Coleman, David L. MD; Moran, Eileen; Serfilippi, Delchi; Mulinski, Paul PhD; Rosenthal, Ronnie MD; Gordon, Bruce; Mogielnicki, R. Peter MD
The productivity of clinicians has come under increasing scrutiny as health care organizations critically examine their efficiency under the constraints of prearranged, contractual care.1–3 Models are needed that promote efficient, high-quality care and equitably capture the breadth of physicians' clinical and administrative activities. The complexity and diversity of physicians' responsibilities in academic health centers such as VA Medical Centers have made it difficult to develop comprehensive productivity models. Indeed, many academic physicians have been discouraged by the failure of existing models to capture the full range of their activities.
The Veterans Health Administration (VA) is the largest health maintenance organization in the United States. Its mission is to provide clinical care for eligible veterans, educate trainees, conduct research, and provide backup to the Department of Defense in the event of a national emergency. VA Medical Centers are affiliated with 107 medical schools, and the VA supports 10% of all graduate medical education in the United States.4 Half of the third- and fourth-year medical students in the United States take clerkships at VAs, and 28,000 residents annually receive all or part of their training in the VA.4
The VA's critical academic and clinical roles warrant a particularly comprehensive approach to the measurement of the productivity of its physician–staff. The Institute of Medicine's report on physician staffing in the VA emphasized the importance of incorporating measures of clinical care, administration, research, and education.5 Moreover, resources are distributed in the VA using the Veterans Equitable Resource Allocation (VERA) model that includes credit for teaching and research, in addition to clinical work.6
The substantial increase in enrollment of eligible veterans in the VA (a 47% increase in patients from 1996 to 2001) and budgetary allocations that do not keep pace with increasing costs of health care have created a compelling need to use personnel resources as efficiently as possible.7,8 The support of physicians' salaries using a strategy that promotes clinical care, administration, medical education, and research in a financially constrained environment is a formidable challenge for the VA and most academic health systems. Therefore, the development of meaningful measures of physicians' productivity in the VA can also serve as a useful model for other health care organizations.
Despite the imperative to develop a model for measuring physicians' productivity in the VA, it has not been undertaken in a manner that consistently reflects both the financial constraints and the breadth of clinical and academic missions of the VA. The development of a meaningful productivity measure and the successful application of any findings require that physicians and institutions agree on a set of basic guiding principles.11,12 The measurement of physicians' productivity has potential to be divisive, particularly if physicians' salaries and opportunities for advancement are affected.11,13 In view of the compelling need to measure physicians' productivity at VA Medical Centers (summarized in List 1), we developed a model productivity measure at the VA Connecticut Healthcare System, a tertiary care referral center affiliated with the Yale School of Medicine and the University of Connecticut School of Medicine. In designing the measure, we used the principles of mission-based management recommended by the American Association of Medical Colleges to measure physicians' productivity in a way that captures clinical, administrative, research, and educational activities.9
THE PHYSICIAN-PRODUCTIVITY MODEL
The model directly measured two domains of physician productivity: clinical care and research. Physicians' productivity in education was derived from the indirect funds for resident positions, and their administrative activities were calculated as a percentage of clinical work.
Measurement of Physicians' Clinical Productivity
The aggregate clinical workload for a clinical unit was obtained, where possible, through the Text Integrated Utilities clinical documents database that interfaces with the VA's computerized patient record system. Current Procedural Terminology (CPT)–coded procedural clinical encounters were extracted from the Veteran's Health Information System Technology Architecture (VISTA) using the Ambulatory Care Reporting Menu options. Operating room (OR) and non-OR CPT-coded procedural workloads were similarly extracted using the Surgical Package menu options available in VISTA.
Inpatient and outpatient encounters that lacked an easily assignable CPT code were derived using a set of standardized assumptions. An “average” evaluation and management CPT code was assigned to each note's title based on the usual level of documentation and clinical work. For example, all medical attending admission notes were assigned CPT code 99223 and follow-up notes were assigned CPT code 99232. Progress notes related to surgical procedures were not credited separately, rather they were included as part of the global CPT code for surgical procedures. Examples of electronic note titles for the pulmonary and critical care section of the medical service are shown in Table 1. The use of note titles to measure clinical productivity required standardizing electronic note titles and educating the clinical staff concerning proper use of the titles.
Each CPT code was converted to a relative value unit (RVU) using the National Physician Fee Schedule Relative Value File (〈http://www.hcfa.gov/stats/cpt/rvudown/htm〉). We converted CPT codes to RVUs and then adjusted the RVUs using the applicable Geographic Practice Cost Index (GPCI). An RVU contains three components: physician work (MD), practice expense (PE), and malpractice expense (MP). Because our goal was to capture physicians' productivity, only the MD component of the RVU was credited. The MD RVUs for an individual clinician, a clinical unit, or a service were totaled and then converted to total clinical dollars “earned” using the 2002 Medicare Fee Schedule conversion factor of $36.20/RVU.
Because the VA is unable to provide comparable salaries for physicians who perform some specialized clinical services, the VA contracts with Yale University School of Medicine to provide a broad range of specialized clinical services. The productivity model was used to establish these contracts with Yale staff or departments for clinical services such as specialized procedures in cardiothoracic surgery, orthopedics, urology, and radiology. When establishing contracts for clinical services, dollar credit for the MD RVU was adjusted according to the prevailing reimbursement rate for the clinical services from third-party payers at Yale University School of Medicine. For example, the Department of Surgery collects an average of 134% of the Medicare Allowable Rate from all third-party payers for cardiothoracic services. Therefore, we adjusted the cardiothoracic surgery contract by 134% of the Medicare Allowable Rate ($36.20/RVU). Moreover, we decided to include the malpractice component of the RVU in contracts because the contracting agency provides its own malpractice coverage.
Measurement of Physicians' Administrative Productivity
We reasoned that physicians' administrative activities are roughly proportional to the quantity of clinical activity in a service or other clinical unit. Therefore, credit for administrative activities was distributed in proportion to the clinical RVUs earned. However, because Services have fixed administrative responsibilities regardless of their sizes, we designed a graduated formula for assigning administrative credit. The formula gives an administrative credit of 15% of clinical work for the first three MD FTEEs (full-time equivalent employees) earned and 10% of clinical work for anything beyond three MD FTEEs. Therefore, for the first 14,817 RVUs earned (the equivalent of three average physician salaries when converted to dollars using the 2002 Medicare Fee Schedule), a supplement of 15% of the earned RVUs was distributed to each service to support administrative activities. Ten percent of any clinical RVUs earned above 14,817 was also distributed to each service.
We based the credit for administrative activity (10% or 15% of clinical work) on the physician-time distribution in the Cost Distribution Report (CDR). Although we did not assign values for specific administrative functions to individual physicians, we developed a model to credit individuals (see Table 2). The credit is assigned on a “point system” to compare administrative activities among physicians in a service according to the following formula:
Measurement of Educational Productivity
Equation (Uncited)Image Tools
In the fiscal year 2002, the VA provided $41,781 in indirect support for each resident position funded by the VA. The CDR was again used to determine the average amount of physicians' time spent in support of education and training as a percentage of all indirect educational support. We determined that 49.6% of the indirect costs for education were attributable to physicians' salaries. Therefore, we credited each service with $20,723 towards physicians' salaries for each VA-funded resident or fellow trained in the service. To assign credit to individual clinicians for educational activity, we developed a point system to credit individual educational activities (see Table 2). As with the scheme for crediting administrative activities, the point system for educational activities credited physician–staff in an individual service by using the following formula:
Measurement of Research Activity
Equation (Uncited)Image Tools
We used the VERA Research Support Accounting Team's formula for crediting research productivity to each service. As shown in Table 3, this formula credits research grants and other research activities to services or individuals by allocating fractions of FTEEs for each activity. These fractions of physicians' salaries correspond to the funding provided under VERA to each VISN for support of the salaries of physicians engaged in research. Allowance was also made for activities such as research committee membership, mentoring trainees, and protected time for faculty in the early phases of their careers or following interruption of grant support. A supplement of 10% of the total research credit was included in the model to support protected time for junior research faculty or other faculty for limited periods of time.
EXAMPLES OF APPLICATION OF THE PHYSICIANS' PRODUCTIVITY SURVEY
A sample physicians' productivity survey of the pulmonary and critical care section is shown in Table 4. To determine how many physician FTEEs were earned by the pulmonary and critical care section in this model, we converted the GPCI-adjusted RVUs to dollars by assuming reimbursement at 100% of the Medicare Fee Schedule and a GPCI in Connecticut of 1.05. Credits for administration, education, and research activities were allocated as described above. The average salary plus fringe benefit of our physician–staff in pulmonary and critical care is $146,615 per year. Using this annual salary figure, the sum of the pulmonary and critical care section's activities earn the equivalent of 3.564 physician FTEEs.
The same model was used to establish a contract for cardiothoracic services with the Department of Surgery at Yale University School of Medicine. The average reimbursement for all cardiothoracic procedures at Yale is 134% of the Medicare allowable rate. Therefore, we modified the conversion of GPCI-adjusted RVUs to dollars by using 134% of the Medicare Allowable Rate. Because the cardiothoracic attending physicians were under contract, VA policy requires the university to provide malpractice coverage. Therefore, we used both the MD and MP components of the RVU to calculate physicians' productivity. Workload from the prior 12 months was used to estimate clinical workload for the period of the contract. The workload was converted to dollars using the formula:
We added $28,560 (0.10 × 7,890 RVUs × $36.20 per RVU) for administrative activities and $47,360 for research credit (primary investigator on two “other national peer-reviewed research”) as well as $41,446 for educational credit (two cardiothoracic surgery residents × $20,723/resident). The compilation of the contract for clinical, research, education, and administrative activities in cardiothoracic surgery is shown in Table 5.
ADVANTAGES OF THE PHYSICIANS' PRODUCTIVITY MODEL
The physicians' productivity model has several useful applications. First, we have not had a model for determining relative physician-staffing levels in the VA that reliably captures the breadth of activities expected of physicians in VA Medical Centers affiliated with medical schools. The quantification of teaching and research activities in measuring physicians' productivity is a critical feature of the model. Moreover, credit is assigned to these activities in direct proportion to the VA's financial investment in support of teaching and research.
Second, the documentation of clinical encounters and resident supervision is inconsistent among practitioners in the VA. Our productivity model was designed, in part, to create additional incentives to improve documentation of clinical care and supervision of trainees. Because the model credits only clinical encounters that are adequately described in the medical record, documentation of clinical care by practitioners has become substantially more consistent as they recognize that their encounters are being measured.
Third, the productivity model has been very useful in formulating contracts for specialized clinical services that more reliably protect the interests of the VA and the affiliate. The VA benefits from using a reimbursement model that provides incentives for care of veterans and through the stipulation in the contracts that compensation will be provided only when standards of documentation are fulfilled. The affiliate may benefit by receiving more realistic, market-driven reimbursement for specialized services.
Fourth, the model has provided useful information to aid in decisions about the distribution of salary support among sections or services. For example, we have transferred resources from one service that was “underperforming” to another service that was achieving a very high level of productivity. In addition to staffing adjustments, we have used the model to make more informed decisions concerning replacement or supplementation of physician–staff.
Fifth, administrative activities by clinicians are vital to providing effective and efficient clinical care. Examples of these activities include peer review, continuing education, coordination of care, quality assurance, supervision and training of employees, budgetary review and preparation, outreach, credentialing and privileging, and fulfillment of regulatory requirements. Inclusion of administrative activities in our productivity model has provided an additional incentive and opportunity for physicians to participate in these activities.
Sixth, the VERA model distributes indirect funds for education according to the number of VA-funded trainees. Therefore, our model credits services or sections with educational productivity based solely on the number of resident or fellow positions paid by the VA. We aligned our model with reimbursement for education under VERA to assure that credit would be assigned in proportion to resources provided for this purpose. Although we have suggested a relatively simple method to credit individuals for educational productivity, U.S. medical schools use widely varying strategies to measure educational activities among faculty14
Finally, RVU-based models for crediting clinical workload are increasingly used in health care organizations.1,2,15,16 Although time-based models of clinical productivity have been used,17 we did not use a time-based model because it would not account for differences in efficiency among physicians, differences in practice among services or specialties, or different market values of physicians' activities, and would not directly measure the amount of work accomplished. Moreover, the RVU-based models appear to provide useful incentives in academic teaching hospitals. For example, an RVU-based incentive plan in an academic teaching hospital that serves indigent patients resulted in a 31% increase in productivity and 50% increase in collections.16 We also selected the RVU model to aid in benchmarking clinical workload to non-VA settings. Additionally, the RVU model provides a sound market-based mechanism for comparing clinical productivity among and within VA facilities.
POTENTIAL LIMITATIONS OF THE PHYSICIANS' PRODUCTIVITY MODEL
Although the RVU-based model for measuring physicians' productivity has several useful applications, it has some potential limitations. For example, although the method has the advantage that it is a standardized, frequently updated instrument, it may not fully recognize decision-making or cognitive activities. Therefore, the RVU-based model may replicate the finding in other health care organizations that procedural specialties such as cardiology, surgery, and gastroenterology are more “productive” than non-procedural specialties such as primary care, endocrinology, and rheumatology. Models that measure “encounter time” and “intensity of the encounter” may be useful alternatives, however, these parameters are difficult to measure on a broad scale in the VA system.18
Our approach to measuring clinical productivity is helpful in services such as internal medicine, surgery, dermatology, and neurology. The model is less useful, however, in pathology and anesthesiology. These services have found the research and education components of the model to be helpful, but have had considerably more difficulty in using the RVU-based method to assess clinical productivity. An alternative method for measuring clinical productivity in anesthesiology is available.19
Although availability of support services and case mix influence physicians' productivity,20 our model presumes that practices and procedures of individual physicians are important determinants of productivity. An illustrative study from a VA primary care setting examined the determinants of physicians' productivity.3 In this study of 2,721 outpatient visits to 56 physicians, productivity was defined as the number of patients seen per physician per hour. The productivities of individual physicians varied substantially. The variability between physicians was attributed to variations in the clinic characteristics, patient mix, and personal practices of individual physicians. Although the comparison of physicians' productivity among and within health care organizations should account for each of these factors, the personal practices of individual physicians are an important determinant of productivity.2,3,20
The desire of managed care organizations to limit resource use has led to the rewarding of physicians who use fewer services.20 Although successful in helping to reduce health care costs, these incentives have proven controversial among both providers and patients.21,22 Therefore, it is increasingly important that future measures of quality of care be incorporated into measures of physicians' productivity. The VA's electronic medical record facilitates extracting information that can be used to assess the quality of care. As shown in the examples in List 2, our model can be adjusted to account for quality of achievement in the four domains of clinical care, education, research, and administration.
Our effort to measure physicians' productivity in the VA has the potential to be a vital component of the process designed to ensure optimum use of resources, improve the quality of care, meet clinical and academic institutional goals, correct factors that adversely affect physicians' productivity, and recognize physicians who either need assistance to become more productive or deserve recognition for outstanding productivity. The relative values ascribed to the activities and accomplishments of physicians have a substantial influence on the character of the medical profession, the vitality of medical education and research, and the cost and quality of health care. Physicians in academic health care organizations such as the VA need to participate actively in this process to ensure continued improvements in medical care, education, and research.
1. Agasse P Jr. Physician productivity measurement, methodology and implementation. J Society Health Systems. 1996;5:41–9.
2. Anonymous. Harness contracting clout by tracking physician productivity. Capitation Management Report. 1997;4:194–7.
3. Smith DM, Martin DK, Langefeld CD, Miller ME, Freedman JA. Primary care physician productivity: the physician factor. J Gen Intern Med. 1995;10:495–503.
4. Office of Academic Affiliations, Department of Veterans Affairs, Washington DC, Government Printing Office, 2001.
5. Institute of Medicine. Physician Staffing for the VA. Volume I. Washington, DC: National Academy Press, 1991.
6. Veterans Equitable Resource Allocation. Equity of Funding and Access to Care across Networks, Department of Veterans Affairs. Washington, DC: Government Printing Office, March 1997.
7. Kizer KW. Health care, not hospitals: transforming the Veterans' Health Administration. In: Dauphinais GW, Price C (eds). Straight from the CEO: The World's Top Business Leaders Reveal Ideas that Every Manager Can Use. New York: Simon & Schuster, 1998:112–20.
8. VA Health Care. Allocation changes would better align resources with workload. General Accounting Office Report. Washington, DC: Government Printing Office, February 2002.
9. D'Alessandri RM, Albertsen P, Atkinson BF, et al. Measuring contributions to the clinical mission of medical schools and teaching hospitals. Acad Med. 2000;75:1232–7.
10. Gold M, Hurley R, Lake T, Ensor T, Berenson R. A national survey of the arrangements managed care plans make with physicians. N Engl J Med. 1995;333:1678–83.
11. Greenfield AR. Physician productivity: a managerial challenge. J Ambulat Care Manage. 1998;12:6–10.
12. Fisher R, Ury W, Patton B. Getting to Yes: Negotiating Agreement without Giving In. 2nd ed. New York: Penguin, 1992.
13. Salmon JW, White W, Feinglass J. The futures of physicians: agency and autonomy reconsidered. Theoretical Med. 1990;11:261–74.
14. Mallon WT, Jones RF. How do medical schools use measurement systems to track faculty activity and productivity in teaching? Acad Med. 2002;77:115–23.
15. Hickey M, Ichter JT. Promoting physician productivity through a variable compensation system. Healthcare Financial Management. 1997;51:38–40.
16. Stewart MG, Jones DB, Garson AT. An incentive plan for professional fee collections at an indigent-care teaching hospital. Acad Med. 2001;76:1094–99.
17. Howell LP, Hogarth M, Anders TF. Creating a mission-based reporting system at an academic health center. Acad Med. 2002;77:130–8.
18. Lasker RD, Marquis MS. The intensity of physicians' work in patient visits. Implications for the coding of patient evaluation and management services. N Engl J Med. 1999;341:337–41.
19. Abouleish AE, Zornow MH, Levy RS, Abate J, Prough DS. Measurement of individual clinical productivity in an academic anesthesiology department. Anesthesiology. 2000;93:1509–16.
20. Hurdle S, Pope GC. Improving physician productivity. J Ambulat Care Manage. 1989;12:11–26.
21. Kennedy KM, Wofford DA. Physician equity in health care delivery systems: three alternative models. J Health Care Finance. 1998;24:36–47.
22. Kuttner R. Must good HMOs go bad? The commercialization of prepaid group health care. N Engl J Med. 1998;338:1558–63.