Statistics from Altmetric.com
In many industrialised populations, adenocarcinoma of the oesophagus holds the dubious distinction of being the cancer for which the proportional rise in incidence has been most rapid.1 The factors conferring increased risks are well established, including obesity, gastro-oesophageal acid reflux, smoking and male sex.2 It has also long been known that Barrett's oesophagus, the metaplastic precursor, has a measurable rate of progressing to cancer. This observation has given rise to the notion that screening or surveillance for Barrett's oesophagus should, in theory, reduce the risks of dying from oesophageal adenocarcinoma. However, screening for any cancer is a deceptively tricky business, and while the concept of screening holds considerable intuitive appeal, it is very difficult to establish that any screening programme confers benefits.
Bhat et al3 attempt to infer the effect of a prior diagnosis of Barrett's oesophagus among patients with oesophageal adenocarcinoma, and to estimate the effects of two biases that are known to affect the interpretation of screening studies, namely, lead time bias and length bias.4 Lead time is the interval by which a diagnosis of cancer has been brought forward by screening; it has the effect of spuriously lengthening the time from diagnosis to death, even if no survival benefit is conferred by the screening activity. For example, it is reasonable to assume that patients with Barrett's oesophagus undergo more frequent endoscopies than other people, and therefore may have their cancers detected at an earlier point in time than would otherwise be the …