Lung cancer is a leading cause of cancer-related death among both men and women. Screening with low-dose computed tomography (LDCT) can shift detection toward earlier stages of lung cancer, enabling more effective treatment. Evidence from randomized controlled trials has confirmed a statistically significant reduction in lung cancer mortality. The National Lung Screening Trial (NLST), a multicenter randomized study, compared LDCT screening with chest radiography. A 20% reduction in lung cancer mortality and a 6.7% reduction in overall mortality were observed in the LDCT group compared to controls [1]. The Dutch–Belgian Lung Cancer Screening Trial (NELSON) similarly reported significantly lower lung cancer mortality in the LDCT group compared to the no-screening group [2]. The U.S. Preventive Services Task Force (USPSTF) and the European Commission recommend lung cancer screening [3,4]. The USPSTF recommends annual LDCT screening for adults aged 50–80 years with a 20 pack-year smoking history who currently smoke or have quit within the past 15 years (Grade B recommendation) [3]. However, outside the setting of clinical trials, it remains unclear whether participation in LDCT screening is associated with improved all-cause survival in real-world populations and what factors may account for the observed benefits. Therefore, this study assesses the comorbidity profiles and 10-year overall survival of individuals undergoing LDCT screening compared with a carefully selected cohort of the general population and provides a comprehensive discussion of potential mechanisms, sources of bias, and data limitations.

Of the 43,686 individuals, 7281 (16.7%) were in the screening group and 36,405 (83.3%) in the control group. The groups were balanced by gender: 50.7% female and 49.3% male (Table 1). After matching, no statistically significant differences were observed for the 32 matched variables (Tables 2 and 3). The most common comorbidities in the study group were hypertension (52.2%), chronic coronary artery disease (20.5%), hypercholesterolemia (13.2%), diabetes (13.0%), COPD (11.4%), unstable coronary artery disease (5.6%), atrial fibrillation (2.8%), and heart failure (2.0%). Most participants (86.5%) lived in urban areas, 4.8% in urban–rural communities, and 8.6% in rural communities (Table 2). Table 3 presents additional county-level indicators used to balance potential social inequalities.

Multivariate analysis identified participation in LDCT screening (HR=0.653), hypertension (HR=0.882), hypercholesterolemia (HR=0.567), and female sex (HR=0.527) as positive predictors of survival. Negative predictors included lung cancer (HR=5.141), ischemic stroke (HR=2.451), colorectal cancer (HR=2.166), heart failure (HR=2.061), chronic kidney disease (HR=2.044), breast cancer (HR=1.837), COPD (HR=1.766), atherosclerosis (HR=1.503), and diabetes (HR=1.471) (Table 4). The 10-year overall survival was higher in the screening group by 5.6 percentage points (86.9%) compared with controls (81.3%, p<0.001). In sex-stratified analyses, overall survival was higher among females by 3.7 percentage points and among males by 7.5 percentage points in the screening group compared with their respective controls (Fig. 1).

Fig. 1.

(A) Survival plot (Kaplan–Meier estimator) for the analyzed population after matching. Sex-specific survival plots, B – females, C – males. Blue line – control group, yellow line – lung cancer screening group.

In this national, propensity score-matched observational study, participation in LDCT screening was associated with a 5.6-percentage point higher 10-year overall survival. Because the design of our study is observational, this association should not be interpreted as causal. Below, we outline potential mechanisms and sources of bias that may explain these findings.

A ‘healthy volunteer effect’ likely contributed to the observed survival advantage. Individuals who participate in preventive programs are often more health-conscious, adhere to treatment, and engage with healthcare services earlier, which can improve outcomes across conditions – not solely lung cancer [9,10].

A recent Cochrane meta-analysis including more than 90,000 participants demonstrated that although LDCT screening substantially reduces lung cancer-specific mortality (by approximately 21%), the associated reduction in all-cause mortality is considerably smaller (approximately 5%). The authors estimated that approximately 226 individuals need to be screened to prevent one lung cancer-related death. In addition, increased rates of false-positive results and an estimated overdiagnosis rate of approximately 18% were reported [11]. These findings indicate that improvements in overall survival are unlikely to be explained solely by cancer-specific mortality reduction and should be interpreted in the context of broader healthcare utilization and engagement. This aligns with prior evidence from randomized trials (e.g., NLST/NELSON) showing cancer-specific mortality benefits from LDCT, while our observational study suggests that participants may also derive broader survival advantages from health-promoting behaviors [1,2].

The apparently favorable associations observed for hypertension and hypercholesterolemia should be interpreted cautiously. These conditions, when diagnosed, often lead to regular medical follow-up and evidence-based therapies, which may reduce cardiovascular risk. As no data on treatment initiation, adherence, or effectiveness were available, these findings most likely reflect increased healthcare engagement rather than a direct protective biological effect. The lack of survival benefit in individuals with combined hypertension and hypercholesterolemia may reflect greater cumulative cardiovascular risk or unmeasured confounding.

An additional factor that prompts lung cancer screening program participants to begin screening for other lifestyle diseases and their subsequent prevention is the fact that the LDCT screening report also includes other incidental findings. Elements such as the coronary artery calcium score, aortic calcification, aneurysms, emphysema, fatty liver disease, osteoporosis, and any tumors and enlarged lymph nodes encourage participants to seek further diagnostic testing, which can lead to effective treatment and prevention of lifestyle diseases such as CVD, COPD, and cancer. We consider LDCT screening as a teachable moment in terms of smoking cessation intervention and LDCT result.

Despite the use of propensity score matched analyses with 32 variables, an unmeasured confounding is probable. Individual-level smoking intensity and duration, BMI, alcohol consumption, occupational exposures, and personal exposure to air pollution were unavailable in the NFZ data-base. Although we used tobacco-related comorbidities and county-level socioeconomic and healthcare proxies, residual confounding may remain. In particular, since all screened participants met heavy-smoking criteria (≥20 pack-years), the control group likely included a lower proportion of heavy smokers, which could bias the association toward the null; conversely, health engagement factors in the screened group could bias results toward benefit. The net direction and magnitude of residual confounding therefore cannot be precisely determined.

The socioeconomic factors, quality of care likely influence survival. We matched on county-level average wages and physician density to approximate access to care, but these aggregates do not capture individual-level income, education, or the quality of services received (public vs private sector, waiting times, diagnostic capacity). Given known socioeconomic gradients in mortality, differential access and quality may have contributed to the observed association.

The environmental exposures vary within counties and cities. Average air quality metrics at the county level may not reflect individual exposure heterogeneity, particularly in regions with marked intra-urban gradients or industrial activity. This limitation may again contribute to residual confounding.

The NFZ data-base are comprehensive but not without limitations. As a payer database, NFZ records may be subject to coding inaccuracies, upcoding, or variable reporting practices across providers. While such misclassification is unlikely to be systematically differential with respect to screening status after matching, both over- and underestimation of comorbidity prevalence are possible. Despite these constraints, NFZ remains the most complete nationwide source of healthcare utilization and mortality for Poland, enabling large-scale, long-term follow-up.

This study is observational and cannot establish causality. Unmeasured confounders (smoking intensity, BMI, alcohol, occupational/environmental exposures, individual socioeconomic status) were not available in NFZ data, potentially leading to residual confounding. County-level proxies for income and physician density do not capture individual-level access or care quality. Potential NFZ coding inaccuracies may affect comorbidity ascertainment. The absence of cause-of-death data precluded direct assessment of lung cancer-specific mortality. In addition, lack of detailed data on smoking intensity, duration, and cessation limits the ability to fully adjust for tobacco-related confounding. Finally, while propensity score matching on 32 variables reduced measurable imbalances, it cannot account for unmeasured factors, including health-seeking behavior.

Participation in LDCT lung cancer screening was associated with higher 10-year overall survival (difference of 5.6 percentage points) compared with matched controls in a national Polish cohort. Given the observational design and potential residual confounding, these results should be interpreted as an association, not proof of causation. Future prospective studies integrating individual-level behavioral, socioeconomic, and environmental data are needed to clarify mechanisms underlying the observed survival advantage.

RD, ES, TZ, JP, WR conceived the concept of the study. All authors were involved in writing manuscript. PK, BK and MK were involved in data collection. BK, MK, JP analyzed the data. All authors edited and approved the final version of the manuscript.

AI tools were used for language polishing, improving sentence structure, and correcting grammar. All intellectual content, research design, data analysis were conducted solely by the authors, who take full responsibility for the final manuscript.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors declare not to have any conflicts of interest that may be considered to influence directly or indirectly the content of the manuscript.