The impact of timing and prioritization on the cost-effectiveness of birth cohort testing and treatment for hepatitis C virus in the United States


  • Potential conflict of interest: Dr. L'Italien owns stock and is an employee of Bristol-Myers Squibb. Dr. McEwan owns stock in Health Economics & Outcomes Research. Dr. Yuan is an employee of Bristol-Myers Squibb.

  • This study was funded by an unrestricted grant from Bristol-Myers Squibb.


Recent United States guidelines recommend one-time birth cohort testing for hepatitis C infection in persons born between 1945 and 1965; this represents a major public health policy undertaking. The purpose of this study was to assess the role of treatment timing and prioritization on predicted cost-effectiveness. The MONARCH hepatitis C lifetime simulation model was used in conjunction with a testing and treatment decision tree to estimate the cost-effectiveness of birth cohort versus risk-based testing incorporating information on age, fibrosis stage and treatment timing. The study used a 1945-1965 birth cohort and included disease progression, testing and treatment-related parameters. Scenario analysis was used to evaluate the impact of hepatitis C virus (HCV) prevalence, treatment eligibility, age, fibrosis stage and timing of treatment initiation on total costs, quality-adjusted life years (QALYs), HCV-related complications and cost-effectiveness. The cost-effectiveness of birth cohort versus risk-based testing was $28,602. Assuming 91% of the population is tested, at least 278,000 people need to be treated for birth cohort testing to maintain cost-effectiveness. Prioritizing treatment toward those with more advanced fibrosis is associated with a decrease in total cost of $7.5 billion and 59,035 fewer HCV-related complications. Total QALYs and complications avoided are maximized when treatment initiation occurs as soon as possible after testing. Conclusion: This study confirms that birth cohort testing is, on average, cost-effective. However, this remains true only when enough tested and HCV-positive subjects are treated to generate sufficient cost offsets and QALY gains. Given the practical and financial challenges associated with implementing birth cohort testing, the greatest return on investment is obtained when eligible patients are treated immediately and those with more advanced disease are prioritized. (HEPATOLOGY 2013)

Hepatitis C virus (HCV) is a major global public health issue. In the United States, chronic HCV infection is the leading cause of hepatocellular carcinoma (HCC) and liver transplantation.1-5 It is estimated that 3.2 million people are living with chronic HCV in the United States,6 and between 45% and 85% of these people are unaware of their infection.7-10 Historically, the two principal modes of HCV transmission are blood transfusion and injection drug use11; however, after the introduction of routine blood testing in 1992, the predominant route of disease transmission in the United States is now among persons who inject drugs12; with an estimated 17,000 new infections occurring annually.13 In the absence of a robust HCV testing and treatment program in the United States, it is estimated that 1.76 million people with chronic HCV will develop cirrhosis; 400,000 will develop HCC, and more than 1 million will die of an HCV-related death.14


CDC, Centers for Disease Control and Prevention; ESLD, end-stage liver disease; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; NHANES, National Health and Nutrition Examination Survey; QALYs, quality-adjusted life years; SVR, sustained virological response

Risk-based testing guidelines to identify HCV infection were first published in 1998.15 These guidelines advocated testing for HCV in persons at high risk of infection, such as those with a history of injection drug use and those receiving a blood transfusion or organ transplant prior to comprehensive blood screening. Recent modeling studies have presented compelling economic analyses demonstrating that birth cohort screening compared with risk-based screening in the United States is cost-effective.16-18 These findings have led to published Centers for Disease Control and Prevention (CDC) guidelines advocating one-time testing for HCV of all persons born between 1945 and 1965.13

Despite demonstrable cost-effectiveness, there are substantial financial and practical barriers to the widespread implementation of a comprehensive birth cohort testing program. Recent evidence shows only a fraction of HCV infections being identified in high-risk subjects compared with expected levels of identification given risk-based testing guidelines.8 The one-time birth cohort testing approach recommended in recent CDC guidelines would result in the testing of 66.9 million people, identifying 1.1 million treatment-eligible patients.

The health care delivery implications of such a testing policy are substantial. In addition to any direct health care costs, there will be an inevitable increased demand for hepatology-specific manpower expertise. The disparity between the availability of appropriately trained practitioners and the growing need for advanced hepatology care has been acknowledged.19 With the expansion of available treatment options, future therapy regimens will become increasingly individualized to specific patient characteristics, ensuring a continued need for specialist expertise. Any increased patient awareness of HCV infection status will inevitably place additional demands on health care providers.

Furthermore, it is unclear how the timing of testing and treatment would impact cost-effectiveness; for example, following identification of subjects with chronic HCV infection, should treatment-eligible patients maintain watchful waiting or would immediate therapy optimize health outcomes? Defining “optimal” is also problematic; is the testing and treatment policy designed to minimize future HCV-related complications, minimize therapy-related expenditure, or maximize life years and quality-adjusted life years (QALYs) gained? From a public health perspective, the emergence of novel therapeutic approaches to the treatment of HCV infection capable of achieving rates of sustained virological response approaching 100%, even in the most difficult-to-treat patients, means that altering the future transmission dynamics of the disease is entirely feasible.20 Importantly, this development may modify the patient's perspective on treatment initiation. The historic acceptance of watchful waiting, given the side effect and efficacy profile associated with pegylated interferon and ribavirin, is understandable; however, presented with an awareness of HCV infection plus the potential for a cure, it would be reasonable to expect that treatment uptake rates would increase among eligible patients.

Treatment uptake and eligibility have important consequences for the future transmission dynamics of the disease in the United States; the widespread treatment of subjects with HCV has the potential to reduce future transmission patterns.20 It is noteworthy, however, that treatment strategies following testing may have a limited impact on the future dynamics of HCV infection, as it is those most likely to contribute to future HCV infections who are least likely to be eligible for treatment; such as persons who inject drugs, the homeless and the incarcerated. While these subjects are likely to be younger and at relatively low risk of developing end-stage liver disease (ESLD) complications, birth cohort screening is unlikely to be of any major relevance to this subpopulation.

The impact of the size of the tested population, the numbers eligible for treatment, disease stage, and the prioritization and timing of treatment on overall cost-effectiveness is not well understood. Therefore, the principal objective of this study was to estimate the relationship between the cost-effectiveness of a one-time birth cohort testing of the population born between 1945 and 1965 and a risk-based testing of the same population to identify whether a phased time-dependent, age-dependent, and fibrosis stage–dependent treatment program offers value from a health economics perspective. We omitted anyone born outside of the birth cohort population from the analysis, because they were assumed to be tested within the risk-based strategy and thus would be unaffected by the birth cohort program. A secondary objective was to understand how the timing of treatment initiation impacts costs, QALYs and HCV-related complications avoided.

Subjects and Methods

An estimation of the natural history of progression from chronic infection to ESLD was conducted using the MONARCH (MOdelling the NAtural histoRy and Cost effectiveness of Hepatitis C) model. This is a cohort-based Markov lifetime simulation that has been described in detail.21 Additionally, we utilized a testing and treatment decision tree in combination with the MONARCH model to assess the lifetime costs, life years, and QALYs associated with number of testing and treatment-related scenarios.

Tested Population.

We modeled a population comprising all individuals born between 1945 and 1965 in the United States (66.9 million people). From this population, we excluded those previously diagnosed with chronic HCV (∼674,480 people).16 Our analysis compared two testing strategies. First, a risk-based strategy in which those at-risk in the population (persons with a history of injection drug use, recipients of blood clotting factor concentrates produced prior to 1978, blood transfusion or organ transplantation prior to 1992, long-term dialysis, children from HCV-infected mothers and those in occupations that expose them to HCV)15 are tested. The risk-based strategy tests approximately 22.34% (14,793,816 members) of the total population and identifies 1.77% (262,260 people) with chronic HCV.17 Second, the birth cohort testing strategy outlined above is implemented assuming 91.21% (60,404,514 members) of the total population are tested, identifying approximately 1.77% (1,070,840 people) with chronic HCV. In both scenarios, we compare the costs and effects of a one-time testing and treatment program. A flow diagram of the two scenarios is shown in Fig. 1.

Figure 1.

Flow diagram showing the derivation of the number of subjects tested and the number of subjects eligible for treatment for birth cohort and risk-based testing strategies.


Following testing, 24% of those positive for HCV and treatment eligible are treated in the first year, whereas the remainder are scheduled to receive treatment in equal proportions over the subsequent 10-year period; this total treatment period is consistent with the analysis reported by McGarry et al.17 Prior to treatment initiation, a percentage of the cohort may progress and become ineligible for treatment, or may die; this is dependent upon the timing of treatment initiation. We model the treatment of patients with genotype 1 using telapravir in combination with peginterferon-alfa plus ribavirin; genotypes 2 and 3 are treated with peginterferon-alfa plus ribavirin. Sustained virological response (SVR) rates among patients infected with genotype 1 HCV commencing treatment in fibrosis stages F0-F2 and F3-F4 were 0.78 and 0.62, respectively22; for genotype 2 and 3, SVR rates were 0.76, 0.61, and 0.57 in those treated in fibrosis stages F0-F2, F3, and F4, respectively.23


The MONARCH HCV model is designed to progress a cohort of subjects, in annual cycles, through the Metavir disease stages: F0 (no fibrosis) through F1 (portal fibrosis with no septa), F2 (portal fibrosis with few septa), F3 (portal fibrosis with numerous septa), and F4 (compensated cirrhosis) and potentially on to ESLD complications. The model flow diagram is shown in Fig. 2. Progression through fibrosis stages is controlled via stage-specific transition probabilities influenced by duration of HCV infection, age, sex, genotype, source of infection (acting as a proxy for post-acquisition behavior), and excessive alcohol consumption (defined as an average daily consumption >20 g).24 Table 1 reports the transition rates and baseline parameters used in the model. We assumed 75% of chronic HCV infections are genotype 1.17

Figure 2.

Flow diagram of the MONARCH model. Annual transition probabilities control progression through disease states.

Table 1. Annual Disease Progression Rates, Distribution Type, and Parameters Used in the Model
Disease ProgressionMeanSEDistributionParametersReference
  • *

    Proportion of cohort with excess alcohol consumption.

  • Observational or trial setting.

  • Abbreviations: BT, blood transfusion; DC, decompensated cirrhosis; GT1, proportion of cohort genotype 1; HCV age, age at HCV infection; HCVD, HCV duration of infection (years); LT, liver transplantation.

F0 to F10.079 Normalexp [−2.0124 − (0.07589 × HCVD) + (0.3247 × Design) + (0.5063 × Male) + (0.4839 × GT1)]24
F1 to F20.073 Normalexp [−1.5387 − (0.06146 × HCVD) + (0.8001 × Alcohol)]24
F2 to F30.111 Normalexp [−1.6038 + (0.0172 × HCV Age) − (0.05939 × HCVD) + (0.4539 × Alcohol)]24
F3 to F40.051 Normalexp [−2.2898 + (0.01689 × HCV Age) − (0.03694 × HCVD) + (0.5963 × IDU) + (1.1682 × BT) − (0.4652 × GT1)]24
F3 to HCC0.0020.013Normal0.0010.02814
F4 to DC0.0300.010Normal0.0100.05014
F4 to HCC0.0170.014Normal0.0080.04514
F4 SVR relapse0.048     
DC to LT0.0310.008Normal0.0160.04729
HCC to LT0.103    17
DC to death0.1000.051Normal0.1000.20014
HCC to death0.427 Normal0.3410.51230
DC to HCC0.0790.011Beta1.930136.11031
LT (year 1)0.1500.005Beta16.28061.23032
LT (year 2+)0.0300.003Beta22.900378.88032
Model parameters Notes 
 Age, years57Estimated from the midpoint of the birth cohort age range as of 2012.17
 Proportion male0.620 28
 Proportion genotype 10.750 33
 Alcohol*0.367Alcohol, blood transfusion, and injection drug use proportions were estimated from the average of all the United States studies presented in the referenced paper that did not present any parameter bias.24
 Blood transfusion0.29724
 Injection drug use0.46824
 HCVD20.000 Assumed
 Design0.000 Assumed

Subjects enter the model immediately after testing and, in the base case, are distributed across fibrosis stages: (15.0% in F0; 29.5% in F1; 20.3% in F2; 17.1% in F3; 18.1% in F4).17 There are three cohort profiles propagated through the model:

  • 1Subjects undiagnosed. These subjects progress through the model potentially incurring costs for ESLD complications and health utility decrements associated with ESLD complications. We assumed no costs of chronic HCV were incurred by these individuals.
  • 2Subjects diagnosed but not treated. These subjects incur chronic HCV-related and ESLD-related costs and disutilities.
  • 3Subjects diagnosed and treated. These subjects incur chronic HCV-related and treatment-related costs and disutilities; successfully treated patients achieving an SVR are removed to a healthy state, and we assume a normal life expectancy. Patients achieving SVR from fibrosis stage F4 have the potential to relapse within the first year and progress to decompensated cirrhosis and HCC.

The model assumes no difference in fibrosis stage progression rates between those diagnosed and not diagnosed.

The model is run over a lifetime and predicts total costs, QALYs, the number of predicted liver-related complications, and deaths per year. Although our base case analysis focuses on comparing a birth cohort testing and treatment strategy with a risk-based testing and treatment strategy, our primary focus was on the effect of stratifying treatment in those tested by age, fibrosis stage, and time.


The model takes a health care payer perspective and considers only direct medical costs; these are presented in Table 2. HCV specific treatment and monitoring costs are derived from weekly estimated costs accommodating duration of treatment; we assume null responders are treated for 12 weeks only. Testing and healthcare costs were estimated from contemporary U.S sources.17 We assumed that undiagnosed patients would only incur the costs of liver related complications. All costs are independent of age and are discounted at 3.5%.

Table 2. Chronic HCV and ESLD Health State Costs and Health State Utility Values
  1. Abbreviations: DC, decompensated cirrhosis; LT, liver transplantation.

Chronic HCV F0-F3$209209Gamma209134,35
Chronic HCV F4$557557Gamma557134,35
LT (year 1)$168,6437,487Gamma3,798,6460.04436
LT (year 2+)$38,0153,797Gamma380,6010.10036
Testing and therapy-related cost
 Positive diagnosis$113    37
 Negative diagnosis$30    37
Genotype 1
 Biopsy$571    34
 Diagnosis-related$1,231    34
 Treatment$70,740    35,38,39
 Treatment failure$56,784    35,38,39
Genotype 2/3
 Diagnosis-related$660    34
 Treatment$14,245    35,38,39
 Treatment failure$7,123    35,38,39
Health utility   MinMax 
 On treatment0.720/0.755  0.730/0.7600.710.75040
 SVR0.860  0.8400.88041
 Chronic HCV F0-F30.790  0.7700.81042
 Chronic HCV F40.760  0.7000.79043
 DC0.690  0.4400.6943
 HCC0.670  0.6000.72043
 LT (year 1)0.500  0.4000.69044
 LT (year 2+)0.770  0.5700.77043


We utilized published health state utility values based on a United States population analysis.18 Utility values are modeled independently of age. The health utilities used in this model are presented in Table 2. Future health benefits, as measured by QALYs, are discounted at 3.5%.


Our analysis focused on three key operational areas that impact the cost-effectiveness of HCV testing and treatment: treatment eligibility, age and fibrosis stage treatment prioritization, and timing of treatment initiation. These are described in further detail:

  • 1The cost-effectiveness of testing for HCV is dependent upon the total number tested (which is a fixed cost for a given number tested), the number of HCV-infected individuals identified, and the numbers eligible for treatment. We therefore assessed the relationship between the proportion of tested and diagnosed subjects treated (including subjects who die or progress before treatment initiation) and the cost-effectiveness of birth cohort testing compared with risk-based testing. We presented the results of varying the proportion of tested and diagnosed people treated (15% to 100%) and compared two scenarios:
    • aThe base case birth cohort and risk-based populations, as reported in the CDC guidelines.
    • bThe birth cohort and risk-based populations reported in the analysis by McGarry et al.17 (a paper modeling the birth cohort born between 1945 and 1970).
    • 2We assessed the impact of prioritizing treatment by age and fibrosis stage on the cost-effectiveness of testing and treatment. We chose to model four average age cohorts consistent with our birth cohort, using ages and fibrosis stage distributions presented in the analysis by McGarry et al. We modeled the birth cohort, partitioning the analysis into four age groups with associated fibrosis stage population distributions17:
    • a45-49 years: F0, 21.0%; F1, 35.6%; F2, 20.6%; F3, 13.8%; F4, 9.0%.
    • b50-54 years: F0, 16.4%; F1, 32.0%; F2, 21.0%; F3, 16.5%; F4, 14.1%.
    • c55-59 years: F0, 12.5%; F1, 27.2%; F2, 20.4%; F3, 18.6%; F4, 21.3%.
    • d60-64 years: F0, 9.8%; F1, 22.9%; F2, 19.0%; F3, 19.4%; F4, 28.9%.
    We took the mid-point estimates for the age of each group (47, 52, 57, and 62 years) and ran three scenarios treating the same number of patients but changing the distribution across fibrosis stages:
    • aPrioritizing treatment toward initial fibrosis stages (“F0 skew”).
    • bBiasing treatment toward late fibrosis stages (“F4 skew”).
    • cNo prioritization (“no skew”).
    • 3We assessed the impact of treatment timing. Assuming all patients are tested and identified in year 1, we analyzed the impact of treating eligible patients across the following five scenarios:
      • aAssumes an exponentially declining profile treating 48% of patients in year 1, and the remaining patients are treated by year 4.
      • bAssumes 1% of patients are treated in year 1, and an additional 2% are treated each year thereafter, up to 19% in year 10.
      • cAssumes 10% of patients are treated each year.
      • dAssumes 19% of patients are treated in year 1, reducing by 2% each year thereafter to 1% in year 10.
      • eAssumes an exponentially increasing profile where treatment initiation is deferred until year 7; all patients are treated by year 10.For all analyses, total costs and number of ESLD-related complications are reported.


Under the base model settings, the predicted cost-effectiveness of birth cohort testing compared with risk-based testing was $28,602. Figure 3 demonstrates the relationship between the percentage of the tested population being treated (including subjects who die or progress before treatment initiation) and the cost-effectiveness of birth cohort testing versus risk-based testing. At a willingness to pay threshold of $50,000, Fig. 3 shows that approximately 278,000 (26%) of the identified population need to be treated for birth cohort testing to be cost-effective when compared with current risk-based testing. By treating at least 143,000 more people than current risk-based testing, enough benefit and cost offsets will be generated to warrant the extra costs related to diagnosing 809,000 people on top of the current risk-based testing; an additional $1.44 billion in testing costs and $3.87 billion in chronic HCV care (assuming no treatment of the 809,000).

Figure 3.

Graph demonstrating the relationship between the proportion of tested and HCV-positive subjects treated (including subjects who die or progress before treatment initiation) and the cost-effectiveness of birth cohort testing versus risk-based testing. The CDC population (bold line) and McGarry population (dotted line) are shown.

Given the need to test, identify, and treat a large number of patients to ensure birth cohort testing is cost-effective, the impact of prioritizing treatment in those identified is illustrated in Fig. 4. This shows the effect on cost, QALYs, and number of complications associated with prioritizing treatment by age and fibrosis stage. In Fig. 4, the y axis at y = 0 represents the “no skew” scenario, with each plot showing differences in total lifetime costs, QALYs, and complications with treatment prioritized to those with less fibrosis (F0 skew) and those with advanced fibrosis (F4 skew). In each scenario, the same numbers of patients are being treated (represented by the darker shaded area in Fig. 5); the differences are driven predominantly by avoided complications. Biasing treatment toward F4 is associated with decreased costs ($4.1 billion compared with “no skew” and up to $7.5 billion compared with “F0 skew” in those aged 57 years), increased QALYs (142,029 for those aged 47 years, 141,342 when aged 52 years, 112,102 when aged 57 years, and 82,603 for those aged 62 years when comparing with “no skew”) and between 29,444 (compared with “no skew”) and 59,035 (compared with “F0 skew”) fewer ESLD-related complications.

Figure 4.

Assessing the impact of prioritizing treatment by fibrosis stage. The three graphs illustrate the effect on total discounted costs, QALYs, and number of complications when prioritizing treatment toward earlier fibrosis stages (F0 skew) or later fibrosis stages (F4 skew) for subjects aged 47, 52, 57, and 62 years, respectively, compared with treating uniformly across fibrosis stages (n = 135,089 for all).

Figure 5.

These six graphs represent the distribution across fibrosis stage and the number of people treated by age group. Dark gray bars represent treated patients; light gray bars represent untreated patients.

Following the identification of treatment-eligible subjects, there are a number of ways in which treatment uptake may be prioritized. Figure 6 illustrates the predicted consequences of treatment initiation across five scenarios that prioritize earlier or later treatment uptake. These treatment scenarios are further stratified by fibrosis stage–based treatment. In all cases, a total of 551,800 HCV treatment–eligible patients are allocated treatments over a 10-year period in the model; for each scenario, total discounted costs, QALYs, and the number of expected HCV-related complications are reported in Fig. 6. Earlier treatment initiation is associated with increased cost, increased QALYs, and the lowest number of ESLD complications.

Figure 6.

The predicted consequences of treatment initiation across five scenarios that prioritize earlier or later treatment uptake (top left). These treatment scenarios are further stratified by fibrosis stage severity-based treatment. Triangles represent treatment skewed toward less severe fibrosis stages; squares represent treatment skewed toward more advanced fibrosis; circles represent no fibrosis stage treatment bias. In all scenarios, a total of 551,800 HCV treatment–eligible patients are allocated treatments over a 10-year-period.


A number of recent publications have demonstrated that birth cohort screening is cost-effective compared with the current practice of risk-based screening. Our base case cost-effectiveness of $28,602 is consistent with previous estimates.16, 17 Our estimates of cost-effectiveness were, however, considerably greater than those estimated by Coffin et al.,18 who reported incremental cost per QALY ratios of $7,900 for screening the general population and $4,200 for the birth cohort population born between 1945 and 1965. This is because our analysis compares a risk-based testing strategy with a birth cohort strategy, whereas Coffin et al. compared a risk-based scenario (that identifies a significantly higher number of infections) to a risk-based plus one-time screening strategy that includes 15% of the population.

Importantly, the implementation of a birth cohort testing program represents a significant logistical and financial undertaking, and the principle objective of our analysis was to estimate how various implementation issues (e.g., the timing and prioritization of treatment) impact future costs and health outcomes. Two important drivers of cost-effectiveness in birth cohort testing are the number of prevalent infections within the tested population and the treatment uptake rate. The cost associated with implementing an HCV testing program is substantial, and achieving cost-effectiveness is conditional upon identifying and treating enough patients to generate sufficient cost offsets and QALY gains. Therefore, adequate commitment focused on attaining the necessary testing and treatment uptake is required to ensure birth cohort testing is cost-effective.

The additional specific cost incurred through implementation of birth cohort testing is relatively small (∼$1.9 billion to test 66.2 million people) compared with the cost associated with treatment (∼$25.9 billion to treat 551,800 people). Therefore, the cost-effectiveness of birth cohort testing is predominantly driven by the cost-effectiveness of treating chronic HCV; which, based on the United States population, is reported to be cost-effective.25-27 Treating patients with more advanced disease is typically more cost-effective, because despite lower efficacy, the potential to avoid the costs and quality of life decrements associated with ESLD-related complications is increased. Our analysis further confirms this within the context of a testing and treatment program. For a fixed number of treated patients, prioritizing therapy initiation in those with more advanced disease has the potential to reduce overall costs by maximizing the cost offsets associated with ESLD complications avoided. Furthermore, this approach also maximises QALYs. Comparing the costs and QALYs gained when prioritizing treatment toward F0 and F4, Fig. 4 suggests that treating patients and prioritizing those in F4 is more cost-effective than treating on a first-come, first-serve basis, and significantly more cost-effective than treating with priority given to those in F0. Furthermore, it appears that treating older patients incurs a greater cost and lower QALY gain than treating younger patients. This is predominantly due to the greater susceptibility to disease progression and higher mortality rate of older patients. Therefore, severity of fibrosis and timing of treatment after diagnosis are both important factors worth considering when optimizing a testing and treatment program. Analysis of the cost-effectiveness of treating patients in specific fibrosis stages as part of a testing and treatment program is challenging. This is because overall cost-effectiveness is influenced by the numbers tested (which represents a fixed cost in the analysis) and the number of people identified within each specific fibrosis stage. Our analysis sought to compare a clinically relevant scenario: having identified a given number of patients with chronic HCV, is a targeted fibrosis stage–specific treatment policy better value than treating across all fibrosis stages? This analysis demonstrates that treatment initiation biased toward F3 and F4 results in reduced cost and increased QALYs compared with a policy of treatment regardless of fibrosis stage.

The timing of treatment initiation is also an important factor. Our analysis indicates that if birth cohort testing and treatment policy is initiated, then immediate treatment prioritized toward those with more advanced disease will minimize cost, minimize complications, and maximize health-related quality of life.

There are a number of limitations to our analysis. We have used estimates of the size of the tested population from a previous economic evaluation of birth cohort screening16; this is, however, likely to be an underestimate. Data published by Chak et al.28 suggest that HCV prevalence estimates derived from the National Health and Nutrition Examination Survey (NHANES) underestimates true prevalence by 500,000 to 1 million based on estimates of unreported cases among the homeless and incarcerated. The rationale for excluding these subjects was to maintain consistency with the cohort and methodology used to inform the CDC guidelines.13 Failure to expand the underlying NHANES population will have limited relevance to the estimation of birth cohort cost-effectiveness. Those subjects not captured in NHANES are described as high-risk28 and would therefore be candidates for inclusion within existing risk-based identification. Other groups underreported in NHANES will be a mixture of those who are eligible and ineligible for treatment. The interpretation of our analysis and findings is therefore conditional upon the birth cohort selected and the subset of treatment-eligible subjects identified.

A further limitation of our analysis is that we did not consider the retreatment of prior null responders or the effects of resistance in those not achieving SVR. This will be an important consideration in the next few years as the number of new antiviral therapies indicated for the treatment of chronic HCV infection increases substantially. The sequencing of initial and subsequent treatment stratified by patient phenotype will present a challenging public health optimization problem. Drug acquisition cost will be a pivotal consideration.

A further limitation is that our projection of future costs and benefits is conditional upon the age-specific distribution of fibrosis stage at diagnosis. The distribution we have used is derived from a previous modeling study17 and is therefore subject to some uncertainty. Consequently, in respect of absolute numbers, our projected future costs, complications, and QALYs should be interpreted with this limitation in mind. Our analysis of the cost-effectiveness of targeted fibrosis stage–specific treatment is, however, unaffected by the shape of the fibrosis stage distribution across the treatment-eligible population.

In conclusion, our study confirms that birth cohort testing compared with risk-based testing is cost-effective. It is imperative that such a program is initiated in full to ensure a sufficient number of HCV cases are identified and, given the practical and financial challenges of implementing such a program, the greatest return on investment is obtained when eligible patients are treated immediately and that those with more advanced disease are prioritized.