Is it defensible to use volume standards for purchasing care?

The emergence of the Leapfrog group initiative to use measures of quality and safety as criteria for contracting for provision of medical services was a watershed event. Major payers would pay for quality, or, more accurately, no longer pay for inferior quality. Not surprisingly, physicians have been skeptical about this enterprise. Battered over the past 2 decades by an never-ending array of purchaser/payer/plan restrictions on medical practice that were clearly designed to save money (and increase profits), many saw this as yet 1 more device—a ruse, perhaps—for cutting costs—at doctors’ expense. Equally galling was the use of volume of services as a measure of quality. For them, the findings by Goodney et al that there are no significant differences overall in length of stay (LOS) or return to the hospital between high-volume and low-volume hospitals comes as welcome news: See, it doesn’t save money after all. But ironically the results are also pleasing for those who welcomed this innovative approach to improving quality of care: See, higher quality doesn’t cost more. In fact, there are several aspects of the study that suggest these measured differences don’t tell the entire cost story. First, deaths in the operating rooms obviously reduce lengths of stay. To the extent that mortality differences result from more of these early deaths, LOS would be lower in low-volume hospitals. If so, increased costs in high-volume hospitals would be worth paying for. Second, academic health centers tend to be high-volume hospitals (one of the reasons there has been little outcry about the Leapfrog requirements from academia). Both costs and lengths of stay are often higher in teaching hospitals, and for complicated patients, quality of care is better too. Unless data are adjusted for this variable, the results could be skewed. Finally, the use of logarithmic transformation to deal with the problem of rare very long stays, though justified for performing regression analysis, could also skew the data. Very long lengths of stay can have a profound impact on the average LOS. If low-volume hospitals with more complications also have a higher fraction overall that have very long stays (a reasonable and testable hypothesis) reducing the impact of those cases by logarithmic transformation favors low-volume hospitals. But, if we accept this current news that there are no significant difference overall in length of stay and returns to the hospital at face value, as well as the authors’ important prior work that demonstrated better outcomes in high-volume hospitals, it is still legitimate to question whether volume of cases is a suitable measure by which to judge quality. Most of us would say it is not. Surgeons, far more than the public, the policy makers, and even our medical colleagues, recognize the profound effect of surgical skill on outcomes, as well as the remarkable variation of its distribution among our colleagues. We also know that these differences are only modestly related to the volume of procedures