Purpose: Acute HIV infection often causes influenza-like illness (ILI) and is associated with high infectivity. Antiretroviral therapy (ART) substantially decreases infectivity and could reduce transmission if people with acute HIV infection could be identified promptly. We estimated the effectiveness and cost-effectiveness of strategies to identify and treat acute HIV infection in men who have sex with men (MSM) in the US.

Method: We developed a dynamic model of the HIV epidemic among MSM aged 13-64 in the US. We estimated the number of new infections, quality-adjusted life-years (QALYs), and costs for three testing approaches: viral load (VL) testing for individuals with ILI, expanded screening with antibody testing, and expanded screening with antibody and VL testing. We included treatment with ART for individuals identified as acutely infected.

Result: At the present rate of HIV-antibody testing, we estimated that 538,000 new infections will occur among MSM over the next 20 years. Expanding antibody screening coverage to 90% of MSM annually reduces new infections by 2.8% and costs $12,582 per QALY gained. Symptom-based VL testing is more expensive than expanded annual antibody testing, but is more effective and costs $22,786 per QALY gained. Combining expanded annual antibody screening with symptom-based VL testing prevents twice as many infections compared to expanded antibody screening alone, at a cost of $29,923 per QALY gained. Adding VL testing to all annual HIV antibody tests costs more than $100,000 per QALY gained. 

Conclusion: Among MSM, use of HIV VL testing in persons with ILI prevents more infections than does expansion of annual antibody screening alone and is inexpensive relative to other screening interventions. Clinicians should consider VL testing in MSM with ILI, in addition to encouraging annual HIV antibody screening.

Purpose: The CDC currently recommends one-time and annual HIV testing regimens for low-risk and high-risk individuals respectively. Since these recommendations were released in 2006, early initiation of highly active antiretroviral therapy (HAART) has become more common.  In light of these developments we re-analyze the optimal HIV testing regimen.

Method: We build a simple mathematical model to find the optimal testing frequency for various risk groups, using annual incidence rates as proxies for risk. We focus on high-risk (1% annual incidence), moderate-risk (0.1% annual incidence) and low-risk (0.01% annual incidence) individuals. The key parameter in our model is the incremental net monetary loss (INML) of delaying detection of an HIV infection by one year.  This parameter incorporates both monetary and health care costs.  We calculate the optimal testing frequency for three values for INML, $4,000, $1000, and $150.  We estimate an INML of $4,000 from a scenario of early HAART initiation and consider an INML of $1000 to be a more conservative value.

Result: With an INML of $4,000, the optimal time between tests is 4 years for low-risk groups, 1.2 years for moderate-risk groups, and 0.4 years for high-risk groups.  For an INML of $1,000, the optimal time between tests is 8 years, 2.4 years, and 0.8 years for low, moderate, and high-risk groups.  The current CDC guidelines are close to the frequencies that would be optimal with an INML equal to $150, an implausibly low value.  For an INML of $150, the optimal time between tests is 20 years for low-risk groups and 2 years for high-risk groups.

Conclusion: With a reasonable INML, our model suggests that HIV testing for low-risk individuals should be more frequent than the one-time testing currently recommended by the CDC.

Purpose: Over 3 million Americans are infected with chronic hepatitis C (HCV), a serious liver disease. Current U.S. guidelines recommend no screening in the general population. There is disagreement among advisory bodies regarding screening of high-risk individuals. We assessed the cost-effectiveness of universal and risk-factor guided HCV screening for asymptomatic U.S. adults (40-60 years old) at a routine medical visit.

Methods: We developed a decision-analytic Markov model that included the natural history of chronic HCV (genotype 1, 2, or 3) and advanced liver disease. We assessed the lifetime costs (2010 USD), quality adjusted life-years (QALYs) gained, and incremental cost-effectiveness ratios (ICERs) of three screening strategies: no screening, risk-factor guided screening, and universal screening. Risk factors included combinations of history of drug use, blood transfusion prior to 1992, and sexual behaviors. Analyses of the (1999-2008) National Health and Nutrition Examination Survey data provided gender- and age-specific HCV and risk factor prevalence estimates among HCV negative and positive individuals. Those individuals identified via screening who are HCV positive and eligible for treatment receive either standard therapy (peginterferon alfa and ribavirin) in the base case or standard therapy in combination with a recently-developed protease inhibitor as a scenario analysis.

Results: For men, universal screening has an ICER of $42,900/QALY compared to no screening. In order for risk-factor guided screening to be cost-effective, ≥80% of high-risk individuals must truthfully report their status. Even if all high-risk individuals reported truthfully, universal screening is still cost-effective ($47,400/QALY). For women, universal screening has an ICER of $69,100/QALY compared to no screening. Risk-based screening has an ICER approaching $100,000/QALY even if 80% of high-risk individuals truthfully reported. Newer treatments improve incremental cost-effectiveness ratios relative to standard therapy. Screening is less cost-effective for individuals above age 50 because HCV prevalence peaks around 50 years.  Low treatment acceptance, disutility of knowing one’s HCV status, and high treatment costs erode screening cost-effectiveness.

Conclusions: Universal screening is likely cost-effective for both men and women at a willingness to pay threshold of $100,000/QALY. The efficiency of risk-factor guided screening depends strongly on efficiently identifying most high-risk individuals. These findings suggest that existing U.S. HCV screening guidelines should be reconsidered.

Purpose: In a recent randomized controlled trial, daily oral preexposure chemoprophylaxis (PrEP) has been shown to be very effective for HIV prevention in men who have sex with men (MSM), and the US Centers for Disease Control and Prevention (CDC) recently provided interim guidance for its use among MSM who are at high risk for sexual acquisition of HIV.  Previous studies failed to reach a consistent estimate of its cost-effectiveness.

Method: We used an epidemic modeling framework combined with detailed economic analysis to estimate costs and health outcomes for various PrEP strategies.  We developed a dynamic model of the HIV epidemic among MSM aged 13-64 in the US, with annual HIV incidence of 0.8% in the base case, representing an average across the US.  We assumed in the base case that PrEP reduces HIV infection risk by 73%, as seen among MSM reporting high adherence to PrEP.  We estimated the number of new infections, quality-adjusted life-years (QALYs), costs, and incremental cost-effectiveness ratio for each strategy. 

Result: If PrEP is initiated in 20% of the MSM population in the US, we estimate a reduction in new HIV infections of 21% and a gain of 893,000 QALYs over 20 years at a cost of $103,000 per QALY, given an effectiveness of 73%.  Initiating PrEP in a larger proportion of the MSM population averts more infections but at increasing cost per QALY gained (more than $120,000 per QALY gained when at least 60% of the population is placed on PrEP).  If PrEP is 44% effective in reducing infection risk, new HIV infections are reduced by 13% and PrEP costs $168,000 per QALY gained.  PrEP has a more favorable incremental cost-effectiveness ratio in sub-groups of MSM with higher incidence, costing less than $50,000 per QALY gained when annual incidence is greater than 1.5%.

Conclusion: Use of PrEP for HIV prevention in the general MSM population is modestly expensive, but PrEP may be cost-effective by conventional standards in high-risk sub-groups of MSM.

Purpose: Calculating cases of illnesses caused by droplet infections and evaluating the influence of interventions requires dynamic simulation models. The aim of the work is to develop a module to simulate social interaction in epidemic disease propagation and show that models using such complex structures can provide different and more accurate results than calculations neglecting social networks.

Method: Data from EU-project POLYMOD (SP22-CT-2004-502084) about contacts between people and their location is thoroughly analyzed. We use agent-based modeling to create the social interaction sub model due to the very inhomogeneous contact structure as well as the necessity to create a flexible, extensible module. Data from Statistik Austria and structural knowledge about places is used to create different work places, schools, households and places for leisure activities. Each place type has its own structure. For example the place type school defines a structure which consist of several classes with pupils and teachers which change classes according to their movement rules. The social interaction model uses many realizations of the place type school with different parameters for school and class sizes as well as the age structure of the pupils according to data from Statistik Austria. The spread of the disease happens through contacts between infected and susceptible people who are at the same place at the same time.

Result: Social networks are established through places where people meet regularly. The model does not only allow simulating the quickness of the spread of diseases but also locate places where many potential infectious contacts occur. It also helps identifying key people who are responsible for many infections as well as simulating the outcome of interventions. Actual this social module is used in a model for influenza. At some places and especially in schools many infections occur. Simulating scenarios where teachers are vaccinated or schools are closed show that the pace of the epidemic can be slowed down.

Conclusion: Spread of diseases through contacts between individuals can be more properly assessed when simulating contact places. Especially the evaluation and simulation of interventions which target only certain population groups or locations and therefore affect some people more than others benefit from social interaction models. Due to the modular design the contact model can be adapted and used for other droplet infections.