Illustrative simulator of annual low-dose CT (LDCT) lung cancer screening adherence — how the gap between trial-setting and real-world repeat-screening rates shapes realized mortality benefit
Randomized lung cancer screening trials such as the National Lung Screening Trial (NLST) and NELSON demonstrated that annual low-dose CT (LDCT) screening reduces lung cancer mortality relative to no screening or chest X-ray. That benefit, however, was measured under trial conditions with structured recall, dedicated coordinators, and high protocol adherence. Community screening programs report meaningfully lower rates of patients returning for their next annual scan — a gap between demonstrated efficacy and realized real-world effectiveness that every screening program has to reckon with.
Randomized controlled trials measure efficacy under close to ideal conditions: participants are recruited into a study, followed by dedicated staff, contacted proactively for their next visit, and often more health-engaged than the general screening-eligible population to begin with. Real-world "effectiveness" reflects what happens once screening leaves that protected environment and is delivered inside routine clinical workflows — where scheduling, insurance coverage, reminder infrastructure, and patient competing priorities all intervene.
This efficacy–effectiveness gap is not unique to lung screening; it appears across cancer screening modalities. But because lung screening's benefit depends on repeated annual imaging (rather than a one-time test), incomplete adherence has an outsized effect on how much of the trial-demonstrated benefit a real-world population actually receives.
Illustrative figures cited in this simulator (trial adherence ~90%+, real-world repeat rates commonly well below that) are simplified, educational approximations meant to illustrate the shape and direction of the gap — not a specific clinical claim for any single program or population.
Programs and registries typically estimate annual adherence by tracking the proportion of screening-eligible patients who return for a subsequent LDCT within a defined window (e.g., 11–15 months) after a prior negative or low-risk result. Data sources include institutional screening registries, the ACR Lung Cancer Screening Registry (LCSR)-style program reporting, electronic health record recall logs, and insurance claims analyses.
Measurement is complicated by patients who move, switch insurance or health systems, are screened elsewhere without record linkage, or become ineligible (age, smoking status, comorbidity) between rounds — all of which can make true adherence hard to pin down precisely, and is one reason reported real-world rates vary so widely across studies and programs.
Because mortality benefit in the underlying trials was generated by a program of repeated annual scans, not a single scan, the benefit a real-world screening program delivers scales with how well it sustains that annual cadence across its enrolled population. A program with strong initial enrollment but poor return rates realizes only a fraction of the population-level benefit that trial data would suggest is achievable — even though every individual scan performed is clinically identical to a trial scan.
The central modeling idea in this simulator: the mortality benefit a program realizes is not simply "screening happened" — it is a function of how many patients stayed on an annual cadence long enough for early detection to matter. Adherence is the multiplier on trial efficacy, not a side detail.
Lung cancer screening works by catching a growing nodule earlier — often across successive annual scans — rather than by any single image alone. A patient who is screened once and never returns receives a meaningfully smaller share of the mortality benefit than a patient sustained on an annual program, because the whole premise of interval imaging is comparing this year's scan to last year's baseline to catch change while it is still early and resectable.
A single LDCT can and does find some cancers at an early stage. But screening trials were designed, and their mortality benefit was measured, around repeated annual rounds: a nodule too small or indeterminate to act on in year one may show a growth pattern by year two that meaningfully changes management. Patients who screen once and disengage forgo this comparative, longitudinal signal entirely — their single scan behaves more like an isolated diagnostic test than a screening program.
In this simulator, "cumulative program benefit realized" is modeled as an illustrative function of average adherence sustained across the years entered into the program — approaching the full trial-demonstrated benefit only when adherence stays high across many consecutive years, and degrading sharply when adherence drops early or often.
The mortality benefit of screening is generally attributed to "stage shift": finding a higher proportion of lung cancers at early, more curable stages (I/II) rather than at the locally advanced or metastatic stages that dominate diagnoses made only after symptoms appear. Early-stage lung cancer is amenable to surgical resection with substantially better 5-year survival than late-stage disease.
Because many early-stage nodules are asymptomatic and slow-growing, catching them depends on the screening program being in place, imaging-ready, and actually re-imaging the patient at the interval when that nodule becomes detectable and actionable — which is precisely the annual cadence that incomplete adherence disrupts.
This simulator uses an illustrative, simplified relationship between average sustained adherence and realized benefit: realized benefit rises faster than adherence at high adherence levels (near-complete annual programs capture close to the full trial benefit) and falls off more steeply at low adherence (a handful of scattered scans capture only a fraction of what a sustained program would). This is a pedagogical approximation, not a validated dose-response curve from a specific published model — its purpose is to make the underlying intuition ("adherence compounds over years") visible and interactive.
Two patients can each be "screened" and still receive very different real-world benefit: one screened once and lost to follow-up, one screened consistently for a decade. Program design should treat the second scan — and every scan after it — as no less important than the first.
Adherence drop-off after an initial screen is rarely one single cause. Scheduling difficulty, the absence of a proactive reminder system, transportation and access burden, and a sharp decline in perceived urgency after a reassuring first result all compound across successive years to pull patients off the annual cadence the program depends on.
Without a streamlined, low-effort path back into the imaging schedule, the burden of initiating a follow-up appointment falls on the patient — competing against work schedules, other medical appointments, and the simple fact that "next year" is easy to defer indefinitely. Programs relying on patients to call in and self-schedule their own annual follow-up consistently see lower return rates than programs with automated scheduling or standing orders.
When no one prompts the patient as their annual window approaches, an appointment that "felt important" at the time of the first scan can simply slip past unnoticed. This is compounded by care fragmentation — a patient screened during a hospitalization, an urgent-care visit, or by a since-departed primary care provider may have no consistent point of contact tracking when their next scan is due.
LDCT screening requires physical access to imaging equipment, which is unevenly distributed — rural patients, those without reliable transportation, and patients balancing caregiving or inflexible work schedules face a disproportionate burden simply getting to and from repeated annual appointments, even when they remain motivated to be screened.
A negative or low-risk first scan can paradoxically reduce a patient's felt urgency to return the following year — "I was checked and I was fine" is an intuitive but incomplete read of a screening program that only works because it keeps checking. This effect is well documented across screening modalities and is one of the more addressable targets for patient education at the point of a negative result.
Every one of these barriers is at least partly addressable by care-delivery design — not by asking more of the patient, but by reducing how much the program depends on unprompted patient initiative to sustain the annual cadence.
Structured recall and reminder infrastructure — the kind long used in mammography and colonoscopy screening programs — meaningfully improves the odds a patient returns for their next annual LDCT. Moving from no system, to basic reminders, to a coordinated multi-channel recall program is one of the most direct levers a screening program has over its own adherence rates.
A structured recall system tracks each enrolled patient's due date for their next annual scan and proactively reaches out — by mail, automated phone call, SMS/text, patient portal message, or a combination — well before that date, with an easy path to confirm or reschedule. The strongest versions combine automated reminders with a live patient navigator who can troubleshoot access barriers (transportation, insurance authorization, conflicting appointments) for patients who do not respond to automated outreach alone.
Critically, recall systems shift the operational default: instead of a patient needing to remember and act, the system needs the patient to simply respond to an outreach that is already scheduled and already accounts for their prior screening history.
No single communication channel reaches every patient reliably — phone numbers change, mail gets missed, portal messages go unread. Programs that layer multiple channels (e.g., an initial mailed letter, a follow-up automated call, and a text reminder closer to the due date) generally reach a larger share of their population than any one channel alone, and can escalate to live outreach for patients who remain unresponsive across channels.
Organized recall infrastructure is well established in other cancer screening programs — national mammography call-and-recall systems and colonoscopy/FIT reminder programs have both been associated with higher sustained participation than opportunistic, patient-initiated re-screening. Lung cancer screening programs adapting these same recall principles are, in effect, applying a proven adherence lever from a different screening context to a newer program.
This simulator models three illustrative recall strength levels — none, basic reminders, and structured multi-channel recall — each associated with a different assumed year-over-year adherence decline rate and a different baseline adherence boost. These are simplified, educational parameters intended to demonstrate the qualitative effect of recall investment on long-run adherence, not a specific published quantitative model for any real program.
The strongest lever a screening program controls is not persuading patients to care more — it is reducing how much the annual cadence depends on the patient remembering and initiating contact in the first place.
Individual patient reminders address one patient at a time. Program-level adherence tracking — aggregating return rates across the whole enrolled population, by year, by referring clinic, by demographic or access subgroup — lets a screening program see systemic gaps that no single patient-level reminder would ever reveal, and target fixes where they matter most.
A program that only reminds individual patients has no way of knowing, in aggregate, whether its overall adherence is improving or declining year over year, or whether particular subgroups — by referring clinic, insurance type, distance from the imaging site, or year of enrollment — are falling off the annual cadence at a much higher rate than others. Program-level tracking answers exactly that question by rolling up individual return data into a dashboard the program can act on.
Aggregated adherence tracking commonly surfaces patterns that are invisible at the individual level: a particular referring clinic with unusually low return rates (perhaps due to a workflow gap in how patients are handed off after their first scan), a cohort enrolled in a particular year that shows a sharper-than-average decline (perhaps coinciding with a staffing change in the recall team), or a demographic subgroup facing disproportionate access barriers that generic reminders do not address.
Registries modeled on structures like the American College of Radiology's Lung Cancer Screening Registry (LCSR) allow programs to benchmark their own adherence against broader program-level data, supporting quality-improvement cycles rather than relying on anecdote.
Program-level tracking is most valuable when it feeds back into recall system design rather than sitting in a static report: a subgroup identified as high-risk for drop-off can be prioritized for live navigator outreach rather than automated reminders alone; a clinic with a workflow gap can have that gap fixed directly; a cohort showing early decline can trigger earlier, more intensive re-engagement before the gap widens further.
Individual-level reminders keep single patients on track. Program-level tracking is what lets a screening program notice — and fix — the systemic patterns that individual reminders alone will never surface.