A Proven New Model for Reimbursing Physicians


About 35-45% of health systems’ total costs stem from practices that don’t benefit and may even hurt patients — such as questionable or unjustified treatments, unproven variation in treatment delivery, redundant or otherwise unnecessary testing, and reimbursement for services that result from bad health outcomes.

At Pennsylvania-based Geisinger Health System, where I was CEO for 15 years, changing our model for physician reimbursement was one way we aimed to combat such practices and achieve value-based care. Physician reimbursement is just part of a complex puzzle, of course, and our experience at Geisinger is not the only viable approach, even to that one piece. But we nonetheless learned some valuable lessons in this important area. I’d like to describe our model, using some of the hard data that illustrate its success, and then to offer my analysis of the underlying elements that made it work.

The Model and Its Results

Geisinger’s physician reimbursement model meant focusing on how we pay the physicians we directly employ and how our insurance company reimburses the impaneled doctors we don’t employ. While our own physicians are still largely reimbursed on a fee-for-service basis from non-Geisinger insurance payers, since 2002, we’ve been using the following 80/20 compensation model:

  • 80% of total cash compensation is based on the usual piecework metrics: panel size, number of patients seen, number of work units performed, and so on.
  • 20% of total cash compensation (an arbitrarily chosen threshold) is linked to how well physicians improved quality and reduced costs. Toward this end, we track both hospital-based care and improvements in health outcomes for outpatients with multiple chronic diseases (who are served in our community-practice service line).

For hospital-based care, each of our 28 areas that provide intervention services, such as coronary artery bypass graft (CABG) surgery, annually choose a specific innovation target. And 20% of their compensation is linked to achieving that strategic goal (unrelated to fee-for-service compensation). For instance, for elective CABG, the selected goal in 2006 was for all patients to successfully meet 120 best-practice treatment requirements that are known to be critical to achieving optimal health outcomes. The goal was reached, and the result was a 67% relative improvement in the mortality rate and 18% lower costs. Physicians received their 20% innovation bonus.

Insight Center

  • Measuring Costs and Outcomes in Healthcare
    Sponsored by Medtronic

    A collaboration of the editors of Harvard Business Review and the New England Journal of Medicine, exploring cutting-edge ways to improve quality and reduce waste.
For the community-practice service line, beginning in 2006, physicians committed to achieving nine best-practice treatment goals that have known associations with better outcomes for patients with type 2 diabetes. Within three years (by 2009), 99% of our 30,000 type 2 diabetes patients achieved at least seven of our nine goals. The result: 141 strokes and 306 heart attacks were prevented, and 166 patients avoided diabetic-related eye disease. Again, the physicians received their 20% performance bonus.

Within several years of offering this mixed-payment incentive, we saw better outcomes at a lower cost for 18 common treatment interventions, due to mitigating unjustified variation in hospital-based care. For outpatient care, optimizing management of high-risk chronic diseases yielded consistent 30% absolute declines in the need for acute-care hospitalization and rehospitalization. Even more important, we saw significant decreases in the percentage of patients with prevalent chronic conditions (such as type 2 diabetes, congestive heart failure, chronic obstructive pulmonary disease, and hypertension) who progressed to “long-term medically ill” status. The reduction in the total cost of care did not stem from restrictions in access to care, but from real improvements in health outcomes.

In 2010 we began to extend our pay-for-value 80/20 model to the physicians we don’t directly employ, both inside and outside Pennsylvania, most notably the private-practice primary care physicians impaneled by our insurance company. For Geisinger-insured patients with multiple chronic diseases, up to a 15% increase in the private practitioners’ total compensation could occur if our version of chronic disease management redesign was accomplished. Within one year, the hospitalization rate for these patients decreased by 30% — similar to our experience with the employed physicians.

Keys to Changing Physician Behavior

These successes weren’t due to compensation changes alone. Our new care pathways were effective because they were led by physicians, enabled by real-time data-based feedback, and primarily focused on improving the quality of patient care. Stepping back, I would highlight six essential elements of our new reimbursement model for physicians:

  1. We worked to encourage payer and provider to work together to create value for patients who traditionally incur high costs and have poor outcomes. For example, we aimed to identify and eliminate erythropoietin (EPO) treatment for the 20% of anemia patients who could just as well be treated with iron supplements, thereby decreasing EPO side effects and dramatically reducing costs.
  2. We used specific data from both the insurance company and patients’ electronic medical records. Admittedly, this process is not a simple one, particularly if the insurance company and the provider group are not connected. But once we had it, this critical data was shared as feedback with providers at the time they were actually caring for their patients, identifying which physicians and which practices were achieving the best results in chronic disease management. We then extended the successful methods to the other caregivers.
  3. We tied a substantial amount of total compensation (15% to 20%) to value-based outcome improvements, not piecework reimbursement for services. And if our fee-for-service reimbursement from the non-Geisinger insurers goes down, there is no reason why the value-based compensation cannot be increased.
  4. The men and women who actually work in the service lines themselves chose which care processes to change. Involving them directly in decision making secured their buy-in and made success more likely.
  5. Patient-care metrics were followed in real time, and quality- and cost-of-care outcomes were monitored annually.
  6. Cost reduction was always a consequence of our practice redesign, but it was the goal of improving patient outcomes that fundamentally motivated our physicians to change their behavior. And it was an engaged physician leadership that drove the care redesign.

Are these ingredients in changing physician behavior the final word on physician reimbursement models? Of course not. Indeed, using reimbursement incentives to change doctors’ behavior is itself only one of the necessary elements in the broader shift from volume-based to value-based care. My hope is that sharing what we did at Geisinger will help other institutions in their own efforts, tailored to their own needs and their own patient populations.