Evaluation of an empiric risk screening score to identify acute and early HIV-1 infection among MSM in Coastal Kenya

Data mining and machine learning in HIV infection risk research: An overview and recommendations

In the contemporary era, the applications of data mining and machine learning have permeated extensively into medical research, significantly contributing to areas such as HIV studies. By reviewing 38 articles published in the past 15 years, the study presents a roadmap based on seven different aspects, utilizing various machine learning techniques for both novice researchers and experienced researchers seeking to comprehend the current state of the art in this area. While traditional regression modeling techniques have been commonly used, researchers are increasingly adopting more advanced fully supervised machine learning and deep learning techniques, which often outperform the traditional methods in predictive performance. Additionally, the study identifies nine new open research issues and outlines possible future research plans to enhance the outcomes of HIV infection risk research. This review is expected to be an insightful guide for researchers, illuminating current practices and suggesting advancements in the field.

Performance of HIV Infection Prediction Models in Men Who Have Sex with Men: A Systematic Review and Meta-Analysis

Effective ways to identify and predict men who have sex with men (MSM) at substantial risk for HIV is a global priority. HIV risk assessment tools can improve individual risk awareness and subsequent health-seeking actions. We sought to identify and characterize the performance of HIV infection risk prediction models in MSM through systematic review and meta-analysis. PubMed, Embase, and The Cochrane Library were searched. Eighteen HIV infection risk assessment models with a total of 151,422 participants and 3643 HIV cases were identified, eight of which have been externally validated by at least one study (HIRI-MSM, Menza Score, SDET Score, Li Model, DHRS, Amsterdam Score, SexPro model, and UMRSS). The number of predictor variables in each model ranged from three to 12, age, the number of male sexual partners, unprotected receptive anal intercourse, recreational drug usage (amphetamines, poppers), and sexually transmitted infections were critical scoring variables. All eight externally validated models performed well in terms of discrimination, with the pooled area under the receiver operating characteristic curve (AUC) ranging from 0.62 (95%CI: 0.51 to 0.73, SDET Score) to 0.83 (95%CI: 0.48 to 0.99, Amsterdam Score). Calibration performance was only reported in 10 studies (35.7%, 10/28). The HIV infection risk prediction models showed moderate-to-good discrimination performance. Validation of prediction models across different geographic and ethnic environments is needed to ensure their real-world application.

The role of machine learning in HIV risk prediction

Despite advances in reducing HIV-related mortality, persistently high HIV incidence rates are undermining global efforts to end the epidemic by 2030. The UNAIDS Fast-track targets as well as other preventative strategies, such as pre-exposure prophylaxis, have been identified as priority areas to reduce the ongoing transmission threatening to undermine recent progress. Accurate and granular risk prediction is critical for these campaigns but is often lacking in regions where the burden is highest. Owing to their ability to capture complex interactions between data, machine learning and artificial intelligence algorithms have proven effective at predicting the risk of HIV infection in both high resource and low resource settings. However, interpretability of these algorithms presents a challenge to the understanding and adoption of these algorithms. In this perspectives article, we provide an introduction to machine learning and discuss some of the important considerations when choosing the variables used in model development and when evaluating the performance of different machine learning algorithms, as well as the role emerging tools such as Shapely Additive Explanations may play in helping understand and decompose these models in the context of HIV. Finally, we discuss some of the potential public health and clinical use cases for such decomposed risk assessment models in directing testing and preventative interventions including pre-exposure prophylaxis, as well as highlight the potential integration synergies with algorithms that predict the risk of sexually transmitted infections and tuberculosis.

Predicting the risk of HIV infection among internal migrant MSM in China: An optimal model based on three variable selection methods

Introduction

Internal migrant Men who have sex with men (IMMSM), which has the dual identity of MSM and floating population, should be more concerned among the vulnerable groups for HIV in society. Establishing appropriate prediction models to assess the risk of HIV infection among IMMSM is of great significance to against HIV infection and transmission.

Methods

HIV and syphilis infection were detected using rapid test kits, and other 30 variables were collected among IMMSM through questionnaire. Taking HIV infection status as the dependent variable, three methods were used to screen predictors and three prediction models were developed respectively. The Hosmer-Lemeshow test was performed to verify the fit of the models, and the net classification improvement and integrated discrimination improvement were used to compare these models to determine the optimal model. Based on the optimal model, a prediction nomogram was developed as an instrument to assess the risk of HIV infection among IMMSM. To quantify the predictive ability of the nomogram, the C-index measurement was performed, and internal validation was performed using bootstrap method. The receiver operating characteristic (ROC) curve, calibration plot and dynamic component analysis (DCA) were respectively performed to assess the efficacy, accuracy and clinical utility of the prediction nomogram.

Results

In this study, 12.52% IMMSMs were tested HIV-positive and 8.0% IMMSMs were tested syphilis-positive. Model A, model B, and model C fitted well, and model B was the optimal model. A nomogram was developed based on the model B. The C-index of the nomogram was 0.757 (95% CI: 0.701-0.812), and the C-index of internal verification was 0.705.

Conclusions

The model established by stepwise selection methods incorporating 11 risk factors (age, education, marriage, monthly income, verbal violence, syphilis, score of CUSS, score of RSES, score of ULS, score of ES and score of DS) was the optimal model that achieved the best predictive power. The risk nomogram based on the optimal model had relatively good efficacy, accuracy and clinical utility in identifying internal migrant MSM at high-risk for HIV infection, which is helpful for developing targeted intervention for them.

Risk-Based Screening Tools to Optimise HIV Testing Services: a Systematic Review

PURPOSE OF REVIEW

Effective ways to diagnose the remaining people living with HIV who do not know their status are a global priority. We reviewed the use of risk-based tools, a set of criteria to identify individuals who would not otherwise be tested (screen in) or excluded people from testing (screen out).

RECENT FINDINGS

Recent studies suggest that there may be value in risk-based tools to improve testing efficiency (i.e. identifying those who need to be tested). However, there has not been any systematic reviews to synthesize these studies. We identified 18,238 citations, and 71 were included. The risk-based tools identified were most commonly from high-income (51%) and low HIV (<5%) prevalence countries (73%). The majority were for "screening in" (70%), with the highest performance tools related to identifying MSM with acute HIV. Screening in tools may be helpful in settings where it is not feasible or recommended to offer testing routinely. Caution is needed for screening out tools, where there is a trade-off between reducing costs of testing with missing cases of people living with HIV.

Predictors of testing history and new HIV diagnosis among adult outpatients seeking care for symptoms of acute HIV infection in coastal Kenya: a cross-sectional analysis of intervention participants in a stepped-wedge HIV testing trial

Background

HIV testing is the first step to stop transmission. We aimed to evaluate HIV testing history and new diagnoses among adult outpatients in Kenya aged 18–39 years seeking care for symptoms of acute HIV infection (AHI).

Methods

The Tambua Mapema Plus study, a stepped-wedge trial, enrolled patients presenting to care at six primary care facilities with symptoms of AHI for a targeted HIV-1 nucleic acid (NA) testing intervention compared with standard provider-initiated testing using rapid antibody tests. Intervention participants underwent a questionnaire and NA testing, followed by rapid tests if NA-positive. Multinomial logistic regression was used to analyse factors associated with never testing or testing > 1 year ago (“late retesting”) relative to testing ≤ 1 year ago (“on-time testers”). Logistic regression was used to analyse factors associated with new diagnosis. All analyses were stratified by sex.

Results

Of 1,500 intervention participants, 613 (40.9%) were men. Overall, 250 (40.8%) men vs. 364 (41.0%) women were late retesters, and 103 (16.8%) men vs. 50 (5.6%) women had never tested prior to enrolment. Younger age, single status, lower education level, no formal employment, childlessness, sexual activity in the past 6 weeks, and > 1 sexual partner were associated with testing history among both men and women. Intimate partner violence > 1 month ago, a regular sexual partner, and concurrency were associated with testing history among women only. New diagnoses were made in 37 (2.5%) participants (17 men and 20 women), of whom 8 (21.6%) had never tested and 16 (43.2%) were late retesters. Newly-diagnosed men were more likely to have symptoms for > 14 days, lower education level and no religious affiliation and less likely to be young, single, and childless than HIV-negative men; newly-diagnosed women were more likely to report fever than HIV-negative women. Among men, never testing was associated with fivefold increased odds (95% confidence interval 1.4–20.9) of new diagnosis relative to on-time testers in adjusted analyses.

Conclusion

Most new HIV diagnoses were among participants who had never tested or tested > 1 year ago. Strengthening provider-initiated testing targeting never testers and late retesters could decrease time to diagnosis among symptomatic adults in coastal Kenya.

Trial registration

ClinicalTrials.gov Identifier: NCT03508908 registered on 26/04/2018.

Clinical signs and symptoms associated with acute HIV infection from an intensely monitored cohort on 2 continents

Define the clinical presentation of acute human immunodeficiency virus infection (AHI) among men and women from 2 continents to create a clinical scoring algorithm.Comparison of incident sign and symptom between those with and without AHI.At-risk human immunodeficiency virus (HIV) negative men and women in Thailand, Kenya, Tanzania, and Uganda underwent twice-weekly testing for HIV. Newly diagnosed participants were evaluated twice weekly for 21 days after infection.Of the 3345 participants enrolled, 56 African females and 36 biological males from Thailand were diagnosed with AHI. Four hundred fifty-two of their encounters were compared to 18,281 HIV negative encounters. Due to a high degree of heterogeneity among incident symptoms, 2 unique subgroups based upon geography and sex were created. Among Thai males, the signs and symptoms with the greatest odds ratio (OR) between AHI and uninfected participants were nausea (OR 16.0, 95% confidence interval [CI] 3.9-60.2, P < .001) and lymphatic abnormalities (OR 11.8, 95% CI 4.2-49.0, P < .001); and among African females were pain behind the eyes (OR 44.4, 95% CI 12.0-158.0, P < .0001) and fatigue (OR 22.7, 95% CI 11.3-44.3, P < .001). The Thai male scoring algorithm had a 66% sensitivity and 84% specificity while the African female algorithm had a sensitivity of 27% and specificity of 98%.The different incident symptoms during AHI necessitated creating 2 different scoring algorithms that can guide diagnostic testing among a particular sex in the appropriate geographic setting. Further research on risk exposure, sex, and demographic specific models is warranted.

Predicting, Diagnosing, and Treating Acute and Early HIV Infection in a Public Sector Facility in Eswatini

BACKGROUND

The lack of acute and early HIV infection (AEHI) diagnosis and care contributes to high HIV incidence in resource-limited settings. We aimed to assess the yield of AEHI, predict and diagnose AEHI, and describe AEHI care outcomes in a public sector setting in Eswatini.

SETTING

This study was conducted in Nhlangano outpatient department from March 2019 to March 2020.

METHODS

Adults at risk of AEHI underwent diagnostic testing for AEHI with the quantitative Xpert HIV-1 viral load (VL) assay. AEHI was defined as the detection of HIV-1 VL on Xpert and either an HIV-seronegative or HIV-serodiscordant third-generation antibody-based rapid diagnostic test (RDT) result. First, the cross-sectional analysis obtained the yield of AEHI and established a predictor risk score for the prediction of AEHI using Lasso logistic regression. Second, diagnostic accuracy statistics described the ability of the fourth-generation antibody/p24 antigen-based Alere HIV-Combo RDT to diagnose AEHI (vs Xpert VL testing). Third, we described acute HIV infection care outcomes of AEHI-positive patients using survival analysis.

RESULTS

Of 795 HIV-seronegative/HIV-serodiscordant outpatients recruited, 30 (3.8%, 95% confidence interval: 2.6% to 5.3%) had AEHI. The predictor risk score contained several factors (HIV-serodiscordant RDT, women, feeling at risk of HIV, swollen glands, and fatigue) and had sensitivity and specificity of 83.3% and 65.8%, respectively, to predict AEHI. The HIV-Combo RDT had sensitivity and specificity of 86.2% and 99.9%, respectively, to diagnose AEHI. Of 30 AEHI-positive patients, the 1-month cumulative treatment initiation was 74% (95% confidence interval: 57% to 88%), and the 3-month viral suppression (<1000 copies/mL) was 87% (67% to 98%).

CONCLUSION

AEHI diagnosis and care seem possible in resource-limited settings.

Derivation and Validation of an HIV Risk Prediction Score Among Gay, Bisexual, and Other Men Who Have Sex With Men to Inform PrEP Initiation in an STD Clinic Setting

BACKGROUND

Clinicians and health departments would ideally undertake targeted efforts to promote HIV pre-exposure prophylaxis (PrEP) and frequent HIV testing using data-based criteria to identify populations at elevated risk for HIV. We developed an HIV risk prediction score for men who have sex with men (MSM) to identify individuals at substantial risk for HIV acquisition.

METHODS

We created a retrospective cohort of MSM who tested HIV-negative at the sexually transmitted disease clinic in Seattle, WA, from 2001 to 2015, and identified seroconversions using HIV surveillance data. We split the cohort randomly 2:1 into derivation and validation data sets, and used Cox proportional hazards to estimate the hazard of acquiring HIV associated with behavioral and clinical predictors, and the Akaike information criterion to determine which variables to retain in our model.

RESULTS

Among 16,448 MSM, 640 seroconverted over a 14.3-year follow-up period. The best prediction model included 13 variables and had an area under the receiver operating characteristic curve of 0.73 (95% confidence interval: 0.71 to 0.76), 76% sensitivity, and 63% specificity at a score cutoff ≥11. A simplified model restricted to 2011-2015 included 4 predictors [methamphetamine use, condomless receptive anal intercourse (CRAI), ≥10 partners, and current diagnosis or self-reported gonorrhea/syphilis in the past year]. This model, the Seattle PrEP Score, had an area under the receiver operating characteristic curve of 0.69 (95% confidence interval: 0.64 to 0.73), 62% sensitivity, and 70% specificity. One-year incidence was 0.5% for a score of 0, 0.7% for a score of 1, and 2.1% for scores ≥2.

CONCLUSIONS

The Seattle PrEP Score was predictive of HIV acquisition and could help clinicians and public health agencies identify MSM who could benefit from PrEP and frequent HIV testing.

Machine Learning to Identify Persons at High-Risk of Human Immunodeficiency Virus Acquisition in Rural Kenya and Uganda

BACKGROUND

In generalized epidemic settings, strategies are needed to prioritize individuals at higher risk of human immunodeficiency virus (HIV) acquisition for prevention services. We used population-level HIV testing data from rural Kenya and Uganda to construct HIV risk scores and assessed their ability to identify seroconversions.

METHODS

During 2013-2017, >75% of residents in 16 communities in the SEARCH study were tested annually for HIV. In this population, we evaluated 3 strategies for using demographic factors to predict the 1-year risk of HIV seroconversion: membership in ≥1 known "risk group" (eg, having a spouse living with HIV), a "model-based" risk score constructed with logistic regression, and a "machine learning" risk score constructed with the Super Learner algorithm. We hypothesized machine learning would identify high-risk individuals more efficiently (fewer persons targeted for a fixed sensitivity) and with higher sensitivity (for a fixed number targeted) than either other approach.

RESULTS

A total of 75 558 persons contributed 166 723 person-years of follow-up; 519 seroconverted. Machine learning improved efficiency. To achieve a fixed sensitivity of 50%, the risk-group strategy targeted 42% of the population, the model-based strategy targeted 27%, and machine learning targeted 18%. Machine learning also improved sensitivity. With an upper limit of 45% targeted, the risk-group strategy correctly classified 58% of seroconversions, the model-based strategy 68%, and machine learning 78%.

CONCLUSIONS

Machine learning improved classification of individuals at risk of HIV acquisition compared with a model-based approach or reliance on known risk groups and could inform targeting of prevention strategies in generalized epidemic settings.

CLINICAL TRIALS REGISTRATION

NCT01864603.

Risk assessment and identification of HIV infection among men who have sex with men: a cross-sectional study in Southwest China

OBJECTIVES

The HIV epidemic is around 7%-20% among men who have sex with men (MSM) in Southwest China. The low HIV-testing rate highlights the need for tools to identify high-risk MSM in resource-limited regions. Our aim was, therefore, to evaluate the HIV RISK Assessment Tool for HIV prediction and to characterise the primary infection among MSM in Southwest China.

DESIGN

A cross-sectional survey was conducted in Guizhou province between January and December 2018. Participants were recruited from gay communities, among whom the HIV RISK Assessment Tool was evaluated. Logistic regression was used to analyse items associated with HIV and the area under the curve (AUC) of the receiver operating curve was calculated to quantify discrimination performance.

PARTICIPANTS

1330 MSM participants, of which 83 (6.2%) tested as HIV positive.

RESULTS

A higher composite score of the tool (adjusted OR (aOR) 9.33, 95% CI 4.57 to 19.05) was independently associated with HIV infection. Items positively associated with HIV infection included having 2-5 same sex partners (aOR 2.43, 95% CI 1.28 to 4.64), always (aOR 5.93, 95% CI 1.59 to 22.13) or sometimes (aOR 4.25, 95% CI 2.09 to 8.64) having unprotected anal intercourse, taking both insertive and receptive sex roles (aOR 4.95, 95% CI 2.57 to 9.53) or only the receptive sex role (aOR 2.26, 95% CI 1.21 to 4.24). The tool showed an optimal discrimination ability (AUC=0.827), with a specificity of 0.747 and sensitivity of 0.785. Five MSM were identified with primary infection and had similar sexual risk behaviors as HIV-positive participants.

CONCLUSIONS

The HIV RISK Assessment Tool showed an overall good performance in predicting HIV risk among MSM in Guizhou province where the prevalence is still severe. This tool also demonstrated a potential to identify primary infection and is worth being promoted in resource-limited regions.

Acute and early HIV infection screening among men who have sex with men, a systematic review and meta-analysis

INTRODUCTION

Screening for acute and early HIV infections (AEHI) among men who have sex with men (MSM) remains uncommon in sub-Saharan Africa (SSA). Yet, undiagnosed AEHI among MSM and subsequent failure to link to care are important drivers of the HIV epidemic. We conducted a systematic review and meta-analysis of AEHI yield among MSM mobilized for AEHI testing; and assessed which risk factors and/or symptoms could increase AEHI yield in MSM.

METHODS

We systematically searched four databases from their inception through May 2020 for studies reporting strategies of mobilizing MSM for testing and their AEHI yield, or risk and/or symptom scores targeting AEHI screening. AEHI yield was defined as the proportion of AEHI cases among the total number of visits. Study estimates for AEHI yield were pooled using random effects models. Predictive ability of risk and/or symptom scores was expressed as the area under the receiver operator curve (AUC).

RESULTS

Twenty-two studies were identified and included a variety of mobilization strategies (eight studies) and risk and/or symptom scores (fourteen studies). The overall pooled AEHI yield was 6.3% (95% CI, 2.1 to 12.4; I2  = 94.9%; five studies); yield varied between studies using targeted strategies (11.1%; 95% CI, 5.9 to 17.6; I2  = 83.8%; three studies) versus universal testing (1.6%; 95% CI, 0.8 to 2.4; two studies). The AUC of risk and/or symptom scores ranged from 0.69 to 0.89 in development study samples, and from 0.51 to 0.88 in validation study samples. AUC was the highest for scores including symptoms, such as diarrhoea, fever and fatigue. Key risk score variables were age, number of sexual partners, condomless receptive anal intercourse, sexual intercourse with a person living with HIV, a sexually transmitted infection, and illicit drug use. No studies were identified that assessed AEHI yield among MSM in SSA and risk and/or symptom scores developed among MSM in SSA lacked validation.

CONCLUSIONS

Strategies mobilizing MSM for targeted AEHI testing resulted in substantially higher AEHI yields than universal AEHI testing. Targeted AEHI testing may be optimized using risk and/or symptom scores, especially if scores include symptoms. Studies assessing AEHI yield and validation of risk and/or symptom scores among MSM in SSA are urgently needed.

A Novel HIV-1 RNA Testing Intervention to Detect Acute and Prevalent HIV Infection in Young Adults and Reduce HIV Transmission in Kenya: Protocol for a Randomized Controlled Trial

Background

Detection and management of acute HIV infection (AHI) is a clinical and public health priority, and HIV infections diagnosed among young adults aged 18 to 39 years are usually recent. Young adults with recent HIV acquisition frequently seek care for symptoms and could potentially be diagnosed through the health care system. Early recognition of HIV infection provides considerable individual and public health benefits, including linkage to treatment as prevention, access to risk reduction counseling and treatment, and notification of partners in need of HIV testing.

Objective

The Tambua Mapema Plus study aims to (1) test 1500 young adults (aged 18-39 years) identified by an AHI screening algorithm for acute and prevalent (ie, seropositive) HIV, linking all newly diagnosed HIV-infected patients to care and offering immediate treatment; (2) offer assisted HIV partner notification services to all patients with HIV, testing partners for acute and prevalent HIV infection and identifying local sexual networks; and (3) model the potential impact of these two interventions on the Kenyan HIV epidemic, estimating incremental costs per HIV infection averted, death averted, and disability-adjusted life year averted using data on study outcomes.

Methods

A modified stepped-wedge design is evaluating the yield of this HIV testing intervention at 4 public and 2 private health facilities in coastal Kenya before and after intervention delivery. The intervention uses point-of-care HIV-1 RNA testing combined with standard rapid antibody tests to diagnose AHI and prevalent HIV among young adults presenting for care, employs HIV partner notification services to identify linked acute and prevalent infections, and follows all newly diagnosed patients and their partners for 12 months to ascertain clinical outcomes, including linkage to care, antiretroviral therapy (ART) initiation and virologic suppression in HIV-infected patients, and pre-exposure prophylaxis uptake in uninfected individuals in discordant partnerships.

Results

Enrollment started in December 2017. As of April 2020, 1374 participants have been enrolled in the observation period and 1500 participants have been enrolled in the intervention period, with 13 new diagnoses (0.95%) in the observation period and 37 new diagnoses (2.47%), including 2 AHI diagnoses, in the intervention period. Analysis is ongoing and will include adjusted comparisons of the odds of the following outcomes in the observation and intervention periods: being tested for HIV infection, newly diagnosed with prevalent or acute HIV infection, linked to care, and starting ART by week 6 following HIV diagnosis. Participants newly diagnosed with acute or prevalent HIV infection in the intervention period are being followed for outcomes, including viral suppression by month 6 and month 12 following ART initiation and partner testing outcomes.

Conclusions

The Tambua Mapema Plus study will provide foundational data on the potential of this novel combination HIV prevention intervention to reduce ongoing HIV transmission in Kenya and other high-prevalence African settings.

Trial Registration

ClinicalTrials.gov NCT03508908; https://clinicaltrials.gov/ct2/show/NCT03508908

International Registered Report Identifier (IRRID)

DERR1-10.2196/16198

HIV incidence among gay men and other men who have sex with men in 2020: where is the epidemic heading?

The goal to effectively prevent new HIV infections among gay, bisexual, and other men who have sex with men (MSM) is more challenging now than ever before. Despite declines in the late 1990s and early 2000s, HIV incidence among MSM is now increasing in many low- and high-income settings including the US, with young, adolescent, and racial/ethnic minority MSM being among those at highest risk. Potentiating HIV risks across all settings are individual-, network-, and structural-level factors such as stigma and lack of access to pre-exposure prophylaxis (PrEP) and antiretroviral treatment as prevention. To make a sustained impact on the epidemic, a concerted effort must integrate all evidence-based interventions that will most proximally decrease HIV acquisition and transmission risks, together with structural interventions that will support improved coverage and retention in care. Universal HIV treatment, increased access to HIV testing, and daily oral PrEP have emerged as integral to the prevention of HIV transmission, and such efforts should be immediately expanded for MSM and other populations disproportionately affected by HIV. Respect for human rights and efforts to combat stigma and improve access to prevention services are needed to change the trajectory of the HIV pandemic among MSM.

Development and Validation of a Personalized Social Media Platform-Based HIV Incidence Risk Assessment Tool for Men Who Have Sex With Men in China

BACKGROUND

Personalized risk assessments can help medical providers determine targeted populations for counseling and risk reduction interventions.

OBJECTIVE

The objective of this study was to develop a social media platform-based HIV risk prediction tool for men who have sex with men (MSM) in China based on an independent MSM cohort to help medical providers determine target populations for counseling and risk reduction treatments.

METHODS

A prospective cohort of MSM from Shenyang, China, followed from 2009 to 2016, was used to develop and validate the prediction model. The eligible MSM were randomly assigned to the training and validation dataset, and Cox proportional hazards regression modeling was conducted using predictors for HIV seroconversion selected by the training dataset. Discrimination and calibration were performed, and the related nomogram and social media platform-based HIV risk assessment tool were constructed.

RESULTS

The characteristics of the sample between the training dataset and the validation dataset were similar. The risk prediction model identified the following predictors for HIV seroconversion: the main venue used to find male sexual partners, had condomless receptive or insertive anal intercourse, and used rush poppers. The model was well calibrated. The bootstrap C-index was 0.75 (95% CI 0.65-0.85) in the training dataset, and 0.60 (95% CI 0.45-0.74) in the validation dataset. The calibration plots showed good agreement between predicted risk and the actual proportion of no HIV infection in both the training and validation datasets. Nomogram and WeChat-based HIV incidence risk assessment tools for MSM were developed.

CONCLUSIONS

This social media platform-based HIV infection risk prediction tool can be distributed easily, improve awareness of personal HIV infection risk, and stratify the MSM population based on HIV risk, thus informing targeted interventions for MSM at greatest risk for HIV infection.

Risk Prediction Score for HIV Infection: Development and Internal Validation with Cross-Sectional Data from Men Who Have Sex with Men in China

Receptive anal intercourse, multiple partners, condomless sex, sexually transmitted infections (STIs), and drug/alcohol addiction are familiar factors that correlate with increased human immunodeficiency virus (HIV) risk among men who have sex with men (MSM). To improve estimation to HIV acquisition, we created a composite score using questions from routine survey of 3588 MSM in Beijing, China. The HIV prevalence was 13.4%. A risk scoring tool using penalized maximum likelihood multivariable logistic regression modeling was developed, deploying backward step-down variable selection to obtain a reduced-form model. The full penalized model included 19 sexual predictors, while the reduced-form model had 12 predictors. Both models calibrated well; bootstrap-corrected c-indices were 0.70 (full model) and 0.71 (reduced-form model). Non-Beijing residence, short-term living in Beijing, illegal drug use, multiple male sexual partners, receptive anal sex, inconsistent condom use, alcohol consumption before sex, and syphilis infection were the strongest predictors of HIV infection. Discriminating higher-risk MSM for targeted HIV prevention programming using a validated risk score could improve the efficiency of resource deployment for educational and risk reduction programs. A valid risk score can also identify higher risk persons into prevention and vaccine clinical trials, which would improve trial cost-efficiency.

Prediction of extended high viremia among newly HIV-1-infected persons in sub-Saharan Africa

Objective

Prompt identification of newly HIV-infected persons, particularly those who are most at risk of extended high viremia (EHV), allows important clinical and transmission prevention benefits. We sought to determine whether EHV could be predicted during early HIV infection (EHI) from clinical, demographic, and laboratory indicators in a large HIV-1 incidence study in Africa.

Design

Adults acquiring HIV-1 infection were enrolled in an EHI study assessing acute retroviral syndrome (ARS) symptoms and viral dynamics.

Methods

Estimated date of infection (EDI) was based on a positive plasma viral load or p24 antigen test prior to seroconversion, or the mid-point between negative and positive serological tests. EHV was defined as mean untreated viral load ≥5 log₁₀ copies/ml 130–330 days post-EDI. We used logistic regression to develop risk score algorithms for predicting EHV based on sex, age, number of ARS symptoms, and CD4 and viral load at diagnosis.

Results

Models based on the full set of five predictors had excellent performance both in the full population (c-statistic = 0.80) and when confined to persons with each of three HIV-1 subtypes (c-statistic = 0.80–0.83 within subtypes A, C, and D). Reduced models containing only 2–4 predictors performed similarly. In a risk score algorithm based on the final full-population model, predictor scores were one for male sex and enrollment CD4<350 cells/mm³, and two for having enrollment viral load >4.9 log₁₀ copies/ml. With a risk score cut-point of two, this algorithm was 85% sensitive (95% CI: 76%-91%) and 61% specific (55%-68%) in predicting EHV.

Conclusions

Simple risk score algorithms can reliably identify persons with EHI in sub-Saharan Africa who are likely to sustain high viral loads if treatment is delayed. These algorithms may be useful for prioritizing intensified efforts around care linkage and retention, treatment initiation, adherence support, and partner services to optimize clinical and prevention outcomes.

Clinical and public health implications of acute and early HIV detection and treatment: a scoping review

INTRODUCTION

The unchanged global HIV incidence may be related to ignoring acute HIV infection (AHI). This scoping review examines diagnostic, clinical, and public health implications of identifying and treating persons with AHI.

METHODS

We searched PubMed, in addition to hand-review of key journals identifying research pertaining to AHI detection and treatment. We focused on the relative contribution of AHI to transmission and the diagnostic, clinical, and public health implications. We prioritized research from low- and middle-income countries (LMICs) published in the last fifteen years.

RESULTS AND DISCUSSION

Extensive AHI research and limited routine AHI detection and treatment have begun in LMIC. Diagnostic challenges include ease-of-use, suitability for application and distribution in LMIC, and throughput for high-volume testing. Risk score algorithms have been used in LMIC to screen for AHI among individuals with behavioural and clinical characteristics more often associated with AHI. However, algorithms have not been implemented outside research settings. From a clinical perspective, there are substantial immunological and virological benefits to identifying and treating persons with AHI - evading the irreversible damage to host immune systems and seeding of viral reservoirs that occurs during untreated acute infection. The therapeutic benefits require rapid initiation of antiretrovirals, a logistical challenge in the absence of point-of-care testing. From a public health perspective, AHI diagnosis and treatment is critical to: decrease transmission via viral load reduction and behavioural interventions; improve pre-exposure prophylaxis outcomes by avoiding treatment initiation for HIV-seronegative persons with AHI; and, enhance partner services via notification for persons recently exposed or likely transmitting.

CONCLUSIONS

There are undeniable clinical and public health benefits to AHI detection and treatment, but also substantial diagnostic and logistical barriers to implementation and scale-up. Effective early ART initiation may be critical for HIV eradication efforts, but widespread use in LMIC requires simple and accurate diagnostic tools. Implementation research is critical to facilitate sustainable integration of AHI detection and treatment into existing health systems and will be essential for prospective evaluation of testing algorithms, point-of-care diagnostics, and efficacious and effective first-line regimens.

Incorporating Acute HIV Screening into Routine HIV Testing at Sexually Transmitted Infection Clinics, and HIV Testing and Counseling Centers in Lilongwe, Malawi

Background and Objectives:

Integrating acute HIV-infection (AHI) testing into clinical settings is critical to prevent transmission, and realize potential treatment-as-prevention benefits. We evaluated acceptability of AHI testing and compared AHI prevalence at sexually transmitted infection (STI) clinics and HIV testing and counseling (HTC) clinics in Lilongwe, Malawi.

Methods:

We conducted HIV RNA testing for HIV-seronegative patients visiting STI and HTC clinics. AHI was defined as positive RNA and negative/discordant rapid antibody tests. We evaluated demographic, behavioral, and transmission-risk differences between STI and HTC patients and assessed performance of a risk-score for targeted screening.

Results:

Nearly two-thirds (62.8%, 9280/14,755) of eligible patients consented to AHI testing. We identified 59 persons with AHI (prevalence = 0.64%)–a 0.9% case-identification increase. Prevalence was higher at STI [1.03% (44/4255)] than at HTC clinics [0.3% (15/5025), P < 0.01], accounting for 2.3% of new diagnoses vs 0.3% at HTC clinic. Median viral load (VL) was 758,050 copies per milliliter; 25% (15/59) had VL ≥10,000,000 copies per milliliter. Median VL was higher at STI (1,000,000 copies/mL) compared with HTC (153,125 copies/mL, P = 0.2). Among persons with AHI, those tested at STI clinics were more likely to report genital sores compared with those tested at HTC clinics (54.6% vs 6.7%, P < 0.01). The risk score algorithm performed well in identifying persons with AHI at HTC clinics (sensitivity = 73%, specificity = 89%).

Conclusions:

The majority of patients consented to AHI testing. AHI prevalence was substantially higher in STI clinics than HTC clinics. Remarkably high VLs and concomitant genital scores demonstrate the potential for transmission. Universal AHI screening at STI clinics, and targeted screening at HTC centers, should be considered.

Targeted screening of at-risk adults for acute HIV-1 infection in sub-Saharan Africa

BACKGROUND

Patients with acute HIV-1 infection (AHI) have elevated infectivity, but cannot be diagnosed using antibody-based testing. Approaches to screen patients for AHI are urgently needed to enable counselling and treatment to reduce onward transmission.

METHODS

We pooled data from four African studies of high-risk adults that evaluated symptoms and signs compatible with acute retroviral syndrome and tested for HIV-1 at each visit. AHI was defined as detectable plasma viral load or p24 antigen in an HIV-1-antibody-negative patient who subsequently seroconverted. Using generalized estimating equation, we identified symptoms, signs, and demographic factors predictive of AHI, adjusting for study site. We assigned a predictor score to each statistically significant predictor based on its beta coefficient, summing predictor scores to calculate a risk score for each participant. We evaluated the performance of this algorithm overall and at each site.

RESULTS

We compared 122 AHI visits with 45 961 visits by uninfected patients. Younger age (18-29 years), fever, fatigue, body pains, diarrhoea, sore throat, and genital ulcer disease were independent predictors of AHI. The overall area under the receiver operating characteristics curve (AUC) for the algorithm was 0.78, with site-specific AUCs ranging from 0.61 to 0.89. A risk score of at least 2 would indicate AHI testing for 5-50% of participants, substantially decreasing the number needing testing.

CONCLUSION

Our targeted risk score algorithm based on seven characteristics reduced the number of patients needing AHI testing and had good performance overall. We recommend this risk score algorithm for use by HIV programs in sub-Saharan Africa with capacity to test high-risk patients for AHI.

The HIV treatment cascade in acutely infected people: informing global guidelines

PURPOSE OF REVIEW

Acute and early HIV (AHI) is a pivotal time during HIV infection, yet there remain major shortfalls in diagnosis, linkage to care, and antiretroviral therapy (ART) initiation during AHI. We introduce an AHI-specific cascade, review recent evidence pertaining to the unique challenges of AHI, and discuss strategies for improving individual and public health outcomes.

RECENT FINDINGS

Presentation during AHI is common. Expanding use of fourth-generation testing and pooled nucleic acid amplification testing has led to improved AHI detection in resource-wealthy settings. Technologies capable of AHI diagnosis are rare in resource-limited settings; further development of point-of-care devices and utilization of targeted screening is needed. Rapid ART initiation during AHI limits reservoir seeding, preserves immunity, and prevents transmission. Reporting of AHI cascade outcomes is limited, but new evidence suggests that impressive rates of diagnosis, linkage to care, rapid ART initiation, and viral suppression can be achieved.

SUMMARY

With advancements in AHI diagnostics and strong evidence for the therapeutic and prevention benefits of ART initiated during AHI, improving AHI cascade outcomes is both crucial and feasible. HIV guidelines should recommend diagnostic algorithms capable of detecting AHI and prescribe rapid, universal ART initiation during AHI.

How can we better identify early HIV infections?

PURPOSE OF REVIEW

Detection of early HIV infections (EHIs), including acute HIV infection (AHI), is important for individual health, prevention of HIV transmission, and measurement of HIV incidence. We describe markers of EHI, diagnostic strategies for detecting these markers, and ways to incorporate these strategies into diagnostic and HIV incidence algorithms.

RECENT FINDINGS

For individual diagnosis in the USA and Europe, laboratory-based diagnostic algorithms increasingly incorporate fourth-generation HIV antigen tests, allowing for earlier detection. In some sub-Saharan African settings, symptom-based screening is being explored to identify subsets of persons at high risk for AHI. Point-of-care diagnostics designed for AHI detection are in the pipeline and, if validated, represent an opportunity for real-time AHI diagnosis. At the population level, multiassay algorithms are promising new strategies for estimating HIV incidence on the basis of several assays applied to cross-sectional samples. These algorithms can be developed to optimize performance, in addition to cost and logistical considerations.

SUMMARY

There are important recent advances in detection of EHIs at the individual and population levels. Applying optimal combinations of tests in diagnostic and HIV incidence algorithms is urgently needed to support the multiple goals derived from enhanced detection and discrimination of EHIs.

Brief Report: Health-Seeking Behavior and Symptoms Associated With Early HIV Infection: Results From a Population-Based Cohort in Southern Malawi

HIV transmission is most likely to occur during the first few months after infection, yet few cases are identified during this period. Using a population-based cohort of young Malawian women, we identify the distinct symptomology and health-seeking behavior marking early HIV infection by comparing it with periods of seronegativity and chronic infection. During early HIV infection, women are more likely to report malaria-like symptoms and visit clinics for malaria care. In malaria-endemic contexts, where acute HIV symptoms are commonly mistaken for malaria, early diagnostic HIV testing and counseling should be integrated into health care settings where people commonly seek treatment for malaria.

Emerging themes for sensitivity training modules of African healthcare workers attending to men who have sex with men: a systematic review

Sensitivity training of front-line African health care workers (HCWs) attending to men who have sex with men (MSM) is actively promoted through national HIV prevention programming in Kenya. Over 970 Kenyan-based HCWs have completed an eight-modular online training free of charge (http://www.marps-africa.org) since its creation in 2011. Before updating these modules, we performed a systematic review of published literature of MSM studies conducted in sub-Saharan Africa (sSA) in the period 2011-2014, to investigate if recent studies provided: important new knowledge currently not addressed in existing online modules; contested information of existing module topics; or added depth to topics covered already. We used learning objectives of the eight existing modules to categorise data from the literature. If data could not be categorised, new modules were suggested. Our review identified 142 MSM studies with data from sSA, including 34 studies requiring module updates, one study contesting current content, and 107 studies reinforcing existing module content. ART adherence and community engagement were identified as new modules. Recent MSM studies conducted in sSA provided new knowledge, contested existing information, and identified new areas of MSM service needs currently unaddressed in the online training.

Impact of nucleic acid testing relative to antigen/antibody combination immunoassay on the detection of acute HIV infection

OBJECTIVE

To assess the addition of HIV nucleic acid testing (NAT) to fourth-generation (4thG) HIV antigen/antibody combination immunoassay in improving detection of acute HIV infection (AHI).

METHODS

Participants attending a major voluntary counseling and testing site in Thailand were screened for AHI using 4thG HIV antigen/antibody immunoassay and sequential less sensitive HIV antibody immunoassay. Samples nonreactive by 4thG antigen/antibody immunoassay were further screened using pooled NAT to identify additional AHI. HIV infection status was verified following enrollment into an AHI study with follow-up visits and additional diagnostic tests.

RESULTS

Among 74 334 clients screened for HIV infection, HIV prevalence was 10.9% and the overall incidence of AHI (N = 112) was 2.2 per 100 person-years. The inclusion of pooled NAT in the testing algorithm increased the number of acutely infected patients detected, from 81 to 112 (38%), relative to 4thG HIV antigen/antibody immunoassay. Follow-up testing within 5 days of screening marginally improved the 4thG immunoassay detection rate (26%). The median CD4 T-cell count at the enrollment visit was 353 cells/μl and HIV plasma viral load was 598 289 copies/ml.

CONCLUSION

The incorporation of pooled NAT into the HIV testing algorithm in high-risk populations may be beneficial in the long term. The addition of pooled NAT testing resulted in an increase in screening costs of 22% to identify AHI: from $8.33 per screened patient to $10.16. Risk factors of the testing population should be considered prior to NAT implementation given the additional testing complexity and costs.

Development and validation of the San Diego Early Test Score to predict acute and early HIV infection risk in men who have sex with men

BACKGROUND

Although men who have sex with men (MSM) represent a dominant risk group for human immunodeficiency virus (HIV), the risk of HIV infection within this population is not uniform. The objective of this study was to develop and validate a score to estimate incident HIV infection risk.

METHODS

Adult MSM who were tested for acute and early HIV (AEH) between 2008 and 2014 were retrospectively randomized 2:1 to a derivation and validation dataset, respectively. Using the derivation dataset, each predictor associated with an AEH outcome in the multivariate prediction model was assigned a point value that corresponded to its odds ratio. The score was validated on the validation dataset using C-statistics.

RESULTS

Data collected at a single HIV testing encounter from 8326 unique MSM were analyzed, including 200 with AEH (2.4%). Four risk behavior variables were significantly associated with an AEH diagnosis (ie, incident infection) in multivariable analysis and were used to derive the San Diego Early Test (SDET) score: condomless receptive anal intercourse (CRAI) with an HIV-positive MSM (3 points), the combination of CRAI plus ≥5 male partners (3 points), ≥10 male partners (2 points), and diagnosis of bacterial sexually transmitted infection (2 points)-all as reported for the prior 12 months. The C-statistic for this risk score was >0.7 in both data sets.

CONCLUSIONS

The SDET risk score may help to prioritize resources and target interventions, such as preexposure prophylaxis, to MSM at greatest risk of acquiring HIV infection. The SDET risk score is deployed as a freely available tool at http://sdet.ucsd.edu.

Acute HIV-1 infection is as common as malaria in young febrile adults seeking care in coastal Kenya

Background:

Febrile adults are usually not tested for acute HIV-1 infection (AHI) in Africa. We assessed a strategy to diagnose AHI among young adult patients seeking care.

Methods:

Young adults (<30 years) who met predefined AHI criteria at care seeking, including fever, sexually transmitted disease symptoms, diarrhoea, body pains or multiple partners were referred from five pharmacies and screened at five health facilities. Prevalent HIV-1 was diagnosed by nationally recommended serial rapid HIV-1 testing. Willing HIV-1-negative patients were evaluated for AHI, defined as a positive p24 antigen test, and subsequent seroconversion or RNA detection. Febrile patients evaluated for AHI were also screened for malaria using a rapid test, with PCR confirmation of positives.

Results:

In 3602 adults seeking care, overall HIV-1 prevalence was 3.9%: 7.6% (68/897) among patients meeting AHI criteria vs. 2.6% (71/2705) among those who did not (P < 0.001). AHI was diagnosed in five of 506 HIV-1-negative or discordant patients who met AHI risk criteria and were completely evaluated [prevalence 1.0%, 95% confidence interval (CI) 0.3–2.3%]. Of these five AHI cases, four were diagnosed among the 241 patients with fever (prevalence 1.7%, 95% CI 0.5–4.2%), vs. one among 265 non-febrile patients (prevalence 0.4%, 95% CI 0.0–2.0%, P = 0.1). Malaria was confirmed by PCR in four (1.7%) of the 241 febrile patients.

Conclusion:

AHI was as common as confirmed malaria in young febrile adults seeking care. An AHI detection strategy targeting young febrile adults seeking care at pharmacies and health facilities is feasible and should be considered as an HIV-prevention strategy in high-transmission settings.