Lung Cancer Screening: What the Trials Show, and Who Is Being Reached

Two large randomised trials show that offering regular low-dose CT scans to people at high risk of lung cancer reduces deaths from the disease, by around a quarter among men in the European NELSON trial. England's national programme is finding most lung cancers at stage 1. But the benefit comes with false alarms and some overdiagnosis, and in the United States only about one in four eligible adults was up to date with screening in 2024.

Lung cancer is usually found late because it often causes no symptoms until it has spread. Screening tries to change that by looking for it in people who feel well, which is exactly why its trade-offs matter: everyone scanned is healthy until the scan says otherwise. For charities and campaigners the question is no longer whether lung screening can work. It is whether programmes reach the people most likely to benefit, and whether everyone invited gets an honest account of what the scan can and cannot do.

Lung cancer kills more people than any other cancer, which is why screening is worth testing

In 2024 there were an estimated 2.6 million new lung cancer diagnoses worldwide, 12.8% of all new cancers, and an estimated 1.9 million deaths from lung cancer, 19.1% of all cancer deaths, according to the GLOBOCAN estimates from the International Agency for Research on Cancer^[3]. No other cancer causes as many deaths. Our lung cancer evidence review covers incidence, survival, lung cancer in people who have never smoked and the stigma research; this piece looks only at screening.

The reason screening is attractive is stage. Lung cancer found while it is still confined to the lung has far better survival than lung cancer found after it has spread, and many lung cancers are diagnosed only after they have spread (see our piece on stage at diagnosis). The trouble is that earlier diagnosis on its own does not prove that lives are saved. A cancer found two years earlier will look as if it has longer survival even if the person dies at the same time (lead-time bias), and some cancers found by screening would never have caused harm. That is why the only convincing evidence is a randomised trial that counts deaths.

Two big trials found fewer lung cancer deaths with CT screening

The two trials that shaped policy are the US National Lung Screening Trial (NLST) and the Dutch-Belgian NELSON trial. A systematic review for the US Preventive Services Task Force assessed the evidence from seven randomised trials involving 86,486 people^[4]. In the NLST, three annual low-dose CT scans reduced lung cancer deaths compared with chest X-ray, with an incidence rate ratio of 0.85 (95% confidence interval 0.75 to 0.96). In NELSON, four rounds of CT reduced lung cancer deaths compared with no screening, with a rate ratio of 0.75 (0.61 to 0.90)^[4].

NELSON enrolled 13,195 men and 2,594 women aged 50–74, and offered scans at the start, after one year, after three years and after five and a half years, with at least ten years of follow-up to the end of 2015^[1]. Average adherence to screening among men was 90.0%. Lung cancer deaths were 2.50 per 1,000 person-years in the screened group and 3.30 in the unscreened group, a cumulative rate ratio at ten years of 0.76 (0.61 to 0.94). The smaller group of women showed a rate ratio of 0.67 at ten years, but the confidence interval (0.38 to 1.14) was wide enough to include no effect, so the trial was not large enough to be sure for women on its own^[1].

On the strength of this evidence the US Preventive Services Task Force in 2021 recommended annual low-dose CT for adults aged 50–80 with a 20 pack-year smoking history who currently smoke or quit within the past 15 years, concluding "with moderate certainty" that screening has "a moderate net benefit"^[5].

A 24% relative reduction means about one death prevented for every 130 people screened over ten years

Relative and absolute figures describe the same result in different ways; confusing them is how screening evidence gets overstated or dismissed.

• Relative reduction. NELSON's rate ratio of 0.76 means men offered screening had lung cancer death rates 24% lower than men not offered it^[1].
• Absolute difference. In the same trial the gap was 0.80 lung cancer deaths per 1,000 person-years (3.30 minus 2.50)^[1]. Put another way, the Task Force review calculated that 130 people needed to be screened in NELSON over ten years to prevent one lung cancer death, and 323 in the NLST over six and a half years^[4].

Both numbers matter. A 24% reduction in deaths from the most lethal cancer is a large public-health gain when applied to a whole high-risk population. For any one person, though, most people screened were never going to die of lung cancer in that period whether scanned or not, and a small minority benefit. The two figures are not directly comparable: the NLST compared CT with chest X-ray rather than with no screening, and the trials differed in follow-up length and screening schedule.

False positives are common; serious harm from follow-up is rare in trials but higher in routine care

A false positive is a scan that shows something that needs further checks but turns out not to be cancer. Small, mostly harmless lung nodules are common, and modern programmes manage many of them with repeat scans rather than biopsies. In NELSON, 9.2% of screened participants needed at least one extra scan for an indeterminate result, and 2.1% were referred for a suspicious nodule^[1]. The Task Force review found that in the NLST, for every 1,000 people screened, false-positive results led to 17 invasive procedures such as biopsies (a number needed to harm of 59), and fewer than one person per 1,000 had a major complication^[4].

Routine care can look different from a trial. A study of 9,266 people screened at five US health systems between 2014 and 2018 found that 31.9% had further imaging and 2.8% had an invasive procedure. Among those who had an invasive procedure after an abnormal result, 30.6% had any complication and 20.6% a major one, compared with 17.7% and 9.4% in the NLST^[6]. The gap is why quality assurance and agreed nodule-management protocols matter once screening leaves the trial setting.

Scans also turn up things that are not lung cancer. The Task Force review reported incidental findings in 4.4% to 40.7% of people screened across studies, and noted that radiation-induced cancers are a rare harm^[4]. In England's programme, among 114,430 people scanned in the early phase, 47.8% had coronary artery calcification, 12.1% moderate or severe emphysema and 0.46% another, non-lung cancer^[2]. Some of these findings are useful; others lead to tests and worry without benefit.

Overdiagnosis: some screen-found cancers would never have caused harm

Overdiagnosis is not a false positive. It is a real cancer, confirmed under the microscope, that would never have caused symptoms or death in the person's lifetime. Nobody can tell which individual cancer is overdiagnosed; it can only be estimated by counting how many extra cancers a screened group has compared with an unscreened group after long follow-up.

The best-known estimate comes from the NLST: after 6.4 years of observing 53,452 people, the authors estimated an 18.5% probability (95% CI 5.4% to 30.6%) that any lung cancer detected by CT screening was an overdiagnosis, rising to 78.9% for the slow-growing type then called bronchioloalveolar carcinoma^[7]. In NELSON, lung cancer was diagnosed at 5.58 per 1,000 person-years in men screened and 4.91 in men not screened after ten years^[1]; the persisting excess is consistent with some overdiagnosis, although part of it could close with longer follow-up. Across the trials reviewed for the Task Force, estimates ranged from a 0% to a 67% chance that a screen-detected lung cancer was overdiagnosed^[4]. The honest answer is that overdiagnosis happens, and its size depends on who is screened, how nodules are managed and how long people are followed.

Overdiagnosis is also why rising survival or a shift to stage 1 cannot, alone, show that a programme saves lives: harmless cancers found by screening make both look better. Our explainer on what survival statistics mean covers lead-time and overdiagnosis bias in more detail.

Programmes are scaling up, but reach is uneven

England offers the most detailed published picture of a national programme. It began in 2019 as the Targeted Lung Health Check, was renamed the NHS England Lung Cancer Screening Programme in February 2025, and is expected to reach full national coverage by 2030. People aged 55–74 who have ever smoked are invited to a lung health check, and those above a risk threshold are offered low-dose CT^[2]. A five-year evaluation published in Nature Medicine in March 2026 reported these figures to March 2025:

Source: Lee et al., Nature Medicine 2026^[2]. The authors report that recent uptake of the lung health check is over 60%.

Of the cancers found, 63.1% were stage 1 and 12.6% stage 2, so about three-quarters were at an early stage; 8.8% were stage 4^[2]. The authors report that the early-stage share of lung cancer in England rose over the five years, most notably in more deprived areas. Yet uptake of the lung health check in the early phase was 33.2% in the most deprived fifth of areas against 41.6% in the least deprived^[2]. The programme also reported limited data on false positives and recall, a gap that matters for judging harms at scale.

Elsewhere in the UK, a 2026 pilot in Scotland identified 2,801 people aged 50–74 as potentially eligible; 668 (23.8%) responded, 503 were found eligible, 436 of those had a scan, and lung cancer was diagnosed in 5 of the people scanned (1.15%). Response was lower in more deprived areas^[8].

In the US, one in four eligible adults is up to date, but only 6% of those without health coverage

The United States has recommended lung screening since 2013^[5], yet use remains low. In 2024, 24.5% of adults eligible under the 2021 criteria reported being up to date with screening, up 6.0 percentage points since 2022, according to national survey data analysed in JAMA Internal Medicine^[9]. The gap by health coverage is stark: 39.1% among people with military-related coverage, 30.6% with Medicare, 23.6% with Medicaid, 19.0% among those with private insurance and 6.1% among uninsured adults, with no statistically significant change for uninsured adults since 2022. Twenty-one states saw significant increases and none a significant fall^[9]. These are self-reported survey figures, which can overstate true screening.

Eligibility rules themselves can build in inequity. In the Southern Community Cohort Study, only 57.6% of Black people who went on to develop lung cancer would have qualified for screening under the 2021 US criteria, compared with 74.0% of White people. Replacing the pack-year rule with a criterion of 20 or more years of smoking raised these to 85.3% and 82.0%^[10]. Pack-years multiply cigarettes per day by years smoked, and the authors conclude that the measure is an inadequate and biased way to decide eligibility. England uses multivariable risk thresholds rather than pack-years alone^[2]; whether any eligibility model is fair also needs testing in each population where it is used.

Uptake of breast, bowel and cervical screening shows similar patterns by deprivation and coverage; see our piece on screening uptake worldwide.

What the evidence asks of us

The trial evidence is strong enough that the debate has moved from whether to screen to how to screen well. The figures point to five priorities for health systems, commissioners and funders:

• Reach the people at highest risk first. Absolute benefit is greatest where risk is greatest, yet uptake is lowest in deprived areas of England and Scotland and among uninsured adults in the US^[2][9][8]. Programmes should be designed and monitored to close that gap, for example through proactive invitation, scanning close to where people live and no cost at the point of care.
• Fix eligibility rules that miss people. Criteria based only on pack-years exclude a larger share of Black people who go on to develop lung cancer than White people^[10]. Eligibility should be tested for fairness in every population where it is used.
• Keep harms close to trial levels. Complication rates after invasive procedures in routine US care were higher than in the NLST^[6]. Nodule-management protocols, quality-assured reporting and audit of false positives, procedures and complications should be built in from the start, and published.
• Measure what matters. A shift to stage 1 is encouraging, but because of overdiagnosis and lead-time bias it is not proof of fewer deaths. Programmes should track lung cancer mortality, overdiagnosis and harms over time, and report them openly; England's evaluation notes that recall and false-positive data are still limited^[2].
• Explain the trade-offs honestly. People invited deserve information that gives both the relative and absolute benefit, the chance of a false alarm and the possibility of overdiagnosis, in plain language, so that taking part or not is a genuinely informed choice. Screening must never become a reason to stigmatise people who smoke or have smoked; the people it serves are those at highest risk, and services should be built around that.

Language: this piece says "people with cancer" and "people living with and after cancer", and describes the illness and its treatment plainly, without combat metaphors. Clinical terms such as "stage", "incidence" and "net survival" are used in the sense the cited source defines them.

Why we're publishing this

Lung cancer causes more cancer deaths than any other cancer, and people affected by it often face stigma as well as late diagnosis. Screening is one of the few tools shown in trials to reduce lung cancer deaths, so charities, support groups and campaigners need clear figures on what it achieves, what it costs people in false alarms and overdiagnosis, and who is being left out. This piece is part of our Cancer research library; the case it supports belongs to the people living with and after lung cancer, their families, and the organisations that stand with them.

How to read this data

The mortality figures come from randomised trials, the strongest design for judging screening, but they describe specific populations: people with a substantial smoking history, aged 50–74 in NELSON, in high-income countries. NELSON's main result is for men; it had too few women for a firm separate answer. Trial rates are per 1,000 person-years (one person followed for one year counts as one person-year), and rate ratios compare whole groups offered screening with groups not offered it. Numbers needed to screen depend on the risk of the group and the length of follow-up, so they cannot be moved from one population to another.

Programme figures from England are observational: they show how many people were invited, scanned and diagnosed, and at what stage, but not by themselves how many deaths were prevented. Stage shifts can partly reflect overdiagnosis. The US uptake figures are self-reported in a national telephone survey, and other studies using different definitions report different levels (for example 20.0% in a separate 2024 analysis of the same survey with a different approach)^[11]. Overdiagnosis estimates vary widely between studies because they depend on long follow-up and on how the comparison is made. Evidence from lower- and middle-income countries, and on screening people who have never smoked, is much thinner, and we have not covered it here.

Use this data

Free to cite with attribution to Better Health For Everyone. Charities, campaigners and journalists are welcome to reuse these figures and charts. For the underlying figures as a spreadsheet, get in touch.

Important: This article summarises published research for general information. It is not medical advice and cannot be used to diagnose, treat or predict anything about an individual. If you have concerns about your own or someone else's health, please speak to a qualified health professional.

Sources

1. de Koning HJ, et al. Reduced lung-cancer mortality with volume CT screening in a randomized trial. New England Journal of Medicine, 2020. europepmc.org/article/MED/31995683
2. Lee RW, et al. Implementation of the NHS England Lung Cancer Screening Programme over 5 years. Nature Medicine, 2026. europepmc.org/article/PMC/PMC13190264
3. Sung H, et al. Global cancer statistics 2024: GLOBOCAN estimates of incidence and mortality worldwide for 34 cancers in 186 countries. CA: A Cancer Journal for Clinicians, 2026. europepmc.org/article/MED/42417444
4. Jonas DE, et al. Screening for lung cancer with low-dose computed tomography: updated evidence report and systematic review for the US Preventive Services Task Force. JAMA, 2021. europepmc.org/article/MED/33687468
5. US Preventive Services Task Force (Krist AH, et al.). Screening for lung cancer: US Preventive Services Task Force recommendation statement. JAMA, 2021. europepmc.org/article/MED/33687470
6. Rendle KA, et al. Rates of downstream procedures and complications associated with lung cancer screening in routine clinical practice: a retrospective cohort study. Annals of Internal Medicine, 2024. europepmc.org/article/MED/38163370
7. Patz EF, et al. Overdiagnosis in low-dose computed tomography screening for lung cancer. JAMA Internal Medicine, 2014. europepmc.org/article/MED/24322569
8. Cavers D, et al. Pilot feasibility study to deliver low-dose CT lung cancer screening in Scotland. BMJ Open, 2026. europepmc.org/article/MED/42744380
9. Burus T, McAfee CR, Knight JR, Mullett TW, Hull PC. Lung cancer screening prevalence and changes in 2024. JAMA Internal Medicine, 2026. jamanetwork.com/journals/jamainternalmedicine/fullarticle/2848035
10. Potter AL, et al. Pack-year smoking history: an inadequate and biased measure to determine lung cancer screening eligibility. Journal of Clinical Oncology, 2024. europepmc.org/article/MED/38537159
11. Li S, et al. Who is more likely to receive up-to-date lung cancer screening? A weighted cross-sectional analysis of the 2024 BRFSS. Frontiers in Public Health, 2026. europepmc.org/article/MED/42395284