Survival is higher after repeat lung metastasectomy than after a first metastasectomy: Too good to be true?

In this article, we raise a note of caution about oversophisticated analyses of observational clinical data that may result in misleading conclusions. The analysis of survival after metastasectomy provides examples. Previous analyses have failed to take account of confounding due to successive selection of patients inherently more likely to survive. Clinical selection is based on multiple unrecorded factors, some subjective or even subliminal, which add up to a multifactorial effect that we call ‘‘survivability.’’ The points can be generalized to other diseases and other types of surgery. The conclusion reached is that when there are complex issues of case selection, and then reselection for subsequent successive treatments, then random assignment to differing protocols with planned intention-to-treat analysis may be the only way to separate causation from association.