» Back to Table of Contents

Moving toward a Patient-centered Performance Management System

As Dr. Francis described in his commentary article, initial performance measures launched in VA and elsewhere nearly two decades ago helped drive quality improvement at a time when performance was universally low.¹ Presently, VA has achieved higher levels of performance on these "one size fits all" measures than most other health systems. As we and others have documented (see references 1 and 2 from Dr. Francis' commentary), when performance on such measures is high, aiming for even higher performance can push clinicians and facilities to overtreatment; such efforts may also ignore legitimate exceptions to the measures.² VA has the opportunity to move beyond those first step measures and develop a performance management system to help guide optimal patient-centered care—a system that takes into account the risks and benefits of tests and treatments for individual Veterans, as well as Veterans' own preferences.³

We propose that to move to such a system, we need to consider the net value in care provided—including expected benefits and risks at the individual patient level—and to elicit individual patient preferences for care. By net value, we mean the benefits or gains that can be expected from receiving care, minus costs and potential harms from the care. A patient-centered performance management system should define a point at which the net value of care for an individual is so high that it should almost always be provided, and not doing so would indicate poor care; and a point at which net value is so low (or even negative, in the instance of harmful care) that provision of that care should be an indicator of lower performance. In between the points of high-value and very-low value care lies a gray zone; in this gray zone, provision of care cannot be determined solely by the evidence and should instead be defined through informed patient-clinician decision-making. Documenting both that an informed discussion took place and what the patient's stated care preferences are should constitute high performance in the gray zone.

How would a patient-centered performance management system work? Consider a common scenario. Mr. John L., a 48-year old non-smoker, with treated hypertension and hyperlipidemia, but no history of cardiovascular (CV) disease presents to clinic with a systolic blood pressure of 144/82 mm Hg. How should the provider decide whether Mr. L needs to have his blood pressure medications adjusted? Currently, because the patient does not meet the "all or nothing" blood pressure < 140/90 measure, his provider would likely add a blood pressure medication to his regimen. In a patient-centered performance management system, the electronic health record (EHR) would automatically calculate, using a Veteran-specific risk model, Mr. L's 10-year risk for CV disease (about 5 percent). Then, the system would present the CV risk reduction (benefit) and side effects of specific medications or other treatments in a way that both provider and patient could understand; the system would also suggest whether the magnitude of the proposed treatment effects was large (high net value), moderate or small (moderate to low net value), or even harmful (negative net value). The EHR would also promote documentation of Mr. L's goals and preferences, such as avoiding side effects, having high levels of energy, and minimizing co-pays. Because the clinical impact from additional treatments in Mr. L's case is low, the provider and patient could decide together not to add an additional antihypertensive medication. The EHR would then prompt for documentation that the patient received information on the risk and benefit of treatment options and chose not to pursue treatment, thus fulfilling the performance measure for hypertension management. In this way, the system would not only guide real-time and patient-centered treatment, but also produce retrospective performance measures that reflect, for example, the efficiency of care (frequency of provision of high value care minus frequency of provision of very low value care) and patient-centeredness of care (documentation of patient preferences in moderate value care). Importantly, it would also incorporate the three elements Dr. Francis suggests, and we agree, are needed to advance performance measurement: risk-tailoring, capturing clinical actions, and incorporating patient preferences and goals.

Given the advances in the last decade in health services research, decision science, implementation research, and informatics, progressing to a patient-centered performance management system is within our grasp, particularly in VA. Success, however, will require committed partnerships between clinical, policy, and operations leaders, informaticians, and health services researchers. Indeed, health services research can help significantly impact VA care delivery by: developing approaches for defining high value, low value, and preference sensitive care; developing and testing use of VA-specific risk prediction models; developing decision support tools that incorporate individual risk and benefit information and testing alternate ways to present this information to patients and providers; conceiving and evaluating new measures to assess patient-centered quality of care, including measures of under-and over-treatment; and assessing how incorporation of patient-centered performance management influences patient experiences, prescribing, time involved in care provision, costs, and provider satisfaction.

References

1. Asch, S.M. et al. "Comparison of Quality of Care for Patients in the Veterans Health Administration and Patients in a National Sample," Annals of Internal Medicine 2004; 141(12):938-45.
2. Kerr, E.A. et al. "Building a Better Quality Measure: Are Some Patients with Poor Intermediate Outcomes Really Getting Good Quality Care?" Medical Care 2003; 41(10):1173-82.
3. Sussman, J.B. et al. "Individual and Population Benefits of Daily Aspirin Therapy: A Proposal for Personalizing National Guidelines," Circulation: Cardiovasc Quality and Outcomes 2011 May; 4(3):268-75.