From The Harms of Screening, by Steven H. Woolf and Russell Harris:
Concerns about the harms of screening might seem exaggerated without closer scrutiny. For instance, if a test with 90% sensitivity and 96% specificity (better than most screening tests) is used to screen for a condition with a prevalence of 0.6% (typical of some cancers), 88% of abnormal results will be erroneous; for every 1000 patients screened, only 6 will have the condition and 40 will have false positive results. That ratio may be acceptable if the benefits obtained by the 6 patients with true disease outweigh the harms incurred by the 1000 patients who undergo screening, but what if there is little evidence that early detection improves their prognosis? If only 1 or 2 of the 6 patients obtain benefit, is it ethical to subject the entire population to screening? The concern about overdiagnosis is justified: by some estimates, overdiagnosis accounts for 15% to 25% of screen-detected lung cancers and potentially more breast and prostate cancers.
Do not confuse this with a “don’t screen” message. It’s a “screen smarter” message.