INTRODUCTION

Ironically, at the very time definitive data are confirming primary care’s essential contributions to health care (i.e., health status is improved and costs are reduced),1,2 adult primary care in the United States finds itself on the brink of crisis.3 Practicing primary care physicians are demoralized, retiring early, and advising others not to go into the field.35 The percentage of recent U.S. medical school graduates and residents planning to enter primary care practice is plummeting to levels that will lead to serious physician shortages.611 The reasons for this decline are multifactorial,35,7,9,1123 but a central factor has been a succession of dysfunctional payment systems that discourage proper delivery of primary care.35,12,1618,2328 We propose a new payment model for primary care that realigns incentives and makes possible the establishment and operation of accountable, modern primary care practices capable of providing the personalized, coordinated, comprehensive care essential to a well-functioning health care system.

FEATURES OF THE MODEL

A risk/needs-adjusted comprehensive payment would be made to the primary care practice for the comprehensive care of each patient. It would replace all encounter-based payments made to the primary care physician under the resource-based relative-value scale (RBRVS) system.29 The payment would be directed to cover all practice expenses and salaries related to operating a robust, modern primary care practice (Table 1), one that would qualify as an “advanced medical home” for adults, a practice structure that enables efficient provision of comprehensive, coordinated, patient-centered care.3032 Included would be monies for essential infrastructures and systems, most importantly, an interoperable electronic health record with decision support.3337 Unlike primary care capitation systems, which also provided an aggregate payment (Table 2),38,39 our model’s comprehensive payment would represent a net investment in primary care practices, not just the actuarially determined consolidation of inadequate RBRVS visit payments, as typically occurred under capitation.24,25,38,39 Total practice revenue would markedly increase compared to that under RBRVS; over two-thirds would be designated for the teams and systems essential to improving care (Table 1). Physician payment would also increase, commensurate with the responsibility assumed and value created (Table 3).

Table 1 Sample allocation formula for comprehensive payment system for adult primary care practice*
Table 2 Comparison of Comprehensive Payment System with other Modes of Payment for Primary Care
Table 3 Examples of Possible Comprehensive Payments, Panel Sizes, and Allocations for Participating Adult Primary Care Practices*

To encourage quality, safety, efficiency, and patient-centered care, we propose that a substantial proportion of the comprehensive payment (e.g., 15–25%) be performance/outcomes-based and paid as a bonus for achieving valued outcomes. Determination of the performance bonus would require consensus goals and use of validated process and outcome measures agreed upon by payers and the profession (e.g., the Starter Set recommended by the National Committee on Quality Assurance).40

Both the comprehensive payment and the performance goals for the bonus would need to be risk- and needs-adjusted so as not to penalize practices for taking on the care of high-risk or psychosocially disadvantaged patients. The adjustments would depend on validated formulas, such as those using principal diagnoses in determining risk for ambulatory care4144 and those taking into account behaviors, psychosocial factors, and social environment to estimate need.45 These formulas would also allow for objective audit to ensure accuracy of the payment adjustment and discourage abuse. The expected manifold differences in actuarially determined care burdens would be reflected in correspondingly scaled payments; variation might be 10-fold or more between the highest and lowest risk/needs categories.

Payment would be made monthly to help smooth cash flow and enable patients to conveniently change primary care practices. A monthly payment schedule would also underscore the shift from piece-work reimbursement to 24/7 physician practice responsibility for patients. Patient copayments for primary care services could remain, but also might switch from per-visit payments to an actuarially determined cost-sharing component of the comprehensive payment, paid as part of the insurance premium.

The payment reform proposed requires concurrent practice transformation (i.e., establishment of the “advanced medical home”) to achieve the necessary savings and improvements in care. Participation in the comprehensive payment system would be dependent on demonstration of the requisite structural and organizational changes. Organizations such as The National Committee for Quality Assurance are developing standards and measures for office practices46 that might be used to determine eligibility for the comprehensive payment. Participating practices would be expected to agree to periodic audit of standards such as these.

Payment for hospital and specialist services and ancillaries such as medications, laboratory tests, and imaging studies would remain the responsibility of payers and not the practices (unlike many prior iterations of primary care capitation, which placed primary care practices at unacceptable financial risk.2327 Minimizing under- and overutilization of such services can be achieved by 1) incorporating evidence-based guidelines for best practices into the practice’s decision-support systems,3336 and 2) by factoring into the pay-for-performance bonus calculation the attainment of consensus goals for cost-effectiveness, efficiency, health outcomes, and patient-centered care. Savings would be stimulated by encouraging best practices and achievement of validated cost-efficiency standards, but not by putting the practice at immediate financial risk for ordering specific tests on a particular patient or for the expenditures of other physicians and providers.

An adjustment to the comprehensive payment might need to be considered when some or all of the responsibility for comprehensive care is transferred to a specialist, as might occur in end-stage renal disease or cancer. Under such circumstances, the specialist might share in or receive the entire payment. Physicians with a specialty who wish to provide comprehensive primary care could participate in the new model if their practices meet advanced medical home standards. Such participating specialists who also perform unique procedural or other services that make a referral unnecessary might be paid an additional reduced fee-for-service payment under selected circumstances. Similarly, primary care physicians might be eligible for fee-for-service reimbursements for some services typically performed by specialists (e.g., skin biopsy).

As alluded to earlier, installation and operation of an interoperable electronic health record (EHR) would be an essential requirement for practices desiring to participate in the comprehensive payment system, given the EHR’s central role in transforming primary care practice30,3337 and in facilitating auditing of care and outcomes. The costs of installation and operation of such a system are considered a legitimate and important component of the comprehensive payment (Table 1).

TESTING AND IMPLEMENTING THE MODEL

Pilot studies conducted in a variety of practice settings and involving a wide spectrum of patients will be needed to validate the proposed model. These investigations will require the collaborative efforts of the physician community, payers, purchasers, and patients, and should utilize an independent research group for data collection, monitoring, and analysis of clinical and economic outcomes. Medicare, as the largest payer and the one whose RBRVS system has been emulated by most other payers, should take the lead and, ideally, collaborate with other large payers to permit a true test of the new payment model.

Demonstration studies will need to address panel size, case mix, and levels of staffing, factors which affect the amount of time available and required for patient care activities (whether office visits, phone calls, record review, or team meetings), key determinants of patient and professional satisfaction.21,47 A range of panel sizes, cases mixes, payment levels, and staffing levels are possible (see Table 3). No single formula is likely to suffice for all settings and populations, but the common denominator needs to be adequate resources to support a comprehensive primary care effort.

There will be formidable research design issues, especially for controlled trials. The first studies might simply test feasibility. These would provide basic observational data comparing financial and patient outcomes pre and post change in reimbursement and relating those outcomes to the various practice models and patterns chosen by participating physicians.

Assuming that pilot studies of the model show promise, the subsequent challenge will be implementation. Most primary care practices do not have the necessary teams or systems in place; new monies will be needed to establish them. The proposed comprehensive payment makes available the financial resources for “tooling-up.” Those primary care practices unsure of being able to make the transition directly to the comprehensive payment system might prefer more evolutionary steps (see below).

STRENGTHS, WEAKNESSES, AND IMPLICATIONS

This payment reform proposal represents an attempt to realign compensation with the primary care mission, providing comprehensive payment for comprehensive care. It frees practices from the growing inadequacy, irrationality, and administrative burdens of the existing RBRVS-based payment system,3,23,28,4951 uncoupling primary care compensation from that of proceduralists, eliminating the zero-sum budgeting game, and overcoming the constraints of a payment system favoring procedure-based care. The model makes possible new payment rules better tailored to the primary care mission and more enabling of practice transformation. It has the potential to establish a new social contract, correcting chronic underpayment in return for accountability and achieving important health outcomes. It acknowledges in explicit financial terms the value primary care can create when properly organized and delivered, an obviously important factor in attracting new physicians to the field and stimulating practice transformation.

Proposing a comprehensive, aggregate payment is likely to evoke memories of primary care capitation with its pejorative connotations; however, there are important differences (Table 2), which lead us to avoid using the term “capitation” to describe our system. The most important differences are risk/needs-adjustment, paying for performance to guard against underservice, and budgeting sufficient monies to support teams and infrastructures. These features are essential to avoiding the withholding of necessary care and the shunning of complex patients that too often occurred under the capitation initiatives of the past decade.2527 In the new model, the gatekeeping of capitation is replaced by coordination and advocacy. Financial risk is borne predominantly by payers, who have the requisite actuarial and capital resources. Nonetheless, practices remain financially accountable, having to work within a global budget, adhere to professional standards of care and referral, and eliminate waste and inefficiency.

This model has some similarities and important differences with salaried models (Table 2). Like salaried models, there is no incentive to inflate the volume of face-to-face visits, but salaried models often have lacked the element of a social contract between the personal physician and the patient, supported in our model by a patient contribution to the retainer. In salaried environments, physicians tend to consider the organization as having the principal accountability to the patient; this has been reflected in lower patient trust of the individual physician.5254

Risk adjustment is a key element of our payment model, both for determining the size of the aggregate monthly payment and for setting pay-for-performance goals. The first operational models of risk/needs adjustment (referred to as “case-mix adjustments”) were diagnosis-based and validated for ambulatory care.55 Because primary care consists largely of ambulatory services, risk adjustment based on diagnoses is likely to be a reasonable approach to predicting the subsequent need for primary care services. Further iterations of risk adjustment by diagnosis have been operational for modifying payments at the health plan level56 and, if modified for application to the practice level, they should facilitate matching payment to care burden. Existing models based on diagnoses 5759 would seem a good fit for the payment system we have outlined. As noted earlier, a validated risk-adjustment framework that incorporates the full spectrum of important risk determinants, including those accounting for patient behaviors45 will be needed.

In a cost-conscious society, it is unlikely that the new payment model will be adopted widely if it is viewed as a give-away to primary care physicians/practices. Conversely, primary care physicians are likely to reject the model if it appears to be yet another attempt to use them as gatekeepers or insurance companies. Our model tries to avoid both pitfalls by assigning most of the financial/actuarial risk to insurers while recognizing the responsibility of primary care practices to be financially and clinically accountable.

To put the financial challenge posed by the model in perspective, it is useful to consider the changes in total health spending that might result from implementation of our payment system and the savings that would be needed to offset them. Physician services currently constitute approximately 25% of all national spending for personal health services; of that amount, depending on the demographics of the population served, a quarter to a third constitutes payment for primary care services60. Thus, only 6–8% of total spending for personal health services currently represents payments to primary care physicians. If we propose a modest comprehensive payment schedule (e.g., an average of $500 per patient per year; see Table 1), it would immediately increase total spending by 2–3%, necessitating a 3%+ reduction in the remaining 88–90% of personal health care spending to offset the increase.

Current estimates of wasteful spending are as high as 30% of total expenditures.61 Studies of electronic medical record systems with decision-support capacity (an essential feature of our payment/practice reform model) have demonstrated substantial savings from reductions in medical errors, pharmacy costs, adverse drug events, unnecessary radiology and laboratory utilization, and avoidable hospital admissions.3336 Additional cost savings in care of the frail elderly are a reasonable expectation from improvements in coordination of care,3032 particularly among the 30% of Medicare beneficiaries who have 4 or more chronic conditions and account for almost 80% of annual program spending.62 Even if our transformed primary care practices eventually achieve only a small fraction of the potential savings, that amount should offset the 2–3% projected increase in costs. Determining financial impact is a critical reason to test the model in pilot study.

In the short run, budget neutrality is unlikely and should not be expected, because upfront investments in practice reorganization and systems will take time to generate the expected savings. Nonetheless, the comprehensive payment model should provide readily apparent early benefits: First and foremost, patients (especially the elderly and the complexly ill) should notice improved access to care made possible by improvements in staffing, scheduling, and infrastructure (unlike concierge practices, which rely heavily on reducing panel size to improve access.63,64 With team practice freeing up the primary physician to perform more thorough patient evaluations, there should be less resorting to otherwise unnecessary tests and referrals. Other early benefits should include those mentioned earlier associated with implementation of the electronic health record.3537 Rapid improvement in care is possible, as suggested by results in the first year of implementing an ambitious pay-for-performance bonus program for British general practitioners.65

Administrative burden has been a major criticism of the RBRVS system.28,4951 Implementation of our model would eliminate claims billing and receiving as well as the onerous documentation and coding requirements associated with RBRVS, supplanting these purely administrative activities with more clinically relevant assessments of practice operations, patient panels, and clinical outcomes. By insisting on an electronic administrative/clinical infrastructure for practice participation in this payment reform, the model makes possible automated audits that should be less disruptive and lower in cost than the administrative demands of RBRVS. Despite the expected lessening of administrative burden, the benefits will not be realized until an electronic health record has been installed, which can be daunting for a small primary care practice, necessitating careful transition planning and budgeting for the changeover.

There is potential for abuses with this model, including misallocation of the comprehensive payment, gaming of the risk/needs-adjustment process, and “dumping” of care onto specialists. These require built-in countermeasures. Siphoning off payments targeted for team salaries and information infrastructure to enrich physician pay can be avoided by developing disbursement guidelines (e.g., Tables 1 and 3) and measures, which can be audited periodically. Use of validated objective measures of risk and need (e.g., principal diagnoses, ejection fraction, creatinine clearance, patterns of care utilization), and independent, random audits of the practice’s electronic health record database should minimize chances of manipulating the risk-adjustment process. Dumping can be reduced by mandating sharing or outright transfer of the comprehensive payment when the specialist assumes most of the responsibility for care and by profiling the referral patterns of physicians, making payment adjustments where overreferrals are occurring. A payment that is adequately risk-adjusted is in itself a powerful disincentive to inappropriate transfer of patients to a specialist. Underutilization of referrals should be discouraged by the clinical outcomes and patient-experience components of our bonus payment determination.

An unintended consequence of this model’s implementation might be practice downsizing. Practices that grew excessively large to meet expenses under fee for service might be tempted to “right-size” their panels to improve care and qualify for performance bonuses (see Table 2). If widely adopted, this could paradoxically reduce access for some64 and trigger a temporary shortage of primary care physicians. Alternatively, the expanded primary care team made possible by our model provides a means of devoting more attention to patients without the need to downsize (which can be painful for both patients and physicians). Moreover, by eliminating the disincentive to care for the complexly ill and needy, we are likely to improve access for those who need it most. With income independent of visit volume, smart delivery strategies (e.g., team care, interoperable medical records, email access, group visits, web-based patient education) can be implemented to reduce individual physician and team workloads, leaving more time for high-value face-to-face encounters, even home visits. Any shortage of primary care physicians that results from initial implementation of our model should be short-lived, as medical school graduates are attracted to the field by the promise of a financially secure, professionally satisfying career and practice environment.

Debate over implementation strategies will be vigorous. The potential for practice and system disruptions from payment reform cause some to argue for incremental approaches, such as increasing the valuation of RBRVS evaluation and management codes or adding a supplemental case management payment for care of high-risk patients.28,48 However, recent experience with capitation showed that if fee-for-service (e.g., under a modified form of RBRVS) continues, then even a comprehensive payment for some patients in a predominantly fee-for-service environment may not alter behavior; practices continue to reward “productivity” (as defined by number of patients seen and procedures performed) rather than develop other performance measures.6668 Adding pay-for-performance to fee-for-service might help counter this behavior, but comprehensive pay for comprehensive care has the potential to be a more straightforward and effective approach to achieving the desired outcomes. Comparative studies are needed.

Primary care in the United States stands at a crossroads. We believe taking the road to recovery requires fundamental reform. It is urgent that new models of payment and practice be developed, tested, and implemented.