Background: Randomized controlled trials (RCTs) of lung cancer screening consistently show an excess number of cancer cases and longer survival in screened groups, but no difference in mortality between screened and control populations.
Methods: The current study reviewed the various types of biases that confuse comparisons based on intermediate endpoints such as stage distribution and survival and the reasons for basing evaluations in RCTs of screening for early cancers on mortality from a specific cancer.
Results: Four RCTs all showed improved stage of disease and survival in screened subjects, but there was no difference in mortality between screened and unscreened populations. The possible explanations for the higher incidence are chance (failed randomization) or "overdiagnosis" (detection of cases by screening that otherwise would never have surfaced). Analysis of the trial results confirmed that chance alone was a very unlikely explanation. Evidence suggests that some overdiagnosis of lung cancer is likely in screened subjects. This is a consistent observation in all other programs of screening for early cancers (breast, prostate, and neuroblastoma).
Conclusions: Overdiagnosis of cancer cases resulting from the screening process itself will give rise to excess cases of disease, and may, at least in part, explain the observations in the randomized trials.