A Simple Risk Prediction Algorithm for HIV Transmission: Results from HIV Prevention Trials in KwaZulu Natal, South Africa (2002-2012)

Geospatial correlations and variations in child mortality and stunting in South Africa: Evaluating distal vs structural determinants

South Africa has one of the highest child mortality and stunting rates in the world. Flexible geoadditive models were used to investigate the geospatial variations in child mortality and stunting in South Africa. We used consecutive rounds of national surveys (2008-2017). The child mortality declined from 31 % to 24 % over time. Lack of medical insurance, black ethnicity, low-socioeconomic conditions, and poor housing conditions were identified as the most significant correlates of child mortality. The model predicted degrees of freedom which was estimated as 19.55 (p < 0.001), provided compelling evidence for sub-geographical level variations in child mortality which ranged from 6 % to 35 % across the country. Population level impact of the distal characteristics on child mortality and stunting exceeded that of other risk factors. Geospatial analysis can help in monitoring trends in child mortality over time and in evaluating the impact of health interventions.

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.

Developing HIV risk prediction tools in four African settings

OBJECTIVE

HIV risk prediction tools are a critical component of efforts to end the HIV pandemic. We aimed to create and validate tools for identifying individuals at highest risk of prevalent and incident HIV in an African setting.

METHODS

We used Logistic regression and Poisson regression to determine risk factors for HIV prevalence and incidence in a multi-country HIV vaccine trial preparedness cohort study among individuals at high risk of HIV, and used the identified factors to create and validate tools that predict HIV risk. We also assessed the performance of the VOICE risk score in predicting HIV incidence among women in the cohort.

RESULTS

The prevalent HIV prediction tool created had good predictive ability [area under the curve (AUC) = 0.70, 95% CI 0.66-0.74]. It included the following participant variables: age, sex, recreational drug use, unprotected male-to-male anal sex, a sexual partner who had other partners, transactional sex and having a partner who was a long-distance truck driver/miner. It was not possible to create a valid HIV incidence prediction tool. Participants with high VOICE risk scores (≥7) had slightly higher HIV incidence but this tool performed poorly within our study (AUC = 0.58, 95% CI 0.51-0.64: Harrell's concordance index = 0.59).

CONCLUSION

We created a prevalent HIV prediction tool that could be used to increase efficiency in diagnosis of HIV and linkage to care in sub-Saharan Africa. Existing incident HIV prediction tools may need modification to include context-specific predictors such as calendar period, participant occupation, study site, before adoption in settings different from those in which they were developed.

Impact of marital status on risk of HIV in South Africa (2002-2017)

After decades of research, South African women continue to have the highest burden of HIV in the world. The aim of the study is to investigate the impact of marital status on HIV using population and community-level data sources. We included data 13,469 and 5682 South African women who participated to South African HIV behavioural surveys in 2002-2012 and 2017, respectively. HIV prevalence and incidence rates were significantly higher among single/not-cohabiting women compared to those married/cohabiting with a partner in all survey participants (adjusted odds ratio (aOR): 1.60 and 1.58 in 2002-2012 and 2017, respectively). A quarter of the HIV diagnoses were attributed to those who were single/not-cohabiting women (population attributable risk (PAR%): 25% and 24%, respectively). More than 40% of the HIV infections were exclusively associated with single women (PAR%: 42%, 95% CI: 33%, 51%). Our results provided strong evidence for the profound impact of marital status on excess HIV infection rates. Targeting and reaching single/not-cohabiting women with multiple risk factors can potentially play a crucial role in the trajectories of the epidemic.

Geospatial variations in socioeconomic conditions and health outcomes in COVID-19 era: insights from South Africa (2020-2022)

South Africa also has the highest burden of coronavirus disease 2019 (COVID-19) related comorbidities in Africa. We aimed to quantify the temporal and geospatial changes in unemployment, food insecurity, and their combined impact on depressive symptoms among South Africans who participated into several rounds of national surveys. We estimated the population-attributable risk percent (P A R % ) for the combinations of the risk factors after accounting for their correlation structure in multifactorial setting. Our study provided compelling evidence for immediate and severe effect of the pandemic where 60% of South Africans reported household food insecurity or household hunger, shortly after the pandemic emerged in 2020. Despite the grants provided by the government, these factors were also identified as the most influential risk factors (adjusted odds ratios (aORs) ranged from 2.06 to 3.10, p < 0.001) for depressive symptoms and collectively associated with 62% and 53% of the mental health symptoms in men and women, respectively. Similar pattern was observed among pregnant women and 41% of the depressive symptoms were exclusively associated with those who reported household hunger. However, aORs associated with the concerns around pandemic and vaccine were mostly not significant and ranged from 1.12 to 1.26 which resulted substantially lower impacts on depressive symptoms (PAR%:7%-and-14%). Our findings suggest that South Africa still has unacceptably high rates of hunger which is accelerated during the pandemic. These results may have significant clinical and epidemiological implications and may also bring partial explanation for the low vaccine coverage in the country, as priorities and concerns are skewed towards economic concerns and food insecurity.

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.

Modeling the Probability of HIV Infection over Time in High-Risk Seronegative Participants Receiving Placebo in Five Randomized Double-Blind Placebo-Controlled HIV Pre-Exposure Prophylaxis Trials: A Patient-Level Pooled Analysis

The World Health Organization recommends pre-exposure prophylaxis (PrEP) for individuals at substantial risk of HIV infection. The aim of this analysis is to quantify the individual risk of HIV infection over time, using a large database of high-risk individuals (n = 5583). We used data from placebo recipients in five phase III PrEP trials: iPrEx, conducted in men who have sex with men and transgender women; VOICE, conducted in young women at high sexual risk; Partners PrEP, conducted in HIV serodiscordant heterosexual couples; TDF2, conducted in high-risk heterosexual men and women; and BTS, conducted in persons who inject drugs. The probability of HIV infection over time was estimated using NONMEM7.4. We identified predictors of HIV risk and found a substantial difference in the risk of infection among and within trial populations, with each study including a mix of low, moderate, and high-risk individuals (p < 0.05). Persons who were female at birth were at a higher risk of HIV infection than people who were male at birth. Final models were integrated in a tool that can assess person-specific risk and simulate cumulative HIV risk over time. These models can be used to optimize future PrEP clinical trials by identifying potential participants at highest risk.

Lifestyle and adherence to antiretrovirals in people with HIV in the COVID-19 pandemic

OBJECTIVES

to assess the lifestyle and adherence to antiretrovirals in people living with HIV (PLHIV) in the COVID-19 pandemic.

METHODS

a cross-sectional study, through telephone interview to 150 patients, using a sociodemographic, epidemiological and clinical form, and questionnaires to assess lifestyle profile and adherence to antiretrovirals. Statistics analysis used Fisher's exact test, odds ratio and 95% confidence interval.

RESULTS

most patients had a satisfactory lifestyle (121; 80.7%) and adequate adherence to antiretrovirals (133; 88.7%). All were in social isolation, without follow-up appointments, with access to the health service only to receive antiretrovirals, and 16 (10.7%) had COVID-19 infection. Evangelicals (p=0.002), Spiritists (p=0.045), patients using atazanavir (p=0.0001) and ritonavir (p=0.002) had a more unsatisfactory lifestyle. Adherence to antiretrovirals was more inadequate in female patients (p=0.009), with two (p=0.004) and three or more children (p=0.006), retired (p=0.029), with serodiscordant partner (p=0.046) and diagnosis time of 5 to 10 years (p=0.027).

CONCLUSIONS

the most PLHIV had a satisfactory lifestyle and adequate adherence to antiretrovirals, but some groups needed intervention to improve medication adherence and lifestyle.

A risk measurement tool for targeted HIV prevention measures amongst young pregnant and lactating women in South Africa

Background

We aimed to develop and validate a tool to identify which pregnant/lactating young South African women (≤ 24 years) are at risk of HIV infection.

Methods

Data from three national South African Prevention of Mother-to-Child Transmission (PMTCT) evaluations were used to internally validate three HIV acquisition risk models for young postpartum women. We used univariate and multivariable logistic regression analysis to determine which risk factors were significant. Model coefficients were rounded and stratified into risk groups and the area under the receiver operating curve (AUROC) was computed. Models were developed to determine which risk factors provided the most predictive accuracy whilst remining clinically meaningful.

Results

Data from 9 456 adult and 4 658 young pregnant and lactating women were included in the development and validation data sets, respectively. The optimal model included the following risk factors: age (20–24 years old), informal house structure, two or more pregnancies, mothers who had knowledge of when they received their last HIV test result, no knowledge of the infant’s father’s HIV status, no knowledge of breastfeeding as a mode of MTCT and knowledge of PMTCT programme. The mean AUROC was 0.71 and 0.72 in the development and validation datasets respectively. The optimum cut off score was ≥ 27, having 84% sensitivity, 44% specificity, and identifying 44% of high-risk women eligible for PrEP.

Conclusion

The optimal model to be used as a possible risk scoring tool to allow for early identification of those pregnant/lactating women most at-risk of HIV acquisition included both statistically as well as clinically meaningful risk factors. A field-based study is needed to test and validate the effectiveness of this targeted approach.

Age-Restriction of a Validated Risk Scoring Tool Better Predicts HIV Acquisition in South African Women: CAPRISA 004

We examined the predictive ability of the VOICE risk screening tool among adolescent girls and young women at heightened HIV risk in urban and peri-urban Kwa-Zulu-Natal, South Africa. Using participant data from CAPRISA 004’s control arm (N = 444), we applied the initial VOICE risk screening score (IRS), a modified risk score (MRS) based on predictive and non-predictive variables in our data, and age-restricted (AIRS and AMRS, respectively). We estimated incidence rates, 95% confidence bounds, sensitivity, specificity, negative and positive predictive values and area under the curve (AUC). The sample’s HIV incidence rate was 9.1/100 Person-Years [95% CI 6.9–11.7], resulting from 60 seroconversions (60/660.7 Person-Years). The IRS’ ≥ 8 cutpoint produced moderate discrimination [AUC = 0.66 (0.54–0.74), sensitivity = 63%, specificity = 57%]. Restricting to age < 25 years improved the score’s predictive ability (AIRS: AUC = 0.69, AMRS: AUC = 0.70), owing mainly to male partner having other partners and HSV-2. The risk tool predicted HIV acquisition at a higher cutpoint in this sample than in the initial VOICE analysis. After age-stratification, fewer variables were needed for maintaining score’s predictiveness. In this high incidence setting, risk screening may still improve the efficiency or effectiveness of prevention counseling services. However, PrEP should be offered to all prevention-seeking individuals, regardless of risk ascertainment.

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.

Risk scores for predicting HIV incidence among adult heterosexual populations in sub-Saharan Africa: a systematic review and meta-analysis

Introduction

Several HIV risk scores have been developed to identify individuals for prioritized HIV prevention in sub‐Saharan Africa. We systematically reviewed HIV risk scores to: (1) identify factors that consistently predicted incident HIV infection, (2) review inclusion of community‐level HIV risk in predictive models and (3) examine predictive performance.

Methods

We searched nine databases from inception until 15 February 2021 for studies developing and/or validating HIV risk scores among the heterosexual adult population in sub‐Saharan Africa. Studies not prospectively observing seroconversion or recruiting only key populations were excluded. Record screening, data extraction and critical appraisal were conducted in duplicate. We used random‐effects meta‐analysis to summarize hazard ratios and the area under the receiver‐operating characteristic curve (AUC‐ROC).

Results

From 1563 initial search records, we identified 14 risk scores in 13 studies. Seven studies were among sexually active women using contraceptives enrolled in randomized‐controlled trials, three among adolescent girls and young women (AGYW) and three among cohorts enrolling both men and women. Consistently identified HIV prognostic factors among women were younger age (pooled adjusted hazard ratio: 1.62 [95% confidence interval: 1.17, 2.23], compared to above 25), single/not cohabiting with primary partners (2.33 [1.73, 3.13]) and having sexually transmitted infections (STIs) at baseline (HSV‐2: 1.67 [1.34, 2.09]; curable STIs: 1.45 [1.17; 1.79]). Among AGYW, only STIs were consistently associated with higher incidence, but studies were limited (n = 3). Community‐level HIV prevalence or unsuppressed viral load strongly predicted incidence but was only considered in 3 of 11 multi‐site studies. The AUC‐ROC ranged from 0.56 to 0.79 on the model development sets. Only the VOICE score was externally validated by multiple studies, with pooled AUC‐ROC 0.626 [0.588, 0.663] (I²: 64.02%).

Conclusions

Younger age, non‐cohabiting and recent STIs were consistently identified as predicting future HIV infection. Both community HIV burden and individual factors should be considered to quantify HIV risk. However, HIV risk scores had only low‐to‐moderate discriminatory ability and uncertain generalizability, limiting their programmatic utility. Further evidence on the relative value of specific risk factors, studies populations not restricted to “at‐risk” individuals and data outside South Africa will improve the evidence base for risk differentiation in HIV prevention programmes.

PROSPERO Number

CRD42021236367

Quantifying Vulnerabilities of Single Women and Sexually Transmitted Infections in South Africa (2002-2016): Is It Getting Better (or Worse)?

Curable sexually transmitted infections (STIs) affect millions of people across the world. Besides unacceptably high HIV rates, South Africa also has the highest burden of STIs in the world. The aim of the study was to investigate temporal changes in STI incidence rates using the data from ~ 10,000 women who enrolled in several HIV prevention trials, KwaZulu-Natal, South Africa (2002-2016). We particularly focused on the changes in distribution of the most influential factors and their population-level impacts on STI incidence rates over time. Characteristics of the women were compared across the study periods: 2002-2004, 2005-2008, 2009-2011, and 2012-2016. Besides multivariable Cox regression models, population attributable risks were calculated for the significant factors. Despite the significant progress in prevention, testing, and treatment programs, infection rates increased substantially from 13.6 to 20.0 per 100 person-year over the study period. Our findings provided a compelling evidence for single/non-cohabiting South African women to be the most vulnerable population who consistently and substantially contributed to increasing STI rates during the 15 years of study duration (PAR%: 44%-47%). We also highlighted the impact of women's lack of knowledge related to their partner, using injectable contraceptives, less parity, and baseline STI positivity which were increased substantially over time. Our findings suggest that a significant proportion of STIs could be prevented by targeting single/non-cohabiting. These results may provide guidance in developing more effective STI prevention programs by targeting women at highest risk of infections and delivering more realistic messages.

Impact of recurrent sexually transmitted infections on HIV seroconversion: Results from multi-state frailty models

After several decades of research, South Africa is still considered to be the epicentre of HIV epidemic. The country also has the highest burden of sexually transmitted infections (STIs) which have been frequently linked to increasing rates of HIV transmission due to biological and behavioural associations between these two pathogeneses. We investigated the cumulative impact of recurrent STIs on subsequent HIV seroconversion among a cohort of South African women. We used the 'frailty' models which can account for the heterogeneity due to the recurrent STIs in a longitudinal setting. The lowest HIV incidence rate was 5.0/100 person-year among women who had no baseline STI and remained negative during the follow-up. This estimate was three times higher among those who had recurrent STIs in the follow-up period regardless of their STI status at baseline (15.8 and 14.0/100 person-year for women with and without STI diagnosis at baseline, respectively). Besides younger age and certain partnership characteristics, our data provided compelling evidence for the impact of recurrent STI. diagnoses on increasing rates of HIV. At the population-level, 65% of HIV infections collectively associated with recurrent STIs. These results have significant clinical and epidemiological implications and may play critical role in the trajectory of the infections in the region.

Characteristics of women who use hormonal contraceptives and their population-level impacts on HIV infection in a cohort of South African women (2002-2016)

PURPOSE

Women's choice of contraceptive can have broader implications and may play a significant role in shaping their sexual practices. We aim to identify the characteristics of women who use hormonal injectables and those at high-risk of HIV infection. We also quantify the population-level impact of their shared characteristics on HIV incidence rates.

MATERIALS AND METHODS

Data from 9948 women who enrolled in six-HIV prevention trials conducted in South Africa (2002-2016) were included. We used logistic and Cox regression models and estimated the population-level impact of the use of injectables on HIV incidence in the multifactorial-model setting.

RESULTS

Using hormonal injectables were associated with increased risk of HIV infection (adjusted Hazard ratio (aHR):1.51, 95% CI:1.22, 1.86). At the population level less than 20% of the infections were associated with injectable contraceptives among younger women (i.e., less than 35 years of age). Factors including being single/not-cohabiting, using condoms at last sex, partner-related factors, and STI diagnosis were all identified as shared characteristics of women who preferred using hormonal injectables and those at high-risk of HIV infection. At the population level, these factors were associated with more than 50% of the infections among women younger than 35 years of age.

CONCLUSIONS

Our analysis presented evidence for the overlapping characteristics of the women who used hormonal injectables and those at high-risk of HIV infection. These findings reinforce the importance of comprehensive contraceptive counselling to women about the importance of dual protection, such as male condoms and hormonal contraceptives use.

Temporal and spatial monitoring of HIV prevalence and incidence rates using geospatial models: Results from South African women

Generalized additive models (GAMs) were used to predict non-linear distributions of HIV prevalence and incidence based on semiparametric methods. The GAMs also provide smooth intensity maps by projecting the predicted HIV prevalence (or incidence) into the contour maps. Two sets of geo-coded data sources were used: (1) population-based cross-sectional data from 10,928 women who participated in four HIV behavioral surveys (2002-2017), (2) clinic-based longitudinal data from 7,557 women who resided in KwaZulu-Natal (2002-2016). Model estimated degrees of freedoms were 15.84,12.17,7.64 and 15.08 (2002-2012), indicating substantial spatial variations in HIV prevalence overtime. At localized-level these HIV incidence ranged from 15 to 18 per 100 person-year and scattered across the relatively homogeneous area within less than 100 km radius. These significant quantitative evidence were further supported by continuous-scale intensity maps. Our findings provided empirical and visual evidence for the changing face of the epidemic in South Africa using geospatial methods.

Quantifying geographical diversity in sexually transmitted infections using population attributable risk: results from HIV prevention trials in South Africa

In this study, we investigated spatial diversities of sexually transmitted infections (STIs) and quantified their impacts on the STIs using population attributable risk (PAR%). The study population was 7,557 women who participated in several HIV prevention trials from KwaZulu-Natal, South Africa. Our results provide compelling evidence for substantial geographical diversities on STI incidence rates in the region. Their population-level impacts on the STIs exceeded the combined impacts of the individual risk factors considered in this study (PAR%: 41% (<25 years), 52% (25-34 years) and 34% (35+ years). When all these factors are considered together, PAR% was the highest among younger women (PAR%: 67%, 82% and 50% for <25, 25-34 and 35+ years old respectively). Results from our study will bring greater insight into the previous research by increasing our understanding of the impacts of the sub-geographical level variations of STI prevalence and incidence rates in the region.

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.

Narrative sexual histories and perceptions of HIV risk among young women taking PrEP in southern Africa: Findings from a novel participatory method

Pre-exposure prophylaxis (PrEP) is an important HIV prevention method for adolescent girls and young women (AGYW) in Africa, who are at heightened risk of HIV infection. HIV risk perception is generally a powerful motivator for adoption of HIV prevention behaviours, including PrEP use. While HIV risk perceptions have been evaluated using quantitative measures, these seldom capture how individuals conceptualize and understand risk within local frameworks of meaning. More nuanced understanding may come from qualitative approaches that map these perceptions across the trajectory of sexual histories. Between 2016 and 2018, we implemented a novel participatory method to investigate risk perceptions in interviews with 32 participants in HPTN 082, a study of AGYW's use of PrEP in South Africa and Zimbabwe. Timelines were used to record narrative sexual histories and perceived HIV risk for each relationship. We found that women assessed HIV risk primarily based on their partners' personal qualities and behaviour (especially relating to infidelity); their subjective experience of being treated respectfully; and the practice of perceived 'risk reduction', including younger partners and condoms, even if used inconsistently. A narrative timeline approach grounded in an understanding of young women's sexual histories may increase critical reflection about HIV risks and facilitate risk-reduction counselling with this group.

Contraceptives and sexual behaviours in predicting pregnancy rates in HIV prevention trials in South Africa: Past, present and future implications

OBJECTIVE

Despite all efforts, high pregnancy rates are often reported in HIV biomedical intervention trials conducted in African countries. We therefore aimed to develop a pregnancy risk scoring algorithm for targeted recruitment and screening strategies among a cohort of women in South Africa.

METHODS

The study population was ~ 10,000 women who enrolled in one of the six biomedical intervention trials conducted in KwaZulu Natal, South Africa. Cox regression models were used to create a pregnancy risk scoring algorithm which was internally validated using standard statistical measures.

RESULTS

Five factors were identified as significant predictors of pregnancy incidence:<25 years old, not using injectable contraceptives, parity (<3), being single/not cohabiting and having ≥ 2 sexual partners in the past three months. Women with total scores of 21-24, 25-35 and 36+ were classified as being at "moderate", "high", "severe" risk of pregnancy. Sensitivity of the development and validation models were reasonably high (sensitivity 76% and 74% respectively).

CONCLUSION

Our risk scoring algorithm can identify and alert researchers to women who need additional non-routine pregnancy assessment and counselling, with statistically acceptable accuracy and robustness.

Predicting the diagnosis of HIV and sexually transmitted infections among men who have sex with men using machine learning approaches

OBJECTIVES

We aimed to develop machine learning models and evaluate their performance in predicting HIV and sexually transmitted infections (STIs) diagnosis based on a cohort of Australian men who have sex with men (MSM).

METHODS

We collected clinical records of 21,273 Australian MSM during 2011-2017. We compared accuracies for predicting HIV and STIs (syphilis, gonorrhoea, chlamydia) diagnosis using four machine learning approaches against a multivariable logistic regression (MLR) model.

RESULTS

Machine learning approaches consistently outperformed MLR. Gradient boosting machine (GBM) achieved the highest area under the receiver operator characteristic curve for HIV (76.3%) and STIs (syphilis, 85.8%; gonorrhoea, 75.5%; chlamydia, 68.0%), followed by extreme gradient boosting (71.1%, 82.2%, 70.3%, 66.4%), random forest (72.0%, 81.9%, 67.2%, 64.3%), deep learning (75.8%, 81.0%, 67.5%, 65.4%) and MLR (69.8%, 80.1%, 67.2%, 63.2%). GBM models demonstrated the ten greatest predictors collectively explained 62.7-73.6% of variations in predicting HIV/STIs. STIs symptoms, past syphilis infection, age, time living in Australia, frequency of condom use with casual male sexual partners during receptive anal sex and the number of casual male sexual partners in the past 12 months were most commonly identified predictors.

CONCLUSIONS

Machine learning approaches are advantageous over multivariable logistic regression models in predicting HIV/STIs diagnosis.

Correlates of Sexually Transmitted Infections Among South African Women Using Individual- and Community-Level Factors: Results from Generalized Additive Mixed Models

South Africa has the highest burden of human immunodeficiency virus (HIV) infections in the world. There is also growing evidence that an individual's risk of contracting HIV is increased by the presence of other sexually transmitted infections (STIs). The primary objective of this study was to examine the association between the prevalence of STIs in a cohort of South African women who enrolled in HIV prevention trials (2002-2012). The current study linked the individual factors with the community-level characteristics using geo-referencing. These multi-level data were analyzed in generalized additive mixed models settings. In the multivariate logistic regression model, younger age (odds ratio [OR] 4.30, 95% CI 3.20, 5.77 and OR 2.72, 95% CI 2.02, 3.66 for age < 25 and 25-29, respectively); being single/not cohabiting (OR 4.57, 95% CI 3.18, 6.53), two + sex partners (OR 1.46, 95% CI 1.18,1.80); parity < 2 (OR 2.04, 95% CI 1.53, 2.72), parity = 2 (OR 1.85, 95% CI 1.37, 2.48), and using injectables (contraceptive) (OR 1.53, 95% CI 1.13, 2.06) were all significantly associated with increased prevalence of STIs. Women who resided in the communities with high proportions of female headed-households were also significantly at higher risk for STIs (OR 1.20, p = .0025). Because these factors may reflect characteristics of the larger groups who share similar cultural norms and social environments, they can provide considerable insight into the spread of STIs. Prevention strategies based on individual and community-level drivers of STIs are likely to be the most effective means of targeting and reaching those at greatest risk of infection. This strategy has the potential to play a significant role in the epidemic's trajectory.

Temporal trends in sexual behaviours and their impacts on HIV incidence among South African women: 2002-2016

South Africa has the highest number of HIV infected individuals in the world. The primary objective of the current study was to describe temporal changes in HIV incidence rates using the data from 9,948 women who enrolled in one of the six HIV prevention trials conducted in KwaZulu-Natal, South Africa. Characteristics of the study population were presented and compared across the four study periods: 2002-2004, 2005-2008, 2009-2011 and 2012-2016. HIV infection rates increased from 6.2 to 9.3 per 100 person-year over the 15 years. These rates were as high as 14 per 100 person-year among women younger than 20 years age. Being single/not cohabiting, using injectable contraceptives, having less than two children, and diagnosed with STI(s) were associated with increased risk of HIV infection. These four factors were associated with 71%, 75%, 80% and 88% of the HIV seroconversions in four study periods. As the research continues to find ways of controlling the spread of the infections, quantifying the temporal trends in risk factors and their population-level impacts on HIV infection may have significant implications. This information may assist in developing effective counselling and education programs by targeting the sexually active single women and delivering more realistic messages.

The impact of self-selection based on HIV risk on the cost-effectiveness of preexposure prophylaxis in South Africa

OBJECTIVES

We explored the impact and cost-effectiveness of preexposure prophylaxis (PrEP) provision to different populations in South Africa, with and without effective self-selection by individuals at highest risk of contracting HIV (through concurrent partnerships and/or commercial sex).

DESIGN AND METHODS

We used a previously developed HIV transmission model to analyse the epidemiological impact of PrEP provision to adolescents, young adults, pregnant women, female sex workers (FSWs) and (MSM), and data from South African PrEP programmes to estimate the cost and cost-effectiveness of PrEP (cost in 2019 USD per HIV infection averted over 20 years, 2019, 38). PrEP uptake followed data from early implementation sites, scaled-up linearly over 3 years, with target coverage set to 18% for adolescents, young adults and pregnant women, 30% for FSW and 54% for MSM.

RESULTS

The annual cost of PrEP provision ranges between $75 and $134 per person. PrEP provision adolescents and young adults, regardless of risk behaviour, will each avert 3.2--4.8% of HIV infections over 20 years; provision to high-risk individuals only has similar impact at lower total cost. The incremental cost per HIV infection averted is lower in high-risk vs. all-risk sub-populations within female adolescents ($507 vs. $4537), male adolescents ($2108 vs. $5637), young women ($1592 vs. $10 323) and young men ($2605 vs. $7715), becoming cost saving within 20 years for high-risk adolescents, young women, MSM and FSWs.

CONCLUSION

PrEP is an expensive prevention intervention but uptake by those at the highest risk of HIV infection will make it more cost-effective, and cost-saving after 14-18 years.

The development of an HIV-specific complexity rating scale

As treatment for HIV improves, an ageing population is experiencing comorbidity which often leads to complex clinical presentations requiring an interdisciplinary care approach. This study sought to quantify clinician assessment of the level of clinical complexity, through the development of a rating scale for people living with HIV (PLHIV), to improve client care through an interdisciplinary care model. An existing alcohol and other drug complexity rating scale was selected and modified for use with PLHIV. HIV-specific items were included through consultation with an interdisciplinary team. A risk-prediction model was developed and validated using clinician ratings of clients attending The Albion Centre, a tertiary HIV clinic in Sydney, Australia, resulting in the development of the Clinical Complexity Rating Scale for HIV (CCRS-HIV). Multivariable logistic regression models identified eight characteristics based on clinician assessment of complexity in PLHIV: financial instability, social isolation, problematic crystal methamphetamine use, mental illness and/or other problematic substance use, cognitive/neurological impairment, polypharmacy, current hepatitis C infection and/or cancer, and other physical health comorbidity. A weighted risk-prediction model was developed and validated. The final model accurately predicted 85% of complex clients, with a sensitivity of 80% and specificity of 91%. This study developed an HIV-specific clinician-rated complexity scale. Further investigations are required to validate the CCRS-HIV with broader HIV populations. This simple complexity screening tool is a promising adjunct to clinical assessment to identify clients with complex physical and psychosocial needs who may benefit from interdisciplinary care interventions and allocation of resources.

Targeted Pregnancy and Human Immunodeficiency Virus Prevention Risk-Reduction Counseling for Young Women: Lessons Learned from Biomedical Prevention Trials

Background

Women enrolled in human immunodeficiency virus (HIV) prevention efficacy trials receive counseling on prevention of HIV, sexually transmitted infections (STIs), and pregnancy during every visit. Incident pregnancy has an impact on efficacy outcomes. Incidence rates of pregnancy and HIV/STIs among women who became pregnant and associated risk factors were assessed.

Methods

Data from 9165 women participating in HIV prevention trials in KwaZulu-Natal, South Africa from 2002-2012 were combined. Demographic and behavioral predictors of incidence pregnancy and incidence HIV and STIs were determined using Cox regression models.

Results

Overall pregnancy incidence was 9.6 per 100 person-year (py) (95% confidence interval [Cl], 9.1-10.3). Human immunodeficiency virus incidence among pregnant women was 5.93 per 100 py (95% Cl, 4.73-7.44). Incidence of STIs among pregnant women for Chlamydia trachomatis, Trichomonas vaginalis, Neisseria gonorrhoeae, and Treponema pallidum (syphilis) were 10.87, 7.42, 3.92, and 1.43 per 100 py, respectively. In the adjusted analyses, we observed overlapping risk factors for HIV acquisition during pregnancy, ie, young age, not married/not cohabitating, and low parity. The risk of pregnancy and HIV acquisition is more than 3 times higher among young women (<20 years of age).

Conclusions

We identified overlapping risk factors for pregnancy and HIV incidence, suggesting an urgent need for appropriate, targeted, individual-centred counseling for women participating in HIV prevention trials.

Investigating spatial disparities in high-risk women and HIV infections using generalized additive models: Results from a cohort of South African women

OBJECTIVE

We identified the geographical clustering of HIV as well as those at highest risk of infection using a decade long data (2002-2012) from KwaZulu-Natal, South Africa.

METHODS

A total of 5,776 women who enrolled in several HIV prevention trials were included in the study. Geo-coded individual-level data were linked to the community-level characteristics using the South African Census. High-risk women were identified using a risk scoring algorithm. Generalized additive models were used to identify the significant geographical clustering of high-risk women and HIV.

RESULTS

Overall, 60% of the women were classified as high risk of HIV. HIV infection rates were estimated as high as 10 to 15 per 100 person year. Areas with high rates of HIV infections were spatially clustered and overlapped particularly in the Northern part of Durban.

CONCLUSION

Targeting multifactorial and complex nature of the epidemic is urgently needed to identify the "high transmission" areas.

Perceived Versus Calculated HIV Risk: Implications for Pre-exposure Prophylaxis Uptake in a Randomized Trial of Men Who Have Sex With Men

BACKGROUND

Inaccurate HIV risk perception by men who have sex with men is a barrier to HIV prevention. Providing information about objective HIV risk could improve pre-exposure prophylaxis (PrEP) uptake.

METHODS

PrEP Accessibility Research & Evaluation 2 (PrEPARE2) was a randomized controlled trial of men who have sex with men to determine whether an objective risk score affects future PrEP uptake. Participants completed a baseline survey to assess demographics, risk behaviors, and HIV self-perceived risk (SPR). The survey generated a calculated HIV risk (CalcR) score, estimating HIV risk based on reported condomless anal intercourse and sexually transmitted infections, and was provided to individuals in the intervention arm. Participants were contacted 8 weeks later to determine whether they initiated PrEP.

RESULTS

Of 171 participants (median age 32 years; 37% Hispanic or non-Hispanic Black; median 5 sexual partners in the past 6 months), 81% had heard of PrEP, and 57% believed they were good PrEP candidates. SPR had poor agreement with CalcR (kappa = 0.176) with 38% underestimating their HIV risk. At week 8, only 14 of 135 participants had initiated PrEP with no difference between arms (CalcR 11%, control 10%, P > 0.99). The most common reason for not starting PrEP was low HIV risk perception. There was a relative decrease in SPR over time (P = 0.06) but no difference between arms (P = 0.29).

CONCLUSION

Providing an objective HIV risk score alone did not increase PrEP uptake. HIV testing performed at testing sites may be a crucial time to correct misperceptions about risk and initiate same-day PrEP, given enthusiasm for PrEP on the testing day to facilitate greater uptake.

Geographical-level contributions of risk factors for HIV infections using generalized additive models: results from a cohort of South African women

South Africa has the highest burden of human immunodeficiency virus (HIV) infections in the world with significant geographical variations. We identified the predictors of HIV infections and their sub-geographical-level contributions to the epidemic using a decade long data (2002-2012) from 6484 South African women. Generalized additive models were used to uncover the most significant features of these estimates across the region. In the overall analysis, younger age, not married or cohabiting with a partner, partner has another partner(s) and null/prim parity, using injectable contraceptives and presence of other sexually transmitted infections (STIs) were identified as independent predictors of HIV seroconversions. Overall, the top three highest contributors to infections were women's marital status (PAR% = 73%, 95% CI: 68%, 77%), parity (PAR% = 47%, 95% CI: 42%, 53%) and partnership factors (PAR% = 37%, 95% CI: 30%, 44%). However, their contributions varied remarkably at sub-geographical level. This was mainly due to the substantial localized variations in their prevalence and hazard ratios across the region. Our results will guide policy makers to develop tailored prevention strategies in order to allocate scarce resources by targeting the most significant contributors of HIV infection at sub-geographical level.

Prevalence of arterial hypertension and risk factors among people with acquired immunodeficiency syndrome

OBJECTIVES

to verify the prevalence of arterial hypertension and its risk factors among people with acquired immunodeficiency syndrome under antiretroviral therapy.

METHOD

cross-sectional study with 208 patients. Data collection was conducted through interviews using a form containing data on sociodemographic, clinical and epidemiological aspects, hypertension risk factors, blood pressure, weight, height, body mass index and abdominal circumference. Mean, standard deviation, odds ratio and confidence interval were calculated, t-test and Chi-square test were used, considering P < 0.05 as statistically significant. Hypertension associated variables were selected for logistic regression.

RESULTS

patients were male (70.7%), self-reported as mixed-race (68.2%), had schooling between 9 and 12 years of study (46.6%), had no children (47.6%), were single (44.2%), in the sexual exposure category (72.1%) and heterosexual (60.6%). The prevalence of people with acquired immunodeficiency syndrome and arterial hypertension was 17.3%. Logistic regression confirmed the influence of age greater than 45 years, family history of hypertension, being overweight and antiretroviral therapy for more than 36 months for hypertension to occur.

CONCLUSION

the prevalence of hypertension was 17.3%. Patients with acquired immunodeficiency syndrome and hypertension were older than 45 years, had family history of hypertension, were overweight and under antiretroviral therapy for more than 36 months.

Predictors of HIV, HIV Risk Perception, and HIV Worry Among Adolescent Girls and Young Women in Lilongwe, Malawi

BACKGROUND

Adolescent girls and young women (AGYW) in sub-Saharan Africa have high HIV prevalence and incidence. We sought to understand which HIV risk factors individually and in combination contribute to risk, and whether these factors are associated with HIV worry and risk perception.

SETTING

This study is ongoing at 4 public health centers in Lilongwe, Malawi (2016-2017).

METHODS

AGYW of 15-24 years old were recruited to participate in a study assessing 4 models of service delivery. At each health center, participants completed a baseline survey assessing socioeconomic, behavioral, biomedical, and partnership characteristics; self-reported HIV status; and, if HIV-uninfected, HIV risk perception (high versus low or none) and HIV worry (any versus none). We analyzed associations between baseline characteristics and HIV prevalence, risk perception, and worry.

RESULTS

Among 1000 AGYW, median age was 19 years (IQR: 17-21). Thirty-three participants reported being HIV-infected. Fifteen characteristics were associated with HIV infection. Having more risk factors was associated with higher HIV prevalence (≤4 factors, 0.5%; 5-8 factors, 6%; >8 factors, 21%). Having more risk factors was also associated with higher risk perception (P < 0.001) and higher worry (P < 0.001). However, among those with ≥8 risk factors, 52% did not consider themselves to be at high risk and 21% did not report any HIV worry.

CONCLUSIONS

Most AGYW perceive little risk of HIV acquisition, even those at highest risk. As a critical gap in the HIV prevention cascade, accurate risk perception is needed to tailor effective and sustained combination prevention strategies for this vulnerable population.