South Africa has over 6,000,000 HIV infected individuals and the province of KwaZulu-Natal (KZN) is the most severely affected. As public health initiatives to better control the HIV epidemic are implemented, timely, detailed and robust surveillance data are needed to monitor, evaluate and inform the programmatic interventions and policies over time. We describe the rationale and design of the HIV Incidence Provincial Surveillance System (HIPSS) to monitor HIV prevalence and incidence.
The household-based survey will include a sample of men and women from two sub-districts of the uMgungundlovu municipality (Vulindlela and the Greater Edendale) of KZN, South Africa. The study is designed as two sequential cross-sectional surveys of 10,000 randomly selected individuals aged 15–49 years to be conducted one year apart. From the cross sectional surveys, two sequential cohorts of HIV negative individuals aged 15–35 years will be followed-up one year later to measure the primary outcome of HIV incidence. Secondary outcomes include the laboratory measurements for pulmonary tuberculosis, sexually transmitted infections and evaluating tests for estimating population-level HIV incidence.
Antiretroviral therapy (ART) access, HIV-1 RNA viral load, and CD4 cell counts in HIV positive individuals will assess the effectiveness of the HIV treatment cascade. Household and individual-level socio-demographic characteristics, exposure to HIV programmatic interventions and risk behaviours will be assessed as predictors of HIV incidence. The incidence rate ratio of the two cohorts will be calculated to quantify the change in HIV incidence between consecutive samples. In anticipation of better availability of population-level HIV prevention and treatment programmes leading to decreases in HIV incidence, the sample size provides 84 % power to detect a reduction of 30 % in the HIV incidence rate between surveys.
The results from HIPSS will provide critical data regarding HIV prevalence and incidence in this community and will establish whether HIV prevention and treatment efforts in a “real world”, non-trial setting have an impact on HIV incidence at a population level. Importantly, the study design and methods will inform future methods for HIV surveillance.
Prisons are recognised internationally as institutions with very high tuberculosis (TB) burdens where transmission is predominantly determined by contact between infectious and susceptible prisoners. A recent South African court case described the conditions under which prisoners awaiting trial were kept. With the use of these data, a mathematical model was developed to explore the interactions between incarceration conditions and TB control measures.
Cell dimensions, cell occupancy, lock-up time, TB incidence and treatment delays were derived from court evidence and judicial reports. Using the Wells-Riley equation and probability analyses of contact between prisoners, we estimated the current TB transmission probability within prison cells, and estimated transmission probabilities of improved levels of case finding in combination with implementation of national and international minimum standards for incarceration.
Levels of overcrowding (230%) in communal cells and poor TB case finding result in annual TB transmission risks of 90% per annum. Implementing current national or international cell occupancy recommendations would reduce TB transmission probabilities by 30% and 50%, respectively. Improved passive case finding, modest ventilation increase or decreased lock-up time would minimally impact on transmission if introduced individually. However, active case finding together with implementation of minimum national and international standards of incarceration could reduce transmission by 50% and 94%, respectively.
Current conditions of detention for awaiting-trial prisoners are highly conducive for spread of drug-sensitive and drug-resistant TB. Combinations of simple well-established scientific control measures should be implemented urgently.
The WHO’s 2013 revisions to its Consolidated Guidelines on ARVs will recommend routine viral load monitoring (VLM), rather than clinical or immunological monitoring, as the preferred monitoring approach on the basis of clinical evidence. However, HIV programmes in resource-limited settings require guidance on the most cost-effective use of resources given other competing priorities, including expansion of ART coverage. Here we assess the cost-effectiveness of alternative patient monitoring strategies.
A range of monitoring strategies was evaluated, including clinical, CD4 and viral load monitoring alone and together at different frequencies and with different criteria for switching to second-line therapies. Three independently-constructed and validated models were analysed simultaneously. Costs were estimated based on resource use projected in the models and associated unit costs; impact was quantified as disability-adjusted life years (DALYs) averted. Alternatives were compared using incremental cost-effectiveness analysis.
All models show that clinical monitoring delivers significant benefit compared to a hypothetical baseline scenario with no monitoring or switching. Regular CD4 cell count monitoring confers a benefit over clinical monitoring alone, at an incremental cost that makes it affordable in more settings than VLM, which is currently more expensive. VLM without CD4 every six to 12 months provides the greatest reductions in morbidity and mortality, but incurs a high cost per DALY averted, resulting in lost opportunities to generate health gains if implemented instead of increasing ART coverage or expanding ART eligibility.
The priority for HIV programmes should be to expand ART coverage, firstly at CD4 <350 cells and then at CD4 <500, using lower-cost clinical or CD4 monitoring. At current costs, VLM should be considered only after high ART coverage has been achieved. Point-of-care technologies and other factors reducing costs may make VLM more affordable in future.
The HIV Modelling Consortium is funded by the Bill and Melinda Gates Foundation. Funding for this work was also provided by the World Health Organization.
Mean duration of recent infection (MDRI) and misclassification of long-term HIV-1 infections, as proportion false recent (PFR), are critical parameters for laboratory-based assays for estimating HIV-1 incidence. Recent review of the data by us and others indicated that MDRI of LAg-Avidity EIA estimated previously required recalibration. We present here results of recalibration efforts using >250 seroconversion panels and multiple statistical methods to ensure accuracy and consensus.
A total of 2737 longitudinal specimens collected from 259 seroconverting individuals infected with diverse HIV-1 subtypes were tested with the LAg-Avidity EIA as previously described. Data were analyzed for determination of MDRI at ODn cutoffs of 1.0 to 2.0 using 7 statistical approaches and sub-analyzed by HIV-1 subtypes. In addition, 3740 specimens from individuals with infection >1 year, including 488 from patients with AIDS, were tested for PFR at varying cutoffs.
Using different statistical methods, MDRI values ranged from 88–94 days at cutoff ODn = 1.0 to 177–183 days at ODn = 2.0. The MDRI values were similar by different methods suggesting coherence of different approaches. Testing for misclassification among long-term infections indicated that overall PFRs were 0.6% to 2.5% at increasing cutoffs of 1.0 to 2.0, respectively. Balancing the need for a longer MDRI and smaller PFR (<2.0%) suggests that a cutoff ODn = 1.5, corresponding to an MDRI of 130 days should be used for cross-sectional application. The MDRI varied among subtypes from 109 days (subtype A&D) to 152 days (subtype C).
Based on the new data and revised analysis, we recommend an ODn cutoff = 1.5 to classify recent and long-term infections, corresponding to an MDRI of 130 days (118–142). Determination of revised parameters for estimation of HIV-1 incidence should facilitate application of the LAg-Avidity EIA for worldwide use.
The estimation of HIV incidence from cross-sectional surveys using tests for recent infection has attracted much interest. It is increasingly recognized that the lack of high performance recent infection tests is hindering the implementation of this surveillance approach. With growing funding opportunities, test developers are currently trying to fill this gap. However, there is a lack of consensus and clear guidance for developers on the evaluation and optimization of candidate tests. A fundamental shift from conventional thinking about test performance is needed: away from metrics relevant in typical public health settings where the detection of a condition in individuals is of primary interest (sensitivity, specificity, and predictive values) and toward metrics that are appropriate when estimating a population-level parameter such as incidence (accuracy and precision). The inappropriate use of individual-level diagnostics performance measures could lead to spurious assessments and suboptimal designs of tests for incidence estimation. In some contexts, such as population-level application to HIV incidence, bias of estimates is essentially negligible, and all that remains is the maximization of precision. The maximization of the precision of incidence estimates provides a completely general criterion for test developers to assess and optimize test designs. Summarizing the test dynamics into the properties relevant for incidence estimation, high precision estimates are obtained when (1) the mean duration of recent infection is large, and (2) the false-recent rate is small. The optimal trade-off between these two test properties will produce the highest precision, and therefore the most epidemiologically useful incidence estimates.
To report the incidence rates of TB and HIV in household contacts of index patients diagnosed with TB.
A prospective cohort study in the Matlosana sub-district of North West Province, South Africa.
Contacts of index TB patients received TB and HIV testing after counseling at their first household visit and were then followed up a year later, in 2010. TB or HIV diagnoses that occurred during the period were determined.
For 2,377 household contacts, the overall observed TB incidence rate was 1.3 per 100 person years (95% CI 0.9–1.9/100py) and TB incidence for individuals who were HIV-infected and HIV seronegative at baseline was 5.4/100py (95% CI 2.9–9.0/100py) and 0.7/100py (95% CI 0.3–1.4/100py), respectively. The overall HIV incidence rate was 2.2/100py (95% CI 1.3–8.4/100py).
In the year following a household case finding visit when household contacts were tested for TB and HIV, the incidence rate of both active TB and HIV infection was found to be extremely high. Clearly, implementing proven strategies to prevent HIV acquisition and preventing TB transmission and progression to disease remains a priority in settings such as South Africa.
It is attractive to estimate disease incidence from cross-sectional surveys, using biomarkers for “recent” infection. Despite considerable interest in applications to HIV, there is currently no consensus on the correct handling of “recent” biomarkers appearing in persons long after infection.
We derive a general expression for a weighted average of recent incidence that – unlike previous estimators – requires no particular assumption about recent infection biomarker dynamics, or about the demographic and epidemiologic context. This is possible through the introduction of an explicit timescale T that truncates the period of averaging implied by the estimator.
The recent infection test dynamics can be summarized into two parameters, similar to those appearing in previous estimators: a mean duration of recent infection and a false-recent rate. We identify a number of dimensionless parameters that capture the bias that arises from working with tractable forms for the resulting estimator, and elucidate the utility of the incidence estimator in terms of the performance of the recency test and the population state. Estimation of test characteristics and incidence is demonstrated using simulated data. The observed confidence interval coverage of the test characteristics and incidence is within 1% of intended coverage.
Biomarker-based incidence estimation can be consistently adapted to a general context without the strong assumptions of previous work about biomarker dynamics and epidemiologic and demographic history.
Efficient HIV prevention requires accurate identification of individuals with risky sexual behaviour. However, self-reported data from sexual behaviour surveys are prone to social desirability bias (SDB). Audio Computer-Assisted Self-Interviewing (ACASI) has been suggested as an alternative to face-to-face interviewing (FTFI), because it may promote interview privacy and reduce SDB. However, little is known about the suitability and accuracy of ACASI in urban communities with high HIV prevalence in South Africa. To test this, we conducted a sexual behaviour survey in Cape Town, South Africa, using ACASI methods.
Participants (n = 878) answered questions about their sexual relationships on a touch screen computer in a private mobile office. We included questions at the end of the ACASI survey that were used to assess participants’ perceived ease of use, privacy, and truthfulness. Univariate logistic regression models, supported by multivariate models, were applied to identify groups of people who had adverse interviewing experiences. Further, we constructed male–female ratios of self-reported sexual behaviours as indicators of SDB. We used these indicators to compare SDB in our survey and in recent FTFI-based Demographic and Health Surveys (DHSs) from Lesotho, Swaziland, and Zimbabwe.
Most participants found our methods easy to use (85.9%), perceived privacy (96.3%) and preferred ACASI to other modes of inquiry (82.5%) when reporting on sexual behaviours. Unemployed participants and those in the 40–70 year old age group were the least likely to find our methods easy to use (OR 0.69; 95% CI: 0.47–1.01 and OR 0.37; 95% CI: 0.23–0.58, respectively). In our survey, the male–female ratio for reporting >2 sexual partners in the past year, a concurrent relationship in the past year, and > 2 sexual partners in a lifetime was 3.4, 2.6, and 1.2, respectively— far lower than the ratios observed in the Demographic and Health Surveys.
Our analysis suggests that most participants in our survey found the ACASI modality to be acceptable, private, and user-friendly. Moreover, our results indicate lower SDB than in FTFI techniques. Targeting older and unemployed participants for ACASI training prior to taking the survey may help to improve their perception of ease and privacy.
ACASI; Sexual behaviour; Social desirability bias; Self-reported data; Gender; South Africa
We studied 1163 sexually-active HIV-infected South African men and women in an urban primary care program to understand patterns of sexual behaviors and whether these behaviors differed by partner HIV status. Overall, 40% reported a HIV-positive partner and 60% a HIV-negative or status unknown partner; and 17.5% reported >2 sex acts in the last 2 weeks, 16.4% unprotected sex in the last 6 months, and 3.7% >1 sex partner in the last 6 months. Antiretroviral therapy (ART) was consistently associated with decreased sexual risk behaviors, as well as with reporting a HIV-negative or status unknown partner. The odds of sexual risk behaviors differed by sex; and were generally higher among participants reporting a HIV-positive partner, but continued among those with a HIV-negative or status unknown partner. These data support ART as a means of HIV prevention. Engaging in sexual risk behaviors primarily with HIV-positive partners was not widely practiced in this setting, emphasizing the need for couples-based prevention.
HIV; AIDS; South Africa; Sexual risk behavior; ART
We derive a new method to estimate the age specific incidence of an infection with a differential mortality, using individual level infection status data from successive surveys. The method consists of a) an SI-type model to express the incidence rate in terms of the prevalence and its derivatives as well as the difference in mortality rate, and b) a maximum likelihood approach to estimate the prevalence and its derivatives. Estimates can in principle be obtained for any chosen age and time, and no particular assumptions are made about the epidemiological or demographic context. This is in contrast with earlier methods for estimating incidence from prevalence data, which work with aggregated data, and the aggregated effect of demographic and epidemiological rates over the time interval between prevalence surveys. Numerical simulation of HIV epidemics, under the presumption of known excess mortality due to infection, shows improved control of bias and variance, compared to previous methods. Our analysis motivates for a) effort to be applied to obtain accurate estimates of excess mortality rates as a function of age and time among HIV infected individuals and b) use of individual level rather than aggregated data in order to estimate HIV incidence rates at times between two prevalence surveys.
Public health responses to HIV epidemics have long relied on epidemiological modelling analyses to help prospectively project and retrospectively estimate the impact, cost-effectiveness, affordability, and investment returns of interventions, and to help plan the design of evaluations. But translating model output into policy decisions and implementation on the ground is challenged by the differences in background and expectations of modellers and decision-makers. As part of the PLoS Medicine Collection “Investigating the Impact of Treatment on New HIV Infections”—which focuses on the contribution of modelling to current issues in HIV prevention—we present here principles of “best practice” for the construction, reporting, and interpretation of HIV epidemiological models for public health decision-making on all aspects of HIV. Aimed at both those who conduct modelling research and those who use modelling results, we hope that the principles described here will become a shared resource that facilitates constructive discussions about the policy implications that emerge from HIV epidemiology modelling results, and that promotes joint understanding between modellers and decision-makers about when modelling is useful as a tool in quantifying HIV epidemiological outcomes and improving prevention programming.
Reliable methods for measuring human immunodeficiency virus (HIV) incidence are a high priority for HIV prevention. They are particularly important to assess the population-level effectiveness of new prevention strategies, to evaluate the community-wide impact of ongoing prevention programs, and to assess whether a proposed prevention trial can be performed in a timely and cost-efficient manner in a particular population and setting. New incidence assays and algorithms that are accurate, rapid, cost-efficient, and can be performed on easily-obtained specimens are urgently needed. On May 4, 2011, the Division of AIDS (DAIDS), National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH), sponsored a 1-day workshop to examine strategies for developing new assays to distinguish recent from chronic HIV infections. Participants included leading investigators, clinicians, public health experts, industry, regulatory specialists, and other stakeholders. Immune-based parameters, markers of viral sequence diversity, and other biomarkers such as telomere length were evaluated. Emerging nanotechnology and chip-based diagnostics, including algorithms for performing diverse assays on a single platform, were also reviewed. This report summarizes the presentations, panel discussions, and the consensus reached for pursuing the development of a new generation of HIV incidence assays.
The relative contributions of transmission and reactivation of latent infection to TB cases observed clinically has been reported in many situations, but always with some uncertainty. Genotyped data from TB organisms obtained from patients have been used as the basis for heuristic distinctions between circulating (clustered strains) and reactivated infections (unclustered strains). Naïve methods previously applied to the analysis of such data are known to provide biased estimates of the proportion of unclustered cases. The hypergeometric distribution, which generates probabilities of observing clusters of a given size as realized clusters of all possible sizes, is analyzed in this paper to yield a formal estimator for genotype cluster sizes. Subtle aspects of numerical stability, bias, and variance are explored. This formal estimator is seen to be stable with respect to the epidemiologically interesting properties of the cluster size distribution (the number of clusters and the number of singletons) though it does not yield satisfactory estimates of the number of clusters of larger sizes. The problem that even complete coverage of genotyping, in a practical sampling frame, will only provide a partial view of the actual transmission network remains to be explored.
Crucial connections between sexual network structure and the distribution of HIV remain inadequately understood, especially in regard to the role of concurrency and age disparity in relationships, and how these network characteristics correlate with each other and other risk factors. Social desirability bias and inaccurate recall are obstacles to obtaining valid, detailed information about sexual behaviour and relationship histories. Therefore, this study aims to use novel research methods in order to determine whether HIV status is associated with age-disparity and sexual connectedness as well as establish the primary behavioural and socio-demographic predictors of the egocentric and community sexual network structures.
We will conduct a cross-sectional survey that uses a questionnaire exploring one-year sexual histories, with a focus on timing and age disparity of relationships, as well as other risk factors such as unprotected intercourse and the use of alcohol and recreational drugs. The questionnaire will be administered in a safe and confidential mobile interview space, using audio computer-assisted self-interview (ACASI) technology on touch screen computers. The ACASI features a choice of languages and visual feedback of temporal information. The survey will be administered in three peri-urban disadvantaged communities in the greater Cape Town area with a high burden of HIV. The study communities participated in a previous TB/HIV study, from which HIV test results will be anonymously linked to the survey dataset. Statistical analyses of the data will include descriptive statistics, linear mixed-effects models for the inter- and intra-subject variability in the age difference between sexual partners, survival analysis for correlated event times to model concurrency patterns, and logistic regression for association of HIV status with age disparity and sexual connectedness.
This study design is intended to facilitate more accurate recall of sensitive sexual history data and has the potential to provide substantial insights into the relationship between key sexual network attributes and additional risk factors for HIV infection. This will help to inform the design of context-specific HIV prevention programmes.
HIV incidence is the rate of new infections in a population over time. HIV incidence is a critical indicator needed to assess the status and trends of the HIV epidemic in populations and guide and assess the impact of prevention interventions.
Several methods exist for estimating population-level HIV incidence: direct observation of HIV incidence through longitudinal follow-up of persons at risk for new HIV infection, indirect measurement of HIV incidence using data on HIV prevalence and mortality in a population, and direct measurement of HIV incidence through use of tests for recent infection (TRIs) that can differentiate “recent” from “non-recent” infections based on biomarkers in cross-sectional specimens. Given the limitations in measuring directly observed incidence and the assumptions needed for indirect measurements of incidence, there is an increasing demand for TRIs for HIV incidence surveillance and program monitoring and evaluation purposes.
Over ten years since the introduction of the first TRI, a number of low-, middle-, and high-income countries have integrated this method into their HIV surveillance systems to monitor HIV incidence in the population. However, the accuracy of these assays for measuring HIV incidence has been unsatisfactory to date, mainly due to misclassification of chronic infections as recent infection on the assay. To improve the accuracy of TRIs for measuring incidence, countries are recommended to apply case-based adjustments, formula-based adjustments using local correction factors, or laboratory-based adjustment to minimize error related to assay misclassification. Multiple tests may be used in a recent infection testing algorithm (RITA) to obtain more accurate HIV incidence estimates.
There continues to be a high demand for improved TRIs and RITAs to monitor HIV incidence, determine prevention priorities, and assess impact of interventions. Current TRIs have noted limitations, but with appropriate adjustments, interpreted in parallel with other epidemiologic data, may still provide useful information on new infections in a population. New TRIs and RITAs with improved accuracy and performance are needed and development of these tools should be supported.
Potent antiretroviral therapy (ART) reduces mortality and morbidity in people living with HIV by reducing viral load and allowing their immune systems to recover. The reduction in viral load soon after starting ART has led to the hypothesis that early and widespread ART could prevent onward transmission and therefore eliminate the HIV epidemic in the long term. While several authors have argued that it is feasible to use HIV treatment as prevention (TasP), provided treatment is started sufficiently early, others have reasonably drawn attention to the many operational difficulties that will need to be overcome if the strategy is to succeed in reducing HIV transmission. Furthermore, international public health policy must be based on more than theoretical studies, no matter how appealing. Community randomized controlled trials provide the gold standard for testing the extent to which early treatment reduces incidence, but much still needs to be understood and the immediate need is for operational studies to explore the practical feasibility of this approach. Here, we examine some of the issues to be addressed, the obstacles to be overcome, and strategies that may be necessary if TasP is to be effective. Studies of this kind will provide valuable information for the design of large-scale trials, as well as essential information that will be needed if early treatment is to be incorporated into public health policy.
Biomarker-based cross-sectional incidence estimation requires a Recent Infection Testing Algorithm (RITA) with an adequately large mean recency duration, to achieve reasonable survey counts, and a low false-recent rate, to minimise exposure to further bias and imprecision. Estimating these characteristics requires specimens from individuals with well-known seroconversion dates or confirmed long-standing infection. Specimens with well-known seroconversion dates are typically rare and precious, presenting a bottleneck in the development of RITAs.
The mean recency duration and a ‘false-recent rate’ are estimated from data on seroconverting blood donors. Within an idealised model for the dynamics of false-recent results, blood donor specimens were used to characterise RITAs by a new method that maximises the likelihood of cohort-level recency classifications, rather than modelling individual sojourn times in recency.
For a range of assumptions about the false-recent results (0% to 20% of biomarker response curves failing to reach the threshold distinguishing test-recent and test-non-recent infection), the mean recency duration of the Vironostika-LS ranged from 154 (95% CI: 96–231) to 274 (95% CI: 234–313) days in the South African donor population (n = 282), and from 145 (95% CI: 67–226) to 252 (95% CI: 194–308) days in the American donor population (n = 106). The significance of gender and clade on performance was rejected (p−value = 10%), and utility in incidence estimation appeared comparable to that of a BED-like RITA. Assessment of the Vitros-LS (n = 108) suggested potentially high false-recent rates.
The new method facilitates RITA characterisation using widely available specimens that were previously overlooked, at the cost of possible artefacts. While accuracy and precision are insufficient to provide estimates suitable for incidence surveillance, a low-cost approach for preliminary assessments of new RITAs has been demonstrated. The Vironostika-LS and Vitros-LS warrant further analysis to provide greater precision of estimates.
Cross-sectional surveys utilizing biomarkers that test for recent infection provide a convenient and cost effective way to estimate HIV incidence. In particular, the BED assay has been developed for this purpose. Controversy surrounding the way in which false positive results from the biomarker should be handled has lead to a number of different estimators that account for imperfect specificity. We compare the estimators proposed by McDougal et al., Hargrove et al. and McWalter & Welte.
The three estimators are analyzed and compared. An identity showing a relationship between the calibration parameters in the McDougal methodology is shown. When the three estimators are tested under a steady state epidemic, which includes individuals who fail to progress on the biomarker, only the McWalter/Welte method recovers an unbiased result.
Our analysis shows that the McDougal estimator can be reduced to a formula that only requires calibration of a mean window period and a long-term specificity. This allows simpler calibration techniques to be used and shows that all three estimators can be expressed using the same set of parameters. The McWalter/Welte method is applicable under the least restrictive assumptions and is the least prone to bias of the methods reviewed.
Highly selective antiretroviral (ARV) regimens such as single dose nevirapine (NVP) used for prevention of mother to child transmission (PMTCT) in resource-limited settings produce transient increases in otherwise marginal subpopulations of cells infected by mutant genomes. The longer term implications for accumulation of further resistance mutations are not fully understood.
We develop a new strain-differentiated hybrid deterministic-stochastic population dynamic type model of healthy and infected cells. We explore how the transient increase in a population of cells transcribed with a common mutation (modelled deterministically), which occurs in response to a short course of monotherapy, has an impact on the risk of appearance of rarer, higher-order, therapy-defeating mutations (modelled stochastically).
Scenarios with a transient of a magnitude and duration such as is known to occur under NVP monotherapy exhibit significantly accelerated viral evolution compared to no-treatment scenarios. We identify a possibly important new biological timescale; namely, the duration of persistence, after a seminal mutation, of a sub-population of cells bearing the new mutant gene, and we show how increased persistence leads to an increased probability that a rare mutant will be present at the moment at which a new treatment regimen is initiated.
Even transient increases in subpopulations of common mutants are associated with accelerated appearance of further rarer mutations. Experimental data on the persistence of small subpopulations of rare mutants, in unfavourable environments, should be sought, as this affects the risk of subverting later regimens.
The BED IgG-Capture Enzyme Immunoassay (cBED assay), a test of recent HIV infection, has been used to estimate HIV incidence in cross-sectional HIV surveys. However, there has been concern that the assay overestimates HIV incidence to an unknown extent because it falsely classifies some individuals with non-recent HIV infections as recently infected. We used data from a longitudinal HIV surveillance in rural South Africa to measure the fraction of people with non-recent HIV infection who are falsely classified as recently HIV-infected by the cBED assay (the long-term false-positive ratio (FPR)) and compared cBED assay-based HIV incidence estimates to longitudinally measured HIV incidence.
We measured the long-term FPR in individuals with two positive HIV tests (in the HIV surveillance, 2003–2006) more than 306 days apart (sample size n = 1,065). We implemented four different formulae to calculate HIV incidence using cBED assay testing (n = 11,755) and obtained confidence intervals (CIs) by directly calculating the central 95th percentile of incidence values. We observed 4,869 individuals over 7,685 person-years for longitudinal HIV incidence estimation. The long-term FPR was 0.0169 (95% CI 0.0100–0.0266). Using this FPR, the cross-sectional cBED-based HIV incidence estimates (per 100 people per year) varied between 3.03 (95% CI 2.44–3.63) and 3.19 (95% CI 2.57–3.82), depending on the incidence formula. Using a long-term FPR of 0.0560 based on previous studies, HIV incidence estimates varied between 0.65 (95% CI 0.00–1.32) and 0.71 (95% CI 0.00–1.43). The longitudinally measured HIV incidence was 3.09 per 100 people per year (95% CI 2.69–3.52), after adjustment to the sex-age distribution of the sample used in cBED assay-based estimation.
In a rural community in South Africa with high HIV prevalence, the long-term FPR of the cBED assay is substantially lower than previous estimates. The cBED assay performs well in HIV incidence estimation if the locally measured long-term FPR is used, but significantly underestimates incidence when a FPR estimate based on previous studies in other settings is used.
Cross-sectional HIV incidence surveillance, using assays that distinguish ‘recent’ from ‘nonrecent’ infections, has been hampered by inadequate performance and characterization of incidence assays. In this study, the Consortium for the Evaluation and Performance of HIV Incidence Assays presents results of the first independent evaluation of five incidence assays (BED, Limiting Antigen Avidity, Less-sensitive Vitros, Vitros Avidity and BioRad Avidity).
A large repository of diverse specimens from HIV-positive patients was established, multiple assays were run on 2500 selected specimens, and data were analyzed to estimate assay characteristics relevant for incidence surveillance.
The mean duration of recent infection (MDRI, average time ‘recent’ while infected for less than some time cut-off T) was estimated from longitudinal data on seroconverters by regression. The false-recent rate (FRR, probability of testing ‘recent’ when infected for longer than T) was explored by measuring the proportions of ‘recent’ results in various subsets of patients.
Assays continue to fail to attain the simultaneously large MDRI and small FRR demanded by existing performance guidelines. All assays produce high FRRs amongst virally suppressed patients (>40%), including elite controllers and treated patients.
Results from this first independent evaluation provide valuable information about the current performance of assays, and suggest the need for further optimization. Variation of ‘recent’/‘nonrecent’ thresholds and the use of multiple antibody-maturation assays, as well as other biomarkers, can now be explored, using the rich data generated by the Consortium for the Evaluation and Performance of HIV Incidence Assays. Consistently high FRRs amongst those virally suppressed suggest that viral load will be a particularly valuable supplementary marker.
biomarkers; HIV; incidence assays; incidence estimation; recent infection