POINT: Volume remains a useful outcome proxy for complex procedures.
High-volume centers have been shown to have better outcomes than low-volume centers for procedures such as pancreatic and esophageal resections and complex aortic repairs, so hospital procedure volume should be used as a proxy measure for surgical outcome quality.
We need to accept that volume affects outcome. You get better at something if you practice it regularly, whether it’s piano, tennis, or surgery. If you don’t see that effect, it probably means that everyone is far enough along on the learning curve that the effect has gone away, or you are not looking at the right outcome. After all, if a surgeon’s parent needed an esophageal resection, that surgeon would probably find the very highest-volume colleague in the area to perform the operation.
But whether or not the volume-outcome relationship should be used to steer patients to high-volume hospitals depends on how steep the relationship is. For complex operations, it’s very steep – 10 percentage points or more in mortality. For those cases, referring patients to high-volume centers and using volume as a proxy outcome measure make sense; there are just not enough cases for additional outcome measures to be statistically meaningful bases for policy decisions. Plus, once you identify the high-volume providers for those operations, you can find out what they’re doing and share those lessons with others.
For more common operations, like coronary artery bypass grafting, inguinal hernia repair, colectomy, and bariatric surgery, the magnitude of difference is much smaller and mortality is rare, so it makes sense to use adverse event rates, functional outcomes, postoperative pain, patient convenience, and similar issues as quality measures, as far as they are known. Registries that track such issues are being built; we should use them.
Some have called the distinction between low-, medium-, and high-volume hospitals arbitrary, but you have to make such distinctions in order to make policy decisions. If you advocate sending complex surgical patients to high-volume hospitals, you have to define what that means. It’s not an easy call; if you require too many procedures for the high-volume designation, you limit the number of institutions that qualify and, therefore, you limit patient access. If too few are required, the distinction is less meaningful. Where you draw the line has trade-offs.
In short, the most rare, complex procedures should be regionalized to high-volume centers of excellence. For more common procedures, we should measure outcomes and share the results so we all keep moving forward.
Dr. Dimick is an associate professor of surgery and chief of the Division of Minimally Invasive Surgery at the University of Michigan in Ann Arbor. He is a cofounder of ArborMetrix, which makes software to measure hospital and surgical outcomes.
COUNTERPOINT: Better indicators of surgical quality need to be identified.
We recently failed to find a statistically significant association between hospital volume alone and in-hospital mortality for pancreatic resection, abdominal aortic aneurysm repair, esophageal resection, and coronary artery bypass grafting (Ann. Surg. 2012;256:606-15). These results suggest that hospital procedure volume alone should not be used as a proxy measure for quality of surgical outcomes.
Hospital volume isn’t entirely irrelevant, but as a stand-alone metric upon which to base referral patterns and other policy decisions, it doesn’t appear to be the most reliable measure. In studies where volume does seem to make a difference, it may be acting as a surrogate for other factors that probably matter more, including individual surgeon experience, careful patient selection, protocol-driven approaches to perioperative care, and well-coordinated surgical and patient-care teams.
These factors may cluster in certain large-volume institutions, but it is also true that some low-volume hospitals have outstanding surgical outcomes and some high-volume hospitals do not.
We used different statistical techniques in our study than those used in previous investigations that found a link between volume and quality; we believe our methods are more robust. One of the limitations of those earlier investigations is that they used arbitrary values to define hospitals into low-, high-, and in-between volume categories, which sacrificed precision.
Instead, using discharge data from 261,412 patient records in the 2008 Nationwide Inpatient Sample, Dr. Benjamin Kozower, Dr. George Stukenborg, and I analyzed volume as a continuous function and accounted for variations in the nonlinear relationships that exist between hospital volume and mortality for the four operations – the hospital volumes of which have been used as measures of surgical quality by various policy and regulatory bodies. When we did that, not only did individual hospital volume not have a significant risk-adjusted association with mortality, but also its relative strength of association with mortality was disproportionately small compared with other factors, including patient-level factors such as nutrition status and renal health.