Performance Measurement 2.0: Time to Raise the Bar

Abstract

In the current climate of American health care reform, effective and accurate performance measurement and reporting of the quality of care provided by caregivers have risen to the top of critical priorities. On one hand, there are more than 100 regional quality reporting initiatives relying on public reporting of performance measures, and yet there is a high degree of variability as to which of these measures are reported, compared, and implemented. Certainly, the increased availability of comparative clinical outcomes among providers of care has proven to be an initial effective catalyst in achieving significant improvements and transparency in the quality of care delivered within the US health system, but high-quality evidence of significant overall impact is still lacking. As a result of these documented inconsistencies, in the current issue of this journal, Roski and Kim have correctly called for a more efficient means of generating nationally consistent performance information. Indeed, the Quality Alliance Steering Committee has already developed a road map and 3-year plan to establish a cohesive infrastructure to achieve this very important national goal. Given the growing mandate for health care providers to demonstrate meaningful improvements in clinical quality, the effective implementation of evidence-based performance measures that are derived from well-constructed clinical practice guidelines has become a critical success factor. When developed by professional society and other health care delivery experts, such guidelines and measures must also ensure both feasibility and improved outcomes, especially to the public and payers. To the extent that core performance measures can be adopted uniformly across the spectrum of payers, individual practitioners and groups will be able to implement delivery schemes more effectively. Because the stakes have become higher than ever to achieve these ends, a more rigorous and consistent approach to these complex issues, such as outlined in the following discussion, must now be established. Clinical practice guidelines require a systematic evaluation of evidence behind the science supporting guideline recommendations for care delivery. Using a taxonomy or framework for grading the quality of evidence and providing a subsequent concomitant “strength of recommendation” is desirable whenever possible. This framework should be published along with guideline recommendations and supporting text, which should also include a summary of the quality of the evidence using graphical and tabular summary statistical techniques. Examples include the evidence grade definitions used by the US Preventive Services Task Force, the modified Grading of Recommendations Assessment, Development and Evaluation (GRADE) system used by the American College of Physicians Clinical Effectiveness Assessment Subcommittee, and the system deployed by the American College of Cardiology and American Heart Association. Another excellent example is the evidence grading system deployed by the Society for Healthcare Epidemiology of America to reduce hospitalacquired infections, but currently, there is no standard “one size fits all” approach to be used by every group of guideline developers. Performance measures should be based on clinical practice guidelines and are useful to assess whether care delivery is achieved in accordance with guideline recommendations. There are 5 critical aspects of performance measures with respect to scientific evidence. First, any performance measure should be developed and accompanied by a structured evaluation of evidence such as those already mentioned for clinical practice guidelines. The most useful and valid performance measures are those that are based on both the highest quality of evidence and the strongest recommendations from guideline developers. Under these circumstances, it is more likely that well-constructed performance measures will actually evaluate the degree to which quality of care has been delivered. Subsequent quality improvement methods and initiatives designed to increase the frequency of care in accordance with higher success of such measures will also be more likely to have a significant impact on the target population. Second, certain performance measures can be independently evaluated as to whether actual improvements in care can be achieved if levels of performance increase. Such an evaluation will, over time, generate a second type