Introduction
HIV counselling and testing (HCT) is critically important, both as a point of entry into HIV care and as an HIV prevention strategy. Mathematical modelling studies suggest that undiagnosed HIV-positive individuals contribute a disproportionately large fraction of HIV transmission [1] and that the impact of antiretroviral treatment (ART) on HIV incidence is likely to be strongly determined by the frequency of HIV testing [2–4]. Recognizing the importance of HCT in both HIV prevention and treatment, UNAIDS has proposed the target of diagnosing 90% of the HIV-positive population by 2020 [5]. It is therefore necessary to monitor the uptake of HCT and to identify the subpopulations that are being missed by HCT services.
However, monitoring the rate of HIV diagnosis in HIV-positive individuals presents a number of practical challenges. National surveys, the most frequent source of HCT uptake estimates in developing countries, usually only report on whether individuals have been tested for HIV, and not whether they have been diagnosed HIV-positive [5,6]. In addition, self-reported HIV testing data from surveys may be inaccurate, given the evidence of over-reporting of health screening from many settings [7–9]. Estimates of numbers of HIV-positive diagnoses, frequently used in high-income countries, may be subject to double-counting due to retesting of previously diagnosed individuals and challenges in linking individual records across systems [10]. As a result of these obstacles, there have been relatively few attempts to estimate the fraction of the HIV-positive population that is undiagnosed, particularly in the developing countries most severely affected by HIV/AIDS.
South Africa is a country with a high HIV prevalence, in which access to HIV testing has historically been limited. However, in recent years, access to HCT has improved dramatically, with 13.3 million South Africans being tested through public health services between April 2010 and June 2011 [11]. By integrating HCT data from the public and private health sectors, together with household survey estimates of the proportion of the population tested for HIV, this study aims to assess South Africa's progress towards increasing HCT uptake and to identify biases associated with different data sources. In addition, this study aims to assess how the undiagnosed fraction of the HIV-positive population is likely to change in future, and whether the UNAIDS target of a 90% diagnosis rate is achievable.
Materials and methods
Uptake of HCT was modelled using Thembisa, a deterministic mathematical model of the South African HIV epidemic. Detailed descriptions of the model have been published previously [12,13]. Briefly, the South African population is stratified by age, sex, marital status, level of sexual risk behaviour, male circumcision status and (in the HIV-positive population) CD4+ cell count and receipt of ART. Changes in the numbers of adults in each stratum are calculated over time, starting in 1985, based on assumptions regarding rates of demographic change and HIV epidemic dynamics. For the purpose of this analysis, assumptions regarding sexual behaviour, HIV transmission and HIV mortality are fixed at the posterior means estimated previously when fitting the model to South African HIV prevalence data and mortality data [12].
The population aged 10 years and older is further divided into three HIV-testing history groups (never tested, previously tested negative and previously tested positive). Three types of HIV testing are modelled: testing in antenatal clinics, testing of HIV patients with opportunistic infections and testing for other reasons. The annual rate at which sexually experienced individuals get tested is assumed to depend on their HIV stage (s), age (x), sex (g), HIV testing history (i) and the calendar year (t):
;)
where b(t) is the base rate of HIV testing in year t, in individuals who do not have any HIV symptoms and are not pregnant; Ag(x,t) is an adjustment factor to represent the effect of age and sex on the base rate of test uptake; ri is an adjustment factor to represent the effect of testing history; Ωs is the annual incidence of opportunistic infections in CD4+ stage s; di(t) is the fraction of patients with opportunistic infection who are tested for HIV in year t; Fg,s(x,t) is the fertility rate in sexually experienced women aged x, in HIV stage s, during year t (set to zero for men); and vi(t) is the proportion of pregnant women who receive HIV testing in year t. The function used to represent the effect of age and sex on the uptake of HIV testing is
where Bg(t) is a time-dependent sex adjustment factor, and αg and λg are coefficients for the effect of age on the rate of HIV test uptake. Bg(t) is set to 1 for women (g = 1), while for men, the ratio is allowed to change over time. This time dependency is modelled by specifying a constant ratio up to 2002 [B0(2002)], a constant ratio after 2010 [B0(2010)] and a linear change in the ratio between 2002 and 2010. The assumed values of the model parameters are summarized in Table 1 and in Table S4 of the supplementary material for the time-varying parameters. In the case of parameters that cannot be quantified precisely, prior distributions are specified to represent ranges of uncertainty.
Table 1: Parameter values and data sources.
The b(t) parameter values are estimated in each year from reported numbers of HIV tests performed in South Africa. These reported numbers include data from public health facilities (from 2002 to 2012), the life insurance industry (from 2002 to 2011), medical schemes (2011) and other private providers (2011). A more detailed explanation of the method used to estimate the total numbers of HIV tests and the method to derive b(t) from these numbers is provided in the supplementary material, https://links.lww.com/QAD/A702. Due to lack of data from the period prior to 2002, the annual numbers of HIV tests were assumed to increase linearly from zero in 1990 to the estimated 2002 total; in a sensitivity analysis, we assess the effect of alternative nonlinear growth assumptions.
A likelihood function was defined to represent the degree of model consistency with two further data sources: the HIV prevalence in individuals tested for HIV in 6 years (2004–2008 and 2010) and the proportion of adults who reported having ever been tested for HIV in three nationally represent household surveys (in 2005, 2008 and 2012) [26,27]. The latter set of survey estimates was stratified by age, sex and HIV status. In defining the likelihood function, allowance was made for mis-reporting of prior HIV testing, with the parameter θh representing the ratio of the odds of reporting prior testing to the actual odds of prior testing, in individuals of HIV status h. Further detail regarding the specification of the likelihood function and the data sets is provided in the supplementary material, https://links.lww.com/QAD/A702.
A Bayesian approach was used to estimate the posterior distribution of model estimates most consistent with the likelihood function after incorporating the prior uncertainty ranges in Table 1. A posterior sample of 1000 parameter combinations was generated using Incremental Mixture Importance Sampling [28], and means and 95% confidence intervals (95% CIs) were calculated from this sample.
The model was used to estimate the change in HCT uptake up to the middle of 2012. In addition, the model was used to project the likely change in the fraction of HIV-positive adults who are undiagnosed over the 2012–2020 period, assuming that the annual number of HIV tests performed in South Africa would remain constant at 10 million per annum, in line with Department of Health targets [29].
Results
Model estimates of the fraction of the South African population ever tested for HIV were reasonably consistent with the survey data, particularly in 2012, although the model tended to underestimate the 2008 proportions (Fig. 1). Proportions reporting prior testing increased substantially between 2005 and 2012, especially in women. Consistent with the survey data, the model estimated a lower rate of prior testing in older HIV-positive women than younger HIV-positive women. The model estimates of numbers of HIV tests and HIV prevalence in adults tested for HIV were also consistent with the empirical data (Figure S2). Posterior estimates of the model parameters were mostly similar to the prior means (Table S6), although after excluding antenatal and opportunistic infection patient testing, the model estimate of the male-to-female testing ratio in 2010 (0.68, 95% CI 0.53–0.84) was lower than that in 2002 (0.83, 95% CI 0.70–0.99). The posterior estimate of the relative rate of testing in previously-diagnosed individuals (0.92, 95% CI: 0.77–1.00) was also different from the prior mean (0.5). Of those testing positive between 2011 and 2012, an estimated that 47% (95% CI 44–50%) were diagnosed positive for the first time.
Fig. 1: Proportions of adults who report having ever been tested for HIV.Model estimates have been adjusted to reflect expected reporting bias. Dashed lines represent 95% confidence intervals around average model estimates, while vertical lines represent 95% confidence intervals around survey estimates.
Bias in reporting of prior HIV testing
The odds ratio relating reported prior testing to actual prior testing was estimated to be 1.51 (95% CI 1.36–1.68) in HIV-negative adults and 0.91 (95% CI 0.83–0.97) in HIV-positive adults. As a result, the model estimate of the true fraction of adults ever tested in 2012 (58.0%, 95% CI 55.5–60.7%) was lower than the fraction who reported having ever tested (65.5%). A more substantial difference was found when comparing the model estimate of the fraction of adults tested between 2011 and 2012 (21.8%, 95% CI 21.4–22.1%) and the proportion who reported having tested for HIV in the last 12 months in 2012 (43.4%).
Proportion of HIV-positive adults who are undiagnosed
The model estimated that there were 5.7 million HIV-positive adults aged 15 years and older in South Africa in the middle of 2012. Of these, 23.7% (95% CI 23.1–24.3) were undiagnosed, with the proportion being substantially higher in men (31.9%, 95% CI 29.7–34.3) than in women (19.0%, 95% CI 17.9–19.9%). In young HIV-positive adults, the fraction undiagnosed was particularly high because HIV was in most cases recently acquired, although in adolescent boys, a significant proportion of infections were diagnosed because they had been vertically acquired (Fig. 2). The fraction of HIV-positive women who were undiagnosed was also particularly high at the older ages.
Fig. 2: Proportions of HIV-positive adults diagnosed and treated, by age and sex (2012).
HIV diagnosis relative to CD4+ progression
Of those HIV-positive adults who were undiagnosed in the middle of 2012, an estimated 31.5% were eligible to receive ART according to the guidelines in place at the time (CD4+ cell count <350 cells/μl) and 51.4% were eligible to receive ART according to recently revised guidelines (CD4+ cell count <500 cells/μl). Of ART-naïve, HIV-positive adults with CD4+ cell counts less than 350 cells/μl, an estimated 33.6% were undiagnosed, and of the ART-naive, HIV-positive adults with CD4+ cell counts of 350–499 cells/μl, the fraction undiagnosed was 32.5%. The probability that a newly infected adult progresses to a CD4+ cell count less than 350 cells/μl without diagnosis was calculated assuming that the rates of HIV testing estimated in 2010–2011 continued to apply in future years. These probabilities were estimated to be significantly higher in adults infected at older ages and in individuals who had never tested for HIV previously, with probabilities above 50% for most adults who acquire HIV after the age of 40 (Fig. 3). At young ages, probabilities of undiagnosed CD4+ progression were also significantly higher in men than in women.
Fig. 3: Probability that a newly infected individual progresses to CD4+ cell count <350 cells/μl without being diagnosed HIV-positive (2010–2011 HIV-testing rates).
Projected changes in undiagnosed fraction
Figure 4a shows that although the number of undiagnosed HIV-positive adults in South Africa is estimated to have declined substantially over the last decade, the number remains substantial (664 000 men and 679 000 women in 2012). If the Department of Health targets of 10 million HIV tests per annum are met, the undiagnosed numbers are expected to decline to 249 000 men and 286 000 women by 2020. This implies an undiagnosed fraction of 8.9%, with the 10% UNAIDS target being met in 2018 (Fig. 4b).
Fig. 4: Projected changes in undiagnosed HIV-positive adults.Projections are calculated assuming that the number of HIV tests performed is 10 million in each year after 2012. The dashed line in (b) represents the UNAIDS target (90% of HIV-positive adults diagnosed, equivalent to 10% undiagnosed).
Discussion
South Africa has made substantial progress in increasing access to HCT, with the estimated fraction of HIV-positive adults who are undiagnosed reducing from more than 80% in the early 2000s to 24% by the middle of 2012. However, the uptake of HCT and the probability of being diagnosed before the CD4+ cell count declines to 350 cells/μl are relatively low in men and older adults. This is clinically important because individuals who remain undiagnosed in late disease are at a high mortality risk, and if they only start ART at an advanced stage of disease, their prognosis is likely to be relatively poor [30]. The low rate of testing in older adults may reflect a perception, among both patients and healthcare providers, that older adults are not at risk of HIV [31,32]. The low rate of testing in men, and its recent decline relative to female testing rates, is more difficult to explain. Although antenatal HIV testing partially explains the higher rate of HIV testing in young women than young men, other South African studies have confirmed our finding that the sex differential remains even after excluding antenatal testing [14,33]. This could be due to gender norms, which often idealize male resilience [34]. Removing barriers to male HIV testing will be important not only for men themselves but also for ART as a female HIV prevention strategy.
Our results suggest that there may be bias in the reporting of prior HIV testing in surveys. HIV-negative adults appear to overreport prior HIV testing, consistent with studies that have examined the accuracy of self-reported screening for other health conditions [7–9]. These studies suggest that overreporting is most substantial in socioeconomically disadvantaged groups [9] and in face-to-face interviews [7]. However, our results suggest minimal bias in the reporting of prior HIV testing in HIV-positive adults. We hypothesize that for adults who perceive themselves to be at risk of HIV infection, any social desirability bias in favour of health-seeking behaviour may be offset by a fear of being marked as ‘likely to be infected’ [34,35]. Reporting bias may also differ depending on whether respondents are asked about having ever tested or having tested in the last 12 months, with ‘telescoping bias’ leading to the recency of previous testing being exaggerated [36]. The latter is particularly apparent in the 43.4% of adults who in 2012 reported having tested for HIV in the last 12 months [26]; if this were true, it would imply at least 15 million HIV tests performed, well in excess of the 10 million HIV tests reported by the public and private sectors combined. These findings have important implications for the monitoring of HCT uptake in developing countries, which relies heavily on self-reported data from surveys.
We estimate that 76% of HIV-positive adults in 2012 had been diagnosed previously, which is consistent with rates of 74–76% in three recent studies in rural South Africa, in which HIV-positive individuals received their test results [37–39]. In contrast, the national household survey of 2012, which did not provide test results to respondents, found that only 49.8% of HIV-positive individuals reported being aware of their HIV status [26]. However, of those HIV-positive adults who reported being unaware of their HIV status in this survey, 26.3% had antiretroviral drugs detectable in their blood specimens (Sean Jooste, personal communication), which suggests substantial underreporting of prior diagnosis, consistent with evidence from high-income settings [40,41]. HIV-diagnosed individuals may fear disclosing their HIV status to researchers, and hence may deny prior testing and diagnosis, particularly when they know that their test results will not be returned to them.
In the South African public sector, the HIV prevalence among individuals tested for HIV between 2010 and 2011 was 16.2% [42]. This may seem surprisingly high if it is assumed that most HIV-diagnosed adults do not get retested. Our model simulations suggest that the high HIV prevalence can best be explained by a high rate of retesting in individuals who have previously been diagnosed positive, and our estimate that only 47% of individuals testing positive in 2011–2012 were newly diagnosed is consistent with estimates of 40–79% in home-based HCT programmes in other African countries [43]. Retesting in previously diagnosed individuals may be due to HIV-diagnosed individuals doubting the accuracy of their test results [21]. Alternatively, HIV-diagnosed individuals who move between health services in different locations without being officially referred may need to get retested in order to obtain a referral letter [44]. It is also possible that previously diagnosed individuals may find it difficult to refuse the offer of HIV testing when it is heavily routinized in certain health settings. Whatever the explanation, it should not be assumed that all individuals who test positive are being diagnosed HIV-positive for the first time. This is particularly important to consider when applying mathematical models to reported numbers of ‘new’ HIV diagnoses [10].
This article proposes an innovative approach to estimating the relative sizes of the diagnosed and undiagnosed HIV epidemics, integrating survey data and routine data from health services within a mathematical modelling framework. Other methods, developed mostly for high-income countries [10], typically require other information (such as numbers of AIDS cases, AIDS deaths and CD4+ cell counts at the time of diagnosis) that are usually not available or reliable in most resource-limited settings. The Bayesian modelling approach has the advantage of weighting appropriately prior knowledge about patterns of HCT uptake and data from different sources, avoiding excessive reliance on a single data set [45]. To our knowledge, this is the first study to apply such a multiparameter synthesis approach to estimating rates of HIV testing and numbers of undiagnosed infections in a developing country.
A limitation of this analysis is that there is uncertainty regarding total numbers of HIV tests performed, especially in the period prior to 2002. However, relatively few of the HIV-positive individuals who were undiagnosed in 2002 would have survived to 2012 in the absence of diagnosis, and the 2012 estimates are therefore relatively robust. In a sensitivity analysis, model estimates were found to be similar when allowing for variation in numbers of HIV tests performed prior to 2002. In a further sensitivity analysis, in which total numbers of HIV tests performed in each year were assumed to be 20% greater than assumed in the baseline model, the model provided a significantly poorer fit to the data, which suggests that an undercount as large as 20% is very unlikely (see supplementary material, https://links.lww.com/QAD/A702). Another limitation of this analysis is that it considers only HIV testing in individuals aged 10 years and older. The model allows for HIV screening in infants born to HIV-positive mothers, but does not explicitly consider HIV testing in older children, though there is a clear need for greater HIV testing in this age group [46].
The proportion of HIV-positive South Africans who are undiagnosed is similar to the rates of 35% and lower in high-income countries [47–50], and it is encouraging that South Africa is on track to meet the UNAIDS target of a less than 10% fraction undiagnosed by 2020 if it continues to test 10 million individuals per annum. However, sustaining this high rate of testing may prove challenging, particularly if testing fatigue sets in. Alternative models may need to be considered to increase HCT uptake further. For example, home-based HCT has been shown to be highly acceptable [37,43,51,52] and may be particularly important in reaching older adults and populations with poorer access to health services. Opportunities also exist to improve provider-initiated testing in South African health services [18,53,54]. Mobile HCT services [51,55–58], incentivized testing [59,60] and self-testing [61] should also be explored. Further cost-effectiveness analysis is required to determine the optimal frequency and mode of HIV screening in the South African setting, balancing the costs of screening against the benefits of earlier treatment and reduced HIV transmission.
Acknowledgements
We are grateful to Peter Barron, Andrew Boulle and Rob Dorrington for providing useful comments on earlier drafts. We also thank the Department of Health and Swiss Re for providing data on HIV testing in the public sector and insurance industry, respectively. This research was funded by the Hasso Plattner Foundation and the National Institutes of Health (grant 1R01AI094586-01).
L.J. developed the mathematical model, as part of a larger project directed by L.B. The study was conceived by L.J., with input from L.B. and T.R. T.R. and S.J. provided household survey data and assisted in its interpretation. L.J. wrote the first draft of the article, and all authors reviewed the draft and assisted in editing.
Conflicts of interest
There are no conflicts of interest.
References
1. Hall HI, Holtgrave DR, Maulsby C.
HIV transmission rates from persons living with HIV who are aware and unaware of their infection.
AIDS 2012; 26:893–896.
2. Granich RM, Gilks CF, Dye C, De Cock KM, Williams BG.
Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model.
Lancet 2009; 373:48–57.
3. Dodd PJ, Garnett GP, Hallett TB.
Examining the promise of HIV elimination by ’test and treat’ in hyperendemic settings.
AIDS 2010; 24:729–735.
4. Long EF, Brandeau ML, Owens DK.
The cost-effectiveness and population outcomes of expanded HIV screening and antiretroviral treatment in the United States.
Ann Intern Med 2010; 153:778–789.
5. UNAIDS. Ambitious treatment targets: writing the final chapter of the AIDS epidemic. Geneva; 2014.
http://www.unaids.org/en/media/unaids/contentassets/documents/unaidspublication/2014/JC2670_UNAIDS_Treatment_Targets_en.pdf [Accessed 19 August 2014].
6. Cremin I, Cauchemez S, Garnett GP, Gregson S.
Patterns of uptake of HIV testing in sub-Saharan Africa in the pretreatment era.
Trop Med Int Health 2012; 17:e26–e37.
7. Rauscher GH, Johnson TP, Cho YI, Walk JA.
Accuracy of self-reported cancer-screening histories: a meta-analysis.
Cancer Epidemiol Biomarkers Prevent 2008; 17:748–757.
8. Klein JD, Graff CA, Santelli JS, Hedberg VA, Allan MJ, Elster AB.
Developing quality measures for adolescent care: validity of adolescents’ self-reported receipt of preventive services.
Health Serv Res 1999; 34:391–404.
9. Burgess DJ, Powell AA, Griffin JM, Partin MR.
Race and the validity of self-reported cancer screening behaviors: development of a conceptual model.
Prevent Med 2009; 48:99–107.
10. Working Group on Estimation of HIV Prevalence in Europe
HIV in hiding: methods and data requirements for the estimation of the number of people living with undiagnosed HIV.
AIDS 2011; 25:1017–1023.
11. Pillay Y, White C, McCornick N.
How times have changed - HIV and AIDS in South Africa in 2011.
South Afr Med J 2012; 102:77–78.
12. Johnson L. THEMBISA version 1.0: a model for evaluating the impact of HIV/AIDS in
South Africa. Centre for Infectious Disease Epidemiology and Research, University of Cape Town; 2014.
http://www.publichealth.uct.ac.za/publication-reports-0 [Accessed 1 December 2014].
13. Bekker LG, Johnson L, Cowan F, Overs C, Besada D, Hillier S, et al.
Combination HIV prevention for female sex workers: what is the evidence?.
Lancet 2015; 385:72–87.
14. Snow RC, Madalane M, Poulsen M.
Are men testing? Sex differentials in HIV testing in Mpumalanga Province, South Africa.
AIDS Care 2010; 22:1060–1065.
15. Actuarial Society of
South Africa. ASSA2008 AIDS and Demographic Model. 2011.
http://aids.actuarialsociety.org.za [Accessed 5 April 2011].
16. Kincaid DL, Parker W, Johnson S, Schierhout G, Kelly K, Connolly C, et al. AIDS Communication Programmes, HIV Prevention, and Living with HIV and AIDS in
South Africa, 2006. Pretoria: Johns Hopkins Health and Education in
South Africa; 2008.
http://jhhesa.org/PDF/National%20Comm%20Report.pdf [Accessed 8 March 2011].
17. Pettifor A, MacPhail C, Suchindran S, Delany-Moretlwe S.
Factors associated with HIV testing among public sector clinic attendees in Johannesburg, South Africa.
AIDS Behav 2010; 14:913–921.
18. Dalal S, Lee CW, Farirai T, Schilsky A, Goldman T, Moore J, et al.
Provider-initiated HIV testing and counseling: increased uptake in two public community health centers in South Africa and implications for scale-up.
PLoS One 2011; 6:e27293.
19. Luseno WK, Wechsberg WM.
Correlates of HIV testing among South African women with high sexual and substance-use risk behaviours.
AIDS Care 2009; 21:178–184.
20. Ng’ang’a A, Waruiru W, Ngare C, Ssempijja V, Gachuki T, Njoroge I, et al.
The status of HIV testing and counseling in Kenya: results from a nationally representative population-based survey.
J Acquir Immune Defic Syndr 2014; 66 (Suppl 1):S27–S36.
21. Shanks L, Klarkowski D, O’Brien DP.
False positive HIV diagnoses in resource limited settings: operational lessons learned for HIV programmes.
PLoS One 2013; 8:e59906.
22. Reniers G, Eaton J.
Refusal bias in HIV prevalence estimates from nationally representative seroprevalence surveys.
AIDS 2009; 23:621–629.
23. Floyd S, Molesworth A, Dube A, Crampin AC, Houben R, Chihana M, et al.
Underestimation of HIV prevalence in surveys when some people already know their status, and ways to reduce the bias.
AIDS 2013; 27:233–242.
24. Holmes CB, Wood R, Badri M, Zilber S, Wang B, Maartens G, et al.
CD4 decline and incidence of opportunistic infections in Cape Town, South Africa: implications for prophylaxis and treatment.
J Acquir Immune Defic Syndr 2006; 42:464–469.
25. Anglaret X, Minga A, Gabillard D, Ouassa T, Messou E, Morris B, et al.
AIDS and non-AIDS morbidity and mortality across the spectrum of CD4 cell counts in HIV-infected adults before starting antiretroviral therapy in Côte d’Ivoire.
Clin Infect Dis 2012; 54:714–723.
26. Shisana O, Rehle T, Simbayi LC, Zuma K, Jooste S, Zungu N, et al. South African national HIV prevalence, incidence, and behaviour survey, 2012. Cape Town: Human Sciences Research Council; 2014.
http://www.hsrc.ac.za/en/research-outputs/view/6871 [Accessed 16 April 2014].
27. Peltzer K, Matseke G, Mzolo T, Majaja M.
Determinants of knowledge of HIV status in South Africa: results from a population-based HIV survey.
BMC Public Health 2009; 9:174.
28. Raftery AE, Bao L.
Estimating and projecting trends in HIV/AIDS generalized epidemics using Incremental Mixture Importance Sampling.
Biometrics 2010; 66:1162–1173.
29. Department of Health. Annual performance plan: 2014/15-2016/17. 2014.
http://www.health.gov.za/docs/strategic/2013/app201415.pdf. [Accessed 20 August 2014].
30. Johnson LF, Mossong J, Dorrington RE, Schomaker M, Hoffmann CJ, Keiser O, et al.
Life expectancies of South African adults starting antiretroviral treatment: collaborative analysis of cohort studies.
PLoS Med 2013; 10:e1001418.
31. Adekeye OA, Heiman HJ, Onyeabor OS, Hyacinth HI.
The new invincibles: HIV screening among older adults in the U.S.
PLoS One 2012; 7:e43618.
32. Naik R, Tabana H, Doherty T, Zembe W, Jackson D.
Client characteristics and acceptability of a home-based HIV counselling and testing intervention in rural South Africa.
BMC Public Health 2012; 12:824.
33. April MD, Walensky RP, Chang Y, Pitt J, Freedberg KA, Losina E, et al.
HIV testing rates and outcomes in a South African community, 2001–2006: implications for expanded screening policies.
J Acquir Immune Defic Syndr 2009; 51:310–316.
34. Siu GE, Wight D, Seeley JA.
Masculinity, social context and HIV testing: an ethnographic study of men in Busia district, rural eastern Uganda.
BMC Public Health 2014; 14:33.
35. Wolff B, Nyanzi B, Katongole G, Ssesanga D, Ruberantwari A, Whitworth J.
Evaluation of a home-based voluntary counselling and testing intervention in rural Uganda.
Health Policy Plan 2005; 20:109–116.
36. McPhee SJ, Nguyen TT, Shema SJ, Nguyen B, Somkin C, Vo P, et al.
Validation of recall of breast and cervical cancer screening by women in an ethnically diverse population.
Prevent Med 2002; 35:463–473.
37. van Rooyen H, Barnabas RV, Baeten JM, Phakathi Z, Joseph P, Krows M, et al.
High HIV testing uptake and linkage to care in a novel program of home-based HIV counseling and testing with facilitated referral in KwaZulu-Natal, South Africa.
J Acquir Immune Defic Syndr 2013; 64:e1–e8.
38. Iwuji C, Orne-Gliemann J, Tanser F, Thiébaut R, Larmarange J, Okesola N, et al.
Feasibility and acceptability of an antiretroviral treatment as prevention (TasP) intervention in rural South Africa: results from the ANRS 12249 TasP cluster-randomised trial [Abstract WEAC0105LB]. 20th International AIDS Conference; 20–25 July 2014. Melbourne, Australia.
39. Huerga H, Maman D, Etard JF, Farhat JB, Bouhenia M.
Mbongolwane and Eshowe HIV impact in population survey. Paris, France: Epicentre and Medecins sans Frontieres; 2014.
40. Marzinke MA, Clarke W, Wang L, Cummings V, Liu TY, Piwowar-Manning E, et al.
Nondisclosure 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.
41. Sullivan AK, Savage EJ, Lowndes CM, Paul G, Murphy G, Carne S, et al.
Nondisclosure of HIV status in UK sexual health clinics - a pilot study to identify nondisclosure within a national unlinked anonymous seroprevalence survey.
Sex Trans Infect 2013; 89:120–121.
43. Sabapathy K, Van den Bergh R, Fidler S, Hayes R, Ford N.
Uptake of home-based voluntary HIV testing in sub-Saharan Africa: a systematic review and meta-analysis.
PLoS Med 2012; 9:e1001351.
44. Govindasamy D, Kranzer K, van Schaik N, Noubary F, Wood R, Walensky RP, et al.
Linkage to HIV, TB and noncommunicable disease care from a mobile testing unit in Cape Town, South Africa.
PLoS One 2013; 8:e80017.
45. Goubar A, Ades AE, De Angelis D, McCarrigle CA, Mercer CH, Tookey PA, et al.
Estimates of human immunodeficiency virus prevalence and proportion diagnosed based on Bayesian multiparameter synthesis of surveillance data.
J Royal Stat Soc Series A 2008; 171:541–567.
46. Kellerman S, Essajee S.
HIV testing for children in resource-limited settings: what are we waiting for?.
PLoS Med 2010; 7:e1000285.
47. Hamers FF, Phillips AN.
Diagnosed and undiagnosed HIV-infected populations in Europe.
HIV Med 2008; 9 (Suppl 2):6–12.
48. Campsmith ML, Rhodes PH, Hall HI, Green TA.
Undiagnosed HIV prevalence among adults and adolescents in the United States at the end of 2006.
J Acquir Immune Defic Syndr 2010; 53:619–624.
49. Supervie V, Ndawinz JD, Lodi S, Costagliola D.
The undiagnosed HIV epidemic in France and its implications for HIV screening strategies.
AIDS 2014; 28:1797–1804.
50. Aghaizu A, Brown AE, Nardone A, Gill ON, Delpech VC. HIV in the United Kingdom: 2013 report. London: Public Health England; 2013.
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/326601/HIV_annual_report_2013.pdf. [Accessed 21 September 2014].
51. Maheswaran H, Thulare H, Stanistreet D, Tanser F, Newell ML.
Starting a home and mobile HIV testing service in a rural area of South Africa.
J Acquir Immune Defic Syndr 2012; 59:e43–e46.
52. Doherty T, Tabana H, Jackson D, Naik R, Zembe W, Lombard C, et al.
Effect of home based HIV counselling and testing intervention in rural South Africa: cluster randomised trial.
Br Med J 2013; 346:f3481.
53. Kharsany AB, Karim QA, Karim SS.
Uptake of provider-initiated HIV testing and counseling among women attending an urban sexually transmitted disease clinic in South Africa: missed opportunities for early diagnosis of HIV infection.
AIDS Care 2010; 22:533–537.
54. Leon N, Naidoo P, Mathews C, Lewin S, Lombard C.
The impact of provider-initiated (opt-out) HIV testing and counseling of patients with sexually transmitted infection in Cape Town, South Africa: a controlled trial.
Implement Sci 2010; 5:8.
55. Sweat M, Morin S, Celentano D, Mulawa M, Singh B, Mbwambo J, et al.
Community-based intervention to increase HIV testing and case detection in people aged 16–32 years in Tanzania, Zimbabwe, and Thailand (NIMH Project Accept, HPTN 043): a randomised study.
Lancet Infect Dis 2011; 11:525–532.
56. van Schaik N, Kranzer K, Wood R, Bekker LG.
Earlier HIV diagnosis: are mobile services the answer?.
South Afr Med J 2010; 100:671–674.
57. Mabuto T, Latka MH, Kuwane B, Churchyard GJ, Charalambous S, Hoffmann CJ.
Four models of HIV counseling and testing: utilization and test results in South Africa.
PLoS One 2014; 9:e102267.
58. Bassett IV, Govindasamy D, Erlwanger AS, Hyle EP, Kranzer K, van Schaik N, et al.
Mobile HIV screening in Cape Town, South Africa: clinical impact, cost and cost-effectiveness.
PLoS One 2014; 9:e85197.
59. Nglazi MD, van Schaik N, Kranzer K, Lawn SD, Wood R, Bekker LG.
An incentivized HIV counseling and testing program targeting hard-to-reach unemployed men in Cape Town, South Africa.
J Acquir Immune Defic Syndr 2012; 59:e28–e34.
60. Kranzer K, Govindasamy D, van Schaik N, Thebus E, Davies N, Zimmermann M, et al.
Incentivized recruitment of a population sample to a mobile HIV testing service increases the yield of newly diagnosed cases, including those in need of antiretroviral therapy.
HIV Med 2012; 13:132–137.
61. Pant Pai N, Sharma J, Shivkumar S, Pillay S, Vadnais C, Joseph L, et al.
Supervised and unsupervised self-testing for HIV in high- and low-risk populations: a systematic review.
PLoS Med 2013; 10:e1001414.
