The 'obesity paradox' and survival after colorectal cancer: true or false?

Body mass index (BMI), as an approximation of general adiposity, is an established risk factor for incidence of several adult cancer types, including colorectal cancer (CRC). There is a common perception that these relationships extrapolate directly as adverse prognostic factors after diagnosis, but evidence for this is lacking. The paper from Sclesinger et al. in this issue of the journal adds a new dimension to this debate focusing on relationships of post-diagnosis BMI (as a marker of the steady-state weight among survivors) and survival, and provides evidence on a decreased mortality risk among overweight (post-diagnosis BMI 25.0-29.9 kg/m(2)) compared with normal weight (post-diagnosis BMI 18.5-24.9 kg/m(2)) CRC survivors-an example of an 'obesity paradox.' The observation of the 'obesity paradox' is well documented in the methodology literature, but perhaps, less familiar to the cancer readership. Three broad classes of explanation are posited: (1) the associations are true and plausible; (2) the associations are false and reflect methodological issues; or (3) the observations represent a specific form of selection bias, known as collider bias. The present author argues that the obesity paradox reflects the latter-a product of a statistical bias-and emphasizes that, while these findings are hypothesis generating, they will not alter clinical practice or recommendations.