Surveillance & follow-up

Surveillance Imaging Yield — Number Needed to Scan

After curative-intent lung cancer treatment, how many surveillance CT scans are needed to detect one recurrence — and how many to detect one actionable event? This calculator expresses that arithmetic to support multidisciplinary discussion. It estimates detection yield, not survival benefit.

Detecting recurrence is not the same as living longer.

Earlier detection is subject to lead-time and length bias. Randomised evidence on more intensive CT-based follow-up after completely resected NSCLC (the IFCT-0302 trial, Westeel et al.) did not demonstrate an overall-survival benefit over minimal follow-up. Use this tool to understand the yield of surveillance and to support a shared, individualised discussion — not to assert that a given surveillance schedule prolongs life.

Calculator

Recurrence hazard is highest in the first 1–2 years, then falls.
Recurrence + second primary, in the early (≈ first 2 years) period.
Share of events found by a scheduled scan before symptoms.
Detected events leading to curative or life-prolonging treatment.

Results

Effective annual hazard
Peak risk × time-since-treatment factor.
Interval event risk
Probability of an event during one interval.
Number needed to scan — any event
Scans per recurrence or new cancer detected.
NNS — CT-detected event
Includes the CT detection fraction.
NNS — actionable event
The most defensible yield metric.
Yield: actionable events / 100 scans
The same number, the intuitive way round.
Enter values and calculate.

What this does — and doesn't — tell you

  • It is a yield calculator, not a benefit calculator. A low "number needed to scan" means recurrences are found efficiently — it does not mean that finding them earlier improves survival. That requires trial evidence, which for routine intensive CT follow-up is, at best, mixed (IFCT-0302).
  • "Number needed to scan" (NNS) is a detection-yield metric. It is deliberately not called NNT (number needed to treat), which is an effectiveness measure from randomised trials. Do not interpret NNS as a treatment effect.
  • The hazard is time-dependent. Recurrence risk peaks within the first two years and declines thereafter, so the tool scales the peak-phase risk by a months-since-treatment factor. The factor is an illustrative profile, not a validated curve.
  • The preset numbers are illustrative placeholders. Replace them with your own registry / cohort data. Risk varies by stage, nodal status, margins, histology, Ki-67 and prior therapy.
  • Interval cancers exist. The CT detection fraction (< 100%) accounts for events that present symptomatically between scans.

Formulae

QuantityFormula
Effective annual hazardh = peak_annual_risk × time_factor
Interval event riskp = 1 − exp(−h × interval_months / 12)
NNS, any event1 / p
NNS, CT-detected event1 / (p × CT_detection_fraction)
NNS, actionable event1 / (p × CT_detection_fraction × actionable_fraction)
Yield (actionable / 100 scans)100 × p × CT_detection_fraction × actionable_fraction

Time-since-treatment factor (illustrative)

PeriodFactor on peak hazardRationale
0–24 months1.0Early peak of the recurrence hazard.
25–48 months0.6Hazard declining after the early peak.
49–60 months0.35Lower late hazard; second primaries persist.
Over 60 months0.2Mostly second primary risk; consider stopping or spacing imaging.

Preset assumptions (illustrative — replace with local data)

GroupPeak-phase annual event riskNotes
Low-risk NSCLC4%Approximate stage IA-like, R0, no adverse features.
Intermediate-risk NSCLC10%Higher local stage or adverse pathological features.
High-risk NSCLC20%N+, stage III, close margin, post-CRT, or otherwise high relapse risk.
SCLC after response45%Very high early recurrence risk.
Typical carcinoid, R0 N01%Low annualised risk; late recurrence possible — long horizon.
Atypical or N+ carcinoid6%Varies by nodal status, Ki-67, necrosis and margins.
Key reference. Westeel V, et al. Results of the IFCT-0302 phase III trial assessing minimal versus CT-scan-based follow-up for completely resected non-small-cell lung cancer. Overall survival was not significantly improved by CT-based surveillance. Recurrence-hazard timing (early peak within ~2 years) is well described in resected NSCLC cohorts.

Clinical framing & assumptions: Nordic Thoracic Oncology Group (NTOG). Interactive engine: powered by Vahtian. Runs entirely in your browser; no data is stored or transmitted. This is an educational decision-support tool. It is not a medical device, does not recommend a surveillance schedule, and does not replace national guidelines, MDT review, or clinical judgement.