Screening for Lung Cancer with Computed Tomography: Systematic Reviews on Effectiveness and Patient Preferences

This systematic review synthesizes evidence from 85 studies to conclude that low-dose CT screening for lung cancer in adults aged 50–74 with significant smoking history reduces lung-cancer mortality but causes substantial harms like overdiagnosis and false positives, resulting in a net benefit that a majority of informed patients likely find acceptable.

The Gist

Imagine your lungs are like a vast, dark forest. Most of the time, the trees (your healthy cells) grow quietly and peacefully. But sometimes, a few trees start growing in a weird, dangerous way (cancer). The problem is that when these "bad trees" are small, you can't see them, and you don't feel them. By the time you notice the forest is on fire, it's often too late to put it out easily.

This paper is a massive, high-tech investigation into a new tool: Low-Dose CT Scans. Think of this tool as a super-powered, all-seeing drone that can fly over the forest and spot those tiny, weird trees long before they become a fire.

The researchers asked three big questions to see if sending this drone over everyone's forest is a good idea.

1. Does the Drone Save Lives? (The Good News)

The Analogy: Imagine you have a net to catch fish. If you cast the net, you catch the big, dangerous sharks (lung cancer) that would have eaten you. But you also catch a lot of small, harmless minnows and some seaweed.

The Findings:

• The Win: The study looked at data from over 94,000 people (mostly smokers or former smokers aged 50–74). They found that using the drone (CT scan) does save lives. For every 1,000 people screened, about 2 to 4 fewer people die from lung cancer compared to those who didn't get scanned. It's like saving a small group of people from a disaster every time you scan a crowd.
• The Catch: It's not a magic wand. The benefit is real, but it's modest. You have to scan people multiple times (3 or 4 times over a few years) to see this result.

2. What's the "Price" of Using the Drone? (The Bad News)

The Analogy: Because the drone is so sensitive, it sees everything. It spots a weird rock that looks like a tree, a bird that looks like a monster, and a cloud that looks like a storm. These are called False Positives.

The Findings:

• The False Alarms: This is the biggest downside. For every 1,000 people scanned, about 225 to 350 people will get a "False Alarm." The drone says, "Hey, there's something suspicious!" but when doctors investigate, it turns out to be nothing.
• The Stress: Getting a false alarm is like being told your house might be on fire. It causes anxiety, worry, and sleepless nights for a few months while you wait for the fire department (doctors) to check.
• The Unnecessary Procedures: About 10% of those with false alarms might need to undergo invasive tests (like a biopsy) just to be sure. It's like breaking a window to check if there's a burglar, only to find out there wasn't one.
• Overdiagnosis: Sometimes, the drone finds a "bad tree" that was actually growing so slowly it would never have hurt the person in their lifetime. We treat it anyway, which is like cutting down a tree that was going to die of old age next year. This happens to about 8 out of every 1,000 people screened.

3. Do People Actually Want This? (The Human Element)

The Analogy: Imagine you are offered a lottery ticket. The prize is saving your life, but the cost is a 25% chance of a false alarm and a small chance of unnecessary surgery. Would you buy the ticket?

The Findings:

• The Verdict: The researchers asked thousands of people, "If you knew the risks and the rewards, would you do it?"
• The Result: A small majority of people said, "Yes, I'd do it." They decided that the chance of saving a life was worth the stress of the false alarms.
• The Nuance: However, if the chance of saving a life is very low (because the person is at low risk), fewer people want the scan. It's all about the balance: Is the potential reward big enough to justify the hassle and worry?

The New Rules of the Game

The paper also looked at how we interpret the drone's images.

• Old Rules (NLST): The old way of looking at the images was very sensitive. It flagged almost everything, leading to lots of false alarms.
• New Rules (LungRADS): The new system is smarter. It's like teaching the drone to ignore the clouds and birds and only flag the really dangerous trees.
• The Result: Using the new rules cuts the number of false alarms in half! But we aren't 100% sure yet if this new system misses any dangerous trees that the old system would have caught.

The Bottom Line

Think of lung cancer screening like a high-stakes insurance policy.

• Who is it for? Mostly people aged 50–74 who have a heavy smoking history.
• Is it worth it? Yes, for most people in this group, the "insurance" (saving lives) is worth the "premium" (the stress of false alarms and extra tests).
• The Catch: It's not perfect. You will have false alarms, and some people will get treated for cancers they didn't need. But, on balance, the researchers say that for the right group of people, the benefits likely outweigh the harms.

In short: The drone works, but it's noisy. It saves lives, but it also scares a lot of people who are actually fine. If you fit the criteria, it's a conversation worth having with your doctor to decide if the trade-off is right for you.

Technical Summary

1. Problem Statement

Lung cancer remains the leading cause of cancer death in Canada and globally, with a majority of cases diagnosed at late stages due to asymptomatic early progression. While Low-Dose Computed Tomography (LDCT) screening has shown promise, previous guidelines (e.g., the 2016 Canadian Task Force on Preventive Health Care recommendations) were based on older data, primarily the National Lung Screening Trial (NLST) with short-term follow-up.

Key gaps necessitating this review included:

• Long-term Evidence: The need for data on benefits and harms after a decade of follow-up.
• Selection Criteria: Uncertainty regarding whether risk prediction models (e.g., PLCOM2012) are superior to trial-based criteria (age and pack-year history) for identifying eligible populations.
• Nodule Management: The impact of newer classification systems (e.g., LungRADS) compared to the sensitive criteria used in major trials (NLST) on false-positive rates and mortality.
• Patient Preferences: A lack of contemporary data on how informed patients value the trade-offs between mortality reduction, overdiagnosis, false positives, and psychosocial harms.

2. Methodology

The authors conducted three linked systematic reviews addressing three Key Questions (KQs), adhering to the Canadian Task Force on Preventive Health Care (CTFPHC) manual and GRADE methodology.

• Scope & Eligibility:
  • KQ1 (Benefits/Harms): Included Randomized Controlled Trials (RCTs) and observational studies on LDCT screening vs. no screening/chest X-ray. Focus on adults (18+), with specific attention to high-risk groups (50–74 years, 20–30 pack-year history).
  • KQ2 (Patient Preferences): Included studies measuring Health State Utility Values (HSUVs/disutilities) and preference-based data (trade-offs, intentions) from 2012 onwards.
  • KQ3 (Comparative Effects): Compared trial-based selection/nodule criteria against risk prediction models and LungRADS.
  • Inclusion Criteria: Studies from Very High Human Development Index countries, published in English or French.
• Search Strategy: Comprehensive searches in MEDLINE, Embase, Cochrane Central, Scopus, and EconLit up to July 2025 (with specific updates for different KQs).
• Data Synthesis:
  • Meta-analysis: Used random-effects models for pooling data where appropriate. Absolute effects were calculated using pooled relative risks and control event rates.
  • Decision Thresholds: A working group of clinicians established pre-specified thresholds for "small but important" effects (e.g., 1 fewer lung-cancer death per 1,000; 75 false positives per 1,000).
  • Certainty Assessment: GRADE framework was used to rate certainty (High, Moderate, Low, Very Low) based on risk of bias, inconsistency, indirectness, imprecision, and reporting bias.
• Outcomes:
  • Critical: Lung-cancer mortality, all-cause mortality, overdiagnosis.
  • Important: False positives (FPs), incidental findings, major complications, psychosocial harms, and Health-Related Quality of Life (HRQoL).

3. Key Contributions

• Updated Evidence Base: Synthesized data from 85 studies (N=640,537) including 13 RCTs, providing the most contemporary assessment of LDCT screening with long-term follow-up (10–12 years).
• Integration of Patient Preferences: Explicitly incorporated patient values and preferences into the evidence synthesis, moving beyond clinical efficacy to assess "net benefit" from a patient perspective.
• Decision Threshold Framework: Applied a novel approach where conclusions are based on whether effects meet pre-defined decision thresholds rather than solely on statistical significance or point estimates.
• Comparative Analysis of Protocols: Directly evaluated the transition from trial-based nodule management (NLST criteria) to LungRADS, quantifying the reduction in false positives.

4. Key Results

A. Screening Benefits and Harms (KQ1)

• Mortality Reduction:
  • Lung-Cancer Mortality: Moderate certainty evidence indicates LDCT screening (3–4 rounds) reduces lung-cancer mortality by 4.0 fewer deaths per 1,000 screened (95% CI: 1.2 to 6.4) over 10–12 years. This exceeds the decision threshold of 1.0.
  • All-Cause Mortality: Low certainty evidence suggests a reduction of 3.7 fewer deaths per 1,000 (95% CI: 8.5 fewer to 1.2 more). The upper confidence interval crossed the threshold, indicating imprecision.
  • Subgroups: Absolute benefits appear higher in women than men and in those with higher baseline risk. Benefits may be negligible for those at the lowest baseline risk (<1.5% 6-year risk).
• Overdiagnosis: Moderate certainty evidence shows an excess of 8.4 cases per 1,000 screened (95% CI: 1.3 to 16.3), exceeding the threshold of 2.5.
• False Positives (FPs): High certainty evidence indicates at least 75 people per 1,000 (likely 225) will experience at least one FP requiring recall or diagnostic follow-up.
• Incidental Findings: High certainty that at least 150 per 1,000 will have incidental findings; moderate certainty that at least 100 per 1,000 will have clinically significant/actionable incidental findings.
• Harms from Invasive Testing: Low certainty evidence suggests screening does not cause at least 2.5 major complications or 0.1 deaths per 1,000 among those without cancer.
• Psychosocial Harms: Undergoing a scan causes little-to-no harm. However, a positive result (especially requiring invasive follow-up) likely causes small-to-moderate psychosocial harm (4–8% change from baseline), which generally resolves within 6 months of diagnostic resolution.

B. Comparative Effects (KQ3)

• Nodule Management: Switching from NLST criteria to LungRADS (among NLST-eligible populations) probably reduces the false-positive rate by approximately 50% (from ~300 to ~136 per 1,000). However, the rate remains above the decision threshold of 75 per 1,000. The impact of this reduction on mortality benefits is currently unknown.
• Selection Criteria: Applying risk prediction models (e.g., PLCOM2012) to trial populations suggests absolute benefits are significantly higher for high-risk individuals and negligible for low-risk individuals, even if they meet age/pack-year criteria.

C. Patient Preferences (KQ2)

• Disutilities:
  • Positive screening tests and FPs generally have little-to-no disutility (<0.04).
  • Stage IIIB/IV cancer diagnosis and treatment carry small-to-moderate disutilities (0.07–0.09).
  • Early-stage (I-IIIA) cancer diagnosis/treatment has little-to-no disutility, supporting the concern that overdiagnosis may not cause significant long-term HRQoL loss, though the psychological burden exists.
• Trade-offs: Low certainty evidence suggests a small majority (51–75%) of informed people would accept 69–122 false positives and 1.3 cases of overdiagnosis to prevent one lung-cancer death.
• Net Benefit: As the "net benefit" scenario improves (higher mortality reduction, lower harms), the proportion of people preferring screening increases from ~25–50% (low benefit) to >75% (high benefit).

5. Significance and Conclusions

• Clinical Recommendation: Screening adults aged 50–74 years with a 20–30 pack-year smoking history using 3–4 rounds of annual LDCT offers a net benefit that a small majority of eligible people likely find acceptable. The benefits (mortality reduction) outweigh the harms for this specific demographic.
• Policy Implications:
  • LungRADS Adoption: While LungRADS reduces false positives, its impact on mortality is unproven. Programs should adopt it to reduce patient anxiety and resource use but must monitor for potential delays in cancer detection.
  • Risk-Based Selection: Eligibility criteria should consider individualized risk prediction models rather than rigid age/pack-year cutoffs to maximize benefit and minimize overdiagnosis in low-risk subgroups.
  • Equity Gaps: Evidence is lacking for racialized groups, Indigenous peoples, and those under 50 or over 74, despite potential equivalent risks.
• Future Research: There is a critical need for prospective studies comparing different screening intervals, nodule management systems, and selection criteria, as well as studies focusing on equity-deserving populations and the long-term management of incidental findings.

This review provides a robust, contemporary foundation for updating lung cancer screening guidelines, emphasizing a balance between mortality reduction and the management of overdiagnosis and false positives through shared decision-making.