This section combines the components of Sections 1-3 to run the model and produce the raw outputs, which will be used for analysis. After loading of model functions (section 4.0 below), analysis takes place over two subcomponents - deterministic model analysis of combinations of interventions and deterministic model analysis of single interventions. We also conduct a probability sensitivity analysis (PSA) for each component. This produces CEA results and quantifies the uncertainty of our inputs and decisions.
\(~\)
The core model functions are automatically loaded when users install the LEMHIVpack R package, and load library(LEMHIVpack):
devtools::install_github("HERU-LEM/LEMHIVpack") library(LEMHIVpack)
Some key base model values (i.e. interventions and CEA parameters, set time period, time steps, discount rate etc.) for CascadeCEA-Interventions-1-LoadBaselineWorkspace.R and what they are currently set to are provided in Table 4.0.
\(~\)
Table 4.0: Sample base model parameters with their R name and value
| Description | R name | Value |
|---|---|---|
| Intervention sustainment duration | int.sus |
10 |
| Start year | int.first.year |
2020 |
| End of projection year | lyr |
2040 |
| Scale up period | scale.up.period |
18 months |
| Discount rate for cost and QALY | Discounting |
0.03 |
\(~\)
CascadeCEA-Interventions-1-LoadParameterWorkspace-Combination.R sets global and free model parameters such as intervention scales and effectiveness, city and state level costs, based on calibrated values for the respective parameter. Also in this script - we call model inputs from files named Evidence-Inputs-Master.xlsx and Evidence-Inputs-Master-Ideal.xlsx stored in the data directory, and initialize the model. Finally, we set parameter names using the CascadeCEA-Interventions-1-ParNamesExportIntModel.R script found in the R directory. Note that the scripts included in 01_Setup and main R folders of the github repository will automatically be called by subsequent modules, and do not need to be run on their own.
\(~\)
In this component (use CascadeCEA-Interventions-2-RunModelAndAccum-Combination-Deterministic.R), we run a deterministic model for estimating accumulated costs and epidemiological outcomes of combinations of interventions for individual cities or all cities. Intervention outcome matrices are produced (outcome.comb.mx), with a list of outcomes and respective R labels are provided in Table 4.1. Note that we used a 20-year time horizon (2020–2040) in our main analysis, to capture the long-term individual benefits of ART and second-order transmission effects (Nosyk et al. 2020). Both costs and QALYs were reported using a 3% annual discount rate.
\(~\)
Table 4.1: Description of model outcomes with R names
| R name | Description |
|---|---|
Infections.total-20Y |
Projected total infections over 20-year time horizon (2020-2040) |
SuscPY-over20Y |
Number of person-years among susceptible population over 20-year time horizon (2020-2040) |
Infections.total-10Y |
Total infections over 10-year time horizon (2020-2030) |
SuscPY-over10Y |
Number of person-years among susceptible population over 10-year time horizon (2020-2030) |
Infections.total-5Y |
Total infections over 5-year time horizon (2020-2025) |
SuscPY-over5Y |
Number of person-years among susceptible population over 10-year time horizon (2020-2030) |
QALYs.sum |
Sum of quality adjusted life years (QALYs) gained |
costs.total.sum |
Sum of total costs |
costs.hru.sum |
Sum of costs for cumulative healthcare resource utilization (HRU) |
costs.art.sum |
Sum of ART costs |
costs.art.ini.sum |
Sum of ART initiation costs |
costs.oat.sum |
Sum of Opioid Agonist Therapy (OAT) costs |
costs.prep.sum |
Sum of PrEP costs |
costs.prep.tests.sum |
Sum of HIV test before beginning PrEP costs |
costs.test.sum |
Sum of HIV testing costs |
int.costs.sum |
Sum of intervention costs |
int.impl.costs.sum |
Sum of intervention costs during implementation period (18 months) |
int.sust.costs.sum |
Sum of sustaining interventions (after 18-month implementation period) costs |
\(~\)
In this component we perform PSA on the deterministic model run done in component 4.2 to evaluate the extent of parameter uncertainty for each intervention. For each city, we used the 2000 best-fitting calibrated parameter sets from 10,000 calibration runs, sampling all non-calibrated parameters simultaneously from distributions that were previously developed for each model parameter (Zang et al. 2020; Krebs et al. 2019). Intervention outcome matrices are produced (outcome.comb.SA.mx), outcomes and respective labels in R are provided in Table 4.1.
To run PSA on the optimal combination of implementation strategy (OCIS) use CascadeCEA-Interventions-2-RunModelAndAccum-Combination-PSA-OCIS.R which can be found in the 02_Run_model directory.
To run PSA on the combinations proximal to the OCIS use CascadeCEA-Interventions-2-RunModelAndAccum-Combination-PSA-Proximal.R which can also be found in the 02_Run_model directory.
\(~\)
This component estimates accumulated costs and epidemiological outcomes for a single intervention rather than a combination of interventions (use CascadeCEA-Interventions-2-RunModelAndAccum-SingleIntervention-Deterministic.R). This component is similar to component 4.2 in the sense that we draw on model functions and data loaded in Sections 1-3. Intervention outcome matrices are produced (outcome.int.mx”), and respective R labels are found in Table 4.1. Users can generate outputs for a single city or all cities.
\(~\)
In this component we perform PSA on the deterministic model run in component 4.4 to evaluate the extent of parameter uncertainty for a single intervention. Intervention outcome matrices are produced (outcome.int.SA.mx), outcomes and respective labels in R are provided in Table 4.1. Use CascadeCEA-Interventions-2-RunModelAndAccum-SingleIntervention-PSA.R found in the R directory.
\(~\)
Krebs, E., B. Enns, L. Wang, X. Zang, D. Panagiotoglou, C. Del Rio, J. Dombrowski, et al. 2019. “Developing a Dynamic Hiv Transmission Model for 6 U.s. Cities: An Evidence Synthesis.” Journal Article. PLoS One 14 (5): e0217559. https://doi.org/10.1371/journal.pone.0217559.
Nosyk, B., X. Zang, E. Krebs, B. Enns, J. E. Min, C. N. Behrends, C. Del Rio, et al. 2020. “Ending the Hiv Epidemic in the Usa: An Economic Modelling Study in Six Cities.” Journal Article. Lancet HIV 7 (7): e491–e503. https://doi.org/10.1016/S2352-3018(20)30033-3.
Zang, X., E. Krebs, J. E. Min, A. Pandya, B. D. L. Marshall, B. R. Schackman, C. N. Behrends, D. J. Feaster, B. Nosyk, and Localized HIV Modeling Study Group. 2020. “Development and Calibration of a Dynamic Hiv Transmission Model for 6 Us Cities.” Journal Article. Med Decis Making 40 (1): 3–16. https://doi.org/10.1177/0272989X19889356.