Specific steps can help hospitals effectively engage physicians in initiatives to reduce clinical variation and, in turn, improve the value of care delivery.
In many U.S. markets, the shift toward value-based health care means that efforts to boost volume must take a back seat to improving quality and reducing costs. Efforts to improve value through clinical variation reduction, however, will not be successful without changes in physician behavior. The prospect of such change will likely generate physician resistance—and overcoming it will require effective leadership.
Hospital leadership must listen to physicians and enlist their involvement in identifying variation reduction opportunities, measuring performance, and improving value. Other key hospital stakeholders, including staff in case management,
quality, health information, nursing, and operations should also be engaged in the process and be given clear roles and accountabilities in support of improvement strategies.
Physician Engagement Strategies
Physician and executive leadership must formally and informally engage the medical staff to communicate the expectations, goals, measures, and accountability for value improvement efforts. This outreach may include, for example, participation in design teams to review data and understand current practice variation for high-volume DRGs or service lines. Physician leaders must partner with staff in a way that is supportive and collaborative and that rewards those who are engaged and achieve the required outcomes.
We have found the following best practices for engaging physicians in these initiatives.
Utilize scorecards and dashboards. Physicians tend not to change their behavior if desired changes are not being measured and reported. Properly designed and communicated scorecards demonstrate not only what is expected to change but also how important the change is to clinical and hospital leadership. Scorecards ultimately foster peer comparisons that encourage transparency, friendly competition, and improvement.
Effective scorecards show individual metrics and overall results based on weighted scoring, by physician, with an option for names to be blinded. These scorecards allow for peer comparison, specifically an understanding of where each physician stands against peers who treat similar patients. Physicians are notably competitive, and blinded public displays of accomplishments, which they can compare against their own personalized scorecards, foster friendly competition among colleagues and account for much of the immediate improvement that is typically seen in the first few months.
We have often found that physicians respond well to simply knowing where they stand in comparison to their colleagues. We audited adult primary care physicians regarding the frequency with which they ordered an MRI or CT scan during a three-month period, and plotted their results on a bell curve (blinded except to their own place on the curve). Their comparative standing shocked the few physicians who ordered these imaging tests more than two standard deviations more frequently than the mean.
Scorecards should have no more than six to eight metrics that are tailored to the desired changes of each specialty or physician group, and should be presented at intervals that are long enough to allow for more than 30 data points (N > 30) to be reported. Smaller sample sizes do not provide statistically significant reports and lead to counterproductive “sawtooth” performance charts.
Typical scorecard metrics include:
- Average length of stay and excess days per discharge
- Readmission rates
- Time gaps in orders
- Numbers of procedures unrelated to the diagnosis
- Clinical documentation improvement (CDI) query response times
- Adjusted mortality
When scorecards utilize objective and transparent metrics that are determined in collaboration with, and approved by, both physicians and executive leadership, they quickly come to be viewed as valuable assessments of performance. Affected physicians should have input into the metrics selected, with appropriate departmental and medical executive approval.
Depending on leadership preference and organizational culture, scorecards can be blinded or unblinded. We strongly urge systems to share feedback in a blinded manner for only about three months, with the stated intention of becoming more transparent thereafter. This approach allows physicians to help correct and refine the early reports and to make some easy course corrections before their performance is made known to their colleagues.
Scorecards are a critical element of any clinical variation reduction program. They allow the medical staff and other impacted clinicians to evaluate their efforts relative to goals, track trends, and identify opportunities for improvement.
Ensure accurate, acuity-adjusted data and analytics. Physicians are data-driven; it is in their DNA as scientists. As part of efforts to reduce clinical variation, they must be presented with reliable, accurate, and acuity-adjusted data by DRG or clinical cohort. Multiple DRGs from the most common Major Diagnostic Categories should be trended over at least a one- to two-year period. Clinical and financial outcomes should also be trended over a minimum of one to two years to gain an accurate representation of the value that the hospital and its physicians have been delivering to patients and the community.
Most physicians prefer to be benchmarked to their own internal best practices as opposed to some external standard. Therefore, another characteristic of effective analytics is the ability to allow physicians to compare their own best outcomes to their less desirable outcomes. Clinical variation analytics should also demonstrate the link between physician behavior and reduced variation. In other words, physicians should be able to control the data elements by changing their clinical practice patterns.
In our experience, three types of variation can be improved through physician behavior change:
- Clinical resource consumption
- Length of stay (LOS)
- Clinical documentation
Reductions in clinical resource consumption and LOS typically result in improved quality outcomes and lower costs. Improvement in clinical documentation ensures that publicly reported quality metrics are accurate and has the added benefit of potentially increasing revenue.
The data should be sufficiently granular to show each physician the amount of variation within each hospital service line (e.g. Lab, Pharmacy, Radiology, etc.). Moreover, physicians should be able to compare their individual variation to that of colleagues who treat the same patient types and severities. And the data should clearly show the cause of the variation in clinical practice patterns. For example, orders unrelated to the principal diagnosis, over- and underutilization, and clinical documentation issues should all be clearly visible in the analytics that are presented. Once causes and sources of variation are identified through effective analytics, the difficult work of actually reducing variation begins.
It is useful to mention that there are benefits to sharing with providers a) the definitions of the metrics and b) the attribution assumptions that are used to assign a case to a physician. When leaders initially share quality or utilization reports with physicians, they should not treat the reports as absolute truth—and in no case should they judge performance before the reports have been validated. The first reports are often flawed. Leaders should approach physicians tentatively with early reports and ask them to help validate the results. As the physicians de-bug the reports, their confidence in the accuracy of the reports increases.
It is crucial that clinical variation analytics be risk-adjusted and physician-specific. Risk-adjusted analytics allow physicians and other clinicians to understand the severity of illness of each practitioner’s patient population. This information tends to reduce resistance and enhance buy-in among physicians.
Align financial incentives. When possible, one of the best ways for health systems to engage physicians is to align the financial incentives of the physicians with those of the hospital. It is certainly difficult to engage physicians when their financial incentives are opposed to those of the hospital.
One real-life example of an incentive mismatch is the push by hospitals and health systems for orthopedists to use only one or two prosthetic joint vendors and select prostheses that reduce the total cost of care. A less expensive prosthesis may require steps that add time to the operation and thus potentially affect payments for surgeons, given that they receive a professional fee for each procedure performed. A prosthesis that saves money for the hospital could limit a surgeon to performing three joint replacements per day rather than four or five.
Aligning physician financial incentives is tricky due to the Stark Law and regulations against inurement. Hospitals cannot pay for referrals and are restricted in providing financial incentives for physicians to work at their facilities. Therefore, incentive alignment must be done within explicit exceptions to the Stark Law or through a mechanism such as a clinically integrated network (CIN), a commercial or government shared savings program, or a co-management agreement. CINs, operating within Federal Trade Commission guidelines, can redistribute savings that accrue from utilization management and quality improvement activities.
Of course, the increasing share of physicians who are employed by hospitals can participate in the hospital’s employee incentive program. Finally, properly structured professional service agreements can include incentives for meeting certain performance targets.
Coming next month: Additional strategies for engaging physicians in efforts to reduce clinical variation.