Dataset Information

Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling study.

ABSTRACT: BACKGROUND:HIV pre-exposure prophylaxis (PrEP) is effective but underused, in part because clinicians do not have the tools to identify PrEP candidates. We developed and validated an automated prediction algorithm that uses electronic health record (EHR) data to identify individuals at increased risk for HIV acquisition. METHODS:We used machine learning algorithms to predict incident HIV infections with 180 potential predictors of HIV risk drawn from EHR data from 2007-15 at Atrius Health, an ambulatory group practice in Massachusetts, USA. We included EHRs of all patients aged 15 years or older with at least one clinical encounter during 2007-15. We used ten-fold cross-validated area under the receiver operating characteristic curve (cv-AUC) with 95% CIs to assess the model's performance at identifying individuals with incident HIV and patients independently prescribed PrEP by clinicians. The best-performing model was validated prospectively with 2016 data from Atrius Health and externally with 2011-16 data from Fenway Health, a community health centre specialising in sexual health care in Boston (MA, USA). We calculated HIV risk scores (ie, probability of an incident HIV diagnosis) for every HIV-uninfected patient not on PrEP during 2007-15 at Atrius Health and assessed the distribution of scores for thresholds to determine possible candidates for PrEP in the three study cohorts. FINDINGS:We included 1?155?966 Atrius Health patients from 2007-15 (150 [<0·1%] patients with incident HIV) in our development cohort, 537?257 Atrius Health patients in 2016 (16 [<0·1%] with incident HIV) in our prospective validation cohort, and 33?404 Fenway Health patients from 2011-16 (423 [1·3%] with incident HIV) in our external validation cohort. The best-performing algorithm was obtained with least absolute shrinkage and selection operator (LASSO) and had a cv-AUC of 0·86 (95% CI 0·82-0·90) for identification of incident HIV infections in the development cohort, 0·91 (0·81-1·00) on prospective validation, and 0·77 (0·74-0·79) on external validation. The LASSO model successfully identified patients independently prescribed PrEP by clinicians at Atrius Health in 2016 (cv-AUC 0·93, 95% CI 0·90-0·96) or Fenway Health (0·79, 0·78-0·80). HIV risk scores increased steeply at the 98th percentile. Using this score as a threshold, we prospectively identified 9515 (1·8%) of 536?384 patients at Atrius Health in 2016 and 4385 (15·3%) of 28?702 Fenway Health patients as potential PrEP candidates. INTERPRETATION:Automated algorithms can efficiently identify patients at increased risk for HIV acquisition. Integrating these models into EHRs to alert providers about patients who might benefit from PrEP could improve prescribing and prevent new HIV infections. FUNDING:Harvard University Center for AIDS Research, Providence/Boston Center for AIDS Research, Rhode Island IDeA-CTR, the National Institute of Mental Health, and the US Centers for Disease Control and Prevention.

SUBMITTER: Krakower DS

PROVIDER: S-EPMC7522919 | biostudies-literature | 2019 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling study.

Krakower Douglas S DS Gruber Susan S Hsu Katherine K Menchaca John T JT Maro Judith C JC Kruskal Benjamin A BA Wilson Ira B IB Mayer Kenneth H KH Klompas Michael M

The lancet. HIV 20190705 10

<h4>Background</h4>HIV pre-exposure prophylaxis (PrEP) is effective but underused, in part because clinicians do not have the tools to identify PrEP candidates. We developed and validated an automated prediction algorithm that uses electronic health record (EHR) data to identify individuals at increased risk for HIV acquisition.<h4>Methods</h4>We used machine learning algorithms to predict incident HIV infections with 180 potential predictors of HIV risk drawn from EHR data from 2007-15 at Atriu ...[more]

PMID: 31285182

Similar Datasets

Project description:BACKGROUND:The limitations of existing HIV risk prediction tools are a barrier to implementation of pre-exposure prophylaxis (PrEP). We developed and validated an HIV prediction model to identify potential PrEP candidates in a large health-care system. METHODS:Our study population was HIV-uninfected adult members of Kaiser Permanente Northern California, a large integrated health-care system, who were not yet using PrEP and had at least 2 years of previous health plan enrolment with at least one outpatient visit from Jan 1, 2007, to Dec 31, 2017. Using 81 electronic health record (EHR) variables, we applied least absolute shrinkage and selection operator (LASSO) regression to predict incident HIV diagnosis within 3 years on a subset of patients who entered the cohort in 2007-14 (development dataset), assessing ten-fold cross-validated area under the receiver operating characteristic curve (AUC) and 95% CIs. We compared the full model to simpler models including only men who have sex with men (MSM) status and sexually transmitted infection (STI) positivity, testing, and treatment. Models were validated prospectively with data from an independent set of patients who entered the cohort in 2015-17. We computed predicted probabilities of incident HIV diagnosis within 3 years (risk scores), categorised as low risk (<0·05%), moderate risk (0·05% to <0·20%), high risk (0·20% to <1·0%), and very high risk (?1·0%), for all patients in the validation dataset. FINDINGS:Of 3?750?664 patients in 2007-17 (3?143?963 in the development dataset and 606?701 in the validation dataset), there were 784 incident HIV cases within 3 years of baseline. The LASSO procedure retained 44 predictors in the full model, with an AUC of 0·86 (95% CI 0·85-0·87) for incident HIV cases in 2007-14. Model performance remained high in the validation dataset (AUC 0·84, 0·80-0·89). The full model outperformed simpler models including only MSM status and STI positivity. For the full model, flagging 13?463 (2·2%) patients with high or very high HIV risk scores in the validation dataset identified 32 (38·6%) of the 83 incident HIV cases, including 32 (46·4%) of 69 male cases and none of the 14 female cases. The full model had equivalent sensitivity by race whereas simpler models identified fewer black than white HIV cases. INTERPRETATION:Prediction models using EHR data can identify patients at high risk of HIV acquisition who could benefit from PrEP. Future studies should optimise EHR-based HIV risk prediction tools and evaluate their effect on prescription of PrEP. FUNDING:Kaiser Permanente Community Benefit Research Program and the US National Institutes of Health.

Project description:Pre-exposure prophylaxis (PrEP) has emerged as a promising strategy for preventing the transmission of HIV. Although only one formulation is currently approved for PrEP, research into both new compounds and new delivery systems for PrEP regimens offer intriguing challenges from the perspective of pharmacokinetic and pharmacodynamic modeling. This review aims to provide an overview the current modeling landscape for HIV PrEP, focused on PK/PD and QSP models relating to antiretroviral agents. Both current PrEP treatments and new compounds that show promise as PrEP agents are highlighted, as well as models of uncommon administration routes, predictions based on models of mechanism of action and viral dynamics, and issues related to adherence to therapy. The spread of human immunodeficiency virus (HIV) remains one of the foremost global health concerns. In the absence of a vaccine, other prophylactic strategies have been developed to prevent HIV transmission. One approach, known as pre-exposure prophylaxis (PrEP), allows HIV-negative individuals who are at high risk of exposure to the virus, be it through an HIV-positive sexual partner or through the shared use of drug injection equipment, to substantially reduce the risk of developing an HIV infection. PrEP is a relatively recent approach to combating the HIV epidemic, with the only currently approved treatment being Truvada, a daily oral antiretroviral (ARV) therapy initially indicated in the treatment of active HIV-1 infections, but approved for HIV PrEP in 2012. Although PrEP therapy has consistently demonstrated high efficacy in preventing HIV infection, this efficacy is dependent on patient adherence to the prescribed treatment regimen. This can present a significant problem in low- and middle-income countries, which may lack the infrastructure to provide sufficient access to PrEP medication to maintain daily dosing regimens. Furthermore, while the conventional approach has generally been to advocate for continuous administration akin to regimens used for viral suppression in infected patients, there has been some discussion of whether a better treatment paradigm might be to push for PrEP therapy primarily during those known periods of heightened exposure risk, while relying on post-exposure prophylaxis regimens to prevent infection after unanticipated exposures during low-risk periods. These considerations have led to a push for the development of long-duration and on-demand PrEP formulations, including subdermal and subcutaneous implants, slow-release intramuscular depot injections, vaginal and rectal antimicrobial gels, and intravaginal rings and dissolving films. PrEP therapy is a quickly evolving field, with a variety of antiretroviral compounds and formulations under investigation. This review aims to report on notable drugs and formulations from a pharmacokinetic/pharmacodynamic (PK/PD) modeling perspective. Given the nature of PrEP as a preventive therapy designed for long-term use, clinical trials for PrEP therapies can last for months or even years, particularly in the case of long-duration formulations. Furthermore, in contrast to antiretroviral trials in infected patients, pharmacodynamic endpoints in PrEP therapies are difficult to quantify, as the primary endpoint for efficacy is generally the rate of seroconversion. Computational modeling approaches offer flexible and powerful tools to provide insight into drug behavior in clinical settings, and can ultimately reduce the time, expense, and patient burden incurred in the development of PrEP therapies.

Project description:INTRODUCTION:Oral pre-exposure prophylaxis (PrEP) provision is a priority intervention for high HIV prevalence settings and populations at substantial risk of HIV acquisition. This mathematical modelling analysis estimated the impact, cost and cost-effectiveness of scaling up oral PrEP in 13 countries. METHODS:We projected the impact and cost-effectiveness of oral PrEP between 2018 and 2030 using a combination of the Incidence Patterns Model and the Goals model. We created four PrEP rollout scenarios involving three priority populations-female sex workers (FSWs), serodiscordant couples (SDCs) and adolescent girls and young women (AGYW)-both with and without geographic prioritization. We applied the model to 13 countries (Eswatini, Ethiopia, Haiti, Kenya, Lesotho, Mozambique, Namibia, Nigeria, Tanzania, Uganda, Zambia and Zimbabwe). The base case assumed achievement of the Joint United Nations Programme on HIV/AIDS 90-90-90 antiretroviral therapy targets, 90% male circumcision coverage by 2020 and 90% efficacy and adherence levels for oral PrEP. RESULTS:In the scenarios we examined, oral PrEP averted 3% to 8% of HIV infections across the 13 countries between 2018 and 2030. For all but three countries, more than 50% of the HIV infections averted by oral PrEP in the scenarios we examined could be obtained by rollout to FSWs and SDCs alone. For several countries, expanding oral PrEP to include medium-risk AGYW in all regions greatly increased the impact. The efficiency and impact benefits of geographic prioritization of rollout to AGYW varied across countries. Variations in cost-effectiveness across countries reflected differences in HIV incidence and expected variations in unit cost. For most countries, rolling out oral PrEP to FSWs, SDCs and geographically prioritized AGYW was not projected to have a substantial impact on the supply chain for antiretroviral drugs. CONCLUSIONS:These modelling results can inform prioritization, target-setting and other decisions related to oral PrEP scale-up within combination prevention programmes. We caution against extensive use given limitations in cost data and implementation approaches. This analysis highlights some of the immediate challenges facing countries-for example, trade-offs between overall impact and cost-effectiveness-and emphasizes the need to improve data availability and risk assessment tools to help countries make informed decisions.

Project description:As pre-exposure prophylaxis (PrEP) effectiveness is strongly linked to adherence, we sought to determine if certain self-report measures could be used to inform objective PrEP adherence. We studied participants from the TAPIR study (a multicenter randomized study of daily text messages to support adherence to PrEP In At-Risk), a 48-week randomized controlled trial of HIV-uninfected men who have sex with men (MSM) randomized to receive text message to support adherence versus standard of care. Self-reported medication adherence was assessed using several validated measures modified for PrEP. Objective PrEP adherence was determined through dried blood spot (DBS) measurement of intracellular tenofovir diphosphate (TFV-DP) and emtricitabine triphosphate (FTC-TP). A summary of adherence was estimated using responses to the seven adherence items at weeks 12 and 48 using confirmatory factor analysis. Correlations between self-report questions and drug concentrations were estimated with Pearson's correlations for continuous outcomes and point-biserial correlations for dichotomous outcomes. Receiver operating characteristic (ROC) analyses were conducted to assess the performance of self-report measures in predicting protective or perfect TFV-DP concentrations. Of the 369 participants who completed week 12 or 48 visits, the mean age was 35 (standard deviation 9 years), with 79% White, 12% Black, and 29% Hispanic. Correlations between self-report measures of adherence (both individual items and the adherence factor) and quantifiable FTC-TP and continuous TFV-DP concentrations showed that all self-report measures were significantly associated with these objective measures. Compared to a summary measure of self-reported adherence, the 4-week percent taken question medication recall was the only self-report item similarly or more strongly associated with recent adherence and long-term protective and perfect adherence at weeks 12 and 48. ROC analysis also showed that 4-week percent taken question had a reasonable AUC (0.798 at week 12 and 0.758 at week 48) in predicting protective TFV-DP concentrations. All single-item self-report questions assessing PrEP adherence were significantly associated with biomarker quantification, with the 4-week percent taken question performing best. Therefore, in the absence of drug concentration measurements, a 4-week self-report percent taken question may be a good single-item measure of PrEP adherence.

Project description:OBJECTIVES:Neisseria gonorrhoeae (NG) and Chlamydia trachomatis (CT) increase the risk of HIV transmission among men who have sex with men (MSM). Diagnosis of NG/CT may provide an efficient entry point for prevention of HIV through the delivery of pre-exposure prophylaxis (PrEP); however, the additional population-level impact of targeting PrEP to MSM diagnosed with NG/CT is unknown. DESIGN:An agent-based simulation model of NG/CT and HIV cocirculation among MSM calibrated against census data, disease surveillance reports and the US National HIV Behavioral Surveillance study. SETTING:Baltimore City, Maryland, USA. INTERVENTIONS:PrEP implementation was modelled under three alternative scenarios: (1) PrEP delivery at NG/CT diagnosis (targeted delivery), (2) PrEP evaluation at NG/CT screening/testing and (3) PrEP evaluation in the general community (untargeted). MAIN OUTCOME:The projected incidence of HIV after 20 years of PrEP delivery under two alternatives: when equal numbers of MSM are (1) screened for PrEP or (2) receive PrEP in each year. RESULTS:Assuming 60% uptake and 60% adherence, targeting PrEP to MSM diagnosed with NG/CT could reduce HIV incidence among MSM in Baltimore City by 12.4% (95% uncertainty range (UR) 10.3% to 14.4%) in 20 years, relative to no PrEP. Expanding the coverage of NG/CT screening (such that individuals experience a 50% annual probability of NG/CT screening and evaluation for PrEP on NG/CT diagnosis) can further increase the impact of targeted PrEP to generate a 22.0% (95% UR 20.1% to 23.9%) reduction in HIV incidence within 20 years. When compared with alternative implementation scenarios, PrEP evaluation at NG/CT diagnosis increased impact of PrEP on HIV incidence by 1.5(95% UR 1.1 to 1.9) times relative to a scenario in which PrEP evaluation happened at the time of NG/CT screening/testing and by 1.6 (95% UR 1.2 to 2.2) times relative to evaluating random MSM from the community. CONCLUSIONS:Targeting MSM infected with NG/CT increases the efficiency and effectiveness of PrEP delivery. If high levels of sexually transmitted infection screening can be achieved at the community level, NG/CT diagnosis may be a highly effective entry point for PrEP initialisation.

Dataset Information

Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling study.

Publications

Development and validation of an automated HIV prediction algorithm to identify candidates for pre-exposure prophylaxis: a modelling study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets