There is a growing sentiment to tie physician performance in the care of certain key disorders to insurance reimbursement.1 Diabetes is one of those disease states. This concept is not only being advocated for Medicare beneficiaries, but also is being considered for many private health insurance contracts.2 Because private insurers are encouraged to offer Medicare beneficiaries alternative insurance plans,1 there is no reason to believe that so-called “pay-for-performance” plans will not be included in the insurance contracts for these Medicare alternatives. If pay for performance becomes a widespread concept in health insurance, there is no doubt that it will be applied to diabetic patients. Therefore, it is important for providers in diabetes care to understand this concept and its practical implications to their work.
Design of Pay for Performance
The concept of pay for performance has gained much momentum in the past few years. There is the intention of applying this scheme to all levels of health care provision, ranging from individual providers to hospitals and nursing homes.1 There is much enthusiasm about how the application of this form of reimbursement will improve the health status of the American population,3-5 and many recognized experts, both health professionals and lay experts, are enthusiastic about this model.
Pay for performance depends on the basic assumption that specific and reliable end points of care may be identified that will measure whether the care rendered to patients is carried out in a reasonably competent manner.1,6 Demonstration projects are in progress.6 Advocates of pay for performance describe end points that would be taken from the official guidelines of national health organizations such as the American Diabetes Association (ADA).7 The idea is that the systematic application of such guidelines can significantly improve diabetes care.8 Pay-for-performance contracts would financially reward providers when their patients' care meets the applicable guidelines and decrease reimbursement for visits when it does not.9 The intention is to focus providers on those aspects of care that the insurers, backed by national organizations, believe are most important and to reduce provider focus on less critical aspects of care. There is an overall belief that this sort of structure for our health system will measurably improve the health status of the general population.10
Application of Pay for Performance to Diabetes Care
Few disease states will be as affected by pay-for-performance plans as diabetes. At least three well-recognized sets of care guidelines apply to most diabetic patients. These are the ADA clinical practice guidelines,7 the National Cholesterol Education Program Adult Treatment Panel III guidelines for management of lipid disorders,11 and the Joint National Committee 7 guidelines for treatment of hypertension.12
These three disease states coexist in many patients with diabetes. The end point measures that are the most likely to apply to the assessment of care in such patients are those three sets of guidelines. The relatively high level of concurrence of the three disease states in most diabetic patients will certainly influence physicians' approach to this patient population versus other patient groups. This will be in part because of the much enhanced financial risk attendant to the care of a patient group that will be assessed by three separate guidelines.
There seems little doubt that priorities of care will change in the approach to diabetic patients. Treatment approaches that are the most effective in bringing blood pressure, blood glucose, and lipids under control will gain the most active attention of treating providers. Therapies that address ancillary needs of these patients are likely to become much less interesting.
The more difficult the application of a treatment proves to be, the less likely it will retain provider loyalty. Drugs for the primary disorders that are readily available on the managed care formulary of a given patient will be much more readily used for those patients than they are even now. This is because the time and effort to bring all three areas of clinical concern under control in diabetic patients will often be quite intense and time consuming. Therefore, the distraction of using complex therapies for less relevant areas or treatments for the primary disorders that are not on the patient's formulary will become much less feasible.
Problems With Pay for Performance in Diabetes Care
The enthusiastic rush to apply the pay-for-performance concept to diabetes care ignores certain critical problems that may beset its application. Perhaps the most important of these rests with the inherent limitations of clinical practice guidelines themselves. Additionally, there is the potential for a serious conflict of interest in health insurers' management of this concept related to the formulary drugs that are available to providers to achieve the required end points.
These two critical issues may result in pay for performance becoming a mechanism for health insurers to achieve a massive reduction in provider reimbursement, framed as a quality improvement process on behalf of patients. If taken to a logical extreme, an unintentional and undesirable result of pay for performance might be a decrease in the availability of health care for those individuals who most need it: the less educated and capable patients, who are most likely to have difficulty achieving excellent clinical outcomes within the usual care environment.
Clinical practice guidelines were never framed with any specific application in mind. They were suggestions for goals of care. How feasible their end points are to achieve in large populations of unselected patients treated in routine clinical settings is only now being tested.
Nevertheless, there are worrisome trends throughout the literature to suggest that in diabetes care or lipid management, there may be limitations to success. For example, some of the most recognized studies on efforts to achieve glycemic control in diabetic patients never achieved the end point of bringing the hemoglobin A1c (A1C) to < 7%. The lowest mean A1C in patients with type 1 diabetes in the Diabetes Control and Complications Trial was 7.2%.13 However, that was achieved with an unusual organization of care, in which an excess of health professionals far beyond what is available in routine clinical settings was applied to gain patient adherence to the treatment protocol. Once the study ended and the clinical protocol was extended within more typical clinical settings, the A1C in the group of patients being intensively treated rose to 7.9%.14 This was not significantly different from that of the control population (8.2%). Similar results have been reported in studies on patients with type 2 diabetes.15,16
A recent study on patients with lipid disorders suggests one important possible flaw in the pay-for-performance scheme. This was a study of > 12,000 patients with hypercholesterolemia.17 The results demonstrated that regardless of the clinical care rendered, the outcomes of care were substantially influenced by the clinical decisions made by patients independent of physician care. Factors such as the cost of drugs and the patients' age and sex were important determinants of whether the patients chose to follow physician advice. These questions, more than the excellence of provider care, determined the outcome of treatment. Thus, a major flaw in the application of pay-for-performance plans to disorders such as diabetes is that these plans ignore how dependent the outcomes of care are on patient behavior independent of the excellence of care provided.
Another critical issue is the potential conflict of interest in the governance of the drugs available to achieve the required outcomes of care by the same insurance organizations that stand to benefit financially by the application of pay for performance. If a health insurer restricts its formulary to less expensive and less potent therapeutic agents, the insurer further transfers the risk for achieving positive clinical outcomes to the efforts of the provider. In this situation, the provider may be required by an insurer to assume infinite risk with no real protection. The alternative of spending excessive amounts of time to gain approval for the use of nonformulary drugs will be infeasible as insurance reimbursements decline under the pay-for-performance plan. This means that the ratio between reward and punishment in the fee schedules for patients under this system will be important questions. For providers who must use highly restricted formularies to achieve required outcomes, the reward for doing so must be high. We expect that the discounts in reimbursement for failure will be steep.
Possible Outcomes of Pay for Performance
We expect pay for performance to stimulate a positive reorganization of the focus of care to those end points that national health organizations have identified as being the most important. We also expect the health status of many patients to improve under this system. In these aspects, pay for performance may be a positive influence.
However, there also may be some very undesirable outcomes of pay for performance. One may be the systematic exclusion of patients with disorders such as diabetes from many clinical practices. In general, the more risk associated with a patient population, the less attractive that group of patients may be to clinicians. A similar worry is that patients perceived as being noncompliant or those who fail to achieve clinical goals quickly and easily may be excluded from clinical practices. Individual practices may shape their patient base to become less financially risky.
What will happen to the more challenging patients is a real question. Unless a tier of high risk is established (similar to assigned risk in the field of auto insurance), no provider, whether primary care or specialist, will be willing to take on such patients. Thus, pay for performance may become a system that includes only the best-performing patients in disease areas such as diabetes.
Steven B. Leichter, MD, FACP, FACE, is co-director of the Columbus Research Foundation and president of Endocrine Consultants, PC, in Columbus, Ga. He is a professor of medicine at Mercer University School of Medicine in Macon, Ga.
- American Diabetes Association