Recommendations for cancer screening would be different if we measured endpoints that are valid, reliable, specific, and important to patients.

Background: Despite enthusiasm for cancer screening, systematic reviews consistently fail to show that screening reduces all-cause mortality. This narrative review explores conceptual issues, and inconsistencies between evidence and opinion about screening.

Review: We examined the interpretation of screening studies in relation to three intellectual traditions: (1) The relationship between prevention and cure; (2) Confirmation bias and the challenge of incorporating new data: less care may produce better outcomes than more care; (3) The answers to three structured questions about efficacy, effectiveness, and value of treatments proposed by Sir Archie Cochrane and Sir Austin Bradford Hill.

Synthesis: When considering extensions of life expectancy or all-cause mortality, systematic reviews typically show cancer screening to have only small effects and often non-significant effects on all-cause mortality. Early diagnosis does not assure application of an intervention that alters the pathway toward demise. The interpretation of screening results is also affected by several known biases. Investigators and advocates are encumbered by an over focus on studies designed to determine if a treatment can work under ideal circumstances. To advance the field, we need a greater emphasis on evaluations that ask 'Does the treatment work under real-world conditions?', and 'Is the treatment worth it?' in terms of outcomes that are meaningful to patients.