In 2022, the New England Journal of Medicine published the results of the NordICC trial — the first randomised controlled study of colonoscopy screening ever conducted. Over 84,000 people were followed for ten years. The trial found an 18% reduction in cancer incidence and no significant reduction in cancer deaths. To prevent a single case of colorectal cancer, 455 people had to be invited for screening. To prevent a single death, the numbers were statistically indistinguishable from zero.¹
This is the pattern.
Across the major screening programmes — mammography, PSA, Pap, colonoscopy, lung CT — when the question is whether the screened population actually outlives the unscreened population, the benefit largely disappears.² The statistic the programmes advertise is disease-specific mortality: deaths from the disease the test is looking for. The statistic they bury is all-cause mortality: whether the screened group, taken as a whole, lives longer. The two numbers are not the same. You can reduce deaths from one disease while total deaths remain flat — because treatment has killed as many people as the disease prevented, or because the disease you found was never going to kill anyone.²
The screened do not live longer than the unscreened. They are more likely to spend their remaining years monitored, biopsied, cut, and medicated for conditions that would not have harmed them. This essay catalogues twelve tests that produce that conversion, organised by the four mechanisms through which it is achieved.
