Skip to main content
NCBI home page
HRB Open Research logoLink to HRB Open Research
. 2025 Jul 23;8:55. Originally published 2025 Apr 15. [Version 2] doi: 10.12688/hrbopenres.14126.2

Protocol for the economic evaluation of LCS in Ireland: modelling costs, eligibility, and outcomes

Tatiana Bezdenezhnykh 1,a, James O'Mahony 2, Benjamin Jacob 1, Deirdre Murray 3, Daniel Ryan 4, Jarushka Naidoo 4, Seamus Cotter 5, Alan Smith 6, Patrick Redmond 1
PMCID: PMC12416482  PMID: 40927357

Version Changes

Revised. Amendments from Version 1

This version includes substantial revisions in response to peer reviewer feedback. In Work Package 1, we clarified the base-case eligibility criteria (≥20 pack-years, quit within 15 years) and outlined comparative analyses using alternative criteria (e.g., NELSON, USPSTF 2013/2021). We expanded the description of the Markov model calibration. In Work Package 2, we addressed limitations in Irish diagnostic cost data by clarifying the use of HIPE and DRG-based estimates for LDCT and biopsy, and outlined the approach to model high-cost drugs using NCRI data by stage. We included references to cBioPortal and Ngo et al. (2023) to inform cost assumptions. We also noted that immunotherapy uptake changes possible in future treatment patterns will be explored through scenario analyses. In Work Package 3, we updated the modelling strategy to reflect the 2022 UK NSC ENaBL model. We clarified the use of stage-shift rather than direct mortality reduction inputs from NLST/NELSON. We added new text on how performance status will potentially be indirectly modelled using survival modifiers by detection mode and how the use of NLST-based stage distribution, while conservative, aligns with UK model assumptions. We included a section on model validation, outlining key metrics (e.g., stage distribution, detection rates, mortality impact) and external benchmarks (e.g., NLST, UKLS).

Abstract

Background

Lung cancer (LC) is the leading cause of cancer death in Ireland, yet no national screening programme exists. While low-dose computed tomography (LDCT) screening reduces lung cancer mortality by approximately 20% in high-risk populations, its cost-effectiveness in Ireland remains uncertain. Evidence on the economic burden of lung cancer care and the feasibility of screening is needed to support policy decisions.

Aim

This research programme will evaluate the economic impact of lung cancer care in Ireland and assess the cost-effectiveness of LDCT screening. By integrating screening eligibility modelling, stage-specific cost analysis, and economic evaluation, the study aims to generate evidence to support resource allocation and policy development.

Methods

The programme consists of three interlinked work packages. First, screening eligibility will be estimated using a dynamic Markov model that integrates demographic data from the Central Statistics Office (CSO), population projections, and smoking history data from Eurobarometer. Second, a stage-specific cost analysis will be conducted using a discrete event simulation (DES) model informed by data from the National Cancer Registry Ireland (NCRI), the Healthcare Pricing Office (HPO), and other healthcare reimbursement sources. Third, a cost-effectiveness analysis will adapt a UK-based LC natural history model (updated ENaBL model 2022) to evaluate alternative screening strategies, incorporating Irish-specific costs, clinical outcomes, and quality-adjusted life-years (QALYs)

Results and Implications:

This programme will generate evidence to inform the design of a cost-effective LCS programme in Ireland. Findings will guide healthcare planning, optimise screening strategies, and support sustainable policy decisions.

Keywords: Lung cancer, Cost analysis, Cost-Effectiveness Analysis, Risk Assessment, Screening eligibility

Introduction

Lung Cancer: A Public Health Burden

In Ireland, lung cancer (LC) remains a leading cause of cancer-related deaths, with no organised national lung cancer screening (LCS) programme currently in place 1 . Unlike some other cancers where survival has improved due to advances in early detection and treatment, LC continues to have poor survival outcomes, primarily due to late-stage diagnosis. The five-year survival rate in Ireland remains below 20%, reflecting the need for improved early detection strategies 2 .

Despite strong international evidence supporting the effectiveness of low-dose computed tomography (LDCT) screening in reducing LC mortality, Ireland has yet to implement a national screening programme. Several barriers, including uncertainties surrounding cost-effectiveness, healthcare capacity, and resource allocation, have delayed policy action. Currently, LC is diagnosed predominantly through symptomatic presentation, which often occurs at an advanced stage when treatment options are more limited, costly, and less effective 3 .

Countries such as the United States, United Kingdom, Australia, China, Portugal, and Hungary have introduced or piloted LDCT-based LCS programmes following the results of major trials, including the National Lung Screening Trial (NLST) and the Dutch-Belgian NELSON study 4, 5 . However, the feasibility and cost-effectiveness of such programmes depend on national healthcare structures, smoking prevalence, participation rates, and economic considerations. These factors remain underexplored in Ireland, creating a critical gap in evidence to inform policy decisions. This research programme aims to address these gaps by conducting a comprehensive economic evaluation of LCS, integrating screening eligibility modelling, stage-specific cost analysis, and cost-effectiveness assessment.

Epidemiology and Mortality Trends in Ireland

LC trends in Ireland differ markedly between men and women. While LC mortality rates in Irish men have declined significantly since the mid-1980s, rates among women have remained stable or increased. In 2012, Ireland had the eighth lowest LC mortality rate among men in Europe but the fifth highest among women 1 . This divergence reflects historical smoking trends, where smoking prevalence among women increased later than in men, leading to a delayed but rising burden of lung cancer.

Projections suggest that these patterns will persist. Between 2015 and 2045, the age-standardised incidence rate of LC in Ireland is expected to decrease by 16% in males but increase by 29% in females 2 . This shift has significant implications for healthcare planning, as the growing burden of LC among women may lead to increased demand for diagnostic and treatment services. Without effective early detection measures, the rising incidence will contribute to further strain on the Irish healthcare system.

Economic Burden of LC in Ireland

LC care is among the most resource-intensive areas of oncology, with treatment costs increasing substantially for patients diagnosed at an advanced stage. The introduction of novel therapies, particularly immunotherapy and targeted treatments, has improved patient outcomes but has also substantially increased costs 6, 7 . Late-stage LC is associated with prolonged hospital stays and intensive treatment regimens, placing further pressure on healthcare budgets. According to the NCRI, in 2016, lung cancer was the leading cancer type with health service costs attributable to modifiable risk factors, totalling approximately €62 million, all of which was linked to smoking 8 .

Studies from other healthcare systems suggest that earlier detection through LDCT screening may reduce overall treatment costs by shifting diagnoses to earlier, more treatable stages. However, there is a lack of Ireland-specific data on the economic burden of LC by stage at diagnosis. Without such evidence, it is difficult for policymakers to determine the potential financial impact of a national screening programme and to allocate resources efficiently. Understanding the cost implications of LC treatment at different stages of disease progression is therefore a critical component of this research.

The Importance of Early Detection

Early detection is the most effective strategy for improving LC survival. LDCT screening has been shown to reduce LC mortality by approximately 20–24% among high-risk populations 9 . The NLST reported a 20% reduction in LC mortality with LDCT screening compared to no screening, while the NELSON trial demonstrated a 24% reduction in men and up to 33% in women 10, 11 . In response to this growing body of evidence, the European Commission’s 2022 Council Recommendations on Cancer Screening identified LC as a priority for early detection, encouraging member states to develop risk-based screening approaches 9 .

Economic evaluations from several countries, including the United Kingdom, China, Australia, and Portugal, indicate that LDCT screening is likely to be cost-effective, particularly for individuals aged 55–75 with a smoking history of at least 20 pack-years 12– 16 . However, implementing a screening programme is not without challenges. LDCT screening introduces additional costs related to follow-up investigations, management of incidental findings, and the risk of overdiagnosis. Designing an efficient and sustainable screening programme requires a balance between clinical benefit and economic feasibility, ensuring that screening efforts are targeted towards those who will benefit most.

According to European Commission Country Cancer Profiles, Ireland has the second-highest rate of new cancer diagnoses in the EU, suggesting that the cost of cancer care is set to rise 17 . Without a coordinated screening programme, Ireland risks failing to achieve the reductions in LC mortality observed in other countries. However, before implementation, critical evidence is needed to evaluate the financial and healthcare implications of screening within the Irish context.

Key Evidence Gaps in Ireland

Despite the growing international consensus on the benefits of LCS, several key knowledge gaps hinder policy development in Ireland. The first major gap concerns the size of the high-risk population eligible for LDCT screening. While national smoking prevalence data exist, there is a lack of detailed pack-year history estimates, making it difficult to determine how many individuals would qualify for screening. Additionally, expected participation rates in an Irish LCS programme are unknown, introducing uncertainty into any cost-effectiveness projections.

The second gap relates to the economic impact of screening. No study has assessed the stage-specific costs of LC care in Ireland or estimated the potential cost savings from earlier detection. Understanding the total financial implications of implementing a national screening programme requires detailed cost data, including screening costs, follow-up procedures, and treatment expenses by disease stage.

Finally, Ireland, like many countries, has limited healthcare resources. A cost-effectiveness study can show whether implementing LCS is a good use of public funds compared to other healthcare interventions. By demonstrating who should be screened, the expected costs, and long-term benefits, it can guide policy decisions, secure funding, and ensure the program is both effective and financially sustainable. Cost-effectiveness analysis will inform a budget impact analysis (BIA), which estimates how much funding the Irish healthcare system (HSE) would need to roll out LCS nationwide.

Aim and objectives

The overarching aim of this research programme is to conduct a comprehensive economic evaluation of LCS in Ireland, integrating screening eligibility modelling, stage-specific cost analysis, and cost-effectiveness assessment. This study will provide evidence to support policy decisions on the feasibility and design of a national screening programme.

The specific objectives are:

1.   To estimate the number of individuals eligible for LCS in Ireland, using demographic and smoking history data to model eligibility and participation rates.

2.   To quantify the stage-specific healthcare costs of lung cancer, capturing costs associated with treatment, follow-up care, and end-of-life management.

3.   To evaluate the cost-effectiveness of different LCS strategies, considering alternative screening frequencies, eligibility criteria, and resource implications.

Methods

Study design

This study is comprised of three interlinked work packages: (1) screening eligibility estimation, (2) stage-specific cost analysis, and (3) cost-effectiveness analysis (CEA). The study follows guidelines for economic evaluations and adheres to ISPOR-SMDM best practices for modelling studies, where relevant 18, 19 . The analysis is conducted from the healthcare payer perspective (Health Service Executive, HSE), capturing direct medical costs associated with LCS, diagnosis, and treatment.

Work Package 1: Estimating the Eligible Population for Screening

This work package will estimate the number of individuals eligible for LDCT screening in Ireland based on smoking history, demographic trends and population projections. The analysis will focus on individuals aged 55–74 years with a smoking history of 20+ pack-years who quit within past 15 years (Irish criteria) and compare it to the eligibility under other wide-used criteria (NELSON, USPSTF2013/21).

Data sources and Inclusion criteria

Three key publicly available datasets will be used:

•   Census of Population 2022 (CSO): Provides age- and sex-stratified demographic data for the Republic of Ireland 20 .

•   CSO Population Projections (2022–2045): Enables long-term forecasting of screening demand 21 .

•   Eurobarometer 87.1 (2017): Provides individual-level data on smoking status, smoking duration, and intensity (pack-years), allowing estimation of the high-risk population meeting LDCT eligibility criteria 22 .

Individuals will be considered eligible if they meet the given pack-year threshold, calculated using smoking duration and intensity data from Eurobarometer. Exclusion criteria include never smokers and individuals outside the eligible age range.

Modelling approach

Using pack-year distributions from Eurobarometer data and Census 2022 data on smoking prevalence, the model will determine the proportion of individuals who exceed the eligibility threshold for screening in the base year (2022). A sensitivity analysis based on English real-world data from the UKLS and TLHC pilot programmes will supplement the base-case analysis on based on Eurobarometer pack-years. This will test the model robustness against known participation patterns (e.g., 50% response rate to invitation and 50% of responders meeting criteria). In addition, population forecasts from the Central Statistics Office (CSO) (2022–2045) will be integrated to estimate the long-term demand for LCS in Ireland. A dynamic Markov-based population model will be used to estimate the number of high-risk individuals eligible for LDCT screening over time. The model will simulate the evolution of smoking behaviours, incorporating rates of smoking initiation, cessation, and relapse to reflect changes in the distribution of current and former smokers. For validating our results, we will compare our Markov-based estimates with those derived from the data-driven method similar to Wide et al., 2024 23 .

Work Package 2: Estimating Stage-Specific Costs of LC Care

This work package will estimate direct medical costs associated with LC care, stratified by disease stage (IA–IV) and treatment phase.

Costing perspective and Data sources

The cost analysis will adopt the healthcare payer perspective (Health Service Executive, HSE) and will focus on direct medical costs incurred within the public healthcare system. All costs will be adjusted to 2023 Euro values using the Irish consumer price index (CPI) for healthcare to ensure comparability with current economic conditions. Where necessary, missing cost components will be estimated using published unit cost data and validated through Delphi survey.

Data will be sourced from:

•   National Cancer Registry Ireland (NCRI): Provides data on incidence, stage distribution, survival, and treatment patterns.

•   Healthcare Pricing Office (HPO): Supplies cost estimates for hospital admissions, outpatient care, and procedures, mapped to Diagnosis-Related Groups (DRGs).

•   Hospital In-Patient Enquiry (HIPE) database: Provides hospital resource utilisation data, including length of stay and procedure frequencies.

•   Pharmaceutical Reimbursement Service (PCRS): Provides cost data for systemic therapies, including chemotherapy, immunotherapy, and targeted treatments.

Ireland’s HSE pricing office does not produce unit costs for individual outpatient procedures like LDCT or biopsy. Instead, costs are bundled into DRGs under the AR-DRG system. We will use average DRG-based costs from the HIPE system to estimate diagnostic costs, supplemented by external sources (e.g. estimates available from National Screening Service Ireland) and tested in sensitivity analysis.

Costing methodology

A discrete event simulation (DES) model will be used to estimate the stage-specific costs of LC care in Ireland 24, 25 . The model will simulate a cohort of LC patients, tracking their diagnosis, treatment pathways, and healthcare utilisation over time.

Costs will be assigned based on the stage at diagnosis (I–IV) and categorised according to the phase of care. The initial treatment phase will include diagnostic investigations, surgical procedures, systemic therapies, and radiotherapy. High-cost drugs utilisation will be modelled based on NCRI data on treatment patterns by cancer stage, allowing us to apply costs only to eligible subgroups. Where Irish data are unavailable, we will use published sources such as cBioPortal or similar studies (i.e. Ngo et al., 2023) 26 . Changes in treatment patterns in future immunotherapy use will be explored through scenario and sensitivity analyses.

The surveillance and follow-up phase will incorporate costs associated with routine imaging, outpatient consultations, and disease monitoring. For patients with advanced or relapsed disease, the model will account for the costs of additional treatment lines, palliative care, and supportive interventions.

A 4% discount rate will be applied in line with HIQA guidance.

Work Package 3: Cost-Effectiveness Analysis (CEA) of LCS

The CEA will compare LDCT screening to no screening, evaluating different screening strategies based on frequency (annual, biennial, one-time) and eligibility criteria (age, smoking history, risk thresholds).

Model structure and inputs

The CEA will use a DES model, adapted from an existing UK-based LC natural history model 27, 28 . Key model components include:

  • Demographic inputs: Irish population structure (Census 2022, CSO projections).

  • Epidemiological parameters: LC incidence, mortality, and stage distribution (NCRI).

  • Screening effectiveness: modelled through stage shift, whereby screening leads to earlier detection and a more favourable stage distribution at diagnosis. Stage-specific survival estimates will be based on NCRI data. As NCRI does not include mode of detection or performance status, survival differences between screen-detected and clinically detected cancers will be informed by published literature or trial data (e.g. NLST, UKLS).

  • Healthcare costs: Stage-specific treatment costs (from Work Package 2).

  • Health outcomes: Life-years gained (LYG) and quality-adjusted life-years (QALYs), derived using Irish EQ-5D-5L utility weights (Hobbins et al., 2018).

  • Screening-related harms: False positives, overdiagnosis, and procedure-related complications, drawn from published meta-analyses 29, 30 .

The model will simulate a cohort of high-risk individuals undergoing screening, tracking LC incidence, stage shifts due to earlier detection, and treatment costs. To reflect real-world differences in screen-detected and clinically detected cancers, we will potentially account for the variation in performance status (PS). While NCRI does not currently capture PS at diagnosis, UK-based evidence (e.g. UKLS, TLHC) shows that screen-detected patients are more likely to be PS 0–1 and eligible for curative treatments 31 . As such, we will apply survival modifiers by detection mode, calibrated from published UK data, to approximate the survival advantage associated with better PS in the screening cohort.

The stage distribution for screen-detected lung cancers will be based on the NLST trial, which was also used to calibrate the ENaBL model adopted by the UK National Screening Committee. While more recent English data (e.g. TLHC programme) suggest a higher proportion of Stage I diagnoses (around 62%), the NLST-based distribution reports approximately 52% at Stage I. We will retain the NLST-based distribution to ensure alignment with the UK model structure. However, we acknowledge that this may underestimate the early-stage detection benefit of LDCT screening. This limitation will be discussed, and where possible, we will examine alternative stage distributions in scenario or sensitivity analyses.

Screening strategies will be compared based on frequency (annual, biennial, one-time) and eligibility criteria (age cut-offs, smoking history, risk stratification).

Economic analysis and uncertainty assessment

The cost-effectiveness analysis will adopt a scenario-based design, reflecting the substantial variation in cost-effectiveness estimates observed internationally. We acknowledge that previous models (e.g. Snowsill 2018) and recent evaluations (e.g. UK NSC 2022) have produced widely differing results. Rather than seeking to provide a binary verdict on whether LDCT screening is cost-effective, our primary aim is to test Ireland-specific assumptions and quantify the sensitivity of results to key parameters (e.g. eligibility thresholds, participation rates, treatment costs).

The economic evaluation will estimate the incremental cost-effectiveness ratios (ICERs) for each LCS strategy, comparing them against a no-screening baseline. These ICERs will be assessed in relation to Ireland’s cost-effectiveness threshold of €45,000 per quality-adjusted life-year (QALY), as defined by the Health Information and Quality Authority (HIQA) 32 . The analysis will adhere to CHEERS, ensuring transparency and methodological consistency 18 . We apply a 4% discount rate for both costs and outcomes, in line with HIQA guidance.

To assess uncertainty in the model outputs, both probabilistic and deterministic sensitivity analyses will be conducted. A probabilistic sensitivity analysis (PSA) will incorporate uncertainty in key model parameters, using Monte Carlo simulations to generate cost-effectiveness acceptability curves (CEACs) (ref). These curves will illustrate the probability that different screening strategies remain cost-effective at varying willingness-to-pay thresholds. Simulations will assess parameter uncertainty, assuming gamma distributions for costs, beta distributions for probabilities, and log-normal distributions for relative risks. In addition, a deterministic one-way sensitivity analysis (DSA) will explore the impact of individual parameter variations, such as screening uptake rates, treatment costs, and survival estimates. This approach will provide insights into how specific assumptions influence cost-effectiveness results, strengthening the robustness of policy recommendations.

Model validation

To support the credibility of the model, we will conduct external validation where data availability permits. This will include comparing the modelled stage distribution of screen-detected versus clinically detected cancers to benchmarks from NLST, NELSON, and UK programmes such as UKLS and TLHC. Detection rates, eligibility proportions, and uptake levels will be cross-checked against trial data and Irish screening pilot data, if available. Survival outcomes by stage and mode of detection will be validated against NCRI and international sources. Mortality reductions and overdiagnosis rates will also be compared to published ranges to ensure outputs are consistent with established evidence.

Data sources and data handling

This study will only use secondary, anonymised data, with no collection of individual-level patient data. All datasets are publicly available or accessed under institutional agreements.

•   NCRI: Aggregated LC incidence, stage, survival data.

•   CSO & Eurobarometer: Demographic and smoking prevalence data.

•   HPO, HIPE, and PCRS: Healthcare cost and resource utilisation data.

Discussion

This research programme addresses a critical evidence gap in the economic evaluation of LCS in Ireland. By integrating screening eligibility modelling, stage-specific cost analysis, and cost-effectiveness evaluation, it provides a comprehensive assessment of the economic and financial implications of LDCT screening. This is the first study of its kind in Ireland, generating policy-relevant evidence to inform decisions on resource allocation, programme feasibility, and long-term cancer control strategies. The findings will support national healthcare planning and align with the strategic objectives of the National Cancer Control Programme (NCCP).

Policy implications

The results of this study will provide empirical evidence to guide policymakers in evaluating the feasibility of a national LCS programme. By estimating the size of the high-risk population, the programme will inform financial planning, workforce requirements, and infrastructure investments needed for sustainable implementation. The stage-specific cost analysis will demonstrate the financial impact of late-stage LC care, reinforcing the economic argument for early detection. Furthermore, the cost-effectiveness model will provide a structured framework for assessing alternative screening strategies, allowing policymakers to balance clinical benefits with economic feasibility.

Beyond lung cancer, this research establishes a methodological foundation for future health economic evaluations in Ireland, particularly in oncology and chronic disease screening. The findings may contribute to broader discussions on risk-based screening policies, resource allocation, and long-term healthcare sustainability. Additionally, as real-world data on screening implementation and outcomes become available, the model can be refined and expanded to guide policy adjustments over time

Strengths and limitations

A key strength of this research programme lies in its comprehensive methodological approach, integrating diverse datasets from the National Cancer Registry Ireland (NCRI), the Hospital In-Patient Enquiry (HIPE) system, the Pharmaceutical Reimbursement Service (PCRS), and the Central Statistics Office (CSO). By leveraging dynamic Markov modelling, discrete event simulation (DES), and cost-effectiveness analysis (CEA), the study ensures robust, policy-relevant estimates. The use of advanced modelling techniques in R enhances the accuracy and transparency of cost-effectiveness calculations, while adherence to international reporting standards, including CHEERS (Consolidated Health Economic Evaluation Reporting Standards), ensures comparability with global research.

Despite these strengths, some limitations must be acknowledged. The screening eligibility analysis relies on Eurobarometer 2017 data for smoking history, as neither the 2022 Census nor the Healthy Ireland survey provides detailed pack-year estimates. While Eurobarometer offers the most comprehensive available data, more recent or Ireland-specific datasets would improve precision.

Another limitation is the availability and granularity of healthcare cost data. While administrative datasets such as HIPE and PCRS provide cost estimates, they lack detailed information on outpatient care, diagnostics, and treatment-specific costs. In the absence of a comprehensive national cost database, this study will estimate certain costs using Diagnosis-Related Groups (DRGs) and published unit costs, which may introduce some uncertainty into cost-effectiveness estimates. Validation through expert consultation, a Delphi survey and sensitivity analyses will help mitigate this issue.

Finally, as with all model-based economic evaluations, assumptions regarding screening uptake, treatment pathways, and long-term health outcomes are necessary. While probabilistic and deterministic sensitivity analyses will be conducted to explore uncertainty, the real-world effectiveness of LCS in Ireland will depend on actual programme implementation, participation rates, and adherence to follow-up care. Future studies should incorporate local pilot screening data to refine these projections.

Dissemination and knowledge translation

The findings of this research programme will be widely disseminated to maximise impact. Peer-reviewed publications will be produced for each work package, ensuring methodological transparency and academic contribution. Results will also be presented at national and international conferences, facilitating engagement with researchers, clinicians, policymakers, and public health experts.

To support real-world policy translation, an evidence summary will be developed and shared with healthcare stakeholders, including the Department of Health, the NCCP, and the HSE. Where appropriate, findings will be adapted for professional and public engagement through policy briefs, stakeholder workshops, and targeted knowledge-sharing initiatives. In addition, opportunities to communicate key insights through social media and professional networks will be explored to enhance visibility and accessibility of the research.

Conclusion

This research programme represents a substantial contribution to health economic research in Ireland, providing the first comprehensive economic evaluation of LCS in the country. By integrating modelling of screening eligibility, stage-specific cost analysis, and cost-effectiveness evaluation, it addresses both immediate policy questions and long-term healthcare planning considerations.

Ethic and consent

This study will be based exclusively on secondary data analysis. All datasets used are either publicly available in anonymised form or were accessed under institutional agreements in compliance with GDPR and data governance requirements.

  • Eurobarometer 87.1 (2017): This dataset contains de-identified, individual-level data and is publicly available through the GESIS data archive. Informed consent was obtained from participants at the time of data collection by the European Commission, and ethical oversight was provided by the data collectors in accordance with EU regulations.

  • National Cancer Registry Ireland (NCRI), Healthcare Pricing Office (HPO), Hospital In-Patient Enquiry (HIPE), and Pharmaceutical Reimbursement Service (PCRS) data were accessed in aggregated or fully anonymised form under institutional agreements. These datasets do not contain identifiable personal information, and no direct contact with participants occurred.

  • CSO Census and Population Projections are fully anonymised and publicly available.

As no new data were collected and all data were anonymised or publicly available, additional ethical approval or participant consent was not required for this study.

Funding Statement

Health Research Board [SDAP-2023-033].

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

[version 2; peer review: 1 approved, 2 approved with reservations]

Data availability

Underlying data

No data are associated with this article.

AI use disclosure

The authors confirm that they have read and agree to comply with the F1000 AI Policy. Generative AI tools, specifically OpenAI’s ChatGPT (version GPT-4, July 2025), were used to support the preparation of this manuscript. The tool was used to assist with drafting, editing, and refining text, and improving clarity in accordance with author intentions. All scientific content, data interpretation, and critical analysis were conducted and verified by the authors.

References

  • 1. O’Brien K, Walsh P, Deady S, et al. : Cancer trends 27 - lung cancer.National Cancer Registry Ireland;2015, [cited 2025 Apr 2]. Reference Source
  • 2. Cancer incidence projections for Ireland 2020–2045.National Cancer Registry;2019, [cited 2025 Apr 25]. Reference Source
  • 3. Kennedy MP, Hall PS, Callister ME: Factors affecting hospital costs in lung cancer patients in the United Kingdom. Lung Cancer. 2016;97:8–14. 10.1016/j.lungcan.2016.04.009 [DOI] [PubMed] [Google Scholar]
  • 4. National Lung Screening Trial Research Team, Aberle DR, Berg CD, et al. : The National Lung Screening Trial: overview and study design. Radiology. 2011;258(1):243–53. 10.1148/radiol.10091808 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Zhao YR, Xie X, de Koning HJ, et al. : NELSON Lung Cancer Screening study. Cancer Imaging. 2011;11 Spec No A(1A):S79–84. 10.1102/1470-7330.2011.9020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Schaft N, Dörrie J, Schuler G, et al. : The future of affordable cancer immunotherapy. Front Immunol. 2023;14: 1248867. 10.3389/fimmu.2023.1248867 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Tran G, Zafar SY: Financial toxicity and implications for cancer care in the era of molecular and immune therapies. Ann Transl Med. 2018;6(9):166. 10.21037/atm.2018.03.28 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. National Cancer Registry: Estimates of the health service costs of cancers associated with smoking, overweight and obesity, and alcohol intake in Ireland during 2016. 2020. Reference Source
  • 9. Council of the European Union: Council recommendation on strengthening prevention through early detection: a new EU approach on cancer screening replacing council recommendation 2003/878/EC. J Eur Union. 2022;100:1–10. Reference Source [Google Scholar]
  • 10. National Lung Screening Trial Research Team, Aberle DR, Adams AM, et al. : Reduced lung-cancer mortality with low-dose Computed Tomographic screening. N Engl J Med. 2011;365(5):395–409. 10.1056/NEJMoa1102873 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. de Koning HJ, van der Aalst CM, de Jong PA, et al. : Reduced lung-cancer mortality with volume CT screening in a randomized trial. N Engl J Med. 2020;382(6):503–13. 10.1056/NEJMoa1911793 [DOI] [PubMed] [Google Scholar]
  • 12. Sun C, Zhang X, Guo S, et al. : Determining cost-effectiveness of Lung Cancer Screening in urban Chinese populations using a state-transition Markov model. BMJ Open. 2021;11(7): e046742. 10.1136/bmjopen-2020-046742 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Ten Berge H, Togka K, Pan X, et al. : Cost–effectiveness of Lung Cancer Screening with volume computed tomography in Portugal. J Comp Eff Res. 2024;13(11): e240102. 10.57264/cer-2024-0102 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Behar Harpaz S, Weber MF, Wade S, et al. : Updated cost-effectiveness analysis of Lung Cancer Screening for Australia, capturing differences in the health economic impact of NELSON and NLST outcomes. Br J Cancer. 2023;128(1):91–101. 10.1038/s41416-022-02026-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Rózsa P, Kerpel-Fronius A, Murányi MP, et al. : Economic evaluation of Low-Dose Computed Tomography for Lung Cancer Screening among high-risk individuals – evidence from Hungary based on the HUNCHEST-II study. BMC Health Serv Res. 2024;24(1): 1537. 10.1186/s12913-024-11828-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Grover H, King W, Bhattarai N, et al. : Systematic review of the cost-effectiveness of screening for lung cancer with Low Dose Computed Tomography. Lung Cancer. 2022;170:20–33. 10.1016/j.lungcan.2022.05.005 [DOI] [PubMed] [Google Scholar]
  • 17. European Cancer Inequalities Registry: Country cancer profile 2023: Ireland.2023. Reference Source
  • 18. Husereau D, Drummond M, Petrou S, et al. : Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement. Int J Technol Assess Health Care. 2013;29(2):117–22. 10.1017/S0266462313000160 [DOI] [PubMed] [Google Scholar]
  • 19. Briggs AH, Weinstein MC, Fenwick EA, et al. : Model parameter estimation and uncertainty: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force--6. Value Health. 2012;15(6):835–42. 10.1016/j.jval.2012.04.014 [DOI] [PubMed] [Google Scholar]
  • 20. Census of Population 2022, Central Statistics Office (CSO): CSO open data platform, profile 4 - disability, health and carers, F4042 – population.2023. Reference Source
  • 21. Population Projections, Central Statistics Office (CSO): Population projections/current population and labour force projections (2022 Based)/PEC19 - projected population based on census 2022.2023. Reference Source
  • 22. Special Eurobarometer 458: attitudes of Europeans towards tobacco and electronic cigarettes. European Commission,2017. Reference Source
  • 23. Wade S, Ngo P, He Y, et al. : Estimates of the eligible population for Australia’s targeted national lung cancer screening program, 2025–2030. Public Health Res Pract. 2024;35(1): PU24004. 10.1071/PU24004 [DOI] [PubMed] [Google Scholar]
  • 24. Crealey GE, Hackett C, Harkin K, et al. : Melanoma-related costs by disease stage and phase of management in Ireland. J Public Health (Oxf). 2023;45(3):714–722. 10.1093/pubmed/fdac154 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Buja A, Rivera M, De Polo A, et al. : Estimated direct costs of Non-Small Cell Lung Cancer by stage at diagnosis and disease management phase: a whole-disease model. Thorac Cancer. 2021;12(1):13–20. 10.1111/1759-7714.13616 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Ngo P, Karikios D, Goldsbury D, et al. : Development and validation of txSim: a model of advanced lung cancer treatment in Australia. Pharmacoeconomics. 2023;41(11):1525–1537. 10.1007/s40273-023-01291-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Snowsill T, Yang H, Griffin E, et al. : Low-Dose Computed Tomography for Lung Cancer Screening in high-risk populations: a systematic review and economic evaluation. Health Technol Assess. 2018;22(69):1–276. 10.3310/hta22690 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. UK National Screening Committee: Final report on the cost-effectiveness of Low-Dose Computed Tomography (LDCT) screening for lung cancer in high-risk individuals.Public Health England,2022. Reference Source
  • 29. Gareen IF, Duan F, Greco EM, et al. : Impact of Lung Cancer Screening results on participant Health–Related Quality of Life and state anxiety in the National Lung Screening Trial. Cancer. 2014;120(21):3401–9. 10.1002/cncr.28833 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Fabbro M, Hahn K, Novaes O, et al. : Cost-Effectiveness Analyses of Lung Cancer Screening using Low-Dose Computed Tomography: a systematic review assessing strategy comparison and risk stratification. Pharmacoecon Open. 2022;6(6):773–786. 10.1007/s41669-022-00346-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Schabath MB, Massion PP, Thompson ZJ, et al. : Differences in patient outcomes of prevalence, interval, and Screen-Detected Lung Cancers in the CT arm of the National Lung Screening Trial. PLoS One. 2016;11(8): e0159880. 10.1371/journal.pone.0159880 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32. Health Information and Quality Authority (HIQA): Guidelines for the economic evaluation of health technologies in Ireland.HIQA,2020. Reference Source
HRB Open Res. 2025 Aug 30. doi: 10.21956/hrbopenres.15636.r48522

Reviewer response for version 2

David Baldwin 1

I am happy with the responses to the reviewers

Is the study design appropriate for the research question?

Yes

Is the rationale for, and objectives of, the study clearly described?

Yes

Are sufficient details of the methods provided to allow replication by others?

Partly

Are the datasets clearly presented in a useable and accessible format?

Partly

Reviewer Expertise:

Lung cancer screening

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

HRB Open Res. 2025 Jul 10. doi: 10.21956/hrbopenres.15525.r47388

Reviewer response for version 1

Geertruida H de Bock 1, Danrong Zhong 1

This is a nice manuscript with a very interesting analysis. Lung cancer screening reduced lung cancer mortality by shifting lung cancer stage distribution. Cost-effectiveness modeling studies can provide critical evidence to inform policymakers and accelerate the implementation of screening programs. I hope these comments contribute to strengthening the manuscript.

1. What can we learn from this study compared to what already have been published? Is the Irish context different from other countries, like UK? How this is different?

Work package 1: eligible population for screening

1. It’s written that “To ensure accuracy, the Markov model will be calibrated using historical smoking prevalence trends and validated against published Irish smoking data.” This is not clear. What is the difference between the calibration and the validation data set?

2.. Clarity pack-year threshold

The analysis focuses on individuals aged 55-74 with a smoking history of 15 or 20 pack-years: which threshold was applied (15 or 20)? For former smokers, is there a cessation time threshold (e.g. quit <10 years or <15years)?

3.. Reference for LCS recommendation

The statement ‘the risk group typically recommended for LCS’ , can you added reference on that? Is based on which guidelines? There are several guidelines as well as lung cancer prediction models. Which one is selected and why?

Work package 2: cost estimation

1. Data source

The listed sources (National Cancer Registry Ireland, Healthcare Pricing Office, etc.) do not include: costs for LDCT scan, follow-up diagnostics (e.g. biopsies for suspicious nodules).

Prices are very political sensitive. Especially for the expensive treatments (immunotherapy) prices are based on negotiation (often per hospital). Where the expensive drugs will be the main driver in the cost-effectiveness. How the authors will deal with this? How to deal with expected changes in prescription behavior? (e.g. in future immunotherapy will be also prescribed for early stage LC, there will be new immunotherapy drugs in the near future etc). And for immunotherapy, not all of the patients were eligible for this expensive treatment, how to estimate the percentage of immunotherapy eligibles in different stages?

2. Discount rate

What discount rate was applied (e.g. 3% or 5%)?

Rationale for the chosen discount rate.

Work package 3: Cost-effectiveness analysis of LCS

1. Mortality reduction estimate

‘Screening effectiveness: mortality reduction estimates from NLST (20%) and NELSON (24%)’, which screening effectiveness are really use in this study?

The effective of screening is critically dependent on:

(1) High-risk population definitions: NLST (aged 55-75 ever smokers, more than 30 pack-years and currently smoke or quit less than 15 years), for NELSON (aged 50-75, smoke more than 15 cigarettes per day for more than 25 years, or 10 cigarettes per day for more than 30 years, currently smoke or quit less than 10 years). These differing eligibility criteria may lead to variations in observed mortality reduction. Eligibility of these 2 trials are different from the definition in this protocol (aged 55-74, with pack year 15-20), is it appropriate to apply mortality reduction estimates from NLST/NELSON to high-risk population in this study (aged 55-74, with pack year 15-20)?

(2) Screening protocol: in NLST, is 3 rounds of the annual screening, for NELSON is 4 rounds of screening (baseline, regular screening after previous round screening with 1-year interval, 2-year interval, and 2.5 year interval). For this protocol, compared screening (annal, biennial and one-time strategies) to no screening, is it appropriate to apply mortality reduction from NLST/NELSON?

(3) For NELSON, 24% mortality reduction is only report on male, for female, is 33% reduction, will this analysis use 24% for both gender or account for the higher female reduction?

2. Screening strategies

The CEA compares LDCT screening to no screening, evaluating annal, biennial and one-time strategies, rationale for selecting these screening strategies.

3. I miss an impact analysis in WP3. What will be the impact of the implementation of LC screening for the budget, for the personal and how this relates to stop-smoking programs? The Budget impact analysis should specify its time horizon, is a 3-year, 5-year or 10-year timeframe more appropriate for LC screening?

Model validation

Different models ( DES: discrete event simulation model; dynamic Markov-based population model) were used in this protocol, will these models undergo validation? If so:

1. which independent datasets or published studies will be used for validation?

2. validation metrics: which outcomes will be compared to assess model accuracy? For example:

Stage distribution (screen-detected vs. clinical detected cancers)

Tumor detection rates

Size distribution of detected tumors…

Is the rationale for, and objective of, the study clearly described?

Yes

Is the study design appropriate for the research question?

Yes

Are sufficient details of the methods provided to allow replication by other?

Partly

Are the datasets clearly presented in a useable and accessible format?

Partly

Competing interests: No competing interests were disclosed.

Reviewer expertise: lung cancer screening, cost-effectiveness evaluation.

Is the study design appropriate for the research question?

Yes

Is the rationale for, and objectives of, the study clearly described?

Yes

Are sufficient details of the methods provided to allow replication by others?

Partly

Are the datasets clearly presented in a useable and accessible format?

Partly

Reviewer Expertise:

epidemiology, screening, lung cancer

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above.

HRB Open Res. 2025 Jul 17.
Tatiana Bezdenezhnykh 1

General comment: 

We thank the reviewer for their positive feedback and thoughtful suggestions. We appreciate the recognition of the study’s potential to inform national policy and accelerate implementation of lung cancer screening in Ireland. The reviewer has helpfully identified several areas where greater clarity, detail, and justification are needed. We have responded to each point in detail below and will revise the manuscript to incorporate the necessary clarifications and additions.

Work Package 1: Eligible Population for Screening

1. Calibration and Validation 

Calibration refers to adjusting model parameters (e.g. smoking transitions) so that simulated outputs align with historical Irish smoking trends and external estimates. For example, we will calibrate smoking transition rates from Eurobarometer against Health Ireland survey smoking prevalence. Validation refers to comparing model outputs, such as future prevalence trends and eligibility estimates against external empirical findings (from Ireland and other jurisdictions: UK, France, Australia)

2. Pack-year thresholds and cessation 

The primary analysis will focus on individuals aged 55–74 years with a smoking history of 20 or more pack-years who have quit within the past 15 years, reflecting current Irish eligibility criteria. We will also compare this group to those eligible under other widely used international guidelines, including NELSON, USPSTF 2013, and USPSTF 2021, which define risk groups typically recommended for lung cancer screening.

3. Reference for risk group selection 

The definition aligns with USPSTF 2021, NELSON, and NCCP guidance. We will revise the protocol to cite these guidelines and clarify why we selected this risk group: it balances comparability with major trials while reflecting Irish population risk profiles.

Work Package 2 – Cost Estimation

We appreciate the reviewer’s comments and agree that estimating accurate costs, especially for diagnostics and high-cost therapies, is a key challenge, particularly in the Irish context. The points below are clarified in the updated protocol.

1. LDCT scans and follow-up diagnostics: Ireland does not have unit costs for individual outpatient procedures like LDCT or biopsy. Instead, costs are bundled into DRGs under the AR-DRG system per case. We will use average DRG-based costs from the HIPE system to estimate diagnostic costs, supplemented by external sources (e.g. estimates available from National Screening Service Ireland) and tested in sensitivity analysis.

2. High-cost drugs and immunotherapy: We will base our estimates on NCRI data on treatment patterns by stage, alongside published drug prices. The proportion of patients eligible for immunotherapy at each stage will be estimated using NCRI treatment data and supported by published evidence on PD-L1 expression and clinical criteria.

3. Future changes and uncertainty: We recognise that treatment guidelines and prescribing patterns will evolve. To account for this, we will include scenario analyses to explore the impact of expanded indications and new therapies over time. 

4. Discount Rate

We apply a 4% discount rate for both costs and outcomes (in the cost-effectiveness analysis), in line with HIQA guidance.

Work Package 3 – Cost-Effectiveness Analysis

1. Mortality reduction estimates 

We agree that trial-based mortality reductions (e.g. NLST 20%, NELSON 24–33%) are specific to their designs and populations. We further clarify in the updated version of the protocol that we do not directly apply these reductions. Instead, screening benefit is modelled through a stage-shift approach, with survival based on stage and detection mode (e.g. screen-detected vs. clinical), consistent with the UK NSC’s ENaBL model.

2. Rationale for screening strategies 

We selected annual, biennial, and one-time screening to mirror strategies evaluated in NLST, NELSON, and UK pilot studies. These reflect realistic programme options for Ireland and enable trade-off analysis between frequency, benefit, and cost.

3. Budget impact and implementation 

We agree that this is an important and policy-relevant extension of the analysis. While our primary focus is on the economic evaluation, which lays the foundation for a BIA, we recognise the relevance of implementation considerations. As such, factors such as workforce capacity, scanner availability, and integration with smoking cessation services will be discussed outside of the scope of this project.

Model validation 

We agree that validation is an essential part of the modelling process. Our protocol involves three distinct models, which will undergo validation appropriate to their structure and intended purpose.

1. Validation datasets: 

For the costing model, which is based on NCRI data, we will validate internal consistency (e.g. cost by stage aligns with expected treatment pathways) and cross-reference cost estimates with external published costing studies (e.g. from UK and EU contexts).

For the CEA model, we will validate model outputs against both Irish and international data: 

  • NCRI data: LC incidence, stage distribution, survival by stage. 

  • Published trial data: NLST and NELSON outcomes (e.g. stage shift, detection rates). 

  • UKLS/TLHC data: where relevant, to validate outputs related to screening effectiveness. 

2. We will assess the external validity of model outputs using the following metrics, where data availability allows: 

  • Stage distribution of lung cancers: among screen-detected vs. symptomatically diagnosed cases. Compare how NCRI data (only clinical cases) aligns with NLST, NELSON, and UKLS data (e.g. % Stage I in NLST = 52%, in TLHC ≈ 62%). 

  • Detection rate: Number of cancers detected per 1,000 screened individuals; benchmark against published trial and pilot data (NLST, NELSON, TLHC).

  • Screening uptake and eligibility proportions: % of population eligible under different criteria; % of eligible individuals screened. Cross-validated with Irish data and UK pilot uptake figures. The Irish LC screening pilot is underway at its early stages. If the data is available early, it will be certainly used in validation. 

  • Mortality outcomes (indirectly): modelled lung cancer mortality reductions compared to NLST (20%), NELSON (24%–33%) under similar screening protocols (as a sense check for reasonableness). 

  • Survival outcomes by stage and detection mode: Check alignment with NCRI.

  • Overdiagnosis estimates: Compare % of cases estimated to be overdiagnosed in the model to published ranges (e.g. NLST estimate ≈ 18–25%).

HRB Open Res. 2025 Jun 3. doi: 10.21956/hrbopenres.15525.r47383

Reviewer response for version 1

Stephen Wade 1

The article describes a protocol for an economic evaluation of a lung cancer screening (LCS) in Ireland. The goals of the analysis are to obtain estimates of 1) the eligible population (as the screening is targeted), 2) the stage-specific costs of lung cancer care, and 3) the incremental cost-effectiveness of screening versus no-screening.

Overall, much of the protocol reflects existing practise; however there are such enormous differences in published estimates of 1) and 3) above that it brings in to question the reliability of "existing practise" and whether this is sufficient to support good decision making in Ireland (or other contexts).

With reservations, the protocol is sound. I have a few comments about methods.

Estimating the Eligible Population for Screening

Data-driven approaches are preferable to Markov models which have stricter assumptions about smoking behaviour and how that behaviour is going to be reported by participants. In Australia, for example, a model-based approach produced estimates that differed by a factor of two from data-driven methods (see "National Lung Cancer Screening Program – Updated participation modelling" https://www.health.gov.au/resources/publications/nlcsp-participation-modelling, and see https://doi.org/10.17061/phrp34342410 for the data-driven approach). This is a large discrepancy!

The budget impact is already subject to great uncertainty because of the difficulty in estimating participation rate; having a faulty denominator would be a terrible start.

Unless absolutely necessary, I think developing a model to determine who attends screening in 2057 does not seem like a useful - or even scientifically valid - exercise.

Estimating Stage-Specific Costs of LC Care

ICIs and targeted therapies are only administered when a specific subtype is detected (and often only reimbursed when the disease is at a specific stage). Ideally, then, some estimate of the prevalence of the relevant subtype (by stage) would be used to inform the cost. I did not find a reference to this in the protocol. The cBioportal has been used in the past to inform the costing of metastatic NSCLC - this would likely be a useful template for costing here (see Ngo et al https://doi.org/10.1007/s40273-023-01291-6).

Cost-Effectiveness Analysis (CEA) of LCS

I note that the model intended as a basis (Snowsill 2018) produced vastly different results from the more recent 2022 evaluation supplied to the UK National Screening Committee (see https://view-health-screening-recommendations.service.gov.uk/lung-cancer/). 

From a review published in 2022: "The range of cost estimates in the literature is broad, from US$1464 to $2 million per quality-adjusted life year (QALY) gained depending on the setting, modelling approach, and policy question." (see https://doi.org/10.1016/S0140-6736(22)01694-4). There is significant (catastrophic) uncertainty that renders these analyses of low value in answering the question of whether LCS is cost-effective compared to no screening. Even within jurisdiction, with shared data sources, a factor of three difference can be observed in estimate of benefit (see Figure 4 of "Evaluation of the Benefits and Harms of Lung Cancer Screening With Low-Dose Computed Tomography: Modeling Study for the US Preventive Services Task Force" https://doi.org/10.1001/jama.2021.1077).

The collection of evidence (all based on modelling) does not convince me that, say, a biennial program is more (or less) cost-effective than an annual program; yet this is still a frequently tested hypothesis. Similarly, there are long-standing questions about pack-year versus risk-model based criteria - frequently tested but no clear answer.

There is no data source or methodological leap presented in this protocol that convinces me that this study will prove different to the others. Perhaps a stronger focus should be placed here on sensitivity of the cost-effectiveness to parameters or Ireland-specific policy questions; with the acknowledgement that these models might not be able to answer "if" something is cost-effective, but they might identify which factors are most significant to future investigations (or pilots) and policy design.

Is the study design appropriate for the research question?

Partly

Is the rationale for, and objectives of, the study clearly described?

Yes

Are sufficient details of the methods provided to allow replication by others?

Partly

Are the datasets clearly presented in a useable and accessible format?

Not applicable

Reviewer Expertise:

bayesian statistics, health economics, cancer epidemiology, tobacco control

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

References

  • 1. : Estimates of the eligible population for Australia’s targeted National Lung Cancer Screening Program, 2025–2030. Public Health Research and Practice .2024;35(1) : 10.17061/phrp34342410 10.17061/phrp34342410 [DOI] [PubMed] [Google Scholar]
  • 2. : Development and Validation of txSim: A Model of Advanced Lung Cancer Treatment in Australia. PharmacoEconomics .2023;41(11) : 10.1007/s40273-023-01291-6 1525-1537 10.1007/s40273-023-01291-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. : Lung cancer screening. The Lancet .2023;401(10374) : 10.1016/S0140-6736(22)01694-4 390-408 10.1016/S0140-6736(22)01694-4 [DOI] [PubMed] [Google Scholar]
  • 4. : Evaluation of the Benefits and Harms of Lung Cancer Screening With Low-Dose Computed Tomography. JAMA .2021;325(10) : 10.1001/jama.2021.1077 10.1001/jama.2021.1077 [DOI] [PMC free article] [PubMed] [Google Scholar]
HRB Open Res. 2025 Jul 17.
Tatiana Bezdenezhnykh 1

General comment: 

We appreciate the reviewer’s insight into methodological uncertainty and agree that while modelling remains a useful tool, its limitations must be carefully acknowledged. Variation in model estimates comes about, in part, from different modelling assumptions. We believe that establishing a domestic model for Ireland is an important part of capacity building. While current practice certainly is imperfect, we feel it is important that we at least establish a baseline modelling capacity up which we can further develop and refine.

  1. Thank you for highlighting the discrepancy observed between model-based and data-driven approaches in the Australian context. The point is well made and it does give us pause for thought regarding the value of simulation vs empirical analysis. Our analysis is anchored on data-derived baseline eligibility estimates and reflects a plausible scenario for future smoking behaviours in Ireland. We believe that modelling is useful in understanding why anticipated and actual participation might differ. Moreover, while modelling might not accurately anticipate the level of actual participation, it hopefully can at least be useful in predicting trends in eligibility over time. Indeed, where eligibility determined by age and smoking history, such trends are arguably quite predictable. That said, significant uncertainties regarding participation will naturally persist. While we have chosen a Markov model to simulate mid-term trends in screening eligibility, we acknowledge the importance of grounding this in real-world data. To address potential discrepancies between modelling approaches, we will validate our Markov-based estimates with those derived from the data-driven method, as suggested by the reviewer. Finally, we agree that projecting as far as 2057 adds limited value given the uncertainty involved and is of modest current policy relevance. Accordingly, we have revised the projection horizon to 2045. 

  2. Thank you for this useful suggestion. We acknowledge the importance of this point, especially when therapies can differ so much in terms of their cost. We will include estimates of histologic subtypes and relevant biomarker prevalence (e.g. EGFR, ALK) by disease stage, informed by international datasets such as cBioPortal , if local registry data (NCRI) on assignment of targeted therapies is not enough or unavailable. We will also revise our costing methodology to weight targeted therapy uptake accordingly. 

  3. We acknowledge the point the reviewer makes regarding the heterogeneity of estimates and the questions this raises regarding the value of simulation evidence. We would make some observations in reply, some of which relate to replies above. One is that there are often good reasons for variations in simulation estimates, often down to fundamental differences in the way that models are constructed, but also to more easily observable differences in the way strategies are simulated and compared. For this reason, the variation in CEA estimates may be less than initially apparent. That said, it is undeniable that estimates vary, leaving decision makers with considerable uncertainty.  We do not believe that such uncertainty is reason to abandon CEA modelling. The process of modelling construction and testing allows the modeller to appreciate what are the important factors that influence CEA estimates, where additional evidence is needed and where models should be refined. Ireland currently does not have a domestic cancer screening modelling platform, applied to lung screening or otherwise. Establishing this as a capability is important, not just for value for money assessment, but also for budget impact analysis and capacity planning.  We do need to frankly admit the limitations of modelling and communicate the uncertainties to decision makers. We feel we would be in a better position to do so with a model than without one, despite the inherent flaws of such approaches.

HRB Open Res. 2025 May 13. doi: 10.21956/hrbopenres.15525.r46935

Reviewer response for version 1

David Baldwin 1

The protocol paper describes the plans for an economic review of lung cancer screening in Ireland. This is an important topic as lung cancer screening has the potential to save many lives.

My comments reflect the experience gained from the health economics analysis undertaken for the UK National Screening Committee.   I appreciate that some of the comments will present considerable difficulty but failure to address some aspects risk getting the wrong results - as indeed the paper you quote by Snowshill et al presented.  Note that the ICER changed some 18 fold once the correct input variables were included.  It is therefore important that the latest report, and not the 2018 paper is both read by the authors and quoted in their protocol.  It is available here please scroll down to the final report link (2022). https://view-health-screening-recommendations.service.gov.uk/lung-cancer/

1. Defining the high risk group and participation.  The authors intend to attempt to select on the basis of pack-years using Eurobarometer.  This is entirely dependent on the accuracy of those data so that needs to be confirmed.  I would suggest that it would be better to model both eligibility and participation using the English real-world data which is unlikely to be very different in Ireland. This should at least be checked again the Eurobarometer and a sensitivity analysis using the English data done.  This would be first to calculate the number of ever smokers in the eligible age range, then model a 50% response to the invite then approximately 50% of these being eligible and most of these having a CT. - 20-25% of the initial invite.  We are seeing improved participation as the programme rolls out but this would be a far better way to estimate unless Eurobarometer is super accurate.

2. The NCRI data is being used to measure stage distribution and survival - how accurate is this.  Does it contain performance status?  The latter is important if you are to model the effect of a screen because studies show that most people are PS 0-1 in screening in contrast to clinically detected.  This translates to better treatment outcomes which means survival is better even accounting for lead time effect.  If you are going to adjust for overdiagnosis then you should also include this, otherwise you underestimate the impact of screening.  Interval cancers should be modelled as clinically detected.  Incidence round cancers will have a greater proportion of early stage than prevalence. 

3. I was confused about the use of mortality derived from NLST and NELSON and then references to stage specific survival.  Which are you using?  The intention to control for over diagnosis  yet use mortality from the studies (which controls for  overdiagnosis, lead and length time implies two seperate analyses?

4. The intention to use LYG and discounts is fine but there will need to be adjustment for the very different fitness levels that we see between screen detected and clinically detected.  For the English analysis we were able to use survival by stage and performance status to show the impact in real world data of the screen - according to whatever stage was detected.  Studies both show the proportion in each stage and the proportion in each PS category. We also knew survival by stage in the screening cohort (UKLS).

5. what stage distribution are you using for screening detected lung cancer. NLST seems to have detected 10 % points fewer stage 1 than the English data ( 52 vs 62).  The UKNSC analysis was via a natural history model that modelled NLST data so likely underestimates the value of screening.

6. incidental findings are a key issue - in the English analysis we simply added a cost derived from the paper by Bartlett et al, but this does not add any benefit, which we now see (in the working programmes) is evident.

I hope these comments have been helpful

Is the study design appropriate for the research question?

Yes

Is the rationale for, and objectives of, the study clearly described?

Yes

Are sufficient details of the methods provided to allow replication by others?

Partly

Are the datasets clearly presented in a useable and accessible format?

Partly

Reviewer Expertise:

Lung cancer screening

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

HRB Open Res. 2025 Jul 17.
Tatiana Bezdenezhnykh 1

General Comment: 

We thank the reviewer for their thoughtful and detailed feedback, drawing on extensive experience with the UK National Screening Committee’s work. The comments raised are insightful and will help refine the protocol to ensure greater methodological robustness and relevance to real-world policy contexts. We note the reviewer’s comments about the importance of using the most up to date iteration of the Snowsill et al work, thus, we will adjust the methodology to align it with 2022 NCS report.  

  1. We appreciate this recommendation. We will use Eurobarometer data only to inform smoking histories to estimate the proportion of eligible people. Eurobarometer data is closely aligned with Irish Census 2022 smoking prevalence and is the only source known to us that reports smoking length and the average number of cigarettes for the Irish quitters. We will further apply UK real-world data to estimate the potential uptake. (e.g., 50% response rate to invitation and 50% of responders meeting criteria). To support the base case methodology, we will provide estimates using the approach suggested above. 

  2. Thank you for this suggestion. The initial protocol indeed did not give consideration as to how health status would vary between those with screen or clinically detected cancers and what implications this might have for survival. We see that performance status (PS) could be one useful way for differentiating health status and prognosis between such individuals. PS is unfortunately not currently captured in the NCRI dataset. We are likely to incorporate such effects by assuming survival probabilities that differ by the mode of presentation, if we can adequately support this with data from either Irish hospital data or screening cohorts such as NLST or UKLS. 

  3. We thank the reviewer for highlighting this important clarification. We acknowledge that using both trial-derived mortality reductions and stage-specific survival curves is not a plausible approach. We have updated the protocol to clarify this modelling approach and avoid confusion.  In line with the 2022 ENaBL model, our model will simulate disease progression through preclinical and clinical stages and rely on stage-specific survival estimates from NCRI and UK sources. The benefit of screening will be estimated via stage shift, i.e. the change in stage distribution due to screening, and heterogeneity in survival for clinically-detected and screen-detected cancers.   

  4. Thank you for raising this important point. We fully agree that performance status (PS) is a key determinant of survival and treatment eligibility, and that screen-detected patients are generally fitter than those diagnosed clinically. Unfortunately, performance status is not currently captured in the NCRI dataset, which limits our ability to directly model PS-specific survival using Irish data. To address this, we plan to follow the approach taken in other economic evaluations, including the updated analysis of the UK NSC’s ENaBL model. 

  5. Thank you for this helpful observation. As our model structure is based on the ENaBL model, which itself was calibrated using NLST data, we will retain these underlying assumptions for consistency. However, we recognise that this may underestimate the potential benefits of screening, particularly in terms of early-stage detection. We will explicitly acknowledge this limitation in the protocol and, where feasible, explore the impact of varying stage distributions in sensitivity analysis to assess the robustness of our results 

  6. We agree that incidental findings are a relevant consideration in the cost-effectiveness analysis. As in the English analysis, we plan to include an estimate of the cost of follow-up investigations for incidental findings (e.g., from Bartlett et al.), but recognise that doing so without capturing potential downstream benefits introduces an inconsistency. We acknowledge that a complete evaluation would ideally account for both the costs and potential health benefits of incidental findings. However, given current evidence limitations and the scope of this analysis, quantifying these benefits is beyond what we can reasonably include at this stage. We will clearly note this as a limitation in the protocol and highlight it as an area for future refinement as more real-world data emerge from screening programmes. 

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Availability Statement

    Underlying data

    No data are associated with this article.

    AI use disclosure

    The authors confirm that they have read and agree to comply with the F1000 AI Policy. Generative AI tools, specifically OpenAI’s ChatGPT (version GPT-4, July 2025), were used to support the preparation of this manuscript. The tool was used to assist with drafting, editing, and refining text, and improving clarity in accordance with author intentions. All scientific content, data interpretation, and critical analysis were conducted and verified by the authors.

    Articles from HRB Open Research are provided here courtesy of Health Research Board Ireland

    RESOURCES