Introduction

This report presents the findings from the Washington HIV/AIDS Modeling for Prevention project (WHAMP). The goal of the project was to develop epidemiological modeling tools for projecting the next 10 years of the HIV epidemic among MSM in Washington State, track the costs of the state’s two major treatment and prevention programs, and create a tool for projecting the impact of changes in external subsidies.

Data sources

In order for this epidemiological model to inform local decision making, it must be based on local data. Six sources of data are used here:

  1. Local demographics

    We estimated the number of MSM in WA State by 10 yr age-group x race (Black, Hispanic, Other) x region (Eastern WA, King Co, Western WA). The data for this come from two sources:

  2. The WHAMP survey

    The WHAMP survey was conducted over a three month period using social media (primarily FaceBook, with a small number of Grindr and Growlr respondents) from Sep 11 - Dec 11 2019. These data were used for parameterizing the sexual network dynamics, sexual behaviors within partnerships (e.g., condom use and act rates), HIV testing rates, and PrEP dynamics. ART dynamics were largely drawn from WADOH surveillance data sources, since ART use was nearly universal among HIV+ respondents.

  3. WADOH Drug Assistance Program Client data

    There are two primary state-funded assistance programs for HIV prevention and care: the ART drug assistance program (ADAP) and the PrEP drug assistance program (PDAP). PDAP also provides subsidies for wrap-around around care (regular checkups and testing).
    DOH provided line-level client data on enrollment days and costs for program participants from 2014-2018 to ZK, and an additional set of data from EHARS on viral load. The client data were used to estimate enrollment/disenrollment dynamics and per person-day program costs. The EHAS data were used to compare VL suppression rates for ADAP particpants to non-particiants.

  4. WADOH Surveillance data on treatment outcomes

    DOH provided aggregate data to DRW for 2018 on time to VL suppression, and rates of engagement in care. These data were used to parameterize the ART continuum.

  5. WADOH Diagnosed HIV: New cases and Prevalent cases

    DOH provided aggregate data on the age x race x region distribution of Dx HIV to MM. These data are used as calibration or validation targets.

This online document collects and organizes reports from analyses conducted as part of the workflow.

Demographics

  • Estimates of MSM by age, race and region Reproducing the demographics of MSM in WA State, and projecting them into the future, raises a number of challenges. The demographics, explanation of the problems and solutions we adopted are found here.

WA State Dx HIV 2014-19

These estimates are based on the WA state surveillance data. They are used for calibration or validation targets.

WHAMP survey

These data are used for constructing both input parameters and validation targets for the simulation.


Care continuum

Testing

  • Descriptive analysis – This includes basic descriptives of the WHAMP survey data, and exploratory modeling. The exploratory models include a latent class mixture model, and fixed effects models comparing Cox PH, exponential, weibull and log-log specifications.

  • Testing parameters – This is the Gomperz model ultimately used for parameterizing the testing hazard for “general testers” (i.e., not on PrEP, and not late-testers).

  • Testing targets

PrEP continuum

  • PrEP parameters – Calculates the input parameters for the PrEP dynamics: awareness, initiation, stopping (quit + pause), and re-intiation.

  • PrEP targets

ART continuum

  • ART parameters – ART dynamics are: intiation, interruption, re-initation.

    ART use among HIV+ in the WHAMP survey is universal, so these data are not used to estimate input parameters, only targets. The input parameters for the ART continuum are taken from DRWs analysis of DOH surveillance data. Her notes can be found here

  • ART targets

    Note: participation in ADAP modifies the rate of VL suppression, and these targets are taken from ZKs analysis of DOH ADAP+EHARS data.

  • Clinical outcome

    The primary outcome is viral load. This is treated as a trajectory, with 3 possible paths: partial suppression, full suppression, and durable full suppression. Parameters are taken from DRWs analysis of DOH VL surveillance data.

DAP Participation & Costs

We have DAP participation estimates from two sources: the WHAMP survey, and the DOH DAP Claims data.

  • WHAMP DAP parameters

    There is only one pair of parameters we use from WHAMP survey data: the overall DAP enrollment fractions (ADAP fraction of ART users, and PDAP fraction of PrEP users). These will serve as calibration parameters (the target will be overall program size, taken from the DOH DAP client dats). The reason WHAMP can’t be used for more is the small subgroup sizes available in the survey. The ADAP subgroup is limited by the number of the HIV+ in the survey sample, and the PDAP subgroup is limited by the number of PrEP users.

  • DOH DAP parameters

    The DOH DAP client database is used for everything else: program disenrollment rates, costs, and target program size and composition.