Research has shown wide variation in per capita spending among different states and among different counties within the same state. Some of this variation can be explained by the health status of the population, local pricing, patient cultural and demographic factors, and the local liability environment.

However, the vast majority of variation in spending is unexplained and likely due to a failure of health care providers to follow established best practice guidelines. This type of variation is associated with unnecessary over-utilization, while reducing variation leads to reductions in utilization and improvements in quality.

Crystal Run Healthcare in New York State implemented a program to reduce practice variation through development of a robust best practice library grounded in published clinical guidelines and expert opinion from professional organizations. The program has allowed Crystal Run to improve quality and patient access while lowering cost for outpatient care. In doing so, it has also increased revenues for the physicians in the practice.

About Crystal Run Healthcare

Crystal Run Healthcare is a mission driven, physician-owned, multispecialty medical group with nearly 350 physicians. The practice consists of over 30 office sites across the lower Hudson Valley of New York State, extending into Manhattan. Crystal Run serves approximately 250,000 patients with a catchment area extending into four states, performing over 1,000,000 patient visits annually across 40 specialties.

In keeping with its mission to provide high-quality, patient-centered care, Crystal Run implemented electronic health records in 1999. In 2006, the group earned Joint Commission Accreditation for Ambulatory Care Services, and in 2009 it became a National Committee for Quality Assurance (NCQA) Level III Patient Centered Medical Home.

After passage of the Affordable Care Act (ACA), the organization embraced the concept of value-based payments, becoming one of the original 27 Acccountable Care Organizations (ACOs) to participate in the Medicare Shared Savings Program. Crystal Run has also engaged many of its commercial payers and managed Medicaid contracts in transitioning to shared savings, with the long term strategy of assuming full risk for its population and expanding its health insurance product, Crystal Run Health Plans.

Practice Variation Reduction At Crystal Run

Crystal Run began its practice variation reduction program with the pilot diagnosis of diabetes. There were three steps in the process.

First, the organization calculated the total cost per person per year for patients with diabetes and then segmented those costs into professional, laboratory, radiology, and procedure charges.

The second step was to compare the cost of care among the providers treating diabetes. The group did this in a transparent manner so that each physician could see how he or she compared to colleagues. The final step was to analyze the variation and determine its source.

The results of this effort are based on the set of providers who were at Crystal Run for the one year duration of the pilot. We found that variation in cost among providers was not related to severity of illness, as such variation was identical among all patients, regardless of the number of diabetic comorbidities they had (e.g., renal insufficiency, coronary artery disease, hypertension, hyperlipidemia).

Likewise, variation was not related to clinical quality; there was no correlation between the cost of care and the percentage of a provider’s patients with Hemoglobin A1c values under seven. Instead, the variation was rooted in failure to adhere to established best practice guidelines. Once providers were made aware of the evidence-based standards, saw how their individual practice style compared to their peers, and were encouraged to follow the established guideline, variation was reduced.

Reducing The Cost Of Care

As shown in Exhibit 1, Crystal Run Healthcare reduced the cost of treating diabetes over a six month period by 9 percent, or almost $450,000 (standardized to Medicare dollars). The providers on the left side of the graph reduced cost by eliminating unnecessary services. The providers on the right side of the graph appropriately increased cost by addressing underutilization. However, for most diagnoses, the net effect is lower health care costs.

Building on the success of the diabetes pilot, Crystal Run‘s Clinical Transformation Officers charged every medical and surgical specialty with performing a similar exercise, instructing them to choose a diagnosis that was both common and amenable to the development of an evidence-based best practice guideline.

The 15 chosen diagnoses—listed in Exhibit 2—were indeed prevalent. Approximately 97 percent of the practice’s distinct adult patients had at least one of these 15 diseases addressed during a visit in that year. Over the course of the year, variation among providers and costs was reduced for 14 of the 15 disease categories, resulting in a total savings of almost $4.2 million. The largest reductions were seen in diabetes, hypertension, hyperlipidemia, and thyroid nodules.

Divisions now meet quarterly to participate in a variation reduction exercise for the diagnosis of their choice. This has resulted in the development of dozens of new best practice guidelines that have been posted on the intranet for easy reference at the point of care.

Other recent achievements include: a 24 percent reduction in charges for iron infusions used to treat iron deficiency anemia by standardizing which intravenous formulation is administered; a 76 percent reduction in laboratory charges for dysuria in urgent care due to the creation of a urinary tract infection protocol; and, a 17 percent reduction in charges for the evaluation of abnormal liver function tests.

Improving Access And Quality

Analysis revealed that the largest source of variation for many of the target diagnoses related to professional charges. Often, this was due to differences in the average number of visits per patient per year for a given diagnosis among providers within the same specialty, and not differences in coding patterns or the distribution of new patients and consults.

In an effort to curtail this form of unnecessary variation, providers were instructed to include recommended follow-up frequency in any best practice guideline created. For some chronic conditions, such as diabetes, hypothyroidism, and heart disease, such recommendations already existed.

For chronic conditions where practice guidelines do not take a stance regarding follow-up frequency, Crystal Run physicians chose a standardized interval by balancing expert opinion and the currently practiced standard of care in the community.

Adherence to diagnosis-specific recommendations for follow-up visits created additional capacity within the organization. For the original 15 diagnoses chosen for variation reduction, nearly 13,000 visits were eliminated between 2010 and 2012. The largest reductions were seen in diabetes, hypertension, renal masses, and chronic obstructive pulmonary disease.

According to the American Academy of Family Physicians, the average physician provides 3,600 patient visits per year. Therefore, the elimination of 13,000 office visits created capacity equivalent to that of 3.6 additional physicians.

A similar phenomenon was seen throughout the entire practice population, not just among patients with one of the original 15 diagnoses. Across the organization, visits per patient were reduced from 3.7 to 3.4 between 2010 and 2012. As a result, the practice eliminated 42,000 visits. These reductions have led to the capacity equivalent of 12 additional physicians, improving access for existing patients and facilitating the growth of the practice in scope and geography.

Crystal Run’s primary objective in promoting increased adherence to practice standards was to improve the quality of clinical outcomes. Since the onset of the variation reduction program, the practice has remained above the NCQA’s benchmark goals for the control of blood sugar levels in patients with diabetes and blood pressure control in patients with hypertension.

Cholesterol levels in patients with coronary artery disease and hyperlipidemia improved and surpassed the NCQA benchmark goal. Furthermore, hospitalizations per 1,000 patients for the pilot diagnosis of diabetes decreased by 15 percent for the Medicare population between 2010-2013, despite the reduction in visits per patient and reductions in cost. So reducing practice variation did improve clinical quality, and the reduction in costs and increased available capacity were simply side effects of the process.

Improving Profitability While Reducing Utilization

Between 2010, the year before the inception of Crystal Run’s variation reduction program, and 2013, overall receipts per patient were reduced by 14.5 percent, translating to a savings for payers of roughly $22.6 million. At the time, although some of the organization’s payers were transitioning to risk based contracting, the vast majority of payers continued to pay fee-for-service.

Therefore, to remain financially stable during the transition in payment models, Crystal Run Healthcare focused on filling freed capacity with new patients. The number of patients increased by 30 percent between 2010-2013, with an increase of 11.6 percent for primary care providers, 11.3 percent for medical specialties, 19.8 percent for surgeons, and 32.7 percent for OBGYNs (all data are for providers who were practicing at Crystal Run Healthcare continuously between 2010 and 2013).

At the onset of this process, some providers were concerned about reduced compensation due to reduced patient volume, particularly the procedural based specialists. However, many were hopeful that by creating new capacity, they would be able to grow panel size and care for more complicated patients, translating to increased revenue per physician.

The percentage of patients seen for a moderate or high risk visit dramatically increased between 2011 and 2013 for many of the organization’s busiest specialties, with increases of 8.5 percent for cardiology, 16.6 percent for neurology, 28.3 percent for orthopedics, 28.2 percent for pulmonology, and 62.2 percent for hematology/oncology.

The net result of growing panel sizes and caring for sicker patients is an increase in total practice receipts of 11.1 percent. Individual physician compensation has increased for most partners and employed physicians between 2010 and today, despite national trends to the contrary.

Furthermore, continuing medical education programs and adherence to best practice guidelines empowered primary care physicians to expand the initial treatment for many conditions that were once referred to procedural specialists, such as orthopedic consultation for osteoarthritis. This translated to fewer visits needed to book one procedure for these specialists, leading to increased physician satisfaction.

Policy Implications

Crystal Run’s experience clearly illustrates that standardizing care not only reduces health care cost, but also increases a practice’s capacity to care for a growing patient population. There are several implications for policymakers and for public and private payers. First, states, particularly those concerned with physician shortages, should promote evidence-based best practice guidelines for the evaluation and treatment of chronic diseases.

Second, providers should be accountable for adherence to such guidelines through public reporting. Taking it one step further, payers could tier providers based on their ability to deliver evidence-based, standardized care, with lower copays assigned to top tier providers. Furthermore, payers should reward providers that successfully reduce variation in care by sharing in the savings achieved.

Lastly, payers should transition from a transaction based, fee-for-service payment model to one that rewards outcomes and total patients under care. Such a transition will incentivize medical practices not only to eliminate unnecessary care, but also to find innovative ways to care for patients.

Examples include telephone protocols for low-risk, sub-acute conditions such as urinary tract infections and upper respiratory infections; scheduled telephone visits or asynchronous email visits to review a diabetic’s blood sugar records; and, group visits to provide patient education for gestational diabetes or congestive heart failure.

There are many challenges that face physicians today, particularly those in independent medical practices such as Crystal Run Healthcare. However, we believe that the process of variation reduction will allow practices to achieve the Triple Aim while simultaneously improving their financial strength and that of the health care system in general.

Exhibit 1: Reduced Variation In Health Care Costs For Diabetes, Crystal Run Healthcare, 2010 vs. 2012

Hines-Exhibit 1_tg

Exhibit 2: Crystal Run Healthcare Variation Reduction Program — Chosen Diagnoses

Hines-Exhibit 2_tg