As the authors have expertise in the impact of hospital transfers on artificially lowering ICU mortality, we wonder if they explored the role of transfers in explaining the association between staffing and mortality seen in their study cohort. Nonphysician ICUs are more common in lower volume, community hospitals. Studies show that hospital concerns about public reporting of mortality result in more acute-care transfers for critically ill patients. It is plausible that sicker patients cared for in nonphysician ICUs in the study by Kerlin and colleagues, given their higher severity of illness, were more likely to be transferred to higher-level care centers, making mortality appear “better” at nonphysician hospitals. This residual confounding by severity of illness in patients transferred from outside hospitals to intensivist hospitals may mask some of the effect of ICU staffing on subsequent mortality.