Cost and cost‐effectiveness of a simplified treatment model with direct‐acting antivirals for chronic hepatitis C in Cambodia

In 2016, Médecins Sans Frontières established the first general population Hepatitis C virus (HCV) screening and treatment site in Cambodia, offering free direct‐acting antiviral (DAA) treatment. This study analysed the cost‐effectiveness of this intervention.


| INTRODUC TI ON
The World Health Organization (WHO) estimated that 71 million people were infected with the Hepatitis C virus (HCV) globally in 2015. 1 Most (80%) HCV infections are in low-and middle-income countries (LMIC), 2 but fewer than 5% of these patients are diagnosed. 3 HCV is a major contributor to liver cancer and overall cancer deaths in Cambodia and Asia. 4,5 Southeast Asia has the second highest burden of viral hepatitis mortality globally. 1 More than 11 million people are estimated to have antibodies to HCV in this region, 6 with 2.3% of Cambodians exposed to HCV. 7 Direct-acting antivirals (DAAs) offer an effective cure for HCV with few side effects. Access to these medicines, however, has been limited by their high cost, alongside the cost of diagnostics and the infrastructure required for scaling up treatment. 3,8 In Cambodia, healthcare expenditure per capita is low ($69 in 2012); of this 60% comes from patient out of pocket expenses. 9  Association for the Study of the Liver guidelines, 11 was subsequently simplified in 2017 by reducing the number of patient visits and treatment monitoring conducted, with data suggesting this did not adversely affect patient outcomes (cure rate and incidence of serious adverse events). 10 In this study, we evaluate the cost-effectiveness of both the full and simplified models of care implemented by MSF. To our knowledge, this is the first study to conduct a full costing and cost-effectiveness analysis of a real-world HCV treatment intervention in a LMIC. This includes Cambodian patient-level data on EQ-5D-5L health states for different HCV disease stages and the cost of healthcare for HCV-related liver disease.

| ME THODS
This study evaluated the cost-effectiveness of MSF's HCV treatment program in Cambodia 10

| Setting and models of care
Patient characteristics, costs and quality of life data were collected from an observational cohort study evaluating the 'real-world' effectiveness of DAAs for the treatment of chronic HCV infection in adults (≥18 years).
This study was conducted at MSF's HCV clinic, embedded within the gastroenterology department of the Preah Kossamak Hospital in Phnom Penh. 10 Clinic staff were employed by MSF, with a small number of nurses and doctors seconded from the Ministry of Health.
The FMC and SMC both included processes for HCV diagnosis and liver disease staging, following which patients either began treatment, were referred for further tests to determine eligibility or control comorbidities, or were determined to be ineligible for treatment. 10 Treatment regimens were based on sofosbuvir with daclatasvir or sofosbuvir with ledipasvir, with sofosbuvir/daclatasvir used as a pan-genotypic regimen in the SMC. Each patient was tested for sustained virological response at 12 weeks after the end of treatment (SVR12) to determine if treatment was successful. The SMC used point of care tests (SD Bioline and GeneXpert) for diagnostics while the FMC used an external lab for ELISA and PCR confirmation. In the SMC, compared to the FMC, the number of sessions where nurse-counsellors discussed risk factors, adherence and lifestyle education with patients were reduced (from ten to two),

Conclusions:
The simplified model of care was cost saving compared to no treatment, emphasizing the importance of simplifying pathways of care for improving access to HCV treatment in low-resource settings.

K E Y W O R D S
cost-effectiveness, direct-acting antiviral treatment, healthcare costs, hepatitis C, low-income population, Markov process, treatment costs

Lay Summary
• Access to treatment for Hepatitis C virus (HCV) has been limited in low and middle-income countries because of the high cost of drugs and complex treatment protocol.
• Médecins Sans Frontières screened and treated patients for HCV in Phnom Penh, Cambodia, with a simple treatment protocol with fewer visits (than standard of care), point of care testing and task shifting from doctors to nurses.
• The simple treatment protocol saved money and had better outcomes projected over the lifetime of the cohort patients compared to if they had not received HCV screening and treatment.
genotype testing was eliminated, three blood tests for monitoring during treatment were removed, patient visits during treatment were reduced from eight to four, with some tasks shifted from doctors to nurses and pharmacists ( Figure 1, Table S1

| Costing methods
The costs of HCV diagnosis and treatment were estimated from the provider's perspective in 2017 US dollars, using a retrospective cohort-based approach over a 9-month period (September 2016-May 2017). Data on costs came from MSF's expenditure records and price lists, which were allocated to activities making up the FMC as described above, with subtotals calculated for clinic visits, laboratory tests, DAA and other medicines and diagnosis costs.

| Patient-level resource use
Detailed patient-level data on the numbers and types of visits and tests undertaken for patients receiving each model of care were collected from electronic individual patient records. 12 Costs of casefinding were included as the cost of testing patients who were not found to be chronically infected. Data on the quantity of DAAs and other medicines received by each patient during the intervention came from pharmacy dispensing records. Staff time was allocated for each activity through the use of staff time sheets completed over the course of one week by patient-facing staff, direct observation and interviews with staff and according to the number of patient interactions in the observation period.

| Valuation of resources
Valuation of the resources used in the intervention was based on detailed financial records provided by MSF using activity-based costing. The cost per activity within the intervention, for example the baseline treatment visit, was estimated as the sum of ingredients (Table S5) (Table S6). In the base case we exclude costs specific to MSF and replace staff costs for international staff with what local staff would be paid by MSF for the F I G U R E 1 Summary of differences between full (top, blue) and simplified (bottom, orange) models of care at each phase of the screening and treatment process. Full details included in Table S1 same role. More details on costing methods can be found in the supplementary material (Methods S1).

| Cost of hepatitis C-related disease
Information on patient access to healthcare for HCV-related liver disease prior to treatment was gathered through a resource-use questionnaire administered to a subset (n = 144, liver disease stages F0-F4) of diagnosed patients at their initial visit. The questionnaire asked patients to recall the number of hospital inpatient or outpatient and clinic visits (health system contacts) in the 6 months prior to coming to the MSF clinic. We assumed that contacts in 6 months represented half of the annual number of contacts. For each type of visit, the patient was further asked to recall the reason and the price paid for the most recent visit. We use patient-reported costs to represent the cost of care in the base case because prior to the MSF program, all relevant costs would be the responsibility of the patient (Methods S2).

| Simulation model
The Liver disease-related mortality (assumed to occur in the DC and HCC states only) and background age-dependent mortality ( Figure S1) were modelled as absorbing health states. The number of liver transplants performed in Cambodia is negligible, and so was not modelled. The model was implemented in R version 3.6.1 using the heemodpackage. 15,16

| Disease progression rates
Parameter values for progression through disease states were sourced from previous studies (Table 1). Each patient started in one of seven liver disease states according to the distribution in the cohort. At the end of each cycle, patients either remain in the same state, move into a more advanced disease state or die from background or HCV-related mortality ( Figure 2). In the base case we assume there is no re-infection following curative treatment.
The SVR12 rate was calculated for the FMC and SMC from patients who were due to complete treatment at least 12 weeks before the data export date (17 July 2018), with those not tested for SVR12 for any reason, including loss to follow-up and death, counted as not achieving SVR12 (treatment failure). Patients who fail treatment return to the infected states and are assumed to face the same risks of liver disease progression as untreated patients, while cured patients move to a susceptible compartment in the same liver disease stage they were treated in. Being cured stops disease progression in precirrhotic patients while it is slowed in cirrhotic patients.

| Cost-effectiveness analysis
We evaluated the cost-effectiveness of the FMC and SMC compared to no HCV treatment in the study population. The model was run for 100 years to cover the full lifetime of the cohort, with a baseline discount rate of 3% for both future costs and outcomes. 19 Cost-effectiveness was evaluated in terms of lifetime costs per QALY gained, or the incremental cost-effectiveness ratio (ICER) for each strategy. This was compared to an empirical opportunity cost-based willingness-to-pay (WTP) threshold of $248 per QALY gained 20 and to the commonly used threshold of GDP per capita ($1270). 21

| Sensitivity analysis
We accounted for uncertainty in key parameters by conducting a probabilistic sensitivity analysis, in which 1000 parameter sets were sampled from their statistical distributions (Table 1). The cost of treatment for each fibrosis stage was varied according to the data in a triangular distribution of the median and inter-quartile range (IQR, Table S4). Cost of care was varied in a triangular distribution according to the bootstrapped 95% confidence interval for F1-F4 (Table 2) and according to the range of consensus values for DC and HCC.
Health state values were varied in a triangular distribution of the mean and 95% confidence interval for each fibrosis stage and time point (Table 3).
In addition, we conducted one-way sensitivity analyses by vary-  April 2018 ( Figure S4). The mean age of those treated was 55.9 (SD 10.6) years and 57% were female. Patients with cirrhosis (F4), DC or HCC made up 32.8% of diagnosed patients and 44.5% of treated patients (Table S2).

| Cohort treatment outcomes
Under the FMC and SMC, 624 and 1324 patients initiated treatment and were followed-up to at least 12 weeks post treatment respectively. These individuals were included in the costing and outcomes analysis, while patients treated during the transitional phase between the models of care were excluded from analysis ( Figure   S4).

| Costs of HCV treatment
The median total cost of HCV testing and treatment for the FMC was $925 (IQR $668-1,631) and $376 (IQR $344-422) for the SMC ( Figure 3, Table S4). The biggest contributor to the overall cost in both models was DAA costs, making up 26% and 42% of the total treatment cost for the FMC and SMC respectively.

| Cost of hepatitis C-related disease
The number of health system contacts reported in the resourceuse questionnaire is presented in Table 2 and Figure S2.

| Health state values
Health state values generated through EQ-5D-5L varied by fibro-  Table 3. The health profiles (proportion of individuals with each level for the five dimensions) reported in EQ-5D-5L are presented in Tables S7-S9.

| Cost-effectiveness
The FMC compared to no treatment was cost-effective ($187/QALY) compared to the opportunity cost WTP threshold ($248) and GDP threshold ($1270). The SMC was cost-saving (−$91/QALY,   being cost-effective compared to the opportunity cost threshold ( Figure S3).

| Sensitivity analysis
When key model parameters are varied the SMC generally remains cost-saving compared to no treatment (Figure 4), while the FMC is never cost-effective compared to the SMC (Table S11) The SMC would remain cost-saving at an antibody prevalence of 3% or a re-infection rate up to 6%/year.

| D ISCUSS I ON
Our analyses suggest that an HCV testing and treatment intervention undertaken among the general population in Cambodia could be cost-saving when a simplified model of care is used. This simplified model of care provides similar high rates of cure to the initially implemented but more costly full model of care. 10

| Strengths and limitations
The  levels of healthcare, and how this will change in the future, is crucial to improving the accuracy of cost-effectiveness estimates for HCV treatment.
Another strength is that we directly estimated health state utility weights for each stage of liver disease and treatment using EQ-5D-5L questionnaires collected from the treated population. Self-reported quality of life improved during treatment and at SVR12 as compared to baseline, similar to what has been observed in a cohort of patients treated with DAAs in Japan, 25 but with our data adding important information for LMICs.
A main limitation of this analysis is that, because of a lack of information on current re-infection risk we did not account for changes in disease incidence over time in our analysis. However, as a result of the advanced age and fibrosis observed in our cohort of patients, it is likely that many individuals were infected decades ago rather than being at current risk of infection and re-infection.
Including a constant re-infection rate made little difference to the result.
Furthermore, the population involved in this intervention were self-selecting, and were therefore highly motivated to receive treatment. The 65% antibody prevalence rate among screened patients indicates many were aware or suspected they were infected with HCV. Globally, fewer than 5% of those infected with HCV are aware of their status, 26 indicating that this high screening yield is not likely to be maintained once individuals who are aware of their status have been treated (warehousing effect).

| Comparison with other studies
Previous evidence on the cost-effectiveness of treatment with DAAonly regimens for HCV in LMIC is limited, but in agreement with our results. Studies in Egypt 28 and India 29-32 have found HCV screening and treatment to be cost-saving, and cost-effective in Lebanon 33 and Indonesia. 34 The WHO Hep C Calculator, 35 which allows a user to input the cost of treatment, finds HCV treatment (not including testing) in Cambodia to be cost-effective when default values are used and cost-saving when input values from this study are used.
However, these previous models do not use data from a real-world intervention, while ours is the first to use data from a local treatment intervention to estimate the costs of testing, treatment and healthcare, as well as quality of life utilities for different stages of HCV disease. Our analysis is therefore an important addition to the literature.

| Implications and conclusions
Much of the effort towards expanding access to HCV treatment in LMIC has focused on reducing and simplifying the cost of testing and DAA medication. In this cohort, simplification of the treatment pathway quadrupled the number of people that could be treated and cured, 10 reduced the cost of treatment by two-thirds and resulted in the intervention becoming cost-saving. This remained so even if the prevalence of HCV was as low as 3%, suggesting this model of care could be used to expand screening and treatment across Cambodia.
Although our study was implemented by MSF, the Ministry of Health in Cambodia and other LMIC can be empowered by these results to scale-up HCV treatment access using a simplified model of care relative to standard international guidelines. 11 Urgent scale-up of treatment access for all infected individuals is necessary to prevent HCV-related mortality and reach HCV elimination as proposed by WHO. 3

ACK N OWLED G EM ENTS
We thank Dr Yvan Hutin (World Health Organization) for providing information on the cost of care for advanced liver disease in Cambodia and for input on a previous version of the manuscript, and Dr Chhit Dimanche of Preah Kossamak hospital for expert input on liver disease care in Cambodia.

CO N FLI C T S O F I NTE R E S T
PV has received unrestricted research grants from Gilead and honoraria from Gilead and Abbvie. MI, JPD, SK, MLP, TM, SB, AL and DM were employed by MSF and/or by Epicentre, an association created by MSF in 1986 to provide epidemiological expertise to underpin MSF operations. They participated in planning this study, carrying out the research and writing the report. The other authors declare no conflicts of interest.