Theory 08/Public Health & Epidemiology /Screening theory/3 min
Screening Works Through Sojourn Time
A better test for the wrong cancer saves nobody. The clock matters more than the lens.
The benefit of a screening programme is set by the sojourn time of the detectable preclinical phase rather than by test sensitivity, so a highly sensitive test for a fast-growing cancer produces no mortality benefit, and programmes should be selected by sojourn time rather than by test performance.
01The evidence
- Simulation of randomised screening trials has been used to compare methods of estimating lead time and benefit, which makes the timing parameters the quantities that determine whether a programme helps.Computational Statistics and Data Analysis
- Lead-time bias correction in screening analysis shows that survival gains can be produced by earlier diagnosis alone, without any change in the time of death.Cancers
02The tension
Screening programmes are chosen and justified by test performance: sensitivity, specificity, and the stage shift they produce. Those metrics are easy to measure and they systematically overstate benefit, because earlier detection is counted as longer survival even when the date of death is unchanged.
03The theory
A screening test can only help if it detects a lesion during a window when treatment is more effective. That window is the sojourn time, the interval between detectability and clinical presentation.
If the sojourn time is short, the lesion is likely to present clinically before or soon after screening, and the programme cannot change the outcome. Sensitivity is irrelevant at that point because there is nothing to find in time.
The prediction is that mortality benefit should track sojourn time across cancers better than it tracks test sensitivity, and that programmes with high sensitivity but short sojourn should show stage shift without mortality benefit. That pattern is already visible in several programmes and is usually attributed to implementation.
The policy consequence is that candidate cancers should be ranked by sojourn time before test development, and that a screening programme should be justified by the window rather than by the assay.
04What this is not
Performance-led policy selects by sensitivity and specificity. This selects by sojourn time, which predicts null mortality benefit from excellent tests and explains stage shift without benefit.
05Predictions
- P1Mortality benefit should track sojourn time better than sensitivity across programmes.
- P2High-sensitivity programmes with short sojourn should show stage shift without mortality benefit.
- P3Ranking by sojourn time should predict trial results better than ranking by test accuracy.
06What would refute it
Benefit tracking sensitivity at controlled sojourn would leave test performance as the criterion.
07What it would change
It would change how cancers are selected for screening, and how trials are designed.
08Evidence base
- Computer simulation of screening trials: lead time and benefit time — CSDAhttps://www.sciencedirect.com/science/article/abs/pii/S0167947396000291
- Lead-time bias correction in cancer screening — Cancershttps://pmc.ncbi.nlm.nih.gov/articles/PMC11010975/pdf/cancers-16-01326.pdf