Impact of lung cancer screening eligibility criteria on participants' characteristics and healthcare resources.

María Olivia Cabrera: Conceptualization; data curation; methodology; formal analysis; writing – original draft; writing – review and editing. Kevin ten Haaf: Conceptualization; methodology; funding acquisition; formal analysis; writing – review and editing; supervision. Josep Maria Borràs: Conceptualization; methodology; project administration; funding acquisition; supervision; writing – review and editing. Rebeca Font: Conceptualization; formal analysis; methodology. Judit Solà: Conceptualization; formal analysis. Juul Hubert: Writing – review and editing. Mónica Ballesteros: Conceptualization; writing – review and editing. Josep Alfons Espinàs: Conceptualization; methodology; project administration; funding acquisition; supervision; writing – review and editing.

This research was funded by the European Union's Horizon 2020 Research and Innovation Program under grant agreement number 848294. Additional support was provided by the Departament de Recerca i Universitats de la Generalitat de Catalunya and AGAUR (grant number 2021 SGR 00808). We thank the CERCA Program/Generalitat de Catalunya for their institutional support.

Data were drawn from the Spanish National Health Survey (ENSE), a cross‐sectional survey on subjects ≥15 years old, conducted by the Ministry of Health. We analyzed data from the 2012 Survey, the most recent with complete information on current and former smokers, while excluding the 2017 survey due to missing essential data on former smokers, such as starting age and years since quitting. This survey constitutes a representative sample of the non‐institutionalized Spanish population. The total population sample corresponded to 21,007 individuals for 2012. Comprehensive details on the ENSE methodology can be found on the Spanish Ministry of Health's website: https://www.sanidad.gob.es/estadEstudios/estadisticas/bancoDatos.htm . The following sociodemographic and smoking‐related variables were obtained: age, sex, smoking status at the time of the interview, number of daily cigarettes consumed by current and former smokers, age at starting smoking and age at quitting smoking for former smokers, body mass index (BMI), personal history of cancer, and presence of chronic obstructive pulmonary disease (COPD). As the survey did not include information on family history of lung cancer, we assumed no additional risk from family history in the population for this analysis. Out of the 21,007 individuals, 20,984 had a smoking status record ( n  = 4762 current smokers, n  = 554 occasional smokers, n  = 4181 former smokers, and n  = 12,041 never smokers); missing values ( n  = 23) were excluded. We defined the following types of smokers for our study: current smokers (daily smokers), former smokers, and never smokers; occasional smokers were grouped under never smokers. For the number of daily cigarettes calculations, only participants who smoked cigarettes were considered, and the number of cigars and pipes was excluded. The variable education was recalculated into six categories, according to the International Standard Classification of Education 26 for use in PLCOm2012norace model (Tables  S1 and S3 , Supporting Information). For this study, four criteria were analyzed in the 2012 ENSE population; two criteria were based on accumulated smoking history (PY‐based criteria): 4‐IN‐THE‐LUNG‐RUN (4ITLR1) 21 and United States Preventive Services Task Force (USPSTF), 27 and the other two types were risk‐based criteria, using the PLCOm2012 no race model: 4‐IN‐THE‐LUNG‐RUN (4ITLR2) 21 and Targeted Lung Health Check (TLHC) 13 (Table  1 ). TLHC contemplates PLCOm2012 with the race component, for consistency; for this study the PLCOm2012 no race was used for 4ITLR2 and TLHC. Overview of the different lung cancer screening (LCS) strategies and their inclusion criteria. Note : 4IN‐THE‐LUNG‐RUN inclusion criteria can be either by smoking history or risk score. The PLCOm2012norace model was developed to predict the risk of lung cancer over 6 years. 16 , 28 It considers the variables: age, smoking history, personal history of cancer, family history of lung cancer, presence of chronic obstructive pulmonary disease (COPD), education level, and body mass index (details on PLCOm2012no race can be found in Data  S1 ). To estimate the number of eligible individuals for lung cancer screening, we first identified the population that met each criterion (Table  1 ) from the 2012 ENSE survey. We then calculated the percentage of individuals by sex from the 2012 survey (in 5‐year age groups) that met each criterion. For 4ITLR1 and 4ITLR2 (60–64, 65–69, 70–74, 75–79) for TLHC (55–59, 60–64, 65–69, 70–74) and for USPSTF (50–54, 55–59, 60–64, 65–69, 70–74, 75–80) (Table  S6 ). Thereby, we obtained the percentage of individuals that met each criterion, for the overall population of 2012. We then extrapolated these percentages to the total Spanish population aged ≥15 years in 2022, based on data from the National Institute of Statistics (INE), separated by the same age groups and sex, to provide a more recent estimate of individuals meeting each screening criterion (Table  S7 ). We assumed an overall 50% lung cancer screening participation rate, based on information reported by the Colorectal Cancer Screening Program Network in Spain 29 from the year 2019 (the last estimation that has been published). Based on these estimations, we calculated the number of CTs needed according to each criterion for annual and biennial screening. Finally, we identified the total number of annual LDCT scans performed in the public system in 2023 30 and the total number of CT scanners based on the Statistical Portal from the Spanish Ministry of Health's report 31 to calculate the percentage increase in LDCT utilization, according to each criterion and annual or biennial interval (Table  3 ). All variables were classified as continuous (ratio) or categorical (nominal or ordinal). All statistical analyses were conducted using IBM SPSS Statistics V. 27 and R Statistical Software (v4.4.2; R Core Team, 2024). Eligibility across different criteria was assessed for each individual with full information on smoking history.

Table  2 shows a description of the ENSE 2012 survey population and variables of interest for each eligibility criterion. USPSTF had the highest number of eligible individuals ( n  = 1809), followed by TLHC ( n  = 914), 4ITLR2 ( n  = 613), and 4ITLR1 ( n  = 506) (Table  2 ). Characteristics of the ENSE population and subpopulations meeting different lung cancer screening (LCS) criteria. Abbreviations: 4ITLR1 and 4ITLR2, 4 IN‐THE‐LUNG‐RUN trial; ENSE, Spanish National Health Survey; TLHC, targeted lung health check; USPSTF, U.S. Preventive Services Task Force. All criteria primarily selected men across all age groups (68.1–82.2% of eligible). The highest percentage of men was observed in 4ITLR2 (82.2%), while USPSTF had the highest percentage of women (31.9%) (Table  2 ). The median age varied greatly across criteria, with a median age of 59 years for USPSTF up to 69 years for the 4ITLR2 (Table  2 ). 4ITLR1, 4ITLR2, and TLHC primarily selected former smokers (54.3%, 50.3%, and 56.9%, respectively) whereas in USPSTF most of the participants (50.6%) were current smokers (Table  2 and Figure  S1 ). Regarding pack‐years (PY), among current smokers, the highest median was observed in 4ITLR1 (50 PY), followed by 4ITLR2 (46 PY), TLHC (43 PY), and USPSTF (37 PY). Former smokers had a higher median PY across all criteria; the highest median was observed in 4ITLR2 (66 PY), followed by 4ITLR1 and TLHC (60 PY each) and USPSTF (44 PY) (Table  2 ). In relation to cigarettes smoked per day (CPD), current smokers across all criteria had a median of 20 CPD, whereas former smokers from 4ITLR1, 4ITLR2, and TLHC had a median of 30 CPD, and USPSTF a median of 20 CPD. Former smokers across all criteria had higher CPD and PY than current smokers, despite having shorter smoking durations. When analyzing time since quitting smoking, among former smokers by age groups (60–79, 55–74, 50–80), the median was 15 years for all groups. However, when analyzed by screening criteria, we found that PY criteria chose a population with fewer median years since quitting smoking: 4ITLR1 (6 years) and USPSTF (8 years). In contrast, risk‐based criteria selected individuals with more years since quitting, 4ITLR2 and TLHC (10 years each) (Table  2 ). Men had the highest pack‐year median compared to women across all criteria; the highest was observed in 4ITLR2 (56PY), followed by 4ITLR1 (55.5PY), TLHC (51.2PY), and USPSTF (43.2PY). In contrast, women had the highest pack‐year median in 4ITLR1 (45.5 PY), followed by 4ITLR2 (43.5 PY), TLHC (40.2PY), and USPSTF (34.5 PY) (Table  S4 ). When analyzed by smoking status, former smoking men had a higher PY number compared to current smoking men across all criteria (51.0–69.5 vs. 39.8–51.0, respectively) (Table  S5 ). Risk prediction model criteria (4ITLR2 and TLHC) selected women with fewer pack‐years than 4ITLR1. USPSTF selected men and women with the lowest pack‐years, since it was the criterion that included a younger population with less tobacco exposure (Table  S4 ). Eligibility varied notably across age groups depending on the criteria applied (Figure  1 ). (A) Eligibility distribution by age and criteria. (B) Eligibility distribution by age and criteria in men. (C) Eligibility distribution by age and criteria in women. Across all criteria, most of the eligible population fell within the 60–64 (19.9–42.4%) and 65–69 age groups (14.3–30.4%). For 4ITLR1 the greatest proportion of eligible individuals was in the age group 60–64 (42.49%); for older age groups, this percentage began to decline. Most of the individuals eligible under 4ITLR2 and TLHC were in the 65–69 age group (29.4% and 29.8%, respectively), whereas USPSTF predominantly selected individuals aged 50–54 (26.2%) (Figure  1 ). When analyzed by sex, eligibility proportions within each criterion were heterogeneous (Figure  2 ). Under 4ITLR1, 30.4% of men and 12.0% of women were eligible in the 60–64 age group. Under the 4ITLR2 criterion, in the 65–69 age group, 22.3% of men and 7.0% of women were eligible. In the TLHC criterion, eligibility was highest for men in the 65–69 age group (23.0%), and in the 60–64 age group for women (7.7%). Under the USPSTF criterion, the highest eligibility for both sexes was in the 50–54‐year age group (men: 16.0%, women: 10.1%). Overall, eligibility in pack‐year‐based strategies (4ITLR1‐USPSTF) decreased with age, while eligibility in risk‐based strategies was stable across age groups (Figures  1 and S2 ). (A) Overall smoking status by criteria. (B) Overall smoking status by criteria in men. (C) Overall smoking status by criteria in women. In relation to the proportion of eligible participants within the population meeting each criterion, under 4ITLR1 criterion 16.9% of men and 3.4% of women were eligible (Table  3 ). For 4ITL2 criterion, 21.6% of men and 3.3% of women were eligible, and under TLHC 26.7% of men and 6.6% of women were eligible. The highest proportions were observed under USPSTF, with 30.7% of men and 11.2% of women eligible (Tables  3 and S6 ). Population eligible for lung cancer screening and additional CT requirements. Note : LCS interval was calculated with 50% of screening attendance. Percentages were calculated based on the total number of individuals who met the eligibility criteria (4ITLR1, 4ITLR2, TLHC, USPSTF), relative to the total population within each age group and criteria. Percentage increase relative to current CT activity in Spain and the number of additional CT scanners was based on Statistical Portal information from the Spanish Ministry of Health. Abbreviations: 4ITLR, 4‐IN‐THE‐LUNG‐RUN; CT, computed tomography; TLHC, targeted lung health check; USPSTF, U.S. Preventive Services Task Force. Globally, most criteria favored selecting former smokers (50.3–56.9%), except for the USPSTF criterion (49.4%) (Figure  2 ). Eligible men were more often former smokers compared to women (men: 53.2–61.5%, women: 33.4–47.3%), while eligible women were more often current smokers across all criteria (men: 38.5–46.8%, women: 52.7–66.5%). Overall, risk‐based strategies (4ITLR2‐TLHC) selected more female current smokers (64.2%–66.5%) than pack‐year‐based (4ITLR1‐USPSTF) (52.7%–58.9%) (Figure  2 ). For men, risk‐based strategies (4ITLR2‐TLHC) and pack‐year‐based (4ITLR1‐USPSTF) selected a similar percentage of former smokers (61.5%–55.6% vs. 56.3%–53.2%, respectively). The Venn diagram (Figure  3 ) illustrates the eligible population according to different lung cancer screening (LCS) criteria and the degree of overlap among them. A total of 330 individuals (17.5%) met all four LCS criteria. There was considerable variability among the criteria; USPSTF had the largest proportion of individuals that met that criterion exclusively (43.2%). In contrast, 4ITLR1 did not have any individuals meeting exclusively this criterion. Moreover, 4ITLR2 and TLHC included 2.6% and 3.9% of individuals exclusively meeting their respective criterion. The second‐largest overlap was observed between TLHC and USPSTF, with 15.9% of individuals meeting both criteria. These overlaps demonstrate how different lung cancer screening criteria include distinct yet sometimes overlapping populations. Venn diagram illustrating the overlap between participants meeting different lung cancer screening (LCS) inclusion criteria: 4ITLR1, 4ITLR2, TLHC, and USPSTF. In 2023, according to the Ministry of Health's Statistical Portal, Spain had 703 CT scanners, which performed 5,573,026 CT scans in the public system, equivalent to 7927 CT scans per machine. From 2020 to 2023 the number of CT scanners grew by 9.8%, and the number of CT scans performed grew by 30.0%. The 4ITLR1 criterion included 998,760 eligible individuals, the majority of whom were male former smokers. Implementing this strategy as an annual screening would thus represent a 9.0% increase in the national radiological activity and 63 extra CT scanners required; in a biennial strategy it would thus mean a 4.5% increase and 31 extra CT scanners; this was the most restrictive criterion analyzed. The 4ITLR 2 criterion included 1,188,790 estimated individuals. In an annual program, this strategy would represent a 10.7% increase in national radiological activity and 75 extra CT scanners required; in a biennial program, this would be a 5.3% increase and 37 extra CT scanners. The TLHC criterion broadened the target population to 1,875,926 individuals; in an annual program, this would represent a 16.8% increase in national radiological activity (937,963 LDCT in annual screening) and it would require 118 extra CT scanners; in a biennial program it would be an 8.4% increase and 59 extra CT scanners. The USPSTF criterion included 3,806,525 individuals; this was the broadest criterion and would represent an increase of 34.2% in national radiological activity (1,903,262 LDCT in annual screening) requiring 240 extra CT scanners; in a biennial program it would mean a 17.0% increase and 120 extra CT scanners (Tables  3 and S7 ).

The aim of this paper was to evaluate the impact of different inclusion criteria for lung cancer screening (LCS), focusing on the number and characteristics of individuals to be screened (age group, sex, and smoking behavior) and to assess the resources required for an LCS program in Spain. Our study showed smoking distribution was very dissimilar in men and women throughout the different criteria. Men were more often eligible compared to women across all criteria (Table  3 and Figure  S2 ). Most eligible women were current smokers from younger ages, while men were mostly former smokers from older ages. Our analysis showed that men had higher pack‐years than women across all criteria analyzed. These differences in smoking behavior can be explained because smoking prevalence in women increased during the period of the 1970s and 1980s in Spain, much later than in men. 32 According to this study, risk‐prediction strategies such as 4ITLR2 and TLHC chose women who had lower pack‐years (compared to 4ITLR1) but still presented a high risk of developing lung cancer (Table  S4 ). We found that former smokers had greater pack‐years than current smokers, despite having a shorter smoking duration. The median number of cigarettes across 4ITLR1, 4ITLR2, and TLHC was 30 cigarettes per day and 20 for USPSTF (Table  2 ). Former smokers reported a higher number of cigarettes per day than current smokers. Given that the number of cigarettes per day smoked decreases as a person ages, 33 this suggests that there might be a bias towards reporting the accurate number of cigarettes smoked per day. Due to the higher number of pack‐years in former smokers, they also have a higher representation in eligibility criteria based on pack‐years, such as 4ITLR1 and USPSTF. However, pack‐year criteria include time since smoking cessation as a component for eligibility. Consequently, as time since quitting smoking increases in former smokers, they will lose eligibility, leading to underrepresentation in older age groups in pack‐year‐based strategies. In contrast, risk‐based criteria incorporate smoking duration, cigarettes per day and time since smoking cessation as separate components of risk. Therefore, risk‐based criteria remain more stable in older age groups, allowing for a better representation of high‐risk individuals with diverse smoking behavior across both sexes. This is of great importance, given that lung cancer screening is most efficient when applied to high‐risk individuals. 34 Furthermore, it has been proven that risk models like PLCOm2012 are more efficient than USPSTF 2013 criteria for selecting individuals for lung cancer screening 35 and that with appropriate thresholds, risk prediction models have better sensitivity and specificity than the NLST eligibility criteria. 36 Additionally, a recent cost‐effectiveness analysis by Toumazis et al. supports risk‐based strategies as more cost‐effective than the 2021 USPSTF recommendations. 37 There have been ongoing discussions on which selection criteria to apply for lung cancer screening. The 2021 USPSTF criterion reduced the age and the number of pack‐years, to include more high‐risk African‐American individuals and reduce race disparities. 27 However, an analysis of the multiethnic cohort study (MEC) suggests that the use of the PLCOm2012 risk model can further reduce the racial disparities and improve screening performance across races compared to USPSTF2021. 38 This suggests that some criteria can be modified to improve screening eligibility across different cohorts. Self‐reported smoking history is subject to recall bias that has the potential to impact eligibility for LCS. 39 Our study indicates that risk‐based models could help reduce bias due to reported smoking behavior, as well as account for variations in smoking behavior across sexes. This is consistent with the CREST study, which found that the PLCOm2012 model was more sensitive overall and reduced sex disparities compared to USPSTF criteria in lung cancer screening. 40 Moreover, analysis from the NLST, NELSON, and LUSI trials suggest that lung cancer screening may be more beneficial for women compared to men, 4 , 41 , 42 possibly due to the rising incidence of adenocarcinoma in women. 43 We estimated the eligible population for LCS in Spain using different selection criteria and 50% participation (assuming adherence for colorectal screening). The most restrictive, 4ITLR1 (PY) identified nearly 1 million individuals, which resulted in a national radiological capacity increase of 9.0%, and 63 extra CT scanners required (in annual screening). In contrast, the least restrictive criterion, USPSTF2021 (also PY), included almost 3.9 million individuals, resulting in a national radiological capacity increase of 34.2% (almost four times the 4ITLR1 criterion), and an increase of 240 extra CT scanners (Tables  3 and S7 ). Estimations for the 4ITLR1 criterion are aligned with findings by Fu et al., who identified 1.04 million high‐risk individuals in Spain in 2014 using the NLST criteria combined with a PLCOm2012 risk of 2.0%. 44 A recent study defined a target population in Galicia of 249,099 individuals based on the criteria of 50–80 years old, current or former smokers who quit <15 years ago with more than 20 PY, and 53,931 as 65–80 years old, current or former smokers who quit <10 years ago with more than 30 PY. 45 While these studies provide valuable insights, one study compared different strategies to identify a high‐risk population for developing cancer in Spain, and the other estimated the eligible population for LCS in Galicia. In contrast, we provide detailed overviews of both healthcare resource requirements and characteristics of the individuals selected by these criteria. In particular, our study highlights the variation in pack‐years in former smokers and screening eligibility across different age groups for women. Therefore, we believe our findings offer a valuable starting point for developing inclusion criteria for lung cancer screening programs in countries that have sex differences in smoking patterns and provide essential information for health policy makers within our health system. This study has some limitations. We used data from the 2012 ENSE, as it was the latest available survey that included details on smoking history for former smokers. Given the high proportion of former smokers across all eligibility criteria, our study shows the importance of eliciting details on smoking history (such as accurate numbers of CPD, smoking duration, etc.) in both current and former smokers in health surveys. Additionally, we had to assume that none of the participants had a family history of lung cancer, as this information was not available from the surveys. Finally, variability in the productivity of CT scanners was not assessed in our estimation. This work provides a framework for further evaluations, such as health‐impact assessment and health‐economic evaluations for an LCS program, taking into account the benefits for high‐risk individuals as well as the health care resources needed. Also, considering flexibility for future adaptations, such as lowering the age ranges and/or threshold, in case of risk prediction approaches. Essential considerations for identifying a target population include addressing sex differences, smoking status characteristics and age groups that would benefit the most.

In conclusion, different LCS inclusion criteria directly impact required healthcare capacity and yield populations with highly varied characteristics. Consequently, balance across these characteristics (age, sex, smoking behavior) should be taken into consideration when choosing the selection criteria. Risk‐based criteria may provide more balanced eligibility across age groups.

Lung cancer was the most diagnosed cancer worldwide in 2022, accounting for 12.4% of all cancers globally and 10.8% in Spain. It was also responsible for over 18.7% of all cancer deaths worldwide and 20.2% in Spain. 1 , 2 Alongside smoking prevention and cessation, two clinical trials have shown that low‐dose CT (LDCT) screening effectively reduces lung cancer mortality. The U.S. National Lung Screening Trial (NLST) reported a 20% reduction in lung cancer mortality 3 and the Dutch–Belgian Lung Cancer Screening Trial (NELSON) showed a 26% reduction in lung cancer mortality in men after 10 years of follow‐up. 4 Currently, eight countries have implemented national LCS programs: three in Europe (Czech Republic, Poland, and Croatia), 5 , 6 , 7 one in the United States (U.S.), 8 one in Oceania (Australia), 9 and three in Asia (China, Taiwan, and South Korea). 10 , 11 , 12 Moreover, the United Kingdom (UK) has a targeted lung health check (TLHC) program solely implemented in some parts of the country. 13 These programs vary in eligibility criteria, since some select participants through dichotomized criteria based on pack‐years (PY) and others through risk‐based approaches. This second strategy uses validated risk prediction models to estimate the personal lung cancer risk 14 , 15 such as PLCOm2012, which can be used to select participants to benefit from LDCT screening. 16 As well as variations in smoking behavior, economic factors and key program components such as budget constraints and limited health‐care resources must be considered for local screening decisions. 17 , 18 Various programs have advised biennial screening rather than annual screening due to CT capacity restraints. 19 , 20 In Spain there are no formal LCS programs, but some pilot studies such as the 4‐IN‐THE‐LUNG‐RUN (4ITLR), 21 the International Lung Screening Trial (ILST), 22 and Cancer Screening, Smoking Cessation and Respiratory Assessment (CASSANDRA) 23 are ongoing. However, while in the United States and Northern Europe, the spread of tobacco smoking began before the 1950s, in Spain, men started smoking more widely in the late 1940s and women in the late 1960s. 24 These changes in smoking patterns in Spain are shown in the smoking prevalence increase among women aged 45–64, from 8.1% in 1989 to 23.9% in 2011, and a decrease for men from 43.8% to 30% over the same period. 25 Consequently, LCS programs in Spain will need to account for larger variations in eligibility between sexes compared to the United States and Northern Europe. Therefore, this study's objective was to analyze the impact of different eligibility criteria for lung cancer screening in Spain, considering different methodologies employed and the characteristics of the population.

María Olivia Cabrera, Josep Maria Borràs, and Josep Alfons Espinàs report funds from the European Union. Kevin ten Haaf reports grants from the European Union (Horizon 2020, 848294) during the conduct of the study. Outside the submitted work, he reports grants from Dutch Research Council/Netherlands Organization of Health Research (ZonMW Grant number 09150161910060) and grants from NIH/National Cancer Institute Grant (1U01CA199284‐01), grants from University of Zurich, Switzerland, grants from Cancer Research UK, grants from Cancer Australia, grants from Medical Services Advisory Committee of the Australian Ministry of Health, grants from Open Mind Call Convergence, speaking and consulting fees, and traveling reimbursements from Centre Hospitalier Universitaire Vaudois, Johnson&Johnson Deutsches Krebsforschungszentrum and SkinVision, paid to their institute. Additionally, he received traveling reimbursements from Rescue Lung Society and International Association for the Study of Lung Cancer, and personal prize from Heineken Prizes Foundations, all outside the submitted work. Juul Hubert reports funds from the European Union. The other authors have no conflict of interest.

Data S1. Supporting Information.