Kramer Barnett S, Croswell Jennifer Miller
Office of Disease Prevention, Office of the Director, National Institutes of Health, Bethesda, Maryland 20892, USA.
Annu Rev Med. 2009;60:125-37. doi: 10.1146/annurev.med.60.101107.134802.
The concept of early detection of cancer holds great promise and intuitive appeal. However, powerful biases can mislead clinicians when evaluating the efficacy of screening tests by clinical observation alone. Selection bias, lead-time bias, length-biased sampling, and overdiagnosis are counterintuitive concepts with critical implications for early-detection efforts. This article explains these biases and other common confounders in cancer screening. The most direct and reliable way to avoid being led astray by intuitions is through the use of randomized controlled trials.
癌症早期检测的概念前景广阔且颇具直观吸引力。然而，在仅通过临床观察评估筛查测试的疗效时，强大的偏差可能会误导临床医生。选择偏差、领先时间偏差、长度偏倚抽样和过度诊断是一些违反直觉的概念，对早期检测工作有着至关重要的影响。本文解释了这些偏差以及癌症筛查中其他常见的混杂因素。避免被直觉误导的最直接、最可靠的方法是采用随机对照试验。
