New example: PrEP intervention for HIV prevention

[smjenness]

Motivation

The existing HIV gallery example explicitly suggests PrEP as a next step. Pre-exposure prophylaxis is the dominant biomedical HIV prevention strategy globally and demonstrates intervention mechanics that are distinct from both vaccination (PrEP is continuous and reversible, not one-shot) and treatment (PrEP targets susceptible individuals, not infected ones). EpiModel is widely used for PrEP modeling in practice — a gallery example would help new users get started.

Suggested Design

Disease Model: SI with HIV Stages + PrEP Cascade

Build on the existing HIV example's stage structure (acute/chronic/AIDS) and add a PrEP cascade for susceptible individuals:

Attribute	States	Description
status	s, i	Susceptible or infected
stage	acute, chronic, AIDS	HIV disease stage (infected only)
prep.status	NA, indicated, active, discontinued	PrEP cascade state (susceptible only)
ART.status	NA, on, off	ART status (infected only)

Custom Modules

Module	Purpose
infection.FUN	S → I transmission; PrEP-active individuals have reduced susceptibility (prep.efficacy)
progress.FUN	HIV stage progression: acute → chronic → AIDS
prep.FUN	PrEP cascade: indication → uptake → adherence/discontinuation. Handles screening for PrEP eligibility, initiation, and loss to follow-up.
treatment.FUN	ART initiation for diagnosed HIV+ individuals
departures.FUN	Background + disease-specific mortality (AIDS stage)
arrivals.FUN	New sexually active individuals entering the population

Parameters

Parameter	Description	Suggested Value
inf.prob.chronic	Base per-act transmission probability (chronic stage)	0.002
inf.prob.acute.mult	Relative infectiousness during acute stage	5–10
inf.prob.AIDS.mult	Relative infectiousness during AIDS	3
act.rate	Acts per partnership per week	2
prep.indication.rate	Weekly probability an eligible susceptible is identified for PrEP	0.01–0.05
prep.uptake.rate	Weekly probability an indicated individual starts PrEP	0.1–0.3
prep.efficacy	Reduction in susceptibility while on PrEP (multiplicative)	0.90–0.96
prep.discontinuation.rate	Weekly probability of stopping PrEP	0.005–0.02
prep.reinitiation.rate	Weekly probability of restarting PrEP after discontinuation	0.005
ART.initiation.rate	Weekly probability an HIV+ individual starts ART	0.01
ART.suppression.efficacy	Reduction in infectiousness while on ART	0.95

ERGM Parameterization

Model a population with heterogeneous sexual activity levels, where PrEP targeting matters:

• ~edges: Mean degree ~0.7
• ~concurrent: Moderate concurrency (characteristic of MSM networks)
• ~nodefactor("risk_group"): Two risk groups (high/low activity). High-risk group has higher mean degree.
• ~nodematch("risk_group"): Moderate assortative mixing by risk group (high-risk individuals preferentially partner with each other, but some bridging occurs)
• Dissolution ~offset(edges): Mean partnership duration ~40 weeks for main, ~10 weeks for casual

This setup makes PrEP targeting interesting: should PrEP be offered to everyone equally, or targeted to the high-risk group? The ERGM structure determines how much targeting matters.

Scenarios

1. Baseline (no PrEP, no ART): Uncontrolled HIV epidemic showing acute/chronic/AIDS dynamics
2. ART only: Treatment-as-prevention baseline — what ART alone achieves
3. ART + untargeted PrEP: PrEP offered to all susceptibles at equal rate
4. ART + targeted PrEP: PrEP preferentially offered to high-risk group (higher indication rate for high-risk)
5. PrEP with high discontinuation: Same as scenario 3 but with higher discontinuation rate — demonstrate that adherence/retention is as important as coverage

Analyses

• Compare HIV incidence and prevalence across scenarios over 10+ years
• Infections averted by PrEP (relative to ART-only baseline)
• PrEP coverage over time (active PrEP users / susceptible population)
• Number needed to treat (NNT): PrEP person-years per infection averted
• Demonstrate that targeted PrEP to the high-risk group is more efficient (lower NNT) than untargeted
• Show that high discontinuation rates severely undermine PrEP effectiveness

Relationship to Existing Examples

• Direct extension of the HIV gallery example (reuses stage structure and ART module)
• Demonstrates a susceptible-side intervention (contrast with treatment-side interventions in TestAndTreatIntervention and HIV ART)
• Introduces a cascade module with multiple states (indicated → active → discontinued → reinitiated)
• Introduces risk group heterogeneity and targeted intervention (not in any current example)