Hirsch and colleagues [5] report a complex analysis aimed at defining the relationship between institutional volume and hospital mortality for congenital heart surgery and, not surprisingly, suggest that high-volume institutions have lower hospital mortality than low-volume institutions. Although the presented data are restricted to two specific procedures, the conclusions are consistent with the majority of reports examining this issue using a wider range of procedures [4, 6]. However, it should be noted that others have not always found a consistent relationship between hospital volume and outcome [7]. Therefore, it is important to recognize the possibility that hospital volume may be a proxy for other aspects of care that are important predictors of survival and may be more commonly associated with high-volume centers [2]. This concern is supported by the observation that in most published analyses there are some low-volume centers that have riskadjusted mortality which is equivalent (or superior) to that of large-volume centers [2, 6]. Comparison of diagnosisspecific cohorts within the Congenital Heart Surgeons Society using databases that are rich in patient-specific and institution-specific data suggests that there is considerable heterogeneity in the relationship between institutional (and surgeon) performance and volume. Some lesions appear to have outcomes that are related to volume, whereas patient and institutional management patterns may be the primary determinants of outcome with other lesions. Also, and importantly, an institution with superb performance with one lesion may have quite mediocre performance with another lesion. In an analysis of the volume–mortality relationship in California, Bazzani et al. [2] demonstrated that the presence of a large center with superb outcomes in California can ‘‘leverage’’ an analysis toward the conclusion that there is an inverse relationship between volume and mortality. In the Bazzani example, elimination of the single large center from the analysis nearly eliminated any demonstrable relationship between volume and mortality. In this regard, a deficit in the article by Hirsch et al. [5] is the lack of a scatter plot depicting the volume–mortality data points and confidence limits around the volume–mortality statistical model to allow us to interpret the strength of the relationship that the authors have reported. We would be better informed if we could examine the frequency with which some low-volume institutions may have performed as well as high-volume institutions. Furthermore, we need some objective determination of the confidence limits supporting the authors’ conclusions that a volume–mortality relationship exists. The observation that some low-volume centers can deliver performance that is equivalent to that of higher volume centers suggests that there is more afoot than simply the number of procedures performed. Although the concept that ‘‘practice makes perfect’’ is intuitive, there may be other factors that influence mortality rates. These other important factors need to be identified for two reasons. First, any effort at regionalization should be designed to move patients toward the center with the characteristics that imbue their patient population with superior survival. These characteristics may be limited to hospital volume but are likely to reflect other more subtle markers of institutional performance. For example, Ashburn et al. [1] demonstrated that optimal survival for patients with C. A. Caldarone (&) O. Al-Radi Division of Cardiovascular Surgery, The Hospital for Sick Children, 525 University Ave, Toronto, ON, Canada M5G 2L3 e-mail: christopher.caldarone@sickkids.ca