Deep learning and social network analysis elucidate drivers of HIV transmission in a high-incidence cohort of people who inject drugs.
Ontology highlight
ABSTRACT:
SUBMITTER: Clipman SJ
PROVIDER: S-EPMC9581475 | biostudies-literature | 2022 Oct
REPOSITORIES: biostudies-literature
Publications
Deep learning and social network analysis elucidate drivers of HIV transmission in a high-incidence cohort of people who inject drugs.
Clipman Steven J SJ Mehta Shruti H SH Mohapatra Shobha S Srikrishnan Aylur K AK Zook Katie J C KJC Duggal Priya P Saravanan Shanmugam S Nandagopal Paneerselvam P Kumar Muniratnam Suresh MS Lucas Gregory M GM Latkin Carl A CA Solomon Sunil S SS
Science advances 20221019 42
Globally, people who inject drugs (PWID) experience some of the fastest-growing HIV epidemics. Network-based approaches represent a powerful tool for understanding and combating these epidemics; however, detailed social network studies are limited and pose analytical challenges. We collected longitudinal social (injection partners) and spatial (injection venues) network information from 2512 PWID in New Delhi, India. We leveraged network analysis and graph neural networks (GNNs) to uncover facto ...[more]
Similar Datasets
Project description:ObjectiveWe investigated the duration of HIV transmission clusters.DesignFifty-four individuals newly infected at enrollment in the ALIVE cohort were included, all of whom had sequences at an intake visit (T1) and from a second (T2) and/or a third (T3) follow-up visit, median 2.9 and 5.4 years later, respectively.MethodsSequences were generated using the 454 DNA sequencing platform for portions of HIV pol and env (HXB2 positions 2717-3230; 7941-8264). Genetic distances were calculated using tn93 and sequences were clustered over a range of thresholds (1--5%) using HIV-TRACE. Analyses were performed separately for individuals with pol sequences for T1 + T2 (n = 40, 'Set 1') and T1 + T3 (n = 25; 'Set 2'), and env sequences for T1 + T2 (n = 47, 'Set 1'), and T1 + T3 (n = 30; 'Set 2').ResultsFor pol, with one exception, a single cluster contained more than 75% of samples at all thresholds, and cluster composition was at least 90% concordant between time points/thresholds. For env, two major clusters (A and B) were observed at T1 and T2/T3, although cluster composition concordance between time points/thresholds was low (<60%) at lower thresholds for both sets 1 and 2. In addition, several individuals were included in clusters at T2/T3, although not at T1.ConclusionCaution should be used in applying a single threshold in population studies where seroconversion dates are unknown. However, the retention of some clusters even after 5 + years is evidence for the robustness of the clustering approach in general.
| S-EPMC9421556 | biostudies-literature
Project description:BackgroundDivergent estimates of HIV incidence among people who inject drugs (PWID) in Ukraine have been reported in modeling studies, longitudinal cohort studies, and recent infection assays used in cross-sectional surveys. Estimates range from 0.65 to 24.8 infections per 100 person-years with substantial regional variation. In this paper, we study the sources of this discrepancy.MethodsWe compared baseline characteristics of study subjects recruited in the cross-sectional integrated bio-behavioral surveillance surveys (IBBS) in 2011 and 2013, with those from the longitudinal network intervention trial (network RCT) conducted between 2010 - 2013, the study that found a remarkably high incidence of HIV among PWID in Ukraine. The analysis was conducted for two cities: Mykolaiv and Odesa.ResultsSignificant differences were found in the characteristics of study subjects recruited in the IBBS surveys and the network RCT, in particular in Odesa, where the mismatch in the estimates of HIV incidence is greatest. In Odesa, recent syringe sharing was about three times as prevalent in the network RCT as in the IBBS; 39% of the network RCT and 16-18% of the IBBS participants indicated stimulants rather than opiates as their drug of choice; 97% of respondents in the network RCT and 45% in the IBBS-2013 reported injecting in a group over half of the time; and the average monthly number of injections in the network RCT was about twice that in the IBBS studies.ConclusionsDifferences in study designs and sampling methodologies may be responsible for the substantial differences in HIV incidence estimates among PWID in Ukraine. The potential sources of selection bias differed between the studies and likely resulted in the recruitment of lower risk individuals into the IBBS studies compared to the network RCT. Risk stratification in the population of PWID may have implications for future surveillance and intervention efforts.
| S-EPMC6899203 | biostudies-literature
Project description:ObjectiveHIV infection in sub-Saharan Africa increasingly occurs among people who inject drugs (PWID). Kenya is one of the first to implement a national needle and syringe program. Our study undertook a baseline assessment as part of evaluating needle and syringe program in a seek, test, treat, and retain approach.MethodsParticipants enrolled between May and December 2012 from 10 sites. Respondent-driven sampling was used to reach 1785 PWID for HIV-1 prevalence and viral load determination and survey data.ResultsEstimated HIV prevalence, adjusted for differential network size and recruitment relationships, was 14.5% in Nairobi (95% CI: 10.8 to 18.2) and 20.5% in the Coast region (95% CI: 17.3 to 23.6). Viral load (log10 transformed) in Nairobi ranged from 1.71 to 6.12 (median: 4.41; interquartile range: 3.51-4.94) and in the Coast from 1.71 to 5.88 (median: 4.01; interquartile range: 3.44-4.72). Using log10 viral load 2.6 as a threshold for HIV viral suppression, the percentage of HIV-infected participants with viral suppression was 4.2% in Nairobi and 4.6% in the Coast. Heroin was the most commonly injected drug in both regions, used by 93% of participants in the past month, typically injecting 2-3 times/day. Receptive needle/syringe sharing at last injection was more common in Nairobi (23%) than in the Coast (4%). Estimated incidence among new injectors was 2.5/100 person-years in Nairobi and 1.6/100 person-years in the Coast.ConclusionsThe HIV epidemic is well established among PWID in both Nairobi and Coast regions. Public health scale implementation of combination HIV prevention has the potential to greatly limit the epidemic in this vulnerable and bridging population.
| S-EPMC4624615 | biostudies-literature
Project description:Estimates for the contribution of transmission arising from acute HIV infections (AHIs) to overall HIV incidence vary significantly. Furthermore, little is known about AHI-attributable transmission among people who inject drugs (PWID), including the extent to which interventions targeting chronic infections (e.g. HAART as prevention) are limited by AHI transmission. Thus, we estimated the proportion of transmission events attributable to AHI within the mature HIV epidemic among PWID in New York City (NYC).Modeling study.We constructed an interactive sexual and injecting transmission network using an agent-based model simulating the HIV epidemic in NYC between 1996 and 2012. Using stochastic microsimulations, we cataloged transmission from PWID based on the disease stage of index agents to determine the proportion of infections transmitted during AHI (in primary analyses, assumed to last 3 months).Our calibrated model approximated the epidemiological features of the mature HIV epidemic in NYC between 1996 and 2012. Annual HIV incidence among PWID dropped from approximately 1.8% in 1996 to 0.7% in 2012. Over the 16-year period, AHI accounted for 4.9% (10th/90th percentile: 0.1-12.3%) of incident HIV cases among PWID. The annualized contribution of AHI increased over this period from 3.6% in 1996 to 5.9% in 2012.Our results suggest that, in mature epidemics such as NYC, between 3% and 6% of transmission events are attributable to AHI among PWID. Current HIV treatment as prevention strategies are unlikely to be substantially affected by AHI-attributable transmission among PWID populations in mature epidemic settings.
| S-EPMC5069170 | biostudies-literature
Project description:IntroductionHIV testing programmes have struggled to reach the most marginalized populations at risk for HIV. Social network methods such as respondent-driven sampling (RDS) and peer-based active case-finding (ACF) may be effective in overcoming barriers to reaching these populations. We compared the client characteristics, proportion testing HIV positive (yield), and number of new cases found through two RDS strategies and an ACF approach to HIV case-finding among people who inject drugs (PWID) in Tajikistan.MethodsRoutine programme data from adult PWID recruited to testing under the HIV Flagship Project in Tajikistan were analysed to compare client demographic and clinical characteristics across the three approaches. We also compared the number of previously untested clients, the number of new HIV cases found, and the yield across the case-finding strategies, and evaluated predictors of new HIV diagnosis using fixed-effects logistic regression.ResultsFrom 24 October 2016 to 30 June 2017, Flagship tested 10,300 PWID for HIV, including 2143 under RDS with unrestricted waves (RDS1, yield: 1.5%), 3517 under restricted RDS (RDS2, yield: 2.6%), and 4640 under ACF (yield: 1.5%). Clients recruited under ACF were similar in age (35.8 vs. 36.8) and gender (91% vs. 90% male) to those recruited through RDS, though ACF clients were more likely to report being a first-time tester (85.1% vs. 68.3%, p < 0.001). After controlling for age, sex, previous testing history and accounting for clustering at the site level, we found that clients tested under both RDS1 (aOR: 1.74, 95% CI: 1.04 to 2.90) and RDS2 (aOR: 1.54, 95% CI: 1.11 to 2.15) had higher odds of testing newly positive for HIV relative to clients recruited through ACF. We did not find significant differences in the odds of new HIV infection between those recruited from RDS1 versus RDS2 (aOR: 1.12, 95% CI: 0.67 to 1.86).ConclusionsRDS-based interventions resulted in higher yields and overall case-finding, especially when recruitment was restricted. However, ACF identified a higher proportion of first-time testers. To find at least 90% of PWID living with HIV in Tajikistan, it may be necessary to implement multiple case-finding approaches concurrently to maximize testing coverage.
| S-EPMC6055120 | biostudies-literature
Project description:BACKGROUND: The risk-related behaviours and practices associated with injection drug use remain a driver of HIV and hepatitis C virus (HCV) transmission throughout the world. Here we evaluated HIV and HCV transmission patterns in the context of social networks of injection drug users (IDU) recruited from a higher incidence region in order to better understand factors that contribute to ongoing transmission among IDU. METHODS: IDU recruited through a chain-referral method provided biological specimens for analysis. HIV and HCV positive specimens were sequenced and analyzed using phylogenetic methods (Neighbour-joining and bayesian) and transmission patterns of HIV and HCV evaluated in the context of the recruitment networks. RESULTS: Among the 407 recruited IDU, HCV and HIV prevalence were 60.6% and 10.1%, respectively; 98% of HIV positive individuals were co-infected with HCV. Thirty-six percent of HCV sequences were associated with clusters, compared to 67% of HIV sequences. Four (16.7%) of the 24 HCV clusters contained membership separated by 2 or fewer recruitment cycles, compared to 10 (41.6%) derived from more than one recruitment component. Two (28.6%) of the 7 HIV clusters contained membership separated by 2 or fewer recruitment cycles while 6 (85.7%) were composed of inter component membership. CONCLUSIONS: Few HIV and HCV transmissions coincided with the recruitment networks, suggesting that they occurred in a different social context or a context not captured by the recruitment network. However, among the complete cohort, a higher degree of HIV clustering indicates many are recent infections originating from within current social networks, whereas a larger proportion of HCV infections may have occurred earlier in injecting history and in the context of a different social environment.
| S-EPMC3140499 | biostudies-literature
Project description:As HIV incidence among people who inject drugs grows in the context of an escalating drug overdose epidemic in North America, investigating how network structure may affect vulnerability to rapid HIV transmission is necessary for preventing outbreaks. We compared the characteristics of the observed contact tracing network from the 2015 outbreak in rural Indiana with 1000 networks generated by an agent-based network model with approximately the same number of individuals (n = 420) and ties between them (n = 913). We introduced an initial HIV infection into the simulated networks and compared the subsequent epidemic behavior (e.g., cumulative HIV infections over 5 years). The model was able to produce networks with largely comparable characteristics and total numbers of incident HIV infections. Although the model was unable to produce networks with comparable cohesiveness (where the observed network had a transitivity value 35.7 standard deviations from the mean of the simulated networks), the structural variability of the simulated networks allowed for investigation into their potential facilitation of HIV transmission. These findings emphasize the need for continued development of injection network simulation studies in tandem with empirical data collection to further investigate how network characteristics played a role in this and future outbreaks.
| S-EPMC7940592 | biostudies-literature
Project description:BackgroundHIV prevalence among people who inject drugs (PWID) in Ukraine is among the highest in the world. In this study, we aimed to assess whether a social network intervention was superior to HIV testing and counselling in affecting HIV incidence among PWID. Although this was not the primary aim of the study, it is associated with reducing drug and sex risk behaviours, which were primary aims.MethodsIn this clustered randomised trial, PWID who were 16 years of age or older, had used self-reported drug injection in the past 30 days, were willing to be interviewed for about 1 hour and tested for HIV, were not too impaired to comprehend and provide informed consent, and, for this paper, who tested HIV negative at baseline were recruited from the streets by project outreach workers in three cities in southern and eastern Ukraine: Odessa, Donetsk, and Nikolayev. Index or peer leaders, along with two of their network members, were randomly assigned (1:1) by the study statistician to the testing and counselling block (control group) or the testing and counselling plus a social network intervention block (intervention group). No stratification or minimisation was done. Participants in the network intervention received five sessions to train their network members in risk reduction. Those participants assigned to the control group received no further intervention after counselling. The main outcome of this study was HIV seroconversion in the intent-to-treat population as estimated with Cox regression and incorporating a γ frailty term to account for clustering. This trial is registered with ClinicalTrial.gov, number NCT01159704.FindingsBetween July 12, 2010, and Nov 23, 2012, 2304 PWIDs were recruited, 1200 of whom were HIV negative and are included in the present study. 589 index or peer leaders were randomly assigned to the control group and 611 were assigned to the intervention group. Of the 1200 HIV-negative participants, 1085 (90%) were retained at 12 months. In 553·0 person-years in the intervention group, 102 participants had seroconversion (incidence density 18·45 per 100 person-years; 95% CI 14·87-22·03); in 497·1 person-years in the control group 158 participants seroconverted (31·78 per 100 person-years; 26·83-36·74). This corresponded to a reduced hazard in the intervention group (hazard ratio 0·53, 95% CI 0·38-0·76, p=0·0003). No study-related adverse events were reported.InterpretationThese data provide strong support for integrating peer education into comprehensive HIV prevention programmes for PWID and suggest the value in developing and testing peer-led interventions to improve access and adherence to pre-exposure prophylaxis and antiretroviral therapy.FundingThe National Institute on Drug Abuse.
| S-EPMC5101021 | biostudies-literature
Project description:ObjectiveTo quantify the effect of opiate substitution treatment in relation to HIV transmission among people who inject drugs.DesignSystematic review and meta-analysis of prospective published and unpublished observational studies.Data sourcesSearch of Medline, Embase, PsychINFO, and the Cochrane Library from the earliest year to 2011 without language restriction.Review methodsWe selected studies that directly assessed the impact of opiate substitution treatment in relation to incidence of HIV and studies that assessed incidence of HIV in people who inject drugs and that might have collected data regarding exposure to opiate substitution treatment but not have reported it. Authors of these studies were contacted. Data were extracted by two reviewers and pooled in a meta-analysis with a random effects model.ResultsTwelve published studies that examined the impact of opiate substitution treatment on HIV transmission met criteria for inclusion, and unpublished data were obtained from three additional studies. All included studies examined methadone maintenance treatment. Data from nine of these studies could be pooled, including 819 incident HIV infections over 23,608 person years of follow-up. Opiate substitution treatment was associated with a 54% reduction in risk of HIV infection among people who inject drugs (rate ratio 0.46, 95% confidence interval 0.32 to 0.67; P<0.001). There was evidence of heterogeneity between studies (I(2)=60%, χ(2)=20.12, P=0.010), which could not be explained by geographical region, site of recruitment, or the provision of incentives. There was weak evidence for greater benefit associated with longer duration of exposure to opiate substitution treatment.ConclusionOpiate substitution treatment provided as maintenance therapy is associated with a reduction in the risk of HIV infection among people who inject drugs. These findings, however, could reflect comparatively high levels of motivation to change behaviour and reduce injecting risk behaviour among people who inject drugs who are receiving opiate substitution treatment.
| S-EPMC3489107 | biostudies-literature
Project description:Co-infections with human immunodeficiency virus type 1 (HIV-1) and human pegivirus (HPgV) are common in hepatitis C virus (HCV)-infected individuals. However, analysis on the evolutionary dynamics and transmission network profiles of these viruses among individuals with multiple infections remains limited. A total of 228 injecting drug users (IDUs), either HCV- and/or HIV-1-infected, were recruited in Kuala Lumpur, Malaysia. HCV, HIV-1 and HPgV genes were sequenced, with epidemic growth rates assessed by the Bayesian coalescent method. Based on the sequence data, mono-, dual- and triple-infection were detected in 38.8%, 40.6% and 20.6% of the subjects, respectively. Fifteen transmission networks involving HCV (subtype 1a, 1b, 3a and 3b), HIV-1 (CRF33_01B) and HPgV (genotype 2) were identified and characterized. Genealogical estimates indicated that the predominant HCV, HIV-1 and HPgV genotypes were introduced into the IDUs population through multiple sub-epidemics that emerged as early as 1950s (HCV), 1980s (HIV-1) and 1990s (HPgV). By determining the difference in divergence times between viral lineages (ΔtMRCA), we also showed that the frequency of viral co-transmission is low among these IDUs. Despite increased access to therapy and other harm reduction interventions, the continuous emergence and coexistence of new transmission networks suggest persistent multiple viral transmissions among IDUs.
| S-EPMC4602306 | biostudies-literature