Lung cancer is the leading cause of cancer death in Croatia, and across Europe the disease is diagnosed late in the majority of patients, when curative treatment is no longer possible. The clinical case for screening with low-dose computed tomography had been established by two landmark trials: the National Lung Screening Trial demonstrated a 20 percent reduction in lung cancer mortality, and the Dutch-Belgian NELSON trial showed a 24 percent reduction overall and 33 percent among women.
Despite this evidence and a 2022 EU Beating Cancer Plan endorsing lung cancer screening, Croatia became the first EU member state to launch a government-funded national lung cancer screening program, beginning in October 2020. By August 2025, more than 50,000 participants had been screened and over 70,000 low-dose CT scans had been performed.
Croatia's decision to act before most of Europe was shaped by its high lung cancer burden, the availability of a national digital health infrastructure, and the practical ability to implement a GP-centered recruitment model across a relatively compact health system. The Croatian model was designed from the outset to be both clinically rigorous and operationally scalable.
The program incorporated lessons from multiple international protocols including I-ELCAP, ACR Lung-RADS, and the European Consortium recommendations, ultimately adapting I-ELCAP criteria for nodule management while setting its own eligibility thresholds calibrated to the Croatian population and healthcare context.
The Croatian program targets individuals aged 50 to 75 with a smoking history of at least 30 pack-years who are either current smokers or former smokers who quit within the last 15 years. These criteria were set to balance sensitivity for detecting lung cancers in the highest-risk population with feasibility given national CT scanning capacity and radiologist workforce constraints.
Croatia's general practitioner network forms the backbone of the recruitment model. Approximately 2,300 GP clinics each serve about 1,800 patients, and approximately 90 percent of the Croatian population visits their GP at least once per year, making general practice the most efficient point of contact for identifying and enrolling eligible participants.
GPs were given three core responsibilities within the program: identifying eligible candidates from their patient lists, conducting motivational interviews that integrate smoking cessation counseling with screening enrollment, and managing follow-up referrals for participants whose scans return positive or indeterminate results. This GP-centered design ensures that screening is embedded in the existing longitudinal patient-physician relationship rather than operating as a standalone parallel service.
The program is fully integrated into Croatia's national IT healthcare platform, enabling paperless referral, result delivery, and follow-up tracking across all participating sites. Participants are enrolled digitally, scans are read and reported electronically, and positive results trigger automated referrals to one of six specialized lung nodule clinics operating in five cities across the country.
All low-dose CT scans are performed at 24 licensed radiology sites in 17 cities across Croatia, with strict technical standards governing image acquisition and radiation dose. The mean radiation dose delivered across more than 70,000 scans was 0.89 millisieverts, well below the 1.5 mSv upper limit specified in the program protocol, confirming that the low-dose requirement is being met consistently in practice.
Nodule management follows modified I-ELCAP criteria adapted for the Croatian context. Each scan result is classified into one of three categories: negative, semi-positive or indeterminate, and positive. The classification is based on volumetric assessment of identified nodules rather than diameter alone, which provides more accurate size characterization and better reproducibility across different scanner models and reconstruction parameters.
Artificial intelligence tools are incorporated into the volumetric analysis workflow to support radiologist assessment of nodule size, shape, and growth over time. The AI-assisted approach is intended to reduce inter-reader variability and improve consistency of nodule characterization across all 24 radiology sites, particularly for borderline cases where volume thresholds determine whether a participant is referred for additional workup.
Radiologist licensing for participation in the program requires completion of dedicated training that includes didactic lectures, hands-on workshops, and independent reading of 30 representative cases covering the full range of nodule morphologies encountered in lung cancer screening. Only radiologists from high-volume centers who interpret more than 300 thorax CT scans per year are eligible for licensing, ensuring a minimum level of practice volume and clinical experience.
The program screened more than 50,000 participants between October 2020 and August 2025, with 54 percent male and 46 percent female participants and a mean age of 62 years. The sex distribution reflects Croatia's smoking epidemiology, where female uptake of tobacco has historically been high relative to other European countries, a demographic fact that influenced the eligibility design to ensure women were not systematically excluded by pack-year criteria alone.
The overall positive scan rate was 4.5 percent, meaning that approximately 1 in 22 participants received a result requiring active clinical follow-up at a lung nodule clinic. This rate is substantially lower than the 5.9 percent positive rate associated with ACR Lung-RADS criteria and higher than the 1.4 percent rate reported with standard I-ELCAP criteria, reflecting the intermediate threshold design adopted by the Croatian program.
The program's biennial follow-up protocol assigns negative-result participants to routine re-screening at two-year intervals. Participants with indeterminate findings undergo shorter-interval CT surveillance, while those with positive results are referred directly to one of the six lung nodule clinics for multidisciplinary evaluation including PET-CT, bronchoscopy, or surgical biopsy as clinically indicated.
Smoking cessation support is integrated into every screening contact. GPs conduct motivational counseling at enrollment, and participants who screen positive receive additional cessation intervention alongside their diagnostic workup. This integration reflects the dual public health rationale of the program: reducing lung cancer mortality through early detection while simultaneously reducing future tobacco-related disease burden through cessation.
The Croatian program occupies a distinct design space relative to other European programs, combining features of the English Targeted Lung Health Checks, the Dutch NELSON protocol, and I-ELCAP into a nationally unified system. Unlike England's program, which uses the PLCOm2012 risk prediction model to stratify eligibility within a broad age range, Croatia uses fixed smoking history thresholds that are simpler to implement consistently across thousands of GP clinics.
The 30 pack-year eligibility criterion is more stringent than the 20 pack-year threshold used in some European pilot programs, which reduces the total eligible population but concentrates resources on the highest-risk individuals where the mortality benefit per scan is greatest. This trade-off was made explicitly to align program capacity with available radiologist workforce and scanner infrastructure.
Croatia's AI-assisted volumetric analysis approach addresses a genuine operational challenge in multi-site programs: ensuring that nodule size thresholds are applied consistently across readers with different levels of experience and scanners with different reconstruction parameters. Volumetric measurement reduces the inter-reader variability that would arise from manual diameter measurement alone, particularly for nodules near the volume thresholds that determine follow-up intensity.
The GP-centered recruitment model has proven effective in achieving broad population coverage, leveraging the near-universal annual contact between Croatian citizens and their family physicians. This approach contrasts with invitation-based models in other countries where participation depends on participants responding to population-level letters, a design that tends to underperform in lower-income and lower-health-literacy populations.
Croatia's five-year experience demonstrates that a national lung cancer screening program can be successfully implemented in a medium-sized EU health system with existing digital infrastructure and a strong primary care network. The program achieved its core operational targets: consistent radiation dose below protocol limits, a positive rate within the expected range for a volume-based assessment system, and broad geographic coverage through 24 licensed radiology sites.
The GP-centered model is the most transferable structural lesson from the Croatian experience. Any country with a functioning primary care registration system and high annual GP contact rates can replicate this recruitment architecture, reducing dependence on population-level invitation mailings and improving uptake in underserved communities who are less likely to self-refer.
Radiologist licensing with center-volume requirements proved essential for quality assurance across a geographically distributed program. Countries planning multi-site implementations should build licensing and retraining mechanisms into program design from the outset rather than relying on informal professional competence claims.
Remaining challenges include long-term funding sustainability as the screened cohort grows and biennial rounds accumulate, integrating the full downstream diagnostic and treatment pathway for positive-result participants, and extending geographic access to nodule clinic services beyond the current five-city network. These operational priorities will shape the next phase of the Croatian program as it moves from launch into long-term maturation.