Questioning a Sensible Result

Contemporary data collection strategies, storage capabilities, and modern statistical methodology have made retrospective analyses of observational databases commonplace. Such databases afford opportunities to learn about the effectiveness and risks of interventions or health behaviors that generally cannot be randomized. In this issue of NEJM Evidence, Cho et al.1 assemble survey data and cohort data from four countries to quantify the association between age–sex-specific smoking cessation and mortality. The authors conclude that smoking cessation at any age is associated with lower excess overall mortality risk and lower death from diseases made more common by smoking. It is difficult to argue with this conclusion — to question the magnitude of the associations is not.