The Microsoft® Excel-based Markov model uses annual disease progression rates to simulate the natural history of HCV through disease stages from acute hepatitis (considering spontaneous clearance) to chronic fibrosis and end-stage outcomes (including hepatocellular carcinoma, decompensated cirrhosis, liver transplantation and liver-related death). (9) The model is seeded with population and background mortality data specific to Croatia (10)as well as country-level HCV epidemiological data (Table 1). New infections entering the model were back-calculated between 1950 and 2018 using known prevalence. Annual case counts were calculated by applying a progression rate to each disease stage and adjusted for all-cause mortality, ageing, and cure. All-cause mortality rates (extracted from the United Nations mortality database) were adjusted to increase incrementally in relation to transfusion and injection drug use trends.

A full list of epidemiological inputs is included in Table 1. Since many of the historical inputs have remained consistent with the original analysis (4), only inputs that have changed are described in detail here.

HCV-RNA Positive Infections – There were an estimated 21,980 Croatians chronically infected with viremic HCV (0.51% prevalence) in 2010. This was calculated by adjusting the reported adult anti-HCV prevalence estimate (0.90% anti-HCV among adults aged 20–85 in 2010) (11) to account for all ages (0.73% anti-HCV among all ages in 2010) and applying a 70% viremic rate. (12) HCV Genotype – Data from the three largest Croatian national laboratories collected over nine years on 3,655 patients reported the majority of infections were genotype 3, 1b and 1a, respectively. (13) Total and Annual Diagnosed – The number of people diagnosed via antibody or RNA testing and reported to the Croatian Institute of Public Health was available annually for 1994–2018. (14) Experts estimated that 4,000 total patients were diagnosed with chronic HCV as of 2014. This is based on an estimated average of 300 diagnoses annually from 1993–2010 and an estimated average of 200 diagnoses annually from 2011–2014, with adjustments for 70% viremia (12), mortality and cured patients. (10) Annual Treated – In 2018, about 470 patients were treated, and treatment eligibility was expanded from ≥F3 to remove all restrictions (expert input). Risk Factors – Two risk factors associated with a higher HCV-related mortality were included in the model. Experts estimated that 23% of 2018 total infections resulted from a blood transfusion and that 14% of the infected population were among people who inject drugs actively (injecting within the last year [PWID]). The PWID population was calculated by first estimating the number of drug users in Croatia (n=15,444) (15), of which experts estimated 74% were PWID. After applying a 35% anti-HCV prevalence among PWID (16)and adjusting for 70% viremia (12), 2,800 HCV-RNA positive PWID are estimated for 2018; 14% of total infections (n=2,800/20,000). The number of liver transplants were available from 1999–2017 (17), of which 15% were estimated to be HCV-related (expert input).

After seeding the model with Croatia-specific epidemiological factors, the Base scenario and four intervention scenarios were modeled for 2015–2030 (Table 2). Data from 2015 is the baseline measure in the World Health Organization’s Global Health Sector Strategy hepatitis elimination targets (WHO elimination targets) and was therefore utilized as the starting point for all scenarios. (2) As of 2018, all treatment restrictions based on fibrosis stage and patient age were dropped, such that for all five scenarios, patients aged 18–84 and fibrosis stages F0-F4 were eligible for treatment with high sustained virologic response (SVR) therapies. (18)

Screening trends were assessed for 2018–2030, under the following assumptions. Base: in 2018 and 2019, HCV screening occurred primarily through blood donation centers, voluntary centers, laboratories and screening vans. As a result, the average prevalence in the population screened is estimated to be the same as the general population prevalence in Croatia. Intervention Scenarios: Future screening campaigns are expected to target mixed-risk populations through primary care centers (estimated to be an average of twice the general population prevalence, in 2020) (Appendix, Table 2). To account for loss to follow-up within screening campaigns, only 90% of newly diagnosed patients were estimated to be initiated on treatment annually. Finally, it was assumed that procedures would be optimized to screen each individual only once.

Base – Objective: Model the current standard of care for HCV in Croatia. Assumptions: In 2018 and 2019, 30,000–40,000 patients were screened annually, of which 170 were diagnosed with viremic HCV and reported to the Croatian Institute of Public Health. The number of patients treated each year with DAAs, estimated at 470 patients in 2018, was expected to drop rapidly starting in 2022 to reflect a decreasing pool of previously diagnosed patients (expert input).

Screen 75K – Objective: Estimate the impact of an awareness campaign to increase screening rates. Assumption: Screen 75,000 Croatians annually, starting in 2020.

Screen 100K – Objective: Estimate the impact of an awareness campaign to ambitiously increase screening rates. Assumption: Screen 100,000 Croatians annually, starting in 2020.

Graduated Screening + Link to Care – Objective: Estimate the impact of an awareness campaign to ramp up screening over time while improving linkage to treatment for previously diagnosed patients. Assumptions: Increase screening to 100,000 people annually, by 2022. Treat patients who were diagnosed in previous years but were lost to follow-up.

WHO Elimination Targets – Objective: Quantify the level of effort needed to eliminate HCV in Croatia based on the four WHO elimination targets. Assumptions: Reduce new infections by 80%, diagnose 90% of all infections, treat 80% of eligible patients, and reduce liver-related mortality by 65%, between 2015 and 2030. (2)

Crystal Ball, an Excel add-in by Oracle, was used to run Monte Carlo simulations to generate 95% uncertainty intervals and sensitivity analysis. Beta-PERT distributions were used for all uncertain inputs.

At the beginning of 2018, there were an estimated 20,000 (95% uncertainty interval (UI): 14,500 – 24,000) viremic infections in Croatia. By the end of 2018, 19% (n=3,700) of infections were diagnosed. Of the total infected population, only 2% (n=470) were initiated on treatment. At an SVR rate of 97%, 460 of these patients were estimated to be cured (Figure 1). Sixty percent of patients with HCV in 2018 were estimated to have no fibrosis or early stage fibrosis (F0-F1), 30% had F2-F3, and 10% had F4 cirrhosis or end-stage liver-disease. Most infections (n=20%) were in adults aged 45–49 years.

Under the Base, the number of viremic cases peaked in 2007 and will continue to decline by 30% between 2015 and 2030, resulting in 14,100 cases by the end of 2030. This decline is largely due to mortality. If no change is made to the HCV treatment paradigm in Croatia, liver-related deaths (LRD), incident hepatocellular carcinoma (HCC), and incident decompensated cirrhosis (DC) will increase by 70%, 80% and 90%, respectively. LRD will increase from 60 in 2015 to 110 in 2030, and incident HCC will increase from 50 in 2015 to 90 in 2030. Annually, new cases of DC will also increase from 40 in 2015 to 70 in 2030 (Table 3, Figure 3). Croatia is estimated to screen a total of 460,000 patients and diagnose 2,700 patients from 2018–2030 (Figure 4).

By screening 75,000 Croatians annually and linking 90% of newly diagnosed patients to treatment, total infections would decline by 45%, but LRDs, incident HCC and incident DC would increase 45–55% from 2015–2030. Alternatively, if 100,000 Croatians are screened each year (with 90% of diagnosed patients linked to treatment), total infections would decline by 50%, compared with a 20–30% increase in LRD, incident HCC and incident DC. With graduated screening and linkage to care for previously diagnosed patients, total infections would decrease 55%, with a 5% reduction in LRD and 13–20% increase in HCC and DC from 2015–2030 (Table 3, Figure 3). All three enhanced screening scenarios require a 3–4-fold increase in total screenings and 2–3-fold increase in total diagnoses, compared to current practices (Figure 4).

Compared with the base scenario, eliminating HCV would avert 500 LRDs, 380 incident cases of HCC and 300 incident cases of DC by 2030 (Table 2). Achieving WHO elimination targets would reduce total infections by 17,000, decrease annual LRD by 40, and decrease incident cases of HCC and DC by 30 and 20, respectively, from 2015–2030 (Table 3, Figure 3). Elimination targets can be achieved by increasing screening, treatment, and linkage to care. Screening must ramp up to about 250,000 tests annually so that around 1,500 patients can be diagnosed and treated each year (Table 2). Compared to the Base, 5–6 times the total number of patients must be screened and diagnosed from 2018–2030 to achieve elimination targets (Figure 4).

This project was supported by Gilead Sciences. Gilead Sciences had no input on the content, study design, data selection, decision to publish, or preparation of the manuscript.

The authors gratefully thank the following organizations for their involvement in consultations in forming this analysis: the Croatian Institute of Public Health, University Hospital of Infectious Disease Zagreb, Split University Hospital, Merkur Hospital Zagreb, Croatian Health Insurance Fund and Hepatos.

Reports of the number of patients diagnosed with chronic HCV were provided by in-country experts for 2014–2018. Case counts were adjusted assuming 80% were viremic (expert input). Experts agree underreporting exists in Croatia, as some patients seek anonymous testing in blood donor clinics and these cases are not consistently reported. The rate of underreporting is unknown; however, experts estimate about 20% of the total infected population is diagnosed. The data shown in Table 1 represent known diagnosed cases, however true estimates are likely higher. Annual treatment estimates were based on the number of patients covered by insurance and therefore received DAAs, and an additional cohort who obtained treatment from alternative sources (expert input).

The screening module was developed to calculate the number of screening tests necessary to diagnose a given number of HCV-infections. The module used aforementioned epidemiological inputs (the infected population stratified by age and the number of annual diagnoses) to calculate the population eligible for screening, i.e., the undiagnosed, HCV-infected population that was either asymptomatic or not linked-to-care.

In the model, the more advanced stages of liver disease were assumed to be diagnosed first. It was also assumed that cases of spontaneous clearance were diagnosed at the same rate as asymptomatic, chronically infected cases.

To calculate the size of the population eligible for screening, the module tracked the populations with a history of screening, diagnosis, or SVR. It was assumed that each individual would receive at most one HCV antibody screen, and populations with a history of diagnosis or SVR would be excluded from future screening. Populations outside of the age range eligible for screening were excluded.

The number of patients needed to screen (NNS) to find one HCV antibody positive case was calculated, as shown in Equations 1–2. We then calculated the annual number of HCV antibody screens performed (Equation 3), of which all HCV antibody-positive tests were assumed to receive confirmatory HCV RNA testing. The case-finding algorithm assumed that persons with advanced liver disease were symptomatic and would seek care with or without active screening campaigns. Thus, each newly diagnosed case of advanced liver disease was assumed to require two screens; one HCV antibody and one confirmatory HCV RNA test.

Equation 1. Number needed to screen to diagnose one HCV antibody-positive case, unadjusted, in year t

, where

Population eligible for screening_t was the population eligible for screening in year t. Eligibility of screening was determined by birth year (by scenario), history of having received a screen within the screening strategy (excluded), history of diagnosis (excluded), and history of SVR (excluded).

Equation 2. Number needed to screen to diagnose one HCV antibody-positive case, adjusted, in year t

, where

is the number needed to screen to diagnose one HCV antibody-positive case, unadjusted, in year t;

k is the ratio of prevalence in the population where the screening is conducted to that in the general population.

Equation 3. Number of HCV antibody screens performed, in year t

Newly diagnosed symptomatic HCV antibody-positive cases_t + NNS_t ×

Newly diagnosed asymptomatic or not yet linked-to-care cases_t, where

NNS_t is the number needed to screen to diagnose one HCV antibody-positive case in year t.

Appendix Table 1 shows the number of patients we would expect to be diagnosed annually, based on the average number of patients screened each year for 2020 through 2030. An average was calculated to account for the depleting pool of infected individuals over time that necessitates additional screens. A range of 1 to 3 times the 2018 viremic prevalence (0.46%–1.48%) was used to account for screening practices in the general and at-risk populations. A sensitivity of +/– 20% was also considered. Results were as follows:

Screening an average of 40,000 people/year results in an average of 200–650 new viremic diagnoses annually.

Screening an average of 75,000 people/year results in an average of 350–900 new viremic diagnoses a year.

Screening an average of 100,000 people/year results in an average of 450–1,1500 new viremic diagnoses a year.

Experts estimate that at least 40,000 Croatians were screened in 2018. The majority were tested in blood banks, with no tests conducted in primary care setting (expert input; Dr. Vince). Experts envisioned a goal of 75,000 Croatians screened annually, starting in 2020 (Appendix Table 2). This would involve initiating a national action plan including an awareness campaign in primary care settings and maintaining testing levels in other settings.