Case-mix Measurement and Assessing Quality of Hospital Care

Patients, insurers, and health care systems are increasingly counseled to consider both quality and cost in purchasing health care services, particularly hospital care, to "buy right." Although far more cost information is available today than 25 years ago, purchasers are no better off when trying to select hospitals of high quality. Nevertheless, quality issues are increasingly in the public mind because of fear that efforts to cut costs will reduce quality as well. The heart of the problem is that research on quality of care has focused on the way care is rendered, the process of care, rather than on outcomes of care. Such process measures are critical to quality assurance activities meant to correct problems, but most consumers and payers are more interested in using quality assessment to select good hospitals and avoid the bad. For such comparison shopping, information on outcomes is more objective and more relevant. Outcome measures such as death and rehospitalization may also be less expensive than process measures because they are available from administrative data. Unfortunately, comparing outcomes intelligently requires that we be able to correct for differences in hospital case mix that might account for differences in outcomes and that we be confident that differences in outcomes that remain after correcting for case mix truly reflect differences in quality of care. One example of using outcomes as a screen for quality of care is the release of hospital-specific Medicare mortality rates, which will occur about the time this supplement appears. Another is the recent decision of the Pennsylvania Health Care Cost Containment Council to require all hospitals to collect case-mix data on all discharges in order to determine quality of care. These public strategies are controversial, principally because of uncertainty as to whether the state of the art in case-mix adjustment really allows meaningful comparisons of outcomes such as death rates. These strategies imply that case-mix measurement is, at least potentially, a key to measuring quality of care. However, case-mix measurement of this kind is at the cutting edge of health services research and is quite controversial. To provide readers with a sense of the diversity of scholarly views on the application of case-mix measurement to measuring quality of care, I asked six investigators to respond to questions related to three areas. In putting together the contributors' comments, I have preserved their arguments as they presented them, even at the expense of occasional repetition, in order to indicate the degree of consensus and disagreement that actually exists in this field.