Simulation Modeling and Metamodeling to Inform National and International HIV Policies for Children and Adolescents

Impact of subgroup-specific heterogeneities and dynamic changes in mortality rates on forecasted population size, deaths, and age distribution of persons receiving antiretroviral treatment in the United States: a computer simulation study

PURPOSE

Model-based forecasts of population size, deaths, and age distribution of people with HIV (PWH) are helpful for public health and clinical services planning but are influenced by subgroup-specific heterogeneities and changes in mortality rates.

METHODS

Using an agent-based simulation of PWH in the United States, we examined the impact of distinct approaches to parametrizing mortality rates on forecasted epidemiology of PWH on antiretroviral treatment (ART). We first estimated mortality rates among (1) all PWH, (2) sex-specific, (3) sex-and-race/ethnicity-specific, and (4) sex-race/ethnicity-and-HIV-acquisition-risk-specific subgroups. We then assessed each scenario by (1) allowing unrestricted reductions in age-specific mortality rates over time and (2) restricting the mortality rates among PWH to subgroup-specific mortality thresholds from the general population.

RESULTS

Among the eight scenarios examined, those lacking subgroup-specific heterogeneities and those allowing unrestricted reductions in future mortality rates forecasted the lowest number of deaths among all PWH and 9 of the 15 subgroups through 2030. The forecasted overall number and age distribution of people with a history of injection drug use were sensitive to inclusion of subgroup-specific mortality rates.

CONCLUSIONS

Our results underscore the potential risk of underestimating future deaths by models lacking subgroup-specific heterogeneities in mortality rates, and those allowing unrestricted reductions in future mortality rates.

Cost-effectiveness of point-of-care versus centralised, laboratory-based nucleic acid testing for diagnosis of HIV in infants: a systematic review of modelling studies

BACKGROUND

Point-of-care (POC) nucleic acid testing for diagnosis of HIV in infants facilitates earlier initiation of antiretroviral therapy (ART) than with centralised (standard-of-care, SOC) testing, but can be more expensive. We evaluated cost-effectiveness data from mathematical models comparing POC with SOC to provide global policy guidance.

METHODS

In this systematic review of modelling studies, we searched PubMed, MEDLINE, Embase, the National Health Service Economic Evaluation Database, Econlit, and conference abstracts, combining terms for "HIV" + "infant"/"early infant diagnosis" + "point-of-care" + "cost-effectiveness" + "mathematical models", without restrictions from database inception to July 15, 2022. We selected reports of mathematical cost-effectiveness models comparing POC with SOC for HIV diagnosis in infants younger than 18 months. Titles and abstracts were independently reviewed, with full-text review for qualifying articles. We extracted data on health and economic outcomes and incremental cost-effectiveness ratios (ICERs) for narrative synthesis. The primary outcomes of interest were ICERs (comparing POC with SOC) for ART initiation and survival of children living with HIV.

FINDINGS

Our search identified 75 records through database search. 13 duplicates were excluded, leaving 62 non-duplicate articles. 57 records were excluded and five were reviewed in full text. One article was excluded as it was not a modelling study, and four qualifying studies were included in the review. These four reports were from two mathematical models from two independent modelling groups. Two reports used the Johns Hopkins model to compare POC with SOC for repeat early infant diagnosis testing in the first 6 months in sub-Saharan Africa (first report, simulation of 25 000 children) and Zambia (second report, simulation of 7500 children). In the base scenario, POC versus SOC increased probability of ART initiation within 60 days of testing from 19% to 82% (ICER per additional ART initiation range US$430-1097; 9-month cost horizon) in the first report; and from 28% to 81% in the second ($23-1609, 5-year cost horizon). Two reports compared POC with SOC for testing at 6 weeks in Zimbabwe using the Cost-Effectiveness of Preventing AIDS Complications-Paediatric model (simulation of 30 million children; lifetime horizon). POC increased life expectancy and was considered cost-effective compared with SOC (ICER $711-850 per year of life saved in HIV-exposed children). Results were robust throughout sensitivity and scenario analyses. In most scenarios, platform cost-sharing (co-use with other programmes) resulted in POC being cost-saving compared with SOC.

INTERPRETATION

Four reports from two different models suggest that POC is a cost-effective and potentially cost-saving strategy for upscaling of early infant testing compared with SOC.

FUNDING

Bill & Melinda Gates Foundation, Unitaid, National Institute of Allergy and Infectious Diseases, National Institute of Child Health and Human Development, WHO, and Massachusetts General Hospital Research Scholars.

Metamodeling for Policy Simulations with Multivariate Outcomes

PURPOSE

Metamodels are simplified approximations of more complex models that can be used as surrogates for the original models. Challenges in using metamodels for policy analysis arise when there are multiple correlated outputs of interest. We develop a framework for metamodeling with policy simulations to accommodate multivariate outcomes.

METHODS

We combine 2 algorithm adaptation methods-multitarget stacking and regression chain with maximum correlation-with different base learners including linear regression (LR), elastic net (EE) with second-order terms, Gaussian process regression (GPR), random forests (RFs), and neural networks. We optimize integrated models using variable selection and hyperparameter tuning. We compare the accuracy, efficiency, and interpretability of different approaches. As an example application, we develop metamodels to emulate a microsimulation model of testing and treatment strategies for hepatitis C in correctional settings.

RESULTS

Output variables from the simulation model were correlated (average ρ = 0.58). Without multioutput algorithm adaptation methods, in-sample fit (measured by R2) ranged from 0.881 for LR to 0.987 for GPR. The multioutput algorithm adaptation method increased R2 by an average 0.002 across base learners. Variable selection and hyperparameter tuning increased R2 by 0.009. Simpler models such as LR, EE, and RF required minimal training and prediction time. LR and EE had advantages in model interpretability, and we considered methods for improving the interpretability of other models.

CONCLUSIONS

In our example application, the choice of base learner had the largest impact on R2; multioutput algorithm adaptation and variable selection and hyperparameter tuning had a modest impact. Although advantages and disadvantages of specific learning algorithms may vary across different modeling applications, our framework for metamodeling in policy analyses with multivariate outcomes has broad applicability to decision analysis in health and medicine.

When Is Mass Prophylaxis Cost-Effective for Epidemic Control? A Comparison of Decision Approaches

BACKGROUND

For certain communicable disease outbreaks, mass prophylaxis of uninfected individuals can curtail new infections. When an outbreak emerges, decision makers could benefit from methods to quickly determine whether mass prophylaxis is cost-effective. We consider 2 approaches: a simple decision model and machine learning meta-models. The motivating example is plague in Madagascar.

METHODS

We use a susceptible-exposed-infectious-removed (SEIR) epidemic model to derive a decision rule based on the fraction of the population infected, effective reproduction ratio, infection fatality rate, quality-adjusted life-year loss associated with death, prophylaxis effectiveness and cost, time horizon, and willingness-to-pay threshold. We also develop machine learning meta-models of a detailed model of plague in Madagascar using logistic regression, random forest, and neural network models. In numerical experiments, we compare results using the decision rule and the meta-models to results obtained using the simulation model. We vary the initial fraction of the population infected, the effective reproduction ratio, the intervention start date and duration, and the cost of prophylaxis.

LIMITATIONS

We assume homogeneous mixing and no negative side effects due to antibiotic prophylaxis.

RESULTS

The simple decision rule matched the SEIR model outcome in 85.4% of scenarios. Using data for a 2017 plague outbreak in Madagascar, the decision rule correctly indicated that mass prophylaxis was not cost-effective. The meta-models were significantly more accurate, with an accuracy of 92.8% for logistic regression, 95.8% for the neural network model, and 96.9% for the random forest model.

CONCLUSIONS

A simple decision rule using minimal information about an outbreak can accurately evaluate the cost-effectiveness of mass prophylaxis for outbreak mitigation. Meta-models of a complex disease simulation can achieve higher accuracy but with greater computational and data requirements and less interpretability.

HIGHLIGHTS

We use a susceptible-exposed-infectious-removed model and net monetary benefit to derive a simple decision rule to evaluate the cost-effectiveness of mass prophylaxis.We use the example of plague in Madagascar to compare the performance of the analytically derived decision rule to that of machine learning meta-models trained on a stochastic dynamic transmission model.We assess the accuracy of each approach for different combinations of disease dynamics and intervention scenarios.The machine learning meta-models are more accurate predictors of mass prophylaxis cost-effectiveness. However, the simple decision rule is also accurate and may be a preferred substitute in low-resource settings.

Socioeconomic inequalities in HIV knowledge, HIV testing, and condom use among adolescent and young women in Latin America and the Caribbean

Objective.

To appraise the presence and magnitude of inter- and intra-country health inequalities related to HIV in Latin America and the Caribbean (LAC) among young females.

Methods.

We analyzed household surveys in twenty LAC countries, that included data from female adolescents and young women (ages 15-24) between 2008 and 2018, measuring inequality with the concentration index of 4 indicators: 1) whether individuals have heard of HIV or not, 2) a composite variable of correct knowledge, 3) reported condom use with the last partner, and 4) whether individuals were ever tested for HIV.

Results.

Participants from households in countries with higher socioeconomic status are more likely to have heard of HIV, have correct knowledge of HIV transmission, and have used condoms during their last sexual intercourse. The inter-country concentration index for those indicators were 0.352, 0.302 and 0.110, respectively.

Conclusions.

Economically disadvantaged female adolescents and young women in LAC face an increased risk for HIV, as they are less aware of HIV and its actual transmission mechanism and are less likely to use condoms with their sexual partners. There is an urgent need to tailor prevention strategies of sexually transmitted infections and HIV for adolescents and young women that are sensitive to their socioeconomic context.

Age-specific mortality rate ratios in adolescents and youth aged 10-24 years living with perinatally versus nonperinatally acquired HIV

OBJECTIVE

To measure mortality incidence rates and incidence rate ratios (IRR) in adolescents and youth living with perinatally acquired HIV (YPHIV) compared with those living with nonperinatally acquired HIV (YNPHIV), by region, by sex, and during the ages of 10-14, 15-19, and 20-24 years in IeDEA.

DESIGN AND METHODS

All those with a confirmed HIV diagnosis, antiretroviral therapy (ART)-naive at enrollment, and who have post-ART follow-up while aged 10-24 years between 2004 and 2016 were included. We estimated post-ART mortality incidence rates and 95% confidence intervals (95% CI) per 100 person-years for YPHIV (enrolled into care <10 years of age) and YNPHIV (enrolled ≥10 years and <25 years). We estimate mortality IRRs in a negative binomial regression model, adjusted for sex, region time-varying age, CD4+ cell count at ART initiation (<350 cells/μl, ≥350 cells/μl, unknown), and time on ART (<12 and ≥12 months).

RESULTS

Overall, 104 846 adolescents and youth were included: 21 340 (20%) YPHIV (50% women) and 83 506 YNPHIV (80% women). Overall mortality incidence ratios were higher among YNPHIV (incidence ratio: 2.3/100 person-years; 95% CI: 2.2-2.4) compared with YPHIV (incidence ratio: 0.7/100 person-years; 95% CI: 0.7-0.8). Among adolescents aged 10-19 years, mortality was lower among YPHIV compared with YNPHIV (all IRRs <1, ranging from 0.26, 95% CI: 0.13-0.49 in 10-14-year-old boys in the Asia-Pacific to 0.51, 95% CI: 0.30-0.87 in 15-19-year-old boys in West Africa).

CONCLUSION

We report substantial amount of deaths occurring during adolescence. Mortality was significantly higher among YNPHIV compared to YPHIV. Specific interventions including HIV testing and early engagement in care are urgently needed to improve survival among YNPHIV.

Optimizing infant HIV diagnosis with additional screening at immunization clinics in three sub-Saharan African settings: a cost-effectiveness analysis

INTRODUCTION

Uptake of early infant HIV diagnosis (EID) varies widely across sub-Saharan African settings. We evaluated the potential clinical impact and cost-effectiveness of universal maternal HIV screening at infant immunization visits, with referral to EID and maternal antiretroviral therapy (ART) initiation.

METHODS

Using the CEPAC-Pediatric model, we compared two strategies for infants born in 2017 in Côte d'Ivoire (CI), South Africa (SA), and Zimbabwe: (1) existing EID programmes offering six-week nucleic acid testing (NAT) for infants with known HIV exposure (EID), and (2) EID plus universal maternal HIV screening at six-week infant immunization visits, leading to referral for infant NAT and maternal ART initiation (screen-and-test). Model inputs included published Ivoirian/South African/Zimbabwean data: maternal HIV prevalence (4.8/30.8/16.1%), current uptake of EID (40/95/65%) and six-week immunization attendance (99/74/94%). Referral rates for infant NAT and maternal ART initiation after screen-and-test were 80%. Costs included NAT ($24/infant), maternal screening ($10/mother-infant pair), ART ($5 to 31/month) and HIV care ($15 to 190/month). Model outcomes included mother-to-child transmission of HIV (MTCT) among HIV-exposed infants, and life expectancy (LE) and mean lifetime per-person costs for children with HIV (CWH) and all children born in 2017. We calculated incremental cost-effectiveness ratios (ICERs) using discounted (3%/year) lifetime costs and LE for all children. We considered two cost-effectiveness thresholds in each country: (1) the per-capita GDP ($1720/6380/2150) per year-of-life saved (YLS), and (2) the CEPAC-generated ICER of offering 2 versus 1 lifetime ART regimens (e.g. offering second-line ART; $520/500/580/YLS).

RESULTS

With EID, projected six-week MTCT was 9.3% (CI), 4.2% (SA) and 5.2% (Zimbabwe). Screen-and-test decreased total MTCT by 0.2% to 0.5%, improved LE by 2.0 to 3.5 years for CWH and 0.03 to 0.07 years for all children, and increased discounted costs by $17 to 22/child (all children). The ICER of screen-and-test compared to EID was $1340/YLS (CI), $650/YLS (SA) and $670/YLS (Zimbabwe), below the per-capita GDP but above the ICER of 2 versus 1 lifetime ART regimens in all countries.

CONCLUSIONS

Universal maternal HIV screening at immunization visits with referral to EID and maternal ART initiation may reduce MTCT, improve paediatric LE, and be of comparable value to current HIV-related interventions in high maternal HIV prevalence settings like SA and Zimbabwe.

The opioid crisis: need for systems science research

The opioid epidemic in the United States has had a devastating impact on millions of people as well as on their families and communities. The increased prevalence of opioid misuse, use disorder and overdose in recent years has highlighted the need for improved public health approaches for reducing the tremendous harms of this illness. In this paper, we explain and call for the need for more systems science approaches, which can uncover the complexities of the opioid crisis, and help evaluate, analyse and forecast the effectiveness of ongoing and new policy interventions. Similar to how a stream of systems science research helped policy development in infectious diseases and obesity, more systems science research is needed in opioids.

Model-Based Methods to Translate Adolescent Medicine Trials Network for HIV/AIDS Interventions Findings Into Policy Recommendations: Rationale and Protocol for a Modeling Core (ATN 161)

Background

The United States Centers for Disease Control and Prevention estimates that approximately 60,000 US youth are living with HIV. US youth living with HIV (YLWH) have poorer outcomes compared with adults, including lower rates of diagnosis, engagement, retention, and virologic suppression. With Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN) support, new trials of youth-centered interventions to improve retention in care and medication adherence among YLWH are underway.

Objective

This study aimed to use a computer simulation model, the Cost-Effectiveness of Preventing AIDS Complications (CEPAC)-Adolescent Model, to evaluate selected ongoing and forthcoming ATN interventions to improve viral load suppression among YLWH and to define the benchmarks for uptake, effectiveness, durability of effect, and cost that will make these interventions clinically beneficial and cost-effective.

Methods

This protocol, ATN 161, establishes the ATN Modeling Core. The Modeling Core leverages extensive data—already collected by successfully completed National Institutes of Health–supported studies—to develop novel approaches for modeling critical components of HIV disease and care in YLWH. As new data emerge from ongoing ATN trials during the award period about the effectiveness of novel interventions, the CEPAC-Adolescent simulation model will serve as a flexible tool to project their long-term clinical impact and cost-effectiveness. The Modeling Core will derive model input parameters and create a model structure that reflects key aspects of HIV acquisition, progression, and treatment in YLWH. The ATN Modeling Core Steering Committee, with guidance from ATN leadership and scientific experts, will select and prioritize specific model-based analyses as well as provide feedback on derivation of model input parameters and model assumptions. Project-specific teams will help frame research questions for model-based analyses as well as provide feedback regarding project-specific inputs, results, sensitivity analyses, and policy conclusions.

Results

This project was funded as of September 2017.

Conclusions

The ATN Modeling Core will provide critical information to guide the scale-up of ATN interventions and the translation of ATN data into policy recommendations for YLWH in the United States.