The cost- effectiveness of case- finding strategies for achieving hepatitis C elimination among men who have sex with men in the UK

Modelling suggests hepatitis C virus (HCV) elimination is possible among men who have sex with men (MSM), with key screening groups including HIV- diagnosed MSM and MSM using pre- exposure prophylaxis (PrEP). Mathematical modelling was used to determine the cost- effectiveness of HCV case- finding strategies among MSM from the provider perspective, and to determine which interventions could achieve a 90% reduction in HCV incidence over 2015– 2030. At baseline, we assumed symptomatic screening in HIV- negative MSM (including PrEP users) and 12- monthly screening among HIV- diagnosed MSM. Improved case- finding strategies included screening alongside HIV testing in HIV- negative MSM not using PrEP (PrEP non- users); 12/6/3- monthly screening in PrEP users; and 6- monthly screening in HIV- diagnosed MSM, with the cost- effectiveness being compared incrementally. Costs (GBP) and quality- adjusted life years (QALYs) were assessed to estimate the mean incremental cost- effectiveness ratio (ICER) with a time horizon to 2050, compared to a willingness- to- pay threshold of £20,000/QALY. From the baseline, the most incrementally cost- effective strategy is to firstly undertake: (1) 12- monthly HCV screening of PrEP users (gaining 6715 QALYs with ICER £1760/QALY), followed by (2) HCV screening among PrEP non- users alongside HIV testing (gaining 7048 QALYs with ICER £4972/QALY). Compared to the baseline, this combined strategy would


| INTRODUC TI ON
The last 10 years have seen a global epidemic of hepatitis C virus (HCV) among men who have sex with men (MSM), with the prevalence of HCV estimated at 1.5% in HIV-negative MSM and 6.3% among HIV-infected MSM. 1,2 Although HCV is a lifelong chronic infection, the development of direct-acting antiviral (DAA) treatments for HCV allow for successful treatment, with cure rates over 90% even among HIV co-infected individuals. 3,4 This has led to the World Health Organization (WHO) developing a Global Health Sector Strategy to eliminate hepatitis as a public health threat by 2030; setting targets to reduce the incidence of new chronic hepatitis B and C infections by 90% and the mortality attributable to hepatitis B and C by 65%. 5 Existing initiatives to increase HCV screening and treatment among MSM in the UK and elsewhere have generally focussed on HIV-diagnosed MSM in contact with care because of their frequent health service contact, 6 and their higher HCV incidence (~sixfold higher in the UK than in HIV-negative MSM not using PrEP). 1 However, significant HCV infection resides in HIV-negative MSM (1.2% prevalence in UK in 2008/2009), 7 especially among those on HIV pre-exposure prophylaxis (2.1% in PROUD study). 8 Although three recent European studies have demonstrated scaling-up HCV treatment can result in substantial reductions (51%-77%) in HCV incidence among HIV-positive MSM, 6,9,10 our modelling suggests that screening and treatment is also needed in HIV-negative MSM to reach the HCV elimination targets among all MSM. 11 The coverage of HIV testing and treatment among MSM has improved over the last decade in the UK, with 92% of people living with HIV being diagnosed in 2017, and the proportion of diagnosed individuals accessing ART increasing from 85% to 98% over 2012-2017. 12 PrEP has also become readily available in Wales and Scotland, 13 and is being rolled out in England. 14 In this changing environment, our previous modelling suggested that numerous screening strategies could achieve HCV elimination among MSM without the need for behavioural change, with options including differing levels of screening among HIV-diagnosed MSM and HIV-negative MSM on or off PrEP. 11 To help inform policy decisions for achieving HCV elimination, this analysis uses modelling to determine which of these HCV screening strategies is the most cost-effective for achieving HCV elimination among MSM in the UK.

| ME THODS
Throughout this work, we make use of the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist. 15

| Model derivation
We adapt a previous deterministic continuous-time model of HIV and HCV transmission among all UK MSM 11,16 to include stages for HCV-related liver disease progression and prior exposure to infection (full details and further parameter discussion in supplementary material Appendix S1). The model ( Figure S1)

K E Y W O R D S
cost-benefit analysis, Hepacivirus, HIV, pre-exposure prophylaxis, sexual and gender minorities

Significance statement
Globally, we face an epidemic of hepatitis C virus (HCV) among men who have sex with men (MSM). However, new direct-acting antivirals for HCV have cure rates over 90% (regardless of HIV co-infection). This has led to the World Health Organization (WHO) setting targets to reduce the incidence of new chronic HCV infections by 90% by 2030. We project that HCV elimination can be achieved cost-effectively among UK MSM. However, policymakers should consider scaling-up HCV screening in HIV-negative MSM, especially PrEP users; with enhanced screening in both these groups necessary for achieving the WHO target. Plus, enhanced screening among HIV-diagnosed MSM is not as cost-effective.
HCV susceptibility, with the remainder becoming chronically infected. Chronically infected MSM progress through liver disease states ( Figure S1) with HCV disease-related death occurring from the decompensated cirrhosis, hepatocellular carcinoma, liver transplant and post-liver transplant stages. HIV co-infection increases liver disease progression, 18 although this is slowed with ART. 19 We model this by assuming a weighted rate of progression within our HIV-diagnosed compartment, dependent on the proportion currently on ART. Following effective HCV treatment, we assume disease progression ceases if individuals are at fibrosis stage F3 or lower 20,21 and is slowed among those with compensated cirrhosis (F4) or decompensated cirrhosis, 21,22 while continuing at the same rate for those with more progressed disease.
HIV-negative individuals may initiate using PrEP, which reduces their risk of HIV acquisition by 86%-97%. 23,24 PrEP was assumed to scale-up from 2018 to give a coverage of PrEP among HIV-negative MSM of 10%-15% by 2020. We also explore a scenario where PrEP coverage reaches between 20% and 30%, as may occur in the UK. 25 The proportion of high-risk and low-risk MSM on PrEP is de- at UK HIV clinics. 6,29,30 Pre-2018, we assume 2.2 years from HCV diagnosis to completing HCV treatment, consistent with UK data for pre-DAA treatments, 29 with this decreasing to six months from 2018 in line with more recent estimates. 6 Before 2015, we assume different HCV sustained viral response (SVR) rates for HIV-positive (SVR of 35%-42%) 31 and HIV-negative MSM (SVR of 59%-69%) 32 based on pre-DAA treatments, but then assume higher cure rates from 2015 for DAA therapies (SVR of 90%-100%), 29 with MSM failing treatment being retreated at the same rate as initial HCV treatment.

| Cost estimations and health utilities
For this analysis, we take the perspective of the UK National Health Service. We assumed UK MSM to number between 650,000-750,000 based on estimates from Natsal 33 and UK data from the European MSM Internet Survey (EMIS-2010). 34 The costs of HCV care for different stages of HCV-related disease were adapted from previously published estimates for the UK (Table 1). [35][36][37] We broadly split these costs into three categories, which we inflate to 2020/21 prices using the hospital and community health services index. The first category comprises the ongoing healthcare costs associated with HCV-infection prior to liver transplants. 35 The second category comprises the cost of a liver transplantation, including the procedural cost and subsequent cost of the patient after a successful transplantation. 36 The final category is the actual cost of a HCV treatment course at £10,000, which aligns with the NHS's negotiated drug price, alongside 12 weeks of treatment and SVR monitoring. Health utilities (quality-adjusted life years [QALYs]) and HCV disease progression rates came from previous studies. 19,22,[35][36][37][38][39] For both QALYs and costs, we apply a discount rate of 3.5% per year from 2020 as recommended by the National Institute for Health and Clinical Excellence (NICE). 40 Health utilities and costs for HIV were not included.
We also consider the intervention costs of HCV screening. We assume that HCV antibody testing is performed on all screened MSM, 37,41 with reflex RNA testing occurring if they test antibody positive. 37,41 We assume these tests are done using blood samples already taken during routine PrEP/HIV/sexual health check-ups.
This means the cost for each HCV test only includes lab testing, plus an assumed 5 min of specialist nurse time for HCV-related discussions around testing. 37 For each positive HCV RNA test, we assume costs for pre-treatment care including ten minutes of phlebotomist and consultant time, and the costs of diagnostics for full blood count, liver function, HCV viral load and genotyping, fibroscan and ultrasound. 37

| Parameterization of sexual risk behaviour
Sexual behaviours were parameterized using data from UK-based respondents to EMIS-2010, an online survey about HIV/STI-related morbidities, behaviours, needs and interventions among MSM across Europe. 42 Over 180,000 men from 38 countries completed the survey, including 18,000 from the UK. From EMIS-2010, we calculated key behavioural parameters given in Table S5 and summarized in Table 1. Briefly, EMIS-2010 data suggest 17.4% of UK MSM are high-risk (≥15 partners), among whom the prevalence of chemsex in the last year is higher than among low-risk MSM (22.6% vs. 11.5%). Chemsex adds an additional risk factor for HCV and HIV infection among the high-risk group above and beyond having more sexual partners. 43,44 The model assumes that MSM have sex more often with others of the same perceived HIV-status, with perceived HIV-positive concordant partnerships having lower condom use (13%) than other partnerships (68%).

| Model calibration
Assuming historic levels of HCV screening and pre-DAA SVR rates with no PrEP, the model was firstly calibrated to give a stable HIV and HCV epidemic in 2012. This is in line with HCV incidence TA B L E 1 Key model parameters with ranges and details of estimation included

HCV-related parameters
Efficacy of HCV treatment with DAAs-after 2015 95% (90%-100%) 3,4 Efficacy is equivalent to the proportion of MSM achieving sustained viral response Average delay from HCV diagnosis to completion of treatment before 2018.

| Model analyses
We firstly determined the best-value-strategy to implement   additional criterion is defined as our optimal-elimination-strategy.

| Sensitivity analyses
To ascertain which parameters are important for determining variability in our cost-effectiveness projections, we performed a linear regression analysis of covariance (ANCOVA) on all model runs of the optimal-elimination-strategy, looking separately at QALYs gained and costs incurred. The proportion of the sum of squares contributed by each parameter was calculated to determine each parameters' importance to the variability in our projections.
We also performed one-way sensitivity analyses on the pro- (11) halving the cost of HCV treatment from £10,000 to £5000; (12) no discounting of costs and QALYs; and (13) doubled PrEP coverage, reaching 20%-30% instead of 10%-15% by 2020. Within these scenarios, we also make the equivalent changes to the baseline scenario when making the new projections. For the doubled PrEP coverage scenario, we also determined if the optimal-elimination-strategy changes.

| Main analyses
Compared to the baseline scenario, our model projections in Figure 1 show that improving screening in HIV-negative MSM on or off PrEP will result in considerable additional impact on HCV incidence or   (Table 2).
In terms of cost-effectiveness, our projections suggest (Figure 2) that starting from the baseline scenario our first priority should be to undertake the following: (1) 12-monthly screening among PrEP users (mean ICER £1760/QALY compared to baseline), then (2) HCV screening among HIV-negative PrEP non-users alongside their HIV testing (mean ICER £4972/QALY compared to (1)). In total, this combined strategy would cost £46.9 (95%CrI £25.3-£66.9) million by 2050, with a mean overall ICER of £3,405/QALY compared to the baseline. This combined intervention also reaches the HCV elimination target in 100% of runs, thus also being the optimal-eliminationstrategy. Interestingly, HCV elimination cannot be achieved among all MSM (in over 95% of model runs) without combining case-finding strategies (1) and (2).
The next most cost-effective scenario is screening HIV-diagnosed MSM every 6 months. Although this strategy is cost-effective at an ICER of £10,090/QALY when just added to the baseline, it is not cost-effective when added to (1) and (2), gaining 201 (95%CrI 137-278) QALYs at a cost of £5.8 (95%CrI 3.9-7.8) million, giving an ICER of £28,845/QALY.

| Sensitivity analysis
Our ANCOVA analysis suggests the main contributors to variation in the projected QALYs gained for our optimal-elimination-strategy were as follows: the proportion of MSM who preferentially mix by risk-status (22.1% of variation) and HIV-status (6.9%); the probabil- that the optimal-elimination-strategy was cost-effective compared to the baseline scenario. Importantly, our model predicts that the intervention will be more cost-effective in scenarios with higher HCV incidence.
Lastly, under the scenario of doubled PrEP coverage, the optimalelimination-strategy remains the same, with a mean ICER of £3802/ QALY compared to the baseline scenario ( Figure 2).

| DISCUSS ION
Our findings suggest that improving HCV case-finding and treatment among MSM in the UK is a cost-effective strategy for reaching HCV elimination without the need for additional risk reduction strategies.
Existing PrEP, HIV care and sexual health appointments provide sufficient opportunity to facilitate the required increase in screening, although reductions to these services could affect the feasibility of these strategies. Our optimal-elimination-strategy adds to the current HCV screening guidance, indicating that 12-monthly testing among PrEP

| Strengths and limitations
We largely draw from the same strengths and limitations of our previous model. 11 The strength of our analysis is in modelling the full co-epidemics of HIV and HCV among MSM in the UK, and using this model to determine optimal screening strategies among different MSM subgroups for achieving HCV elimination.
With regards to limitations, our model is not generalizable to all settings, especially low-/middle-income countries. However, given the robustness of our findings to many sensitivity analyses, including different coverages of PrEP and variations in HCV epidemic dynamics, they should be generalizable to many highincome countries.
Secondly, simplifications were made when modelling the historic HCV epidemic. As discussed in our previous paper, these included assuming a stable HCV epidemic in 2012 (approximating incidence and prevalence data from that time), 11 not explicitly incorporating the impacts of injecting drug use, and assuming MSM have constant numbers of anal sex partners over their lifetime.
We also acknowledge that HCV screening occurs among HIVnegative MSM, but at frequencies that vary based on local testing procedures at sexual health services, country-level guidelines, and whether an individual accesses PrEP formally or informally. To bypass this complexity and give guidance on what HCV screening should occur among HIV-negative MSM, we assumed symptomatic screening at baseline. Similar variation is also likely in current HCV screening frequencies among HIV-diagnosed MSM, which we based on national guidance as it seems to capture what is happening in some HIV clinics. 6 We also did not include health utilities or treatment costs associated with HIV infection, firstly because of an absence of explicit health utilities for HIV/HCV co-infection, and secondly for equity reasons. We do not think these factors should play a role in deciding whether HIV-infected MSM should be treated for HCV or not. Also, as DAA prices were confidentiality agreed with the NHS, we were unable to provide a reference to these costs. However, we believe these costs to be reflective of the actual price per treatment and have considered different prices in our sensitivity analyses. We also did not include the more limited access to liver transplants experienced by HIV-diagnosed MSM.
Lastly, uncertainty exists in the data used to parameterize and calibrate the model, resulting in uncertainty in our model projections. This includes uncertainty in sexual behaviour data obtained from the online EMIS-2010 survey. Although EMIS-2010 was self-selecting and so the data may be biased towards more sexually active MSM, 42 it is likely to be less biased than other much smaller surveys undertaken in gay venues or STI clinic settings.
There is also uncertainty in the likely scale-up of PrEP across different UK regions, and the level of risk compensation that may occur among PrEP users. Encouragingly, incorporating risk compensation improves the overall cost-effectiveness of our optimal strategy compared to the baseline scenario and does not affect whether this strategy achieves HCV elimination. Higher levels of PrEP coverage also resulted in a similar ICER and the same optimal-elimination-strategy.

| CON CLUS IONS
Our findings have direct implications for any high-income country attempting to eliminate HCV among MSM. They strongly advocate for undertaking frequent HCV screening among HIV-negative MSM, especially among those on PrEP. This is currently not the focus of most HCV elimination initiatives occurring among MSM but is likely to be crucial for fully eliminating HCV among MSM. Fortunately, the screening costs for doing so are not large because additional HCV testing can be incorporated within existing screening practices, with added testing and staff time costs being largely offset by reduced future costs in HCV care and treatment.

NKM and PV have received unrestricted research grants from
Gilead. NKM has also received unrestricted research grants from

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from