This tool is currently under development.
HCV Policy Simulations for Prisons (HCV-PSP) is a web-based tool that allows US state prisons or prison systems to estimate and compare the population health benefits and budgetary impact of a variety of strategies for hepatitis C (HCV) testing and treatment in their facilities. All costs are from the prison perspective.
The findings and conclusions described in this web application and linked journal article are those of the author(s) and do not necessarily represent the views of the U.S. Centers for Disease Control and Prevention. This web tool was funded by the CDC, National Center for HIV, Viral Hepatitis, STD, and TB Prevention Epidemiologic and Economic Modeling Agreement(NEEMA, # 5U38PS004644).
US state prison systems, policy makers, public health researchers and practitioners.
Users can modify a set of input variables (defined below), including population characteristics, intervention parameters and costs. Results are displayed graphically, and graphs update automatically to reflect new selections by the user. Visualizations can be downloaded in png file formats, and the estimates underlying the graphs can be downloaded as tables in xlsx format.
All testing strategies assume that testing is offered to all individuals already incarcerated at the time of implementing the policy, as well as to all future intakes to the prison.
Definitions for HCV testing strategies in the tool:
Risk-based testing: Offer HCV testing to all individuals born from 1945-1965 as well as to all individuals who report a history of injection drug use in the past six months.
Test all: Offer HCV test to all individuals currently incarcerated and to all those entering the prison, without further effort to identify risk factors.
All treatment strategies assume that in order for a person to be treated for HCV in prison, that individual must meet the user-specified clinical criteria for treatment eligibility AND be expected to remain in the facility for at least long enough for the individual to complete the medication for HCV if (s)he is deemed clinically eligible for treatment.
For “treat F3+” and “treat F2+” strategies, persons who are not eligible for treatment based on fibrosis stage are restaged annually until they either reach eligibility based on fibrosis, or are released from prison.
Here are the definitions for HCV treatment strategies in the tool:
Treat F3+: Offer HCV treatment only to those individuals who are chronically infected and who have evidence of at least Metavir fibrosis stage 3 (F3).
Treat F2+: Offer HCV treatment only to those individuals who are chronically infected and who have evidence of at least Metavir fibrosis stage 2 (F2).
Treat all: Offer HCV treatment to all infected individuals regardless of liver fibrosis stage at the time when considering treatment.
Strategy 1 (S1): No testing, no treatment
Strategy 2 (S2): Risk-based testing, treat F3+
Strategy 3 (S3): Risk-based testing, treat F2+
Strategy 4 (S4): Risk-based testing, treat all
Strategy 5 (S5): Test all, treat F3+
Strategy 6 (S6): Test all, treat F2+
Strategy 7 (S7): Test all, treat all
Population Size: The number of individuals incarcerated in your prison or your prison system at baseline. This input represents the size of the existing population who could be tested for HCV and treated for HCV today.
Number of new intakes per year: On average, the number of intakes to the prison or prison system each year. This input represents the annual number of future entrants to the system who would need to be tested and potentially treated for HCV.
Chronic HCV prevalence (%): Among those who are currently in prison or prison system, your best estimate of prevalence of chronic HCV infection (RNA+).
Percent of prison population reporting any history of illicit drug use (%): The percentage of the incarcerated population that reports either having a current, active substance use disorder (injection or non-injection), or having a history of substance use disorder. Based on a study of drug use behaviors among incarcerated persons (Rowell-Cunsolo 2016), the model assigns a portion of this total illicit drug use population to having a history of injection drug use in the previous 6 months. Only those who have a history of specifically injection drug use in the past six months are targeted to testing when implementing “risk-based testing.”
Percent of patients who are expected to reside within the prison system for long enough to be considered for HCV treatment (%): TSome prison systems require that an individual be present for a threshold number of weeks/months if they are going to initiate HCV treatment. The intention is to avoid situations in which an individual begins disease staging, preparations for therapy, and treatment initiation and is then discharged such that (s)he cannot complete the treatment course. This threshold time requirement varies by system. This input asks the user to define the percentage of the entire prisons population that is expected to reside in the prison for longer than that threshold.
Linkage to HCV-specific care within prison (%): Of those identified as having chronic HCV infection, the percentage that will likely present to the prison HCV provider such that (s)he could be disease staged and treated. Not all patients who are identified as infected will necessarily seek treatment.
Cost of HCV antibody test ($): Cost to the prison or prision system of one HCV antibody testing. This cost includes both the cost of the test reagents themselves, as well as any staff, laboratory costs, and overhead employed for testing.
Cost of HCV RNA test ($): Cost to the prison or prision system of one HCV RNA testing. This cost includes both the cost of the test reagents themselves, as well as any staff, laboratory costs, and overhead employed for testing.
HCV treatment cost, first treatment, non-cirrhotic ($): Total cost of medications for treating HCV among patients who have never been treated before and who do not have cirrhosis. In the simulation, we assume that this is a 8-week regimen.
HCV treatment cost, first treatment, cirrhotic ($): Total cost of HCV medications for treating HCV among patients who have never been treated before and who do have cirrhosis. In the simulation, we assume that this is a 12-week regimen.
HCV treatment cost, retreatment, non-cirrhotic ($): Total cost of HCV medications for treating HCV among patients who have previously been treated but did not attain HCV cure and who do not have cirrhosis. In the simulation, we assume that this is a 12-week regimen.
HCV treatment cost, retreatment, cirrhotic ($): Total cost of HCV medications for treating HCV among patients who have previously been treated but did not attain HCV cure and who do have cirrhosis. In the simulation, we assume that this is a 12-week regimen.
Number of HCV diagnoses in prison: The projected number of HCV cases that the prison or prison system will identify under a given screening strategy. Each infected individual is counted only one time. Individuals who are potentially re-screened and re-identified count as only one case identified.
Number of HCV infections cured in prison: The projected number of infections who will attain HCV sustained virologic response (cure) as the result of being treated while in prison, under a given screening and treatment strategy. A person who is infected twice and is cured twice will count as two infections cured.
Total healthcare cost in prison ($): The total cost of healthcare utilization for all individuals while in prison - including HCV screening, linkage to care, and treatment, and all other healthcare needs. Cost assumes the prison/prison system budgetary perspective.
HCV screening cost in prison ($): One component of “total healthcare cost in prison” focused only on the cumulative cost of HCV screening under a given screening strategy, from the prison/prison system perspective.
HCV medications cost in prison ($): The cost of all HCV medications when implementing the given HCV screening and treatment strategy.
This web-based tool is based upon the HEP-CE microsimulation model, which has been published in Clinical Infectious Diseases (Assoumou et al (2019): Cost-effectiveness and Budgetary Impact of Hepatitis C Virus Testing, Treatment, and Linkage to Care in US Prisons, Clinical Infectious Diseases, https://academic.oup.com/cid/advance-article-abstract/doi/10.1093/cid/ciz383/5490662?redirectedFrom=fulltext). For additional methodological details, see the “Methods” tab.
The tool includes two download buttons to download graphs (.png) and the numerical estimates underlying the graphs (.xlsx).
Section 508 requires Federal agencies and grantees receiving Federal funds to ensure that individuals with disabilities who are members of the public or Federal employees have access to and use of electronic and information technology (EIT) that is comparable to that provided to individuals without disabilities, unless an undue burden would be imposed on the agency. If you need assistance with this web application, please contact email@example.com
The estimates shown by HCV-PSP are based on a metamodel of the original HEP-CE microsimulation model (Assoumou et al (2019). Cost-effectiveness and Budgetary Impact of Hepatitis C Virus Testing, Treatment, and Linkage to Care in US Prisons, Clinical Infectious Diseases, https://academic.oup.com/cid/advance-article-abstract/doi/10.1093/cid/ciz383/5490662?redirectedFrom=fulltext). The computational complexities in the original model limit its application in real time prediction. We choose to use a metamodeling strategy to approximate the relationships between inputs and outputs in the full simulation model while also allowing instant predictions of key outcomes based on a small number of customized input values. To balance simplicity and performance, we utilized elastic net (a combination of ridge and lasso regression) with interaction and second order terms to predict each outcome variable individually based on the assumption that all outcome variables are independent. We further tuned the weights for ridge and lasso in order to improve prediction performance. Due to the difficulty of constructing confidence intervals for elastic net predictors, we utilized the root mean squared errors of predictions in the testing dataset as proxies for prediction uncertainty intervals. In addition, our models are constructed from a simulation dataset in which the population size is 1 Million, but we provide predictions and uncertainty intervals according to the user input population size and yearly prisoner intake number.
Please see the full paper (https://academic.oup.com/cid/advance-article-abstract/doi/10.1093/cid/ciz383/5490662?redirectedFrom=fulltext) for the sources used to select input variable ranges.