Cost‐effectiveness of a decentralized, community‐based “one‐stop‐shop” hepatitis C testing and treatment program in Yangon, Myanmar

Abstract Background and Aim The availability of direct‐acting antiviral (DAA) treatment and point‐of‐care diagnostic testing has made hepatitis C (HCV) elimination possible even in low‐ and middle‐income countries (LMICs); however, testing and treatment costs remain a barrier. We estimated the cost and cost‐effectiveness of a decentralized community‐based HCV testing and treatment program (CT2) in Myanmar. Methods Primary cost data included the costs of DAAs, investigations, medical supplies and other consumables, staff salaries, equipment, and overheads. A deterministic cohort‐based Markov model was used to estimate the average cost of care, the overall quality‐adjusted life years (QALYs) gained, and the incremental cost‐effectiveness ratio (ICER) of providing testing and DAA treatment compared with a modeled counterfactual scenario of no testing and no treatment. Results From 30 January to 30 September 2019, 633 patients were enrolled, of whom 535 were HCV RNA‐positive, 489 were treatment eligible, and 488 were treated. Lifetime discounted costs and QALYs of the cohort in the counterfactual no testing and no treatment scenario were estimated to be USD61790 (57 898–66 898) and 6309 (5682–6363) respectively, compared with USD123 248 (122 432–124 101) and 6518 (5894–6671) with the CT2 model of care, giving an ICER of USD294 (192–340) per QALY gained. This “one‐stop‐shop” model of care has a 90% likelihood of being cost‐effective if benchmarked against a willingness to pay of US$300, which is 20% of Myanmar's GDP per capita (2020). Conclusions The CT2 model of HCV care is cost‐effective in Myanmar and should be expanded to meet the National Hepatitis Control Program's 2030 target, alongside increasing the affordability and accessibility of services.


Introduction
Hepatitis C virus (HCV) infection is highly curable, meaning that liver cirrhosis, liver failure, hepatocellular carcinoma (HCC), and premature mortality from untreated HCV can be averted.Globally, an estimated 58 million people have chronic HCV infection. 1In Myanmar, an estimated 1.5 million people have been exposed to HCV 2 ; national HCV antibody prevalence is 2.7%. 3 Population-based RNA prevalence studies have not been conducted, but assuming a 26% spontaneous clearance rate, at least 1.1 million people in Myanmar have chronic HCV infection. 2echnology and treatment advances have changed HCV testing and treatment.Point-of-care laboratory tests and simple direct-acting antiviral (DAA) treatment with minimal side effects have made HCV elimination goals realistic, 2 and public sector HCV elimination programs can now be established in low-and middle-income countries (LMICs). 4In 2016, the World Health Organization (WHO) proposed the elimination of HCV as a major public health threat by 2030, targeting 90% of people with HCV diagnosed and 80% treated. 5yanmar was one of the first South-East Asian countries to provide free DAAs to the public, 6 although demand far exceeds supply.In 2017, the National Hepatitis Control Program (NHCP) released simplified treatment guidelines for HCV infection by specialist and nonspecialist doctors, 7 enabling general practitioners (GPs) to prescribe DAAs and use the Cepheid GeneXpert ® viral load test to assess current infection. 7Myanmar also set targets of diagnosing 50% of people with HCV and treating 50% of patients by 2030, 8 with modeling estimating that approximately 55 000 people should be treated annually between 2020 and 2030 to reach these targets, and that doing so could avert 40 000 new HCV infections and 25 000 HCV-related deaths. 2 In Myanmar, DAAs became available in 2016 through the NHCP, from tertiary hospitals, some international and local nongovernmental organizations (NGOs), and research projects focusing on HIV/HCV coinfection. 9In 2018, NHCP, in collaboration with Clinton Health Access Initiative, expanded treatment availability through a public-private partnership model permitting discounted DAAs and laboratory costs. 10DAAs are also available in some private specialist clinics and hospitals, and prices are decreasing over time with a 28-day supply of Sofosbuvir (SOF) and Daclatasvir (DCV) costing USD110 in 2019. 11espite this progress, challenges remain; with only an estimated 4000 people per year being treated before the COVID-19 pandemic, significant expansion is needed to reach the 55 000 per year required to achieve the national targets. 2 Reduced diagnostic and treatment costs, as well as simplified and affordable models of care, are required to diagnose and treat patients at this scale.
Diagnostic testing and DAA treatment were initially unaffordable to most private patients and governments in LMICs. 12In Myanmar, the Ministry of Health remains the main funder of the public HCV testing and treatment program that began in July 2017. 13Recent efforts to reduce commodity prices by allowing generics manufacturing and sale in many LMICs have increased public access to treatments dramatically. 10However, despite reductions in prices, the costs of generic DAAs and diagnostics continue to prevent expansion of the NHCP program. 13Multiple diagnostic products for HCV antibody and RNA testing have also received WHO prequalification, 4 increasing diagnostic testing options and access, but to date, interest in reducing the cost of diagnostics globally is weak.Challenges to reducing the cost of diagnostics include complex patent protections, the dominance of a few companies, and a general lack of transparency in global diagnostics markets preventing understanding of the true costs of manufacturing, procuring, and distributing tests, reagents, and the platforms on which they run. 14rices for HCV diagnostics changed little in recent years, in contrast to rapidly falling DAA prices. 10Therefore, efforts to further simplify the diagnostic pathway and reduce the costs of diagnostics should be prioritized.
In 2019, the Burnet Institute, in collaboration with Myanmar Liver Foundation (MLF), conducted the Hepatitis C Community-based Testing and Treatment Study (CT2 Study) in Myanmar.The project was implemented in partnership with the Foundation for Innovative New Diagnostics (FIND) as part of the Hepatitis C Elimination through Access to Diagnostics (HEAD-Start) program led by FIND and supported by Unitaid/ WHO.The CT2 Study assessed the feasibility of a decentralized HCV model of care in a resource-limited country. 9,15It showed high retention in care, with most participants initiating treatment within two visits (median time from RDT to DAA initiation 3 days [IQR 2 -5]) and high cure rates (non-cirrhotic patients 93%, cirrhotic patients 83%). 16However, to support further rollout of these programs and encourage investment in decentralized models of HCV care, they must be cost-effective.
We analyzed the costs associated with GP-led HCV clinics (CT2 Study clinics) to estimate the cost-effectiveness of simplified, decentralized models of care versus no access to testing and treatment.This is useful to inform the scale-up of communitybased programs to increase testing and treatment where access is currently insufficient.

Methods
We used a cohort-based Markov model to evaluate the costeffectiveness of a decentralized model of HCV testing and treatment.Cost and effectiveness inputs were collected from the CT2 Study, and disease progression, mortality, and health utility parameters were obtained from the literature.
The CT2 study.The study was conducted in Yangon, the business capital of Myanmar.We collected data from the CT2 clinics: the Burnet Institute clinic, serving mainly people who inject drugs (PWID), and the MLF clinic, serving mainly the general population with liver disease-related health concerns.Each clinic included a trained GP, nurse, and laboratory technician; the Burnet clinic also had a peer support worker.Patients were ineligible to enroll if <18 years or had previously tested HCV RNA-positive, taken treatment for HCV, were coinfected with HIV, hepatitis B, or tuberculosis, had renal function impairment, or were pregnant.CT2 used the WHO-prequalified SD BIOLINE hepatitis C test (Standard Diagnostics Inc., South Korea) for POC antibody detection.The Cepheid GeneXpert ® hepatitis C RNA test was used to assess patients' current infection status.After enrolling in CT2, patients underwent HCV diagnostic testing and other pretreatment assessments, including calculating the aspartate aminotransferase to platelet ratio index (APRI) score to assess for cirrhosis.The APRI score was used to inform duration of treatment (12 or 24 weeks), and with other pretreatment assessments, including a physical examination and liver function test results, guided GPs in specialist referral.DAA therapy used was generic SOF (400 mg) and DCV (60 mg).Patients were tested for sustained viral response (SVR-equivalent to cure) 12 weeks after completing treatment, also using GeneXpert ® hepatitis C RNA test.DAAs and tests were at no cost for patients in the study.
Ethics approval for CT2 Model description.We used a deterministic cohort-based Markov model to compare total lifetime costs and QALYs for CT2 participants between two scenarios: (1) no testing and no treatment; and (2) testing and treatment as occurred in the CT2 clinics (Fig. 1).
The no testing/no treatment scenario (1) was a modeled counterfactual where CT2 study participants were not tested.In this scenario, the cohort entered the model in the first year (2019), of which a percentage had HCV infection and were distributed across stages F0-F4 of the METAVIR fibrosis scoring system, decompensated cirrhosis (DC), and HCC disease, based on estimates from CT2 participants (Table 1).Each year in the model, patients could develop more severe liver disease or die from age-specific all-cause mortality or HCV-related mortality.All-cause mortality was estimated based on the cohort age distribution (CT2 data) and increased over time as the cohort aged.The model was run until everyone in the cohort had died (from HCV-related or all-cause mortality).
In the testing/treatment scenario (2), participants from the clinic sites were all tested on enrollment as per the CT2 study design.Of those with HCV, a proportion were assumed to be cured within the first year (based on CT2 data).Following cure, liver disease progression stopped (apart from F4 to either DC or HCC, at a reduced rate); however, individuals with DC or HCC could still experience liver-related mortality.Figure 1 summarizes the modeled disease progression, and Table 1 lists the parameters used.
Cost estimates.Cost data were collected from the CT2 study's expenditure records and price lists.Average cost per person antibody tested, RNA tested, and treated were calculated by summing the cost of all drugs, medical supplies, diagnostic tests (and auxiliary tests such as liver function tests), staff time, and fractional equipment and overheads used for each step.Table 2 shows line-listed costs from CT2 and unit cost calculations.Staff costs were recorded as minutes of  Model outcomes.Total QALYs, HCV-related deaths, and cumulative costs were calculated for testing/treatment and no testing/no treatment scenarios, and were used to calculate the incremental cost-effectiveness ratio (ICER).Costs and QALYs were discounted at 3% per annum.
Uncertainty and sensitivity analysis.Our probabilistic multivariate uncertainty analysis drew 100 random parameter sets based on the uncertainty ranges of individual parameters in Table 1.Parameters were randomly sampled from uniform distributions across their ranges, or +/À10% if upper/lower bounds were unavailable.Reported results represent point estimates and central 95th percentiles.Myanmar health financing depends mainly on out-of-pocket payments, and spending power differs hugely by income level and region. 23Therefore, instead of using a fixed estimate for willingness to pay (unavailable for Myanmar), we used the distribution of outcomes to estimate the probability that the CT2 model was cost-effective for different willingness-to-pay thresholds.
In addition, we conducted univariate sensitivity analyses to assess the impact on cost-effectiveness by varying the following model parameters: average age, HCV seroprevalence, treatment cost, and initial fibrosis stage distribution.
Of 456 patients who received an SVR test, 421 achieved SVR. 16he cascade of testing and treatment is shown in Figure 2.
In the testing/treatment scenario, with 92% of infected individuals cured at simulation start, discounted QALYs over the remaining cohort lifetime totaled 6518 (95%UI 5886-6581), the total discounted cost of testing, treatment, and disease management was USD123 248 (95%UI 122 440-124 155), and nine HCV deaths occurred.
Willingness to pay.The intervention had a 90% chance of being cost-effective if benchmarked against a willingness-to-pay threshold of USD300 (Fig. 3); USD300 is 20% of GDP per capita in Myanmar in 2020 (USD1468). 24nsitivity analysis.Table 3 shows changes in ICER values depending on the scenarios used for sensitivity analysis.The biggest drivers of cost-effectiveness were average age and initial liver disease distribution; treatment provision to younger cohorts was more cost-effective because they lived longer following cure, and treatment provision to people with more severe disease was more cost-effective because it was more likely to prevent deaths and other severe outcomes.Treatment costs and test positivity rates had some influence on cost-effectiveness, suggesting that continual efforts to reduce prices remain important, and models of care should be targeted to areas of highest burden.

Discussion
This study provides clear evidence that "one-stop-shop" decentralized models for testing and DAA-based HCV treatment among PWID and the general population can be cost-effective in Myanmar.HCV testing and treatment with DAAs produced an estimated ICER of USD294 per QALY gained.The QALYs gained come from lives saved (an estimated 54 vs. 9 deaths in the no testing/no treatment vs. testing/treatment cohorts), as well as prevention of severe liver disease, DC, or HCC through early treatment.This model of care had an estimated 90% probability of being cost-effective against a willingness-to-pay threshold of USD300.While the willingnessto-pay threshold in Myanmar is unknown, USD300 is 20% of GDP per capita in Myanmar, an ICER that would be cost-effective in many LMICs. 25,26he CT2 study has already shown that a one-stop-shop model of care with simplified HCV clinical pathway is safe and feasible in Myanmar and acceptable to healthcare providers and patients. 15High retention in care was also achieved, with cure rates above 90%. 16Our study adds to the evidence by showing that this model of care is a cost-effective way of providing testing and treatment in Myanmar to people with poor current access.Such a model of decentralized, nonspecialist-led care with point-of-care diagnostics is supported by the WHO 27 and is vital to scaling up treatment access to achieve national HCV targets of 50% of HCV patients diagnosed and 50% treated by 2030. 8Future work should consider additional steps required for scale-up, including the number, locations, and types of sites required.9][30] Within Myanmar, MSF proposed a simplified model of HCV care to the Myanmar Ministry of Health, which incorporated fewer patient visits, task-shifting from doctors to nurses and using local staff, and was estimated to be highly costeffective (ICER <USD400/DALY averted compared with no treatment in HCV/HIV coinfected patients). 31e used WHO-prequalified generic DAAs ordered and imported via pooled procurement, so our DAA price was lower than private clinics in Myanmar (SOF/DCV price USD399 (2018) to USD330 (2019) for 12-week treatment versus USD86.76 for 12-week treatment in CT2).Therefore, we conducted sensitivity analyses for 12-week treatment prices of USD300, 20 USD150 based on the DAA prices in Myanmar, and USD28 (from a neighboring country [Pakistan]), 10 finding that the CT2 model of care had a 90% probability of being costeffective if measured against a willingness-to-pay thresholds of 60%, 30%, or 12% per capita GDP respectively (see Appendix).This suggests that while the CT2 model of care is likely to be cost-effective in any of these scenarios, reducing DAA prices would greatly improve the affordability of treatment scale-up.
In cost-effectiveness analyses, willingness-to-pay thresholds are often proposed as per capita GDP, 31,32 following WHO's recommendation. 33However, the cost-effectiveness of health interventions should not be based solely on GDP per capita, but also on the impact of existing health expenditure (unknown for Myanmar), because funding a new intervention may require something else to receive less funding.From a health economics perspective, this change in funding prioritization must achieve better overall outcomes.Aside from cost-effectiveness, factors such as affordability, feasibility, and availability of staff and other resources should be considered. 34he strength of our study is the use of real-life data from an implementation program.Our cost analysis includes full activity-based cost data to calculate the costs of screening, treatment, and follow-up for participants as completely as possible.In addition, few other studies use SOF + DAC alone as DAA treatment in Myanmar.Nonetheless, our study has limitations.First, we explored the cost-effectiveness of HCV infection only, although the prevalence of HCV/HIV coinfection was 20.1% and HIV/HBV/HCV coinfection was 20.7% among PWID 35 ; costeffectiveness of treating HIV and HCV coinfection in Myanmar is published elsewhere.Second, the HCV RNA-positive rate among participants (88%) was higher than the commonly reported 75% 36 ; future projects may not encounter such a high rate.Moreover, the anti-HCV positivity rate was high (95%) in the CT2 cohort, indicating many people "willing and waiting" 37 for fully funded treatment.This proportion positive may change  as more people receive treatment and HCV cure.We also had a high retention rate, thanks to active follow-up by clinic staff, who phoned patients frequently near their appointments.However, in real-world settings, assigning staff to remind patients like this may be difficult to implement. 38To understand how these limitations would influence results, we conducted sensitivity analysis around positivity rates and follow-up rates.Third, care must be taken when generalizing these findings because CT2 was conducted in the business capital of Myanmar where laboratory facilities, electricity, and human resources are more available than in rural and remote areas. 15Future projects should be conducted in areas with high HCV burden, for example, where National HIV/TB Program sites or NGO branches are situated or at private GP clinics, to understand the cost-effectiveness of models of care in other decentralized and less well-served settings.Evidence from the CT2 study is applicable to multiple types of decentralized sites, especially where Xpert machines are already available, as many patients are willing and waiting but require access to affordable treatment through an accessible model of care.
The COVID-19 pandemic and civil unrest have hampered HCV programs in Myanmar via limitation of services, reduced access to clinics due to travel restrictions and security concerns, and difficulties in importation of and transportation of DAAs to services.However, telehealth services and increasing use of full course dispensation alleviated the impacts.These developments make the need for decentralized community-based models of care even more important.
To conclude, we have provided crucial estimates of costeffectiveness of a one-stop-shop community-based model of HCV testing and treatment with DAAs in Myanmar.Strong political will and ongoing negotiations to reduce the costs of DAAs will be essential to eliminate HCV in Myanmar.

Figure 1
Figure 1 Disease progression pathway in model.

Figure 2
Figure 2 Study procedures (minimum visits needed for a patient of 12 weeks DAA treatment).

Figure 3
Figure 3 Probability that the CT2 model of care is cost-effective for different willingness-to-pay thresholds.

Table 1
List of parameters used in the model

Table 2 (
Continued) + unit cost for SVR12 visit + lab staff cost + HCV RNA (SVR12) + fractional overhead cost for RNA test + staff training cost*7 times + office setup cost*7 times + drug cost for 24-week regime † Lab staff time for other tests was assumed to be approximately the same as for HCV RNA test.

Table 3
Sensitivity analysis results