, Erbelding E, Jacob CA, et al. Rapid HIV testing of clients of a mobile STD/HIV clinic. AIDS Patient Care STDS