Health Reform and Delivery System Organization

August 2010

Identifying ways of providing both effective and efficient healthcare is the holy grail of health services research. The recently passed Patient Protection and Affordable Care Act (PPACA, P.L. 111-148 and P.L. 111-152) calls for delivery system reforms that promote coordinated, accountable, high-quality and low-cost care. Mechanisms of providing such services included accountable care organizations (ACOs), medical homes, and payment for care coordination. Several HCFO grantees are studying delivery system strategies designed to engage providers and provider groups to achieve the goals of PPACA. Lessons learned from this research can inform efforts to improve the quality and efficiency of healthcare as more Americans gain health insurance coverage.

In anticipation of these changes, the Robert Wood Johnson Foundation’s Health Care Financing and Organization (HCFO) initiative convened a panel of three of its grantees on “Health Reform and Delivery System Organization” at AcademyHealth’s Annual Research Meeting (ARM) in Boston, MA, June 26-28, 2010. Each speaker discussed preliminary findings from their respective efforts to identify characteristics of effective and efficient systems of care in light of PPACA. Kelly Devers, Ph.D. from the Urban Institute moderated the discussion, which featured John Kralewski, Ph.D. and Douglas Wholey, Ph.D., M.B.A. from the University of Minnesota, and James Reschovsky, Ph.D., M.P.P., from the Center for Studying Health System Change in Washington, D.C.

Accountable Care Organizations

The concept of accountable care organizations (ACOs) has been touted as a potential solution to the fragmentation and overuse of medical services. ACOs “consist of providers who are jointly held accountable for achieving measured quality improvements and reductions in the rate of spending growth.”1 The driving force behind ACOs is the idea of shared patients that are referred within partnering medical groups, hospitals and specialists. ACOs are a mechanism to support value-based purchasing and align payment incentives with the quality of care and resource use.

ACOs are comprised of Accountable Primary Care Groups (APCGs) and Accountable Primary Care Systems (APCSs). By identifying APCGs working within APCSs, researchers can reliably measure quality and resource use. Ideally, APCGs should be a small and cohesive working unit in order to promote care coordination and drive the culture change necessary for accountability. Financial policies can incent providers as a group and reward overall performance, which is the most effective strategy for improving care. Small, closed panels of physicians also circumvent the challenge of attribution, which researchers face when studying larger practices. Attribution refers to the ability, within a system of integrated care, to appropriately link a patient’s health outcomes to particular providers for the purpose of financial accountability. APCGs can scale up into full medical homes, or simply augment primary care with social services.

Douglas Wholey, David Knutson and colleagues at the University of Minnesota have developed a technique for identifying ACPGs and ACPSs based on “network methodology.” The procedure involves creating matrices that array primary care providers against hospital specialists and other primary care providers. By examining the resulting vectors, researchers can locate provider “clusters” that serve the same patients. This approach to identifying potential candidates for ACOs works because patients tend to group around common providers based on reputation, insurance coverage, and proximity. In some instances these ACPGs and ACPSs will be existing organizations; in other situations they will be synthetic. In order to effectively use the network analysis instrument for locating ACOs, researchers will need more provider-level data from Medicaid, which is currently sparse.

The Characteristics of Best Practices

Current research suggests that the cost and quality of care vary across geographical areas and health plans within geographical areas. The causes of this variation at the practice level, however, are not well understood. Several factors may influence the quality and efficiency of medical practices, such as the size, ownership, and composition of the organization, market supply and structure, and payment polices. Two HCFO grantees sought to examine contributors to quality and cost at the practice level. 

John Kralewski and colleagues at the University of Minnesota set out to identify “best medical group practices” that provide high quality care at low cost and determine cultural and structural factors driving their success. They analyzed claims data in a sample of 59 medical group practices with 263,000 enrollees to estimate the relationship among structure, process, and cost measures. There was wide variability in the number of physicians per practice (10-1,500) as well as the ownership and specialty concentrations within the practices. There is also a considerable range in costs per patient per year and quality indicators. Using frontier analysis, researchers found only 7 practices that met the criteria for high quality and low cost. There were no common characteristics among these practices, nor where there recurring themes among the worst performing practices. Essentially, quality was not correlated with cost, and after a certain threshold more spending actually may have decreased quality. 

Preliminary findings suggest that the most efficient physician-owned practices have more functionality in electronic health records (EHRs) than the least efficient physician-owned practices. Efficient practices also make greater use of registries than inefficient practices. All best practices employ nurse practitioners and rank highly on measures and quality, information, and culture. There is a sense of feedback and learning from mistakes, distributing information to and discussing information with physicians, and cooperation in successful clinics. The investigators conclude that health care costs can be reduced by 20 percent at a provider level without eroding the quality of or access to care.

At the Center for Studying Health System Change (CSHSC), James Reschovsky and colleagues looked to identify actionable factors associated with treating high-cost Medicare patients. They explored the influence of the primary care provider (PCP) market, care patterns, and Medicare fee generosity on cost to the Medicare program. Using the 2004-2005 Community Tracking Study Physician Survey cross-referenced with claims data from the Center for Medicare and Medicaid Studies (CMS), researchers calculated the relationship between market, physician, patient, and policy variables on standardized costs. They found that patient and demographic factors explained the vast majority of variation for both high-cost and low-cost groups, and that patients receiving care in multiple census areas accounted for higher spending. Few physician-level contributors were significant, but patients with a medical specialty as their usual source of care and providers that reported inadequate time for office visits were associated with greater costs. The market supply of providers was at best weakly related to costs, whereas the percentage of for-profit providers was linked with higher costs but low elasticities. The county mean fragmentation score was associated with greater expenditures, but only in low-cost beneficiaries. Higher reimbursement rates were linked to higher spending, especially in low-cost beneficiaries.

In terms of overall findings and implications, the researchers found few actionable factors related to Medicare spending. They observed that a high percentage of beneficiaries received care in multiple census divisions, which could complicate payment and delivery reforms, such as pay-for-performance (P4P) and ACOs since provider attribution would be difficult. This underscores the need for patient-centered medical homes (PCMH) as well as patient incentives to improve quality and lower costs. The physician-reported inadequacy of time with patient as a cost driver also lends support to PCMH model with financial incentives to coordinate care. Physician payment reform can alter treatment patterns, but incentives need to be significant. 

Reform efforts may not be able to achieve their objectives simply through reduced care fragmentation, especially among high-cost Medicare beneficiaries. In the most expensive patients, interventions must be targeted to the very sick and complex, and there is no “one size fits all” approach. High-cost chronic conditions may be better prevented through lifestyle changes on the part of the patient. Primary care providers can help to manage chronic conditions, but the time frame between the intervention and results is important. Ultimately, if changes are applied to the entire healthcare system to be more patient-oriented and proactive for preventing chronic conditions, spending in high-cost beneficiaries may see future reductions.


Critical to ensuring the success of health reform is a well-functioning delivery system. HCFO-funded studies help to inform policies that promote the development of new models for delivering high quality care.

Details on these and other related HCFO studies are available at

Title: Accountable Care Teams for Disabled Medicaid Beneficiaries
Grantee Institution: University of Minnesota
Principal Investigator: David J. Knutson, M.S. and Douglas Wholey, Ph.D.
Grant Period: November 1, 2008–July 31, 2009

The researchers identified health care providers who share a common patient population (accountable care teams), using a technique known as “network methodology.” These teams could serve as a medical home for their patients. The process of identifying accountable groups of providers could also support value-based purchasing. The researchers employed the network methodology to identify provider “clusters” – those who are densely interconnected through serving a common patient population, in this case disabled Medicaid patients in Minnesota. They then (1) assessed the “fit” of the clustering with observations about care organization in Minnesota, and (2) developed measures of cluster cohesiveness. The objective of the project was to introduce a new technology for identifying accountable groups of providers that can be the basis of medical homes/care teams and inform efforts to implement value-based purchasing of health care.

Title: The Characteristics of Best Medical Practices
Grantee Institution: University of Minnesota
Principal Investigator: John Kralewski, Ph.D.
Grant Period: December 1, 2008–September 30, 2010

The researchers will identify the organizational characteristics of medical group practices that achieve high quality, low cost care. Current research suggests that the cost and quality of care vary across geographic areas and health plans within geographic areas. The causes of this variation at the practice level, however, are not well understood. Using a sample of physician practices from Medical Group Management Association (MGMA) membership, the researchers will examine how financial incentives within the practice, structural, and cultural attributes at the practice-level influence costs and quality of care, and the linkages between cost and quality. They will examine the effects of practice size, physician workload, and collegiality on cost and quality independently and then examine the effects on cost and quality jointly. Finally, they will calculate the configuration that results in the optimal performance of best medical practices. The objective of this project is to provide benchmarks that policymakers and health insurance plan and medical practice administrators can use to promote a cost-effective health care system.

Title: Cost and Efficiency in Treating High-Cost Medicare Beneficiaries: The Role of Physician Practice and Health System Factors
Grantee Institution: Center for Studying Health System Change
Principal Investigator: James D. Reschovsky, Ph.D.
Grant Period: March 1, 2008–August 31, 2009

The researchers will examine key physician practice and market characteristics that may contribute to high costs and inefficient care in the Medicare program. The study is composed of three phases. In phase one, they will analyze the treatment of high-cost Medicare beneficiaries in order to identify key physician, practice, and market characteristics associated with differences between actual and predicted Medicare payments and medical care use. In phase two, they will examine whether the factors associated with greater than predicted resource use affect high-cost beneficiaries’ health outcomes. Finally, the researchers will examine possible sources of geographic cost variations for high-cost beneficiaries and the extent to which these variations reflect differences in patient characteristics or supply-related factors and practice patterns of providers in a particular region. The objective of this project is to identify potential policy levers that can influence cost effectiveness in the delivery of medical care to high-cost Medicare patients.

Title: Resource Use and Efficiency in Episodes of Care
Grantee Institution: Palo Alto Medical Foundation Research Institute
Principal Investigator: Hal Luft, Ph.D.
Grant Period: July 1, 2008–September 30, 2009

The researchers sought to examine some of the underlying assumptions of episode-based payments, which are hypothesized to encourage more clinically and economically efficient practices by primary care physicians (PCPs). Specifically, they assessed whether episode-based measures of resource use at the individual PCP level, rather than the physician group or medical staff level, are statistically reliable and appropriate. Using data from the Palo Alto Medical Foundation (PAMF), a large multi-specialty physician group that uses electronic medical records (EMRs), the researchers examined whether some PCPs have practice patterns significantly more (or less) expensive than the average at either the episode level or with groups of acute or chronic episodes. The researchers explored the role of components, such as PCP office visits, referrals, imaging, lab tests, and drugs, in these differences in PCP practices. They also studied clinicians’ explanations for differences in practice patterns, such as unmeasured severity, location, or other factors. They compared PAMF-based patterns of care with overall patterns at an episode level (but not physician level) from a large national data set. The objective of the study was to determine whether consistent styles of practice across PCPs within a large medical group can be detected, which would help inform policymakers about whether physician-oriented incentives are worth pursuing.

Title: Identifying Best Practices in the Coordination of Care
Grantee Institution: Center for Studying Health System Change
Principal Investigator: Ann S. O’Malley, M.D., M.P.H.
Grant Period: October 1, 2007 – May 31, 2009

The researchers examined how care is coordinated in ambulatory care settings. Specifically, they identified and documented “best practices” in physician offices that have developed care coordination processes and determined the financial implications of increased coordination. For example, the researchers assessed whether a periodic care coordination fee or itemized billing for coordination activities was more efficient. They also examined a group of “average practices” to assess how they set priorities for coordination activities and what barriers they encountered. The objective of the proposed project was to better inform the replication of organized care coordination processes in medical practices.

1McClellan M et al. 2010. A National Strategy to Put Accountable Care Into Practice. Health Affairs, 29(5): 982-990.