Academic medical centers have traditionally supported primary care in order to meet the comprehensive health care needs of patients from surrounding communities, as well as for its pedagogic value. However, with the spread of managed care, the delivery of primary care at academic centers has evolved significantly over the past decade, as has the need for a comprehensive salary incentive system that promotes successful primary care practice in a dynamic payer environment.
Primary care clinics at academic medical centers began by treating the poor and offered mostly free care, provided by part-time attending physicians and residents. As academic general internal medicine developed as a discipline, formal division status was achieved at many centers. More recently, the proliferation of managed care has given rise to new practices of full-time primary care internists employed by academic centers, both on campus and in nearby communities. Many of these physicians focus nearly exclusively on patient care, rather than on research or teaching. Devising a uniform and comprehensive system of compensation for all primary care physicians (PCPs) at academic medical centers is desirable from the perspectives of equity and the achievement of divisional objectives.
The influence of managed care has magnified fiscal and management pressures. Academic centers may invest in maintaining successful primary care networks in order to participate in large managed care capitation risk contracts, where the provider organization is “at risk” for the medical expenses of patients. The spectrum of full risk applies to all inpatient, outpatient, and pharmacy expenses for a defined panel of patients. The role of “gatekeeper” has been delegated to the PCP. However, in spite of the proliferation of capitation in many areas, much primary care is still paid for on a fee-for-service basis. Hence, compensation incentive systems for salaried PCPs must accommodate this mixed-payer environment, where clinical revenue to the institution is split between fee-for-service and capitation payment.1
These financial imperatives have only been heightened by the impact of the Balanced Budget Act of 1997.2 Many academic centers are under substantial pressure to make all lines of business, including primary care, financially self-sufficient. Opportunities for cross-subsidies from inpatient revenue are rapidly disappearing. To achieve economic stability it is important for PCPs to have a thoughtfully conceived set of incentives that remains consistent with the mission of the academic center and helps to motivate and reward professional behaviors needed for success.
The goals of simultaneously increasing productivity while optimizing resource utilization were central to the development of our academic PCP compensation system at Brigham and Women's Hospital, a founding member of Partners Health Care System, an affiliate of Harvard Medical School. Complicating the task at our academic referral center is the challenge of adjusting measures of performance for the impact of seriously ill patients on productivity and resource utilization. We also recognize the need to reward quality of care, teaching, and institutional seniority.3 In order to promote these organizational objectives we structured a productivity-based salary system with accompanying bonus incentives, which was implemented in 1999. In the rest of this article we describe and discuss the first year of this system's implementation.
Historically many academic centers have relied on compensation systems for PCPs based on guaranteed salary with defined increases for seniority. These systems derive from the tenure model, where institutional seniority and scholarship are highly valued. In such settings, compensation is conveyed partially in the form of academic promotion and job security. However, these systems may face inherent limitations due to tenure and promotion criteria that may under-value clinical work and fail to adequately reward clinical productivity, resulting in varied clinical revenue generation. This model of compensation has grown increasingly out of step with traditional community-based practice, particularly at a time when academic primary care is expected to become financially self-sufficient. Even in the community setting, productive internists are having difficulty maintaining prior income levels.4 The legal ramifications of altering salaries of tenured faculty may cause further complexity.5
Compensation systems on the other end of the spectrum, based solely on patient-visit volume, may establish a bias toward visit “churning” and an incentive to care for healthier patients who require only brief visits.6 We chose to pursue a more balanced option, basing salaries on a volume scale that incorporates the relative work component of clinical visits as the standard measure of productivity.
The resource-based relative-value scale (RBRVS), developed by Hsiao in 1986, assigns a standard measure of work performed to each type of clinical encounter.7 Visit complexity depends on a variety of objective and subjective criteria, as described by Medicare. For example, a moderate-complexity visit is assigned more productivity units under this system than is a low-complexity visit. These units are standardized across all types of physicians' services within a given specialty. Measuring work in this way takes into account the time, clinical intensity, and complexity of decision making required to provide clinical services.
In our model, salaries are based on actual performance that is compared with a uniform annual performance target of work relative-value units (wRVUs). These wRVU targets, based on the relative-value weights established by the Health Care Financing Administration for evaluation and management coding, provide clear incentives to be as productive as possible, without penalizing physicians for seeing more complex and time-consuming patients. The assignment of relative-value units is based primarily on patient contact, though the criteria also take into account review of outside records and testing. A baseline salary (e.g., $130,000/year) is established and then adjusted annually based on actual clinical productivity compared with target wRVU goals. The baseline salary level at the time of implementation of the new salary plan was set to be equivalent to the existing average compensation for all of our employed PCPs. Adjustment for visit complexity provided by the relative-value system is particularly important to our academic PCPs, who are likely to see patients who have highly complex multi-system illnesses. Relative work-unit measurement is consistent with actual revenues received from many payers.
Annual adjustments are made in salary and target RVUs based on the prior year's performance. For example, if the target is 3,000 wRVU/year and 2,700 are performed in year one (10% below target), then year two salary and target wRVUs will be reduced by 10% (a reduction of $13,000, to a salary of $117,000). During year two, if actual wRVUs performed exceed the new target, then quarterly supplemental payments are made based on the number of wRVUs performed that exceed the prorated target. For year three, the projected annual salary would be increased based on the “excess” RVUs performed in year two. Hence, rewards for increased productivity are paid both quarterly and with annual adjustments of salary. Salary decreases for under-performance are limited to annual adjustments. More frequent decreases in salary were felt to be both operationally and organizationally unacceptable.
All wRVUs performed in both ambulatory and inpatient settings are included in a PCP's total. Some academic centers have introduced inpatient-rounding systems where hospitalist physicians care for medical inpatients. Under these systems PCPs may be relieved of the responsibility of caring for hospitalized patients, enhancing time available for outpatient medicine. The benefit of the hospitalist in terms of ambulatory clinic productivity may, however, be partially offset by lost inpatient productivity, with a reduction in inpatient wRVUs.
Productivity-based salary was supplemented with a series of bonus criteria focused on organizational goals of medical management, quality of care, teaching, and seniority. The magnitude of total bonus payments is based on a percentage of the wRVU productivity-derived salary, to a maximum bonus of 10%.
Academic medical centers that employ PCPs who participate in capitated managed care contracts bear financial risk for patient resource utilization. Our academic PCP compensation model incorporates bonus incentives based on resource utilization for our panel of 19,142 managed care patients, divided between two separate full-risk capitation contracts. Our hospital and physicians participate as part of the large integrated delivery network mentioned earlier, Partners Community HealthCare, Inc. (PCHI). PCP performance measures for the capitated-patient panel include total medical expenses as well as laboratory and radiology utilization.
Our employed PCPs are divided into several groups ranging from five to 15 physicians. We evaluate medical management performance at the PCP group level rather than the individual PCP level. Group-level measurement minimizes the influence of sporadic high-cost cases and helps enhance actuarial stability, especially for PCPs whose patient panels are small. Most of our medical management programs and meetings are based on coordinated, group-level efforts. If a group of PCPs is successful in caring for their patients cost-effectively, then bonus payments (e.g., 5% of productivity-derived salary) are made to all PCP group members.
Providing bonus incentives based on medical-management performance is complicated by the issue of case mix. Capitation budgets set by insurers are seldom adjusted for patient illness level. Therefore, it is possible for PCPs who have sicker patients to have a more difficult task of managing care under their predetermined budget. This is a particular problem at academic medical centers, where overall severity of illness can be substantial, yet distributed unevenly between PCP groups and individual PCPs.
Addressing the issue of severity of illness for internal performance measurement requires a system of health-status adjustment of medical expenses.8. Unfortunately, the best of these systems can retrospectively explain only 50% of the variance in the expenses of health care claims. Nonetheless, when calculating our PCP groups' utilization performances for the bonus incentive, we chose to internally risk-adjust total medical expenses using Ambulatory Care Group (ACG) scores.9,10,11,12,13,14,15,16 The ACG system assigns each patient to one of 52 diagnosis groups, based on medical claims coding. Claims data are essential for performing these calculations and are obtained from our insurers as part of our capitation agreements. ACG groups are then ranked according to the intensities of resources required to care for patients with the medical problems that define the groups. For example, a PCP group that has relatively healthy patients will have its actual utilization expense adjusted upward, while a PCP group with sicker patients will have its actual expense decreased by such an adjustment. PCPs can then be evaluated on a more level playing field, by comparison of “risk-adjusted” utilization performance.17 While this is a small component of overall pay, it is consistent with the small volume of fully capitated patients at our institution in general.
Quality of care
Overall quality of care is another important goal but is difficult to measure in an objective and reliable manner. However, we can measure performance of health maintenance tests, such as mammography and Pap tests, relative to such performance carried out by institutional peers. These measures are highly valued by managed care payers in our area. Patient satisfaction data and board certification status are also considered to be proxies of quality. A defined percentage of productivity-derived salary (e.g., 2-3%) is offered as a bonus for achieving these measures of quality.
The teaching role is a core function of the academic medical center physician. Some centers have developed formal payment schemes based on the relative valuation of different teaching activities.18 Clinical productivity is clearly decreased for physicians who teach students in the ambulatory setting.19 Our compensation system rewards those who teach as a significant part of their professional activities by assigning additional RVU credits for teaching. For example, for each four-hour clinical session that a PCP is accompanied by a third-year medical student, credit for three additional RVUs is added to the PCP's annual total. This credit is intended to allow PCPs to see fewer patients and have more time for high-quality teaching. Funds are available from our medical school to subsidize the cost of assigning these teaching RVUs. Unfortunately, in the current environment many insurers are reluctant to fund teaching activities.
We also decided to reward seniority as a component of the bonus incentive. An explicit payment to PCPs based on seniority (e.g., $1,000 per year of seniority at the institution, after the first six years, to a maximum of $16,000 per physician) is included in the compensation calculation. This payment is then adjusted based on productivity performance compared with target. At the other end of the spectrum, PCPs who have recently finished training have salary-guarantee protection for up to two years to allow time for practice development.
In addition to these bonus measures, each PCP group director has a small discretionary payment to PCPs (1% of productivity-derived salary) based on individual PCP's commitment to group activities (e.g., arranging continuing medical education for the group, leading medical management meetings). Discretionary payment by the group leader, as opposed to the division head, was adopted to decentralize management efforts and extend incentives for practice-group—level activities. Table 1 presents a brief compilation of the components of our plan, as discussed above.
IMPACT OF THE NEW SYSTEM
In the wRVU-based salary system described above, we measured pre- and post-incentive productivity in terms of wRVUs and patient visits. PCPs who significantly changed their clinical-session time commitments between the pre-and post-incentive measurement periods were excluded from this analysis. Similarly, we excluded any new PCPs who did not have baseline data, and whose practice productivity might be expected to grow on the basis of routine practice development, regardless of other incentives. After these exclusions, we evaluated the performances of 64 academic PCPs, the durations of whose time spent each week in ambulatory patient care ranged from one to eight clinical sessions (each lasting four hours).
During the two years prior to introduction of the new salary incentive plan we maintained measurements of PCP productivity. Our productivity performance prior to the new salary plan was well below national averages for PCPs, including hospital-based academic physicians. Despite limited management interventions (e.g., verbally encouraging PCPs to schedule more patients), overall productivity was essentially unchanged over the two-year period prior to implementation of our incentive program.
After implementation of the new salary plan, overall visit volume increased by 17% in the first 12 months, compared with the same 12-month period during the previous year (see Table 2). Overall, wRVU volume increased by 20%. We observed these increases for all categories of PCPs: full-time (eight clinical sessions per week) and part-time (fewer than eight sessions per week) physicians, on-campus and off-campus PCPs, and those with more or less than six years of clinical experience (see Table 2). Even for PCP groups that had relatively “mature” practices, we observed significant increases in overall productivity.
The impact of ACG adjustment on total medical expense data for bonus payments was significant. For example, in Group E, the unadjusted total medical expense performance was $186.58 per member per month (pmpm)—see Table 3. However, this group had a particularly high number of seriously ill patients, with an ACG score 19% above average. The ACG risk-adjusted utilization performance became $157.18 pmpm, below our target budget for bonus payment set at $165.00 pmpm. As a result, a medical management bonus payment was made to all group members. Meanwhile, there were other groups, such as Group B, that had low ACG scores, due to relatively healthier patients. This group had an increase in expenditures when adjusted by ACG score (from $164.88 pmpm to $177.11 pmpm), resulting in no medical management bonus payment. Since the ACG scores are relative to the scores of other PCP group panels in our academic center, there is no major net effect from adjusting performance in this manner—some groups improve and some worsen (see Table 3).
In terms of managed care capitation performance, we have had mixed results after one year under the new salary plan. In one of our plans (10,007 patients) we experienced a 1.1% decline in total medical expenses. The rest of our integrated delivery system (including another academic medical center and several community-based primary care groups and hospitals) experienced an overall 1.5% increase in total medical expenses in this contract. In our other capitated health plan (9,135 patients), total medical expenses increased by 12.3%, while expenses in our entire network increased by 6.6%.
We have observed no clear relationship in our risk-patient panels between PCP productivity and total medical expenses. Some of our PCP groups with high annual PCP visit rates (per patient per year) have low total medical expenses. Even when risk-adjusted by ACG score, there is still no clear relationship between total medical expenses and PCP visit frequency.
Overall improvement of performance and reporting of health screening tests has been observed since we began distributing these data in anticipation of full implementation of the new compensation plan. Performance was measured at the individual PCP and group levels, based on data from our electronic medical record. All clinically eligible patients were included, regardless of payer. For example, our mammography rate, based on group average performance, improved steadily from 66% to 76% compliance with screening guidelines during the years preceding and during full implementation of the new salary plan. A similar improvement was observed for Pap test rates.
REFLECTIONS ON OUR EXPERIENCE SO FAR
As described earlier, the comprehensive compensation plan we have developed for employed academic medical center PCPs incorporates criteria based on our organizational goals, with principal focus on productivity, as measured by work relative-value units. The strategy is designed to encourage commitment to clinical practice while not penalizing physicians for treating complex patients whose cases may require greater physician time and effort. Bonus payments for cost-effective medical management (based on risk-adjusted resource utilization) and explicit incentives for quality of care, teaching, and institutional seniority are important components of the plan (see Table 1). A small incentive payment is reserved for the PCP group leader's discretion, based on individual physicians' participation in group-oriented activities. The magnitudes of actual bonus payments are calculated as a percentage of productivity-derived salary, since medical management and quality accomplishments require more effort for physicians who are also highly productive.
We faced significant concerns from several members of our physician staff prior to implementation of the plan; they felt that our PCPs were already performing at their maximum clinical capacity. However, comparison with national bench-mark data of primary care groups suggested otherwise.20 Since implementation of this system, we have seen a significant increase in clinical productivity. After stratifying PCPs by site of practice, seniority, and clinical time commitment, we observed substantial increases in all cohorts, especially “full-time” clinicians.
In the aggregate, physicians' salaries have not decreased under this plan, but rather have increased slightly, as has our overall clinical revenue. We are unaware of any PCPs who have left our medical center for employment elsewhere based solely on their reactions to the implementation of this system, though this is difficult to determine with certainty. Patient-satisfaction survey measures have revealed improvement or have been unchanged since introduction of this system.
It is interesting that the overall growth in wRVUs (our chosen metric of productivity) increased slightly out of proportion to visit volume. We have an active coding—education and coding—auditing program to ensure that evaluation and management coding is reliable and accurate. Changes in wRVU totals based on compliance audits are incorporated in salary calculations.
We believe that medical management incentives have helped us to maintain an early focus on important managed care issues that might otherwise have garnered less attention. Our physician hospital organization has been able to implement an elaborate medical management plan that includes weekly PCP group medical management meetings, provider profiling, and active case management. While many academic provider groups have faced steadily rising total medical expenses, in at least one of our large capitation panels we have managed to decrease total medical expenses. In the other contract, we believe a small group of outlier cases and rising obstetrical frequency, both out of PCP control, were responsible for rising medical expenses.
The mixed-payer environment mandates that we establish concurrent incentives to increase productivity while also rewarding effective medical management. We have not observed a correlation between frequency of PCP visits and total medical expenses. In our experience, PCP professional fees constitute only a small part of our total medical expenses (approximately 5-8%). Although PCPs are themselves a small capitation “cost center,” they do control other cost drivers such as high-cost tests and specialty referrals. Frequent PCP visits may help replace care by more costly specialist providers and lower total medical expenses. Since we provide group-level incentives for efficient medical management performance, we chose not to compensate our PCPs based on capitation panel size alone.
Incentives for cost-effective medical management apply equally to all PCPs, even those who have only small fractions of capitation patients in their practices. Since physicians within the same PCP group often cover for each other, overall performance of the entire primary care group seems the best measure. This also greatly diminishes the impact of any single patient or physician decision. We believe an early focus on medical management, while the market share of full-risk plans is low, will be valuable as our insurers consider greater provider risk-sharing.
Improvement in the performance and recording of health screening measures was observed prior to the full implementation of our incentive plan. The collection and distribution of Health Employer Data Information Set (HEDIS) criteria data to our PCPs was initiated several years prior to the clinical productivity incentives and salary changes, as part of the phased introduction of the overall plan. No actual impact on salary occurred until the entire plan was fully implemented. However, health screening performance data and patient out-of-compliance lists were distributed to PCPs annually. We believe that distribution of these data led to a heightened awareness of the measures that were ultimately incorporated into our salary bonus calculations. While confounding factors may have contributed to the observed improvement (e.g., focus on HEDIS measures by local health maintenance organizations), its occurrence during the development and implementation of our compensation plan (and sharing of performance data) is suggestive evidence of the impact of quality incentives.
Incentives for teaching are a critical aspect of the plan, and distinguish it from incentive arrangements that exist in the non-academic setting. It is important not to penalize physicians who teach as part of their careers if we are to retain excellent teachers among our ranks. Similarly, retention of experienced, seasoned clinicians is the rationale behind the seniority bonus, which is designed to give an explicit reward to physicians who have made a career commitment to our institution. These staff help provide program and departmental stability as well as mentoring for more junior faculty. Some of the more senior staff, covered by previous salary plans for the longest period of time, faced the greatest potential reductions under our new incentive plan. By rewarding seniority we have mitigated some of these changes.
The significant increase in PCP productivity that we observed is critical if financial independence and self-sufficiency are to be achieved by primary care at our institution. In future years we anticipate gradually increasing wRVU targets, while softening the impact of the increases with increased payment per wRVU (in coordination with improving payer contracts). We also intend to regularly reevaluate our participation in capitation contracts and provider risk sharing, a source of significant loss for many academic centers. Regarding quality measures, we anticipate including more sophisticated criteria in the future, such as treatment of diabetes and counseling of smokers. We will maintain bonus payments for institutional seniority and group-level activities as valuable components of the plan.
We feel confident that our complementary salary-incentive program based on clinical productivity and bonus measures will continue to allow us to address the complex goals of academic primary care in the mixed-payer environment. It has proved its value in providing incentives for academic primary care physicians to achieve under burgeoning demands for revenue self-sufficiency, managed care performance, quality of care, and academic commitment. We hope that institutions similar to ours will find the preceding description of our program useful.