Potential Risks and Benefits of Low-Barrier Access and Monitoring for HIV Preexposure Prophylaxis: A Modeling Study
A stochastic network-based modeling study of over-the-counter (OTC) tenofovir disoproxil fumarate/emtricitabine (TDF/FTC) pre-exposure prophylaxis (PrEP) among gay, bisexual, and other men who have sex with men (GBMSM) in Atlanta.
OFFSCRIPT, NIH/NIMH R21 MH136855. Built on the EpiModel / EpiModelHIV-p platform.
Introduction. PrEP is highly effective for HIV prevention, but its impact is limited by access barriers. Low-barrier delivery strategies for daily oral TDF/FTC, such as over-the-counter availability, could improve access by eliminating clinic visits, though they may increase risks from reduced monitoring, including drug resistance, renal toxicity, and hepatitis B virus (HBV) reactivation. We estimated the population-level benefits and risks of introducing OTC PrEP to complement clinic-based care.
Methods. Using a stochastic network-based model of HIV and sexually transmitted infection (STI) transmission among GBMSM in Atlanta, we simulated two delivery pathways: standard prescription PrEP (clinic-based PrEP) and OTC PrEP with simplified initiation and reduced monitoring. We modeled 10% to 50% increases in total PrEP coverage over 10 years. Outcomes included HIV infections averted, STI incidence, contraindicated PrEP use with subnormal renal function (GFR < 60 mL/min/1.73 m2), HBV reactivations, and drug resistance.
Results. OTC PrEP increased coverage and reduced HIV incidence across all scenarios. A 50% coverage increase averted 15.4% of HIV infections over 10 years. However, clinical risks increased: renal impairment exposure rose from 2.5% to 4.8%, HBV reactivations reached 2.27 per 100 person-years, and TDF/FTC resistance increased to 8.40 events per 100 HIV infections. Sensitivity analyses showed that confirming HIV-negative status at initiation reduced resistance by more than 90% without diminishing prevention benefits, while frequent renal monitoring substantially reduced renal risk.
Conclusions. OTC PrEP could produce meaningful population-level HIV prevention benefits if it expands PrEP coverage beyond clinic-based care. Its net public health impact depends on implementation choices that preserve key safety functions, particularly HIV testing at initiation and targeted renal monitoring, without undermining gains in access.
The analysis is organized around one benefit and three safety levers: prevention benefit scales with net coverage, while each clinical risk is governed by a specific implementation parameter rather than by OTC access itself.
- Benefit scales with net coverage. Introducing OTC PrEP alongside clinic-based PrEP reduced HIV incidence in every scenario. The percentage of infections averted (PIA) rose from 3.2% at a +10% coverage increase to 15.4% at +50% over 10 years, with the number needed to treat roughly stable (about 45 to 48). Prevention benefit was governed primarily by coverage and persistence rather than by per-user adherence.
- Resistance is governed by HIV testing at initiation. Combined TDF/FTC resistance rose to 8.40 events per 100 HIV infections at +50% coverage, concentrated among OTC initiations without HIV status confirmation. Confirming HIV-negative status at initiation was the dominant lever: raising confirmation from 0% to 100% cut FTC resistance by more than 90% (9.39 to 0.82 per 100 infections) without reducing prevention benefits.
- Renal risk is governed by monitoring frequency. Exposure to TDF/FTC with subnormal renal function (GFR < 60 mL/min/1.73 m2) rose from 2.5% to 4.8% as coverage increased, and renal monitoring frequency was the dominant lever for containing it. HBV reactivations reached 2.27 per 100 person-years among individuals with chronic HBV.
| Path | Contents |
|---|---|
| R/ | EpiModelHIV-p analysis pipeline: A networks, B and C netsim exploration and scenarios, D restart point, E and F intervention exploration and production scenarios, Z calibration (swfcalib), plus 00-setup.R, shared_variables.R, hpc_configs.R, netsim_settings.R, and z-gfr_optim.R for the renal module |
data/input/ |
Model input parameter files (model_parameters.csv, scenarios.csv) |
Runtime environment: R 4.5, with dependencies managed through renv (renv.lock for local, renv.lock.hpc for the cluster) and pixi (pixi.toml). Simulations run locally for development and on a SLURM high-performance computing cluster for production scenarios.
- Samuel M. Jenness (PI), Emory University
- Adrien Le Guillou (analyst), Emory University and University of Reims
- Julia L. Marcus, Harvard Medical School
- Douglas S. Krakower, Harvard Medical School and Beth Israel Deaconess Medical Center
- Lauren A. Violette, Merck
- Modeling platform: EpiModel and EpiModelHIV-p.
- Data sources: ARTnet for sexual network parameterization; 2019 Atlanta and Georgia HIV and STI surveillance for calibration targets; external clinical literature for the renal, HBV, and resistance modules.
- Funding: NIH/NIMH R21 MH136855.