Mass hepatitis C virus screening among persons in custody in Hong Kong: a modelling cost-effectiveness analysis based on a prospective cohort study

Summary

Background

Persons in custody (PICs) have a disproportionately high rate of hepatitis C virus (HCV) infection, yet correctional elimination programmes remain scarce. We evaluated the cost-effectiveness of different prison-targeted tailored strategies.

Methods

Using prison-based cohort data, we modelled 10,000 PICs via a decision tree and age-stratified Markov framework to compare one-off screening or periodic screening, against no screening. Each ran under hepatologist-led teleconsultation, outreach face-to-face consultation (on-site visits) and hospital-based consultation. Costs and quality-adjusted life years (QALYs) were estimated from a societal perspective, with incremental cost-effectiveness ratios (ICERs) benchmarked to the per-capita GDP (USD 54,134).

Findings

Compared with no screening, one-off screening strategies reduced HCV-related compensated cirrhosis by 0.4–5.0%, decompensated cirrhosis by 0.6–6.8%, hepatocellular carcinoma by 0.4–5.2%, and HCV-related mortality by 0.4–3.7%. Mean costs increased from US $1326 to US $1375 and QALYs from 7.536 to 7.554 per person. ICERs for one-off strategies ranged from US $2680 to 2920 per QALY. Triennial and annual screening yielded higher health gains but with higher ICERs of US $7837 and 15,370 per QALY. Results were robust across care delivery models, prevalence scenarios, and sensitivity analyses.

Interpretation

Prison-based HCV screening with one-time screening strategies is cost-effective in a metropolitan setting. Periodic screening can further reduce disease burden but requires a higher commitment of the healthcare budget. Integration of telemedicine-supported, prison-based screening within the national hepatitis elimination strategy could accelerate HCV micro-elimination and deliver a substantial public health impact.

Funding

Health Bureau, Government of HKSAR under Health and Medical Research Fund (reference: 20211881).

Research in context.

Evidence before this study

We searched PubMed and Google Scholar for articles published in English between Jan 1, 2000, and Oct 31, 2025, using the terms “hepatitis C” AND “prison” OR “incarcerated” OR “correctional” AND “screening” AND “cost-effectiveness”. Evidence consistently shows that hepatitis C virus (HCV) prevalence among persons in custody is substantially higher than in the general population, often exceeding 10% in settings with high injection drug use. Studies from Europe, Australia, and North America demonstrate that prison-based HCV screening and treatment are feasible, achieve high cure rates, and reduce community transmission. However, most published cost-effectiveness analyses have focused on the general population, overlooking incarcerated populations despite their higher disease burden. The few prison-based economic studies mainly come from high-prevalence, high-income countries, often assessing new entrants or opt-in testing only, without long-term modelling or operational evaluation (e.g., telemedicine). Evidence from Asian high-income settings remains scarce, and no prior study has assessed the cost-effectiveness of prison HCV screening in Hong Kong or comparable moderate-prevalence contexts.

Added value of this study

This study provides new evidence on the cost-effectiveness of different prison-based HCV screening strategies and models of care delivery with moderate HCV prevalence and universal access to direct-acting antivirals. An integrated decision tree Markov model was developed to capture both short-term screening and treatment pathways and long-term HCV disease progression, incorporating inmate entry and release dynamics. One-off universal screening that includes existing inmates, new entrants, or both was found to be highly cost-effective, with incremental cost-effectiveness ratios well below the local willingness-to-pay threshold. Screening of existing inmates produced the lowest ICER, while combined screening achieved the greatest total health gains. Periodic screening identified new and reinfection cases, but at a higher cost. The study quantified how one-off versus periodic strategies differ in their long-term disease burden reduction, including downstream complications and mortality, effectively shifting treatment costs from late to curative therapy. Teleconsultation emerged as a cost-efficient and scalable model for specialist care. Sensitivity analyses confirmed robust cost-effectiveness across HCV prevalences of 1–20%, supporting generalisability to other high-income jurisdictions. A global cross-country analysis further identified data-driven thresholds (gross domestic product per capita ≥ USD 20,000; HCV prevalence ≥5%) associated with higher adoption of prison-based HCV screening and treatment strategies.

Implications of all the available evidence

Correctional facilities, housing disproportionately vulnerable and underserved populations, are critical settings for advancing HCV elimination. Implementing universal, opt-out prison-based screening and treatment can markedly reduce morbidity, mortality, and transmission while promoting equitable access to care. Telemedicine-supported models of care can overcome logistical constraints and enhance the cost-efficiency of service delivery in correctional systems. The findings suggest that prison-based screening is particularly valuable in settings with moderate to high HCV prevalence and adequate healthcare resources, where structured programmes can yield substantial health and economic benefits. Integrating correctional settings into national hepatitis plans will be essential to accelerate progress toward global elimination targets.

Introduction

Hepatitis C virus (HCV) infection affects an estimated 56 million people worldwide.¹ If it remains untreated, chronic HCV infection can progress to cirrhosis in 10–20% of patients over 20–30 years, with hepatocellular carcinoma (HCC) developing annually in 1–5% once cirrhosis is present.^2,3 Advances in understanding the HCV life cycle have enabled the development of direct-acting antivirals (DAAs), achieving sustained virological response rates >98% with favourable safety profiles.⁴ In 2016, the World Health Organization (WHO) launched the Global Health Sector Strategy on Viral Hepatitis, targeting to reduce the incidence and mortality related to chronic hepatitis virus infections including HCV by 90% and 65%, respectively, by 2030.⁵

Despite these global efforts, HCV infection remains a leading cause of cirrhosis, HCC, and liver-related deaths.⁶ By 2019, only 36% of people living with HCV had received a diagnosis, and only 20% received curative treatment.⁷ While different countries face unparalleled challenges to implement HCV elimination strategies, key barriers to eliminating HCV that are applicable to virtually all healthcare settings include low disease awareness, stigmatisation, lack of universal screening programmes, and suboptimal linkage to care. These gaps are particularly acute in underserved and high-risk populations, including persons in custody (PICs).

HCV-related mortality in Hong Kong, a global representative metropolitan city, has more than doubled over the past decade (26.2–54.8 per 1000 person-years), mirroring trends in other high-income regions where injection drug use remains the main transmission route.8, 9, 10 The high concentration of people who inject drugs (PWID) within correctional institutions, driven by the criminalisation of illicit substances, amplifies HCV prevalence among PICs, a pattern consistently reported across diverse settings, with estimates from 5% to over 20% worldwide.¹¹ Thus, while shaped by local epidemiology, the Hong Kong experience exemplifies a broader global reality in which prisons serve as critical hotspots of HCV infection and a unique setting for healthcare contact. Targeted screening and linkage-to-care in correctional institutions could therefore deliver disproportionate benefits, both within institutions and through their collateral impact on surrounding communities. In Hong Kong, routine HCV screening among PWID in the community is very limited, as shown by our recent observational study in a correctional institution.^9,12

Correctional institutions are increasingly recognised as priority sites for HCV elimination, as the cyclical movement of individuals between incarceration and the community, driven by release and high rates of reincarceration, supports ongoing bidirectional transmission and maintains the burden of infection within both prisons and the broader population.¹³ Many countries have acknowledged the importance of prison health within national hepatitis strategies, yet specific interventions and detailed epidemiological data remain scarce. The Hong Kong Government's Viral Hepatitis Action Plan 2025–2030 outlines measures for elimination but lacks targeted strategies for prisons due to insufficient local data.¹⁴

To address this knowledge gap, our group has initiated a government-funded programme in one of the correctional institutions to implement HCV screening and linkage-to-care, generating robust prevalence data and assessing the feasibility of a prison-based micro-elimination strategy.¹² Building on these findings, we modelled the cost-effectiveness of mass HCV screening in Hong Kong prisons to inform policies on elimination in correctional institutions, with relevance to similar contexts worldwide.

Methods

Study design and overall approach

We conducted a cost-effectiveness analysis of HCV screening and treatment strategies among PICs in Hong Kong, using a decision tree for the screening-treatment cascade and an age-stratified Markov model for long-term disease progression (Fig. 1). The comparator was no screening, where diagnoses generally occur only when advanced clinical manifestations (decompensated cirrhosis [DC] or HCC) prompt medical evaluation. Analyses adopted the societal perspective over the lifetime horizon of the incarcerated population, with costs and quality-adjusted life years (QALYs) discounted annually. Outcomes included disease burden, cumulative costs, QALYs, and incremental cost-effectiveness ratios (ICERs).

Fig. 1.

Decision Tree and Markov Model Simulation for Hepatitis C Virus Screening and Treatment Pathways. The decision tree illustrates the screening process and classification of individuals based on anti-HCV test results (A). Red circles indicate those who test anti-HCV negative and are excluded from further follow-up, forming a separate status category. The colours in the decision tree correspond to the respective health states in the Markov model. In the Markov model (B), all health states except death can progress to hepatocellular carcinoma, and all can transition to death. The model includes dynamic processes including the inflow of inmates representing the continuous entry of new prisoners with varying HCV statuses, reinfection capturing the potential for re-infection after sustained virological response, and regular follow-up reflecting post-treatment monitoring for recurrence or new infection. The HCV infection process represents the transition from no infection or no active infection to active HCV infection, after which individuals enter the model as undiagnosed cases starting from fibrosis stage F0. Abbreviations: HCV, hepatitis C virus; SVR, sustained virological response; DC, decompensated cirrhosis; HCC, hepatocellular carcinoma.

Construction of the simulated cohort

To generate a representative incarcerated population for model simulation, we used epidemiological and clinical data from a prison-based micro-elimination programme conducted in collaboration with the Correctional Services Department of the Hong Kong Government initiated in 2023.¹² The programme systematically implemented point-of-care anti-HCV antibody testing, reflex ribonucleic acid (RNA) confirmation, liver fibrosis assessment by non-invasive test, and initiation of fully-subsidised DAA for PICs with diagnosed HCV infection under this programme. The programme allowed the estimation of anti-HCV seroprevalence, viraemic rates, genotype distribution, fibrosis burden, and treatment uptake. Demographic and behavioural characteristics, including age, education, smoking, tattooing, intravenous drug use, past medical history and expected duration of stay in prison, were also documented.

The model simulated individuals aged 19–82 years, consistent with our prison-based cohort programme. Using official statistics on the average daily penal population (10,000 PICs), a synthetic cohort of equal size was constructed (Supplementary Materials and Supplementary Table S1). Each individual was probabilistically assigned demographic, behavioural, and clinical attributes aligned with observed data. HCV status was assigned as anti-HCV negative, anti-HCV positive but RNA negative (past exposure), or RNA positive (active infection), with viraemic cases stratified by age-specific fibrosis stage based on elastography and natural history estimates (Supplementary Table S2). Annual inflow of new entrants was modelled separately, with stability of population characteristics over time (Supplementary Materials and Supplementary Figure S1). Because access to race and ethnicity variables in Hong Kong prison health records is restricted and screening included all people in custody irrespective of these characteristics, these data were not analysed.

Screening strategies

Five alternative strategies were evaluated against no screening: (1) one-off screening of new entrants, in which all incoming prisoners were tested upon admission; (2) one-off screening of existing prisoners, involving a single round of testing for all currently incarcerated individuals; (3) combined one-off screening, covering all existing prisoners once and all new entrants at the time of admission; (4) triennial repeat screening conducted every three years; and (5) annual repeat screening, whereby the entire incarcerated population was retested yearly following an initial universal screen. Each prison admission was treated as a new screening opportunity, and entrants were retested at every admission, irrespective of previous incarceration or HCV status. Across all strategies, individuals who were screened subsequently followed a common diagnostic and treatment cascade (described below), with differences between strategies arising only from screening coverage and frequency.

Diagnostic and treatment pathways

The initial diagnostic cascade was modelled as a decision tree incorporating test accuracy, confirmatory testing, and treatment allocation. Individuals who screened anti-HCV antibody positive underwent reflex HCV RNA testing to confirm current infection. Antibody and HCV RNA test parameters were based on published sensitivity and specificity. Liver fibrosis was assessed using the fibrosis-4 index as a non-invasive test to inform clinical management and to parameterise fibrosis-stage-specific treatment outcomes. Individuals with HCV RNA positivity entered treatment with first-line pan-genotypic DAAs, delivered under direct observation, with second-line regimens provided for treatment failures (see Supplementary Materials for details). Sustained virologic response (SVR) was assessed 12 weeks after treatment completion. Treatment efficacy was varied by fibrosis stage, and the distributions of fibrosis stage and post-screening HCV status were documented (Supplementary Table S3). Patients achieving SVR transitioned into post-cure health states within the Markov framework, whereas undiagnosed or treatment-failed cases followed the natural disease progression. Based on our pilot data, coinfections with human immunodeficiency virus (HIV) or hepatitis B virus (HBV) were uncommon and incorporated as background assumptions without separate stratification.^12,14

Care delivery mode

The principal model of care was teleconsultation, with specialist hepatology assessments provided via secure videoconferencing and supported by correctional services staff stationed in the hospital (Supplementary Figure S2). This approach reduced transport and staffing costs and was considered the primary delivery modality. Scenario analyses considered two alternatives: physical outreach, with hepatology teams visiting prisons, and hospital-based consultation, involving escorted transport of PICs for in-person assessment, both incurring additional travel and manpower costs (Supplementary Materials).

Global HCV prevalence and prison programme implementation

To parameterise baseline infection prevalence, we estimated the anti-HCV antibody-positive and HCV RNA-positive rates from the local pilot programme, which were 4.4% and 4.1%, respectively (Supplementary Table S3).¹² Published country-level prison anti-HCV and HCV RNA estimates were used to define the baseline HCV RNA prevalence range,¹⁵ which was varied in scenario analyses to assess effects on costs and QALYs (Supplementary Figures S3–S4). Using the same global sources, we assembled a country-level dataset linking prison anti-HCV seroprevalence with gross domestic product (GDP) per capita in 2024 USD and the presence of an organised prison HCV screening and treatment programme. Prespecified candidate GDP thresholds were compared using logistic regression model fit, and associations with programme implementation were assessed using univariable and multivariable logistic regression (Supplementary Tables S4–S5).

Markov model structure

The Markov model captured annual transitions across health states representing the full HCV disease, including HCC, liver transplantation, and post-transplant phases. Transition probabilities and reinfection rates were derived from published estimates and converted to annual transition probabilities using an exponential model with constant hazard over each cycle to reflect disease progression, treatment outcomes, and mortality risks (Supplementary Table S6). We applied a uniform annual reinfection rate to all post-SVR health states from a de novo random-effects meta-analysis of published prison-based cohorts (Supplementary Table S6, Supplementary Figure S5). Reinfection uncertainty was examined in deterministic sensitivity analyses by varying this rate across its 95% confidence interval. Half-cycle correction was applied. The model dynamically incorporated prison entry and release, allowing evaluation of screening strategies that facilitate earlier diagnosis and treatment during incarceration and after release through community linkage. The cohort was age-stratified by observed fibrosis variation, with each defining the simulation horizon, up to 90 years, consistent with life expectancy in Hong Kong (Supplementary Material).¹⁶ Given the long incarceration duration among HCV-positive PICs,¹² most persons in custody were assumed to complete DAA treatment during incarceration, with or without SVR assessment. For persons whose HCV care extended beyond release, we modelled referral to a structured, hospital-based linkage-to-care programme (Supplementary Figure S2) to ensure continuity of HCV management after release and included all related outpatient consultations, investigations (including SVR testing), DAA treatment, and transportation costs in the analysis (Supplementary Table S7).

Costs

The analysis was conducted from a societal perspective, incorporating direct medical, direct non-medical, and indirect costs. Cost components included diagnostic testing; antiviral therapy with DAAs; consultations delivered via teleconsultation, physical outreach, and hospital-based services (including associated manpower and logistics); and long-term disease management (DC, HCC, liver transplantation, post-transplant care, and end-stage management). We applied a bottom-up costing approach. Resource quantities under each strategy were generated from the decision tree and Markov model and multiplied by unit costs. Unit prices for tests, imaging, outpatient and inpatient care, and liver transplantation were obtained from Hospital Authority fee schedules and hospital finance data. Staff costs for healthcare and Correctional Services Department personnel were based on government wage scales and converted into staff time costs. Long-term disease-management costs were implemented as annual, state-specific costs in the Markov model, derived from Hospital Authority casemix tariffs and hospital finance data and aligned with published Hong Kong economic evaluations (Supplementary Tables S6–S7).¹⁷ Persons in custody with advanced liver disease are referred to Hospital Authority hepatology services through established arrangements. After release, patients were assumed to access the same care pathways and utilisation patterns as other residents with HCV-related advanced liver disease. Indirect costs reflected productivity losses after release among individuals who developed advanced liver disease and required HCV-related healthcare utilisation, valued using the statutory minimum wage (Supplementary Table S8). All costs were accumulated over the model horizon and discounted at 3% annually, expressed in 2025 Hong Kong dollars (inflated to 2025 values where necessary), and converted to USD using an exchange rate of 7.8 Hong Kong dollars per USD to facilitate comparability.

Model inputs and parameter uncertainty

Key clinical and economic inputs are summarised in the Supplementary Materials (Supplementary Table S6). Briefly, point-of-care anti-HCV antibody testing and HCV RNA confirmation were parameterised using published sensitivity and specificity estimates, while annual fibrosis progression and downstream complication risks were taken from established natural history sources. Post-SVR reinfection was modelled as an annual probability and examined in sensitivity analyses as described above. To propagate parameter uncertainty, we assigned Beta distributions to bounded probabilities (including test performance, progression risks, reinfection, SVR, and utilities), Dirichlet distributions to the baseline age-specific fibrosis-stage distribution, and Gamma distributions to unit and state-specific cost inputs (Supplementary Table S6).¹⁸

Health outcomes and cost-effectiveness analysis

The model projected key clinical outcomes using age- and sex-specific Hong Kong life tables with baseline mortality rates adjusted to reflect the higher all-cause mortality among people who experience incarceration (hazard ratio 1.39) and with excess risks applied to DC, active HCC, and persistent HCV infection.¹⁹ Primary outcomes were incremental costs, QALYs, and ICERs. QALYs were derived from published utility weights, with decrements for DC, HCC, and post-transplant states; patients achieving sustained virologic response at early fibrosis stages regained near-normal utility. ICERs were estimated as the incremental cost per QALY gained, compared with either no screening or the next less costly alternative, excluding dominated options, and interpreted against a willingness-to-pay (WTP) threshold equal to one times Hong Kong's per-capita GDP (USD 54,134 in 2024). ICERs below this threshold were considered cost-effective. This economic evaluation was reported in line with the Consolidated Health Economic Evaluation Reporting Standards, and the completed checklist is available in the Supplementary Materials.²⁰

Model validation

Model validity was assessed by three levels through expert review, internal validity assessed by checking model consistency, and external validity by comparing simulated disease burden with real-world data in Hong Kong. External validation included sex- and custody-specific trends showing a stable male predominance (75–82% overall) and standardised all-cause mortality over time, with the model estimate (1018.2 per 100,000 person-years) within the adjusted observed annual range (973–1168) (Supplementary Material and Supplementary Figures S6–S7).

Ethics approval

The study protocol involving human participants conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the Institutional Review Board/Ethics Committee of the University of Hong Kong and the Hospital Authority Hong Kong West Cluster (reference number: UW 22-741). Informed consent in writing was obtained for the part that involved human participants.

Role of the funding source

The funders had no role in the study design, data collection, data analysis, interpretation, or writing of this report.

Results

Disease burden projections

Relative to no screening, one-off combined screening was projected to reduce compensated cirrhosis incidence by 5.0% (80.5–76.4 per 100,000 person-years in the prison population), DC by 6.8% (22.9–21.4), and HCC by 5.2% (19.3–18.3), with a 3.7% decline in HCV-related mortality (15.6–15.1) (Fig. 2). All-cause mortality in the prison population fell marginally by 0.2% (1018.2–1015.7 per 100,000 person-years). The greatest reductions were observed with annual periodic screening, lowering compensated cirrhosis by 39.9% (to 48.4), DC by 38.8% (to 14.0), HCC by 39.7% (to 11.6), and HCV-related mortality by 39.2% (to 9.5), alongside a 1.9% decrease in all-cause mortality (to 998.6).

Fig. 2.

Impact of alternative hepatitis C virus screening strategies among persons in custody in Hong Kong on projected disease burden. (A) Estimated incidence or mortality rates for hepatitis C virus-related compensated cirrhosis, decompensated cirrhosis, hepatocellular carcinoma, and all-cause mortality under each screening strategy. (B) Relative percentage changes compared with no screening. Abbreviations: HCV, hepatitis C virus; CC, compensated cirrhosis; DC, decompensated cirrhosis; HCC, hepatocellular carcinoma.

Total and age-specific costs

Age-specific analyses (Supplementary Table S9) indicated that costs were heavily concentrated among prisoners aged 40–59 years, where annual totals escalating from USD 147,359 in the 40–44 year group to peaks of USD 903,172 in those aged 50–54 years, and remaining substantial at USD 654,792 and USD 247,440 for the 60–64 and 65–69 year groups, respectively. Costs for new entrants were markedly lower across all ages, reaching at most USD 119,801 in the 55–59 year group.

Cost-component decomposition (Supplementary Table S10 and Supplementary Figures S8–S12) revealed a redistribution of expenditures from late-stage disease management to earlier diagnosis and treatment. Compared with the no-screening scenario (total USD 84.3 million), the combined one-off screening strategy incurred slightly higher total costs (USD 87.4 million), largely reflecting greater spending on DAA therapy (USD 4.75 million) and diagnostic activities. In contrast, downstream costs decreased by roughly USD 1.30 million for DC, USD 1.19 million for HCC management, and USD 0.72 million for liver transplantation.

Base-case cost-effectiveness outcomes

In the base-case analysis using the teleconsultation model, one-off HCV screening strategies substantially improved health outcomes at modest incremental cost compared with no screening. The average cost per person increased slightly from USD 1326 under no screening to about USD 1375 with one-off combined screening, while QALYs per person rose from 7.536 to 7.554 (Table 1). Across the three one-off options, ICERs ranged from USD 2680–2920 per QALY gained, with screening of existing inmates showing the lowest ICER, followed closely by the combined population and new entrants. Although triennial and annual screening yielded incremental gains in health benefits, their ICERs increased substantially to USD 7837 and USD 15,370 per QALY, respectively, indicating reduced cost-effectiveness compared with one-off screening. Based on the sequential ICERs (Table 1 and Fig. 3), one-off screening for new entrants was excluded from the cost-effectiveness frontier due to extended dominance, as its ICER (USD 2920/QALY) exceeded that of the next more effective strategy (one-off screening of existing inmates; USD 2651/QALY). All remaining non-dominated strategies lay on the cost-effectiveness frontier and were cost-effective at the WTP.

Table 1.

Cost-effectiveness outcomes of alternative HCV screening strategies among persons-in-custody in the teleconsultation model.

	Total cost (USD)	Total QALYs	Cost/person (USD) (C)	QALY/person (Q)	ICERa (USD/QALY gained) Δ(C)/Δ(Q)	ICERb (USD/QALY gained) Δ(C)/Δ(Q)
No screening	84,333,326	479,204.4	1326.2	7.536	–
One-off screening for new entrants	84,660,359	479,316.4	1331.3	7.537	2920	Extendedly dominated
One-off screening for existing inmates	87,100,940	480,236.9	1369.7	7.552	2680	2651
One-off screening for combined population	87,427,973	480,349	1374.8	7.554	2704	2917
Periodic screening every three years	117,719,005	483,464.5	1851.2	7.603	7837	9723
Periodic screening annually	162,983,328	484,321.5	2563.0	7.616	15,370	52,817

Teleconsultation model: hepatology assessments delivered remotely by videoconference with correctional health staff support (primary delivery modality). Per-person outcomes were calculated as total outcomes divided by the cumulative number of individuals included in the open cohort over the analytic time horizon (baseline inmates plus incident entrants). Results are based on an open-cohort simulation of 63,591 individuals in custody in Hong Kong over the analytic time horizon. Extendedly dominated indicates a strategy that is not on the efficiency frontier, as an alternative strategy is more effective and has a lower incremental cost per QALY gained.

Abbreviations: ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life year.

^aICERs calculated as incremental cost divided by incremental QALYs compared with no screening.

^bICERs calculated as incremental cost divided by incremental QALYs compared with the previous least costly non-dominated strategy.

Fig. 3.

Cost-effectiveness outcomes of hepatitis C virus screening strategies in Hong Kong among the incarceration-experienced population under varying HCV RNA prevalences and care delivery models. (A) Scatterplots show total cost versus quality-adjusted life-years (left panels) and incremental cost-effectiveness planes (right panels) for different screening strategies across three care models: teleconsultation, physical outreach, and hospital-based consultation. Each point represents the outcome at a given HCV RNA prevalence (1%, 4.1% [base case], 10%, and 20%). In the right panels, incremental costs and incremental quality-adjusted life-years are expressed relative to no screening, and the dashed line indicates the willingness-to-pay threshold of 54,134 United States dollars per quality-adjusted life-year gained. The efficient frontier is shown by a black line connecting strategies that are non-dominated under the base-case prevalence after exclusion of extendedly dominated strategies (one-off screening for the new entrants). (B) Incremental per-person and total costs of alternative care delivery models versus the teleconsultation mode at varying HCV RNA prevalences. Abbreviations: HCV, hepatitis C virus; RNA, ribonucleic acid; QALY, quality-adjusted life-year.

Scenario analysis

In scenario analyses, the physical outreach and hospital-based models showed cost-effectiveness patterns consistent with the base-case teleconsultation model, with strategy rankings and health gains largely preserved. At the base HCV RNA prevalence of 4.1%, one-off combined screening achieved similar outcomes across modes, with QALYs per person at 7.554 and mean costs rising slightly from USD 1375 to 1378 (Fig. 3A). Periodic screening every three years or annually provided added benefits but required markedly higher expenditure, up to USD 2576 per person in the hospital-based model. Across all modes, incremental costs for one-off screening remained well below the WTP threshold. Fig. 3B compares incremental costs between consultation modes across varying HCV RNA prevalence levels. Relative to the teleconsultation model, both the physical outreach and hospital-based models showed progressively higher per-person and total incremental costs as prevalence increased from 1% to 20%. The differences between modes remained modest at lower prevalence but widened at higher prevalence, particularly for the hospital-based model, which incurred the greatest total incremental cost.

Globally, mapped distributions of anti-HCV and HCV RNA prevalence showed pronounced geographic heterogeneity, with higher burdens concentrated in parts of Eastern Europe, South-East Asia, and Oceania (Supplementary Figure S3). To reflect this observed variation, HCV RNA prevalence was varied from 1% to 20%, reflecting estimates from 21 countries and territories where most estimates fall within this range. Total and incremental costs increased proportionally, whereas ICERs and strategy rankings remained unchanged, confirming model robustness, indicating that model outcomes were robust to variation in underlying epidemiology.

Cross-country thresholds for prison HCV programme implementation

Global cross-country analyses identified data-driven thresholds for organised prison HCV programmes: GDP per capita ≥USD 20,000 and anti-HCV prevalence ≥5%. In the multivariable logistic regression model using these dichotomisations, programme implementation was positively associated with GDP per capita (odds ratio [OR]: 4.85, 95% CI: 1.05–27.66) and prison anti-HCV seroprevalence (OR: 7.46, 95% CI: 1.28–65.58) (Supplementary Table S5).

Sensitivity analysis

Probabilistic analyses showed that nearly all one-off screening strategies were almost universally cost-effective at WTP thresholds below USD 10,000 per QALY, whereas periodic screening required higher thresholds, with the three-year strategy approaching full probability near USD 20,000 and annual screening reaching about 80% at that level and nearly 100% close to the threshold (Fig. 4). Deterministic sensitivity analyses showed that initial DAA treatment cost and reinfection rate were key ICER drivers across strategies (combined screening of both existing inmates and new entrants: reinfection rate, ICER USD 1439–3882; initial DAA treatment cost, ICER USD 1130–4313). However, the main conclusions were unchanged across the ranges examined (Supplementary Figure S13).

Fig. 4.

Probabilistic Cost-Effectiveness Analysis of Alternative Hepatitis C Virus Screening Strategies Compared with No Screening. Scatter plots in Panel A show the joint distribution of incremental costs and quality-adjusted life-years across 1000 probabilistic simulations for each screening strategy relative to no intervention. Ellipses represent 95% confidence regions. Curves in Panel B display the probability that each strategy is cost-effective relative to the status quo across a range of willingness-to-pay thresholds. The vertical dashed line marks a threshold of one times the gross domestic product per capita (USD 54,134). Abbreviations: QALY, quality-adjusted life-year; GDP, gross domestic product; USD, United States dollars.

Discussion

HCV elimination remains a global health priority, yet incarcerated populations continue to be underserved despite their disproportionately high infection burden and potential for rapid micro-elimination through targeted intervention.⁵ This study provides strong evidence that one-off HCV screening in custodial institutions, whether targeting existing PICs, new entrants, or both groups, offers measurable health gains at a modest incremental cost, with ICERs well below the local willingness-to-pay threshold. Among the one-off strategies, screening of existing inmates yielded the lowest ICER, indicating the most efficient use of resources, while combined screening of both existing inmates and new entrants achieved greater overall population benefit and broader preventive impact. While annual or triennial periodic screening produces greater total health gains by capturing incident and reinfection cases, these approaches require substantially higher investment and may be more suitable as a second phase intervention once the initial elimination gains from one-off screening have been achieved.

International evidence supports phased approaches in settings where resources or operational capacity are constrained. Opt-out reception screening for new prison entrants has achieved high uptake and timely treatment initiation across diverse contexts. In England, a nurse-led protocol across 47 prisons increased antibody testing coverage from 32% to 86% over three years, with 97% of HCV RNA-positive cases commencing DAAs;²¹ similar initiatives in the United States have delivered sustained improvements.²² In our Hong Kong modelling, a one-off new entrant screening strategy remained well within cost-effectiveness thresholds despite smaller total health gains than combined universal screening. This evidence positions reception-based opt-out screening as a pragmatic, transferable micro-elimination option, particularly valuable in resource-limited correctional systems.

Implementation feasibility is central to success. The teleconsultation model in our analysis leveraged specialist hepatology expertise without the logistical and security burdens associated with in-person outreach, maintaining clinical quality while reducing manpower and transport costs. Such digital health integration is highly transferable to other settings that face specialist shortages and logistical complexity, and positions telemedicine as a key enabler of both one-off and periodic prison-based elimination strategies if such infrastructure is available.

For high-income settings where DAA treatment is readily available, the greatest barrier to elimination lies in timely diagnosis. A recent Australian analysis showed that optimised point-of-care testing algorithms can lower the cost per treatment initiation across a range of HCV antibody prevalences,²³ whereas our analysis focuses on screening frequency and target population selection in an open custodial cohort. Taken together, these findings indicate that both diagnostic pathway design and screening intensity are key determinants of the cost-effectiveness of prison-based HCV micro-elimination strategies, and the preferred approach will vary across settings according to local epidemiology, service capacity, and available resources. Given the high prevalence of HCV among people who inject drugs in correctional institutions, delayed case finding and linkage to care in these environments risks perpetuating ongoing transmission and missing critical opportunities for intervention.²⁴ The existing curative approach, characterised by late-stage detection of many diseases, sustains unnecessary mortality and drives rising demand for tertiary care interventions.²⁵ Our modelling demonstrates that proactive universal screening reorients spending toward earlier phase of disease and can shift resources from costly end-stage management toward curative care, achieving both economic and clinical benefits, including over 30% reduction in projected HCC and HCV-related mortality. This strategy represents a WHO-aligned micro-elimination option that highlights the critical role of correctional health systems in advancing regional and national hepatitis C elimination goals.

The impact of these strategies extends beyond prison health. Incarcerated populations offer a unique opportunity for micro-elimination, where structured screening and treatment programmes can yield disproportionately high returns in reducing transmission chains.¹¹ The treatment of hepatitis C infection among incarcerated populations prevents progression to severe liver disease, such as cirrhosis and HCC, and, by reducing the pool of infectious individuals, helps to disrupt transmission chains and decrease reinfection rates among people who inject drugs and other vulnerable groups after release to the community, known as the ‘treatment-as-prevention’ effect.^26,27 Periodic screening magnifies the impact of secondary prevention by facilitating the early identification of incident cases.²⁶ This reflects the principle of dynamic micro-elimination, whereby both the prevalence and incidence of HCV are proactively reduced within targeted high-risk populations through ongoing surveillance and intervention.²⁸

From a global perspective, these findings highlight challenges in aligning prison-based strategies with international elimination frameworks, particularly where high-risk groups remain underserved.^5,29 Among 35 countries and territories with data available, only 20 (57.1%) had implemented prison-based HCV screening models, and these strategies remain unavailable in many regions with very high (>10%) anti-HCV prevalence among prisoners (Supplementary Figure S4). Epidemiological data for HCV infection in the incarcerated population are also lacking in many settings. Despite advances in antiviral therapies, global diagnostic and treatment coverage remains far from elimination thresholds, particularly within marginalised settings such as prison.³⁰ Our cross-country analysis suggests that implementation of organised prison HCV programmes aligns with both health-system capacity and infection burden. In Hong Kong, GDP per capita exceeds the economic threshold (USD 20,000) identified in the cross-country stratification, indicating capacity to support prison-based HCV screening and treatment, and although the prevalence of anti-HCV among PICs in Hong Kong (4.4%) is slightly below cut-off (5.0%), prison-based HCV screening is still cost-effective across a range of prevalence from 1% to 20% (Fig. 3). Our results support a tiered sequencing model, starting with one-off combined screening for cost-effective, rapid case finding, and then transitioning to periodic screening, conducted annually in high-income jurisdictions and triennially in settings where resources are more limited, to sustain gains and capture incident infections. This blueprint can be adapted across diverse health system contexts, balancing efficiency with elimination ambition.

There are several limitations to this study. Although our modelling was informed by local pilot programme data from a correctional institution, some clinical and epidemiological parameters, such as reinfection probabilities, were derived from published studies conducted in other settings. To address this, we prioritised the use of meta-analyses where possible to ensure more reliable estimates. The analysis did not explicitly model the impact of treatment on onward HCV transmission within prisons or after release into the community. This may lead to an underestimation of the broader benefits of screening and treatment, particularly for periodic strategies. Community-level transmission was not modelled because the analysis focused on persons in custody with long incarceration durations and high treatment and linkage-to-care coverage, and this omission is unlikely to materially affect the conclusion that prison-based HCV screening and treatment are cost-effective. Co-infections, particularly HIV, were not modelled in detail due to their low prevalence in the study population and the anticipated limited effect on the overall results. Nevertheless, it is worth exploring the same approach for PWIDs with co-infection with HIV and/or HBV in the future. We also did not include post-release criminal justice or social care costs due to lack of setting-specific evidence, which may make our estimates conservative. The incarcerated population in Hong Kong is predominantly male and our cohort consisted mainly of men, so sex-stratified analyses were not conducted and gender identity was not collected. Despite these limitations, our findings provide important evidence to inform policy decisions regarding HCV screening and treatment strategies among persons in custody in Hong Kong, which may also serve as an example for regions with similar societal and health economic backgrounds.

In many regions, routine HCV screening within correctional institutions remains limited, fragmented, and largely dependent on risk-based or voluntary testing, which fails to identify asymptomatic infections and perpetuates missed opportunities for timely treatment. Establishing a structured universal opt-out screening programme, integrated with telemedicine consultation and DAA therapy, could fundamentally transform the current care paradigm. Such integration would ensure equitable access to diagnosis and treatment regardless of facility resources, streamline specialist input, and minimize logistical and security barriers associated with in-person outreach. By embedding HCV screening and treatment into prison health systems as a standard of care, correctional institutions could evolve into crucial public-health platforms that not only safeguard the health of persons in custody but also yield substantial community-level benefits. Sustained viral suppression and reduced post-release transmission would contribute directly to national elimination efforts, demonstrating how correctional health reforms can act as catalysts for broader health-system strengthening and infectious-disease control.

In conclusion, one-off universal screening of all PICs in Hong Kong was found to be cost-effective for HCV elimination in correctional institutions, achieving health gains at modest incremental cost compared with no screening. This approach enabled earlier diagnosis and treatment, reduced projected HCC incidence, and shifted health expenditures from end-stage disease management to curative therapy. While annual periodic screening offers greater health gains, its higher cost demands careful prioritisation. Embedding systematic prison-based screening supported by telemedicine into the national hepatitis action plan may accelerate micro-elimination and contribute to achieving the WHO 2030 targets.

Contributors

ZD, XL and LYM designed the study and developed the economic model. KTW and EC coordinated and conducted the prospective cohort in the correctional institutions and collected clinical and programme data. ZD curated the data and performed the model simulations and statistical analyses under the supervision of XL and LYM. WKS and MFY provided clinical expertise in hepatology and contributed to the design and interpretation of the screening and treatment pathways. ZD drafted the first version of the manuscript. XL, LYM, WKS, and MFY critically revised the manuscript for important intellectual content, and all authors reviewed and approved the final version. All authors had full access to and verified the underlying data. XL and LYM, as co-corresponding authors, had final responsibility for the decision to submit the manuscript for publication.

Data sharing statement

The data supporting the findings of this study are available within the article and its supplementary materials. Anonymised participant-level data will be made available on reasonable request from the corresponding author, subject to approval by the relevant institutional ethics committee and in accordance with applicable regulatory requirements.

Editor note

The Lancet Group takes a neutral position with respect to territorial claims in published maps and institutional affiliations.

Declaration of interests

XL received research grants or contracts from the Health and Medical Research Fund (HMRF Main Scheme, HMRF Fellowship Scheme, and Hong Kong Special Administrative Region), and from the Research Grants Council Early Career Scheme, Research Impact Fund (HKSAR); is also the former nonexecutive director of ADAMS Hong Kong; received commission grants from Hospital Authority of Hong Kong, internal funding from the University of Hong Kong, and research or education grants from Pfizer, Janssen and Bristol Myers Squibb (BMS), Novartis; received consultancy fees from Merck Sharp & Dohme, Pfizer, Open Health and The Office of Health Economics; and received honoraria for associate editorship from Nature Springer.

LY Mak received funding support from Gilead Sciences and Roche Diagnostics. WKS received speaker's fees from Echosens and MSD, is an advisory board member and received speaker's fees of Abbott, received research funding from AstraZeneca, Alexion Pharmaceuticals, Boehringer Ingelheim, Pfizer and Ribo Life Science, and is an advisory board member, received speaker's fees and research funding from Gilead Sciences. MF Yuen serves as advisor/consultant for AbbVie, Assembly Biosciences, Aligos Therapeutics, Arbutus Biopharma, Bristol Myers Squibb, Clear B Therapeutics, Dicerna Pharmaceuticals, Finch Therapeutics, GlaxoSmithKline, Gilead Sciences, Immunocore, Janssen, Merck Sharp and Dohme, Hoffmann-La Roche and Springbank Pharmaceuticals, Vir Biotechnology and receives grant/research support from Assembly Biosciences, Aligos Therapeutics, Arrowhead Pharmaceuticals, Bristol Myers Squibb, Fujirebio Incorporation, Gilead Sciences, Immunocore, Merck Sharp and Dohme, Hoffmann-La Roche, Springbank Pharmaceuticals and Sysmex Corporation. The remaining authors have no conflicts of interest.